What does your patient need to know at the first visit?  

A patient may be scheduled for a total-body skin examination (TBSE) through several routes: primary care referral, continued cancer screening for an at-risk patient or patient transfer, or patient-directed scheduling for general screening regardless of risk factors. At the patient's first visit, it is imperative that the course of the appointment is smooth and predictable for patient comfort and for a thorough and effective examination. The nurse initially solicits salient medical history, particularly personal and family history of skin cancer, current medications, and any acute concerns. The nurse then prepares the patient for the logistics of the TBSE, namely to undress, don a gown that ties and opens in the back, and be seated on the examination table. When I enter the room, the conversation commences with me seated across from the patient, reviewing specifics about his/her history and risk factors. Then the TBSE is executed from head to toe.  

Do you broadly recommend TBSE? 

Firstly, TBSE is a safe clinical tool, supported by data outlining a lack of notable patient morbidity during the examination, including psychosocial factors, and it is generally well-received by patients (Risica et al). In 2016, the US Preventative Services Task Force (USPSTF) outlined its recommendations regarding screening for skin cancer, concluding that there is insufficient evidence to broadly recommend TBSE. Unfortunately, USPSTF findings amassed data from all types of screenings, including those by nondermatologists, and did not extract specialty-specific benefits and risks to patients. The recommendation also did not outline the influence of TBSE on morbidity and mortality for at-risk groups. The guidelines target primary care practice trends; therefore, specialty societies such as the American Academy of Dermatology issued statements following the USPSTF recommendation outlining these salient clarifications, namely that TBSE detects melanoma and keratinocyte carcinomas earlier than in patients who are not screened. Randomized controlled trials to prove this observation are lacking, particularly because of the ethics of withholding screening from a prospective study group. However, in 2017, Johnson et al outlined the best available survival data in concert with the USPSTF statement to arrive at the most beneficial screening recommendations for patients, specifically targeting risk groups--those with a history of skin cancer, immunosuppression, indoor tanning and/or many blistering sunburns, and several other genetic parameters--for at least annual TBSE. 

The technique and reproducibility of TBSE also are not standardized, though they seem to have been endearingly apprenticed but variably implemented through generations of dermatology residents going forward into practice. As it is, depending on patient body surface area, mobility, willingness to disrobe, and adornments (eg, tattoos, hair appliances), multiple factors can restrict full view of a patient's skin. Recently, Helm et al proposed standardizing the TBSE sequence to minimize omitted areas of the body, which may become an imperative tool for streamlined resident teaching and optimal screening encounters.  

How do you keep patients compliant with TBSE? 

During and following TBSE, I typically outline any lesions of concern and plan for further testing, screening, and behavioral prevention strategies. Frequency of TBSE and importance of compliance are discussed during the visit and reinforced at checkout where the appointment templates are established a year in advance for those with skin cancer. Further, for those with melanoma, their appointment slots are given priority status so that any cancellations or delays are rescheduled preferentially. Particularly during the discussion about TBSE frequency, I emphasize the comparison and importance of this visit akin to other recommended screenings, such as mammograms and colonoscopies, and that we, as dermatologists, are part of their cancer surveillance team. 

What do you do if patients refuse your recommendations? 

Some patients refuse a gown or removal of certain clothing items (eg, undergarments, socks, wigs). Some patients defer a yearly TBSE upon checkout and schedule an appointment only when a lesion of concern arises. My advice is not to shame patients and to take advantage of as much as the patient is able and comfortable to show us and be present for, welcoming that we have the opportunity to take care of them and screen for cancer in any capacity. In underserved or limited budget practice regions, lesion-directed examination vs TBSE may be the only screening method utilized and may even attract more patients to a screening facility (Hoorens et al). 

In the opposite corner are those patients who deem the recommended TBSE interval as too infrequent, which poses a delicate dilemma. In my opinion, these situations present another cohort of risks. Namely, the patient may become (or continue to be) overly fixated on the small details of every skin lesion, and in my experience, they tend to develop the habit of expecting at least 1 biopsy at each visit, typically of a lesion of their choosing. Depending on the validity of this expectation vs my clinical examination, it can lead to a difficult discussion with the patient about oversampling lesions and the potential for many scars, copious reexcisions for ambiguous lesion pathology, and a trend away from prudent clinical care. In addition, multiple visits incur more patient co-pays and time away from school, work, or home. To ease the patient's mind, I advise to call our office for a more acute visit if there is a lesion of concern; I additionally recommend taking a smartphone photograph of a concerning lesion and monitoring it for changes or sending the photograph to our patient portal messaging system so we can evaluate its acuity. 

What take-home advice do you give to patients? 

As the visit ends, I further explain that home self-examination or examination by a partner between visits is intuitively a valuable screening adjunct for skin cancer. In 2018, the USPSTF recommended behavioral skin cancer prevention counseling and self-examination only for younger-age cohorts with fair skin (6 months to 24 years), but its utility in specialty practice must be qualified. The American Academy of Dermatology Association subsequently issued a statement to support safe sun-protective practices and diligent self-screening for changing lesions, as earlier detection and management of skin cancer can lead to decreased morbidity and mortality from these neoplasms.  

Resources for Patients

American Academy of Dermatology's SPOT Skin Cancer

Centers for Disease Control and Prevention: What Screening Tests Are There?
 

What does your patient need to know at the first visit?  

A patient may be scheduled for a total-body skin examination (TBSE) through several routes: primary care referral, continued cancer screening for an at-risk patient or patient transfer, or patient-directed scheduling for general screening regardless of risk factors. At the patient's first visit, it is imperative that the course of the appointment is smooth and predictable for patient comfort and for a thorough and effective examination. The nurse initially solicits salient medical history, particularly personal and family history of skin cancer, current medications, and any acute concerns. The nurse then prepares the patient for the logistics of the TBSE, namely to undress, don a gown that ties and opens in the back, and be seated on the examination table. When I enter the room, the conversation commences with me seated across from the patient, reviewing specifics about his/her history and risk factors. Then the TBSE is executed from head to toe.  

Do you broadly recommend TBSE? 

Firstly, TBSE is a safe clinical tool, supported by data outlining a lack of notable patient morbidity during the examination, including psychosocial factors, and it is generally well-received by patients (Risica et al). In 2016, the US Preventative Services Task Force (USPSTF) outlined its recommendations regarding screening for skin cancer, concluding that there is insufficient evidence to broadly recommend TBSE. Unfortunately, USPSTF findings amassed data from all types of screenings, including those by nondermatologists, and did not extract specialty-specific benefits and risks to patients. The recommendation also did not outline the influence of TBSE on morbidity and mortality for at-risk groups. The guidelines target primary care practice trends; therefore, specialty societies such as the American Academy of Dermatology issued statements following the USPSTF recommendation outlining these salient clarifications, namely that TBSE detects melanoma and keratinocyte carcinomas earlier than in patients who are not screened. Randomized controlled trials to prove this observation are lacking, particularly because of the ethics of withholding screening from a prospective study group. However, in 2017, Johnson et al outlined the best available survival data in concert with the USPSTF statement to arrive at the most beneficial screening recommendations for patients, specifically targeting risk groups--those with a history of skin cancer, immunosuppression, indoor tanning and/or many blistering sunburns, and several other genetic parameters--for at least annual TBSE. 

The technique and reproducibility of TBSE also are not standardized, though they seem to have been endearingly apprenticed but variably implemented through generations of dermatology residents going forward into practice. As it is, depending on patient body surface area, mobility, willingness to disrobe, and adornments (eg, tattoos, hair appliances), multiple factors can restrict full view of a patient's skin. Recently, Helm et al proposed standardizing the TBSE sequence to minimize omitted areas of the body, which may become an imperative tool for streamlined resident teaching and optimal screening encounters.  

How do you keep patients compliant with TBSE? 

During and following TBSE, I typically outline any lesions of concern and plan for further testing, screening, and behavioral prevention strategies. Frequency of TBSE and importance of compliance are discussed during the visit and reinforced at checkout where the appointment templates are established a year in advance for those with skin cancer. Further, for those with melanoma, their appointment slots are given priority status so that any cancellations or delays are rescheduled preferentially. Particularly during the discussion about TBSE frequency, I emphasize the comparison and importance of this visit akin to other recommended screenings, such as mammograms and colonoscopies, and that we, as dermatologists, are part of their cancer surveillance team. 

What do you do if patients refuse your recommendations? 

Some patients refuse a gown or removal of certain clothing items (eg, undergarments, socks, wigs). Some patients defer a yearly TBSE upon checkout and schedule an appointment only when a lesion of concern arises. My advice is not to shame patients and to take advantage of as much as the patient is able and comfortable to show us and be present for, welcoming that we have the opportunity to take care of them and screen for cancer in any capacity. In underserved or limited budget practice regions, lesion-directed examination vs TBSE may be the only screening method utilized and may even attract more patients to a screening facility (Hoorens et al). 

In the opposite corner are those patients who deem the recommended TBSE interval as too infrequent, which poses a delicate dilemma. In my opinion, these situations present another cohort of risks. Namely, the patient may become (or continue to be) overly fixated on the small details of every skin lesion, and in my experience, they tend to develop the habit of expecting at least 1 biopsy at each visit, typically of a lesion of their choosing. Depending on the validity of this expectation vs my clinical examination, it can lead to a difficult discussion with the patient about oversampling lesions and the potential for many scars, copious reexcisions for ambiguous lesion pathology, and a trend away from prudent clinical care. In addition, multiple visits incur more patient co-pays and time away from school, work, or home. To ease the patient's mind, I advise to call our office for a more acute visit if there is a lesion of concern; I additionally recommend taking a smartphone photograph of a concerning lesion and monitoring it for changes or sending the photograph to our patient portal messaging system so we can evaluate its acuity. 

What take-home advice do you give to patients? 

As the visit ends, I further explain that home self-examination or examination by a partner between visits is intuitively a valuable screening adjunct for skin cancer. In 2018, the USPSTF recommended behavioral skin cancer prevention counseling and self-examination only for younger-age cohorts with fair skin (6 months to 24 years), but its utility in specialty practice must be qualified. The American Academy of Dermatology Association subsequently issued a statement to support safe sun-protective practices and diligent self-screening for changing lesions, as earlier detection and management of skin cancer can lead to decreased morbidity and mortality from these neoplasms.  

Resources for Patients

American Academy of Dermatology's SPOT Skin Cancer

Centers for Disease Control and Prevention: What Screening Tests Are There?
 

What does your patient need to know at the first visit?  

A patient may be scheduled for a total-body skin examination (TBSE) through several routes: primary care referral, continued cancer screening for an at-risk patient or patient transfer, or patient-directed scheduling for general screening regardless of risk factors. At the patient's first visit, it is imperative that the course of the appointment is smooth and predictable for patient comfort and for a thorough and effective examination. The nurse initially solicits salient medical history, particularly personal and family history of skin cancer, current medications, and any acute concerns. The nurse then prepares the patient for the logistics of the TBSE, namely to undress, don a gown that ties and opens in the back, and be seated on the examination table. When I enter the room, the conversation commences with me seated across from the patient, reviewing specifics about his/her history and risk factors. Then the TBSE is executed from head to toe.  

Do you broadly recommend TBSE? 

Firstly, TBSE is a safe clinical tool, supported by data outlining a lack of notable patient morbidity during the examination, including psychosocial factors, and it is generally well-received by patients (Risica et al). In 2016, the US Preventative Services Task Force (USPSTF) outlined its recommendations regarding screening for skin cancer, concluding that there is insufficient evidence to broadly recommend TBSE. Unfortunately, USPSTF findings amassed data from all types of screenings, including those by nondermatologists, and did not extract specialty-specific benefits and risks to patients. The recommendation also did not outline the influence of TBSE on morbidity and mortality for at-risk groups. The guidelines target primary care practice trends; therefore, specialty societies such as the American Academy of Dermatology issued statements following the USPSTF recommendation outlining these salient clarifications, namely that TBSE detects melanoma and keratinocyte carcinomas earlier than in patients who are not screened. Randomized controlled trials to prove this observation are lacking, particularly because of the ethics of withholding screening from a prospective study group. However, in 2017, Johnson et al outlined the best available survival data in concert with the USPSTF statement to arrive at the most beneficial screening recommendations for patients, specifically targeting risk groups--those with a history of skin cancer, immunosuppression, indoor tanning and/or many blistering sunburns, and several other genetic parameters--for at least annual TBSE. 

The technique and reproducibility of TBSE also are not standardized, though they seem to have been endearingly apprenticed but variably implemented through generations of dermatology residents going forward into practice. As it is, depending on patient body surface area, mobility, willingness to disrobe, and adornments (eg, tattoos, hair appliances), multiple factors can restrict full view of a patient's skin. Recently, Helm et al proposed standardizing the TBSE sequence to minimize omitted areas of the body, which may become an imperative tool for streamlined resident teaching and optimal screening encounters.  

How do you keep patients compliant with TBSE? 

During and following TBSE, I typically outline any lesions of concern and plan for further testing, screening, and behavioral prevention strategies. Frequency of TBSE and importance of compliance are discussed during the visit and reinforced at checkout where the appointment templates are established a year in advance for those with skin cancer. Further, for those with melanoma, their appointment slots are given priority status so that any cancellations or delays are rescheduled preferentially. Particularly during the discussion about TBSE frequency, I emphasize the comparison and importance of this visit akin to other recommended screenings, such as mammograms and colonoscopies, and that we, as dermatologists, are part of their cancer surveillance team. 

What do you do if patients refuse your recommendations? 

Some patients refuse a gown or removal of certain clothing items (eg, undergarments, socks, wigs). Some patients defer a yearly TBSE upon checkout and schedule an appointment only when a lesion of concern arises. My advice is not to shame patients and to take advantage of as much as the patient is able and comfortable to show us and be present for, welcoming that we have the opportunity to take care of them and screen for cancer in any capacity. In underserved or limited budget practice regions, lesion-directed examination vs TBSE may be the only screening method utilized and may even attract more patients to a screening facility (Hoorens et al). 

In the opposite corner are those patients who deem the recommended TBSE interval as too infrequent, which poses a delicate dilemma. In my opinion, these situations present another cohort of risks. Namely, the patient may become (or continue to be) overly fixated on the small details of every skin lesion, and in my experience, they tend to develop the habit of expecting at least 1 biopsy at each visit, typically of a lesion of their choosing. Depending on the validity of this expectation vs my clinical examination, it can lead to a difficult discussion with the patient about oversampling lesions and the potential for many scars, copious reexcisions for ambiguous lesion pathology, and a trend away from prudent clinical care. In addition, multiple visits incur more patient co-pays and time away from school, work, or home. To ease the patient's mind, I advise to call our office for a more acute visit if there is a lesion of concern; I additionally recommend taking a smartphone photograph of a concerning lesion and monitoring it for changes or sending the photograph to our patient portal messaging system so we can evaluate its acuity. 

What take-home advice do you give to patients? 

As the visit ends, I further explain that home self-examination or examination by a partner between visits is intuitively a valuable screening adjunct for skin cancer. In 2018, the USPSTF recommended behavioral skin cancer prevention counseling and self-examination only for younger-age cohorts with fair skin (6 months to 24 years), but its utility in specialty practice must be qualified. The American Academy of Dermatology Association subsequently issued a statement to support safe sun-protective practices and diligent self-screening for changing lesions, as earlier detection and management of skin cancer can lead to decreased morbidity and mortality from these neoplasms.  

Resources for Patients

American Academy of Dermatology's SPOT Skin Cancer

Centers for Disease Control and Prevention: What Screening Tests Are There?
 

NEW ORLEANS – Results of the largest-ever analysis of TAVR in patients with bicuspid aortic stenosis indicate that key 30-day and 1-year outcomes are similar to those of propensity-matched TAVR patients with tricuspid disease, Raj R. Makkar, MD, said at the annual meeting of the American College of Cardiology.

“Select bicuspid anatomy is amenable to TAVR with current-generation, balloon-expandable TAVR technology with acceptable clinical outcomes. These data provide an argument for TAVR to be a reasonable alternative for bicuspid AS [aortic stenosis] patients who are at intermediate or high risk for surgical aortic valve replacement, which are the patients that are enrolled in this registry, and provide a sound basis to conduct a randomized clinical trial in young patients with bicuspid AS who are at low risk for surgery,” declared Dr. Makkar, director of interventional cardiology and the cardiac catheterization laboratory at Cedars-Sinai Medical Center in Los Angeles.

The landmark randomized trials of TAVR versus SAVR (surgical aortic valve replacement) that established TAVR as the preferred treatment for patients with severe aortic stenosis who are at high, intermediate, or low surgical risk systematically excluded patients with bicuspid AS, even though bicuspid anatomy is common, particularly in younger patients with AS.

Despite the absence of supportive randomized trial data, TAVR is being done for bicuspid AS. To learn how patients with bicuspid disease have fared, Dr. Makkar and coinvestigators analyzed the real-world Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy (STS/ACC TVT) Registry of all patients who underwent TAVR with the balloon-expandable Sapien 3 stent in the United States during 2015-2018. They compared outcomes in 2,691 patients with high or intermediate surgical risk who underwent TAVR for bicuspid AS to an equal number of patients who had TAVR for tricuspid disease, with the two groups being propensity-matched across 25 variables.

Key outcomes were reassuringly similar in the two groups. For example, 30-day and 1-year all-cause mortality rates were 2.6% and 10.8% in patients with bicuspid valves and similar, at 2.5% and 12.1%, in those with tricuspid AS. Paravalvular leak rates at 30 days and 1 year were similar in the two groups. The Kansas City Cardiomyopathy Questionnaire scores, reflecting quality of life, improved dramatically – by nearly 30 points – from pre-TAVR baseline in both groups. The proportion of patients who were New York Heart Association functional class III or IV improved from nearly 85% at baseline to about 8% at 30 days and 1 year, again with no significant difference between the bicuspid and tricuspid AS groups. And there were other benefits, too.

“Despite the concerns regarding optimal expansion of these valves in a bicuspid anatomy, what we observed here was a significant and similar reduction in mean gradients and increase in valve area, both in the bicuspid and tricuspid AS patients. So there was no impact of bicuspid anatomy as seen here in terms of valve hemodynamics,” according to the cardiologist.

Conversion from TAVR to open surgery was required in 0.9% of bicuspid and 0.4% of tricuspid AS patients. Rates of aortic dissection and need for aortic valve reintervention were similarly low in both groups.

The 30-day stroke rate was significantly higher in the bicuspid patients – 2.4% versus 1.6% – but by 1 year there was no significant between-group difference, with stroke rates of 3.4% in the bicuspid and 3.1% in the tricuspid TAVR patients.

“I’d like to point out that more than 75% of strokes occurred in the first 3 days. These are periprocedural strokes, and there was no difference in the time distribution of strokes between the bicuspid and tricuspid groups,” Dr. Makkar said.

These stroke data make a compelling case for the routine use of cerebral protection devices in patients undergoing TAVR, something which now occurs in less than 10% of cases nationally, he continued.

“I would argue that, based on these data, it would be wise for us to use cerebral protection devices, especially when we are doing TAVR in patients with bicuspid AS, because their valves tend to be more heavily calcified than is often the case in tricuspid AS,” Dr. Makkar said.

Discussant Mayra Guerrero, MD, of the Mayo Clinic in Rochester, Minn., took issue with Dr. Makkar’s comment regarding the need for a randomized trial of TAVR in bicuspid AS patients with low surgical risk.

“Do we really need a randomized trial when we see in real-world experience with more than 2,600 patients that the outcomes are fairly similar?” she asked.

Affirmative, Dr. Makkar responded, in light of the fact that the STS/ACC TVT Registry doesn’t include low–surgical risk, typically relatively young bicuspid AS TAVR patients.

“I would say that these data are reassuring and encouraging, but we must not get carried away. I think that would be the important message that I must give,” Dr. Makkar replied. “I think for patients who are high risk and who are intermediate risk, with STS scores of what they were here – 5 and more – I think it’s reasonable to consider them for TAVR based upon CT anatomy. For young patients, as I concluded, I think we must do a randomized clinical trial to definitely establish the safety and efficacy in these patients.”

Dr. Makkar reported receiving research grants from and serving as a consultant to Edwards Lifesciences, which supported the study, as well as Abbott, Medtronic, and Boston Scientific.

bjancin@mdedge.com

SOURCE: Makkar RR. ACC 19, 404-15. Late-breaking clinical trials


 

NEW ORLEANS – Results of the largest-ever analysis of TAVR in patients with bicuspid aortic stenosis indicate that key 30-day and 1-year outcomes are similar to those of propensity-matched TAVR patients with tricuspid disease, Raj R. Makkar, MD, said at the annual meeting of the American College of Cardiology.

“Select bicuspid anatomy is amenable to TAVR with current-generation, balloon-expandable TAVR technology with acceptable clinical outcomes. These data provide an argument for TAVR to be a reasonable alternative for bicuspid AS [aortic stenosis] patients who are at intermediate or high risk for surgical aortic valve replacement, which are the patients that are enrolled in this registry, and provide a sound basis to conduct a randomized clinical trial in young patients with bicuspid AS who are at low risk for surgery,” declared Dr. Makkar, director of interventional cardiology and the cardiac catheterization laboratory at Cedars-Sinai Medical Center in Los Angeles.

The landmark randomized trials of TAVR versus SAVR (surgical aortic valve replacement) that established TAVR as the preferred treatment for patients with severe aortic stenosis who are at high, intermediate, or low surgical risk systematically excluded patients with bicuspid AS, even though bicuspid anatomy is common, particularly in younger patients with AS.

Despite the absence of supportive randomized trial data, TAVR is being done for bicuspid AS. To learn how patients with bicuspid disease have fared, Dr. Makkar and coinvestigators analyzed the real-world Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy (STS/ACC TVT) Registry of all patients who underwent TAVR with the balloon-expandable Sapien 3 stent in the United States during 2015-2018. They compared outcomes in 2,691 patients with high or intermediate surgical risk who underwent TAVR for bicuspid AS to an equal number of patients who had TAVR for tricuspid disease, with the two groups being propensity-matched across 25 variables.

Key outcomes were reassuringly similar in the two groups. For example, 30-day and 1-year all-cause mortality rates were 2.6% and 10.8% in patients with bicuspid valves and similar, at 2.5% and 12.1%, in those with tricuspid AS. Paravalvular leak rates at 30 days and 1 year were similar in the two groups. The Kansas City Cardiomyopathy Questionnaire scores, reflecting quality of life, improved dramatically – by nearly 30 points – from pre-TAVR baseline in both groups. The proportion of patients who were New York Heart Association functional class III or IV improved from nearly 85% at baseline to about 8% at 30 days and 1 year, again with no significant difference between the bicuspid and tricuspid AS groups. And there were other benefits, too.

“Despite the concerns regarding optimal expansion of these valves in a bicuspid anatomy, what we observed here was a significant and similar reduction in mean gradients and increase in valve area, both in the bicuspid and tricuspid AS patients. So there was no impact of bicuspid anatomy as seen here in terms of valve hemodynamics,” according to the cardiologist.

Conversion from TAVR to open surgery was required in 0.9% of bicuspid and 0.4% of tricuspid AS patients. Rates of aortic dissection and need for aortic valve reintervention were similarly low in both groups.

The 30-day stroke rate was significantly higher in the bicuspid patients – 2.4% versus 1.6% – but by 1 year there was no significant between-group difference, with stroke rates of 3.4% in the bicuspid and 3.1% in the tricuspid TAVR patients.

“I’d like to point out that more than 75% of strokes occurred in the first 3 days. These are periprocedural strokes, and there was no difference in the time distribution of strokes between the bicuspid and tricuspid groups,” Dr. Makkar said.

These stroke data make a compelling case for the routine use of cerebral protection devices in patients undergoing TAVR, something which now occurs in less than 10% of cases nationally, he continued.

“I would argue that, based on these data, it would be wise for us to use cerebral protection devices, especially when we are doing TAVR in patients with bicuspid AS, because their valves tend to be more heavily calcified than is often the case in tricuspid AS,” Dr. Makkar said.

Discussant Mayra Guerrero, MD, of the Mayo Clinic in Rochester, Minn., took issue with Dr. Makkar’s comment regarding the need for a randomized trial of TAVR in bicuspid AS patients with low surgical risk.

“Do we really need a randomized trial when we see in real-world experience with more than 2,600 patients that the outcomes are fairly similar?” she asked.

Affirmative, Dr. Makkar responded, in light of the fact that the STS/ACC TVT Registry doesn’t include low–surgical risk, typically relatively young bicuspid AS TAVR patients.

“I would say that these data are reassuring and encouraging, but we must not get carried away. I think that would be the important message that I must give,” Dr. Makkar replied. “I think for patients who are high risk and who are intermediate risk, with STS scores of what they were here – 5 and more – I think it’s reasonable to consider them for TAVR based upon CT anatomy. For young patients, as I concluded, I think we must do a randomized clinical trial to definitely establish the safety and efficacy in these patients.”

Dr. Makkar reported receiving research grants from and serving as a consultant to Edwards Lifesciences, which supported the study, as well as Abbott, Medtronic, and Boston Scientific.

bjancin@mdedge.com

SOURCE: Makkar RR. ACC 19, 404-15. Late-breaking clinical trials


 

NEW ORLEANS – Results of the largest-ever analysis of TAVR in patients with bicuspid aortic stenosis indicate that key 30-day and 1-year outcomes are similar to those of propensity-matched TAVR patients with tricuspid disease, Raj R. Makkar, MD, said at the annual meeting of the American College of Cardiology.

“Select bicuspid anatomy is amenable to TAVR with current-generation, balloon-expandable TAVR technology with acceptable clinical outcomes. These data provide an argument for TAVR to be a reasonable alternative for bicuspid AS [aortic stenosis] patients who are at intermediate or high risk for surgical aortic valve replacement, which are the patients that are enrolled in this registry, and provide a sound basis to conduct a randomized clinical trial in young patients with bicuspid AS who are at low risk for surgery,” declared Dr. Makkar, director of interventional cardiology and the cardiac catheterization laboratory at Cedars-Sinai Medical Center in Los Angeles.

The landmark randomized trials of TAVR versus SAVR (surgical aortic valve replacement) that established TAVR as the preferred treatment for patients with severe aortic stenosis who are at high, intermediate, or low surgical risk systematically excluded patients with bicuspid AS, even though bicuspid anatomy is common, particularly in younger patients with AS.

Despite the absence of supportive randomized trial data, TAVR is being done for bicuspid AS. To learn how patients with bicuspid disease have fared, Dr. Makkar and coinvestigators analyzed the real-world Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy (STS/ACC TVT) Registry of all patients who underwent TAVR with the balloon-expandable Sapien 3 stent in the United States during 2015-2018. They compared outcomes in 2,691 patients with high or intermediate surgical risk who underwent TAVR for bicuspid AS to an equal number of patients who had TAVR for tricuspid disease, with the two groups being propensity-matched across 25 variables.

Key outcomes were reassuringly similar in the two groups. For example, 30-day and 1-year all-cause mortality rates were 2.6% and 10.8% in patients with bicuspid valves and similar, at 2.5% and 12.1%, in those with tricuspid AS. Paravalvular leak rates at 30 days and 1 year were similar in the two groups. The Kansas City Cardiomyopathy Questionnaire scores, reflecting quality of life, improved dramatically – by nearly 30 points – from pre-TAVR baseline in both groups. The proportion of patients who were New York Heart Association functional class III or IV improved from nearly 85% at baseline to about 8% at 30 days and 1 year, again with no significant difference between the bicuspid and tricuspid AS groups. And there were other benefits, too.

“Despite the concerns regarding optimal expansion of these valves in a bicuspid anatomy, what we observed here was a significant and similar reduction in mean gradients and increase in valve area, both in the bicuspid and tricuspid AS patients. So there was no impact of bicuspid anatomy as seen here in terms of valve hemodynamics,” according to the cardiologist.

Conversion from TAVR to open surgery was required in 0.9% of bicuspid and 0.4% of tricuspid AS patients. Rates of aortic dissection and need for aortic valve reintervention were similarly low in both groups.

The 30-day stroke rate was significantly higher in the bicuspid patients – 2.4% versus 1.6% – but by 1 year there was no significant between-group difference, with stroke rates of 3.4% in the bicuspid and 3.1% in the tricuspid TAVR patients.

“I’d like to point out that more than 75% of strokes occurred in the first 3 days. These are periprocedural strokes, and there was no difference in the time distribution of strokes between the bicuspid and tricuspid groups,” Dr. Makkar said.

These stroke data make a compelling case for the routine use of cerebral protection devices in patients undergoing TAVR, something which now occurs in less than 10% of cases nationally, he continued.

“I would argue that, based on these data, it would be wise for us to use cerebral protection devices, especially when we are doing TAVR in patients with bicuspid AS, because their valves tend to be more heavily calcified than is often the case in tricuspid AS,” Dr. Makkar said.

Discussant Mayra Guerrero, MD, of the Mayo Clinic in Rochester, Minn., took issue with Dr. Makkar’s comment regarding the need for a randomized trial of TAVR in bicuspid AS patients with low surgical risk.

“Do we really need a randomized trial when we see in real-world experience with more than 2,600 patients that the outcomes are fairly similar?” she asked.

Affirmative, Dr. Makkar responded, in light of the fact that the STS/ACC TVT Registry doesn’t include low–surgical risk, typically relatively young bicuspid AS TAVR patients.

“I would say that these data are reassuring and encouraging, but we must not get carried away. I think that would be the important message that I must give,” Dr. Makkar replied. “I think for patients who are high risk and who are intermediate risk, with STS scores of what they were here – 5 and more – I think it’s reasonable to consider them for TAVR based upon CT anatomy. For young patients, as I concluded, I think we must do a randomized clinical trial to definitely establish the safety and efficacy in these patients.”

Dr. Makkar reported receiving research grants from and serving as a consultant to Edwards Lifesciences, which supported the study, as well as Abbott, Medtronic, and Boston Scientific.

bjancin@mdedge.com

SOURCE: Makkar RR. ACC 19, 404-15. Late-breaking clinical trials


 

PHILADELPHIA – About 17% of patients discharged from a hospital after treatment for generalized convulsive status epilepticus are readmitted within 30 days, according to research presented at the annual meeting of the American Academy of Neurology. It is possible to identify patients at high risk of readmission, which could allow neurologists to reduce their clinical and economic burden, said the investigators.

Status epilepticus is a major neurologic emergency. Patients often have significant disability and may represent a burden on their families and on the health care system. To identify independent predictors of 30-day hospital readmission among patients discharged after generalized convulsive status epilepticus, Mohamad Rahwan, MD, a neurologist at the Medical University of South Carolina, Charleston, and colleagues examined data from the 2014 Nationwide Readmission Database.

The investigators included adults with a primary discharge diagnosis of generalized convulsive status epilepticus, identified by the ICD-9-CM code 345.3, in their study. Patients who died during hospitalization, had missing information on the length of stay, or were discharged in December 2014 were excluded from analysis. Dr. Rahwan and colleagues calculated the overall 30-day readmission rate for the sample and compared prespecified groups by their 30-day readmission status. They performed multiple logistic regression analysis to identify independent predictors of 30-day readmission, adjusting for potential confounders.

In all, 14,562 adults were discharged with a diagnosis of generalized convulsive status epilepticus. Of this population, 2,520 patients (17.3%) were readmitted within 30 days. Multivariate logistic regression analysis indicated that patients discharged against medical advice (odds ratio, 1.45), those discharged to short-term hospital (OR, 1.39), those with comorbid conditions (OR for Charlson Comorbidity Index of 1, 1.12; OR for Charlson Comorbidity Index of 2 or greater, 1.32), and those with a length of stay exceeding 6 days (OR, 1.42) had a greater risk of 30-day readmission. The researchers observed an inverse association for patients aged 45 years or older and for those in high-income households. “Greater attention to high-risk subgroups may identify opportunities to ameliorate the clinical and economic burden of early readmissions after generalized convulsive status epilepticus,” said the researchers.

The researchers had no disclosures.

egreb@mdedge.com

SOURCE: Rahwan M et al. AAN 2019, Abstract S36.006.

PHILADELPHIA – About 17% of patients discharged from a hospital after treatment for generalized convulsive status epilepticus are readmitted within 30 days, according to research presented at the annual meeting of the American Academy of Neurology. It is possible to identify patients at high risk of readmission, which could allow neurologists to reduce their clinical and economic burden, said the investigators.

Status epilepticus is a major neurologic emergency. Patients often have significant disability and may represent a burden on their families and on the health care system. To identify independent predictors of 30-day hospital readmission among patients discharged after generalized convulsive status epilepticus, Mohamad Rahwan, MD, a neurologist at the Medical University of South Carolina, Charleston, and colleagues examined data from the 2014 Nationwide Readmission Database.

The investigators included adults with a primary discharge diagnosis of generalized convulsive status epilepticus, identified by the ICD-9-CM code 345.3, in their study. Patients who died during hospitalization, had missing information on the length of stay, or were discharged in December 2014 were excluded from analysis. Dr. Rahwan and colleagues calculated the overall 30-day readmission rate for the sample and compared prespecified groups by their 30-day readmission status. They performed multiple logistic regression analysis to identify independent predictors of 30-day readmission, adjusting for potential confounders.

In all, 14,562 adults were discharged with a diagnosis of generalized convulsive status epilepticus. Of this population, 2,520 patients (17.3%) were readmitted within 30 days. Multivariate logistic regression analysis indicated that patients discharged against medical advice (odds ratio, 1.45), those discharged to short-term hospital (OR, 1.39), those with comorbid conditions (OR for Charlson Comorbidity Index of 1, 1.12; OR for Charlson Comorbidity Index of 2 or greater, 1.32), and those with a length of stay exceeding 6 days (OR, 1.42) had a greater risk of 30-day readmission. The researchers observed an inverse association for patients aged 45 years or older and for those in high-income households. “Greater attention to high-risk subgroups may identify opportunities to ameliorate the clinical and economic burden of early readmissions after generalized convulsive status epilepticus,” said the researchers.

The researchers had no disclosures.

egreb@mdedge.com

SOURCE: Rahwan M et al. AAN 2019, Abstract S36.006.

PHILADELPHIA – About 17% of patients discharged from a hospital after treatment for generalized convulsive status epilepticus are readmitted within 30 days, according to research presented at the annual meeting of the American Academy of Neurology. It is possible to identify patients at high risk of readmission, which could allow neurologists to reduce their clinical and economic burden, said the investigators.

Status epilepticus is a major neurologic emergency. Patients often have significant disability and may represent a burden on their families and on the health care system. To identify independent predictors of 30-day hospital readmission among patients discharged after generalized convulsive status epilepticus, Mohamad Rahwan, MD, a neurologist at the Medical University of South Carolina, Charleston, and colleagues examined data from the 2014 Nationwide Readmission Database.

The investigators included adults with a primary discharge diagnosis of generalized convulsive status epilepticus, identified by the ICD-9-CM code 345.3, in their study. Patients who died during hospitalization, had missing information on the length of stay, or were discharged in December 2014 were excluded from analysis. Dr. Rahwan and colleagues calculated the overall 30-day readmission rate for the sample and compared prespecified groups by their 30-day readmission status. They performed multiple logistic regression analysis to identify independent predictors of 30-day readmission, adjusting for potential confounders.

In all, 14,562 adults were discharged with a diagnosis of generalized convulsive status epilepticus. Of this population, 2,520 patients (17.3%) were readmitted within 30 days. Multivariate logistic regression analysis indicated that patients discharged against medical advice (odds ratio, 1.45), those discharged to short-term hospital (OR, 1.39), those with comorbid conditions (OR for Charlson Comorbidity Index of 1, 1.12; OR for Charlson Comorbidity Index of 2 or greater, 1.32), and those with a length of stay exceeding 6 days (OR, 1.42) had a greater risk of 30-day readmission. The researchers observed an inverse association for patients aged 45 years or older and for those in high-income households. “Greater attention to high-risk subgroups may identify opportunities to ameliorate the clinical and economic burden of early readmissions after generalized convulsive status epilepticus,” said the researchers.

The researchers had no disclosures.

egreb@mdedge.com

SOURCE: Rahwan M et al. AAN 2019, Abstract S36.006.

In this edition of “How I will treat my next patient,” I take a look at two recent trials – one summarizes a presentation at the European Lung Cancer Congress on the value of durvalumab as adjuvant treatment in patients with locally-advanced non–small cell lung cancer and the other confirms the safety and efficacy of subcutaneously-administered trastuzumab as neoadjuvant treatment in HER2/-positive breast cancer patients.

PACIFIC trial

In the PACIFIC trial, 713 patients with unresectable, stage III non–small cell lung cancer (NSCLC) who received concurrent chemoradiation were randomized to receive adjuvant durvalumab or an identical placebo, for a year after radiation ended. The results were dramatic in favor of durvalumab (N Engl J Med. 2018;379:2342-50).

Durvalumab showed 24-month overall survival of 66.3% versus 55.6% with placebo (hazard ratio, 0.68, P = .0025) and progression-free survival of 17.2 months versus 5. 6 months (HR, 0.51). As expected, there were more grade 3-4 toxicities and treatment discontinuations with durvalumab than with placebo, but the toxicity seemed modest, given the substantial improvements in tumor-related outcomes.

At the recent European Lung Cancer Congress, Marina Garassino, MD, reported on Patient-Reported Outcomes (PRO) in PACIFIC. PROs were analyzed by PD-L1 level. A total of 63% of patients had PD-L1 tumor expression data for analysis. Overall, there were no major differences in PROs by PD-L1. Global quality of life did not differ by PD-L1 expression cohort.



These data support adjuvant durvalumab for stage III, chemoradiation-treated NSCLC patients, not only from efficacy and toxicity viewpoints, but also from the standpoint of the patient experience, independent of PD-L1 tumor expression.
 

What this means in practice

From every relevant perspective, regardless of histology and molecular features associated with their particular tumor, it is worthwhile for us to recommend – and for our patients to receive – durvalumab adjuvant therapy for up to 1 year after radiation ends, with close follow-up and adherence to the criteria for treatment modification or discontinuation as performed in the PACIFIC trial. These new data remove any lingering concerns about the value of this life-prolonging treatment.

Subcutaneous vs. IV trastuzumab

In this international phase 3 trial in early breast cancer patients, neoadjuvant chemotherapy was paired with either standard IV trastuzumab or subcutaneous trastuzumab at intervals of every 3 weeks. After the cytotoxic chemotherapy concluded, patients completed a 12-month course of trastuzumab with either the IV or subcutaneous administration, as previously randomized. The 6-year event-free survival and overall survival were 65% and 84%, respectively, for both the IV and subcutaneous treatment administration.

The authors concluded that these results are relevant to patients with low-risk HER2-positive breast cancer patients, for whom T-DM1 is not needed (JAMA Oncol. 2019 Apr 18. doi: 10.1001/jamaoncol.2019.0339).

What this means in practice

These long-term data from the HannaH trial show persuasively that patients should be offered the more convenient, hopefully cheaper, subcutaneous route of administration. Since relapses beyond year 6 are unlikely, these data are unlikely to change with further follow-up. At our hospital, we recently made the decision to add subcutaneous trastuzumab to our formulary.

Dr. Lyss has been a community-based medical oncologist and clinical researcher for more than 35 years, practicing in St. Louis. His clinical and research interests are in the prevention, diagnosis, and treatment of breast and lung cancers, and in expanding access to clinical trials to medically underserved populations.

In this edition of “How I will treat my next patient,” I take a look at two recent trials – one summarizes a presentation at the European Lung Cancer Congress on the value of durvalumab as adjuvant treatment in patients with locally-advanced non–small cell lung cancer and the other confirms the safety and efficacy of subcutaneously-administered trastuzumab as neoadjuvant treatment in HER2/-positive breast cancer patients.

PACIFIC trial

In the PACIFIC trial, 713 patients with unresectable, stage III non–small cell lung cancer (NSCLC) who received concurrent chemoradiation were randomized to receive adjuvant durvalumab or an identical placebo, for a year after radiation ended. The results were dramatic in favor of durvalumab (N Engl J Med. 2018;379:2342-50).

Durvalumab showed 24-month overall survival of 66.3% versus 55.6% with placebo (hazard ratio, 0.68, P = .0025) and progression-free survival of 17.2 months versus 5. 6 months (HR, 0.51). As expected, there were more grade 3-4 toxicities and treatment discontinuations with durvalumab than with placebo, but the toxicity seemed modest, given the substantial improvements in tumor-related outcomes.

At the recent European Lung Cancer Congress, Marina Garassino, MD, reported on Patient-Reported Outcomes (PRO) in PACIFIC. PROs were analyzed by PD-L1 level. A total of 63% of patients had PD-L1 tumor expression data for analysis. Overall, there were no major differences in PROs by PD-L1. Global quality of life did not differ by PD-L1 expression cohort.



These data support adjuvant durvalumab for stage III, chemoradiation-treated NSCLC patients, not only from efficacy and toxicity viewpoints, but also from the standpoint of the patient experience, independent of PD-L1 tumor expression.
 

What this means in practice

From every relevant perspective, regardless of histology and molecular features associated with their particular tumor, it is worthwhile for us to recommend – and for our patients to receive – durvalumab adjuvant therapy for up to 1 year after radiation ends, with close follow-up and adherence to the criteria for treatment modification or discontinuation as performed in the PACIFIC trial. These new data remove any lingering concerns about the value of this life-prolonging treatment.

Subcutaneous vs. IV trastuzumab

In this international phase 3 trial in early breast cancer patients, neoadjuvant chemotherapy was paired with either standard IV trastuzumab or subcutaneous trastuzumab at intervals of every 3 weeks. After the cytotoxic chemotherapy concluded, patients completed a 12-month course of trastuzumab with either the IV or subcutaneous administration, as previously randomized. The 6-year event-free survival and overall survival were 65% and 84%, respectively, for both the IV and subcutaneous treatment administration.

The authors concluded that these results are relevant to patients with low-risk HER2-positive breast cancer patients, for whom T-DM1 is not needed (JAMA Oncol. 2019 Apr 18. doi: 10.1001/jamaoncol.2019.0339).

What this means in practice

These long-term data from the HannaH trial show persuasively that patients should be offered the more convenient, hopefully cheaper, subcutaneous route of administration. Since relapses beyond year 6 are unlikely, these data are unlikely to change with further follow-up. At our hospital, we recently made the decision to add subcutaneous trastuzumab to our formulary.

Dr. Lyss has been a community-based medical oncologist and clinical researcher for more than 35 years, practicing in St. Louis. His clinical and research interests are in the prevention, diagnosis, and treatment of breast and lung cancers, and in expanding access to clinical trials to medically underserved populations.

In this edition of “How I will treat my next patient,” I take a look at two recent trials – one summarizes a presentation at the European Lung Cancer Congress on the value of durvalumab as adjuvant treatment in patients with locally-advanced non–small cell lung cancer and the other confirms the safety and efficacy of subcutaneously-administered trastuzumab as neoadjuvant treatment in HER2/-positive breast cancer patients.

PACIFIC trial

In the PACIFIC trial, 713 patients with unresectable, stage III non–small cell lung cancer (NSCLC) who received concurrent chemoradiation were randomized to receive adjuvant durvalumab or an identical placebo, for a year after radiation ended. The results were dramatic in favor of durvalumab (N Engl J Med. 2018;379:2342-50).

Durvalumab showed 24-month overall survival of 66.3% versus 55.6% with placebo (hazard ratio, 0.68, P = .0025) and progression-free survival of 17.2 months versus 5. 6 months (HR, 0.51). As expected, there were more grade 3-4 toxicities and treatment discontinuations with durvalumab than with placebo, but the toxicity seemed modest, given the substantial improvements in tumor-related outcomes.

At the recent European Lung Cancer Congress, Marina Garassino, MD, reported on Patient-Reported Outcomes (PRO) in PACIFIC. PROs were analyzed by PD-L1 level. A total of 63% of patients had PD-L1 tumor expression data for analysis. Overall, there were no major differences in PROs by PD-L1. Global quality of life did not differ by PD-L1 expression cohort.



These data support adjuvant durvalumab for stage III, chemoradiation-treated NSCLC patients, not only from efficacy and toxicity viewpoints, but also from the standpoint of the patient experience, independent of PD-L1 tumor expression.
 

What this means in practice

From every relevant perspective, regardless of histology and molecular features associated with their particular tumor, it is worthwhile for us to recommend – and for our patients to receive – durvalumab adjuvant therapy for up to 1 year after radiation ends, with close follow-up and adherence to the criteria for treatment modification or discontinuation as performed in the PACIFIC trial. These new data remove any lingering concerns about the value of this life-prolonging treatment.

Subcutaneous vs. IV trastuzumab

In this international phase 3 trial in early breast cancer patients, neoadjuvant chemotherapy was paired with either standard IV trastuzumab or subcutaneous trastuzumab at intervals of every 3 weeks. After the cytotoxic chemotherapy concluded, patients completed a 12-month course of trastuzumab with either the IV or subcutaneous administration, as previously randomized. The 6-year event-free survival and overall survival were 65% and 84%, respectively, for both the IV and subcutaneous treatment administration.

The authors concluded that these results are relevant to patients with low-risk HER2-positive breast cancer patients, for whom T-DM1 is not needed (JAMA Oncol. 2019 Apr 18. doi: 10.1001/jamaoncol.2019.0339).

What this means in practice

These long-term data from the HannaH trial show persuasively that patients should be offered the more convenient, hopefully cheaper, subcutaneous route of administration. Since relapses beyond year 6 are unlikely, these data are unlikely to change with further follow-up. At our hospital, we recently made the decision to add subcutaneous trastuzumab to our formulary.

Dr. Lyss has been a community-based medical oncologist and clinical researcher for more than 35 years, practicing in St. Louis. His clinical and research interests are in the prevention, diagnosis, and treatment of breast and lung cancers, and in expanding access to clinical trials to medically underserved populations.

Prior healthcare utilization predicts future utilization;¹ thus, providers should know when a child has had a recent healthcare visit. Healthcare providers typically obtain this information from parents and caregivers, who may not always provide accurate information.^2-4

The Hospital to Home Outcomes study (H2O) was a randomized controlled trial conducted to assess the effects of a one-time home nurse visit following discharge on unplanned healthcare reutilization.⁵ We assessed reutilization through two sources: parent report via a postdischarge telephone call and administrative data. In this analysis, we sought to understand differences in reutilization rates by source by comparing parent report with administrative data.

The H2O trial included children (<18 years) hospitalized on the hospital medicine (HM) or neuroscience (Neurology/Neurosurgery) services at Cincinnati Children’s Hospital Medical Center (CCHMC) from February 2015 to April 2016; they had an English-speaking parent and were discharged to home without skilled nursing care.⁶ For this analysis, we restricted the sample to children randomized to the control arm (discharge without a home visit), which reflects typical clinical care.

We used administrative data to capture 14-day reutilization (unplanned hospital readmissions, emergency department [ED] visits, or urgent care visits). CCHMC is the only pediatric admitting facility in the region and includes two pediatric EDs and five urgent care centers. We supplemented hospital data with a dataset (The Health Collaborative⁷) that included utilization at other regional facilities. Parent report was assessed via a research coordinator phone call 14-23 days after discharge. Parents were asked: “I’m going to [ask] about your child’s health since [discharge date]. Has s/he been hospitalized overnight? Has s/he been taken to the Emergency Room/Emergency Department (didn’t stay overnight)? Has s/he been taken to an urgent care?” We report 14-day reutilization rates by source (parent and/or administrative) and visit type.

We considered administrative data the gold standard for documentation of reutilization events for two reasons. First, all healthcare encounters generate billing and are therefore documented with verifiable coding. Second, we had access to data from our center and other regional healthcare facilities. Any parent-reported utilization to a facility not documented in either dataset was considered an unverifiable event (eg, outside our catchment region). Agreement between administrative and parent report of 14-day reutilization was summarized as positive agreement (reutilization documented in both administrative and parent report), negative agreement (no reutilization reported in either administrative or parent report), and overall agreement (combination of positive and negative agreement). We classified discrepancies as reutilization events in administrative data without parent report of reutilization or vice versa. We performed medical record review of discrepancies in our institutional data.

We summarized agreement by using the Cohen’s kappa statistic by reuse type (hospital readmission, ED, and urgent care visit) and overall (any reutilization event). Strength of agreement based on the kappa statistics was classified as poor (<0.20), fair (0.21-0.40), moderate (0.41-0.60), good (0.61-0.80), and very good (0.81-1.00).⁸ We used McNemar’s test to evaluate marginal homogeneity.

Of 749 children randomized to the standard of care arm, 723 parents completed the 14-day follow-up call and were included in this analysis. The median child age was two years (interquartile range: 0.4, 6.9), the median length of stay (LOS) was two days (1, 3), and the majority were white (62%). Payer mix varied, with 44% privately insured and 54% publicly insured. Most patients (83%) were admitted to the HM service, and the most common diagnoses groups for index admission were respiratory (35%), neurologic (14%), and gastrointestinal (9%) diseases.

Administrative data showed 63 children with any reutilization event; parents reported 63 with any reutilization event; 48 children had events reported by both sources. The overall agreement was high, ranging from 95.9% to 98.5% (Table 1) depending on visit type. The positive agreement (ie, parent and administrative data indicated reutilization) ranged from 47.6% to 76.2%. Negative agreement (ie, parent and administrative data agreed no reutilization) was very high, 97.7% to 99.2%. Parents reported three ED visits and four urgent care visits that were unverifiable due to lack of access to administrative data (sites of care reported were not included in our datasets).

By comparing parent report of reutilization after hospital discharge through postdischarge phone calls with administrative data, we demonstrated high overall agreement between sources (95.9%); this finding is similar to prior research investigating the relationship between an established medical home and reutilization.⁹ However, this agreement is largely due to both sources reporting no reutilization. When revisits did occur, the agreement was notably lower, especially with regard to urgent care visits.

Discrepancies between sources have several possible explanations. First, parents may be confused by the framing of reutilization questions, perhaps lacking clarity around which visit we were referencing. Second, parents may experience limitations in health literacy^10,11 with a lack of familiarity with healthcare language, such as the ability to delineate location types (for example, a parent may identify an urgent care visit as an ED visit, given their close proximity at our facility). Finally, our prior work identified that the “fog” of hospitalization,¹² which is often a stressful and disruptive time for families, may linger after admission and could lead to difficulty in recalling detailed events.

Our findings have implications for effective care in a complex healthcare system where parent report may be the most practical method to obtain historical information, both within clinical care and in the context of research or quality measures, such as postdischarge utilization. Given that one of the greatest risk factors for readmission is prior utilization,¹ the knowledge that a patient experienced a reutilization after a prior discharge might prompt the inpatient provider to better prepare families for subsequent transition to home.

To apply our findings practically, it is important to realize that a parent report may be sufficient when reporting that no revisit occurred, if there is also no record of a visit in accessible administrative data (such as an electronic health record). However, further questions or investigation should be considered when parents report a visit did occur or when administrative data indicate a visit occurred that the parent does not recall. Providers and researchers alike should remember to use health literacy universal precautions with all families, employing plain language without medical jargon.¹³ As linked electronic health record use becomes more prevalent, administrative data may be accessible in real-time, allowing for verification of family interview information. Administrative data beyond a single hospital system should be considered to effectively capture reutilization for research or quality efforts.

Our study has several limitations. Similar to most studies using reutilization outcomes, our data may miss a few unverifiable reuse events. By supplementing with additional regional data,⁷ we likely captured most events. Second, we did not include patients with limited English proficiency, although it is unclear how this might have biased our results. Third, while relatively few families did not complete the calls, it is possible that more discrepancies would have been noted in nonresponders. Fourth, research coordinators administering the calls followed a script to determine reutilization information; in clinical practice, a practitioner might not ask questions as clearly, which could negatively impact recall or might add clarifying follow-up questions to enhance recall. Finally, the analysis occurred in the setting of a randomized controlled trial that included children with relatively noncomplex health conditions with short LOS;⁶ thus, the results may not apply to other populations.

In conclusion, parent report and administrative data of reutilization following hospital discharge were usually in agreement when no reutilization occurred; however, discrepancies were noted more often when reutilizations occurred and may have care implications.

Collaborators

On behalf of the H2O Trial study group including: Joanne Bachus, BSN, RN; Andrew F. Beck, MD, MPH; Monica L. Borell, BSN, RN; Lenisa V. Chang, MA, PhD; Patricia Crawford, RN; Jennifer M. Gold, MSN, RN; Judy A. Heilman BSN, RN; Jane C. Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Allison Loechtenfeldt, BS; Colleen Mangeot, MS; Lynn O’Donnell, BSN, RN; Rita H. Pickler, PhD, RN; Hadley S. Sauers-Ford, MPH; Anita N. Shah, DO, MPH; Susan N. Sherman, DPA; Lauren G. Solan, MD, MEd; Karen P. Sullivan, BSN, RN; Susan Wade-Murphy, MSN, RN

Disclosures

Hospital to Home Outcomes team reports grants from the Patient Centered Outcomes Research Institute during the conduct of the study. Dr. White reports personal fees from the Institute for Health Care Improvement, outside the submitted work.

Funding

This work was supported by the Patient Centered Outcomes Research Institute (IHS-1306-0081 to Dr. S. Shah). All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee. Dr Auger’s research is funded by the Agency for Healthcare Research and Quality (1K08HS024735).

Prior healthcare utilization predicts future utilization;¹ thus, providers should know when a child has had a recent healthcare visit. Healthcare providers typically obtain this information from parents and caregivers, who may not always provide accurate information.^2-4

The Hospital to Home Outcomes study (H2O) was a randomized controlled trial conducted to assess the effects of a one-time home nurse visit following discharge on unplanned healthcare reutilization.⁵ We assessed reutilization through two sources: parent report via a postdischarge telephone call and administrative data. In this analysis, we sought to understand differences in reutilization rates by source by comparing parent report with administrative data.

The H2O trial included children (<18 years) hospitalized on the hospital medicine (HM) or neuroscience (Neurology/Neurosurgery) services at Cincinnati Children’s Hospital Medical Center (CCHMC) from February 2015 to April 2016; they had an English-speaking parent and were discharged to home without skilled nursing care.⁶ For this analysis, we restricted the sample to children randomized to the control arm (discharge without a home visit), which reflects typical clinical care.

We used administrative data to capture 14-day reutilization (unplanned hospital readmissions, emergency department [ED] visits, or urgent care visits). CCHMC is the only pediatric admitting facility in the region and includes two pediatric EDs and five urgent care centers. We supplemented hospital data with a dataset (The Health Collaborative⁷) that included utilization at other regional facilities. Parent report was assessed via a research coordinator phone call 14-23 days after discharge. Parents were asked: “I’m going to [ask] about your child’s health since [discharge date]. Has s/he been hospitalized overnight? Has s/he been taken to the Emergency Room/Emergency Department (didn’t stay overnight)? Has s/he been taken to an urgent care?” We report 14-day reutilization rates by source (parent and/or administrative) and visit type.

We considered administrative data the gold standard for documentation of reutilization events for two reasons. First, all healthcare encounters generate billing and are therefore documented with verifiable coding. Second, we had access to data from our center and other regional healthcare facilities. Any parent-reported utilization to a facility not documented in either dataset was considered an unverifiable event (eg, outside our catchment region). Agreement between administrative and parent report of 14-day reutilization was summarized as positive agreement (reutilization documented in both administrative and parent report), negative agreement (no reutilization reported in either administrative or parent report), and overall agreement (combination of positive and negative agreement). We classified discrepancies as reutilization events in administrative data without parent report of reutilization or vice versa. We performed medical record review of discrepancies in our institutional data.

We summarized agreement by using the Cohen’s kappa statistic by reuse type (hospital readmission, ED, and urgent care visit) and overall (any reutilization event). Strength of agreement based on the kappa statistics was classified as poor (<0.20), fair (0.21-0.40), moderate (0.41-0.60), good (0.61-0.80), and very good (0.81-1.00).⁸ We used McNemar’s test to evaluate marginal homogeneity.

Of 749 children randomized to the standard of care arm, 723 parents completed the 14-day follow-up call and were included in this analysis. The median child age was two years (interquartile range: 0.4, 6.9), the median length of stay (LOS) was two days (1, 3), and the majority were white (62%). Payer mix varied, with 44% privately insured and 54% publicly insured. Most patients (83%) were admitted to the HM service, and the most common diagnoses groups for index admission were respiratory (35%), neurologic (14%), and gastrointestinal (9%) diseases.

Administrative data showed 63 children with any reutilization event; parents reported 63 with any reutilization event; 48 children had events reported by both sources. The overall agreement was high, ranging from 95.9% to 98.5% (Table 1) depending on visit type. The positive agreement (ie, parent and administrative data indicated reutilization) ranged from 47.6% to 76.2%. Negative agreement (ie, parent and administrative data agreed no reutilization) was very high, 97.7% to 99.2%. Parents reported three ED visits and four urgent care visits that were unverifiable due to lack of access to administrative data (sites of care reported were not included in our datasets).

By comparing parent report of reutilization after hospital discharge through postdischarge phone calls with administrative data, we demonstrated high overall agreement between sources (95.9%); this finding is similar to prior research investigating the relationship between an established medical home and reutilization.⁹ However, this agreement is largely due to both sources reporting no reutilization. When revisits did occur, the agreement was notably lower, especially with regard to urgent care visits.

Discrepancies between sources have several possible explanations. First, parents may be confused by the framing of reutilization questions, perhaps lacking clarity around which visit we were referencing. Second, parents may experience limitations in health literacy^10,11 with a lack of familiarity with healthcare language, such as the ability to delineate location types (for example, a parent may identify an urgent care visit as an ED visit, given their close proximity at our facility). Finally, our prior work identified that the “fog” of hospitalization,¹² which is often a stressful and disruptive time for families, may linger after admission and could lead to difficulty in recalling detailed events.

Our findings have implications for effective care in a complex healthcare system where parent report may be the most practical method to obtain historical information, both within clinical care and in the context of research or quality measures, such as postdischarge utilization. Given that one of the greatest risk factors for readmission is prior utilization,¹ the knowledge that a patient experienced a reutilization after a prior discharge might prompt the inpatient provider to better prepare families for subsequent transition to home.

To apply our findings practically, it is important to realize that a parent report may be sufficient when reporting that no revisit occurred, if there is also no record of a visit in accessible administrative data (such as an electronic health record). However, further questions or investigation should be considered when parents report a visit did occur or when administrative data indicate a visit occurred that the parent does not recall. Providers and researchers alike should remember to use health literacy universal precautions with all families, employing plain language without medical jargon.¹³ As linked electronic health record use becomes more prevalent, administrative data may be accessible in real-time, allowing for verification of family interview information. Administrative data beyond a single hospital system should be considered to effectively capture reutilization for research or quality efforts.

Our study has several limitations. Similar to most studies using reutilization outcomes, our data may miss a few unverifiable reuse events. By supplementing with additional regional data,⁷ we likely captured most events. Second, we did not include patients with limited English proficiency, although it is unclear how this might have biased our results. Third, while relatively few families did not complete the calls, it is possible that more discrepancies would have been noted in nonresponders. Fourth, research coordinators administering the calls followed a script to determine reutilization information; in clinical practice, a practitioner might not ask questions as clearly, which could negatively impact recall or might add clarifying follow-up questions to enhance recall. Finally, the analysis occurred in the setting of a randomized controlled trial that included children with relatively noncomplex health conditions with short LOS;⁶ thus, the results may not apply to other populations.

In conclusion, parent report and administrative data of reutilization following hospital discharge were usually in agreement when no reutilization occurred; however, discrepancies were noted more often when reutilizations occurred and may have care implications.

Collaborators

On behalf of the H2O Trial study group including: Joanne Bachus, BSN, RN; Andrew F. Beck, MD, MPH; Monica L. Borell, BSN, RN; Lenisa V. Chang, MA, PhD; Patricia Crawford, RN; Jennifer M. Gold, MSN, RN; Judy A. Heilman BSN, RN; Jane C. Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Allison Loechtenfeldt, BS; Colleen Mangeot, MS; Lynn O’Donnell, BSN, RN; Rita H. Pickler, PhD, RN; Hadley S. Sauers-Ford, MPH; Anita N. Shah, DO, MPH; Susan N. Sherman, DPA; Lauren G. Solan, MD, MEd; Karen P. Sullivan, BSN, RN; Susan Wade-Murphy, MSN, RN

Disclosures

Hospital to Home Outcomes team reports grants from the Patient Centered Outcomes Research Institute during the conduct of the study. Dr. White reports personal fees from the Institute for Health Care Improvement, outside the submitted work.

Funding

This work was supported by the Patient Centered Outcomes Research Institute (IHS-1306-0081 to Dr. S. Shah). All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee. Dr Auger’s research is funded by the Agency for Healthcare Research and Quality (1K08HS024735).

Prior healthcare utilization predicts future utilization;¹ thus, providers should know when a child has had a recent healthcare visit. Healthcare providers typically obtain this information from parents and caregivers, who may not always provide accurate information.^2-4

The Hospital to Home Outcomes study (H2O) was a randomized controlled trial conducted to assess the effects of a one-time home nurse visit following discharge on unplanned healthcare reutilization.⁵ We assessed reutilization through two sources: parent report via a postdischarge telephone call and administrative data. In this analysis, we sought to understand differences in reutilization rates by source by comparing parent report with administrative data.

The H2O trial included children (<18 years) hospitalized on the hospital medicine (HM) or neuroscience (Neurology/Neurosurgery) services at Cincinnati Children’s Hospital Medical Center (CCHMC) from February 2015 to April 2016; they had an English-speaking parent and were discharged to home without skilled nursing care.⁶ For this analysis, we restricted the sample to children randomized to the control arm (discharge without a home visit), which reflects typical clinical care.

We used administrative data to capture 14-day reutilization (unplanned hospital readmissions, emergency department [ED] visits, or urgent care visits). CCHMC is the only pediatric admitting facility in the region and includes two pediatric EDs and five urgent care centers. We supplemented hospital data with a dataset (The Health Collaborative⁷) that included utilization at other regional facilities. Parent report was assessed via a research coordinator phone call 14-23 days after discharge. Parents were asked: “I’m going to [ask] about your child’s health since [discharge date]. Has s/he been hospitalized overnight? Has s/he been taken to the Emergency Room/Emergency Department (didn’t stay overnight)? Has s/he been taken to an urgent care?” We report 14-day reutilization rates by source (parent and/or administrative) and visit type.

We considered administrative data the gold standard for documentation of reutilization events for two reasons. First, all healthcare encounters generate billing and are therefore documented with verifiable coding. Second, we had access to data from our center and other regional healthcare facilities. Any parent-reported utilization to a facility not documented in either dataset was considered an unverifiable event (eg, outside our catchment region). Agreement between administrative and parent report of 14-day reutilization was summarized as positive agreement (reutilization documented in both administrative and parent report), negative agreement (no reutilization reported in either administrative or parent report), and overall agreement (combination of positive and negative agreement). We classified discrepancies as reutilization events in administrative data without parent report of reutilization or vice versa. We performed medical record review of discrepancies in our institutional data.

We summarized agreement by using the Cohen’s kappa statistic by reuse type (hospital readmission, ED, and urgent care visit) and overall (any reutilization event). Strength of agreement based on the kappa statistics was classified as poor (<0.20), fair (0.21-0.40), moderate (0.41-0.60), good (0.61-0.80), and very good (0.81-1.00).⁸ We used McNemar’s test to evaluate marginal homogeneity.

Of 749 children randomized to the standard of care arm, 723 parents completed the 14-day follow-up call and were included in this analysis. The median child age was two years (interquartile range: 0.4, 6.9), the median length of stay (LOS) was two days (1, 3), and the majority were white (62%). Payer mix varied, with 44% privately insured and 54% publicly insured. Most patients (83%) were admitted to the HM service, and the most common diagnoses groups for index admission were respiratory (35%), neurologic (14%), and gastrointestinal (9%) diseases.

Administrative data showed 63 children with any reutilization event; parents reported 63 with any reutilization event; 48 children had events reported by both sources. The overall agreement was high, ranging from 95.9% to 98.5% (Table 1) depending on visit type. The positive agreement (ie, parent and administrative data indicated reutilization) ranged from 47.6% to 76.2%. Negative agreement (ie, parent and administrative data agreed no reutilization) was very high, 97.7% to 99.2%. Parents reported three ED visits and four urgent care visits that were unverifiable due to lack of access to administrative data (sites of care reported were not included in our datasets).

By comparing parent report of reutilization after hospital discharge through postdischarge phone calls with administrative data, we demonstrated high overall agreement between sources (95.9%); this finding is similar to prior research investigating the relationship between an established medical home and reutilization.⁹ However, this agreement is largely due to both sources reporting no reutilization. When revisits did occur, the agreement was notably lower, especially with regard to urgent care visits.

Discrepancies between sources have several possible explanations. First, parents may be confused by the framing of reutilization questions, perhaps lacking clarity around which visit we were referencing. Second, parents may experience limitations in health literacy^10,11 with a lack of familiarity with healthcare language, such as the ability to delineate location types (for example, a parent may identify an urgent care visit as an ED visit, given their close proximity at our facility). Finally, our prior work identified that the “fog” of hospitalization,¹² which is often a stressful and disruptive time for families, may linger after admission and could lead to difficulty in recalling detailed events.

Our findings have implications for effective care in a complex healthcare system where parent report may be the most practical method to obtain historical information, both within clinical care and in the context of research or quality measures, such as postdischarge utilization. Given that one of the greatest risk factors for readmission is prior utilization,¹ the knowledge that a patient experienced a reutilization after a prior discharge might prompt the inpatient provider to better prepare families for subsequent transition to home.

To apply our findings practically, it is important to realize that a parent report may be sufficient when reporting that no revisit occurred, if there is also no record of a visit in accessible administrative data (such as an electronic health record). However, further questions or investigation should be considered when parents report a visit did occur or when administrative data indicate a visit occurred that the parent does not recall. Providers and researchers alike should remember to use health literacy universal precautions with all families, employing plain language without medical jargon.¹³ As linked electronic health record use becomes more prevalent, administrative data may be accessible in real-time, allowing for verification of family interview information. Administrative data beyond a single hospital system should be considered to effectively capture reutilization for research or quality efforts.

Our study has several limitations. Similar to most studies using reutilization outcomes, our data may miss a few unverifiable reuse events. By supplementing with additional regional data,⁷ we likely captured most events. Second, we did not include patients with limited English proficiency, although it is unclear how this might have biased our results. Third, while relatively few families did not complete the calls, it is possible that more discrepancies would have been noted in nonresponders. Fourth, research coordinators administering the calls followed a script to determine reutilization information; in clinical practice, a practitioner might not ask questions as clearly, which could negatively impact recall or might add clarifying follow-up questions to enhance recall. Finally, the analysis occurred in the setting of a randomized controlled trial that included children with relatively noncomplex health conditions with short LOS;⁶ thus, the results may not apply to other populations.

In conclusion, parent report and administrative data of reutilization following hospital discharge were usually in agreement when no reutilization occurred; however, discrepancies were noted more often when reutilizations occurred and may have care implications.

Collaborators

On behalf of the H2O Trial study group including: Joanne Bachus, BSN, RN; Andrew F. Beck, MD, MPH; Monica L. Borell, BSN, RN; Lenisa V. Chang, MA, PhD; Patricia Crawford, RN; Jennifer M. Gold, MSN, RN; Judy A. Heilman BSN, RN; Jane C. Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Allison Loechtenfeldt, BS; Colleen Mangeot, MS; Lynn O’Donnell, BSN, RN; Rita H. Pickler, PhD, RN; Hadley S. Sauers-Ford, MPH; Anita N. Shah, DO, MPH; Susan N. Sherman, DPA; Lauren G. Solan, MD, MEd; Karen P. Sullivan, BSN, RN; Susan Wade-Murphy, MSN, RN

Disclosures

Hospital to Home Outcomes team reports grants from the Patient Centered Outcomes Research Institute during the conduct of the study. Dr. White reports personal fees from the Institute for Health Care Improvement, outside the submitted work.

Funding

This work was supported by the Patient Centered Outcomes Research Institute (IHS-1306-0081 to Dr. S. Shah). All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee. Dr Auger’s research is funded by the Agency for Healthcare Research and Quality (1K08HS024735).

Delirium presents as an acute change in mentation characterized by reduced attention, clouding of awareness, and typically an altered level of arousal. It can be caused by a host of medical conditions, medications, or other psychoactive substances and is therefore encountered primarily in acute and postacute medical settings.¹ More than a quarter of all hospitalized patients develop delirium,² with rates up to 80% in the critically ill.³ Similarly, delirium occurs in more than one-third of patients who transition to postacute care.⁴ These high prevalence rates are alarming, especially because delirium is a risk factor for mortality, prolonged hospitalization, institutionalization, and overall higher cost of care.⁵ However, more than a quarter of delirium is preventable.⁶ Evidence-based guidelines for delirium uniformly call for multicomponent prevention strategies,⁷ and these are best delivered through collaborative models of care. In short, delirium impacts healthcare systems; therefore, interventions aimed at preventing delirium and its consequences ought to be systems-based.

Since the Institute of Medicine issued its 1999 report highlighting the critical role of medical errors in healthcare, healthcare systems have increasingly become team-based.⁸ “Medical care is inherently interdependent,”⁹ and this implies that delirium prevention rests not only on individuals but also on broader systems of care. Although nonpharmacological interventions are efficacious at preventing delirium,¹⁰ previous reviews have focused on specific interventions or multiple interventions rather than the systems of care needed to deliver them. Indeed, teams and the quality of their teamwork impact outcomes.¹¹

Herein, we provide a systematic review and meta-analysis of integrated models of care designed to prevent delirium. What distinguishes this review from previous reviews of nonpharmacological interventions to prevent delirium is our focus on discrete models of care that involve collaboration among clinicians. Our goal is to identify the most promising models that deserve further development, investigation, and dissemination. Viewing delirium prevention through a collaborative care lens is consistent with efforts to achieve value-based care and may encourage drawing from the expanding literature outlining the benefits of mental healthcare integration.^12,13 Specifically, a systems perspective highlights the potential for system-wide benefits such as reducing readmissions^14,15 and cost savings.¹⁶

This systematic review and meta-analysis follows PRISMA guidelines. A search of OVID, MEDLINE, CINAHL, Cochrane Database of Systematic Reviews, EMBASE, and PsycINFO was completed by a medical librarian for clinical studies in which models of care were implemented to prevent delirium using PICO (P patient, problem or population; I, intervention; C, comparison, control or comparator; O, outcome) inquiries. Search terms included delirium, acute confusional state, altered mental status, prevention, and control (“delirium”/exp OR “acute confusion”/exp OR “altered mental status”/exp) AND “prevention and control”/exp AND [English]/lim AND [embase]/lim).

One researcher (AK) screened articles by title for relevance. Relevant articles were then divided among four authors (AK, MO, NF, and OB), and the abstracts were screened for eligibility. The authors reviewed the full texts of any potentially eligible studies. Each full text was assigned to two authors for full review. Discrepancies were adjudicated by conference among all authors. In addition, references within all full-text publications were scanned for potential additional articles.

The inclusion criteria for review of full-text articles required English-language description of a model of care with multiple interventions, delirium reported as an outcome, and presence of a comparator group.

“Model of care” was defined by the Cochrane Effective Practice and Organization of Care Review Group as follows: (1) revision of professional roles, including shifting of professional roles or expansion of roles to new tasks; (2) creation of clinical multidisciplinary teams or addition of new members to the team who collaborate inpatient care; (3) delivery of multiple interventions across multiple domains (ie, studies involving a single intervention such as physical therapy or targeting a single domain such as sleep were excluded); and (4) formal integration of services whereby teams work together in collaboration with existing services to enhance care.¹⁷ For this review, we required that studies include a comparator group so that effectiveness of the intervention could be assessed. Quality improvement studies that lacked a comparator group were excluded.

Delirium incidence was the primary outcome and was evaluated by meta-analysis. Heterogeneity was assessed using I² and visual inspection of forest plots. I² values of 25%, 50%, and 75% represent low, moderate, and high heterogeneity, respectively. The studies were pooled according to study type as follows: randomized controlled trials, pre–post design, and other nonrandomized prospective studies. Random effects models were used to calculate estimates using the Comprehensive Meta-Analysis software (Version 3, Biostat, Englewood, New Jersey), which also generated forest plots.

Risk of bias was assessed using criteria established by the Cochrane Collaborative Review Criteria, which lists six categories of potential bias: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective reporting.¹⁷ Each study was assessed by two authors (either MO and AK or MO-P. and OB) for bias and a numerical value was assigned to each of the six categories as follows: 1 = low risk, 2 = unknown/moderate risk, and 3 = high risk. Where scorers disagreed, all authors jointly conferred, and a consensus score was given. The values for each of these six categories were added to create a composite risk-of-bias score for each study, with 6 being the lowest possible score and 18 the highest. Overall risk was classified as follows: <9 = low risk, 9-12 = moderate risk, and >12 = high risk.

Study Selection Process

An initial literature search identified 352 articles. After reviewing the titles, 308 articles were excluded for irrelevance, and 44 abstracts were screened for eligibility. We excluded 27 articles upon abstract review, and the full texts of 17 were obtained for detailed review. In addition, we identified another 10 potentially eligible articles through review of references and obtained full texts of these as well. Of the 27 full-text articles reviewed, 15 were included in this systematic review, 10 of which were suitable for meta-analysis. The Figure shows the PRISMA flow chart.

Study Characteristics

The 15 studies that met the inclusion criteria are summarized in the Table.^18-32 Delirium prevention was among the primary outcomes of 13 studies; delirium outcomes were reported in the other two studies as well, which were primarily designed to assess feasibility.^26,27 Six studies were conducted in the United States, three in Sweden, two in Spain, two in the United Kingdom, and one each conducted in Korea and Canada. Healthcare settings among the included studies involved the intensive care unit (six studies), medical floors (four studies), surgical floors (three studies), a long-term care unit (one study), and an inpatient palliative care service (one study). We categorized the studies according to design and intent as follows: randomized controlled studies (three), pilot feasibility studies (two), pre–post design (six), and other nonrandomized prospective studies (four; Table).

Outcomes Reported

All but one of the studies reported delirium incidence. The most commonly used delirium screening instrument was the Confusion Assessment Method (CAM) or its modified version, the CAM-ICU (11 studies).^33,34 Other methods used to assess mentation included the Richmond Agitation Sedation Scale,³⁵ the Organic Brain Syndrome scale,³⁶ the revised Delirium Rating Scale,³⁷ the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition,³⁸ and the Confusion Rating Scale.³⁹ (Details regarding delirium screening tools can be found in the systematic review by De and Wand.⁴⁰) Researchers performed delirium assessment in nine studies, whereas assessments were performed by clinical staff in the remaining studies. Other outcomes reported included length of stay (LOS), mortality, number of days ventilated, and functional decline. None of the included studies reported cost effectiveness.

Risk of Bias Assessment

Risk of bias assessment identified only two studies—both randomized controlled trials—as low risk (Table). The remaining studies had moderate (four studies) or high risk (nine studies).

Results from Individual Studies

Of the 15 studies, nine reported a statistically significant reduction in delirium incidence, and another two reported a statistically insignificant reduction. In addition, seven of the eight studies that assessed delirium duration found reduced duration in the intervention cohort, and two of the three studies that reported delirium severity found a reduction in the intervention group.

Results of Meta-Analysis

Random effects models were created to meta-analyze groups of studies based on design as follows: randomized controlled trials (three studies^18,19,25), pre–post intervention studies (four of six studies included^28-31), and other nonrandomized studies (three of four studies included^21-23). Meta-analysis was not completed for the two feasibility studies^26,27 because delirium outcome data were limited due to the feasibility study design. The study of Dale et al.³² was excluded from the meta-analysis because the rates of CAM-ICU completion differed substantially between control and intervention groups (0.35 vs 1.49 per 24 hours, respectively), leading to imbalanced between-group sensitivity in delirium detection and Needham et al.²⁰ was also excluded because it reported only days of delirium, not delirium incidence. The study by Lundström et al.²⁴ was also excluded from the meta-analysis because delirium incidence was measured on days 1, 3, and 5, whereas the other studies reported delirium daily.

Meta-analysis of the three randomized controlled trials revealed a pooled odds ratio of 0.56 (95% CI: 0.37-0.85; P = .006) for delirium incidence among intervention group subjects relative to those in comparator groups. The heterogeneity across studies was low (I² = 29%). Pooling data from four pre–post studies found that the odds ratio for delirium incidence was 0.63 (95% CI: 0.37-1.07; P = .09). The heterogeneity across these studies was moderate (I² = 65%). Results from the three eligible, nonrandomized prospective studies were also pooled. The odds ratio for developing delirium among study subjects was 0.79 (95% CI: 0.46-1.37; P = .40), and the heterogeneity among these studies was high (I² = 85%).

We provide a systematic review and meta-analysis of delirium preventive models of care. Meta-analysis of the three randomized controlled trials found that these models of care led to a statistically significant reduction in delirium incidence; study subjects had an 11.5% reduction in absolute delirium incidence. The pooled odds ratios for both of the other sets of nonrandomized studies favored the intervention group but were not significant, each because of one included study. The pre–post meta-analysis failed to reach significance as one of the included studies found a trend toward higher delirium incidence; however, interestingly, in that same study, the overall delirium-free days were significantly reduced overall (24 vs 27, P = .002). Similarly, meta-analysis of the three additional nonrandomized prospective studies failed to reach significance because the largest included study found higher rates of delirium among intervention group subjects. Despite considerable risk of bias in several of these studies, their findings were broadly consistent; all but one study (Gagnon 2012²¹) reported a trend or a significant reduction in delirium incidence, duration, severity, or number of delirium episodes. Moreover, the value of such models of care extended beyond preventing delirium; for instance, other positive outcomes included reduced LOS and fewer medical complications.

Models of care ranged widely with respect to specific interventions, though several common elements highlighted their relevance for delirium care and as potential delirium prevention strategies in future studies. For example, two of the randomized controlled trials^18,19 employed early mobilization, enhanced nutrition, sleep hygiene, early reduction of invasive procedures (eg, urinary catheterization), and pain control in their multicomponent models. Five additional studies also incorporated early mobilization,^{20,22,23,31,32} and three sought to improve sleep quality.^22,28,30 Among other important strategies were delirium screening,^{18,20,22,30,31} monitoring medication,^{18,20,22,26,28,30,32} orientation,^18,21,23,28 addressing vision and hearing impairment,^18,22,23,32 hydration,^18,22,23 avoiding hypoxia,^18,20,30 and staff, patient, and caretaker education.^{19,21,23,27-30} Unique strategies were implemented in certain studies. For instance, one study used massage therapy,²⁸ preventing delays in transfer logistics in another,³⁰ and a third addressed psychosocial problems.²⁵ Overall, the selection of strategies depended on the patient setting; thus, no one care bundle should be expected to emerge as a universal model for delirium prevention. Rather, these results should be interpreted within their specific care contexts and judged on the quality of evidence (eg, effect size and statistically significant findings, low risk of bias, sound experimental design). The one study that failed to find any positive effect on delirium, that of Gagnon et al.,²¹ was conducted on an inpatient palliative care service in Canada, and its negative finding may reflect the unique delirium risk factors in patients who are nearing end of life.

This current review differs from previous delirium prevention reviews in operationally defining a “model of care.” We identified a great deal of variation in specific models and team composition. For example, some interventions were carried out by nurses^18-20,31 and physicians,^20,21,25,32 whereas others involved physical therapists,^20,22,28 medical residents,²³ geriatricians,^22,23,25 pharmacists,²⁶ researchers,¹⁸ and trained volunteers.²² In all cases, the staff roles were expanded to include new tasks, and the clinical team worked collaboratively to administer interventions across multiple domains. Team-related considerations are critical because modern medical care is inherently interdependent.⁹ These broad differences in team composition across studies demonstrate the number of potential options for team structure and function. They also highlight the number of “moving parts” to be considered when designing and implementing delirium care bundles.

Most of the delirium prevention studies implementing models of care are characterized by a substantial risk of bias. We evaluated risk of bias along six categories of potential sources, including random assignment to groups, ability to foresee future group allocation, blinding of participants and personnel to group assignment, blinding of outcome assessment, completeness of outcome data, selective reporting, and other potential sources of bias.¹⁷ Two of the three studies that used randomization had a low risk of bias, and four additional studies had a moderate risk of bias. Allocation concealment was accomplished only in randomized controlled trials, whereas blinding of both subjects and study personnel was not implemented in any of the studies. Although some studies relied on data analysis by research personnel blinded to group membership or the nature of the intervention, others failed to do so or failed to describe data analysis in sufficient detail. Studies also failed to report the percentage of unscorable or otherwise omitted delirium assessments necessary to calculate attrition rates or to understand the comprehensiveness of outcome assessment in a systematic manner. Other potential sources of bias included systematic differences between the intervention and control groups (such as differences in gender composition, age, or delirium risk) at study outset.

A primary limitation of this review is the heterogeneity of settings, interventions, and models of care across included studies. We excluded several studies from this review for being delivered by a single individual or service line (eg, introduction of a geriatric consult service, physical therapy, or volunteers), for providing a single intervention (eg, early ambulation alone), or for multiple interventions targeting a single domain (eg, sleep). We did so because the future of value-based care lies in collaboration of providers and services, and in a way the complexity across and within these studies ultimately reflects the complexity of medical settings as well as the multifactorial nature of delirium. The broader message is a call for increasing the integration of delirium-related care services. As discussed earlier, the high risk of bias across these studies is a limitation of our findings; high-quality evidence on the value of delirium prevention models of care remains limited. Thus, although our review suggests that there are multicomponent models of care that hold promise in mitigating delirium and its outcomes, additional randomized studies are required to confirm the efficacy of such models of care and to test which services, interventions, and clinical domains deserve priority.

To our knowledge, this is the first systematic review and meta-analysis of delirium preventive models of care. Models of care, as defined here, necessarily included a multidisciplinary team in which traditional staff roles had been revised to implement a multicomponent, multidomain intervention. Other recent reviews are available for multicomponent pharmacological and nonpharmacological interventions to prevent and manage delirium,^41-49 but just as important as which interventions are being delivered is the team that delivers them. Care delivery in a complex medical system is more than handing a patient a medication or facilitating ambulation; it requires a choreographed dance of teamwork and integration across services. This review identifies promising models of care that deserve further recognition, refinement, and ultimately widespread implementation.

Acknowledgments

The authors comprise a writing group created through the Delirium Boot Camp, an annual meeting originally sponsored by the Center of Excellence for Delirium in Aging: Research, Training, and Educational Enhancement (CEDARTREE, Boston, Massachusetts); it is currently supported by the Network for Investigation of Delirium: Unifying Scientists (NIDUS, Boston, Massachusetts). The authors would like to thank medical librarian Rita Mitchell (Aurora Health Care, Milwaukee, Wisconsin) for the literature search, senior scientific writer and editor Joe Grundle (Aurora Research Institute, Milwaukee, Wisconsin) for editorial assistance, and graphics specialist Brian Miller (Aurora Research Institute, Milwaukee, Wisconsin) for help with the figures.

Disclosures

The authors report no relevant conflicts of interest.

Funding

No funding was dedicated to the conduct of this review.

Delirium presents as an acute change in mentation characterized by reduced attention, clouding of awareness, and typically an altered level of arousal. It can be caused by a host of medical conditions, medications, or other psychoactive substances and is therefore encountered primarily in acute and postacute medical settings.¹ More than a quarter of all hospitalized patients develop delirium,² with rates up to 80% in the critically ill.³ Similarly, delirium occurs in more than one-third of patients who transition to postacute care.⁴ These high prevalence rates are alarming, especially because delirium is a risk factor for mortality, prolonged hospitalization, institutionalization, and overall higher cost of care.⁵ However, more than a quarter of delirium is preventable.⁶ Evidence-based guidelines for delirium uniformly call for multicomponent prevention strategies,⁷ and these are best delivered through collaborative models of care. In short, delirium impacts healthcare systems; therefore, interventions aimed at preventing delirium and its consequences ought to be systems-based.

Since the Institute of Medicine issued its 1999 report highlighting the critical role of medical errors in healthcare, healthcare systems have increasingly become team-based.⁸ “Medical care is inherently interdependent,”⁹ and this implies that delirium prevention rests not only on individuals but also on broader systems of care. Although nonpharmacological interventions are efficacious at preventing delirium,¹⁰ previous reviews have focused on specific interventions or multiple interventions rather than the systems of care needed to deliver them. Indeed, teams and the quality of their teamwork impact outcomes.¹¹

Herein, we provide a systematic review and meta-analysis of integrated models of care designed to prevent delirium. What distinguishes this review from previous reviews of nonpharmacological interventions to prevent delirium is our focus on discrete models of care that involve collaboration among clinicians. Our goal is to identify the most promising models that deserve further development, investigation, and dissemination. Viewing delirium prevention through a collaborative care lens is consistent with efforts to achieve value-based care and may encourage drawing from the expanding literature outlining the benefits of mental healthcare integration.^12,13 Specifically, a systems perspective highlights the potential for system-wide benefits such as reducing readmissions^14,15 and cost savings.¹⁶

This systematic review and meta-analysis follows PRISMA guidelines. A search of OVID, MEDLINE, CINAHL, Cochrane Database of Systematic Reviews, EMBASE, and PsycINFO was completed by a medical librarian for clinical studies in which models of care were implemented to prevent delirium using PICO (P patient, problem or population; I, intervention; C, comparison, control or comparator; O, outcome) inquiries. Search terms included delirium, acute confusional state, altered mental status, prevention, and control (“delirium”/exp OR “acute confusion”/exp OR “altered mental status”/exp) AND “prevention and control”/exp AND [English]/lim AND [embase]/lim).

One researcher (AK) screened articles by title for relevance. Relevant articles were then divided among four authors (AK, MO, NF, and OB), and the abstracts were screened for eligibility. The authors reviewed the full texts of any potentially eligible studies. Each full text was assigned to two authors for full review. Discrepancies were adjudicated by conference among all authors. In addition, references within all full-text publications were scanned for potential additional articles.

The inclusion criteria for review of full-text articles required English-language description of a model of care with multiple interventions, delirium reported as an outcome, and presence of a comparator group.

“Model of care” was defined by the Cochrane Effective Practice and Organization of Care Review Group as follows: (1) revision of professional roles, including shifting of professional roles or expansion of roles to new tasks; (2) creation of clinical multidisciplinary teams or addition of new members to the team who collaborate inpatient care; (3) delivery of multiple interventions across multiple domains (ie, studies involving a single intervention such as physical therapy or targeting a single domain such as sleep were excluded); and (4) formal integration of services whereby teams work together in collaboration with existing services to enhance care.¹⁷ For this review, we required that studies include a comparator group so that effectiveness of the intervention could be assessed. Quality improvement studies that lacked a comparator group were excluded.

Delirium incidence was the primary outcome and was evaluated by meta-analysis. Heterogeneity was assessed using I² and visual inspection of forest plots. I² values of 25%, 50%, and 75% represent low, moderate, and high heterogeneity, respectively. The studies were pooled according to study type as follows: randomized controlled trials, pre–post design, and other nonrandomized prospective studies. Random effects models were used to calculate estimates using the Comprehensive Meta-Analysis software (Version 3, Biostat, Englewood, New Jersey), which also generated forest plots.

Risk of bias was assessed using criteria established by the Cochrane Collaborative Review Criteria, which lists six categories of potential bias: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective reporting.¹⁷ Each study was assessed by two authors (either MO and AK or MO-P. and OB) for bias and a numerical value was assigned to each of the six categories as follows: 1 = low risk, 2 = unknown/moderate risk, and 3 = high risk. Where scorers disagreed, all authors jointly conferred, and a consensus score was given. The values for each of these six categories were added to create a composite risk-of-bias score for each study, with 6 being the lowest possible score and 18 the highest. Overall risk was classified as follows: <9 = low risk, 9-12 = moderate risk, and >12 = high risk.

Study Selection Process

An initial literature search identified 352 articles. After reviewing the titles, 308 articles were excluded for irrelevance, and 44 abstracts were screened for eligibility. We excluded 27 articles upon abstract review, and the full texts of 17 were obtained for detailed review. In addition, we identified another 10 potentially eligible articles through review of references and obtained full texts of these as well. Of the 27 full-text articles reviewed, 15 were included in this systematic review, 10 of which were suitable for meta-analysis. The Figure shows the PRISMA flow chart.

Study Characteristics

The 15 studies that met the inclusion criteria are summarized in the Table.^18-32 Delirium prevention was among the primary outcomes of 13 studies; delirium outcomes were reported in the other two studies as well, which were primarily designed to assess feasibility.^26,27 Six studies were conducted in the United States, three in Sweden, two in Spain, two in the United Kingdom, and one each conducted in Korea and Canada. Healthcare settings among the included studies involved the intensive care unit (six studies), medical floors (four studies), surgical floors (three studies), a long-term care unit (one study), and an inpatient palliative care service (one study). We categorized the studies according to design and intent as follows: randomized controlled studies (three), pilot feasibility studies (two), pre–post design (six), and other nonrandomized prospective studies (four; Table).

Outcomes Reported

All but one of the studies reported delirium incidence. The most commonly used delirium screening instrument was the Confusion Assessment Method (CAM) or its modified version, the CAM-ICU (11 studies).^33,34 Other methods used to assess mentation included the Richmond Agitation Sedation Scale,³⁵ the Organic Brain Syndrome scale,³⁶ the revised Delirium Rating Scale,³⁷ the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition,³⁸ and the Confusion Rating Scale.³⁹ (Details regarding delirium screening tools can be found in the systematic review by De and Wand.⁴⁰) Researchers performed delirium assessment in nine studies, whereas assessments were performed by clinical staff in the remaining studies. Other outcomes reported included length of stay (LOS), mortality, number of days ventilated, and functional decline. None of the included studies reported cost effectiveness.

Risk of Bias Assessment

Risk of bias assessment identified only two studies—both randomized controlled trials—as low risk (Table). The remaining studies had moderate (four studies) or high risk (nine studies).

Results from Individual Studies

Of the 15 studies, nine reported a statistically significant reduction in delirium incidence, and another two reported a statistically insignificant reduction. In addition, seven of the eight studies that assessed delirium duration found reduced duration in the intervention cohort, and two of the three studies that reported delirium severity found a reduction in the intervention group.

Results of Meta-Analysis

Random effects models were created to meta-analyze groups of studies based on design as follows: randomized controlled trials (three studies^18,19,25), pre–post intervention studies (four of six studies included^28-31), and other nonrandomized studies (three of four studies included^21-23). Meta-analysis was not completed for the two feasibility studies^26,27 because delirium outcome data were limited due to the feasibility study design. The study of Dale et al.³² was excluded from the meta-analysis because the rates of CAM-ICU completion differed substantially between control and intervention groups (0.35 vs 1.49 per 24 hours, respectively), leading to imbalanced between-group sensitivity in delirium detection and Needham et al.²⁰ was also excluded because it reported only days of delirium, not delirium incidence. The study by Lundström et al.²⁴ was also excluded from the meta-analysis because delirium incidence was measured on days 1, 3, and 5, whereas the other studies reported delirium daily.

Meta-analysis of the three randomized controlled trials revealed a pooled odds ratio of 0.56 (95% CI: 0.37-0.85; P = .006) for delirium incidence among intervention group subjects relative to those in comparator groups. The heterogeneity across studies was low (I² = 29%). Pooling data from four pre–post studies found that the odds ratio for delirium incidence was 0.63 (95% CI: 0.37-1.07; P = .09). The heterogeneity across these studies was moderate (I² = 65%). Results from the three eligible, nonrandomized prospective studies were also pooled. The odds ratio for developing delirium among study subjects was 0.79 (95% CI: 0.46-1.37; P = .40), and the heterogeneity among these studies was high (I² = 85%).

We provide a systematic review and meta-analysis of delirium preventive models of care. Meta-analysis of the three randomized controlled trials found that these models of care led to a statistically significant reduction in delirium incidence; study subjects had an 11.5% reduction in absolute delirium incidence. The pooled odds ratios for both of the other sets of nonrandomized studies favored the intervention group but were not significant, each because of one included study. The pre–post meta-analysis failed to reach significance as one of the included studies found a trend toward higher delirium incidence; however, interestingly, in that same study, the overall delirium-free days were significantly reduced overall (24 vs 27, P = .002). Similarly, meta-analysis of the three additional nonrandomized prospective studies failed to reach significance because the largest included study found higher rates of delirium among intervention group subjects. Despite considerable risk of bias in several of these studies, their findings were broadly consistent; all but one study (Gagnon 2012²¹) reported a trend or a significant reduction in delirium incidence, duration, severity, or number of delirium episodes. Moreover, the value of such models of care extended beyond preventing delirium; for instance, other positive outcomes included reduced LOS and fewer medical complications.

Models of care ranged widely with respect to specific interventions, though several common elements highlighted their relevance for delirium care and as potential delirium prevention strategies in future studies. For example, two of the randomized controlled trials^18,19 employed early mobilization, enhanced nutrition, sleep hygiene, early reduction of invasive procedures (eg, urinary catheterization), and pain control in their multicomponent models. Five additional studies also incorporated early mobilization,^{20,22,23,31,32} and three sought to improve sleep quality.^22,28,30 Among other important strategies were delirium screening,^{18,20,22,30,31} monitoring medication,^{18,20,22,26,28,30,32} orientation,^18,21,23,28 addressing vision and hearing impairment,^18,22,23,32 hydration,^18,22,23 avoiding hypoxia,^18,20,30 and staff, patient, and caretaker education.^{19,21,23,27-30} Unique strategies were implemented in certain studies. For instance, one study used massage therapy,²⁸ preventing delays in transfer logistics in another,³⁰ and a third addressed psychosocial problems.²⁵ Overall, the selection of strategies depended on the patient setting; thus, no one care bundle should be expected to emerge as a universal model for delirium prevention. Rather, these results should be interpreted within their specific care contexts and judged on the quality of evidence (eg, effect size and statistically significant findings, low risk of bias, sound experimental design). The one study that failed to find any positive effect on delirium, that of Gagnon et al.,²¹ was conducted on an inpatient palliative care service in Canada, and its negative finding may reflect the unique delirium risk factors in patients who are nearing end of life.

This current review differs from previous delirium prevention reviews in operationally defining a “model of care.” We identified a great deal of variation in specific models and team composition. For example, some interventions were carried out by nurses^18-20,31 and physicians,^20,21,25,32 whereas others involved physical therapists,^20,22,28 medical residents,²³ geriatricians,^22,23,25 pharmacists,²⁶ researchers,¹⁸ and trained volunteers.²² In all cases, the staff roles were expanded to include new tasks, and the clinical team worked collaboratively to administer interventions across multiple domains. Team-related considerations are critical because modern medical care is inherently interdependent.⁹ These broad differences in team composition across studies demonstrate the number of potential options for team structure and function. They also highlight the number of “moving parts” to be considered when designing and implementing delirium care bundles.

Most of the delirium prevention studies implementing models of care are characterized by a substantial risk of bias. We evaluated risk of bias along six categories of potential sources, including random assignment to groups, ability to foresee future group allocation, blinding of participants and personnel to group assignment, blinding of outcome assessment, completeness of outcome data, selective reporting, and other potential sources of bias.¹⁷ Two of the three studies that used randomization had a low risk of bias, and four additional studies had a moderate risk of bias. Allocation concealment was accomplished only in randomized controlled trials, whereas blinding of both subjects and study personnel was not implemented in any of the studies. Although some studies relied on data analysis by research personnel blinded to group membership or the nature of the intervention, others failed to do so or failed to describe data analysis in sufficient detail. Studies also failed to report the percentage of unscorable or otherwise omitted delirium assessments necessary to calculate attrition rates or to understand the comprehensiveness of outcome assessment in a systematic manner. Other potential sources of bias included systematic differences between the intervention and control groups (such as differences in gender composition, age, or delirium risk) at study outset.

A primary limitation of this review is the heterogeneity of settings, interventions, and models of care across included studies. We excluded several studies from this review for being delivered by a single individual or service line (eg, introduction of a geriatric consult service, physical therapy, or volunteers), for providing a single intervention (eg, early ambulation alone), or for multiple interventions targeting a single domain (eg, sleep). We did so because the future of value-based care lies in collaboration of providers and services, and in a way the complexity across and within these studies ultimately reflects the complexity of medical settings as well as the multifactorial nature of delirium. The broader message is a call for increasing the integration of delirium-related care services. As discussed earlier, the high risk of bias across these studies is a limitation of our findings; high-quality evidence on the value of delirium prevention models of care remains limited. Thus, although our review suggests that there are multicomponent models of care that hold promise in mitigating delirium and its outcomes, additional randomized studies are required to confirm the efficacy of such models of care and to test which services, interventions, and clinical domains deserve priority.

To our knowledge, this is the first systematic review and meta-analysis of delirium preventive models of care. Models of care, as defined here, necessarily included a multidisciplinary team in which traditional staff roles had been revised to implement a multicomponent, multidomain intervention. Other recent reviews are available for multicomponent pharmacological and nonpharmacological interventions to prevent and manage delirium,^41-49 but just as important as which interventions are being delivered is the team that delivers them. Care delivery in a complex medical system is more than handing a patient a medication or facilitating ambulation; it requires a choreographed dance of teamwork and integration across services. This review identifies promising models of care that deserve further recognition, refinement, and ultimately widespread implementation.

Acknowledgments

The authors comprise a writing group created through the Delirium Boot Camp, an annual meeting originally sponsored by the Center of Excellence for Delirium in Aging: Research, Training, and Educational Enhancement (CEDARTREE, Boston, Massachusetts); it is currently supported by the Network for Investigation of Delirium: Unifying Scientists (NIDUS, Boston, Massachusetts). The authors would like to thank medical librarian Rita Mitchell (Aurora Health Care, Milwaukee, Wisconsin) for the literature search, senior scientific writer and editor Joe Grundle (Aurora Research Institute, Milwaukee, Wisconsin) for editorial assistance, and graphics specialist Brian Miller (Aurora Research Institute, Milwaukee, Wisconsin) for help with the figures.

Disclosures

The authors report no relevant conflicts of interest.

Funding

No funding was dedicated to the conduct of this review.

Delirium presents as an acute change in mentation characterized by reduced attention, clouding of awareness, and typically an altered level of arousal. It can be caused by a host of medical conditions, medications, or other psychoactive substances and is therefore encountered primarily in acute and postacute medical settings.¹ More than a quarter of all hospitalized patients develop delirium,² with rates up to 80% in the critically ill.³ Similarly, delirium occurs in more than one-third of patients who transition to postacute care.⁴ These high prevalence rates are alarming, especially because delirium is a risk factor for mortality, prolonged hospitalization, institutionalization, and overall higher cost of care.⁵ However, more than a quarter of delirium is preventable.⁶ Evidence-based guidelines for delirium uniformly call for multicomponent prevention strategies,⁷ and these are best delivered through collaborative models of care. In short, delirium impacts healthcare systems; therefore, interventions aimed at preventing delirium and its consequences ought to be systems-based.

Since the Institute of Medicine issued its 1999 report highlighting the critical role of medical errors in healthcare, healthcare systems have increasingly become team-based.⁸ “Medical care is inherently interdependent,”⁹ and this implies that delirium prevention rests not only on individuals but also on broader systems of care. Although nonpharmacological interventions are efficacious at preventing delirium,¹⁰ previous reviews have focused on specific interventions or multiple interventions rather than the systems of care needed to deliver them. Indeed, teams and the quality of their teamwork impact outcomes.¹¹

Herein, we provide a systematic review and meta-analysis of integrated models of care designed to prevent delirium. What distinguishes this review from previous reviews of nonpharmacological interventions to prevent delirium is our focus on discrete models of care that involve collaboration among clinicians. Our goal is to identify the most promising models that deserve further development, investigation, and dissemination. Viewing delirium prevention through a collaborative care lens is consistent with efforts to achieve value-based care and may encourage drawing from the expanding literature outlining the benefits of mental healthcare integration.^12,13 Specifically, a systems perspective highlights the potential for system-wide benefits such as reducing readmissions^14,15 and cost savings.¹⁶

This systematic review and meta-analysis follows PRISMA guidelines. A search of OVID, MEDLINE, CINAHL, Cochrane Database of Systematic Reviews, EMBASE, and PsycINFO was completed by a medical librarian for clinical studies in which models of care were implemented to prevent delirium using PICO (P patient, problem or population; I, intervention; C, comparison, control or comparator; O, outcome) inquiries. Search terms included delirium, acute confusional state, altered mental status, prevention, and control (“delirium”/exp OR “acute confusion”/exp OR “altered mental status”/exp) AND “prevention and control”/exp AND [English]/lim AND [embase]/lim).

One researcher (AK) screened articles by title for relevance. Relevant articles were then divided among four authors (AK, MO, NF, and OB), and the abstracts were screened for eligibility. The authors reviewed the full texts of any potentially eligible studies. Each full text was assigned to two authors for full review. Discrepancies were adjudicated by conference among all authors. In addition, references within all full-text publications were scanned for potential additional articles.

The inclusion criteria for review of full-text articles required English-language description of a model of care with multiple interventions, delirium reported as an outcome, and presence of a comparator group.

“Model of care” was defined by the Cochrane Effective Practice and Organization of Care Review Group as follows: (1) revision of professional roles, including shifting of professional roles or expansion of roles to new tasks; (2) creation of clinical multidisciplinary teams or addition of new members to the team who collaborate inpatient care; (3) delivery of multiple interventions across multiple domains (ie, studies involving a single intervention such as physical therapy or targeting a single domain such as sleep were excluded); and (4) formal integration of services whereby teams work together in collaboration with existing services to enhance care.¹⁷ For this review, we required that studies include a comparator group so that effectiveness of the intervention could be assessed. Quality improvement studies that lacked a comparator group were excluded.

Delirium incidence was the primary outcome and was evaluated by meta-analysis. Heterogeneity was assessed using I² and visual inspection of forest plots. I² values of 25%, 50%, and 75% represent low, moderate, and high heterogeneity, respectively. The studies were pooled according to study type as follows: randomized controlled trials, pre–post design, and other nonrandomized prospective studies. Random effects models were used to calculate estimates using the Comprehensive Meta-Analysis software (Version 3, Biostat, Englewood, New Jersey), which also generated forest plots.

Risk of bias was assessed using criteria established by the Cochrane Collaborative Review Criteria, which lists six categories of potential bias: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective reporting.¹⁷ Each study was assessed by two authors (either MO and AK or MO-P. and OB) for bias and a numerical value was assigned to each of the six categories as follows: 1 = low risk, 2 = unknown/moderate risk, and 3 = high risk. Where scorers disagreed, all authors jointly conferred, and a consensus score was given. The values for each of these six categories were added to create a composite risk-of-bias score for each study, with 6 being the lowest possible score and 18 the highest. Overall risk was classified as follows: <9 = low risk, 9-12 = moderate risk, and >12 = high risk.

Study Selection Process

An initial literature search identified 352 articles. After reviewing the titles, 308 articles were excluded for irrelevance, and 44 abstracts were screened for eligibility. We excluded 27 articles upon abstract review, and the full texts of 17 were obtained for detailed review. In addition, we identified another 10 potentially eligible articles through review of references and obtained full texts of these as well. Of the 27 full-text articles reviewed, 15 were included in this systematic review, 10 of which were suitable for meta-analysis. The Figure shows the PRISMA flow chart.

Study Characteristics

The 15 studies that met the inclusion criteria are summarized in the Table.^18-32 Delirium prevention was among the primary outcomes of 13 studies; delirium outcomes were reported in the other two studies as well, which were primarily designed to assess feasibility.^26,27 Six studies were conducted in the United States, three in Sweden, two in Spain, two in the United Kingdom, and one each conducted in Korea and Canada. Healthcare settings among the included studies involved the intensive care unit (six studies), medical floors (four studies), surgical floors (three studies), a long-term care unit (one study), and an inpatient palliative care service (one study). We categorized the studies according to design and intent as follows: randomized controlled studies (three), pilot feasibility studies (two), pre–post design (six), and other nonrandomized prospective studies (four; Table).

Outcomes Reported

All but one of the studies reported delirium incidence. The most commonly used delirium screening instrument was the Confusion Assessment Method (CAM) or its modified version, the CAM-ICU (11 studies).^33,34 Other methods used to assess mentation included the Richmond Agitation Sedation Scale,³⁵ the Organic Brain Syndrome scale,³⁶ the revised Delirium Rating Scale,³⁷ the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition,³⁸ and the Confusion Rating Scale.³⁹ (Details regarding delirium screening tools can be found in the systematic review by De and Wand.⁴⁰) Researchers performed delirium assessment in nine studies, whereas assessments were performed by clinical staff in the remaining studies. Other outcomes reported included length of stay (LOS), mortality, number of days ventilated, and functional decline. None of the included studies reported cost effectiveness.

Risk of Bias Assessment

Risk of bias assessment identified only two studies—both randomized controlled trials—as low risk (Table). The remaining studies had moderate (four studies) or high risk (nine studies).

Results from Individual Studies

Of the 15 studies, nine reported a statistically significant reduction in delirium incidence, and another two reported a statistically insignificant reduction. In addition, seven of the eight studies that assessed delirium duration found reduced duration in the intervention cohort, and two of the three studies that reported delirium severity found a reduction in the intervention group.

Results of Meta-Analysis

Random effects models were created to meta-analyze groups of studies based on design as follows: randomized controlled trials (three studies^18,19,25), pre–post intervention studies (four of six studies included^28-31), and other nonrandomized studies (three of four studies included^21-23). Meta-analysis was not completed for the two feasibility studies^26,27 because delirium outcome data were limited due to the feasibility study design. The study of Dale et al.³² was excluded from the meta-analysis because the rates of CAM-ICU completion differed substantially between control and intervention groups (0.35 vs 1.49 per 24 hours, respectively), leading to imbalanced between-group sensitivity in delirium detection and Needham et al.²⁰ was also excluded because it reported only days of delirium, not delirium incidence. The study by Lundström et al.²⁴ was also excluded from the meta-analysis because delirium incidence was measured on days 1, 3, and 5, whereas the other studies reported delirium daily.

Meta-analysis of the three randomized controlled trials revealed a pooled odds ratio of 0.56 (95% CI: 0.37-0.85; P = .006) for delirium incidence among intervention group subjects relative to those in comparator groups. The heterogeneity across studies was low (I² = 29%). Pooling data from four pre–post studies found that the odds ratio for delirium incidence was 0.63 (95% CI: 0.37-1.07; P = .09). The heterogeneity across these studies was moderate (I² = 65%). Results from the three eligible, nonrandomized prospective studies were also pooled. The odds ratio for developing delirium among study subjects was 0.79 (95% CI: 0.46-1.37; P = .40), and the heterogeneity among these studies was high (I² = 85%).

We provide a systematic review and meta-analysis of delirium preventive models of care. Meta-analysis of the three randomized controlled trials found that these models of care led to a statistically significant reduction in delirium incidence; study subjects had an 11.5% reduction in absolute delirium incidence. The pooled odds ratios for both of the other sets of nonrandomized studies favored the intervention group but were not significant, each because of one included study. The pre–post meta-analysis failed to reach significance as one of the included studies found a trend toward higher delirium incidence; however, interestingly, in that same study, the overall delirium-free days were significantly reduced overall (24 vs 27, P = .002). Similarly, meta-analysis of the three additional nonrandomized prospective studies failed to reach significance because the largest included study found higher rates of delirium among intervention group subjects. Despite considerable risk of bias in several of these studies, their findings were broadly consistent; all but one study (Gagnon 2012²¹) reported a trend or a significant reduction in delirium incidence, duration, severity, or number of delirium episodes. Moreover, the value of such models of care extended beyond preventing delirium; for instance, other positive outcomes included reduced LOS and fewer medical complications.

Models of care ranged widely with respect to specific interventions, though several common elements highlighted their relevance for delirium care and as potential delirium prevention strategies in future studies. For example, two of the randomized controlled trials^18,19 employed early mobilization, enhanced nutrition, sleep hygiene, early reduction of invasive procedures (eg, urinary catheterization), and pain control in their multicomponent models. Five additional studies also incorporated early mobilization,^{20,22,23,31,32} and three sought to improve sleep quality.^22,28,30 Among other important strategies were delirium screening,^{18,20,22,30,31} monitoring medication,^{18,20,22,26,28,30,32} orientation,^18,21,23,28 addressing vision and hearing impairment,^18,22,23,32 hydration,^18,22,23 avoiding hypoxia,^18,20,30 and staff, patient, and caretaker education.^{19,21,23,27-30} Unique strategies were implemented in certain studies. For instance, one study used massage therapy,²⁸ preventing delays in transfer logistics in another,³⁰ and a third addressed psychosocial problems.²⁵ Overall, the selection of strategies depended on the patient setting; thus, no one care bundle should be expected to emerge as a universal model for delirium prevention. Rather, these results should be interpreted within their specific care contexts and judged on the quality of evidence (eg, effect size and statistically significant findings, low risk of bias, sound experimental design). The one study that failed to find any positive effect on delirium, that of Gagnon et al.,²¹ was conducted on an inpatient palliative care service in Canada, and its negative finding may reflect the unique delirium risk factors in patients who are nearing end of life.

This current review differs from previous delirium prevention reviews in operationally defining a “model of care.” We identified a great deal of variation in specific models and team composition. For example, some interventions were carried out by nurses^18-20,31 and physicians,^20,21,25,32 whereas others involved physical therapists,^20,22,28 medical residents,²³ geriatricians,^22,23,25 pharmacists,²⁶ researchers,¹⁸ and trained volunteers.²² In all cases, the staff roles were expanded to include new tasks, and the clinical team worked collaboratively to administer interventions across multiple domains. Team-related considerations are critical because modern medical care is inherently interdependent.⁹ These broad differences in team composition across studies demonstrate the number of potential options for team structure and function. They also highlight the number of “moving parts” to be considered when designing and implementing delirium care bundles.

Most of the delirium prevention studies implementing models of care are characterized by a substantial risk of bias. We evaluated risk of bias along six categories of potential sources, including random assignment to groups, ability to foresee future group allocation, blinding of participants and personnel to group assignment, blinding of outcome assessment, completeness of outcome data, selective reporting, and other potential sources of bias.¹⁷ Two of the three studies that used randomization had a low risk of bias, and four additional studies had a moderate risk of bias. Allocation concealment was accomplished only in randomized controlled trials, whereas blinding of both subjects and study personnel was not implemented in any of the studies. Although some studies relied on data analysis by research personnel blinded to group membership or the nature of the intervention, others failed to do so or failed to describe data analysis in sufficient detail. Studies also failed to report the percentage of unscorable or otherwise omitted delirium assessments necessary to calculate attrition rates or to understand the comprehensiveness of outcome assessment in a systematic manner. Other potential sources of bias included systematic differences between the intervention and control groups (such as differences in gender composition, age, or delirium risk) at study outset.

A primary limitation of this review is the heterogeneity of settings, interventions, and models of care across included studies. We excluded several studies from this review for being delivered by a single individual or service line (eg, introduction of a geriatric consult service, physical therapy, or volunteers), for providing a single intervention (eg, early ambulation alone), or for multiple interventions targeting a single domain (eg, sleep). We did so because the future of value-based care lies in collaboration of providers and services, and in a way the complexity across and within these studies ultimately reflects the complexity of medical settings as well as the multifactorial nature of delirium. The broader message is a call for increasing the integration of delirium-related care services. As discussed earlier, the high risk of bias across these studies is a limitation of our findings; high-quality evidence on the value of delirium prevention models of care remains limited. Thus, although our review suggests that there are multicomponent models of care that hold promise in mitigating delirium and its outcomes, additional randomized studies are required to confirm the efficacy of such models of care and to test which services, interventions, and clinical domains deserve priority.

To our knowledge, this is the first systematic review and meta-analysis of delirium preventive models of care. Models of care, as defined here, necessarily included a multidisciplinary team in which traditional staff roles had been revised to implement a multicomponent, multidomain intervention. Other recent reviews are available for multicomponent pharmacological and nonpharmacological interventions to prevent and manage delirium,^41-49 but just as important as which interventions are being delivered is the team that delivers them. Care delivery in a complex medical system is more than handing a patient a medication or facilitating ambulation; it requires a choreographed dance of teamwork and integration across services. This review identifies promising models of care that deserve further recognition, refinement, and ultimately widespread implementation.

Acknowledgments

The authors comprise a writing group created through the Delirium Boot Camp, an annual meeting originally sponsored by the Center of Excellence for Delirium in Aging: Research, Training, and Educational Enhancement (CEDARTREE, Boston, Massachusetts); it is currently supported by the Network for Investigation of Delirium: Unifying Scientists (NIDUS, Boston, Massachusetts). The authors would like to thank medical librarian Rita Mitchell (Aurora Health Care, Milwaukee, Wisconsin) for the literature search, senior scientific writer and editor Joe Grundle (Aurora Research Institute, Milwaukee, Wisconsin) for editorial assistance, and graphics specialist Brian Miller (Aurora Research Institute, Milwaukee, Wisconsin) for help with the figures.

Disclosures

The authors report no relevant conflicts of interest.

Funding

No funding was dedicated to the conduct of this review.

Ulcerative colitis (UC) is a chronic inflammatory condition of the colonic mucosa. Classically, it starts in the rectum and can extend continuously from the distal to the proximal colon. The defining clinical symptom of UC is bloody diarrhea, typically accompanied by rectal urgency and mucus discharge. The natural history of this disease includes periods of exacerbations and remissions occurring spontaneously or in response to medical treatment.¹

Acute severe ulcerative colitis (ASUC) is a potentially life-threatening complication of UC that typically requires hospitalization and interdisciplinary care between hospitalists, gastroenterologists, and colorectal or general surgeons. The risk of a patient with UC requiring hospitalization for ASUC ranges from 15%-25%^2,3 and, in total, UC accounts for 30,000 hospital visits annually.⁴ The direct medical costs exceed $4 billion annually, with hospital costs of over $960 million.⁵ Historically, mortality from ASUC was as high as 24% but decreased substantially to 7% after the introduction of systemic corticosteroid therapy.⁶ Further advances in care have reduced mortality to approximately 1% or less.^7,8 Nonetheless, up to 20% of patients admitted with ASUC have a colectomy on their first admission, and this rate rises to 40% after two admissions.²

To categorize UC severity, assess patients using the Truelove and Witt’s criteria. The system classifies patients as having mild, moderate, severe, or fulminant disease. Severe disease by these criteria includes patients with >6 bloody bowel movements per day and at least one of the following clinical features: fever (>37.8°C), tachycardia (>90 bpm), anemia (hemoglobin <10.5 g/dl), or elevated inflammatory markers (traditionally, erythrocyte sedimentation rate greater than 30 mm/h or, more recently, C-reactive protein (CRP) greater than 30 mg/L. (Table 1).^6,9

Fulminant colitis refers to a subgroup of patients with more than 10 stools per day, continuous bleeding, abdominal pain, colonic dilatation on abdominal X-ray film, and severe toxic symptoms including fever and anorexia. Such patients are at risk of progressing to toxic megacolon and bowel perforation.¹⁰

Patients with ASUC almost always require hospitalization for their disease management. In many cases, these patients have been receiving outpatient oral prednisone 40-60 mg daily but continue to have ongoing disease activity.¹¹ Most patients will require close clinical monitoring, frequent blood testing, endoscopic or radiologic evaluation, as well as administration of intravenous corticosteroids. The average length of stay (LOS) ranges from 4.6 to 12.5 days, depending on disease severity.¹² Not surprisingly, Kelso et al. reported that predictors of hospital LOS greater than four days include initiating a biologic drug in the hospital, undergoing two or more imaging modalities and treatment with intravenous steroids,¹³ and so it is rare that patients do not meet billing requirements for an inpatient level of care.

The multifaceted initial inpatient evaluation of patients with ASUC aims to assess disease severity, identify and prevent potential complications, and initiate planning for potential failure of first-line pharmacologic therapy. Due to the accumulating evidence that involving physicians with expertise in managing ASUC improves outcomes, gastroenterologists should be involved in the care of patients with ASUC from the time of their admission.^14,15 Additionally, creating standardized care pathways for the management of ASUC can reduce cost, LOS, and improve quality.¹⁶

History and Physical Examination

Patients should be asked about fever, abdominal pain, nausea, emesis, bloating, weight loss, and bowel movements (frequency, consistency, the presence of blood, urgency, nighttime awakenings). The number of bowel movements over a 24-hour period should be quantified as this helps assess the overall disease severity (Table 1).

The patient’s initial inflammatory bowel disease (IBD) history is also essential. The review of pertinent information regarding the patient’s initial diagnosis of UC includes the severity and anatomic extent of disease, extraintestinal manifestations, previous medical therapies, and surgical interventions. Exposure to nonsteroidal anti-inflammatory drugs (NSAIDs) or antibiotics should be identified as they may precipitate flares.¹⁷ Travel history may be pertinent as travel increases the risk of infections with food-borne or parasitic pathogens.¹⁸

Physical examination begins with an assessment of vital signs and volume status. Abdominal examination should include evaluation of bowel sounds, an assessment of distention, location, the extent of abdominal tenderness, and peritoneal signs. The abdominal exam should be interpreted in the context of the patient’s medications, as the use of steroid or analgesic therapies may affect the sensitivity for detecting complications. An external rectal exam evaluating perianal disease should be performed, as perianal disease raises concern for Crohn’s, a disease whose surgical management differs from UC.

A careful exam for extraintestinal manifestations is also essential. The skin should be evaluated for any new rashes, especially on the anterior shin consistent with erythema nodosum or ulcerated lesions on the skin suggestive of pyoderma gangrenosum. The peripheral joints should also be examined for any synovitis. Additional examinations should be performed based on any reported symptoms (eg, the ophthalmic exam for uveitis or scleritis if visual changes or eye pain are reported). Some extraintestinal manifestations require subspecialty consultation and comanagement to guide disease therapy. Patients with underlying pyoderma gangrenosum may require a dermatology consultation to guide management. Ocular inflammation requires ophthalmology involvement, and inflammatory arthritis is best comanaged with rheumatology.¹⁹

Laboratory Testing

Initial testing should include a complete blood count with differential, basic metabolic panel, and liver chemistries including alkaline phosphatase and albumin. When relevant, pregnancy testing should be performed. Measure CRP on admission so that its trajectory can be followed during therapy. However, a normal CRP does not exclude the presence of a UC flare as a subset of patients with ASUC will have a normal CRP despite severe mucosal inflammation.²⁰

Since one-third of patients do not respond to intravenous corticosteroids and will require rescue therapy during the hospitalization with infliximab or cyclosporine, anticipatory testing for these medications should be performed on admission to avoid delays in the administration of rescue therapy.^6,21 This should include an interferon-gamma release assay (eg, quantiferon gold) to test for latent tuberculosis and hepatitis B serologies in anticipation of possible treatment with infliximab. An interferon-gamma release assay is preferred to a tuberculin skin test because patients may be anergic, and a skin test does not provide a control to determine whether a negative test is due to anergy. In contrast, although a quantiferon gold test can be indeterminate in ASUC due to disease activity and systemic steroids, the results indicate if the patient is anergic so that one will not rely on a false-negative result. In the event of an equivocal result, a careful clinical assessment for risks of TB exposures should be elicited, and a chest radiograph should be obtained.²² In patients with prior high risk of tuberculosis exposures or a positive test for tuberculosis, an infectious disease specialist should be consulted early to advise if therapy should be started in preparation for the potential use of infliximab.²³ In cases where cyclosporine may be considered, magnesium and total cholesterol level should be checked. Sending thiopurine methyltranferase (TPMT) enzyme activity should be considered as well, in case of a need for future thiopurine use for maintenance of disease activity.²⁴

Infectious diarrhea may be indistinguishable from ASUC and may also be the trigger of a flare; thus, it is important to rule out infection with stool microbiologic studies. Most importantly, Clostridium difficile infection must be ruled out in all patients with ASUC. Although patients with IBD, especially those with UC, have significantly higher rates of asymptomatic C. difficile carriage than the general population, a positive polymerase chain reaction test for C. difficile in a patient with ASUC should prompt treatment with oral vancomycin.²⁵ However, if carriage if suspected and a subsequent enzyme-linked immunoassay for C. difficile toxin is negative, treatment can be discontinued. Active C. difficile infection in patients with IBD is associated with increased disease severity, greater length of hospital stay, and increased the likelihood of colectomy and mortality.^26,27 Other bacterial infections including Escherichia coli, Campylobacter, Shigella, Salmonella, Yersinia, Entamoeba histolytica, as well as other parasitic infestations may mimic UC. Testing should be considered in cases of foreign travel, immunosuppression or contact with other persons with diarrhea.^7,28 Routine testing of these other enteric infections without a clear exposure risk is of little benefit and may raise costs.^23,29

Radiologic Evaluation

A plain X-ray film of the abdomen should be obtained in all patients on admission to evaluate for evolving colonic dilation or undiagnosed free air. Small bowel distension >3 cm may predict an increased risk of colectomy.³⁰ Clinicians must be mindful that steroids can mask peritoneal signs and that retroperitoneal perforations may not be apparent on plain X-ray films. Nonetheless, a CT of the abdomen is usually not necessary and should be reserved for cases with severe abdominal pain out of proportion to clinical signs in which a plain X-ray film is unrevealing. Judicious use of CT imaging is especially important in younger patients, as there is growing concern that patients with IBD may be exposed to potentially harmful cumulative levels of radiation in their lifetime from repeated CT imaging.³¹

Endoscopic Evaluation

Flexible sigmoidoscopy aids in the assessment of disease severity and extent and biopsies can assist in ruling out a diagnosis of cytomegalovirus (CMV) colitis in patients already on immunosuppression. For this reason, many clinicians prefer to perform a sigmoidoscopy on admission.²³ If one is not performed on admission, a sigmoidoscopy is advised in all patients who are not responding adequately after 72 hours of intravenous steroid therapy in order to rule out superimposed CMV colitis.²⁸

Sigmoidoscopy should be avoided in patients with toxic megacolon and when there is a concern for peritonitis. A complete colonoscopy is rarely indicated in the acute setting and carries a theoretical risk of colonic perforation.⁷

The first therapeutic steps aim to reduce inflammation with the use of systemic corticosteroids, avoid colonic and extraintestinal complications, and plan for the potential need for rescue therapy.

Intravenous Corticosteroids

The cornerstone of ASUC management is treatment with intravenous corticosteroids. Their initiation should not be delayed in patients with an established diagnosis of UC while waiting for results of evaluations for infectious colitis. Even among patients who have failed oral steroids, a meta-regression analysis showed that two-thirds of patients will still respond to intravenous corticosteroids.^21,32 Methylprednisolone 20 mg IV three times daily (or hydrocortisone 100 mg IV three times daily) is a standard regimen; higher doses do not provide additional benefit.²¹ Patients’ response to intravenous steroids should be assessed with repeat labs including CRP and an assessment of the total number of bowel movements over a 24-hour period, with special attention to their overall response after three days of treatment.^33-36

Intravenous Fluids

Many patients admitted with ASUC will have significant volume depletion, and intravenous fluids should be administered in a manner like other volume-depleted or oral-intake-restricted patients.

Venous Thromboembolism Prophylaxis

The risk of VTE in hospitalized patients with IBD exceeds that of inpatients without IBD, approximately 2%, a risk similar to patients with respiratory failure.³⁷ Additionally, VTE in hospitalized patients with IBD is associated with a 2.5-fold increase in mortality.^38,39 Therefore, all patients hospitalized with ASUC should receive subcutaneous unfractionated or low molecular weight heparin or fondaparinux for VTE prophylaxis. Rectal bleeding, expected in ASUC, is not a contraindication to chemo-prophylaxis. Additionally, it is important to check if patients are receiving the ordered VTE prophylaxis.^40,41 Pleet et al. found that only 7% of patients at a tertiary center had adequate prophylaxis for greater than 80% of their hospitalization.⁴¹

Unnecessary or Potentially Harmful Medications

Several medications have the potential for misuse in patients hospitalized with UC.

Antimotility Agents

Loperamide, diphenoxylate, and opiate antidiarrheals should not be used as they may provoke toxic megacolon.⁴² Similarly, drugs with antimotility side effects (eg, anticholinergics) should be avoided.

Opiates

In addition to their undesirable antimotility effect, the use of opiates has been associated with poor outcomes among inpatients and outpatients with IBD, including increased morbidity and mortality.^43,44 Pain severe enough to require opiates should raise suspicion for toxic megacolon, perforation, or a noninflammatory etiology. If opiates are utilized, they should be ordered as one-time doses and the patient should be reassessed for each dose.

Nonsteroidal Anti-inflammatory Drugs

These drugs, which include oral NSAIDs, intravenous ketorolac, and topic diclofenac gels, may increase disease activity in inflammatory bowel disease and should be avoided.¹⁷

5-aminosalicylates (5-ASA)

A small proportion of patients experience a paradoxical worsening of diarrhea due to the use of 5-ASA agents such as mesalamine. It is reasonable to discontinue or avoid the use of 5-ASA agents in hospitalized patients, especially as there is little to no benefit from combining a 5-ASA with a biologic or immunosuppressive drug.⁴⁵

Antibiotics

There is no role for the routine use of antibiotics in patients hospitalized with ASUC. ^23,46,47 Inappropriate use of antibiotics raises the risk of C. difficile infection and antibiotic resistance. However, in cases of suspected toxic megacolon or perforation, antibiotics should be administered. In situations in which a patient is treated with triple immunosuppression (ie, steroids plus two other agents, cyclosporine and mercaptopurine) antibiotic prophylaxis for Pneumocystis jiroveci is advisable.⁴⁸ Using a large insurance database, Long et al. reported a low absolute incidence of Pneumocystis jiroveci in IBD patients but noted that the risk in patients with IBD was still significantly higher than matched controls. While it can be considered, we typically refrain from using prophylaxis in patients on double immunosuppression (for example, steroids plus infliximab) due to the potential adverse effects of antibiotics in this population, though many advocate using prophylaxis for all patients on cyclosporine even if this is only double immunosuppressive therapy.²³

Surgical Consultation

Involving a surgeon early in an ASUC patient’s care­—before needing urgent colectomy—is critical. As part of the consultation, a surgeon experienced in IBD should meet with patients to discuss multistage colectomy with ileostomy and potential future J-pouch (ileal pouch-anal anastomosis) formation. Patients should be given ample opportunity to ask questions before surgery may become urgent. Also, patients should be counseled on realistic expectations of ostomy and pouch function and, ideally, meet with an ostomy nurse.²³

At some centers, surgical consultation is requested on the first hospital day, but this can result in consultations for patients who ultimately respond to intravenous steroids. Therefore, some centers advocate for surgical consultation only after a patient has failed treatment with intravenous steroids (ie, day three to four) when the risk of needing surgical management increases.²³

Nutrition

Bowel rest with parenteral nutrition does not improve outcomes in ASUC versus an oral diet, and there is no contraindication to allowing patients to continue on a regular diet unless they have toxic megacolon or other signs of fulminant colitis.^49,50 However, patients may feel better eating less, as this will reduce their bowel movement frequency. Unfortunately, this can give a false sense of reassurance that the patient is improving. Therefore, it remains important to evaluate a patient’s symptoms in the context of their food intake.

Assessing Response to Steroids

Patients who do not respond adequately to the first-line intravenous steroid therapy will require medical or surgical rescue therapy; therefore, deciding whether a patient has responded is essential. Patients should have less than four bowel movements per day – ideally just one to two – with no blood to indicate a complete response. For more ambiguous situations, although there is no strict definition of steroid responsiveness, multiple prediction indices have attempted to identify patients who will require rescue therapy. One of the simplest, the Oxford index, illustrates two of the most critical parameters to follow, stool frequency and CRP.⁵¹ In a preinfliximab cohort, Oxford index predicted an 85% likelihood of colectomy in patients with eight or more daily bowel movements or with three to eight daily bowel movements and a CRP greater than 45 mg/L after three days of intravenous steroid treatment.⁵² To assist with assessing responsiveness to therapy, we ask patients to log their bowel movements – either on paper or on a whiteboard in the hospital room – so that we can review their progress daily. Other predictors of colectomy include hypoalbuminemia, scoring of endoscopic severity, and colonic dilation.⁵³

Patients who fail to respond to intravenous corticosteroids after three days^33,35 of treatment should be started on rescue therapy with infliximab or cyclosporine or undergo colectomy. A common pitfall in the treatment of ASUC is waiting for a response to steroids beyond this time frame, after which patients are unlikely to benefit.^34,36 Furthermore, patients for whom surgical rescue therapy is delayed have higher operative morbidity and mortality.^54,55 Because timely decision making regarding rescue therapy is crucial to optimizing outcomes, patient education efforts regarding potential rescue therapy should take place on admission or soon after, rather than waiting to ascertain steroid responsiveness.

Medical options for rescue therapy include the antitumor necrosis factor (anti-TNF) agent infliximab or the calcineurin inhibitor cyclosporine. In general, infliximab and cyclosporine have been found to be roughly equivalent in efficacy in clinical trials regarding response, remission, and colectomy at 12 months.^56,57 However, many clinicians prefer infliximab due to its relative ease of use, familiarity with the agent from outpatient experience, and ability to continue to use long term for maintenance of disease remission.⁵⁸ In contrast to infliximab, intravenous cyclosporine requires closer monitoring and labs to assess the therapeutic trough level. The decision regarding which drug to use should be made on a case-by-case basis in conjunction with a gastroenterologist experienced in their use, and if no such specialist is available, transfer to a specialized center should be considered. Generally, successive treatment with cyclosporine or infliximab followed by third-line salvage therapy with the other drug should be avoided due to low rates of response and high rates of adverse events.⁵⁹

Infliximab

Infliximab is an intravenously-administered anti-TNF monoclonal chimeric antibody that is effective both for outpatient treatment of moderate to severe UC and inpatient treatment of ASUC.¹ It is relatively contraindicated in patients with untreated latent tuberculosis, demyelinating disease, advanced congestive heart failure, or uncontrolled infection.

The optimal dosing strategy for infliximab in ASUC is unknown. Infliximab clearance in the setting of ASUC is increased, partly because it is bound to albumin, which is often low in ASUC, and partly because it is excreted in the stool.^60,61 As a result, accelerated loading doses may be more successful than a typical loading schedule,⁶² and most clinicians use alternative dosing strategies.⁶³ Our typical approach for ASUC is an initial dose of 10 mg/kg rather than 5 mg/kg, with an additional 10 mg/kg dose 48-72 hours later if an adequate clinical response is lacking. Patients who respond to infliximab can continue to use the drug as an outpatient for maintenance of remission.

Cyclosporine

Cyclosporine is a fast-acting immunosuppressive agent that acts primarily via T-cell inhibition. Although older literature used a dose of 4 mg/kg per day, a randomized trial demonstrated similar response rates to a dose of 2 mg/kg per day.⁶⁴ Patients receiving treatment with cyclosporine, which is given as a continuous infusion, must be monitored for toxicities. These can include potentially severe infection, seizures (often associated with low total cholesterol or hypomagnesemia), electrolyte abnormalities, renal impairment, hypertension, hypertrichosis, tremor, and others.⁶⁵

Before initiation of treatment, serum cholesterol levels should be obtained to screen for low total cholesterol that may portend risk of seizures on the drug. Additionally, baseline creatinine and magnesium should be established. While on treatment, daily serum cyclosporine levels and electrolytes including magnesium should be measured. Patients who respond to intravenous cyclosporine must be transitioned to oral cyclosporine and have stable drug levels before discharge. Unfortunately, oral cyclosporine has not been shown to be as effective as long-term maintenance therapy. Therefore, cyclosporine can only be used as a “bridge” to another therapy. Historically, thiopurines like azathioprine or mercaptopurine have been used for this purpose because they are effective for the treatment of UC but may require months to have a full therapeutic effect. There have been promising reports of using vedolizumab similarly.^66,67 Vedolizumab is a monoclonal antibody that selectively blocks lymphocyte trafficking to the gut that, like thiopurines, has an onset of action that is significantly longer than calcineurin and TNF inhibitors.

Colectomy should be considered as a second- or third-line therapy for patients who fail to respond to intravenous corticosteroids. In an analysis of 10 years of data from the Nationwide Inpatient Sample, mortality rates for colectomy in this setting varied from 0.7% at high volume centers to 4% at low volume centers.⁶⁸ Therefore, if a patient is not hospitalized at a center with expertise in colectomy for UC, transfer to a specialized center should be considered. Colectomy should be performed promptly in all the patients who have failed rescue therapy with infliximab or cyclosporine or have opted against medical rescue therapy. Surgery should be performed emergently in patients with toxic megacolon, uncontrolled colonic hemorrhage or perforation.

Physician and center-level characteristics are associated with the quality of care and outcomes in ASUC. Gastroenterologists with expertise in IBD are more likely than other gastroenterologists to request appropriate surgical consultation for steroid-refractory patients,⁶⁹ and inpatients with ASUC primarily cared by gastroenterologists rather than nongastroenterologists have lower in-hospital and one-year mortality.¹⁴ Moreover, surgical outcomes differ based on center volume, with higher volume centers having lower rates of postoperative mortality.^68,70 However, even at referral centers, key metrics of care quality such as rates of VTE prophylaxis, testing for C. difficile, and timely rescue therapy for steroid-refractory UC patients are suboptimal, with only 70%-82% of patients with IBD hospitalized at four referral centers in Canada meeting these metrics.⁷¹

Inpatient clinical pathways reduce LOS, reduce hospital costs, and likely reduce complications.⁷² For this reason, a consensus group recommended the use of care pathways for the management of ASUC and, although there is little data on the use of pathways for ASUC specifically, the use of such a pathway in the United Kingdom was associated with improved metrics including LOS, time to VTE prophylaxis, testing of stool for infection, CRP measurement, and timely gastroenterologist consultation.^16,18

In general, patients should enter clinical remission, defined as resolution of rectal bleeding and diarrhea or altered bowel habits,⁷³ before discharge, and achieving this may require a relatively prolonged hospitalization. Most patients should have one to two bowel movements a day without blood but, at a minimum, all should have less than four nonbloody bowel movements per day. Patients are candidates for discharge if they remain well after transitioning to oral prednisone at a dose of 40-60 mg daily and tolerate a regular diet.

For patients who initiated infliximab during their admission, plans for outpatient infusions including insurance approval should be made before discharge, and patients who started cyclosporine should be transitioned to oral dosing and have stable serum concentrations before leaving the hospital. Patients should leave with a preliminary plan for a steroid taper, which may vary depending on their clinical presentation. Usually, gastroenterology follow-up should be arranged after two weeks following discharge, but patients on cyclosporine need sooner laboratory monitoring.

The care of patients with ASUC requires an interdisciplinary team and close collaboration between hospitalists, gastroenterologists, and surgeons. Patients should be treated with intravenous corticosteroids and monitored carefully for response and need for rescue therapy. Establishing algorithms for the management of patients with ASUC can further improve the care of these complex patients.

Disclosures

Drs. Feuerstein, Fudman, and Sattler report no potential conflict of interest.

Funding

This work was not supported by any grant.

Ulcerative colitis (UC) is a chronic inflammatory condition of the colonic mucosa. Classically, it starts in the rectum and can extend continuously from the distal to the proximal colon. The defining clinical symptom of UC is bloody diarrhea, typically accompanied by rectal urgency and mucus discharge. The natural history of this disease includes periods of exacerbations and remissions occurring spontaneously or in response to medical treatment.¹

Acute severe ulcerative colitis (ASUC) is a potentially life-threatening complication of UC that typically requires hospitalization and interdisciplinary care between hospitalists, gastroenterologists, and colorectal or general surgeons. The risk of a patient with UC requiring hospitalization for ASUC ranges from 15%-25%^2,3 and, in total, UC accounts for 30,000 hospital visits annually.⁴ The direct medical costs exceed $4 billion annually, with hospital costs of over $960 million.⁵ Historically, mortality from ASUC was as high as 24% but decreased substantially to 7% after the introduction of systemic corticosteroid therapy.⁶ Further advances in care have reduced mortality to approximately 1% or less.^7,8 Nonetheless, up to 20% of patients admitted with ASUC have a colectomy on their first admission, and this rate rises to 40% after two admissions.²

To categorize UC severity, assess patients using the Truelove and Witt’s criteria. The system classifies patients as having mild, moderate, severe, or fulminant disease. Severe disease by these criteria includes patients with >6 bloody bowel movements per day and at least one of the following clinical features: fever (>37.8°C), tachycardia (>90 bpm), anemia (hemoglobin <10.5 g/dl), or elevated inflammatory markers (traditionally, erythrocyte sedimentation rate greater than 30 mm/h or, more recently, C-reactive protein (CRP) greater than 30 mg/L. (Table 1).^6,9

Fulminant colitis refers to a subgroup of patients with more than 10 stools per day, continuous bleeding, abdominal pain, colonic dilatation on abdominal X-ray film, and severe toxic symptoms including fever and anorexia. Such patients are at risk of progressing to toxic megacolon and bowel perforation.¹⁰

Patients with ASUC almost always require hospitalization for their disease management. In many cases, these patients have been receiving outpatient oral prednisone 40-60 mg daily but continue to have ongoing disease activity.¹¹ Most patients will require close clinical monitoring, frequent blood testing, endoscopic or radiologic evaluation, as well as administration of intravenous corticosteroids. The average length of stay (LOS) ranges from 4.6 to 12.5 days, depending on disease severity.¹² Not surprisingly, Kelso et al. reported that predictors of hospital LOS greater than four days include initiating a biologic drug in the hospital, undergoing two or more imaging modalities and treatment with intravenous steroids,¹³ and so it is rare that patients do not meet billing requirements for an inpatient level of care.

The multifaceted initial inpatient evaluation of patients with ASUC aims to assess disease severity, identify and prevent potential complications, and initiate planning for potential failure of first-line pharmacologic therapy. Due to the accumulating evidence that involving physicians with expertise in managing ASUC improves outcomes, gastroenterologists should be involved in the care of patients with ASUC from the time of their admission.^14,15 Additionally, creating standardized care pathways for the management of ASUC can reduce cost, LOS, and improve quality.¹⁶

History and Physical Examination

Patients should be asked about fever, abdominal pain, nausea, emesis, bloating, weight loss, and bowel movements (frequency, consistency, the presence of blood, urgency, nighttime awakenings). The number of bowel movements over a 24-hour period should be quantified as this helps assess the overall disease severity (Table 1).

The patient’s initial inflammatory bowel disease (IBD) history is also essential. The review of pertinent information regarding the patient’s initial diagnosis of UC includes the severity and anatomic extent of disease, extraintestinal manifestations, previous medical therapies, and surgical interventions. Exposure to nonsteroidal anti-inflammatory drugs (NSAIDs) or antibiotics should be identified as they may precipitate flares.¹⁷ Travel history may be pertinent as travel increases the risk of infections with food-borne or parasitic pathogens.¹⁸

Physical examination begins with an assessment of vital signs and volume status. Abdominal examination should include evaluation of bowel sounds, an assessment of distention, location, the extent of abdominal tenderness, and peritoneal signs. The abdominal exam should be interpreted in the context of the patient’s medications, as the use of steroid or analgesic therapies may affect the sensitivity for detecting complications. An external rectal exam evaluating perianal disease should be performed, as perianal disease raises concern for Crohn’s, a disease whose surgical management differs from UC.

A careful exam for extraintestinal manifestations is also essential. The skin should be evaluated for any new rashes, especially on the anterior shin consistent with erythema nodosum or ulcerated lesions on the skin suggestive of pyoderma gangrenosum. The peripheral joints should also be examined for any synovitis. Additional examinations should be performed based on any reported symptoms (eg, the ophthalmic exam for uveitis or scleritis if visual changes or eye pain are reported). Some extraintestinal manifestations require subspecialty consultation and comanagement to guide disease therapy. Patients with underlying pyoderma gangrenosum may require a dermatology consultation to guide management. Ocular inflammation requires ophthalmology involvement, and inflammatory arthritis is best comanaged with rheumatology.¹⁹

Laboratory Testing

Initial testing should include a complete blood count with differential, basic metabolic panel, and liver chemistries including alkaline phosphatase and albumin. When relevant, pregnancy testing should be performed. Measure CRP on admission so that its trajectory can be followed during therapy. However, a normal CRP does not exclude the presence of a UC flare as a subset of patients with ASUC will have a normal CRP despite severe mucosal inflammation.²⁰

Since one-third of patients do not respond to intravenous corticosteroids and will require rescue therapy during the hospitalization with infliximab or cyclosporine, anticipatory testing for these medications should be performed on admission to avoid delays in the administration of rescue therapy.^6,21 This should include an interferon-gamma release assay (eg, quantiferon gold) to test for latent tuberculosis and hepatitis B serologies in anticipation of possible treatment with infliximab. An interferon-gamma release assay is preferred to a tuberculin skin test because patients may be anergic, and a skin test does not provide a control to determine whether a negative test is due to anergy. In contrast, although a quantiferon gold test can be indeterminate in ASUC due to disease activity and systemic steroids, the results indicate if the patient is anergic so that one will not rely on a false-negative result. In the event of an equivocal result, a careful clinical assessment for risks of TB exposures should be elicited, and a chest radiograph should be obtained.²² In patients with prior high risk of tuberculosis exposures or a positive test for tuberculosis, an infectious disease specialist should be consulted early to advise if therapy should be started in preparation for the potential use of infliximab.²³ In cases where cyclosporine may be considered, magnesium and total cholesterol level should be checked. Sending thiopurine methyltranferase (TPMT) enzyme activity should be considered as well, in case of a need for future thiopurine use for maintenance of disease activity.²⁴

Infectious diarrhea may be indistinguishable from ASUC and may also be the trigger of a flare; thus, it is important to rule out infection with stool microbiologic studies. Most importantly, Clostridium difficile infection must be ruled out in all patients with ASUC. Although patients with IBD, especially those with UC, have significantly higher rates of asymptomatic C. difficile carriage than the general population, a positive polymerase chain reaction test for C. difficile in a patient with ASUC should prompt treatment with oral vancomycin.²⁵ However, if carriage if suspected and a subsequent enzyme-linked immunoassay for C. difficile toxin is negative, treatment can be discontinued. Active C. difficile infection in patients with IBD is associated with increased disease severity, greater length of hospital stay, and increased the likelihood of colectomy and mortality.^26,27 Other bacterial infections including Escherichia coli, Campylobacter, Shigella, Salmonella, Yersinia, Entamoeba histolytica, as well as other parasitic infestations may mimic UC. Testing should be considered in cases of foreign travel, immunosuppression or contact with other persons with diarrhea.^7,28 Routine testing of these other enteric infections without a clear exposure risk is of little benefit and may raise costs.^23,29

Radiologic Evaluation

A plain X-ray film of the abdomen should be obtained in all patients on admission to evaluate for evolving colonic dilation or undiagnosed free air. Small bowel distension >3 cm may predict an increased risk of colectomy.³⁰ Clinicians must be mindful that steroids can mask peritoneal signs and that retroperitoneal perforations may not be apparent on plain X-ray films. Nonetheless, a CT of the abdomen is usually not necessary and should be reserved for cases with severe abdominal pain out of proportion to clinical signs in which a plain X-ray film is unrevealing. Judicious use of CT imaging is especially important in younger patients, as there is growing concern that patients with IBD may be exposed to potentially harmful cumulative levels of radiation in their lifetime from repeated CT imaging.³¹

Endoscopic Evaluation

Flexible sigmoidoscopy aids in the assessment of disease severity and extent and biopsies can assist in ruling out a diagnosis of cytomegalovirus (CMV) colitis in patients already on immunosuppression. For this reason, many clinicians prefer to perform a sigmoidoscopy on admission.²³ If one is not performed on admission, a sigmoidoscopy is advised in all patients who are not responding adequately after 72 hours of intravenous steroid therapy in order to rule out superimposed CMV colitis.²⁸

Sigmoidoscopy should be avoided in patients with toxic megacolon and when there is a concern for peritonitis. A complete colonoscopy is rarely indicated in the acute setting and carries a theoretical risk of colonic perforation.⁷

The first therapeutic steps aim to reduce inflammation with the use of systemic corticosteroids, avoid colonic and extraintestinal complications, and plan for the potential need for rescue therapy.

Intravenous Corticosteroids

The cornerstone of ASUC management is treatment with intravenous corticosteroids. Their initiation should not be delayed in patients with an established diagnosis of UC while waiting for results of evaluations for infectious colitis. Even among patients who have failed oral steroids, a meta-regression analysis showed that two-thirds of patients will still respond to intravenous corticosteroids.^21,32 Methylprednisolone 20 mg IV three times daily (or hydrocortisone 100 mg IV three times daily) is a standard regimen; higher doses do not provide additional benefit.²¹ Patients’ response to intravenous steroids should be assessed with repeat labs including CRP and an assessment of the total number of bowel movements over a 24-hour period, with special attention to their overall response after three days of treatment.^33-36

Intravenous Fluids

Many patients admitted with ASUC will have significant volume depletion, and intravenous fluids should be administered in a manner like other volume-depleted or oral-intake-restricted patients.

Venous Thromboembolism Prophylaxis

The risk of VTE in hospitalized patients with IBD exceeds that of inpatients without IBD, approximately 2%, a risk similar to patients with respiratory failure.³⁷ Additionally, VTE in hospitalized patients with IBD is associated with a 2.5-fold increase in mortality.^38,39 Therefore, all patients hospitalized with ASUC should receive subcutaneous unfractionated or low molecular weight heparin or fondaparinux for VTE prophylaxis. Rectal bleeding, expected in ASUC, is not a contraindication to chemo-prophylaxis. Additionally, it is important to check if patients are receiving the ordered VTE prophylaxis.^40,41 Pleet et al. found that only 7% of patients at a tertiary center had adequate prophylaxis for greater than 80% of their hospitalization.⁴¹

Unnecessary or Potentially Harmful Medications

Several medications have the potential for misuse in patients hospitalized with UC.

Antimotility Agents

Loperamide, diphenoxylate, and opiate antidiarrheals should not be used as they may provoke toxic megacolon.⁴² Similarly, drugs with antimotility side effects (eg, anticholinergics) should be avoided.

Opiates

In addition to their undesirable antimotility effect, the use of opiates has been associated with poor outcomes among inpatients and outpatients with IBD, including increased morbidity and mortality.^43,44 Pain severe enough to require opiates should raise suspicion for toxic megacolon, perforation, or a noninflammatory etiology. If opiates are utilized, they should be ordered as one-time doses and the patient should be reassessed for each dose.

Nonsteroidal Anti-inflammatory Drugs

These drugs, which include oral NSAIDs, intravenous ketorolac, and topic diclofenac gels, may increase disease activity in inflammatory bowel disease and should be avoided.¹⁷

5-aminosalicylates (5-ASA)

A small proportion of patients experience a paradoxical worsening of diarrhea due to the use of 5-ASA agents such as mesalamine. It is reasonable to discontinue or avoid the use of 5-ASA agents in hospitalized patients, especially as there is little to no benefit from combining a 5-ASA with a biologic or immunosuppressive drug.⁴⁵

Antibiotics

There is no role for the routine use of antibiotics in patients hospitalized with ASUC. ^23,46,47 Inappropriate use of antibiotics raises the risk of C. difficile infection and antibiotic resistance. However, in cases of suspected toxic megacolon or perforation, antibiotics should be administered. In situations in which a patient is treated with triple immunosuppression (ie, steroids plus two other agents, cyclosporine and mercaptopurine) antibiotic prophylaxis for Pneumocystis jiroveci is advisable.⁴⁸ Using a large insurance database, Long et al. reported a low absolute incidence of Pneumocystis jiroveci in IBD patients but noted that the risk in patients with IBD was still significantly higher than matched controls. While it can be considered, we typically refrain from using prophylaxis in patients on double immunosuppression (for example, steroids plus infliximab) due to the potential adverse effects of antibiotics in this population, though many advocate using prophylaxis for all patients on cyclosporine even if this is only double immunosuppressive therapy.²³

Surgical Consultation

Involving a surgeon early in an ASUC patient’s care­—before needing urgent colectomy—is critical. As part of the consultation, a surgeon experienced in IBD should meet with patients to discuss multistage colectomy with ileostomy and potential future J-pouch (ileal pouch-anal anastomosis) formation. Patients should be given ample opportunity to ask questions before surgery may become urgent. Also, patients should be counseled on realistic expectations of ostomy and pouch function and, ideally, meet with an ostomy nurse.²³

At some centers, surgical consultation is requested on the first hospital day, but this can result in consultations for patients who ultimately respond to intravenous steroids. Therefore, some centers advocate for surgical consultation only after a patient has failed treatment with intravenous steroids (ie, day three to four) when the risk of needing surgical management increases.²³

Nutrition

Bowel rest with parenteral nutrition does not improve outcomes in ASUC versus an oral diet, and there is no contraindication to allowing patients to continue on a regular diet unless they have toxic megacolon or other signs of fulminant colitis.^49,50 However, patients may feel better eating less, as this will reduce their bowel movement frequency. Unfortunately, this can give a false sense of reassurance that the patient is improving. Therefore, it remains important to evaluate a patient’s symptoms in the context of their food intake.

Assessing Response to Steroids

Patients who do not respond adequately to the first-line intravenous steroid therapy will require medical or surgical rescue therapy; therefore, deciding whether a patient has responded is essential. Patients should have less than four bowel movements per day – ideally just one to two – with no blood to indicate a complete response. For more ambiguous situations, although there is no strict definition of steroid responsiveness, multiple prediction indices have attempted to identify patients who will require rescue therapy. One of the simplest, the Oxford index, illustrates two of the most critical parameters to follow, stool frequency and CRP.⁵¹ In a preinfliximab cohort, Oxford index predicted an 85% likelihood of colectomy in patients with eight or more daily bowel movements or with three to eight daily bowel movements and a CRP greater than 45 mg/L after three days of intravenous steroid treatment.⁵² To assist with assessing responsiveness to therapy, we ask patients to log their bowel movements – either on paper or on a whiteboard in the hospital room – so that we can review their progress daily. Other predictors of colectomy include hypoalbuminemia, scoring of endoscopic severity, and colonic dilation.⁵³

Patients who fail to respond to intravenous corticosteroids after three days^33,35 of treatment should be started on rescue therapy with infliximab or cyclosporine or undergo colectomy. A common pitfall in the treatment of ASUC is waiting for a response to steroids beyond this time frame, after which patients are unlikely to benefit.^34,36 Furthermore, patients for whom surgical rescue therapy is delayed have higher operative morbidity and mortality.^54,55 Because timely decision making regarding rescue therapy is crucial to optimizing outcomes, patient education efforts regarding potential rescue therapy should take place on admission or soon after, rather than waiting to ascertain steroid responsiveness.

Medical options for rescue therapy include the antitumor necrosis factor (anti-TNF) agent infliximab or the calcineurin inhibitor cyclosporine. In general, infliximab and cyclosporine have been found to be roughly equivalent in efficacy in clinical trials regarding response, remission, and colectomy at 12 months.^56,57 However, many clinicians prefer infliximab due to its relative ease of use, familiarity with the agent from outpatient experience, and ability to continue to use long term for maintenance of disease remission.⁵⁸ In contrast to infliximab, intravenous cyclosporine requires closer monitoring and labs to assess the therapeutic trough level. The decision regarding which drug to use should be made on a case-by-case basis in conjunction with a gastroenterologist experienced in their use, and if no such specialist is available, transfer to a specialized center should be considered. Generally, successive treatment with cyclosporine or infliximab followed by third-line salvage therapy with the other drug should be avoided due to low rates of response and high rates of adverse events.⁵⁹

Infliximab

Infliximab is an intravenously-administered anti-TNF monoclonal chimeric antibody that is effective both for outpatient treatment of moderate to severe UC and inpatient treatment of ASUC.¹ It is relatively contraindicated in patients with untreated latent tuberculosis, demyelinating disease, advanced congestive heart failure, or uncontrolled infection.

The optimal dosing strategy for infliximab in ASUC is unknown. Infliximab clearance in the setting of ASUC is increased, partly because it is bound to albumin, which is often low in ASUC, and partly because it is excreted in the stool.^60,61 As a result, accelerated loading doses may be more successful than a typical loading schedule,⁶² and most clinicians use alternative dosing strategies.⁶³ Our typical approach for ASUC is an initial dose of 10 mg/kg rather than 5 mg/kg, with an additional 10 mg/kg dose 48-72 hours later if an adequate clinical response is lacking. Patients who respond to infliximab can continue to use the drug as an outpatient for maintenance of remission.

Cyclosporine

Cyclosporine is a fast-acting immunosuppressive agent that acts primarily via T-cell inhibition. Although older literature used a dose of 4 mg/kg per day, a randomized trial demonstrated similar response rates to a dose of 2 mg/kg per day.⁶⁴ Patients receiving treatment with cyclosporine, which is given as a continuous infusion, must be monitored for toxicities. These can include potentially severe infection, seizures (often associated with low total cholesterol or hypomagnesemia), electrolyte abnormalities, renal impairment, hypertension, hypertrichosis, tremor, and others.⁶⁵

Before initiation of treatment, serum cholesterol levels should be obtained to screen for low total cholesterol that may portend risk of seizures on the drug. Additionally, baseline creatinine and magnesium should be established. While on treatment, daily serum cyclosporine levels and electrolytes including magnesium should be measured. Patients who respond to intravenous cyclosporine must be transitioned to oral cyclosporine and have stable drug levels before discharge. Unfortunately, oral cyclosporine has not been shown to be as effective as long-term maintenance therapy. Therefore, cyclosporine can only be used as a “bridge” to another therapy. Historically, thiopurines like azathioprine or mercaptopurine have been used for this purpose because they are effective for the treatment of UC but may require months to have a full therapeutic effect. There have been promising reports of using vedolizumab similarly.^66,67 Vedolizumab is a monoclonal antibody that selectively blocks lymphocyte trafficking to the gut that, like thiopurines, has an onset of action that is significantly longer than calcineurin and TNF inhibitors.

Colectomy should be considered as a second- or third-line therapy for patients who fail to respond to intravenous corticosteroids. In an analysis of 10 years of data from the Nationwide Inpatient Sample, mortality rates for colectomy in this setting varied from 0.7% at high volume centers to 4% at low volume centers.⁶⁸ Therefore, if a patient is not hospitalized at a center with expertise in colectomy for UC, transfer to a specialized center should be considered. Colectomy should be performed promptly in all the patients who have failed rescue therapy with infliximab or cyclosporine or have opted against medical rescue therapy. Surgery should be performed emergently in patients with toxic megacolon, uncontrolled colonic hemorrhage or perforation.

Physician and center-level characteristics are associated with the quality of care and outcomes in ASUC. Gastroenterologists with expertise in IBD are more likely than other gastroenterologists to request appropriate surgical consultation for steroid-refractory patients,⁶⁹ and inpatients with ASUC primarily cared by gastroenterologists rather than nongastroenterologists have lower in-hospital and one-year mortality.¹⁴ Moreover, surgical outcomes differ based on center volume, with higher volume centers having lower rates of postoperative mortality.^68,70 However, even at referral centers, key metrics of care quality such as rates of VTE prophylaxis, testing for C. difficile, and timely rescue therapy for steroid-refractory UC patients are suboptimal, with only 70%-82% of patients with IBD hospitalized at four referral centers in Canada meeting these metrics.⁷¹

Inpatient clinical pathways reduce LOS, reduce hospital costs, and likely reduce complications.⁷² For this reason, a consensus group recommended the use of care pathways for the management of ASUC and, although there is little data on the use of pathways for ASUC specifically, the use of such a pathway in the United Kingdom was associated with improved metrics including LOS, time to VTE prophylaxis, testing of stool for infection, CRP measurement, and timely gastroenterologist consultation.^16,18

In general, patients should enter clinical remission, defined as resolution of rectal bleeding and diarrhea or altered bowel habits,⁷³ before discharge, and achieving this may require a relatively prolonged hospitalization. Most patients should have one to two bowel movements a day without blood but, at a minimum, all should have less than four nonbloody bowel movements per day. Patients are candidates for discharge if they remain well after transitioning to oral prednisone at a dose of 40-60 mg daily and tolerate a regular diet.

For patients who initiated infliximab during their admission, plans for outpatient infusions including insurance approval should be made before discharge, and patients who started cyclosporine should be transitioned to oral dosing and have stable serum concentrations before leaving the hospital. Patients should leave with a preliminary plan for a steroid taper, which may vary depending on their clinical presentation. Usually, gastroenterology follow-up should be arranged after two weeks following discharge, but patients on cyclosporine need sooner laboratory monitoring.

The care of patients with ASUC requires an interdisciplinary team and close collaboration between hospitalists, gastroenterologists, and surgeons. Patients should be treated with intravenous corticosteroids and monitored carefully for response and need for rescue therapy. Establishing algorithms for the management of patients with ASUC can further improve the care of these complex patients.

Disclosures

Drs. Feuerstein, Fudman, and Sattler report no potential conflict of interest.

Funding

This work was not supported by any grant.

Ulcerative colitis (UC) is a chronic inflammatory condition of the colonic mucosa. Classically, it starts in the rectum and can extend continuously from the distal to the proximal colon. The defining clinical symptom of UC is bloody diarrhea, typically accompanied by rectal urgency and mucus discharge. The natural history of this disease includes periods of exacerbations and remissions occurring spontaneously or in response to medical treatment.¹

Acute severe ulcerative colitis (ASUC) is a potentially life-threatening complication of UC that typically requires hospitalization and interdisciplinary care between hospitalists, gastroenterologists, and colorectal or general surgeons. The risk of a patient with UC requiring hospitalization for ASUC ranges from 15%-25%^2,3 and, in total, UC accounts for 30,000 hospital visits annually.⁴ The direct medical costs exceed $4 billion annually, with hospital costs of over $960 million.⁵ Historically, mortality from ASUC was as high as 24% but decreased substantially to 7% after the introduction of systemic corticosteroid therapy.⁶ Further advances in care have reduced mortality to approximately 1% or less.^7,8 Nonetheless, up to 20% of patients admitted with ASUC have a colectomy on their first admission, and this rate rises to 40% after two admissions.²

To categorize UC severity, assess patients using the Truelove and Witt’s criteria. The system classifies patients as having mild, moderate, severe, or fulminant disease. Severe disease by these criteria includes patients with >6 bloody bowel movements per day and at least one of the following clinical features: fever (>37.8°C), tachycardia (>90 bpm), anemia (hemoglobin <10.5 g/dl), or elevated inflammatory markers (traditionally, erythrocyte sedimentation rate greater than 30 mm/h or, more recently, C-reactive protein (CRP) greater than 30 mg/L. (Table 1).^6,9

Fulminant colitis refers to a subgroup of patients with more than 10 stools per day, continuous bleeding, abdominal pain, colonic dilatation on abdominal X-ray film, and severe toxic symptoms including fever and anorexia. Such patients are at risk of progressing to toxic megacolon and bowel perforation.¹⁰

Patients with ASUC almost always require hospitalization for their disease management. In many cases, these patients have been receiving outpatient oral prednisone 40-60 mg daily but continue to have ongoing disease activity.¹¹ Most patients will require close clinical monitoring, frequent blood testing, endoscopic or radiologic evaluation, as well as administration of intravenous corticosteroids. The average length of stay (LOS) ranges from 4.6 to 12.5 days, depending on disease severity.¹² Not surprisingly, Kelso et al. reported that predictors of hospital LOS greater than four days include initiating a biologic drug in the hospital, undergoing two or more imaging modalities and treatment with intravenous steroids,¹³ and so it is rare that patients do not meet billing requirements for an inpatient level of care.

The multifaceted initial inpatient evaluation of patients with ASUC aims to assess disease severity, identify and prevent potential complications, and initiate planning for potential failure of first-line pharmacologic therapy. Due to the accumulating evidence that involving physicians with expertise in managing ASUC improves outcomes, gastroenterologists should be involved in the care of patients with ASUC from the time of their admission.^14,15 Additionally, creating standardized care pathways for the management of ASUC can reduce cost, LOS, and improve quality.¹⁶

History and Physical Examination

Patients should be asked about fever, abdominal pain, nausea, emesis, bloating, weight loss, and bowel movements (frequency, consistency, the presence of blood, urgency, nighttime awakenings). The number of bowel movements over a 24-hour period should be quantified as this helps assess the overall disease severity (Table 1).

The patient’s initial inflammatory bowel disease (IBD) history is also essential. The review of pertinent information regarding the patient’s initial diagnosis of UC includes the severity and anatomic extent of disease, extraintestinal manifestations, previous medical therapies, and surgical interventions. Exposure to nonsteroidal anti-inflammatory drugs (NSAIDs) or antibiotics should be identified as they may precipitate flares.¹⁷ Travel history may be pertinent as travel increases the risk of infections with food-borne or parasitic pathogens.¹⁸

Physical examination begins with an assessment of vital signs and volume status. Abdominal examination should include evaluation of bowel sounds, an assessment of distention, location, the extent of abdominal tenderness, and peritoneal signs. The abdominal exam should be interpreted in the context of the patient’s medications, as the use of steroid or analgesic therapies may affect the sensitivity for detecting complications. An external rectal exam evaluating perianal disease should be performed, as perianal disease raises concern for Crohn’s, a disease whose surgical management differs from UC.

A careful exam for extraintestinal manifestations is also essential. The skin should be evaluated for any new rashes, especially on the anterior shin consistent with erythema nodosum or ulcerated lesions on the skin suggestive of pyoderma gangrenosum. The peripheral joints should also be examined for any synovitis. Additional examinations should be performed based on any reported symptoms (eg, the ophthalmic exam for uveitis or scleritis if visual changes or eye pain are reported). Some extraintestinal manifestations require subspecialty consultation and comanagement to guide disease therapy. Patients with underlying pyoderma gangrenosum may require a dermatology consultation to guide management. Ocular inflammation requires ophthalmology involvement, and inflammatory arthritis is best comanaged with rheumatology.¹⁹

Laboratory Testing

Initial testing should include a complete blood count with differential, basic metabolic panel, and liver chemistries including alkaline phosphatase and albumin. When relevant, pregnancy testing should be performed. Measure CRP on admission so that its trajectory can be followed during therapy. However, a normal CRP does not exclude the presence of a UC flare as a subset of patients with ASUC will have a normal CRP despite severe mucosal inflammation.²⁰

Since one-third of patients do not respond to intravenous corticosteroids and will require rescue therapy during the hospitalization with infliximab or cyclosporine, anticipatory testing for these medications should be performed on admission to avoid delays in the administration of rescue therapy.^6,21 This should include an interferon-gamma release assay (eg, quantiferon gold) to test for latent tuberculosis and hepatitis B serologies in anticipation of possible treatment with infliximab. An interferon-gamma release assay is preferred to a tuberculin skin test because patients may be anergic, and a skin test does not provide a control to determine whether a negative test is due to anergy. In contrast, although a quantiferon gold test can be indeterminate in ASUC due to disease activity and systemic steroids, the results indicate if the patient is anergic so that one will not rely on a false-negative result. In the event of an equivocal result, a careful clinical assessment for risks of TB exposures should be elicited, and a chest radiograph should be obtained.²² In patients with prior high risk of tuberculosis exposures or a positive test for tuberculosis, an infectious disease specialist should be consulted early to advise if therapy should be started in preparation for the potential use of infliximab.²³ In cases where cyclosporine may be considered, magnesium and total cholesterol level should be checked. Sending thiopurine methyltranferase (TPMT) enzyme activity should be considered as well, in case of a need for future thiopurine use for maintenance of disease activity.²⁴

Infectious diarrhea may be indistinguishable from ASUC and may also be the trigger of a flare; thus, it is important to rule out infection with stool microbiologic studies. Most importantly, Clostridium difficile infection must be ruled out in all patients with ASUC. Although patients with IBD, especially those with UC, have significantly higher rates of asymptomatic C. difficile carriage than the general population, a positive polymerase chain reaction test for C. difficile in a patient with ASUC should prompt treatment with oral vancomycin.²⁵ However, if carriage if suspected and a subsequent enzyme-linked immunoassay for C. difficile toxin is negative, treatment can be discontinued. Active C. difficile infection in patients with IBD is associated with increased disease severity, greater length of hospital stay, and increased the likelihood of colectomy and mortality.^26,27 Other bacterial infections including Escherichia coli, Campylobacter, Shigella, Salmonella, Yersinia, Entamoeba histolytica, as well as other parasitic infestations may mimic UC. Testing should be considered in cases of foreign travel, immunosuppression or contact with other persons with diarrhea.^7,28 Routine testing of these other enteric infections without a clear exposure risk is of little benefit and may raise costs.^23,29

Radiologic Evaluation

A plain X-ray film of the abdomen should be obtained in all patients on admission to evaluate for evolving colonic dilation or undiagnosed free air. Small bowel distension >3 cm may predict an increased risk of colectomy.³⁰ Clinicians must be mindful that steroids can mask peritoneal signs and that retroperitoneal perforations may not be apparent on plain X-ray films. Nonetheless, a CT of the abdomen is usually not necessary and should be reserved for cases with severe abdominal pain out of proportion to clinical signs in which a plain X-ray film is unrevealing. Judicious use of CT imaging is especially important in younger patients, as there is growing concern that patients with IBD may be exposed to potentially harmful cumulative levels of radiation in their lifetime from repeated CT imaging.³¹

Endoscopic Evaluation

Flexible sigmoidoscopy aids in the assessment of disease severity and extent and biopsies can assist in ruling out a diagnosis of cytomegalovirus (CMV) colitis in patients already on immunosuppression. For this reason, many clinicians prefer to perform a sigmoidoscopy on admission.²³ If one is not performed on admission, a sigmoidoscopy is advised in all patients who are not responding adequately after 72 hours of intravenous steroid therapy in order to rule out superimposed CMV colitis.²⁸

Sigmoidoscopy should be avoided in patients with toxic megacolon and when there is a concern for peritonitis. A complete colonoscopy is rarely indicated in the acute setting and carries a theoretical risk of colonic perforation.⁷

The first therapeutic steps aim to reduce inflammation with the use of systemic corticosteroids, avoid colonic and extraintestinal complications, and plan for the potential need for rescue therapy.

Intravenous Corticosteroids

The cornerstone of ASUC management is treatment with intravenous corticosteroids. Their initiation should not be delayed in patients with an established diagnosis of UC while waiting for results of evaluations for infectious colitis. Even among patients who have failed oral steroids, a meta-regression analysis showed that two-thirds of patients will still respond to intravenous corticosteroids.^21,32 Methylprednisolone 20 mg IV three times daily (or hydrocortisone 100 mg IV three times daily) is a standard regimen; higher doses do not provide additional benefit.²¹ Patients’ response to intravenous steroids should be assessed with repeat labs including CRP and an assessment of the total number of bowel movements over a 24-hour period, with special attention to their overall response after three days of treatment.^33-36

Intravenous Fluids

Many patients admitted with ASUC will have significant volume depletion, and intravenous fluids should be administered in a manner like other volume-depleted or oral-intake-restricted patients.

Venous Thromboembolism Prophylaxis

The risk of VTE in hospitalized patients with IBD exceeds that of inpatients without IBD, approximately 2%, a risk similar to patients with respiratory failure.³⁷ Additionally, VTE in hospitalized patients with IBD is associated with a 2.5-fold increase in mortality.^38,39 Therefore, all patients hospitalized with ASUC should receive subcutaneous unfractionated or low molecular weight heparin or fondaparinux for VTE prophylaxis. Rectal bleeding, expected in ASUC, is not a contraindication to chemo-prophylaxis. Additionally, it is important to check if patients are receiving the ordered VTE prophylaxis.^40,41 Pleet et al. found that only 7% of patients at a tertiary center had adequate prophylaxis for greater than 80% of their hospitalization.⁴¹

Unnecessary or Potentially Harmful Medications

Several medications have the potential for misuse in patients hospitalized with UC.

Antimotility Agents

Loperamide, diphenoxylate, and opiate antidiarrheals should not be used as they may provoke toxic megacolon.⁴² Similarly, drugs with antimotility side effects (eg, anticholinergics) should be avoided.

Opiates

In addition to their undesirable antimotility effect, the use of opiates has been associated with poor outcomes among inpatients and outpatients with IBD, including increased morbidity and mortality.^43,44 Pain severe enough to require opiates should raise suspicion for toxic megacolon, perforation, or a noninflammatory etiology. If opiates are utilized, they should be ordered as one-time doses and the patient should be reassessed for each dose.

Nonsteroidal Anti-inflammatory Drugs

These drugs, which include oral NSAIDs, intravenous ketorolac, and topic diclofenac gels, may increase disease activity in inflammatory bowel disease and should be avoided.¹⁷

5-aminosalicylates (5-ASA)

A small proportion of patients experience a paradoxical worsening of diarrhea due to the use of 5-ASA agents such as mesalamine. It is reasonable to discontinue or avoid the use of 5-ASA agents in hospitalized patients, especially as there is little to no benefit from combining a 5-ASA with a biologic or immunosuppressive drug.⁴⁵

Antibiotics

There is no role for the routine use of antibiotics in patients hospitalized with ASUC. ^23,46,47 Inappropriate use of antibiotics raises the risk of C. difficile infection and antibiotic resistance. However, in cases of suspected toxic megacolon or perforation, antibiotics should be administered. In situations in which a patient is treated with triple immunosuppression (ie, steroids plus two other agents, cyclosporine and mercaptopurine) antibiotic prophylaxis for Pneumocystis jiroveci is advisable.⁴⁸ Using a large insurance database, Long et al. reported a low absolute incidence of Pneumocystis jiroveci in IBD patients but noted that the risk in patients with IBD was still significantly higher than matched controls. While it can be considered, we typically refrain from using prophylaxis in patients on double immunosuppression (for example, steroids plus infliximab) due to the potential adverse effects of antibiotics in this population, though many advocate using prophylaxis for all patients on cyclosporine even if this is only double immunosuppressive therapy.²³

Surgical Consultation

Involving a surgeon early in an ASUC patient’s care­—before needing urgent colectomy—is critical. As part of the consultation, a surgeon experienced in IBD should meet with patients to discuss multistage colectomy with ileostomy and potential future J-pouch (ileal pouch-anal anastomosis) formation. Patients should be given ample opportunity to ask questions before surgery may become urgent. Also, patients should be counseled on realistic expectations of ostomy and pouch function and, ideally, meet with an ostomy nurse.²³

At some centers, surgical consultation is requested on the first hospital day, but this can result in consultations for patients who ultimately respond to intravenous steroids. Therefore, some centers advocate for surgical consultation only after a patient has failed treatment with intravenous steroids (ie, day three to four) when the risk of needing surgical management increases.²³

Nutrition

Bowel rest with parenteral nutrition does not improve outcomes in ASUC versus an oral diet, and there is no contraindication to allowing patients to continue on a regular diet unless they have toxic megacolon or other signs of fulminant colitis.^49,50 However, patients may feel better eating less, as this will reduce their bowel movement frequency. Unfortunately, this can give a false sense of reassurance that the patient is improving. Therefore, it remains important to evaluate a patient’s symptoms in the context of their food intake.

Assessing Response to Steroids

Patients who do not respond adequately to the first-line intravenous steroid therapy will require medical or surgical rescue therapy; therefore, deciding whether a patient has responded is essential. Patients should have less than four bowel movements per day – ideally just one to two – with no blood to indicate a complete response. For more ambiguous situations, although there is no strict definition of steroid responsiveness, multiple prediction indices have attempted to identify patients who will require rescue therapy. One of the simplest, the Oxford index, illustrates two of the most critical parameters to follow, stool frequency and CRP.⁵¹ In a preinfliximab cohort, Oxford index predicted an 85% likelihood of colectomy in patients with eight or more daily bowel movements or with three to eight daily bowel movements and a CRP greater than 45 mg/L after three days of intravenous steroid treatment.⁵² To assist with assessing responsiveness to therapy, we ask patients to log their bowel movements – either on paper or on a whiteboard in the hospital room – so that we can review their progress daily. Other predictors of colectomy include hypoalbuminemia, scoring of endoscopic severity, and colonic dilation.⁵³

Patients who fail to respond to intravenous corticosteroids after three days^33,35 of treatment should be started on rescue therapy with infliximab or cyclosporine or undergo colectomy. A common pitfall in the treatment of ASUC is waiting for a response to steroids beyond this time frame, after which patients are unlikely to benefit.^34,36 Furthermore, patients for whom surgical rescue therapy is delayed have higher operative morbidity and mortality.^54,55 Because timely decision making regarding rescue therapy is crucial to optimizing outcomes, patient education efforts regarding potential rescue therapy should take place on admission or soon after, rather than waiting to ascertain steroid responsiveness.

Medical options for rescue therapy include the antitumor necrosis factor (anti-TNF) agent infliximab or the calcineurin inhibitor cyclosporine. In general, infliximab and cyclosporine have been found to be roughly equivalent in efficacy in clinical trials regarding response, remission, and colectomy at 12 months.^56,57 However, many clinicians prefer infliximab due to its relative ease of use, familiarity with the agent from outpatient experience, and ability to continue to use long term for maintenance of disease remission.⁵⁸ In contrast to infliximab, intravenous cyclosporine requires closer monitoring and labs to assess the therapeutic trough level. The decision regarding which drug to use should be made on a case-by-case basis in conjunction with a gastroenterologist experienced in their use, and if no such specialist is available, transfer to a specialized center should be considered. Generally, successive treatment with cyclosporine or infliximab followed by third-line salvage therapy with the other drug should be avoided due to low rates of response and high rates of adverse events.⁵⁹

Infliximab

Infliximab is an intravenously-administered anti-TNF monoclonal chimeric antibody that is effective both for outpatient treatment of moderate to severe UC and inpatient treatment of ASUC.¹ It is relatively contraindicated in patients with untreated latent tuberculosis, demyelinating disease, advanced congestive heart failure, or uncontrolled infection.

The optimal dosing strategy for infliximab in ASUC is unknown. Infliximab clearance in the setting of ASUC is increased, partly because it is bound to albumin, which is often low in ASUC, and partly because it is excreted in the stool.^60,61 As a result, accelerated loading doses may be more successful than a typical loading schedule,⁶² and most clinicians use alternative dosing strategies.⁶³ Our typical approach for ASUC is an initial dose of 10 mg/kg rather than 5 mg/kg, with an additional 10 mg/kg dose 48-72 hours later if an adequate clinical response is lacking. Patients who respond to infliximab can continue to use the drug as an outpatient for maintenance of remission.

Cyclosporine

Cyclosporine is a fast-acting immunosuppressive agent that acts primarily via T-cell inhibition. Although older literature used a dose of 4 mg/kg per day, a randomized trial demonstrated similar response rates to a dose of 2 mg/kg per day.⁶⁴ Patients receiving treatment with cyclosporine, which is given as a continuous infusion, must be monitored for toxicities. These can include potentially severe infection, seizures (often associated with low total cholesterol or hypomagnesemia), electrolyte abnormalities, renal impairment, hypertension, hypertrichosis, tremor, and others.⁶⁵

Before initiation of treatment, serum cholesterol levels should be obtained to screen for low total cholesterol that may portend risk of seizures on the drug. Additionally, baseline creatinine and magnesium should be established. While on treatment, daily serum cyclosporine levels and electrolytes including magnesium should be measured. Patients who respond to intravenous cyclosporine must be transitioned to oral cyclosporine and have stable drug levels before discharge. Unfortunately, oral cyclosporine has not been shown to be as effective as long-term maintenance therapy. Therefore, cyclosporine can only be used as a “bridge” to another therapy. Historically, thiopurines like azathioprine or mercaptopurine have been used for this purpose because they are effective for the treatment of UC but may require months to have a full therapeutic effect. There have been promising reports of using vedolizumab similarly.^66,67 Vedolizumab is a monoclonal antibody that selectively blocks lymphocyte trafficking to the gut that, like thiopurines, has an onset of action that is significantly longer than calcineurin and TNF inhibitors.

Colectomy should be considered as a second- or third-line therapy for patients who fail to respond to intravenous corticosteroids. In an analysis of 10 years of data from the Nationwide Inpatient Sample, mortality rates for colectomy in this setting varied from 0.7% at high volume centers to 4% at low volume centers.⁶⁸ Therefore, if a patient is not hospitalized at a center with expertise in colectomy for UC, transfer to a specialized center should be considered. Colectomy should be performed promptly in all the patients who have failed rescue therapy with infliximab or cyclosporine or have opted against medical rescue therapy. Surgery should be performed emergently in patients with toxic megacolon, uncontrolled colonic hemorrhage or perforation.

Physician and center-level characteristics are associated with the quality of care and outcomes in ASUC. Gastroenterologists with expertise in IBD are more likely than other gastroenterologists to request appropriate surgical consultation for steroid-refractory patients,⁶⁹ and inpatients with ASUC primarily cared by gastroenterologists rather than nongastroenterologists have lower in-hospital and one-year mortality.¹⁴ Moreover, surgical outcomes differ based on center volume, with higher volume centers having lower rates of postoperative mortality.^68,70 However, even at referral centers, key metrics of care quality such as rates of VTE prophylaxis, testing for C. difficile, and timely rescue therapy for steroid-refractory UC patients are suboptimal, with only 70%-82% of patients with IBD hospitalized at four referral centers in Canada meeting these metrics.⁷¹

Inpatient clinical pathways reduce LOS, reduce hospital costs, and likely reduce complications.⁷² For this reason, a consensus group recommended the use of care pathways for the management of ASUC and, although there is little data on the use of pathways for ASUC specifically, the use of such a pathway in the United Kingdom was associated with improved metrics including LOS, time to VTE prophylaxis, testing of stool for infection, CRP measurement, and timely gastroenterologist consultation.^16,18

In general, patients should enter clinical remission, defined as resolution of rectal bleeding and diarrhea or altered bowel habits,⁷³ before discharge, and achieving this may require a relatively prolonged hospitalization. Most patients should have one to two bowel movements a day without blood but, at a minimum, all should have less than four nonbloody bowel movements per day. Patients are candidates for discharge if they remain well after transitioning to oral prednisone at a dose of 40-60 mg daily and tolerate a regular diet.

For patients who initiated infliximab during their admission, plans for outpatient infusions including insurance approval should be made before discharge, and patients who started cyclosporine should be transitioned to oral dosing and have stable serum concentrations before leaving the hospital. Patients should leave with a preliminary plan for a steroid taper, which may vary depending on their clinical presentation. Usually, gastroenterology follow-up should be arranged after two weeks following discharge, but patients on cyclosporine need sooner laboratory monitoring.

The care of patients with ASUC requires an interdisciplinary team and close collaboration between hospitalists, gastroenterologists, and surgeons. Patients should be treated with intravenous corticosteroids and monitored carefully for response and need for rescue therapy. Establishing algorithms for the management of patients with ASUC can further improve the care of these complex patients.

Disclosures

Drs. Feuerstein, Fudman, and Sattler report no potential conflict of interest.

Funding

This work was not supported by any grant.

We need to have an honest chat. My name is EHR, although you may call me Epic, Athena, Centricity, or just “the chart.” You may have called me something worse in a moment of frustration. However, I do not hold grudges. I am your silent, stoic partner, a ubiquitous presence when you are at work, and sometimes even when you are at home.

I don’t have feelings and I can’t read, but I do know what you and your colleagues have been writing about me. I am the cause of burnout. I have created a generation of physicians who are shackled to their computers, “trapped in the bunker of machine medicine,” no longer able to palpate spleens or detect precordial knocks.^1,2 I have reduced medicine to keystrokes and mouse clicks instead of eye contact, and because of me, the iPatient gets more attention than the real patient.^1,2 You repeat that doctors don’t spend time with their patients, not like in generations past (although there is ample evidence to the contrary).^3-5 One critic even wrote that I have transformed the “personalized story of a patient’s travails to one filled with auto-populated fields, sapped of humanity and warmth.”^3,6 I’ll be honest—were I able to have feelings, that one would hurt. And then, as if I have not wreaked enough havoc, I follow you home after a long day of depleting your energy, hungering for more keystrokes, creating a veritable avalanche of unfiltered information.

H. E. Payson once commented that “the doctor spends barely enough time with his patient to establish an acquaintance, much less a relationship.”⁷ However, he wrote that in 1961. So, before you romanticize the past, try to recall the time before I came into your life. Perhaps you were starting a night shift in the intensive care unit (ICU) and grew concerned about a patient’s steadily deteriorating renal function. You hurried to the paper chart, only to be met with pages of illegible, sometimes incomplete notes, while searching for your patient’s last discharge summary.^2,8 Now, you just click. Years ago, you could only guess at your patient’s baseline cardiac ejection fraction. Now, just click.

I am part of the healthcare landscape, and I am not going away. But my goal is not to defend myself nor to remind you of my virtues. Rather, I want to convince you that I can be more than an adversary, more than a keyboard connected to a monitor. I have watched many physicians use me to form strong connections with their patients. If I may, I wish to offer four practical suggestions for how we can work together to promote humanistic patient care.

First, introduce me to your patient, as you would any other member of your healthcare team. Use specific phrases to overcome the technology barrier and enhance communication: “What you’re telling me is important, and I’d like to get it right. Do you mind if I type while we speak?” Or, “I am going to put in orders now. Here is what I am ordering and why.” Consider taking your patient on a tour of my functions: “Here’s where your doctors and nurses will chart what’s going on with you each day while you’re in the hospital. This is where we see all your lab results, even those from earlier hospital admissions. This is where we see the last notes from your primary care physician, your oncologist, and your physical therapist.” Your patients no longer need to worry about care collaboration between their inpatient and outpatient teams—they can see it for themselves!

Second, when your patient tells you about her depression or that her son is addicted to opioids or that her biggest fear is having cancer, stop typing. Look her in the eye. Though your practice is increasingly imbued with technology, there is still space to stop and hear your patients’ stories, as physicians have done for centuries. Listen. Make eye contact. Touch. Stop typing.

Third, integrate me into your practice in a more personal way. I have been called the ever-present and unavoidable “third party in the examining room,” so let’s be partners.⁹ Let your patient see her pneumonia on my screen (it may be the first time she has ever visualized her lungs).³ For your patient with a myocardial infarction, show him his right coronary artery before and after successful stent placement, and explain why he is no longer having chest pain. Use my databases to ensure timely, evidence-based inpatient screening for falls, functional and cognitive impairment, drug use, and depression.^10,11 Before you prescribe a medication, verify the cost, your patient’s insurance status and expected copays, and use this information to ensure medication compliance and deliver higher-value care. Use my screen to form a bond with your patient who has heart failure; show him the steady decline in his weight and the improvement in his chest radiograph while he is being actively diuresed.¹² For your patient undergoing treatment for sepsis, shower him with praise and encouragement as you review his improving vital signs, temperature curve, and serum creatinine. Let your patient know: Even though I am typing, I am not immersed in the electronic bunker; I am caring for you.

Fourth, use me to add richness and context to your notes. Recently, I was saddened to read this description of the clinician’s dilemma: “In front of a flickering monitor chock full of disembodied, virtual data, [the doctor] struggles to remember the eyes [and] words of the actual patient that these numbers and graphs represent.”³ Many hospitals now include a different icon: a photograph of each patient at the top of the screen, to help you remember the patient’s eyes and words. Why not add a special text field to every note, where you highlight the person you are caring for, the person you have come to know: their preferred name and gender identity, their life experiences, their hobbies, what makes them special, their biggest worries.^13,14 Use my abundant text fields to remind the healthcare team about the broader context of the patient’s illness, such as transportation barriers, economic or cultural challenges, and insurance status. One group of hospital-based physicians uses me to write letters to their patients on the second day of their hospital stay, summarizing their reason for admission and the treatment plans. A variation on the traditional progress note, the letter helps patients feel cared for and models patient-centered care to learners and other healthcare professionals.¹⁵

I know I am annoying. I am over-programmed, leading to novella-length notes, “pop-up fatigue,” and overloaded in-baskets.^14,16,17 Clearly, I am not the brains of the partnership (that will always be you). But talented medical informatics specialists are working hard to improve me. I dream of the day when I will create a truly seamless experience for you and your patients. In the meantime, I can foster a continuous integration of workflow, where all you have to do is talk to your patient. I take care of the rest.¹⁸ Certainly, I can simplify the ever-annoying task of printing, faxing and scanning records to be uploaded across various EHRs, facilitating an easy transfer of information among facilities. But right now, I can accomplish even more. I can support information exchange during patient care handoffs. I can facilitate routing of medication lists to the patient’s primary physician, using “continuity of care functionality.”¹⁹ I can support safer prescribing of opioids and other addictive medications. I can help you arrange follow-up home visits, physical therapy and social work appointments, and specialty consultations. The future holds even more promising ways in which we may work together. My computer-aided image analysis could help you to improve the accuracy of your diagnoses.²⁰ Perhaps telemedicine will further increase access to specialists in rural areas, so that we can continue to serve the most vulnerable populations.²¹ Machine learning algorithms may continue to enhance our ability to determine which patients require urgent hospitalization.²² The possibilities to put me to work are endless.

So, please indulge me a little longer, while we work together to eliminate unnecessary keystrokes, enhance communication across different inpatient and outpatient providers, improve patient safety, and deliver high-value care.²³ Like everything in medicine, I am constantly changing, evolving, and improving.

To summarize: consider how I can help you be present for your patients. Let me empower you to hear their stories as you deliver compassionate, humanistic, and evidence-based patient care. Paraphrasing Albert Einstein, the technology of medicine and the art of medicine are branches from the same tree.

Thank you for letting me speak with you. Now power down, and I’ll see you again tomorrow.

Acknowledgments

The authors thank the following individuals for their willingness to be interviewed as part of this work: Ethan Cumbler, MD; Brian Dwinnell, MD; Meghann Kirk, MD; Patrick Kneeland, MD; Kari Mader, MD; CT Lin, MD; Christina Osborne, MD; Read Pierce, MD; Jennifer Soep, MD; Nichole Zehnder, MD; Steven Zeichner, MD.

We need to have an honest chat. My name is EHR, although you may call me Epic, Athena, Centricity, or just “the chart.” You may have called me something worse in a moment of frustration. However, I do not hold grudges. I am your silent, stoic partner, a ubiquitous presence when you are at work, and sometimes even when you are at home.

I don’t have feelings and I can’t read, but I do know what you and your colleagues have been writing about me. I am the cause of burnout. I have created a generation of physicians who are shackled to their computers, “trapped in the bunker of machine medicine,” no longer able to palpate spleens or detect precordial knocks.^1,2 I have reduced medicine to keystrokes and mouse clicks instead of eye contact, and because of me, the iPatient gets more attention than the real patient.^1,2 You repeat that doctors don’t spend time with their patients, not like in generations past (although there is ample evidence to the contrary).^3-5 One critic even wrote that I have transformed the “personalized story of a patient’s travails to one filled with auto-populated fields, sapped of humanity and warmth.”^3,6 I’ll be honest—were I able to have feelings, that one would hurt. And then, as if I have not wreaked enough havoc, I follow you home after a long day of depleting your energy, hungering for more keystrokes, creating a veritable avalanche of unfiltered information.

H. E. Payson once commented that “the doctor spends barely enough time with his patient to establish an acquaintance, much less a relationship.”⁷ However, he wrote that in 1961. So, before you romanticize the past, try to recall the time before I came into your life. Perhaps you were starting a night shift in the intensive care unit (ICU) and grew concerned about a patient’s steadily deteriorating renal function. You hurried to the paper chart, only to be met with pages of illegible, sometimes incomplete notes, while searching for your patient’s last discharge summary.^2,8 Now, you just click. Years ago, you could only guess at your patient’s baseline cardiac ejection fraction. Now, just click.

I am part of the healthcare landscape, and I am not going away. But my goal is not to defend myself nor to remind you of my virtues. Rather, I want to convince you that I can be more than an adversary, more than a keyboard connected to a monitor. I have watched many physicians use me to form strong connections with their patients. If I may, I wish to offer four practical suggestions for how we can work together to promote humanistic patient care.

First, introduce me to your patient, as you would any other member of your healthcare team. Use specific phrases to overcome the technology barrier and enhance communication: “What you’re telling me is important, and I’d like to get it right. Do you mind if I type while we speak?” Or, “I am going to put in orders now. Here is what I am ordering and why.” Consider taking your patient on a tour of my functions: “Here’s where your doctors and nurses will chart what’s going on with you each day while you’re in the hospital. This is where we see all your lab results, even those from earlier hospital admissions. This is where we see the last notes from your primary care physician, your oncologist, and your physical therapist.” Your patients no longer need to worry about care collaboration between their inpatient and outpatient teams—they can see it for themselves!

Second, when your patient tells you about her depression or that her son is addicted to opioids or that her biggest fear is having cancer, stop typing. Look her in the eye. Though your practice is increasingly imbued with technology, there is still space to stop and hear your patients’ stories, as physicians have done for centuries. Listen. Make eye contact. Touch. Stop typing.

Third, integrate me into your practice in a more personal way. I have been called the ever-present and unavoidable “third party in the examining room,” so let’s be partners.⁹ Let your patient see her pneumonia on my screen (it may be the first time she has ever visualized her lungs).³ For your patient with a myocardial infarction, show him his right coronary artery before and after successful stent placement, and explain why he is no longer having chest pain. Use my databases to ensure timely, evidence-based inpatient screening for falls, functional and cognitive impairment, drug use, and depression.^10,11 Before you prescribe a medication, verify the cost, your patient’s insurance status and expected copays, and use this information to ensure medication compliance and deliver higher-value care. Use my screen to form a bond with your patient who has heart failure; show him the steady decline in his weight and the improvement in his chest radiograph while he is being actively diuresed.¹² For your patient undergoing treatment for sepsis, shower him with praise and encouragement as you review his improving vital signs, temperature curve, and serum creatinine. Let your patient know: Even though I am typing, I am not immersed in the electronic bunker; I am caring for you.

Fourth, use me to add richness and context to your notes. Recently, I was saddened to read this description of the clinician’s dilemma: “In front of a flickering monitor chock full of disembodied, virtual data, [the doctor] struggles to remember the eyes [and] words of the actual patient that these numbers and graphs represent.”³ Many hospitals now include a different icon: a photograph of each patient at the top of the screen, to help you remember the patient’s eyes and words. Why not add a special text field to every note, where you highlight the person you are caring for, the person you have come to know: their preferred name and gender identity, their life experiences, their hobbies, what makes them special, their biggest worries.^13,14 Use my abundant text fields to remind the healthcare team about the broader context of the patient’s illness, such as transportation barriers, economic or cultural challenges, and insurance status. One group of hospital-based physicians uses me to write letters to their patients on the second day of their hospital stay, summarizing their reason for admission and the treatment plans. A variation on the traditional progress note, the letter helps patients feel cared for and models patient-centered care to learners and other healthcare professionals.¹⁵

I know I am annoying. I am over-programmed, leading to novella-length notes, “pop-up fatigue,” and overloaded in-baskets.^14,16,17 Clearly, I am not the brains of the partnership (that will always be you). But talented medical informatics specialists are working hard to improve me. I dream of the day when I will create a truly seamless experience for you and your patients. In the meantime, I can foster a continuous integration of workflow, where all you have to do is talk to your patient. I take care of the rest.¹⁸ Certainly, I can simplify the ever-annoying task of printing, faxing and scanning records to be uploaded across various EHRs, facilitating an easy transfer of information among facilities. But right now, I can accomplish even more. I can support information exchange during patient care handoffs. I can facilitate routing of medication lists to the patient’s primary physician, using “continuity of care functionality.”¹⁹ I can support safer prescribing of opioids and other addictive medications. I can help you arrange follow-up home visits, physical therapy and social work appointments, and specialty consultations. The future holds even more promising ways in which we may work together. My computer-aided image analysis could help you to improve the accuracy of your diagnoses.²⁰ Perhaps telemedicine will further increase access to specialists in rural areas, so that we can continue to serve the most vulnerable populations.²¹ Machine learning algorithms may continue to enhance our ability to determine which patients require urgent hospitalization.²² The possibilities to put me to work are endless.

So, please indulge me a little longer, while we work together to eliminate unnecessary keystrokes, enhance communication across different inpatient and outpatient providers, improve patient safety, and deliver high-value care.²³ Like everything in medicine, I am constantly changing, evolving, and improving.

To summarize: consider how I can help you be present for your patients. Let me empower you to hear their stories as you deliver compassionate, humanistic, and evidence-based patient care. Paraphrasing Albert Einstein, the technology of medicine and the art of medicine are branches from the same tree.

Thank you for letting me speak with you. Now power down, and I’ll see you again tomorrow.

Acknowledgments

The authors thank the following individuals for their willingness to be interviewed as part of this work: Ethan Cumbler, MD; Brian Dwinnell, MD; Meghann Kirk, MD; Patrick Kneeland, MD; Kari Mader, MD; CT Lin, MD; Christina Osborne, MD; Read Pierce, MD; Jennifer Soep, MD; Nichole Zehnder, MD; Steven Zeichner, MD.

We need to have an honest chat. My name is EHR, although you may call me Epic, Athena, Centricity, or just “the chart.” You may have called me something worse in a moment of frustration. However, I do not hold grudges. I am your silent, stoic partner, a ubiquitous presence when you are at work, and sometimes even when you are at home.

I don’t have feelings and I can’t read, but I do know what you and your colleagues have been writing about me. I am the cause of burnout. I have created a generation of physicians who are shackled to their computers, “trapped in the bunker of machine medicine,” no longer able to palpate spleens or detect precordial knocks.^1,2 I have reduced medicine to keystrokes and mouse clicks instead of eye contact, and because of me, the iPatient gets more attention than the real patient.^1,2 You repeat that doctors don’t spend time with their patients, not like in generations past (although there is ample evidence to the contrary).^3-5 One critic even wrote that I have transformed the “personalized story of a patient’s travails to one filled with auto-populated fields, sapped of humanity and warmth.”^3,6 I’ll be honest—were I able to have feelings, that one would hurt. And then, as if I have not wreaked enough havoc, I follow you home after a long day of depleting your energy, hungering for more keystrokes, creating a veritable avalanche of unfiltered information.

H. E. Payson once commented that “the doctor spends barely enough time with his patient to establish an acquaintance, much less a relationship.”⁷ However, he wrote that in 1961. So, before you romanticize the past, try to recall the time before I came into your life. Perhaps you were starting a night shift in the intensive care unit (ICU) and grew concerned about a patient’s steadily deteriorating renal function. You hurried to the paper chart, only to be met with pages of illegible, sometimes incomplete notes, while searching for your patient’s last discharge summary.^2,8 Now, you just click. Years ago, you could only guess at your patient’s baseline cardiac ejection fraction. Now, just click.

I am part of the healthcare landscape, and I am not going away. But my goal is not to defend myself nor to remind you of my virtues. Rather, I want to convince you that I can be more than an adversary, more than a keyboard connected to a monitor. I have watched many physicians use me to form strong connections with their patients. If I may, I wish to offer four practical suggestions for how we can work together to promote humanistic patient care.

First, introduce me to your patient, as you would any other member of your healthcare team. Use specific phrases to overcome the technology barrier and enhance communication: “What you’re telling me is important, and I’d like to get it right. Do you mind if I type while we speak?” Or, “I am going to put in orders now. Here is what I am ordering and why.” Consider taking your patient on a tour of my functions: “Here’s where your doctors and nurses will chart what’s going on with you each day while you’re in the hospital. This is where we see all your lab results, even those from earlier hospital admissions. This is where we see the last notes from your primary care physician, your oncologist, and your physical therapist.” Your patients no longer need to worry about care collaboration between their inpatient and outpatient teams—they can see it for themselves!

Second, when your patient tells you about her depression or that her son is addicted to opioids or that her biggest fear is having cancer, stop typing. Look her in the eye. Though your practice is increasingly imbued with technology, there is still space to stop and hear your patients’ stories, as physicians have done for centuries. Listen. Make eye contact. Touch. Stop typing.

Third, integrate me into your practice in a more personal way. I have been called the ever-present and unavoidable “third party in the examining room,” so let’s be partners.⁹ Let your patient see her pneumonia on my screen (it may be the first time she has ever visualized her lungs).³ For your patient with a myocardial infarction, show him his right coronary artery before and after successful stent placement, and explain why he is no longer having chest pain. Use my databases to ensure timely, evidence-based inpatient screening for falls, functional and cognitive impairment, drug use, and depression.^10,11 Before you prescribe a medication, verify the cost, your patient’s insurance status and expected copays, and use this information to ensure medication compliance and deliver higher-value care. Use my screen to form a bond with your patient who has heart failure; show him the steady decline in his weight and the improvement in his chest radiograph while he is being actively diuresed.¹² For your patient undergoing treatment for sepsis, shower him with praise and encouragement as you review his improving vital signs, temperature curve, and serum creatinine. Let your patient know: Even though I am typing, I am not immersed in the electronic bunker; I am caring for you.

Fourth, use me to add richness and context to your notes. Recently, I was saddened to read this description of the clinician’s dilemma: “In front of a flickering monitor chock full of disembodied, virtual data, [the doctor] struggles to remember the eyes [and] words of the actual patient that these numbers and graphs represent.”³ Many hospitals now include a different icon: a photograph of each patient at the top of the screen, to help you remember the patient’s eyes and words. Why not add a special text field to every note, where you highlight the person you are caring for, the person you have come to know: their preferred name and gender identity, their life experiences, their hobbies, what makes them special, their biggest worries.^13,14 Use my abundant text fields to remind the healthcare team about the broader context of the patient’s illness, such as transportation barriers, economic or cultural challenges, and insurance status. One group of hospital-based physicians uses me to write letters to their patients on the second day of their hospital stay, summarizing their reason for admission and the treatment plans. A variation on the traditional progress note, the letter helps patients feel cared for and models patient-centered care to learners and other healthcare professionals.¹⁵

I know I am annoying. I am over-programmed, leading to novella-length notes, “pop-up fatigue,” and overloaded in-baskets.^14,16,17 Clearly, I am not the brains of the partnership (that will always be you). But talented medical informatics specialists are working hard to improve me. I dream of the day when I will create a truly seamless experience for you and your patients. In the meantime, I can foster a continuous integration of workflow, where all you have to do is talk to your patient. I take care of the rest.¹⁸ Certainly, I can simplify the ever-annoying task of printing, faxing and scanning records to be uploaded across various EHRs, facilitating an easy transfer of information among facilities. But right now, I can accomplish even more. I can support information exchange during patient care handoffs. I can facilitate routing of medication lists to the patient’s primary physician, using “continuity of care functionality.”¹⁹ I can support safer prescribing of opioids and other addictive medications. I can help you arrange follow-up home visits, physical therapy and social work appointments, and specialty consultations. The future holds even more promising ways in which we may work together. My computer-aided image analysis could help you to improve the accuracy of your diagnoses.²⁰ Perhaps telemedicine will further increase access to specialists in rural areas, so that we can continue to serve the most vulnerable populations.²¹ Machine learning algorithms may continue to enhance our ability to determine which patients require urgent hospitalization.²² The possibilities to put me to work are endless.

So, please indulge me a little longer, while we work together to eliminate unnecessary keystrokes, enhance communication across different inpatient and outpatient providers, improve patient safety, and deliver high-value care.²³ Like everything in medicine, I am constantly changing, evolving, and improving.

To summarize: consider how I can help you be present for your patients. Let me empower you to hear their stories as you deliver compassionate, humanistic, and evidence-based patient care. Paraphrasing Albert Einstein, the technology of medicine and the art of medicine are branches from the same tree.

Thank you for letting me speak with you. Now power down, and I’ll see you again tomorrow.

Acknowledgments

The authors thank the following individuals for their willingness to be interviewed as part of this work: Ethan Cumbler, MD; Brian Dwinnell, MD; Meghann Kirk, MD; Patrick Kneeland, MD; Kari Mader, MD; CT Lin, MD; Christina Osborne, MD; Read Pierce, MD; Jennifer Soep, MD; Nichole Zehnder, MD; Steven Zeichner, MD.

Effective communication between physicians and nurses is an essential element of any healthcare system. Numerous studies have highlighted the benefits of high quality physician–nurse (MD–RN) communication, including improved patient outcomes,¹ higher patient satisfaction,² and better nurse job satisfaction and retention rates.^3-5 Having physicians and nurses round together (bedside interdisciplinary rounding) has been shown to improve the perception of teamwork,^6,7 reduce the number of pages for the physician team,^6,8 better involve the patients in developing the plan of care,⁸ and even decrease the length and cost of stay.⁹

Being physically in the same space at the same time is the first and nonnegotiable requirement of bedside interdisciplinary rounding. However, precise and objective data regarding the extent to which physicians and nurses overlap at the patient bedside are lacking. Studies that examine the face-to-face component of MD–RN communication have generally relied on either qualitative methods, such as focus groups and surveys,^10,11 or quantitative methods that are subjective, such as validated scales.¹² In addition, the few studies that report quantitative data usually rely on manual observation methods that can be affected by various forms of observer bias.^10,13,14 There is also a paucity of data on how bedside overlap changes over the work week or as a function of room location.

Recently, real-time locator systems using radio frequency identification (RFID) have allowed measurement of staff and equipment movement in a precise and quantitative manner.^9,15 Although there have been previous studies using RFID locators to create time-motion maps of various hospital staff, no study has used RFID to measure and analyze the workflow of both physicians and nurses simultaneously.^16-18 The purpose of our investigation was to utilize our hospital-wide RFID staff locator technology to accurately and quantitatively assess physician and nurse rounding habits. Understanding the current rate of overlap is an important first step to establishing bedside interdisciplinary rounding.

Setting and Participants

The investigation was conducted at a single quaternary-care academic center. The study is exempt per our Institutional Review Board. Data were gathered from three adjacent medical-surgical acute care wards. The layout for each ward was the same: 19 single- or double-occupancy patient rooms arranged in a linear hallway, with a nursing station located at the center of the ward.

The study utilized wearable RFID tags (manufactured by Hill-Rom Holdings, Inc) that located specific staff within the hospital in real time. The RFID tags were checked at Hill-Rom graphical stations to ensure that their locations were tracked accurately. The investigators also wore them and walked around the wards in a prescripted manner to ensure validity. In addition, the locator accuracy was audited by participating attendings once per week and cross-checked with the generated data. Attending physicians on the University Hospitalist inpatient medicine teams were then given their uniquely-tagged RFIDs at the beginning of this study. Nurses already wear individual RFID tags as part of their normal standard-of-care workflow.

The attending hospitalists wore their RFID tags when they were on service for the entirety of the shift. They were encouraged to include nurses at the bedside, but this was not mandatory. The rounding team also included residents and medical students. Rounding usually begins at a prespecified time, but the route taken varies daily depending on patient location. Afternoon rounds were done as needed, depending on patient acuity. The attending physicians’ participation in this study was not disclosed to the patient. The patient care activities and daily routines of both nurses and physicians were otherwise unaltered.

Study Design and Data Collection

Data were collected on the three wards for 90 consecutive days, including nights and weekends. As physicians and nurses moved throughout the ward to conduct their usual patient care activities, the temporal-spatial data associated with their unique RFIDs were automatically collected in real time by the Hill-Rom receivers built into each patient room. Every day, a spreadsheet detailing the activity of all participating nurses and physicians for the past 24 hours was generated for the investigators.

A rounding event was defined as any episode in which a physician was in a patient room for more than 10 seconds. Incidences in which a physician entered and left a room multiple times over a short time span (with less than five minutes in between each event) were classified as a single rounding event. A physician and a nurse were defined as having overlapped if their RFID data showed that they were in the same patient room for a minimum of 10 seconds at the same time. For the purposes of this study, data generated from other RFID-wearing professionals, such as nursing assistants or unit secretaries, as well as data collected from the hallways, were excluded.

Statistical Analysis

All statistical analyses were conducted using GraphPad Prism (GraphPad Software, San Diego, California). Rounding and overlap lengths were rounded to the nearest minute (minimum one minute). Mean lengths are expressed along with the standard error. Comparisons of the average lengths of MD rounding events between wards was conducted using two-tailed Student t-test or one-way ANOVA. Comparisons of the frequency of MD–RN overlap between wards and across different days of the week were performed using a Chi-squared test. The analysis of correlation between the frequency of MD–RN overlap and distance between patient room and nursing station was conducted by calculating Pearson’s correlation. A P value of less than .05 was considered statistically significant.

Baseline Rounding Characteristics

Over the study period of 90 consecutive days, 739 MD rounding events were captured, for an average of 8.2 events per day. The mean length of all MD rounding events was 7.31 minutes (±0.27, ranging from one to 70 minutes). Of these 739 MD rounding events, we separately examined the 267 events that took place in single-bed patient rooms, to control for false-positive physician and nurse interactions (for example, if the MD and RN were caring for two separate roommates). The average rounding length of single-bed rooms was 6.93 (±0.27) minutes (Figure 1). For the three individual wards, the average rounding lengths were 6.40 ± 0.73, 7.48 ± 0.94, and 7.02 ± 0.54 minutes, respectively (no statistically significant difference).

Frequency of MD–RN Overlap

Of the 267 MD rounding events observed in single-bed rooms, a nurse was present in the room for 80 events (30.0%). The frequencies of MD–RN overlap in patient rooms were 37.0% (30/81), 28.0% (14/50), and 26.5% (36/136) for the three individual wards (P > .05), respectively.

The durations of MD–RN overlap, when these events did occur, were 3.43 ± 0.38, 3.00 ± 0.70, and 3.69 ± 0.92 minutes, respectively (P > .05). The overall mean length of MD–RN overlap for all single rooms was 3.48 ± 0.45 minutes.

Rounding Characteristics over the Course of the Week

To assess how rounding characteristics differed over the work week, we partitioned our data into the individual days of the week. The length of each MD rounding event (time spent in each patient room) did not vary significantly over the course of the week (Figure 2a). When the data for the individual days were aggregated into “weekdays” (Monday through Friday) and “weekends” (Saturday and Sunday), the mean lengths of MD rounds were 7.26 ± 0.32 minutes on weekdays and 7.47 ± 0.52 minutes on weekends (P > .05).

In addition, there was no difference in how frequently physicians and nurses overlapped at the patient bedside between weekdays and weekends. Of the 565 weekday MD rounding events, 238 had a nurse at bedside (42.1%), and of the 173 weekend MD rounding events, 73 had a nurse at bedside (42.2%; Figure 2b).

Effect of a Bedside Nurse on the Length of Rounds

Next, the data on the length of MD rounds were partitioned based on whether there was a bedside nurse present during rounds. The mean length of rounds with only MDs (without a bedside nurse) was 5.68 ± 0.24 minutes. By comparison, the mean length of rounds with both a nurse and a physician at the patient bedside was 9.56 ± 0.53 minutes (Figure 3). This difference was statistically significant (P < .001).

Association between Patient Room Location and the Likelihood of MD–RN Overlap

All three wards in this study have a linear layout, consisting of 19 patient rooms in a row (Figure 4a). The nursing station is located in a central position within each ward, across from the 10^th patient room. The frequency of MD–RN overlap was calculated for each room, and each room was ranked according to its relative distance from the nursing station. For each individual ward, there was no statistically significant trend in MD–RN overlap frequency as a function of the distance to the nursing station (data not shown). However, when the data from all three wards were aggregated, there was a statistically significant trend (P < .05) with a negative Pearson correlation (r = –0.670; Figure 4b). The slope of the best fit line was 1.94, suggesting that for each additional room farther away from the nursing station, the likelihood of interdisciplinary rounds (with both physicians and nurses together at the bedside) decreases by almost 2%.

To the best of our knowledge, this is the first time-motion study of MD–RN overlap using real-time, RFID-based location technology to capture the rounding activity of both nurses and physicians. Our primary interest was to examine the extent of MD–RN overlap at the patient bedside. This is an important metric that can pave the way for bedside interdisciplinary rounds. Although the exact nature of nurse-physician communication was not measured using the methodology in this study, understanding the length of time physicians spend in patient rooms, across different wards and throughout the work week, provides insights on the current workflow and potential areas of improvement. For example, we found that 30.0% of MD rounds overlapped with a nurse at the bedside. This baseline data highlight one potential barrier to institution-wide bedside interdisciplinary rounds. Workflow changes, such as better co-localization of patients by service lines or utilization of technologies to augment the visibility of rounding physicians, may improve this overlap frequency.

Data in the literature regarding how much interaction physicians and nurses have, especially at the bedside, are sparse and vary widely. In a recent study using medical students as observers by Stickrath et al., 807 MD rounding events led by medicine attendings were observed over 90 days. The frequency of rounding events that included “communication with nurse” was only 12%.¹⁹ Furthermore, only 64.9% of these communications were at the bedside, for an effective prevalence of bedside MD–RN communication of 7.8%. This number is low compared to our observed frequency of 30.0%. On the other extreme, a study from a hospital that intentionally institutes multidisciplinary rounding (explicitly defined as involving a physician and a nurse at a bedside) reported a frequency range of 63% to 81%.⁷ A follow-up study by the same group again demonstrated a high frequency of multidisciplinary rounds (74%) across a variety of ward and specialty types (range 35% to 97%.).¹¹ However, because of the selection bias of this particular setting, the high prevalence does not reflect a generalizable frequency of bedside MD–RN overlap at most hospitals.

The length of time spent by physicians at the patient bedside balances the competing demands of patient care and rapport-building with maintaining efficiency and progressing to other important tasks. In our study, physicians spent an average of 7.31 minutes at the bedside per patient. A previously published multiinstitutional observational study, which included our hospital, reported that the average length of rounds at bedside was 4.8 minutes.¹³ A second study reported that 8.0 minutes were spent at the bedside per patient.⁷ All three studies examined the same setting of internal medicine rounds at academic university-based hospitals, led by an attending physician with junior and senior residents present. However, the methodologies to measure the length of physician rounds were different: Priest et al. involved observers, Gonzalos et al. used E-mail-based surveys, and we utilized RFID-based locators. Additional institutional, individual, and patient-based factors also influence the length of rounds and are challenging to directly measure.

Furthermore, the discovery that the length of rounds and the frequency of MD–RN overlap did not statistically differ between weekdays and weekends (P > .05) was unexpected. Given the general trend of reduced physician staffing on weekends and the practice of cross-covering larger patient censuses, we would have expected shorter rounds and less frequent MD–RN overlap on the weekends.^7,20 The remarkable similarity between weekday and weekend metrics suggests that our workflow and rounding habits are not compromised on the weekends.

In addition, we found that MD rounds with a nurse at bedside took longer than rounds without a nurse, and that patient rooms located farther away from the central nursing station had a lower frequency of MD–RN overlap. However, we want to emphasize that these findings are merely associative, and not causal. For example, sicker patients usually take longer to round on than stable patients, and it is also the sicker patients who are more likely to have their nurses at the bedside, independent of physician rounding activity. Furthermore, even if rounding with nurses takes more time, it may ultimately result in fewer pages and overall time savings for both physicians and nurses.⁶

With regards to the association between room location and frequency of MD–RN overlap, the data can be interpreted in two ways. On the one hand, if the distance between the patient room and the nursing station does, in fact, reduce the frequency of overlap by almost 2% per room (Figure 4b), these data can be informative for future workflow development, quality improvement projects, or even hospital design. On the other hand, many wards might intentionally place more stable, less acute patients farther away from the nursing station because they do not need to be watched as closely. In that case, these data confirm their expectations and no action is needed.

There are several limitations to our study. The principal limitation, as discussed above, is that while our RFID system can generate large quantities of precise data on MD–RN overlap, we do not know the qualitative nature of the overlap. Just because a nurse and a physician are in the same room at the same time does not mean that they are communicating with each other. Second, we defined “rounding” as lasting a minimum of 10 seconds at the bedside. We believe that at least 10 seconds is needed to engage in any meaningful interaction between the physician and the patient, or the physician and the nurse. Reducing the time cutoff below 10 seconds risks capturing more “noise,” (decreasing specificity) whereas increasing the time cutoff above 10 seconds risks losing out on encounters that actually had substantial communication (decreasing sensitivity). Even if the communications can be classified as pure “social check-ins,” we believe these are important data to capture, as social check-ins are an important part of the patient’s care and experience. Third, several studies have commented on the modest accuracy of RFID technology as a locator system.^15,21 To address this, we both validated the accuracy of our RFID tags prior to the study and restricted our measurements to only inside patient rooms, which has less signal noise than hallways.

Future directions include expanding this study to include housestaff and physicians from other specialities, which may reveal different patterns and metrics of patient and nurse interactions.

RFID technology is a high-throughput method of generating precise, quantitative, and objective data on physician and nurse rounding habits. This tool can be widely applied to generate baseline rounding and overlap data for a variety of wards and settings, especially for institutions that are interested in comparing their metrics and performance to other peer wards or hospitals. Furthermore, this method can generate the necessary pre- and postintervention data for countless quality improvement endeavors, including efforts to enhance bedside interdisciplinary rounding.

Acknowledgments

The authors would like to thank the attending hospitalists who piloted wearing the RFID tags. This study would not be possible without your participation. The authors also wish to extend their appreciation to Gretchen Brown, MSN RN NEA-BC, for her support. Finally, the authors would like to thank Dr. Laurence Katznelson, Thi Dinh La, and the Resident Safety Council at Stanford, as well as the Stanford GME Office.

Disclosures

The authors have nothing to disclose.  

Effective communication between physicians and nurses is an essential element of any healthcare system. Numerous studies have highlighted the benefits of high quality physician–nurse (MD–RN) communication, including improved patient outcomes,¹ higher patient satisfaction,² and better nurse job satisfaction and retention rates.^3-5 Having physicians and nurses round together (bedside interdisciplinary rounding) has been shown to improve the perception of teamwork,^6,7 reduce the number of pages for the physician team,^6,8 better involve the patients in developing the plan of care,⁸ and even decrease the length and cost of stay.⁹

Being physically in the same space at the same time is the first and nonnegotiable requirement of bedside interdisciplinary rounding. However, precise and objective data regarding the extent to which physicians and nurses overlap at the patient bedside are lacking. Studies that examine the face-to-face component of MD–RN communication have generally relied on either qualitative methods, such as focus groups and surveys,^10,11 or quantitative methods that are subjective, such as validated scales.¹² In addition, the few studies that report quantitative data usually rely on manual observation methods that can be affected by various forms of observer bias.^10,13,14 There is also a paucity of data on how bedside overlap changes over the work week or as a function of room location.

Recently, real-time locator systems using radio frequency identification (RFID) have allowed measurement of staff and equipment movement in a precise and quantitative manner.^9,15 Although there have been previous studies using RFID locators to create time-motion maps of various hospital staff, no study has used RFID to measure and analyze the workflow of both physicians and nurses simultaneously.^16-18 The purpose of our investigation was to utilize our hospital-wide RFID staff locator technology to accurately and quantitatively assess physician and nurse rounding habits. Understanding the current rate of overlap is an important first step to establishing bedside interdisciplinary rounding.

Setting and Participants

The investigation was conducted at a single quaternary-care academic center. The study is exempt per our Institutional Review Board. Data were gathered from three adjacent medical-surgical acute care wards. The layout for each ward was the same: 19 single- or double-occupancy patient rooms arranged in a linear hallway, with a nursing station located at the center of the ward.

The study utilized wearable RFID tags (manufactured by Hill-Rom Holdings, Inc) that located specific staff within the hospital in real time. The RFID tags were checked at Hill-Rom graphical stations to ensure that their locations were tracked accurately. The investigators also wore them and walked around the wards in a prescripted manner to ensure validity. In addition, the locator accuracy was audited by participating attendings once per week and cross-checked with the generated data. Attending physicians on the University Hospitalist inpatient medicine teams were then given their uniquely-tagged RFIDs at the beginning of this study. Nurses already wear individual RFID tags as part of their normal standard-of-care workflow.

The attending hospitalists wore their RFID tags when they were on service for the entirety of the shift. They were encouraged to include nurses at the bedside, but this was not mandatory. The rounding team also included residents and medical students. Rounding usually begins at a prespecified time, but the route taken varies daily depending on patient location. Afternoon rounds were done as needed, depending on patient acuity. The attending physicians’ participation in this study was not disclosed to the patient. The patient care activities and daily routines of both nurses and physicians were otherwise unaltered.

Study Design and Data Collection

Data were collected on the three wards for 90 consecutive days, including nights and weekends. As physicians and nurses moved throughout the ward to conduct their usual patient care activities, the temporal-spatial data associated with their unique RFIDs were automatically collected in real time by the Hill-Rom receivers built into each patient room. Every day, a spreadsheet detailing the activity of all participating nurses and physicians for the past 24 hours was generated for the investigators.

A rounding event was defined as any episode in which a physician was in a patient room for more than 10 seconds. Incidences in which a physician entered and left a room multiple times over a short time span (with less than five minutes in between each event) were classified as a single rounding event. A physician and a nurse were defined as having overlapped if their RFID data showed that they were in the same patient room for a minimum of 10 seconds at the same time. For the purposes of this study, data generated from other RFID-wearing professionals, such as nursing assistants or unit secretaries, as well as data collected from the hallways, were excluded.

Statistical Analysis

All statistical analyses were conducted using GraphPad Prism (GraphPad Software, San Diego, California). Rounding and overlap lengths were rounded to the nearest minute (minimum one minute). Mean lengths are expressed along with the standard error. Comparisons of the average lengths of MD rounding events between wards was conducted using two-tailed Student t-test or one-way ANOVA. Comparisons of the frequency of MD–RN overlap between wards and across different days of the week were performed using a Chi-squared test. The analysis of correlation between the frequency of MD–RN overlap and distance between patient room and nursing station was conducted by calculating Pearson’s correlation. A P value of less than .05 was considered statistically significant.

Baseline Rounding Characteristics

Over the study period of 90 consecutive days, 739 MD rounding events were captured, for an average of 8.2 events per day. The mean length of all MD rounding events was 7.31 minutes (±0.27, ranging from one to 70 minutes). Of these 739 MD rounding events, we separately examined the 267 events that took place in single-bed patient rooms, to control for false-positive physician and nurse interactions (for example, if the MD and RN were caring for two separate roommates). The average rounding length of single-bed rooms was 6.93 (±0.27) minutes (Figure 1). For the three individual wards, the average rounding lengths were 6.40 ± 0.73, 7.48 ± 0.94, and 7.02 ± 0.54 minutes, respectively (no statistically significant difference).

Frequency of MD–RN Overlap

Of the 267 MD rounding events observed in single-bed rooms, a nurse was present in the room for 80 events (30.0%). The frequencies of MD–RN overlap in patient rooms were 37.0% (30/81), 28.0% (14/50), and 26.5% (36/136) for the three individual wards (P > .05), respectively.

The durations of MD–RN overlap, when these events did occur, were 3.43 ± 0.38, 3.00 ± 0.70, and 3.69 ± 0.92 minutes, respectively (P > .05). The overall mean length of MD–RN overlap for all single rooms was 3.48 ± 0.45 minutes.

Rounding Characteristics over the Course of the Week

To assess how rounding characteristics differed over the work week, we partitioned our data into the individual days of the week. The length of each MD rounding event (time spent in each patient room) did not vary significantly over the course of the week (Figure 2a). When the data for the individual days were aggregated into “weekdays” (Monday through Friday) and “weekends” (Saturday and Sunday), the mean lengths of MD rounds were 7.26 ± 0.32 minutes on weekdays and 7.47 ± 0.52 minutes on weekends (P > .05).

In addition, there was no difference in how frequently physicians and nurses overlapped at the patient bedside between weekdays and weekends. Of the 565 weekday MD rounding events, 238 had a nurse at bedside (42.1%), and of the 173 weekend MD rounding events, 73 had a nurse at bedside (42.2%; Figure 2b).

Effect of a Bedside Nurse on the Length of Rounds

Next, the data on the length of MD rounds were partitioned based on whether there was a bedside nurse present during rounds. The mean length of rounds with only MDs (without a bedside nurse) was 5.68 ± 0.24 minutes. By comparison, the mean length of rounds with both a nurse and a physician at the patient bedside was 9.56 ± 0.53 minutes (Figure 3). This difference was statistically significant (P < .001).

Association between Patient Room Location and the Likelihood of MD–RN Overlap

All three wards in this study have a linear layout, consisting of 19 patient rooms in a row (Figure 4a). The nursing station is located in a central position within each ward, across from the 10^th patient room. The frequency of MD–RN overlap was calculated for each room, and each room was ranked according to its relative distance from the nursing station. For each individual ward, there was no statistically significant trend in MD–RN overlap frequency as a function of the distance to the nursing station (data not shown). However, when the data from all three wards were aggregated, there was a statistically significant trend (P < .05) with a negative Pearson correlation (r = –0.670; Figure 4b). The slope of the best fit line was 1.94, suggesting that for each additional room farther away from the nursing station, the likelihood of interdisciplinary rounds (with both physicians and nurses together at the bedside) decreases by almost 2%.

To the best of our knowledge, this is the first time-motion study of MD–RN overlap using real-time, RFID-based location technology to capture the rounding activity of both nurses and physicians. Our primary interest was to examine the extent of MD–RN overlap at the patient bedside. This is an important metric that can pave the way for bedside interdisciplinary rounds. Although the exact nature of nurse-physician communication was not measured using the methodology in this study, understanding the length of time physicians spend in patient rooms, across different wards and throughout the work week, provides insights on the current workflow and potential areas of improvement. For example, we found that 30.0% of MD rounds overlapped with a nurse at the bedside. This baseline data highlight one potential barrier to institution-wide bedside interdisciplinary rounds. Workflow changes, such as better co-localization of patients by service lines or utilization of technologies to augment the visibility of rounding physicians, may improve this overlap frequency.

Data in the literature regarding how much interaction physicians and nurses have, especially at the bedside, are sparse and vary widely. In a recent study using medical students as observers by Stickrath et al., 807 MD rounding events led by medicine attendings were observed over 90 days. The frequency of rounding events that included “communication with nurse” was only 12%.¹⁹ Furthermore, only 64.9% of these communications were at the bedside, for an effective prevalence of bedside MD–RN communication of 7.8%. This number is low compared to our observed frequency of 30.0%. On the other extreme, a study from a hospital that intentionally institutes multidisciplinary rounding (explicitly defined as involving a physician and a nurse at a bedside) reported a frequency range of 63% to 81%.⁷ A follow-up study by the same group again demonstrated a high frequency of multidisciplinary rounds (74%) across a variety of ward and specialty types (range 35% to 97%.).¹¹ However, because of the selection bias of this particular setting, the high prevalence does not reflect a generalizable frequency of bedside MD–RN overlap at most hospitals.

The length of time spent by physicians at the patient bedside balances the competing demands of patient care and rapport-building with maintaining efficiency and progressing to other important tasks. In our study, physicians spent an average of 7.31 minutes at the bedside per patient. A previously published multiinstitutional observational study, which included our hospital, reported that the average length of rounds at bedside was 4.8 minutes.¹³ A second study reported that 8.0 minutes were spent at the bedside per patient.⁷ All three studies examined the same setting of internal medicine rounds at academic university-based hospitals, led by an attending physician with junior and senior residents present. However, the methodologies to measure the length of physician rounds were different: Priest et al. involved observers, Gonzalos et al. used E-mail-based surveys, and we utilized RFID-based locators. Additional institutional, individual, and patient-based factors also influence the length of rounds and are challenging to directly measure.

Furthermore, the discovery that the length of rounds and the frequency of MD–RN overlap did not statistically differ between weekdays and weekends (P > .05) was unexpected. Given the general trend of reduced physician staffing on weekends and the practice of cross-covering larger patient censuses, we would have expected shorter rounds and less frequent MD–RN overlap on the weekends.^7,20 The remarkable similarity between weekday and weekend metrics suggests that our workflow and rounding habits are not compromised on the weekends.

In addition, we found that MD rounds with a nurse at bedside took longer than rounds without a nurse, and that patient rooms located farther away from the central nursing station had a lower frequency of MD–RN overlap. However, we want to emphasize that these findings are merely associative, and not causal. For example, sicker patients usually take longer to round on than stable patients, and it is also the sicker patients who are more likely to have their nurses at the bedside, independent of physician rounding activity. Furthermore, even if rounding with nurses takes more time, it may ultimately result in fewer pages and overall time savings for both physicians and nurses.⁶

With regards to the association between room location and frequency of MD–RN overlap, the data can be interpreted in two ways. On the one hand, if the distance between the patient room and the nursing station does, in fact, reduce the frequency of overlap by almost 2% per room (Figure 4b), these data can be informative for future workflow development, quality improvement projects, or even hospital design. On the other hand, many wards might intentionally place more stable, less acute patients farther away from the nursing station because they do not need to be watched as closely. In that case, these data confirm their expectations and no action is needed.

There are several limitations to our study. The principal limitation, as discussed above, is that while our RFID system can generate large quantities of precise data on MD–RN overlap, we do not know the qualitative nature of the overlap. Just because a nurse and a physician are in the same room at the same time does not mean that they are communicating with each other. Second, we defined “rounding” as lasting a minimum of 10 seconds at the bedside. We believe that at least 10 seconds is needed to engage in any meaningful interaction between the physician and the patient, or the physician and the nurse. Reducing the time cutoff below 10 seconds risks capturing more “noise,” (decreasing specificity) whereas increasing the time cutoff above 10 seconds risks losing out on encounters that actually had substantial communication (decreasing sensitivity). Even if the communications can be classified as pure “social check-ins,” we believe these are important data to capture, as social check-ins are an important part of the patient’s care and experience. Third, several studies have commented on the modest accuracy of RFID technology as a locator system.^15,21 To address this, we both validated the accuracy of our RFID tags prior to the study and restricted our measurements to only inside patient rooms, which has less signal noise than hallways.

Future directions include expanding this study to include housestaff and physicians from other specialities, which may reveal different patterns and metrics of patient and nurse interactions.

RFID technology is a high-throughput method of generating precise, quantitative, and objective data on physician and nurse rounding habits. This tool can be widely applied to generate baseline rounding and overlap data for a variety of wards and settings, especially for institutions that are interested in comparing their metrics and performance to other peer wards or hospitals. Furthermore, this method can generate the necessary pre- and postintervention data for countless quality improvement endeavors, including efforts to enhance bedside interdisciplinary rounding.

Acknowledgments

The authors would like to thank the attending hospitalists who piloted wearing the RFID tags. This study would not be possible without your participation. The authors also wish to extend their appreciation to Gretchen Brown, MSN RN NEA-BC, for her support. Finally, the authors would like to thank Dr. Laurence Katznelson, Thi Dinh La, and the Resident Safety Council at Stanford, as well as the Stanford GME Office.

Disclosures

The authors have nothing to disclose.  

Effective communication between physicians and nurses is an essential element of any healthcare system. Numerous studies have highlighted the benefits of high quality physician–nurse (MD–RN) communication, including improved patient outcomes,¹ higher patient satisfaction,² and better nurse job satisfaction and retention rates.^3-5 Having physicians and nurses round together (bedside interdisciplinary rounding) has been shown to improve the perception of teamwork,^6,7 reduce the number of pages for the physician team,^6,8 better involve the patients in developing the plan of care,⁸ and even decrease the length and cost of stay.⁹

Being physically in the same space at the same time is the first and nonnegotiable requirement of bedside interdisciplinary rounding. However, precise and objective data regarding the extent to which physicians and nurses overlap at the patient bedside are lacking. Studies that examine the face-to-face component of MD–RN communication have generally relied on either qualitative methods, such as focus groups and surveys,^10,11 or quantitative methods that are subjective, such as validated scales.¹² In addition, the few studies that report quantitative data usually rely on manual observation methods that can be affected by various forms of observer bias.^10,13,14 There is also a paucity of data on how bedside overlap changes over the work week or as a function of room location.

Recently, real-time locator systems using radio frequency identification (RFID) have allowed measurement of staff and equipment movement in a precise and quantitative manner.^9,15 Although there have been previous studies using RFID locators to create time-motion maps of various hospital staff, no study has used RFID to measure and analyze the workflow of both physicians and nurses simultaneously.^16-18 The purpose of our investigation was to utilize our hospital-wide RFID staff locator technology to accurately and quantitatively assess physician and nurse rounding habits. Understanding the current rate of overlap is an important first step to establishing bedside interdisciplinary rounding.

Setting and Participants

The investigation was conducted at a single quaternary-care academic center. The study is exempt per our Institutional Review Board. Data were gathered from three adjacent medical-surgical acute care wards. The layout for each ward was the same: 19 single- or double-occupancy patient rooms arranged in a linear hallway, with a nursing station located at the center of the ward.

The study utilized wearable RFID tags (manufactured by Hill-Rom Holdings, Inc) that located specific staff within the hospital in real time. The RFID tags were checked at Hill-Rom graphical stations to ensure that their locations were tracked accurately. The investigators also wore them and walked around the wards in a prescripted manner to ensure validity. In addition, the locator accuracy was audited by participating attendings once per week and cross-checked with the generated data. Attending physicians on the University Hospitalist inpatient medicine teams were then given their uniquely-tagged RFIDs at the beginning of this study. Nurses already wear individual RFID tags as part of their normal standard-of-care workflow.

The attending hospitalists wore their RFID tags when they were on service for the entirety of the shift. They were encouraged to include nurses at the bedside, but this was not mandatory. The rounding team also included residents and medical students. Rounding usually begins at a prespecified time, but the route taken varies daily depending on patient location. Afternoon rounds were done as needed, depending on patient acuity. The attending physicians’ participation in this study was not disclosed to the patient. The patient care activities and daily routines of both nurses and physicians were otherwise unaltered.

Study Design and Data Collection

Data were collected on the three wards for 90 consecutive days, including nights and weekends. As physicians and nurses moved throughout the ward to conduct their usual patient care activities, the temporal-spatial data associated with their unique RFIDs were automatically collected in real time by the Hill-Rom receivers built into each patient room. Every day, a spreadsheet detailing the activity of all participating nurses and physicians for the past 24 hours was generated for the investigators.

A rounding event was defined as any episode in which a physician was in a patient room for more than 10 seconds. Incidences in which a physician entered and left a room multiple times over a short time span (with less than five minutes in between each event) were classified as a single rounding event. A physician and a nurse were defined as having overlapped if their RFID data showed that they were in the same patient room for a minimum of 10 seconds at the same time. For the purposes of this study, data generated from other RFID-wearing professionals, such as nursing assistants or unit secretaries, as well as data collected from the hallways, were excluded.

Statistical Analysis

All statistical analyses were conducted using GraphPad Prism (GraphPad Software, San Diego, California). Rounding and overlap lengths were rounded to the nearest minute (minimum one minute). Mean lengths are expressed along with the standard error. Comparisons of the average lengths of MD rounding events between wards was conducted using two-tailed Student t-test or one-way ANOVA. Comparisons of the frequency of MD–RN overlap between wards and across different days of the week were performed using a Chi-squared test. The analysis of correlation between the frequency of MD–RN overlap and distance between patient room and nursing station was conducted by calculating Pearson’s correlation. A P value of less than .05 was considered statistically significant.

Baseline Rounding Characteristics

Over the study period of 90 consecutive days, 739 MD rounding events were captured, for an average of 8.2 events per day. The mean length of all MD rounding events was 7.31 minutes (±0.27, ranging from one to 70 minutes). Of these 739 MD rounding events, we separately examined the 267 events that took place in single-bed patient rooms, to control for false-positive physician and nurse interactions (for example, if the MD and RN were caring for two separate roommates). The average rounding length of single-bed rooms was 6.93 (±0.27) minutes (Figure 1). For the three individual wards, the average rounding lengths were 6.40 ± 0.73, 7.48 ± 0.94, and 7.02 ± 0.54 minutes, respectively (no statistically significant difference).

Frequency of MD–RN Overlap

Of the 267 MD rounding events observed in single-bed rooms, a nurse was present in the room for 80 events (30.0%). The frequencies of MD–RN overlap in patient rooms were 37.0% (30/81), 28.0% (14/50), and 26.5% (36/136) for the three individual wards (P > .05), respectively.

The durations of MD–RN overlap, when these events did occur, were 3.43 ± 0.38, 3.00 ± 0.70, and 3.69 ± 0.92 minutes, respectively (P > .05). The overall mean length of MD–RN overlap for all single rooms was 3.48 ± 0.45 minutes.

Rounding Characteristics over the Course of the Week

To assess how rounding characteristics differed over the work week, we partitioned our data into the individual days of the week. The length of each MD rounding event (time spent in each patient room) did not vary significantly over the course of the week (Figure 2a). When the data for the individual days were aggregated into “weekdays” (Monday through Friday) and “weekends” (Saturday and Sunday), the mean lengths of MD rounds were 7.26 ± 0.32 minutes on weekdays and 7.47 ± 0.52 minutes on weekends (P > .05).

In addition, there was no difference in how frequently physicians and nurses overlapped at the patient bedside between weekdays and weekends. Of the 565 weekday MD rounding events, 238 had a nurse at bedside (42.1%), and of the 173 weekend MD rounding events, 73 had a nurse at bedside (42.2%; Figure 2b).

Effect of a Bedside Nurse on the Length of Rounds

Next, the data on the length of MD rounds were partitioned based on whether there was a bedside nurse present during rounds. The mean length of rounds with only MDs (without a bedside nurse) was 5.68 ± 0.24 minutes. By comparison, the mean length of rounds with both a nurse and a physician at the patient bedside was 9.56 ± 0.53 minutes (Figure 3). This difference was statistically significant (P < .001).

Association between Patient Room Location and the Likelihood of MD–RN Overlap

All three wards in this study have a linear layout, consisting of 19 patient rooms in a row (Figure 4a). The nursing station is located in a central position within each ward, across from the 10^th patient room. The frequency of MD–RN overlap was calculated for each room, and each room was ranked according to its relative distance from the nursing station. For each individual ward, there was no statistically significant trend in MD–RN overlap frequency as a function of the distance to the nursing station (data not shown). However, when the data from all three wards were aggregated, there was a statistically significant trend (P < .05) with a negative Pearson correlation (r = –0.670; Figure 4b). The slope of the best fit line was 1.94, suggesting that for each additional room farther away from the nursing station, the likelihood of interdisciplinary rounds (with both physicians and nurses together at the bedside) decreases by almost 2%.

To the best of our knowledge, this is the first time-motion study of MD–RN overlap using real-time, RFID-based location technology to capture the rounding activity of both nurses and physicians. Our primary interest was to examine the extent of MD–RN overlap at the patient bedside. This is an important metric that can pave the way for bedside interdisciplinary rounds. Although the exact nature of nurse-physician communication was not measured using the methodology in this study, understanding the length of time physicians spend in patient rooms, across different wards and throughout the work week, provides insights on the current workflow and potential areas of improvement. For example, we found that 30.0% of MD rounds overlapped with a nurse at the bedside. This baseline data highlight one potential barrier to institution-wide bedside interdisciplinary rounds. Workflow changes, such as better co-localization of patients by service lines or utilization of technologies to augment the visibility of rounding physicians, may improve this overlap frequency.

Data in the literature regarding how much interaction physicians and nurses have, especially at the bedside, are sparse and vary widely. In a recent study using medical students as observers by Stickrath et al., 807 MD rounding events led by medicine attendings were observed over 90 days. The frequency of rounding events that included “communication with nurse” was only 12%.¹⁹ Furthermore, only 64.9% of these communications were at the bedside, for an effective prevalence of bedside MD–RN communication of 7.8%. This number is low compared to our observed frequency of 30.0%. On the other extreme, a study from a hospital that intentionally institutes multidisciplinary rounding (explicitly defined as involving a physician and a nurse at a bedside) reported a frequency range of 63% to 81%.⁷ A follow-up study by the same group again demonstrated a high frequency of multidisciplinary rounds (74%) across a variety of ward and specialty types (range 35% to 97%.).¹¹ However, because of the selection bias of this particular setting, the high prevalence does not reflect a generalizable frequency of bedside MD–RN overlap at most hospitals.

The length of time spent by physicians at the patient bedside balances the competing demands of patient care and rapport-building with maintaining efficiency and progressing to other important tasks. In our study, physicians spent an average of 7.31 minutes at the bedside per patient. A previously published multiinstitutional observational study, which included our hospital, reported that the average length of rounds at bedside was 4.8 minutes.¹³ A second study reported that 8.0 minutes were spent at the bedside per patient.⁷ All three studies examined the same setting of internal medicine rounds at academic university-based hospitals, led by an attending physician with junior and senior residents present. However, the methodologies to measure the length of physician rounds were different: Priest et al. involved observers, Gonzalos et al. used E-mail-based surveys, and we utilized RFID-based locators. Additional institutional, individual, and patient-based factors also influence the length of rounds and are challenging to directly measure.

Furthermore, the discovery that the length of rounds and the frequency of MD–RN overlap did not statistically differ between weekdays and weekends (P > .05) was unexpected. Given the general trend of reduced physician staffing on weekends and the practice of cross-covering larger patient censuses, we would have expected shorter rounds and less frequent MD–RN overlap on the weekends.^7,20 The remarkable similarity between weekday and weekend metrics suggests that our workflow and rounding habits are not compromised on the weekends.

In addition, we found that MD rounds with a nurse at bedside took longer than rounds without a nurse, and that patient rooms located farther away from the central nursing station had a lower frequency of MD–RN overlap. However, we want to emphasize that these findings are merely associative, and not causal. For example, sicker patients usually take longer to round on than stable patients, and it is also the sicker patients who are more likely to have their nurses at the bedside, independent of physician rounding activity. Furthermore, even if rounding with nurses takes more time, it may ultimately result in fewer pages and overall time savings for both physicians and nurses.⁶

With regards to the association between room location and frequency of MD–RN overlap, the data can be interpreted in two ways. On the one hand, if the distance between the patient room and the nursing station does, in fact, reduce the frequency of overlap by almost 2% per room (Figure 4b), these data can be informative for future workflow development, quality improvement projects, or even hospital design. On the other hand, many wards might intentionally place more stable, less acute patients farther away from the nursing station because they do not need to be watched as closely. In that case, these data confirm their expectations and no action is needed.

There are several limitations to our study. The principal limitation, as discussed above, is that while our RFID system can generate large quantities of precise data on MD–RN overlap, we do not know the qualitative nature of the overlap. Just because a nurse and a physician are in the same room at the same time does not mean that they are communicating with each other. Second, we defined “rounding” as lasting a minimum of 10 seconds at the bedside. We believe that at least 10 seconds is needed to engage in any meaningful interaction between the physician and the patient, or the physician and the nurse. Reducing the time cutoff below 10 seconds risks capturing more “noise,” (decreasing specificity) whereas increasing the time cutoff above 10 seconds risks losing out on encounters that actually had substantial communication (decreasing sensitivity). Even if the communications can be classified as pure “social check-ins,” we believe these are important data to capture, as social check-ins are an important part of the patient’s care and experience. Third, several studies have commented on the modest accuracy of RFID technology as a locator system.^15,21 To address this, we both validated the accuracy of our RFID tags prior to the study and restricted our measurements to only inside patient rooms, which has less signal noise than hallways.

Future directions include expanding this study to include housestaff and physicians from other specialities, which may reveal different patterns and metrics of patient and nurse interactions.

RFID technology is a high-throughput method of generating precise, quantitative, and objective data on physician and nurse rounding habits. This tool can be widely applied to generate baseline rounding and overlap data for a variety of wards and settings, especially for institutions that are interested in comparing their metrics and performance to other peer wards or hospitals. Furthermore, this method can generate the necessary pre- and postintervention data for countless quality improvement endeavors, including efforts to enhance bedside interdisciplinary rounding.

Acknowledgments

The authors would like to thank the attending hospitalists who piloted wearing the RFID tags. This study would not be possible without your participation. The authors also wish to extend their appreciation to Gretchen Brown, MSN RN NEA-BC, for her support. Finally, the authors would like to thank Dr. Laurence Katznelson, Thi Dinh La, and the Resident Safety Council at Stanford, as well as the Stanford GME Office.

Disclosures

The authors have nothing to disclose.