Natalie Lee, MD, MPH

Individuals aged 65 years and over represent 13% of the United States population and account for nearly 40% of hospital discharges.¹ Bedrest hastens the functional decline of older patients^2-5 and is associated with risk of serious complications, such as falls, delirium, venous thrombosis, and skin breakdown.^6,7 Ambulation is widely recognized as important for improving hospital outcomes.^8-10 Observational studies suggest that increases of 600 steps per day are associated with shortened length of hospital stay.⁹ However, randomized trials of assisted ambulation have not demonstrated consistent benefit.^11-14 As a result, usual care at most hospitals in the United States does not include assisted ambulation. Even when ambulation is ordered, execution of the orders is inconsistent.^15-17

Studies have demonstrated the benefits of various exercise protocols for older patients in rehabilitation facilities,^18,19 medical intensive care units,²⁰ and medical and surgical wards.^13,18,21 These interventions are usually nursing centered; however, assisting patients with ambulation multiple times per day may be a burdensome addition to the myriad responsibilities of nurses.^19,22,23 In fact, ambulation orders are the most frequently overlooked nursing task.²⁴

We designed a graded protocol of assisted ambulation implemented by a dedicated patient care nursing assistant (PCNA) multiple times daily to increase patient mobility. The objective of this study was to assess the feasibility and effectiveness of such an intervention for older inpatients. We hypothesized that the intervention would prove feasible and improve hospital outcomes, including less need for inpatient rehabilitation and shorter length of stay.

We conducted a single-blind randomized controlled trial of patients aged ≥60 years and admitted as medical inpatients to the Cleveland Clinic Main Campus, a tertiary care center with over 1,440 inpatient beds. The consent form and study protocol were approved by the Cleveland Clinic Institutional Review Board, and the study was registered with ClinicalTrials.gov (NCT02757131).

Patients

All patients who were admitted to study wards for a medical illness and evaluated by Physical Therapy (PT) were eligible for the study. PT evaluations were ordered by the medical team if deemed necessary on the basis of factors, such as age, estimated mobility, and concerns raised by the ancillary staff. All patients who were expected to be discharged to a skilled nursing facility placement or who required home PT received a PT evaluation. Assessment of mobility was documented via Activity Measure for Postacute Care Inpatient Basic Mobility “six-clicks” short form, hereafter abbreviated as “six-clicks.” Based on past experience, patients with scores <16 rarely go home (<20% of the time), and those with scores >20 usually go home regardless of ambulation. Therefore, only patients with scores of 16-20 were invited to participate in the study. Although patients who were not evaluated by PT might also benefit from the intervention, we required a six-clicks score to assess eligibility. The exclusion criteria included anticipated remaining length of stay less than three days, admission under observation status, admission to the intensive care unit (ICU,) patients receiving comfort care measures, and patients with medical conditions precluding ambulation, such as decompensated heart failure or unstable angina.

Randomization

Patients were randomized to “usual care” or “mobility technician” after baseline evaluation using a computerized system. A block randomization scheme with a size of four was used to ensure an approximately equal number of patients per group.

Intervention

Patients randomized to the intervention group were asked to participate in the ambulation protocol outlined by the PT three times daily under the supervision of the mobility technician. The protocol involved four exercise levels (sitting, standing, walking, and stairs), which were implemented depending on the patient’s physical capacity. The mobility technicians, who were PCNAs, were trained by the PT team. PCNAs have no specific degrees or certification. They are taught safe handling techniques during their job orientation, so they already had an understanding of how to transfer and assist a patient with ambulation. The mobility technician training consisted of one four-hour session run by the PT team in the physical therapy department and the nursing unit. The training included safe handling practices and basic mobility, such as transfers from bed to chair, bed to standing, walking with assistance, and walking independently with equipment such as cane, rolling walker, and walking belt. All instruction was demonstrated by the trainer, and the mobility technician was then able to practice. The mobility technician then shadowed the trainer and practiced the techniques under supervision. Competency was assessed by the trainer.

The cohort of patients randomized to “usual care” was not seen by the mobility technicians but was not otherwise restricted in nursing’s baseline ability to execute recommendations placed by the PT team. Compliance with the recommendations is highly variable and dependent on patient acuity during the shift, staffing issues, and competing duties. Cleveland Clinic promotes a “culture of mobility,” and nurses are encouraged to get patients out of bed and assist with ambulation.

Study Instruments—Measures of Mobility

The six-clicks instrument is a tool for measuring basic mobility. It was adapted from the Activity Measures for Post-Acute Care (AM-PAC) instrument.²⁵ Although initially created for self-report in the post-acute care setting, six-clicks has been validated for use by PTs in the acute care setting²⁶ and is currently in use at more than 1,000 US hospitals. Cleveland Clinic PTs have used this measure for routine evaluation since 2011. The instrument has high interrater reliability and can predict discharge disposition.^27-29

Each patient was provided with a tracking device (Fitbit) attached at the wrist to record daily steps for measuring mobility. The use of Fitbit has been validated in ambulatory and inpatient settings.³⁰ The device produces step counts within 3% of the observed step count for most patients but may undercount steps in patients with very slow gait.³¹ The device was provided to each enrollee and collected at discharge.

Variables

Demographic information, comorbid diagnoses, and discharge destination were extracted from the electronic medical record. Information on prehospitalization physical activity level was obtained from the initial PT assessment. Falls were tracked through the safety event reporting system.

Outcomes

The primary outcomes were discharge disposition and hospital length of stay. The secondary outcomes included average steps per day, change in six-clicks score from admission to discharge, inpatient mortality, admission to ICU, falls, deep vein thrombosis, pulmonary embolism, or pneumonia, and readmission within 30 days.

Statistical Analysis

Patient characteristics were summarized as means and standard deviations or medians and interquartile ranges for continuous variables and as frequencies and percentages for categorical variables. The t-test or Wilcoxon rank sum test was applied to compare continuous characteristics between the intervention and control groups. Chi-squared test or Fisher’s exact test was applied to compare categorical characteristics. Given its skewed distribution, the length of stay was log-transformed and compared between the two groups using Student’s t-test. Chi-squared test was used to compare categorical outcomes. The analysis of final six-clicks scores was adjusted for baseline scores, and the least-square estimates are provided. A linear mixed effects model was used to compare the number of daily steps taken because each participant had multiple steps measured. Results were adjusted for prehospital activity. In addition to comparing the total steps taken by each group, we determined the proportion of patients who exceeded a particular threshold by taking the average number of steps per day for all subjects and relating it to home discharge using the Receiver Operating Characteristics (ROC) curve. An optimal cut-off was determined to maximize the Youden index. We also compared the proportion of patients who exceeded 900 steps because this value was previously reported as an important threshold.³² All analyses were conducted using intention-to-treat principles. We also conducted a per-protocol analysis in which we limited the intervention group to those who received at least one assisted ambulation session. A dose-response analysis was also performed, in which patients were categorized as not receiving the therapy, receiving sessions on one or two days, or receiving them on more than two days.

All analyses were conducted using R-studio (Boston, MA). Statistical significance was defined as a P-value < .05. Given that this is a pilot study, the results were not adjusted for multiple comparisons.

Characteristics of patients in the intervention and control groups are shown in Table 1. The patients were mostly white and female, with an average age in the mid-70s (range 61-98). All measures evaluated were not significantly different between the intervention and control groups. However, more patients in the intervention group had a prehospital activity level classified as independent.

Table 2 demonstrates the feasibility of the intervention. Of patients randomized to the intervention group, 74% were ambulated at least once. Once enrolled, the patients successfully participated in assisted ambulation for about two-thirds of their hospital stay. However, the intervention was delivered for only one-third of the total length of stay because most patients were not enrolled on admission. On average, the mobility technicians made 11 attempts to ambulate each patient and 56% of these attempts were successful. The proportion of unsuccessful attempts did not change over the course of the study. The reasons for unsuccessful attempts included patient refusal (n = 102) or unavailability (n = 68), mobility technicians running out of time (n = 2), and other (n = 12).

In this pilot study conducted with older medical inpatients, we found that assisted ambulation provided by a dedicated mobility technician was feasible and increased the number of steps taken by patients. Not all patients in the treatment group received the intervention partly due to the fact that the program initially did not include weekends and the mobility technicians were sometimes redirected to other nursing duties. Both issues were addressed during the course of the study. In the per protocol analysis, the intervention increased the average six-clicks score and there was a nonsignificant reduction in the percentage of patients discharged to a rehabilitation facility.

A range of hospital-based mobility interventions have been described. Several of which were complex multidisciplinary interventions that included a mobility component. The compliance rates have ranged from 82% to 93.7%,12,13 although a systematic review noted that many studies do not provide this level of information.¹¹ Interventions were carried out by nursing staff and PT with support from family members and social workers.^33-35 Ambulation-specific programs have also relied on nurses and PT13,14,36 and, occasionally, on research assistants to implement assisted ambulation protocols.^12,37 A recent study that employed research assistants to deliver inhospital ambulation reported achieving 51.3% of intended walks.³⁷

In contradistinction to previous studies, we created a new role, employing PCNAs as dedicated mobility technicians. We did this for two reasons. First, the approach is less expensive than deploying registered nurses or PTs to ambulate patients and therefore more likely to be adopted by hospitals, especially if it can decrease the cost of an episode of care by avoiding subsequent inpatient rehabilitation. Mobility technicians have no degree or certification requirements and are therefore paid less than nurses or physical therapists. Second, by having a single responsibility, mobility technicians were more likely to engage in their task than nurses, who have competing responsibilities. However, when nurse staffing was short, nurse managers were tempted to recall the PCNAs for other nursing duties. It took time before PCNAs and supervisors prioritized this new responsibility. When they did, the number of attempted walks increased substantially, but the percentage of successful attempts remained constant at 56%, highlighting the difficulty of getting hospitalized patients to walk.

On average, patients who received the intervention engaged in 72 minutes of additional physical activity and averaged 990 steps per day. Observational data suggest patients accrue about 1,100 steps in the day before discharge, with older patients accruing closer to 900.21 One study found that older patients with fewer than 900 steps per day were likely to experience a functional decline.32 We also found that patients who achieved at least 900 steps were more likely to go home. However, we found that a lower threshold, namely, 400 steps, offered better discrimination between patients who go home and those who do not. Future prospective studies are needed to establish the appropriate goal for exercise interventions. A lower step goal could dramatically enhance the efficiency of the intervention.

A Cochrane review found that pooled analysis of multidisciplinary interventions that included exercise, often in the form of walking, achieved a small but significant increase in the proportion of patients discharged to home (RR 1.08, 95%CI 1.03 to 1.14).¹¹ We found no significant change in the discharge disposition, but our study was underpowered for this endpoint. The six-clicks score showed a small but significant change in the per protocol analysis. The six-clicks score has been shown to correlate with discharge disposition,^28,29 and an improvement in the score suggests that discharge disposition may be influenced.

The intervention may also not have been implemented for long enough. On average, visits were achieved for one-third of the hospital stay partly because of the delay in PT evaluation, which we required for eligibility. In practice, PT evaluation can occur just a few days before the anticipated discharge. We observed a dose dependent response among patients in the intervention group, suggesting that earlier intervention could be more effective. Earlier intervention might be achieved if the MT performed the six-clicks on potentially eligible patients.

The effects of hospitalization on mobility may be the most pronounced in the long term; one study found that 40% of hospitalized older patients manifested new ADL or IADL disability three months after discharge compared with 31% at discharge.⁷ Hospital-based mobility interventions may continue to affect subjects’ independence for weeks or months. In one RCT, an inpatient ambulation intervention improved mobility in the community one month after discharge.³⁷ A hospital-based exercise program that included ambulation achieved better functional outcomes one month later.¹³ One RCT that combined inpatient exercise with outpatient care coordination also decreased readmission rates.³⁴ We found that the intervention did not affect readmission.

This pilot study has several limitations. The sample size was small, and the findings need to be replicated in a larger randomized controlled trial. This is particularly important because the two study arms were not balanced in terms of their prehospital activity. After adjustment for prehospital activity, the differences in the step count in the intention-to-treat analysis were no longer significant. As we adjusted the intervention to hospital workflow, the intervention changed over time. The intention-to-treat analysis may therefore underestimate the effect of the intervention. This work provides a basis for future trial. Finally, discharge disposition depends on a complex interplay of factors, including social factors and preferences, which may not be affected by a mobility intervention.

In summary, an inhospital mobility protocol of attempting ambulation delivered by dedicated mobility technicians three times daily successfully increased the daily step counts and mobility scores of the patients. Studies with a larger sample size are needed to determine whether the proposed approach can affect length of hospital stay, discharge disposition, and overall cost of an episode of care.

Disclosures

Mary Stilphen reports consulting for CreCare and Adeo, which license and distribute AM-PAC short forms, including 6 clicks. All other authors report no conflicts of interest.

Funding

This study was supported by a Research Program Committee grant from the Cleveland Clinic.

Individuals aged 65 years and over represent 13% of the United States population and account for nearly 40% of hospital discharges.¹ Bedrest hastens the functional decline of older patients^2-5 and is associated with risk of serious complications, such as falls, delirium, venous thrombosis, and skin breakdown.^6,7 Ambulation is widely recognized as important for improving hospital outcomes.^8-10 Observational studies suggest that increases of 600 steps per day are associated with shortened length of hospital stay.⁹ However, randomized trials of assisted ambulation have not demonstrated consistent benefit.^11-14 As a result, usual care at most hospitals in the United States does not include assisted ambulation. Even when ambulation is ordered, execution of the orders is inconsistent.^15-17

Studies have demonstrated the benefits of various exercise protocols for older patients in rehabilitation facilities,^18,19 medical intensive care units,²⁰ and medical and surgical wards.^13,18,21 These interventions are usually nursing centered; however, assisting patients with ambulation multiple times per day may be a burdensome addition to the myriad responsibilities of nurses.^19,22,23 In fact, ambulation orders are the most frequently overlooked nursing task.²⁴

We designed a graded protocol of assisted ambulation implemented by a dedicated patient care nursing assistant (PCNA) multiple times daily to increase patient mobility. The objective of this study was to assess the feasibility and effectiveness of such an intervention for older inpatients. We hypothesized that the intervention would prove feasible and improve hospital outcomes, including less need for inpatient rehabilitation and shorter length of stay.

We conducted a single-blind randomized controlled trial of patients aged ≥60 years and admitted as medical inpatients to the Cleveland Clinic Main Campus, a tertiary care center with over 1,440 inpatient beds. The consent form and study protocol were approved by the Cleveland Clinic Institutional Review Board, and the study was registered with ClinicalTrials.gov (NCT02757131).

Patients

All patients who were admitted to study wards for a medical illness and evaluated by Physical Therapy (PT) were eligible for the study. PT evaluations were ordered by the medical team if deemed necessary on the basis of factors, such as age, estimated mobility, and concerns raised by the ancillary staff. All patients who were expected to be discharged to a skilled nursing facility placement or who required home PT received a PT evaluation. Assessment of mobility was documented via Activity Measure for Postacute Care Inpatient Basic Mobility “six-clicks” short form, hereafter abbreviated as “six-clicks.” Based on past experience, patients with scores <16 rarely go home (<20% of the time), and those with scores >20 usually go home regardless of ambulation. Therefore, only patients with scores of 16-20 were invited to participate in the study. Although patients who were not evaluated by PT might also benefit from the intervention, we required a six-clicks score to assess eligibility. The exclusion criteria included anticipated remaining length of stay less than three days, admission under observation status, admission to the intensive care unit (ICU,) patients receiving comfort care measures, and patients with medical conditions precluding ambulation, such as decompensated heart failure or unstable angina.

Randomization

Patients were randomized to “usual care” or “mobility technician” after baseline evaluation using a computerized system. A block randomization scheme with a size of four was used to ensure an approximately equal number of patients per group.

Intervention

Patients randomized to the intervention group were asked to participate in the ambulation protocol outlined by the PT three times daily under the supervision of the mobility technician. The protocol involved four exercise levels (sitting, standing, walking, and stairs), which were implemented depending on the patient’s physical capacity. The mobility technicians, who were PCNAs, were trained by the PT team. PCNAs have no specific degrees or certification. They are taught safe handling techniques during their job orientation, so they already had an understanding of how to transfer and assist a patient with ambulation. The mobility technician training consisted of one four-hour session run by the PT team in the physical therapy department and the nursing unit. The training included safe handling practices and basic mobility, such as transfers from bed to chair, bed to standing, walking with assistance, and walking independently with equipment such as cane, rolling walker, and walking belt. All instruction was demonstrated by the trainer, and the mobility technician was then able to practice. The mobility technician then shadowed the trainer and practiced the techniques under supervision. Competency was assessed by the trainer.

The cohort of patients randomized to “usual care” was not seen by the mobility technicians but was not otherwise restricted in nursing’s baseline ability to execute recommendations placed by the PT team. Compliance with the recommendations is highly variable and dependent on patient acuity during the shift, staffing issues, and competing duties. Cleveland Clinic promotes a “culture of mobility,” and nurses are encouraged to get patients out of bed and assist with ambulation.

Study Instruments—Measures of Mobility

The six-clicks instrument is a tool for measuring basic mobility. It was adapted from the Activity Measures for Post-Acute Care (AM-PAC) instrument.²⁵ Although initially created for self-report in the post-acute care setting, six-clicks has been validated for use by PTs in the acute care setting²⁶ and is currently in use at more than 1,000 US hospitals. Cleveland Clinic PTs have used this measure for routine evaluation since 2011. The instrument has high interrater reliability and can predict discharge disposition.^27-29

Each patient was provided with a tracking device (Fitbit) attached at the wrist to record daily steps for measuring mobility. The use of Fitbit has been validated in ambulatory and inpatient settings.³⁰ The device produces step counts within 3% of the observed step count for most patients but may undercount steps in patients with very slow gait.³¹ The device was provided to each enrollee and collected at discharge.

Variables

Demographic information, comorbid diagnoses, and discharge destination were extracted from the electronic medical record. Information on prehospitalization physical activity level was obtained from the initial PT assessment. Falls were tracked through the safety event reporting system.

Outcomes

The primary outcomes were discharge disposition and hospital length of stay. The secondary outcomes included average steps per day, change in six-clicks score from admission to discharge, inpatient mortality, admission to ICU, falls, deep vein thrombosis, pulmonary embolism, or pneumonia, and readmission within 30 days.

Statistical Analysis

Patient characteristics were summarized as means and standard deviations or medians and interquartile ranges for continuous variables and as frequencies and percentages for categorical variables. The t-test or Wilcoxon rank sum test was applied to compare continuous characteristics between the intervention and control groups. Chi-squared test or Fisher’s exact test was applied to compare categorical characteristics. Given its skewed distribution, the length of stay was log-transformed and compared between the two groups using Student’s t-test. Chi-squared test was used to compare categorical outcomes. The analysis of final six-clicks scores was adjusted for baseline scores, and the least-square estimates are provided. A linear mixed effects model was used to compare the number of daily steps taken because each participant had multiple steps measured. Results were adjusted for prehospital activity. In addition to comparing the total steps taken by each group, we determined the proportion of patients who exceeded a particular threshold by taking the average number of steps per day for all subjects and relating it to home discharge using the Receiver Operating Characteristics (ROC) curve. An optimal cut-off was determined to maximize the Youden index. We also compared the proportion of patients who exceeded 900 steps because this value was previously reported as an important threshold.³² All analyses were conducted using intention-to-treat principles. We also conducted a per-protocol analysis in which we limited the intervention group to those who received at least one assisted ambulation session. A dose-response analysis was also performed, in which patients were categorized as not receiving the therapy, receiving sessions on one or two days, or receiving them on more than two days.

All analyses were conducted using R-studio (Boston, MA). Statistical significance was defined as a P-value < .05. Given that this is a pilot study, the results were not adjusted for multiple comparisons.

Characteristics of patients in the intervention and control groups are shown in Table 1. The patients were mostly white and female, with an average age in the mid-70s (range 61-98). All measures evaluated were not significantly different between the intervention and control groups. However, more patients in the intervention group had a prehospital activity level classified as independent.

Table 2 demonstrates the feasibility of the intervention. Of patients randomized to the intervention group, 74% were ambulated at least once. Once enrolled, the patients successfully participated in assisted ambulation for about two-thirds of their hospital stay. However, the intervention was delivered for only one-third of the total length of stay because most patients were not enrolled on admission. On average, the mobility technicians made 11 attempts to ambulate each patient and 56% of these attempts were successful. The proportion of unsuccessful attempts did not change over the course of the study. The reasons for unsuccessful attempts included patient refusal (n = 102) or unavailability (n = 68), mobility technicians running out of time (n = 2), and other (n = 12).

In this pilot study conducted with older medical inpatients, we found that assisted ambulation provided by a dedicated mobility technician was feasible and increased the number of steps taken by patients. Not all patients in the treatment group received the intervention partly due to the fact that the program initially did not include weekends and the mobility technicians were sometimes redirected to other nursing duties. Both issues were addressed during the course of the study. In the per protocol analysis, the intervention increased the average six-clicks score and there was a nonsignificant reduction in the percentage of patients discharged to a rehabilitation facility.

A range of hospital-based mobility interventions have been described. Several of which were complex multidisciplinary interventions that included a mobility component. The compliance rates have ranged from 82% to 93.7%,12,13 although a systematic review noted that many studies do not provide this level of information.¹¹ Interventions were carried out by nursing staff and PT with support from family members and social workers.^33-35 Ambulation-specific programs have also relied on nurses and PT13,14,36 and, occasionally, on research assistants to implement assisted ambulation protocols.^12,37 A recent study that employed research assistants to deliver inhospital ambulation reported achieving 51.3% of intended walks.³⁷

In contradistinction to previous studies, we created a new role, employing PCNAs as dedicated mobility technicians. We did this for two reasons. First, the approach is less expensive than deploying registered nurses or PTs to ambulate patients and therefore more likely to be adopted by hospitals, especially if it can decrease the cost of an episode of care by avoiding subsequent inpatient rehabilitation. Mobility technicians have no degree or certification requirements and are therefore paid less than nurses or physical therapists. Second, by having a single responsibility, mobility technicians were more likely to engage in their task than nurses, who have competing responsibilities. However, when nurse staffing was short, nurse managers were tempted to recall the PCNAs for other nursing duties. It took time before PCNAs and supervisors prioritized this new responsibility. When they did, the number of attempted walks increased substantially, but the percentage of successful attempts remained constant at 56%, highlighting the difficulty of getting hospitalized patients to walk.

On average, patients who received the intervention engaged in 72 minutes of additional physical activity and averaged 990 steps per day. Observational data suggest patients accrue about 1,100 steps in the day before discharge, with older patients accruing closer to 900.21 One study found that older patients with fewer than 900 steps per day were likely to experience a functional decline.32 We also found that patients who achieved at least 900 steps were more likely to go home. However, we found that a lower threshold, namely, 400 steps, offered better discrimination between patients who go home and those who do not. Future prospective studies are needed to establish the appropriate goal for exercise interventions. A lower step goal could dramatically enhance the efficiency of the intervention.

A Cochrane review found that pooled analysis of multidisciplinary interventions that included exercise, often in the form of walking, achieved a small but significant increase in the proportion of patients discharged to home (RR 1.08, 95%CI 1.03 to 1.14).¹¹ We found no significant change in the discharge disposition, but our study was underpowered for this endpoint. The six-clicks score showed a small but significant change in the per protocol analysis. The six-clicks score has been shown to correlate with discharge disposition,^28,29 and an improvement in the score suggests that discharge disposition may be influenced.

The intervention may also not have been implemented for long enough. On average, visits were achieved for one-third of the hospital stay partly because of the delay in PT evaluation, which we required for eligibility. In practice, PT evaluation can occur just a few days before the anticipated discharge. We observed a dose dependent response among patients in the intervention group, suggesting that earlier intervention could be more effective. Earlier intervention might be achieved if the MT performed the six-clicks on potentially eligible patients.

The effects of hospitalization on mobility may be the most pronounced in the long term; one study found that 40% of hospitalized older patients manifested new ADL or IADL disability three months after discharge compared with 31% at discharge.⁷ Hospital-based mobility interventions may continue to affect subjects’ independence for weeks or months. In one RCT, an inpatient ambulation intervention improved mobility in the community one month after discharge.³⁷ A hospital-based exercise program that included ambulation achieved better functional outcomes one month later.¹³ One RCT that combined inpatient exercise with outpatient care coordination also decreased readmission rates.³⁴ We found that the intervention did not affect readmission.

This pilot study has several limitations. The sample size was small, and the findings need to be replicated in a larger randomized controlled trial. This is particularly important because the two study arms were not balanced in terms of their prehospital activity. After adjustment for prehospital activity, the differences in the step count in the intention-to-treat analysis were no longer significant. As we adjusted the intervention to hospital workflow, the intervention changed over time. The intention-to-treat analysis may therefore underestimate the effect of the intervention. This work provides a basis for future trial. Finally, discharge disposition depends on a complex interplay of factors, including social factors and preferences, which may not be affected by a mobility intervention.

In summary, an inhospital mobility protocol of attempting ambulation delivered by dedicated mobility technicians three times daily successfully increased the daily step counts and mobility scores of the patients. Studies with a larger sample size are needed to determine whether the proposed approach can affect length of hospital stay, discharge disposition, and overall cost of an episode of care.

Disclosures

Mary Stilphen reports consulting for CreCare and Adeo, which license and distribute AM-PAC short forms, including 6 clicks. All other authors report no conflicts of interest.

Funding

This study was supported by a Research Program Committee grant from the Cleveland Clinic.

Individuals aged 65 years and over represent 13% of the United States population and account for nearly 40% of hospital discharges.¹ Bedrest hastens the functional decline of older patients^2-5 and is associated with risk of serious complications, such as falls, delirium, venous thrombosis, and skin breakdown.^6,7 Ambulation is widely recognized as important for improving hospital outcomes.^8-10 Observational studies suggest that increases of 600 steps per day are associated with shortened length of hospital stay.⁹ However, randomized trials of assisted ambulation have not demonstrated consistent benefit.^11-14 As a result, usual care at most hospitals in the United States does not include assisted ambulation. Even when ambulation is ordered, execution of the orders is inconsistent.^15-17

Studies have demonstrated the benefits of various exercise protocols for older patients in rehabilitation facilities,^18,19 medical intensive care units,²⁰ and medical and surgical wards.^13,18,21 These interventions are usually nursing centered; however, assisting patients with ambulation multiple times per day may be a burdensome addition to the myriad responsibilities of nurses.^19,22,23 In fact, ambulation orders are the most frequently overlooked nursing task.²⁴

We designed a graded protocol of assisted ambulation implemented by a dedicated patient care nursing assistant (PCNA) multiple times daily to increase patient mobility. The objective of this study was to assess the feasibility and effectiveness of such an intervention for older inpatients. We hypothesized that the intervention would prove feasible and improve hospital outcomes, including less need for inpatient rehabilitation and shorter length of stay.

We conducted a single-blind randomized controlled trial of patients aged ≥60 years and admitted as medical inpatients to the Cleveland Clinic Main Campus, a tertiary care center with over 1,440 inpatient beds. The consent form and study protocol were approved by the Cleveland Clinic Institutional Review Board, and the study was registered with ClinicalTrials.gov (NCT02757131).

Patients

All patients who were admitted to study wards for a medical illness and evaluated by Physical Therapy (PT) were eligible for the study. PT evaluations were ordered by the medical team if deemed necessary on the basis of factors, such as age, estimated mobility, and concerns raised by the ancillary staff. All patients who were expected to be discharged to a skilled nursing facility placement or who required home PT received a PT evaluation. Assessment of mobility was documented via Activity Measure for Postacute Care Inpatient Basic Mobility “six-clicks” short form, hereafter abbreviated as “six-clicks.” Based on past experience, patients with scores <16 rarely go home (<20% of the time), and those with scores >20 usually go home regardless of ambulation. Therefore, only patients with scores of 16-20 were invited to participate in the study. Although patients who were not evaluated by PT might also benefit from the intervention, we required a six-clicks score to assess eligibility. The exclusion criteria included anticipated remaining length of stay less than three days, admission under observation status, admission to the intensive care unit (ICU,) patients receiving comfort care measures, and patients with medical conditions precluding ambulation, such as decompensated heart failure or unstable angina.

Randomization

Patients were randomized to “usual care” or “mobility technician” after baseline evaluation using a computerized system. A block randomization scheme with a size of four was used to ensure an approximately equal number of patients per group.

Intervention

Patients randomized to the intervention group were asked to participate in the ambulation protocol outlined by the PT three times daily under the supervision of the mobility technician. The protocol involved four exercise levels (sitting, standing, walking, and stairs), which were implemented depending on the patient’s physical capacity. The mobility technicians, who were PCNAs, were trained by the PT team. PCNAs have no specific degrees or certification. They are taught safe handling techniques during their job orientation, so they already had an understanding of how to transfer and assist a patient with ambulation. The mobility technician training consisted of one four-hour session run by the PT team in the physical therapy department and the nursing unit. The training included safe handling practices and basic mobility, such as transfers from bed to chair, bed to standing, walking with assistance, and walking independently with equipment such as cane, rolling walker, and walking belt. All instruction was demonstrated by the trainer, and the mobility technician was then able to practice. The mobility technician then shadowed the trainer and practiced the techniques under supervision. Competency was assessed by the trainer.

The cohort of patients randomized to “usual care” was not seen by the mobility technicians but was not otherwise restricted in nursing’s baseline ability to execute recommendations placed by the PT team. Compliance with the recommendations is highly variable and dependent on patient acuity during the shift, staffing issues, and competing duties. Cleveland Clinic promotes a “culture of mobility,” and nurses are encouraged to get patients out of bed and assist with ambulation.

Study Instruments—Measures of Mobility

The six-clicks instrument is a tool for measuring basic mobility. It was adapted from the Activity Measures for Post-Acute Care (AM-PAC) instrument.²⁵ Although initially created for self-report in the post-acute care setting, six-clicks has been validated for use by PTs in the acute care setting²⁶ and is currently in use at more than 1,000 US hospitals. Cleveland Clinic PTs have used this measure for routine evaluation since 2011. The instrument has high interrater reliability and can predict discharge disposition.^27-29

Each patient was provided with a tracking device (Fitbit) attached at the wrist to record daily steps for measuring mobility. The use of Fitbit has been validated in ambulatory and inpatient settings.³⁰ The device produces step counts within 3% of the observed step count for most patients but may undercount steps in patients with very slow gait.³¹ The device was provided to each enrollee and collected at discharge.

Variables

Demographic information, comorbid diagnoses, and discharge destination were extracted from the electronic medical record. Information on prehospitalization physical activity level was obtained from the initial PT assessment. Falls were tracked through the safety event reporting system.

Outcomes

The primary outcomes were discharge disposition and hospital length of stay. The secondary outcomes included average steps per day, change in six-clicks score from admission to discharge, inpatient mortality, admission to ICU, falls, deep vein thrombosis, pulmonary embolism, or pneumonia, and readmission within 30 days.

Statistical Analysis

Patient characteristics were summarized as means and standard deviations or medians and interquartile ranges for continuous variables and as frequencies and percentages for categorical variables. The t-test or Wilcoxon rank sum test was applied to compare continuous characteristics between the intervention and control groups. Chi-squared test or Fisher’s exact test was applied to compare categorical characteristics. Given its skewed distribution, the length of stay was log-transformed and compared between the two groups using Student’s t-test. Chi-squared test was used to compare categorical outcomes. The analysis of final six-clicks scores was adjusted for baseline scores, and the least-square estimates are provided. A linear mixed effects model was used to compare the number of daily steps taken because each participant had multiple steps measured. Results were adjusted for prehospital activity. In addition to comparing the total steps taken by each group, we determined the proportion of patients who exceeded a particular threshold by taking the average number of steps per day for all subjects and relating it to home discharge using the Receiver Operating Characteristics (ROC) curve. An optimal cut-off was determined to maximize the Youden index. We also compared the proportion of patients who exceeded 900 steps because this value was previously reported as an important threshold.³² All analyses were conducted using intention-to-treat principles. We also conducted a per-protocol analysis in which we limited the intervention group to those who received at least one assisted ambulation session. A dose-response analysis was also performed, in which patients were categorized as not receiving the therapy, receiving sessions on one or two days, or receiving them on more than two days.

All analyses were conducted using R-studio (Boston, MA). Statistical significance was defined as a P-value < .05. Given that this is a pilot study, the results were not adjusted for multiple comparisons.

Characteristics of patients in the intervention and control groups are shown in Table 1. The patients were mostly white and female, with an average age in the mid-70s (range 61-98). All measures evaluated were not significantly different between the intervention and control groups. However, more patients in the intervention group had a prehospital activity level classified as independent.

Table 2 demonstrates the feasibility of the intervention. Of patients randomized to the intervention group, 74% were ambulated at least once. Once enrolled, the patients successfully participated in assisted ambulation for about two-thirds of their hospital stay. However, the intervention was delivered for only one-third of the total length of stay because most patients were not enrolled on admission. On average, the mobility technicians made 11 attempts to ambulate each patient and 56% of these attempts were successful. The proportion of unsuccessful attempts did not change over the course of the study. The reasons for unsuccessful attempts included patient refusal (n = 102) or unavailability (n = 68), mobility technicians running out of time (n = 2), and other (n = 12).

In this pilot study conducted with older medical inpatients, we found that assisted ambulation provided by a dedicated mobility technician was feasible and increased the number of steps taken by patients. Not all patients in the treatment group received the intervention partly due to the fact that the program initially did not include weekends and the mobility technicians were sometimes redirected to other nursing duties. Both issues were addressed during the course of the study. In the per protocol analysis, the intervention increased the average six-clicks score and there was a nonsignificant reduction in the percentage of patients discharged to a rehabilitation facility.

A range of hospital-based mobility interventions have been described. Several of which were complex multidisciplinary interventions that included a mobility component. The compliance rates have ranged from 82% to 93.7%,12,13 although a systematic review noted that many studies do not provide this level of information.¹¹ Interventions were carried out by nursing staff and PT with support from family members and social workers.^33-35 Ambulation-specific programs have also relied on nurses and PT13,14,36 and, occasionally, on research assistants to implement assisted ambulation protocols.^12,37 A recent study that employed research assistants to deliver inhospital ambulation reported achieving 51.3% of intended walks.³⁷

In contradistinction to previous studies, we created a new role, employing PCNAs as dedicated mobility technicians. We did this for two reasons. First, the approach is less expensive than deploying registered nurses or PTs to ambulate patients and therefore more likely to be adopted by hospitals, especially if it can decrease the cost of an episode of care by avoiding subsequent inpatient rehabilitation. Mobility technicians have no degree or certification requirements and are therefore paid less than nurses or physical therapists. Second, by having a single responsibility, mobility technicians were more likely to engage in their task than nurses, who have competing responsibilities. However, when nurse staffing was short, nurse managers were tempted to recall the PCNAs for other nursing duties. It took time before PCNAs and supervisors prioritized this new responsibility. When they did, the number of attempted walks increased substantially, but the percentage of successful attempts remained constant at 56%, highlighting the difficulty of getting hospitalized patients to walk.

On average, patients who received the intervention engaged in 72 minutes of additional physical activity and averaged 990 steps per day. Observational data suggest patients accrue about 1,100 steps in the day before discharge, with older patients accruing closer to 900.21 One study found that older patients with fewer than 900 steps per day were likely to experience a functional decline.32 We also found that patients who achieved at least 900 steps were more likely to go home. However, we found that a lower threshold, namely, 400 steps, offered better discrimination between patients who go home and those who do not. Future prospective studies are needed to establish the appropriate goal for exercise interventions. A lower step goal could dramatically enhance the efficiency of the intervention.

A Cochrane review found that pooled analysis of multidisciplinary interventions that included exercise, often in the form of walking, achieved a small but significant increase in the proportion of patients discharged to home (RR 1.08, 95%CI 1.03 to 1.14).¹¹ We found no significant change in the discharge disposition, but our study was underpowered for this endpoint. The six-clicks score showed a small but significant change in the per protocol analysis. The six-clicks score has been shown to correlate with discharge disposition,^28,29 and an improvement in the score suggests that discharge disposition may be influenced.

The intervention may also not have been implemented for long enough. On average, visits were achieved for one-third of the hospital stay partly because of the delay in PT evaluation, which we required for eligibility. In practice, PT evaluation can occur just a few days before the anticipated discharge. We observed a dose dependent response among patients in the intervention group, suggesting that earlier intervention could be more effective. Earlier intervention might be achieved if the MT performed the six-clicks on potentially eligible patients.

The effects of hospitalization on mobility may be the most pronounced in the long term; one study found that 40% of hospitalized older patients manifested new ADL or IADL disability three months after discharge compared with 31% at discharge.⁷ Hospital-based mobility interventions may continue to affect subjects’ independence for weeks or months. In one RCT, an inpatient ambulation intervention improved mobility in the community one month after discharge.³⁷ A hospital-based exercise program that included ambulation achieved better functional outcomes one month later.¹³ One RCT that combined inpatient exercise with outpatient care coordination also decreased readmission rates.³⁴ We found that the intervention did not affect readmission.

This pilot study has several limitations. The sample size was small, and the findings need to be replicated in a larger randomized controlled trial. This is particularly important because the two study arms were not balanced in terms of their prehospital activity. After adjustment for prehospital activity, the differences in the step count in the intention-to-treat analysis were no longer significant. As we adjusted the intervention to hospital workflow, the intervention changed over time. The intention-to-treat analysis may therefore underestimate the effect of the intervention. This work provides a basis for future trial. Finally, discharge disposition depends on a complex interplay of factors, including social factors and preferences, which may not be affected by a mobility intervention.

In summary, an inhospital mobility protocol of attempting ambulation delivered by dedicated mobility technicians three times daily successfully increased the daily step counts and mobility scores of the patients. Studies with a larger sample size are needed to determine whether the proposed approach can affect length of hospital stay, discharge disposition, and overall cost of an episode of care.

Disclosures

Mary Stilphen reports consulting for CreCare and Adeo, which license and distribute AM-PAC short forms, including 6 clicks. All other authors report no conflicts of interest.

Funding

This study was supported by a Research Program Committee grant from the Cleveland Clinic.

Whether robbing banks or reducing healthcare spending, it makes sense to go where the money is. In the case of healthcare, 32% of spending goes to inpatient care, so hospitals represent a logical target for cost-reduction efforts. Because most hospital costs are fixed, there are basically 2 approaches to reducing spending—shorten length of stay or keep patients out of the hospital altogether. The government has tried both, using the power of financial incentives to spur adoption.

Faced with soaring hospital costs in the 1980s, Medicare introduced its prospective payment system, offering hospitals a fixed payment for each specific Diagnosis-Related Group. Hospitals responded by discharging patients sooner, with a resultant rise in admissions to skilled nursing facilities (SNFs) and rapid growth of the home care industry. Length of stay fell dramatically, dropping 9% in 1984 alone.¹ It continued to decline through the 1990s, falling by almost 20% between 1993 and 2000. In the following decade, despite the rise of hospital medicine, the rate of decrease slowed to 0.2% per year.²

Attention then turned to readmissions. In 2008, the Medicare Payment Advisory Committee proposed that hospitals with high risk-adjusted readmission rates receive lower payments, arguing that readmissions accounted for $15 billion in Medicare spending and that many were preventable. Thus the Hospital Readmissions Reduction Program was born, introducing readmission penalties in 2012.

Numerous interventions emerged from government and nongovernment parties to reduce readmissions. Many used intensive transitional care programs focusing on early follow-up or medication safety, and some even went as far as providing transitional housing.³ Shortly after passage of the Affordable Care Act, readmission rates fell rapidly. Within a few years, however, the rate of decline slowed dramatically and may have reached a plateau.⁴ Many have argued that only a small proportion of readmissions are preventable and that there are more direct ways to promote improved discharge planning without diverting resources from other areas.⁵ It seems that readmissions may not be feasibly reduced much further.

With the advent of accountable care organizations, health systems are now turning their focus to the small population of patients who consume a disproportionate share of healthcare dollars. Because the top 1% of patients—so-called super-utilizers—account for 21% of spending, efforts to reduce their utilization could produce outsized returns.⁶ Initial anecdotal reports described patients with complex physical, behavior, and social needs receiving fragmented care resulting in myriad expensive admissions. The response comprised teams of social workers and community health workers coupled with robust primary care, formulating individualized solutions. However, data supporting the effectiveness of this common-sense approach are lacking. In addition, our understanding of high-cost patients is evolving. For one thing, being a super-utilizer is often temporary, as just over one-quarter are still in that category a year later.⁷ Moreover, not all high-cost patients are frequently admitted.⁸

In this issue of The Journal of Hospital Medicine, Wick et al.⁹ provide additional insight into high utilizers of hospital services. The authors compare definitions of high utilizers based on cost, number of admissions, or cumulative length of stay over one year. Only 10 percent of high utilizers met all 3 definitions. The overlap between high utilizers by cost and length of stay was twice the overlap between high utilizers by number of admissions and either group. This finding is not surprising because hospitals have high fixed costs, so total cost tends to mirror length of stay.

The study was performed in Canada, and the overlap among these groups may be different in the US. The Canadian patients were hospitalized less frequently than their American counterparts, perhaps reflecting better access to primary care in the Canadian system. Regardless, Wick et al.⁹ add to the growing literature suggesting that the terms “high utilization” and “high cost” do not always describe the same population. This finding is important because strategies aimed at patients who are frequently admitted may not be effective for those who generate the highest costs. In trying to reduce overall costs, it may be time to revisit length of stay.

Given the long history of prospective payment in the US and the stagnation in length of stay over the past decade, it is reasonable to wonder whether further reductions are possible. In the study by Wick et al.,⁹ patients with longer lengths of stay were discharged to long-term care facilities. This observation is consistent with others’ reports. Studies of delays in care show that at least 10% of all hospital days can be attributed to delays in discharge, especially to SNFs. In the most recent study, 11% of hospital days were deemed unnecessary by hospitalists, with one-third of those delays due to lack of availability at an extended care facility.¹⁰ Six years earlier, Carey et al. found that 13.5% of inpatient days were unnecessary, with more than 60% of delays attributable to waiting for discharge to a SNF.¹¹

How, then, might we curtail unnecessary waiting, and whose job is it to solve the problem? The prospective payment system should reward hospitals for eliminating waiting—particularly those hospitals operating at capacity, for which the opportunity costs of occupied beds are most acute. Hospitalists, per se, have no incentive to discharge patients who are waiting; these patients are easy to round on, rarely have emergencies, and generate daily bills. Even when hospitalists are employed by the hospital and incentives for both are aligned, hospitalists may still be powerless to discharge waiting patients, summon busy consultants, or create extra slots in the endoscopy suite.

The move to value at the system level may offer hope. As health systems become responsible for the total cost of care, their focus must shift from the individual areas where care is provided to the transitions of care between treatment areas. It is in these transitions that US healthcare has failed most spectacularly, and consequently, it is where the greatest opportunity lies.

To date, most discharge interventions have focused on communication, with a goal of improving patient safety and, to a lesser extent, preventing readmissions. Partnering with SNFs can reduce the rate of readmissions,¹² but for the most part, the incentives for hospitals and post-acute care facilities remain misaligned. Because post-acute care facilities are paid per diem, they have little incentive to reduce patients’ stays or to admit new patients, who are more expensive to care for than existing ones. Physicians round on SNF patients infrequently and have no incentive to discharge patients, exacerbating the problem. Because post-acute care represents a growing proportion of costs for both medical and surgical patients, health systems will need to either have their own facilities or enter into contracts that align the incentives.

What can hospitalists do? As the predominant coordinators of hospitalized patients’ care both for medical and surgical teams, hospitalists meaningfully impact readmissions and lengths of stay through the care they provide.¹³ More important, as their roles in optimizing hospital throughput¹⁴ continue to expand, hospitalists are perhaps best positioned to observe a diverse range of inefficiencies and inadequacies in inpatient practice and translate those observations into new systems of care. Through thoughtful participation in hospital operations, administration, and health services research, hospitalists hold the key to improving the value of care we provide.

Disclosure 

Nothing to report.

Whether robbing banks or reducing healthcare spending, it makes sense to go where the money is. In the case of healthcare, 32% of spending goes to inpatient care, so hospitals represent a logical target for cost-reduction efforts. Because most hospital costs are fixed, there are basically 2 approaches to reducing spending—shorten length of stay or keep patients out of the hospital altogether. The government has tried both, using the power of financial incentives to spur adoption.

Faced with soaring hospital costs in the 1980s, Medicare introduced its prospective payment system, offering hospitals a fixed payment for each specific Diagnosis-Related Group. Hospitals responded by discharging patients sooner, with a resultant rise in admissions to skilled nursing facilities (SNFs) and rapid growth of the home care industry. Length of stay fell dramatically, dropping 9% in 1984 alone.¹ It continued to decline through the 1990s, falling by almost 20% between 1993 and 2000. In the following decade, despite the rise of hospital medicine, the rate of decrease slowed to 0.2% per year.²

Attention then turned to readmissions. In 2008, the Medicare Payment Advisory Committee proposed that hospitals with high risk-adjusted readmission rates receive lower payments, arguing that readmissions accounted for $15 billion in Medicare spending and that many were preventable. Thus the Hospital Readmissions Reduction Program was born, introducing readmission penalties in 2012.

Numerous interventions emerged from government and nongovernment parties to reduce readmissions. Many used intensive transitional care programs focusing on early follow-up or medication safety, and some even went as far as providing transitional housing.³ Shortly after passage of the Affordable Care Act, readmission rates fell rapidly. Within a few years, however, the rate of decline slowed dramatically and may have reached a plateau.⁴ Many have argued that only a small proportion of readmissions are preventable and that there are more direct ways to promote improved discharge planning without diverting resources from other areas.⁵ It seems that readmissions may not be feasibly reduced much further.

With the advent of accountable care organizations, health systems are now turning their focus to the small population of patients who consume a disproportionate share of healthcare dollars. Because the top 1% of patients—so-called super-utilizers—account for 21% of spending, efforts to reduce their utilization could produce outsized returns.⁶ Initial anecdotal reports described patients with complex physical, behavior, and social needs receiving fragmented care resulting in myriad expensive admissions. The response comprised teams of social workers and community health workers coupled with robust primary care, formulating individualized solutions. However, data supporting the effectiveness of this common-sense approach are lacking. In addition, our understanding of high-cost patients is evolving. For one thing, being a super-utilizer is often temporary, as just over one-quarter are still in that category a year later.⁷ Moreover, not all high-cost patients are frequently admitted.⁸

In this issue of The Journal of Hospital Medicine, Wick et al.⁹ provide additional insight into high utilizers of hospital services. The authors compare definitions of high utilizers based on cost, number of admissions, or cumulative length of stay over one year. Only 10 percent of high utilizers met all 3 definitions. The overlap between high utilizers by cost and length of stay was twice the overlap between high utilizers by number of admissions and either group. This finding is not surprising because hospitals have high fixed costs, so total cost tends to mirror length of stay.

The study was performed in Canada, and the overlap among these groups may be different in the US. The Canadian patients were hospitalized less frequently than their American counterparts, perhaps reflecting better access to primary care in the Canadian system. Regardless, Wick et al.⁹ add to the growing literature suggesting that the terms “high utilization” and “high cost” do not always describe the same population. This finding is important because strategies aimed at patients who are frequently admitted may not be effective for those who generate the highest costs. In trying to reduce overall costs, it may be time to revisit length of stay.

Given the long history of prospective payment in the US and the stagnation in length of stay over the past decade, it is reasonable to wonder whether further reductions are possible. In the study by Wick et al.,⁹ patients with longer lengths of stay were discharged to long-term care facilities. This observation is consistent with others’ reports. Studies of delays in care show that at least 10% of all hospital days can be attributed to delays in discharge, especially to SNFs. In the most recent study, 11% of hospital days were deemed unnecessary by hospitalists, with one-third of those delays due to lack of availability at an extended care facility.¹⁰ Six years earlier, Carey et al. found that 13.5% of inpatient days were unnecessary, with more than 60% of delays attributable to waiting for discharge to a SNF.¹¹

How, then, might we curtail unnecessary waiting, and whose job is it to solve the problem? The prospective payment system should reward hospitals for eliminating waiting—particularly those hospitals operating at capacity, for which the opportunity costs of occupied beds are most acute. Hospitalists, per se, have no incentive to discharge patients who are waiting; these patients are easy to round on, rarely have emergencies, and generate daily bills. Even when hospitalists are employed by the hospital and incentives for both are aligned, hospitalists may still be powerless to discharge waiting patients, summon busy consultants, or create extra slots in the endoscopy suite.

The move to value at the system level may offer hope. As health systems become responsible for the total cost of care, their focus must shift from the individual areas where care is provided to the transitions of care between treatment areas. It is in these transitions that US healthcare has failed most spectacularly, and consequently, it is where the greatest opportunity lies.

To date, most discharge interventions have focused on communication, with a goal of improving patient safety and, to a lesser extent, preventing readmissions. Partnering with SNFs can reduce the rate of readmissions,¹² but for the most part, the incentives for hospitals and post-acute care facilities remain misaligned. Because post-acute care facilities are paid per diem, they have little incentive to reduce patients’ stays or to admit new patients, who are more expensive to care for than existing ones. Physicians round on SNF patients infrequently and have no incentive to discharge patients, exacerbating the problem. Because post-acute care represents a growing proportion of costs for both medical and surgical patients, health systems will need to either have their own facilities or enter into contracts that align the incentives.

What can hospitalists do? As the predominant coordinators of hospitalized patients’ care both for medical and surgical teams, hospitalists meaningfully impact readmissions and lengths of stay through the care they provide.¹³ More important, as their roles in optimizing hospital throughput¹⁴ continue to expand, hospitalists are perhaps best positioned to observe a diverse range of inefficiencies and inadequacies in inpatient practice and translate those observations into new systems of care. Through thoughtful participation in hospital operations, administration, and health services research, hospitalists hold the key to improving the value of care we provide.

Disclosure 

Nothing to report.

Whether robbing banks or reducing healthcare spending, it makes sense to go where the money is. In the case of healthcare, 32% of spending goes to inpatient care, so hospitals represent a logical target for cost-reduction efforts. Because most hospital costs are fixed, there are basically 2 approaches to reducing spending—shorten length of stay or keep patients out of the hospital altogether. The government has tried both, using the power of financial incentives to spur adoption.

Faced with soaring hospital costs in the 1980s, Medicare introduced its prospective payment system, offering hospitals a fixed payment for each specific Diagnosis-Related Group. Hospitals responded by discharging patients sooner, with a resultant rise in admissions to skilled nursing facilities (SNFs) and rapid growth of the home care industry. Length of stay fell dramatically, dropping 9% in 1984 alone.¹ It continued to decline through the 1990s, falling by almost 20% between 1993 and 2000. In the following decade, despite the rise of hospital medicine, the rate of decrease slowed to 0.2% per year.²

Attention then turned to readmissions. In 2008, the Medicare Payment Advisory Committee proposed that hospitals with high risk-adjusted readmission rates receive lower payments, arguing that readmissions accounted for $15 billion in Medicare spending and that many were preventable. Thus the Hospital Readmissions Reduction Program was born, introducing readmission penalties in 2012.

Numerous interventions emerged from government and nongovernment parties to reduce readmissions. Many used intensive transitional care programs focusing on early follow-up or medication safety, and some even went as far as providing transitional housing.³ Shortly after passage of the Affordable Care Act, readmission rates fell rapidly. Within a few years, however, the rate of decline slowed dramatically and may have reached a plateau.⁴ Many have argued that only a small proportion of readmissions are preventable and that there are more direct ways to promote improved discharge planning without diverting resources from other areas.⁵ It seems that readmissions may not be feasibly reduced much further.

With the advent of accountable care organizations, health systems are now turning their focus to the small population of patients who consume a disproportionate share of healthcare dollars. Because the top 1% of patients—so-called super-utilizers—account for 21% of spending, efforts to reduce their utilization could produce outsized returns.⁶ Initial anecdotal reports described patients with complex physical, behavior, and social needs receiving fragmented care resulting in myriad expensive admissions. The response comprised teams of social workers and community health workers coupled with robust primary care, formulating individualized solutions. However, data supporting the effectiveness of this common-sense approach are lacking. In addition, our understanding of high-cost patients is evolving. For one thing, being a super-utilizer is often temporary, as just over one-quarter are still in that category a year later.⁷ Moreover, not all high-cost patients are frequently admitted.⁸

In this issue of The Journal of Hospital Medicine, Wick et al.⁹ provide additional insight into high utilizers of hospital services. The authors compare definitions of high utilizers based on cost, number of admissions, or cumulative length of stay over one year. Only 10 percent of high utilizers met all 3 definitions. The overlap between high utilizers by cost and length of stay was twice the overlap between high utilizers by number of admissions and either group. This finding is not surprising because hospitals have high fixed costs, so total cost tends to mirror length of stay.

The study was performed in Canada, and the overlap among these groups may be different in the US. The Canadian patients were hospitalized less frequently than their American counterparts, perhaps reflecting better access to primary care in the Canadian system. Regardless, Wick et al.⁹ add to the growing literature suggesting that the terms “high utilization” and “high cost” do not always describe the same population. This finding is important because strategies aimed at patients who are frequently admitted may not be effective for those who generate the highest costs. In trying to reduce overall costs, it may be time to revisit length of stay.

Given the long history of prospective payment in the US and the stagnation in length of stay over the past decade, it is reasonable to wonder whether further reductions are possible. In the study by Wick et al.,⁹ patients with longer lengths of stay were discharged to long-term care facilities. This observation is consistent with others’ reports. Studies of delays in care show that at least 10% of all hospital days can be attributed to delays in discharge, especially to SNFs. In the most recent study, 11% of hospital days were deemed unnecessary by hospitalists, with one-third of those delays due to lack of availability at an extended care facility.¹⁰ Six years earlier, Carey et al. found that 13.5% of inpatient days were unnecessary, with more than 60% of delays attributable to waiting for discharge to a SNF.¹¹

How, then, might we curtail unnecessary waiting, and whose job is it to solve the problem? The prospective payment system should reward hospitals for eliminating waiting—particularly those hospitals operating at capacity, for which the opportunity costs of occupied beds are most acute. Hospitalists, per se, have no incentive to discharge patients who are waiting; these patients are easy to round on, rarely have emergencies, and generate daily bills. Even when hospitalists are employed by the hospital and incentives for both are aligned, hospitalists may still be powerless to discharge waiting patients, summon busy consultants, or create extra slots in the endoscopy suite.

The move to value at the system level may offer hope. As health systems become responsible for the total cost of care, their focus must shift from the individual areas where care is provided to the transitions of care between treatment areas. It is in these transitions that US healthcare has failed most spectacularly, and consequently, it is where the greatest opportunity lies.

To date, most discharge interventions have focused on communication, with a goal of improving patient safety and, to a lesser extent, preventing readmissions. Partnering with SNFs can reduce the rate of readmissions,¹² but for the most part, the incentives for hospitals and post-acute care facilities remain misaligned. Because post-acute care facilities are paid per diem, they have little incentive to reduce patients’ stays or to admit new patients, who are more expensive to care for than existing ones. Physicians round on SNF patients infrequently and have no incentive to discharge patients, exacerbating the problem. Because post-acute care represents a growing proportion of costs for both medical and surgical patients, health systems will need to either have their own facilities or enter into contracts that align the incentives.

What can hospitalists do? As the predominant coordinators of hospitalized patients’ care both for medical and surgical teams, hospitalists meaningfully impact readmissions and lengths of stay through the care they provide.¹³ More important, as their roles in optimizing hospital throughput¹⁴ continue to expand, hospitalists are perhaps best positioned to observe a diverse range of inefficiencies and inadequacies in inpatient practice and translate those observations into new systems of care. Through thoughtful participation in hospital operations, administration, and health services research, hospitalists hold the key to improving the value of care we provide.

Disclosure 

Nothing to report.