Richard K. Albert, MD

METHODS

RESULTS

DISCUSSION

METHODS

RESULTS

DISCUSSION

METHODS

RESULTS

DISCUSSION

In 2010, the Centers for Medicare and Medicaid Services implemented value‐based purchasing, a payment model that incentivizes hospitals for reaching certain quality and patient experience thresholds and penalizes those that do not, in part on the basis of patient satisfaction scores.[1] Although low patient satisfaction scores will adversely affect institutions financially, they also reflect patients' perceptions of their care. Some studies suggest that hospitals with higher patient satisfaction scores score higher overall on clinical care processes such as core measures compliance, readmission rates, lower mortality rates, and other quality‐of‐care metrics.[2, 3, 4, 5, 6, 7, 8, 9, 10, 11]

The Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey assesses patients' experience following their hospital stay.[1] The percent of top box scores (ie, response of always on a four point scale, or scores of 9 or 10 on a 10‐point scale) are utilized to compare hospitals and determine the reimbursement or penalty a hospital will receive. Although these scores are available to the public on the Hospital Compare website,[12] physicians may not know how their hospital is ranked or how they are individually perceived by their patients. Additionally, these surveys are typically conducted 48 hours to 6 weeks after patients are discharged, and the results are distributed back to the hospitals well after the time that care was provided, thereby offering providers no chance of improving patient satisfaction during a given hospital stay.

Institutions across the country are trying to improve their HCAHPS scores, but there is limited research identifying specific measures providers can implement. Some studies have suggested that utilizing etiquette‐based communication and sitting at the bedside[13, 14] may help improve patient experience with their providers, and more recently, it has been suggested that providing real‐time deidentified patient experience survey results with education and a rewards/emncentive system to residents may help as well.[15]

Surveys conducted during a patient's hospitalization can offer real‐time actionable feedback to providers. We performed a quality‐improvement project that was designed to determine if real‐time feedback to hospitalist physicians, followed by coaching, and revisits to the patients' bedside could improve the results recorded on provider‐specific patient surveys and/or patients' HCAHPS scores or percentile rankings.

METHODS

RESULTS

From April 14, 2014 to September 19, 2014, we enrolled 227 patients in the control group and 228 in the intervention group (Figure 1). Patient demographics are summarized in Table 1. Of the 132 patients in the intervention group who reported anything less than top box scores for any of the 3 questions (thus prompting a revisit by their provider), 106 (80%) were revisited by their provider at least once during their hospitalization.

Figure 1

Daily Surveys

The proportion of patients in both study groups reporting top box scores tended to increase from the first day to the last day of the survey (Figure 2); however, we found no statistically significant differences between the proportion of patients who reported top box scores on first day or last day in the intervention group compared to the control group. The comments made by the patients are summarized in Supporting Table 1 in the online version of this article.

Figure 2

DISCUSSION

The important findings of this study were that (1) daily patient satisfaction scores improved from first day to last day regardless of study group, (2) patients whose providers received real‐time feedback had a trend toward higher HCAHPS proportions for the 3 provider‐related questions as well as the overall rating of the hospital but were not statistically significant, (3) the percentile differences in these 3 questions as well as the overall rating of the hospital were significantly higher in the intervention group as was the median score for the overall hospital rating.

Our original sample size calculation was based upon our own preliminary data, indicating that our baseline top box scores for the daily survey was around 75%. The daily survey top box score on the first day was, however, much lower (Figure 2). Accordingly, although we did not find a significant difference in these daily scores, we were underpowered to find such a difference. Additionally, because only a small percentage of patients are selected for the HCAHPS survey, our ability to detect a difference in this secondary outcome was also limited. We felt that it was important to analyze the percentile comparisons in addition to the proportion of top box scores on the HCAHPS, because the metrics for value‐based purchasing are based upon, in part, how a hospital system compares to other systems. Finally, to improve our power to detect a difference given a small sample size, we converted the scoring system for overall hospital ranking to a continuous variable, which again was noted to be significant.

To our knowledge, this is the first randomized investigation designed to assess the effect of real‐time, patient‐specific feedback to physicians. Real‐time feedback is increasingly being incorporated into medical practice, but there is only limited information available describing how this type of feedback affects outcomes.[22, 23, 24] Banka et al.[15] found that HCAHPS scores improved as a result of real‐time feedback given to residents, but the study was not randomized, utilized a pre‐post design that resulted in there being differences between the patients studied before and after the intervention, and did not provide patient‐specific data to the residents. Tabib et al.[25] found that operating costs decreased 17% after instituting real‐time feedback to providers about these costs. Reeves et al.[26] conducted a cluster randomized trial of a patient feedback survey that was designed to improve nursing care, but the results were reviewed by the nurses several months after patients had been discharged.

The differences in median top box scores and percentile rank that we observed could have resulted from the real‐time feedback, the educational coaching, the fact that the providers revisited the majority of the patients, or a combination of all of the above. Gross et al.[27] found that longer visits lead to higher satisfaction, though others have not found this to necessarily be the case.[28, 29] Lin et al.[30] found that patient satisfaction was affected by the perceived duration of the visit as well as whether expectations on visit length were met and/or exceeded. Brown et al.[31] found that training providers in communication skills improved the providers perception of their communication skills, although patient experience scores did not improve. We feel that the results seen are more likely a combination thereof as opposed to any 1 component of the intervention.

The most commonly reported complaints or concerns in patients' undirected comments often related to communication issues. Comments on subsequent surveys suggested that patient satisfaction improved over time in the intervention group, indicating that perhaps physicians did try to improve in areas that were highlighted by the real‐time feedback, and that patients perceived the physician efforts to do so (eg, They're doing better than the last time you asked. They sat down and talked to me and listened better. They came back and explained to me about my care. They listened better. They should do this survey at the clinic. See Supporting Table 1 in the online version of this article).

Our study has several limitations. First, we did not randomize providers, and many of our providers (approximately 65%) participated in both the control group and also in the intervention group, and thus received real‐time feedback at some point during the study, which could have affected their overall practice and limited our ability to find a difference between the 2 groups. In an attempt to control for this possibility, the study was conducted on an intermittent basis during the study time frame. Furthermore, the proportion of patients who reported top box scores at the beginning of the study did not have a clear trend of change by the end of the study, suggesting that overall clinician practices with respect to patient satisfaction did not change during this short time period.

Second, only a small number of our patients were randomly selected for the HCAHPS survey, which limited our ability to detect significant differences in HCAHPS proportions. Third, the HCAHPS percentiles at our institution at that time were low. Accordingly, the improvements that we observed in patient satisfaction scores might not be reproducible at institutions with higher satisfactions scores. Fourth, time and resources were needed to obtain patient feedback to provide to providers during this study. There are, however, other ways to obtain feedback that are less resource intensive (eg, electronic feedback, the utilization of volunteers, or partnering this with manager rounding). Finally, the study was conducted at a single, university‐affiliated public teaching hospital and was a quality‐improvement initiative, and thus our results are not generalizable to other institutions.

In conclusion, real‐time feedback of patient experience to their providers, coupled with provider education, coaching, and revisits, seems to improve satisfaction of patients hospitalized on general internal medicine units who were cared for by hospitalists.

Disclosure: Nothing to report.

In 2010, the Centers for Medicare and Medicaid Services implemented value‐based purchasing, a payment model that incentivizes hospitals for reaching certain quality and patient experience thresholds and penalizes those that do not, in part on the basis of patient satisfaction scores.[1] Although low patient satisfaction scores will adversely affect institutions financially, they also reflect patients' perceptions of their care. Some studies suggest that hospitals with higher patient satisfaction scores score higher overall on clinical care processes such as core measures compliance, readmission rates, lower mortality rates, and other quality‐of‐care metrics.[2, 3, 4, 5, 6, 7, 8, 9, 10, 11]

The Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey assesses patients' experience following their hospital stay.[1] The percent of top box scores (ie, response of always on a four point scale, or scores of 9 or 10 on a 10‐point scale) are utilized to compare hospitals and determine the reimbursement or penalty a hospital will receive. Although these scores are available to the public on the Hospital Compare website,[12] physicians may not know how their hospital is ranked or how they are individually perceived by their patients. Additionally, these surveys are typically conducted 48 hours to 6 weeks after patients are discharged, and the results are distributed back to the hospitals well after the time that care was provided, thereby offering providers no chance of improving patient satisfaction during a given hospital stay.

Institutions across the country are trying to improve their HCAHPS scores, but there is limited research identifying specific measures providers can implement. Some studies have suggested that utilizing etiquette‐based communication and sitting at the bedside[13, 14] may help improve patient experience with their providers, and more recently, it has been suggested that providing real‐time deidentified patient experience survey results with education and a rewards/emncentive system to residents may help as well.[15]

Surveys conducted during a patient's hospitalization can offer real‐time actionable feedback to providers. We performed a quality‐improvement project that was designed to determine if real‐time feedback to hospitalist physicians, followed by coaching, and revisits to the patients' bedside could improve the results recorded on provider‐specific patient surveys and/or patients' HCAHPS scores or percentile rankings.

METHODS

RESULTS

From April 14, 2014 to September 19, 2014, we enrolled 227 patients in the control group and 228 in the intervention group (Figure 1). Patient demographics are summarized in Table 1. Of the 132 patients in the intervention group who reported anything less than top box scores for any of the 3 questions (thus prompting a revisit by their provider), 106 (80%) were revisited by their provider at least once during their hospitalization.

Figure 1

Daily Surveys

The proportion of patients in both study groups reporting top box scores tended to increase from the first day to the last day of the survey (Figure 2); however, we found no statistically significant differences between the proportion of patients who reported top box scores on first day or last day in the intervention group compared to the control group. The comments made by the patients are summarized in Supporting Table 1 in the online version of this article.

Figure 2

DISCUSSION

The important findings of this study were that (1) daily patient satisfaction scores improved from first day to last day regardless of study group, (2) patients whose providers received real‐time feedback had a trend toward higher HCAHPS proportions for the 3 provider‐related questions as well as the overall rating of the hospital but were not statistically significant, (3) the percentile differences in these 3 questions as well as the overall rating of the hospital were significantly higher in the intervention group as was the median score for the overall hospital rating.

Our original sample size calculation was based upon our own preliminary data, indicating that our baseline top box scores for the daily survey was around 75%. The daily survey top box score on the first day was, however, much lower (Figure 2). Accordingly, although we did not find a significant difference in these daily scores, we were underpowered to find such a difference. Additionally, because only a small percentage of patients are selected for the HCAHPS survey, our ability to detect a difference in this secondary outcome was also limited. We felt that it was important to analyze the percentile comparisons in addition to the proportion of top box scores on the HCAHPS, because the metrics for value‐based purchasing are based upon, in part, how a hospital system compares to other systems. Finally, to improve our power to detect a difference given a small sample size, we converted the scoring system for overall hospital ranking to a continuous variable, which again was noted to be significant.

To our knowledge, this is the first randomized investigation designed to assess the effect of real‐time, patient‐specific feedback to physicians. Real‐time feedback is increasingly being incorporated into medical practice, but there is only limited information available describing how this type of feedback affects outcomes.[22, 23, 24] Banka et al.[15] found that HCAHPS scores improved as a result of real‐time feedback given to residents, but the study was not randomized, utilized a pre‐post design that resulted in there being differences between the patients studied before and after the intervention, and did not provide patient‐specific data to the residents. Tabib et al.[25] found that operating costs decreased 17% after instituting real‐time feedback to providers about these costs. Reeves et al.[26] conducted a cluster randomized trial of a patient feedback survey that was designed to improve nursing care, but the results were reviewed by the nurses several months after patients had been discharged.

The differences in median top box scores and percentile rank that we observed could have resulted from the real‐time feedback, the educational coaching, the fact that the providers revisited the majority of the patients, or a combination of all of the above. Gross et al.[27] found that longer visits lead to higher satisfaction, though others have not found this to necessarily be the case.[28, 29] Lin et al.[30] found that patient satisfaction was affected by the perceived duration of the visit as well as whether expectations on visit length were met and/or exceeded. Brown et al.[31] found that training providers in communication skills improved the providers perception of their communication skills, although patient experience scores did not improve. We feel that the results seen are more likely a combination thereof as opposed to any 1 component of the intervention.

The most commonly reported complaints or concerns in patients' undirected comments often related to communication issues. Comments on subsequent surveys suggested that patient satisfaction improved over time in the intervention group, indicating that perhaps physicians did try to improve in areas that were highlighted by the real‐time feedback, and that patients perceived the physician efforts to do so (eg, They're doing better than the last time you asked. They sat down and talked to me and listened better. They came back and explained to me about my care. They listened better. They should do this survey at the clinic. See Supporting Table 1 in the online version of this article).

Our study has several limitations. First, we did not randomize providers, and many of our providers (approximately 65%) participated in both the control group and also in the intervention group, and thus received real‐time feedback at some point during the study, which could have affected their overall practice and limited our ability to find a difference between the 2 groups. In an attempt to control for this possibility, the study was conducted on an intermittent basis during the study time frame. Furthermore, the proportion of patients who reported top box scores at the beginning of the study did not have a clear trend of change by the end of the study, suggesting that overall clinician practices with respect to patient satisfaction did not change during this short time period.

Second, only a small number of our patients were randomly selected for the HCAHPS survey, which limited our ability to detect significant differences in HCAHPS proportions. Third, the HCAHPS percentiles at our institution at that time were low. Accordingly, the improvements that we observed in patient satisfaction scores might not be reproducible at institutions with higher satisfactions scores. Fourth, time and resources were needed to obtain patient feedback to provide to providers during this study. There are, however, other ways to obtain feedback that are less resource intensive (eg, electronic feedback, the utilization of volunteers, or partnering this with manager rounding). Finally, the study was conducted at a single, university‐affiliated public teaching hospital and was a quality‐improvement initiative, and thus our results are not generalizable to other institutions.

In conclusion, real‐time feedback of patient experience to their providers, coupled with provider education, coaching, and revisits, seems to improve satisfaction of patients hospitalized on general internal medicine units who were cared for by hospitalists.

Disclosure: Nothing to report.

In 2010, the Centers for Medicare and Medicaid Services implemented value‐based purchasing, a payment model that incentivizes hospitals for reaching certain quality and patient experience thresholds and penalizes those that do not, in part on the basis of patient satisfaction scores.[1] Although low patient satisfaction scores will adversely affect institutions financially, they also reflect patients' perceptions of their care. Some studies suggest that hospitals with higher patient satisfaction scores score higher overall on clinical care processes such as core measures compliance, readmission rates, lower mortality rates, and other quality‐of‐care metrics.[2, 3, 4, 5, 6, 7, 8, 9, 10, 11]

The Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey assesses patients' experience following their hospital stay.[1] The percent of top box scores (ie, response of always on a four point scale, or scores of 9 or 10 on a 10‐point scale) are utilized to compare hospitals and determine the reimbursement or penalty a hospital will receive. Although these scores are available to the public on the Hospital Compare website,[12] physicians may not know how their hospital is ranked or how they are individually perceived by their patients. Additionally, these surveys are typically conducted 48 hours to 6 weeks after patients are discharged, and the results are distributed back to the hospitals well after the time that care was provided, thereby offering providers no chance of improving patient satisfaction during a given hospital stay.

Institutions across the country are trying to improve their HCAHPS scores, but there is limited research identifying specific measures providers can implement. Some studies have suggested that utilizing etiquette‐based communication and sitting at the bedside[13, 14] may help improve patient experience with their providers, and more recently, it has been suggested that providing real‐time deidentified patient experience survey results with education and a rewards/emncentive system to residents may help as well.[15]

Surveys conducted during a patient's hospitalization can offer real‐time actionable feedback to providers. We performed a quality‐improvement project that was designed to determine if real‐time feedback to hospitalist physicians, followed by coaching, and revisits to the patients' bedside could improve the results recorded on provider‐specific patient surveys and/or patients' HCAHPS scores or percentile rankings.

METHODS

RESULTS

From April 14, 2014 to September 19, 2014, we enrolled 227 patients in the control group and 228 in the intervention group (Figure 1). Patient demographics are summarized in Table 1. Of the 132 patients in the intervention group who reported anything less than top box scores for any of the 3 questions (thus prompting a revisit by their provider), 106 (80%) were revisited by their provider at least once during their hospitalization.

Figure 1

Daily Surveys

The proportion of patients in both study groups reporting top box scores tended to increase from the first day to the last day of the survey (Figure 2); however, we found no statistically significant differences between the proportion of patients who reported top box scores on first day or last day in the intervention group compared to the control group. The comments made by the patients are summarized in Supporting Table 1 in the online version of this article.

Figure 2

DISCUSSION

The important findings of this study were that (1) daily patient satisfaction scores improved from first day to last day regardless of study group, (2) patients whose providers received real‐time feedback had a trend toward higher HCAHPS proportions for the 3 provider‐related questions as well as the overall rating of the hospital but were not statistically significant, (3) the percentile differences in these 3 questions as well as the overall rating of the hospital were significantly higher in the intervention group as was the median score for the overall hospital rating.

Our original sample size calculation was based upon our own preliminary data, indicating that our baseline top box scores for the daily survey was around 75%. The daily survey top box score on the first day was, however, much lower (Figure 2). Accordingly, although we did not find a significant difference in these daily scores, we were underpowered to find such a difference. Additionally, because only a small percentage of patients are selected for the HCAHPS survey, our ability to detect a difference in this secondary outcome was also limited. We felt that it was important to analyze the percentile comparisons in addition to the proportion of top box scores on the HCAHPS, because the metrics for value‐based purchasing are based upon, in part, how a hospital system compares to other systems. Finally, to improve our power to detect a difference given a small sample size, we converted the scoring system for overall hospital ranking to a continuous variable, which again was noted to be significant.

To our knowledge, this is the first randomized investigation designed to assess the effect of real‐time, patient‐specific feedback to physicians. Real‐time feedback is increasingly being incorporated into medical practice, but there is only limited information available describing how this type of feedback affects outcomes.[22, 23, 24] Banka et al.[15] found that HCAHPS scores improved as a result of real‐time feedback given to residents, but the study was not randomized, utilized a pre‐post design that resulted in there being differences between the patients studied before and after the intervention, and did not provide patient‐specific data to the residents. Tabib et al.[25] found that operating costs decreased 17% after instituting real‐time feedback to providers about these costs. Reeves et al.[26] conducted a cluster randomized trial of a patient feedback survey that was designed to improve nursing care, but the results were reviewed by the nurses several months after patients had been discharged.

The differences in median top box scores and percentile rank that we observed could have resulted from the real‐time feedback, the educational coaching, the fact that the providers revisited the majority of the patients, or a combination of all of the above. Gross et al.[27] found that longer visits lead to higher satisfaction, though others have not found this to necessarily be the case.[28, 29] Lin et al.[30] found that patient satisfaction was affected by the perceived duration of the visit as well as whether expectations on visit length were met and/or exceeded. Brown et al.[31] found that training providers in communication skills improved the providers perception of their communication skills, although patient experience scores did not improve. We feel that the results seen are more likely a combination thereof as opposed to any 1 component of the intervention.

The most commonly reported complaints or concerns in patients' undirected comments often related to communication issues. Comments on subsequent surveys suggested that patient satisfaction improved over time in the intervention group, indicating that perhaps physicians did try to improve in areas that were highlighted by the real‐time feedback, and that patients perceived the physician efforts to do so (eg, They're doing better than the last time you asked. They sat down and talked to me and listened better. They came back and explained to me about my care. They listened better. They should do this survey at the clinic. See Supporting Table 1 in the online version of this article).

Our study has several limitations. First, we did not randomize providers, and many of our providers (approximately 65%) participated in both the control group and also in the intervention group, and thus received real‐time feedback at some point during the study, which could have affected their overall practice and limited our ability to find a difference between the 2 groups. In an attempt to control for this possibility, the study was conducted on an intermittent basis during the study time frame. Furthermore, the proportion of patients who reported top box scores at the beginning of the study did not have a clear trend of change by the end of the study, suggesting that overall clinician practices with respect to patient satisfaction did not change during this short time period.

Second, only a small number of our patients were randomly selected for the HCAHPS survey, which limited our ability to detect significant differences in HCAHPS proportions. Third, the HCAHPS percentiles at our institution at that time were low. Accordingly, the improvements that we observed in patient satisfaction scores might not be reproducible at institutions with higher satisfactions scores. Fourth, time and resources were needed to obtain patient feedback to provide to providers during this study. There are, however, other ways to obtain feedback that are less resource intensive (eg, electronic feedback, the utilization of volunteers, or partnering this with manager rounding). Finally, the study was conducted at a single, university‐affiliated public teaching hospital and was a quality‐improvement initiative, and thus our results are not generalizable to other institutions.

In conclusion, real‐time feedback of patient experience to their providers, coupled with provider education, coaching, and revisits, seems to improve satisfaction of patients hospitalized on general internal medicine units who were cared for by hospitalists.

Disclosure: Nothing to report.

Gender disparities still exist for women in academic medicine.[1, 2, 3, 4, 5, 6, 7, 8, 9] The most recent data from the Association of American Medical Colleges (AAMC) show that although gender disparities are decreasing, women are still under‐represented in the assistant, associate, and full‐professor ranks as well as in leadership positions.[1]

Some studies indicate that gender differences are less evident when examining younger cohorts.[1, 10, 11, 12, 13] Hospital medicine emerged around 1996, when the term hospitalist was first coined.[14] The gender distribution of academic hospitalists is likely nearly equal,[15, 16] and they are generally younger physicians.[15, 17, 18, 19, 20] Accordingly, we questioned whether gender disparities existed in academic hospital medicine (HM) and, if so, whether these disparities were greater than those that might exist in academic general internal medicine (GIM).

METHODS

This study consisted of both prospective and retrospective observation of data collected for academic adult hospitalists and general internists who practice in the United States. It was approved by the Colorado Multiple Institutional Review Board.

Gender distribution was assessed with respect to: (1) academic HM and GIM faculty, (2) leadership (ie, division or section heads), and (3) scholarly work (ie, speaking opportunities and publications). Data were collected between October 1, 2012 and August 31, 2014.

RESULTS

DISCUSSION

The important findings of this study are that, despite an equal gender distribution of academic HM and GIM faculty, fewer women were HM division/section chiefs, fewer women were speakers at the 2 selected national meetings, and fewer women were first or last authors of publications in 2 selected journals in comparison with general internists.

Previous studies have found that women lag behind their male counterparts with respect to academic productivity, leadership, and promotion.[1, 5, 7] Some studies suggest, however, that gender differences are reduced when younger cohorts are examined.[1, 10, 11, 12, 13] Surveys indicate that that the mean age of hospitalists is younger than most other specialties.[15, 19, 20, 24] The mean age of academic GIM physicians is unknown, but surveys of GIM (not differentiating academic from nonacademic) suggest that it is an older cohort than that of HM.[24] Despite hospitalists being a younger cohort, we found gender disparities in all areas investigated.

Our findings with respect to gender disparities in HM division or section leadership are consistent with the annual AAMC Women in US Academic Medicine and Science Benchmarking Report that found only 22% of all permanent division or section heads were women.[1]

Gender disparities with respect to authorship of medical publications have been previously noted,[3, 6, 15, 25] but to our knowledge, this is the first study to investigate the gender of authors who were hospitalists. Although we found a higher proportion of women hospitalists who were first or last authors than was observed by Jagsi and colleagues,[3] women hospitalists were still under‐represented with respect to this measure of academic productivity. Erren et al. reviewed 6 major journals from 2010 and 2011, and found that first authorship of original research by women ranged from 23.7% to 46.7%, and for last authorship from 18.3% to 28.8%.[25] Interestingly, we found no significant gender difference for first authors who were general internists, and there was a trend toward more women general internists being first authors than men for original research, reviews, and brief reports (data not shown).

Our study did not attempt to answer the question of why gender disparities persist, but many previous studies have explored this issue.[4, 8, 12, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42] Issues raised by others include the quantity of academic work (ie, publications and grants obtained), differences in hours worked and allocation of time, lack of mentorship, family responsibilities, discrimination, differences in career motivation, and levels of institutional support, to name a few.

The under‐representation of women hospitalists in leadership, authorship, and speaking opportunities may be consistent with gender‐related differences in research productivity. Fewer publications could lead to fewer national presentations, which could lead to fewer leadership opportunities. Our findings with respect to general internists are not consistent with this idea, however, as whereas women were under‐represented in GIM leadership positions, we found no disparities with respect to the gender of first authors or speakers at national meetings for general internists. The finding that hospitalists had gender disparities with respect to first authors and national speakers but general internists did not, argues against several hypotheses (ie, that women lack mentorship, have less career motivation, fewer career building opportunities).

One notable hypothesis, and perhaps one that is often discussed in the literature, is that women shoulder the majority of family responsibilities, and this may result in women having less time for their careers. Jolly and colleagues studied physician‐researchers and noted that women were more likely than men to have spouses or domestic partners who were fully employed, spent 8.5 more hours per week on domestic activities, and were more likely to take time off during disruptions of usual child care.[33] Carr and colleagues found that women with children (compared to men with children) had fewer publications, slower self‐perceived career progress, and lower career satisfaction, but having children had little effect on faculty aspirations and goals.[2] Kaplan et al., however, found that family responsibilities do not appear to account for sex differences in academic advancement.[4] Interestingly, in a study comparing physicians from Generation X to those of the Baby Boomer age, Generation X women reported working more than their male Generation X counterparts, and both had more of a focus on worklife balance than the older generation.[12]

The nature the of 2 specialties' work environment and job requirements could have also resulted in some of the differences seen. Primary care clinical work is typically conducted Monday through Friday, and hospitalist work frequently includes some weekend, evening, night, and holiday coverage. Although these are known differences, both specialties have also been noted to offer many advantages to women and men alike, including collaborative working environments and flexible work hours.[16]

Finally, finding disparity in leadership positions in both specialties supports the possibility that those responsible for hiring could have implicit gender biases. Under‐representation in entry‐level positions is also not a likely explanation for the differences we observed, because nearly an equal number of men and women graduate from medical school, pursue residency training in internal medicine, and become either academic hospitalists or general internists at university settings.[1, 15, 24] This hypothesis could, however, explain why disparities exist with respect to senior authorship and leadership positions, as typically, these individuals have been in practice longer and the current trends of improved gender equality have not always been the case.

Our study has a number of limitations. First, we only examined publications in 2 journals and presentations at 2 national conferences, although the journals and conferences selected are considered to be the major ones in the 2 specialties. Second, using Internet searches may have resulted in inaccurate gender and specialty assignment, but previous studies have used similar methodology.[3, 43] Additionally, we also attempted to contact individuals for direct confirmation when the information we obtained was not clear and had a second investigator independently verify the gender and specialty data. Third, we utilized division/department websites when available to determine the gender of HM divisions/sections. If not recently updated, these websites may not have reflected the most current leader of the unit, but this concern would seemingly pertain to both hospitalists and general internists. Fourth, we opted to only study faculty and division/section heads at university hospitals, as typically these institutions had GIM and hospitalist groups and also typically had websites. Because we only studied faculty and leadership at university hospitals, our data are not generalizable to all hospitalist and GIM groups. Finally, we excluded pediatric hospitalists, and thus, this study is representative of adult hospitalists only. Including pediatric hospitalists was out of the scope of this project.

Our study also had a number of strengths. To our knowledge, this is the first study to provide an estimate of the gender distribution in academic HM, of hospitalists as speakers at national meetings, as first and last authors, and of HM division or section heads, and is the first to compare these results with those observed for general internists. In addition, we examined 7 years of data from 2 of the major journals and national conferences for these specialties.

In summary, despite HM being a newer field with a younger cohort of physicians, we found that gender disparities exist for women with respect to authorship, national speaking opportunities, and division or section leadership. Identifying why these gender differences exist presents an important next step.

Disclosures: Nothing to report. Marisha Burden, MD and Maria G. Frank, MD are coprincipal authors.

Gender disparities still exist for women in academic medicine.[1, 2, 3, 4, 5, 6, 7, 8, 9] The most recent data from the Association of American Medical Colleges (AAMC) show that although gender disparities are decreasing, women are still under‐represented in the assistant, associate, and full‐professor ranks as well as in leadership positions.[1]

Some studies indicate that gender differences are less evident when examining younger cohorts.[1, 10, 11, 12, 13] Hospital medicine emerged around 1996, when the term hospitalist was first coined.[14] The gender distribution of academic hospitalists is likely nearly equal,[15, 16] and they are generally younger physicians.[15, 17, 18, 19, 20] Accordingly, we questioned whether gender disparities existed in academic hospital medicine (HM) and, if so, whether these disparities were greater than those that might exist in academic general internal medicine (GIM).

METHODS

This study consisted of both prospective and retrospective observation of data collected for academic adult hospitalists and general internists who practice in the United States. It was approved by the Colorado Multiple Institutional Review Board.

Gender distribution was assessed with respect to: (1) academic HM and GIM faculty, (2) leadership (ie, division or section heads), and (3) scholarly work (ie, speaking opportunities and publications). Data were collected between October 1, 2012 and August 31, 2014.

RESULTS

DISCUSSION

The important findings of this study are that, despite an equal gender distribution of academic HM and GIM faculty, fewer women were HM division/section chiefs, fewer women were speakers at the 2 selected national meetings, and fewer women were first or last authors of publications in 2 selected journals in comparison with general internists.

Previous studies have found that women lag behind their male counterparts with respect to academic productivity, leadership, and promotion.[1, 5, 7] Some studies suggest, however, that gender differences are reduced when younger cohorts are examined.[1, 10, 11, 12, 13] Surveys indicate that that the mean age of hospitalists is younger than most other specialties.[15, 19, 20, 24] The mean age of academic GIM physicians is unknown, but surveys of GIM (not differentiating academic from nonacademic) suggest that it is an older cohort than that of HM.[24] Despite hospitalists being a younger cohort, we found gender disparities in all areas investigated.

Our findings with respect to gender disparities in HM division or section leadership are consistent with the annual AAMC Women in US Academic Medicine and Science Benchmarking Report that found only 22% of all permanent division or section heads were women.[1]

Gender disparities with respect to authorship of medical publications have been previously noted,[3, 6, 15, 25] but to our knowledge, this is the first study to investigate the gender of authors who were hospitalists. Although we found a higher proportion of women hospitalists who were first or last authors than was observed by Jagsi and colleagues,[3] women hospitalists were still under‐represented with respect to this measure of academic productivity. Erren et al. reviewed 6 major journals from 2010 and 2011, and found that first authorship of original research by women ranged from 23.7% to 46.7%, and for last authorship from 18.3% to 28.8%.[25] Interestingly, we found no significant gender difference for first authors who were general internists, and there was a trend toward more women general internists being first authors than men for original research, reviews, and brief reports (data not shown).

Our study did not attempt to answer the question of why gender disparities persist, but many previous studies have explored this issue.[4, 8, 12, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42] Issues raised by others include the quantity of academic work (ie, publications and grants obtained), differences in hours worked and allocation of time, lack of mentorship, family responsibilities, discrimination, differences in career motivation, and levels of institutional support, to name a few.

The under‐representation of women hospitalists in leadership, authorship, and speaking opportunities may be consistent with gender‐related differences in research productivity. Fewer publications could lead to fewer national presentations, which could lead to fewer leadership opportunities. Our findings with respect to general internists are not consistent with this idea, however, as whereas women were under‐represented in GIM leadership positions, we found no disparities with respect to the gender of first authors or speakers at national meetings for general internists. The finding that hospitalists had gender disparities with respect to first authors and national speakers but general internists did not, argues against several hypotheses (ie, that women lack mentorship, have less career motivation, fewer career building opportunities).

One notable hypothesis, and perhaps one that is often discussed in the literature, is that women shoulder the majority of family responsibilities, and this may result in women having less time for their careers. Jolly and colleagues studied physician‐researchers and noted that women were more likely than men to have spouses or domestic partners who were fully employed, spent 8.5 more hours per week on domestic activities, and were more likely to take time off during disruptions of usual child care.[33] Carr and colleagues found that women with children (compared to men with children) had fewer publications, slower self‐perceived career progress, and lower career satisfaction, but having children had little effect on faculty aspirations and goals.[2] Kaplan et al., however, found that family responsibilities do not appear to account for sex differences in academic advancement.[4] Interestingly, in a study comparing physicians from Generation X to those of the Baby Boomer age, Generation X women reported working more than their male Generation X counterparts, and both had more of a focus on worklife balance than the older generation.[12]

The nature the of 2 specialties' work environment and job requirements could have also resulted in some of the differences seen. Primary care clinical work is typically conducted Monday through Friday, and hospitalist work frequently includes some weekend, evening, night, and holiday coverage. Although these are known differences, both specialties have also been noted to offer many advantages to women and men alike, including collaborative working environments and flexible work hours.[16]

Finally, finding disparity in leadership positions in both specialties supports the possibility that those responsible for hiring could have implicit gender biases. Under‐representation in entry‐level positions is also not a likely explanation for the differences we observed, because nearly an equal number of men and women graduate from medical school, pursue residency training in internal medicine, and become either academic hospitalists or general internists at university settings.[1, 15, 24] This hypothesis could, however, explain why disparities exist with respect to senior authorship and leadership positions, as typically, these individuals have been in practice longer and the current trends of improved gender equality have not always been the case.

Our study has a number of limitations. First, we only examined publications in 2 journals and presentations at 2 national conferences, although the journals and conferences selected are considered to be the major ones in the 2 specialties. Second, using Internet searches may have resulted in inaccurate gender and specialty assignment, but previous studies have used similar methodology.[3, 43] Additionally, we also attempted to contact individuals for direct confirmation when the information we obtained was not clear and had a second investigator independently verify the gender and specialty data. Third, we utilized division/department websites when available to determine the gender of HM divisions/sections. If not recently updated, these websites may not have reflected the most current leader of the unit, but this concern would seemingly pertain to both hospitalists and general internists. Fourth, we opted to only study faculty and division/section heads at university hospitals, as typically these institutions had GIM and hospitalist groups and also typically had websites. Because we only studied faculty and leadership at university hospitals, our data are not generalizable to all hospitalist and GIM groups. Finally, we excluded pediatric hospitalists, and thus, this study is representative of adult hospitalists only. Including pediatric hospitalists was out of the scope of this project.

Our study also had a number of strengths. To our knowledge, this is the first study to provide an estimate of the gender distribution in academic HM, of hospitalists as speakers at national meetings, as first and last authors, and of HM division or section heads, and is the first to compare these results with those observed for general internists. In addition, we examined 7 years of data from 2 of the major journals and national conferences for these specialties.

In summary, despite HM being a newer field with a younger cohort of physicians, we found that gender disparities exist for women with respect to authorship, national speaking opportunities, and division or section leadership. Identifying why these gender differences exist presents an important next step.

Disclosures: Nothing to report. Marisha Burden, MD and Maria G. Frank, MD are coprincipal authors.

Gender disparities still exist for women in academic medicine.[1, 2, 3, 4, 5, 6, 7, 8, 9] The most recent data from the Association of American Medical Colleges (AAMC) show that although gender disparities are decreasing, women are still under‐represented in the assistant, associate, and full‐professor ranks as well as in leadership positions.[1]

Some studies indicate that gender differences are less evident when examining younger cohorts.[1, 10, 11, 12, 13] Hospital medicine emerged around 1996, when the term hospitalist was first coined.[14] The gender distribution of academic hospitalists is likely nearly equal,[15, 16] and they are generally younger physicians.[15, 17, 18, 19, 20] Accordingly, we questioned whether gender disparities existed in academic hospital medicine (HM) and, if so, whether these disparities were greater than those that might exist in academic general internal medicine (GIM).

METHODS

This study consisted of both prospective and retrospective observation of data collected for academic adult hospitalists and general internists who practice in the United States. It was approved by the Colorado Multiple Institutional Review Board.

Gender distribution was assessed with respect to: (1) academic HM and GIM faculty, (2) leadership (ie, division or section heads), and (3) scholarly work (ie, speaking opportunities and publications). Data were collected between October 1, 2012 and August 31, 2014.

RESULTS

DISCUSSION

The important findings of this study are that, despite an equal gender distribution of academic HM and GIM faculty, fewer women were HM division/section chiefs, fewer women were speakers at the 2 selected national meetings, and fewer women were first or last authors of publications in 2 selected journals in comparison with general internists.

Previous studies have found that women lag behind their male counterparts with respect to academic productivity, leadership, and promotion.[1, 5, 7] Some studies suggest, however, that gender differences are reduced when younger cohorts are examined.[1, 10, 11, 12, 13] Surveys indicate that that the mean age of hospitalists is younger than most other specialties.[15, 19, 20, 24] The mean age of academic GIM physicians is unknown, but surveys of GIM (not differentiating academic from nonacademic) suggest that it is an older cohort than that of HM.[24] Despite hospitalists being a younger cohort, we found gender disparities in all areas investigated.

Our findings with respect to gender disparities in HM division or section leadership are consistent with the annual AAMC Women in US Academic Medicine and Science Benchmarking Report that found only 22% of all permanent division or section heads were women.[1]

Gender disparities with respect to authorship of medical publications have been previously noted,[3, 6, 15, 25] but to our knowledge, this is the first study to investigate the gender of authors who were hospitalists. Although we found a higher proportion of women hospitalists who were first or last authors than was observed by Jagsi and colleagues,[3] women hospitalists were still under‐represented with respect to this measure of academic productivity. Erren et al. reviewed 6 major journals from 2010 and 2011, and found that first authorship of original research by women ranged from 23.7% to 46.7%, and for last authorship from 18.3% to 28.8%.[25] Interestingly, we found no significant gender difference for first authors who were general internists, and there was a trend toward more women general internists being first authors than men for original research, reviews, and brief reports (data not shown).

Our study did not attempt to answer the question of why gender disparities persist, but many previous studies have explored this issue.[4, 8, 12, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42] Issues raised by others include the quantity of academic work (ie, publications and grants obtained), differences in hours worked and allocation of time, lack of mentorship, family responsibilities, discrimination, differences in career motivation, and levels of institutional support, to name a few.

The under‐representation of women hospitalists in leadership, authorship, and speaking opportunities may be consistent with gender‐related differences in research productivity. Fewer publications could lead to fewer national presentations, which could lead to fewer leadership opportunities. Our findings with respect to general internists are not consistent with this idea, however, as whereas women were under‐represented in GIM leadership positions, we found no disparities with respect to the gender of first authors or speakers at national meetings for general internists. The finding that hospitalists had gender disparities with respect to first authors and national speakers but general internists did not, argues against several hypotheses (ie, that women lack mentorship, have less career motivation, fewer career building opportunities).

One notable hypothesis, and perhaps one that is often discussed in the literature, is that women shoulder the majority of family responsibilities, and this may result in women having less time for their careers. Jolly and colleagues studied physician‐researchers and noted that women were more likely than men to have spouses or domestic partners who were fully employed, spent 8.5 more hours per week on domestic activities, and were more likely to take time off during disruptions of usual child care.[33] Carr and colleagues found that women with children (compared to men with children) had fewer publications, slower self‐perceived career progress, and lower career satisfaction, but having children had little effect on faculty aspirations and goals.[2] Kaplan et al., however, found that family responsibilities do not appear to account for sex differences in academic advancement.[4] Interestingly, in a study comparing physicians from Generation X to those of the Baby Boomer age, Generation X women reported working more than their male Generation X counterparts, and both had more of a focus on worklife balance than the older generation.[12]

The nature the of 2 specialties' work environment and job requirements could have also resulted in some of the differences seen. Primary care clinical work is typically conducted Monday through Friday, and hospitalist work frequently includes some weekend, evening, night, and holiday coverage. Although these are known differences, both specialties have also been noted to offer many advantages to women and men alike, including collaborative working environments and flexible work hours.[16]

Finally, finding disparity in leadership positions in both specialties supports the possibility that those responsible for hiring could have implicit gender biases. Under‐representation in entry‐level positions is also not a likely explanation for the differences we observed, because nearly an equal number of men and women graduate from medical school, pursue residency training in internal medicine, and become either academic hospitalists or general internists at university settings.[1, 15, 24] This hypothesis could, however, explain why disparities exist with respect to senior authorship and leadership positions, as typically, these individuals have been in practice longer and the current trends of improved gender equality have not always been the case.

Our study has a number of limitations. First, we only examined publications in 2 journals and presentations at 2 national conferences, although the journals and conferences selected are considered to be the major ones in the 2 specialties. Second, using Internet searches may have resulted in inaccurate gender and specialty assignment, but previous studies have used similar methodology.[3, 43] Additionally, we also attempted to contact individuals for direct confirmation when the information we obtained was not clear and had a second investigator independently verify the gender and specialty data. Third, we utilized division/department websites when available to determine the gender of HM divisions/sections. If not recently updated, these websites may not have reflected the most current leader of the unit, but this concern would seemingly pertain to both hospitalists and general internists. Fourth, we opted to only study faculty and division/section heads at university hospitals, as typically these institutions had GIM and hospitalist groups and also typically had websites. Because we only studied faculty and leadership at university hospitals, our data are not generalizable to all hospitalist and GIM groups. Finally, we excluded pediatric hospitalists, and thus, this study is representative of adult hospitalists only. Including pediatric hospitalists was out of the scope of this project.

Our study also had a number of strengths. To our knowledge, this is the first study to provide an estimate of the gender distribution in academic HM, of hospitalists as speakers at national meetings, as first and last authors, and of HM division or section heads, and is the first to compare these results with those observed for general internists. In addition, we examined 7 years of data from 2 of the major journals and national conferences for these specialties.

In summary, despite HM being a newer field with a younger cohort of physicians, we found that gender disparities exist for women with respect to authorship, national speaking opportunities, and division or section leadership. Identifying why these gender differences exist presents an important next step.

Disclosures: Nothing to report. Marisha Burden, MD and Maria G. Frank, MD are coprincipal authors.

The period immediately following hospital discharge is particularly hazardous for patients.[1, 2, 3, 4, 5] Problems occurring after discharge may result in high rates of rehospitalization and unscheduled visits to healthcare providers.[6, 7, 8, 9, 10] Numerous investigators have tried to identify patients who are at increased risk for rehospitalizations within 30 days of discharge, and many studies have examined whether various interventions could decrease these adverse events (summarized in Hansen et al.[11]). An increasing fraction of patients discharged by medicine and surgery services have some or all of their care supervised by hospitalists. Thus, hospitals increasingly look to hospitalists for ways to reduce rehospitalizations.

Patients discharged from our hospital are instructed to call an advice line (AL) if and when questions or concerns arise. Accordingly, we examined when these calls were made and what issues were raised, with the idea that the information collected might identify aspects of our discharge processes that needed improvement.

METHODS

RESULTS

During the 1‐year study period, 19,303 unique patients were discharged home with instructions regarding the use of the AL. A total of 310 patients called the AL and reported being hospitalized or having surgery within the preceding 30 days. Of these, 2 were excluded (1 who was incarcerated and 1 who was discharged to a skilled nursing facility), leaving 308 patients in the cohort. This represented 1.5% of the total number of unduplicated patients discharged during this same time period (minus the exclusions described above). The large majority of the calls (277/308, 90%) came directly from patients. The remaining 10% came from a proxy, usually a patient's family member. Compared with patients who were discharged during the same time period who did not call the AL, those who called were more likely to speak English, less likely to speak Spanish, more likely to be medically indigent, had slightly longer lengths of stays for their index hospitalization, and were more likely to be discharged from surgery than medicine services (particularly following inpatient surgery) (Table 1).

The median time from hospital discharge to the call was 3 days (interquartile range [IQR], 16), but 31% and 47% of calls occurred within 24 or 48 hours of discharge, respectively. Ten percent of patients called the AL the same day of discharge (Figure 1). We found no difference in timing of the calls as a function of discharging service.

Figure 1

The 308 patients reported a total of 612 problems or concerns (meanstandard deviation number of complaints per caller=21), the large majority of which (71%) were symptom‐related (Table 2). The most common symptom was uncontrolled pain, reported by 33% and 40% of patients discharged from medicine and surgery services, respectively. The next most common symptoms related to the gastrointestinal system and to surgical site issues in medicine and surgery patients, respectively (data not shown).

Sixty‐five patients, representing 21% of the cohort, reported 81 problems understanding or executing discharge instructions. No difference was observed between the fraction of these problems reported by patients from medicine versus surgery (19% and 22%, respectively, P=0.54).

Sixty‐five patients, again representing 21% of the cohort, reported 87 medication‐related problems, 20% from both the medicine and surgery services (P=0.99). Medicine patients more frequently reported difficulties understanding their medication instructions, whereas surgery patients more frequently reported lack of efficacy of medications, particularly with respect to pain control (data not shown).

Thirty percent of patients who called the AL were advised by the nurse to go to the emergency department immediately. Medicine patients were more likely to be triaged to the emergency department compared with surgery patients (45% vs 22%, P<0.0001).

The 30‐day readmission rates and the rates of unscheduled urgent or emergent care visits were higher for patients calling the AL compared with those who did not call (46/308, 15% vs 706/18,995, 4%, and 92/308, 30% vs 1303/18,995, 7%, respectively, both P<0.0001). Similar differences were found for patients discharged from medicine or surgery services who called the AL compared with those who did not (data not shown, both P<0.0001). The median number of days between AL call and rehospitalization was 0 (IQR, 02) and 1 (IQR, 08) for medicine and surgery patients, respectively. Ninety‐three percent of rehospitalizations were related to the index hospitalization, and 78% of patients who were readmitted had no outpatient encounter in the interim between discharge and rehospitalization.

DISCUSSION

We investigated the source and nature of patient telephone calls to an AL following a hospitalization or surgery, and our data revealed the following important findings: (1) nearly one‐half of the calls to the AL occurred within the first 48 hours following discharge; (2) the majority of the calls came from surgery patients, and a greater fraction of patients discharged from surgery services called the AL than patients discharged from medicine services; (3) the most common issues were uncontrolled pain, questions about medications, and problems understanding or executing aftercare instructions (particularly pertaining to the care of surgical wounds); and (4) patients calling the AL had higher rates of 30‐day rehospitalization and of unscheduled urgent or emergent care visits.

The utilization of our patient‐initiated call line was only 1.5%, which was on the low end of the 1% to 10% reported in the literature.[7, 12] This can be attributed to a number of issues that are specific to our system. First, the discharge instructions provided to our patients stated that they should call their primary care provider or the AL if they had questions. Accordingly, because approximately 50% of our patients had a primary care provider in our system, some may have preferentially contacted their primary care provider rather than the AL. Second, the instructions stated that the patients should call if they were experiencing the symptoms listed on the instruction sheet, so those with other problems/complaints may not have called. Third, AL personnel identified patients as being in our cohort by asking if they had been discharged or underwent a surgical procedure within 30‐days of their call. This may have resulted in the under‐reporting of patients who were hospitalized or had outpatient surgical procedures. Fourth, there may have been a number of characteristics specific to patients in our system that reduced the frequency with which they utilized the AL (eg, access to telephones or other community providers).

Most previous studies of patient‐initiated call lines have included them as part of multi‐intervention pre‐ and/or postdischarge strategies.[7, 8, 9, 10, 11, 12, 13] One prior small study compared the information reported by 37 patients who called an AL with that elicited by nurse‐initiated patient contact.[12] The most frequently reported problems in this study were medication‐related issues (43%). However, this study only included medicine patients and did not document the proportion of calls occurring at various time intervals.

The problems we identified (in both medicine and surgery patients) have previously been described,[2, 3, 4, 13, 14, 15, 16] but all of the studies reporting these problems utilized calls that were initiated by health care providers to patients at various fixed intervals following discharge (ie, 730 days). Most of these used a scripted approach seeking responses to specific questions or outcomes, and the specific timing at which the problems arose was not addressed. In contrast, we examined unsolicited concerns expressed by patients calling an AL following discharge whenever they felt sufficient urgency to address whatever problems or questions arose. We found that a large fraction of calls occurred on the day of or within the first 48 hours following discharge, much earlier than when provider‐initiated calls in the studies cited above occurred. Accordingly, our results cannot be used to compare the utility of patient‐ versus provider‐initiated calls, or to suggest that other hospitals should create an AL system. Rather, we suggest that our findings might be complementary to those reported in studies of provider‐initiated calls and only propose that by examining calls placed by patients to ALs, problems with hospital discharge processes (some of which may result in increased rates of readmission) may be discovered.

The observation that such a large fraction of calls to our AL occurred within the first 48 hours following discharge, together with the fact that many of the questions asked or concerns raised pertained to issues that should have been discussed during the discharge process (eg, pain control, care of surgical wounds), suggests that suboptimal patient education was occurring prior to discharge as was suggested by Henderson and Zernike.[17] This finding has led us to expand our patient education processes prior to discharge on both medicine and surgery services. Because our hospitalists care for approximately 90% of the patients admitted to medicine services and are increasingly involved in the care of patients on surgery services, they are integrally involved with such quality improvement initiatives.

To our knowledge this is the first study in the literature that describes both medicine and surgery patients who call an AL because of problems or questions following hospital discharge, categorizes these problems, determines when the patients called following their discharge, and identifies those who called as being at increased risk for early rehospitalizations and unscheduled urgent or emergent care visits. Given the financial penalties issued to hospitals with high 30‐day readmission rates, these patients may warrant more attention than is customarily available from telephone call lines or during routine outpatient follow‐up. The majority of patients who called our AL had Medicare, Medicaid, or a commercial insurance, and, accordingly, may have been eligible for additional services such as home visits and/or expedited follow‐up appointments.

Our study has a number of limitations. First, it is a single‐center study, so the results might not generalize to other institutions. Second, because the study was performed in a university‐affiliated, public safety‐net hospital, patient characteristics and the rates and types of postdischarge concerns that we observed might differ from those encountered in different types of hospitals and/or from those in nonteaching institutions. We would suggest, however, that the idea of using concerns raised by patients discharged from any type of hospital in calls to ALs may similarly identify problems with that specific hospital's discharge processes. Third, the information collected from the AL came from summaries provided by nurses answering the calls rather than from actual transcripts. This could have resulted in insufficient or incorrect information pertaining to some of the variables assessed in Table 2. The information presented in Table 1, however, was obtained from our data warehouse after matching medical record numbers. Fourth, we could have underestimated the number of patients who had 30‐day rehospitalizations and/or unplanned for urgent or emergent care visits if patients sought care at other hospitals. Fifth, the number of patients calling the AL was too small to allow us to do any type of robust matching or multivariable analysis. Accordingly, the differences that appeared between patients who called and those who did not (ie, English speakers, being medically indigent, the length of stay for the index hospitalization and the discharging service) could be the result of inadequate matching or interactions among the variables. Although matching or multivariate analysis might have yielded different associations between patients who called the AL versus those who did not, those who called the AL still had an increased risk of readmission and urgent or emergent visits and may still benefit from targeted interventions. Finally, the fact that only 1.5% of unique patients who were discharged called the AL could have biased our results. Because only 55% and 53% of the patients who did or did not call the AL, respectively, saw primary care physicians within our system within the 3 years prior to their index hospitalization (P=0.679), the frequency of calls to the AL that we observed could have underestimated the frequency with which patients had contact with other care providers in the community.

In summary, information collected from patient‐initiated calls to our AL identified several aspects of our discharge processes that needed improvement. We concluded that our predischarge educational processes for both medicine and surgery services needed modification, especially with respect to pain management, which problems to expect after hospitalization or surgery, and how to deal with them. The high rates of 30‐day rehospitalization and of unscheduled urgent or emergent care visits among patients calling the AL identifies them as being at increased risk for these outcomes, although the likelihood of these events may be related to factors other than just calling the AL.

The period immediately following hospital discharge is particularly hazardous for patients.[1, 2, 3, 4, 5] Problems occurring after discharge may result in high rates of rehospitalization and unscheduled visits to healthcare providers.[6, 7, 8, 9, 10] Numerous investigators have tried to identify patients who are at increased risk for rehospitalizations within 30 days of discharge, and many studies have examined whether various interventions could decrease these adverse events (summarized in Hansen et al.[11]). An increasing fraction of patients discharged by medicine and surgery services have some or all of their care supervised by hospitalists. Thus, hospitals increasingly look to hospitalists for ways to reduce rehospitalizations.

Patients discharged from our hospital are instructed to call an advice line (AL) if and when questions or concerns arise. Accordingly, we examined when these calls were made and what issues were raised, with the idea that the information collected might identify aspects of our discharge processes that needed improvement.

METHODS

RESULTS

During the 1‐year study period, 19,303 unique patients were discharged home with instructions regarding the use of the AL. A total of 310 patients called the AL and reported being hospitalized or having surgery within the preceding 30 days. Of these, 2 were excluded (1 who was incarcerated and 1 who was discharged to a skilled nursing facility), leaving 308 patients in the cohort. This represented 1.5% of the total number of unduplicated patients discharged during this same time period (minus the exclusions described above). The large majority of the calls (277/308, 90%) came directly from patients. The remaining 10% came from a proxy, usually a patient's family member. Compared with patients who were discharged during the same time period who did not call the AL, those who called were more likely to speak English, less likely to speak Spanish, more likely to be medically indigent, had slightly longer lengths of stays for their index hospitalization, and were more likely to be discharged from surgery than medicine services (particularly following inpatient surgery) (Table 1).

The median time from hospital discharge to the call was 3 days (interquartile range [IQR], 16), but 31% and 47% of calls occurred within 24 or 48 hours of discharge, respectively. Ten percent of patients called the AL the same day of discharge (Figure 1). We found no difference in timing of the calls as a function of discharging service.

Figure 1

The 308 patients reported a total of 612 problems or concerns (meanstandard deviation number of complaints per caller=21), the large majority of which (71%) were symptom‐related (Table 2). The most common symptom was uncontrolled pain, reported by 33% and 40% of patients discharged from medicine and surgery services, respectively. The next most common symptoms related to the gastrointestinal system and to surgical site issues in medicine and surgery patients, respectively (data not shown).

Sixty‐five patients, representing 21% of the cohort, reported 81 problems understanding or executing discharge instructions. No difference was observed between the fraction of these problems reported by patients from medicine versus surgery (19% and 22%, respectively, P=0.54).

Sixty‐five patients, again representing 21% of the cohort, reported 87 medication‐related problems, 20% from both the medicine and surgery services (P=0.99). Medicine patients more frequently reported difficulties understanding their medication instructions, whereas surgery patients more frequently reported lack of efficacy of medications, particularly with respect to pain control (data not shown).

Thirty percent of patients who called the AL were advised by the nurse to go to the emergency department immediately. Medicine patients were more likely to be triaged to the emergency department compared with surgery patients (45% vs 22%, P<0.0001).

The 30‐day readmission rates and the rates of unscheduled urgent or emergent care visits were higher for patients calling the AL compared with those who did not call (46/308, 15% vs 706/18,995, 4%, and 92/308, 30% vs 1303/18,995, 7%, respectively, both P<0.0001). Similar differences were found for patients discharged from medicine or surgery services who called the AL compared with those who did not (data not shown, both P<0.0001). The median number of days between AL call and rehospitalization was 0 (IQR, 02) and 1 (IQR, 08) for medicine and surgery patients, respectively. Ninety‐three percent of rehospitalizations were related to the index hospitalization, and 78% of patients who were readmitted had no outpatient encounter in the interim between discharge and rehospitalization.

DISCUSSION

We investigated the source and nature of patient telephone calls to an AL following a hospitalization or surgery, and our data revealed the following important findings: (1) nearly one‐half of the calls to the AL occurred within the first 48 hours following discharge; (2) the majority of the calls came from surgery patients, and a greater fraction of patients discharged from surgery services called the AL than patients discharged from medicine services; (3) the most common issues were uncontrolled pain, questions about medications, and problems understanding or executing aftercare instructions (particularly pertaining to the care of surgical wounds); and (4) patients calling the AL had higher rates of 30‐day rehospitalization and of unscheduled urgent or emergent care visits.

The utilization of our patient‐initiated call line was only 1.5%, which was on the low end of the 1% to 10% reported in the literature.[7, 12] This can be attributed to a number of issues that are specific to our system. First, the discharge instructions provided to our patients stated that they should call their primary care provider or the AL if they had questions. Accordingly, because approximately 50% of our patients had a primary care provider in our system, some may have preferentially contacted their primary care provider rather than the AL. Second, the instructions stated that the patients should call if they were experiencing the symptoms listed on the instruction sheet, so those with other problems/complaints may not have called. Third, AL personnel identified patients as being in our cohort by asking if they had been discharged or underwent a surgical procedure within 30‐days of their call. This may have resulted in the under‐reporting of patients who were hospitalized or had outpatient surgical procedures. Fourth, there may have been a number of characteristics specific to patients in our system that reduced the frequency with which they utilized the AL (eg, access to telephones or other community providers).

Most previous studies of patient‐initiated call lines have included them as part of multi‐intervention pre‐ and/or postdischarge strategies.[7, 8, 9, 10, 11, 12, 13] One prior small study compared the information reported by 37 patients who called an AL with that elicited by nurse‐initiated patient contact.[12] The most frequently reported problems in this study were medication‐related issues (43%). However, this study only included medicine patients and did not document the proportion of calls occurring at various time intervals.

The problems we identified (in both medicine and surgery patients) have previously been described,[2, 3, 4, 13, 14, 15, 16] but all of the studies reporting these problems utilized calls that were initiated by health care providers to patients at various fixed intervals following discharge (ie, 730 days). Most of these used a scripted approach seeking responses to specific questions or outcomes, and the specific timing at which the problems arose was not addressed. In contrast, we examined unsolicited concerns expressed by patients calling an AL following discharge whenever they felt sufficient urgency to address whatever problems or questions arose. We found that a large fraction of calls occurred on the day of or within the first 48 hours following discharge, much earlier than when provider‐initiated calls in the studies cited above occurred. Accordingly, our results cannot be used to compare the utility of patient‐ versus provider‐initiated calls, or to suggest that other hospitals should create an AL system. Rather, we suggest that our findings might be complementary to those reported in studies of provider‐initiated calls and only propose that by examining calls placed by patients to ALs, problems with hospital discharge processes (some of which may result in increased rates of readmission) may be discovered.

The observation that such a large fraction of calls to our AL occurred within the first 48 hours following discharge, together with the fact that many of the questions asked or concerns raised pertained to issues that should have been discussed during the discharge process (eg, pain control, care of surgical wounds), suggests that suboptimal patient education was occurring prior to discharge as was suggested by Henderson and Zernike.[17] This finding has led us to expand our patient education processes prior to discharge on both medicine and surgery services. Because our hospitalists care for approximately 90% of the patients admitted to medicine services and are increasingly involved in the care of patients on surgery services, they are integrally involved with such quality improvement initiatives.

To our knowledge this is the first study in the literature that describes both medicine and surgery patients who call an AL because of problems or questions following hospital discharge, categorizes these problems, determines when the patients called following their discharge, and identifies those who called as being at increased risk for early rehospitalizations and unscheduled urgent or emergent care visits. Given the financial penalties issued to hospitals with high 30‐day readmission rates, these patients may warrant more attention than is customarily available from telephone call lines or during routine outpatient follow‐up. The majority of patients who called our AL had Medicare, Medicaid, or a commercial insurance, and, accordingly, may have been eligible for additional services such as home visits and/or expedited follow‐up appointments.

Our study has a number of limitations. First, it is a single‐center study, so the results might not generalize to other institutions. Second, because the study was performed in a university‐affiliated, public safety‐net hospital, patient characteristics and the rates and types of postdischarge concerns that we observed might differ from those encountered in different types of hospitals and/or from those in nonteaching institutions. We would suggest, however, that the idea of using concerns raised by patients discharged from any type of hospital in calls to ALs may similarly identify problems with that specific hospital's discharge processes. Third, the information collected from the AL came from summaries provided by nurses answering the calls rather than from actual transcripts. This could have resulted in insufficient or incorrect information pertaining to some of the variables assessed in Table 2. The information presented in Table 1, however, was obtained from our data warehouse after matching medical record numbers. Fourth, we could have underestimated the number of patients who had 30‐day rehospitalizations and/or unplanned for urgent or emergent care visits if patients sought care at other hospitals. Fifth, the number of patients calling the AL was too small to allow us to do any type of robust matching or multivariable analysis. Accordingly, the differences that appeared between patients who called and those who did not (ie, English speakers, being medically indigent, the length of stay for the index hospitalization and the discharging service) could be the result of inadequate matching or interactions among the variables. Although matching or multivariate analysis might have yielded different associations between patients who called the AL versus those who did not, those who called the AL still had an increased risk of readmission and urgent or emergent visits and may still benefit from targeted interventions. Finally, the fact that only 1.5% of unique patients who were discharged called the AL could have biased our results. Because only 55% and 53% of the patients who did or did not call the AL, respectively, saw primary care physicians within our system within the 3 years prior to their index hospitalization (P=0.679), the frequency of calls to the AL that we observed could have underestimated the frequency with which patients had contact with other care providers in the community.

In summary, information collected from patient‐initiated calls to our AL identified several aspects of our discharge processes that needed improvement. We concluded that our predischarge educational processes for both medicine and surgery services needed modification, especially with respect to pain management, which problems to expect after hospitalization or surgery, and how to deal with them. The high rates of 30‐day rehospitalization and of unscheduled urgent or emergent care visits among patients calling the AL identifies them as being at increased risk for these outcomes, although the likelihood of these events may be related to factors other than just calling the AL.

The period immediately following hospital discharge is particularly hazardous for patients.[1, 2, 3, 4, 5] Problems occurring after discharge may result in high rates of rehospitalization and unscheduled visits to healthcare providers.[6, 7, 8, 9, 10] Numerous investigators have tried to identify patients who are at increased risk for rehospitalizations within 30 days of discharge, and many studies have examined whether various interventions could decrease these adverse events (summarized in Hansen et al.[11]). An increasing fraction of patients discharged by medicine and surgery services have some or all of their care supervised by hospitalists. Thus, hospitals increasingly look to hospitalists for ways to reduce rehospitalizations.

Patients discharged from our hospital are instructed to call an advice line (AL) if and when questions or concerns arise. Accordingly, we examined when these calls were made and what issues were raised, with the idea that the information collected might identify aspects of our discharge processes that needed improvement.

METHODS

RESULTS

During the 1‐year study period, 19,303 unique patients were discharged home with instructions regarding the use of the AL. A total of 310 patients called the AL and reported being hospitalized or having surgery within the preceding 30 days. Of these, 2 were excluded (1 who was incarcerated and 1 who was discharged to a skilled nursing facility), leaving 308 patients in the cohort. This represented 1.5% of the total number of unduplicated patients discharged during this same time period (minus the exclusions described above). The large majority of the calls (277/308, 90%) came directly from patients. The remaining 10% came from a proxy, usually a patient's family member. Compared with patients who were discharged during the same time period who did not call the AL, those who called were more likely to speak English, less likely to speak Spanish, more likely to be medically indigent, had slightly longer lengths of stays for their index hospitalization, and were more likely to be discharged from surgery than medicine services (particularly following inpatient surgery) (Table 1).

The median time from hospital discharge to the call was 3 days (interquartile range [IQR], 16), but 31% and 47% of calls occurred within 24 or 48 hours of discharge, respectively. Ten percent of patients called the AL the same day of discharge (Figure 1). We found no difference in timing of the calls as a function of discharging service.

Figure 1

The 308 patients reported a total of 612 problems or concerns (meanstandard deviation number of complaints per caller=21), the large majority of which (71%) were symptom‐related (Table 2). The most common symptom was uncontrolled pain, reported by 33% and 40% of patients discharged from medicine and surgery services, respectively. The next most common symptoms related to the gastrointestinal system and to surgical site issues in medicine and surgery patients, respectively (data not shown).

Sixty‐five patients, representing 21% of the cohort, reported 81 problems understanding or executing discharge instructions. No difference was observed between the fraction of these problems reported by patients from medicine versus surgery (19% and 22%, respectively, P=0.54).

Sixty‐five patients, again representing 21% of the cohort, reported 87 medication‐related problems, 20% from both the medicine and surgery services (P=0.99). Medicine patients more frequently reported difficulties understanding their medication instructions, whereas surgery patients more frequently reported lack of efficacy of medications, particularly with respect to pain control (data not shown).

Thirty percent of patients who called the AL were advised by the nurse to go to the emergency department immediately. Medicine patients were more likely to be triaged to the emergency department compared with surgery patients (45% vs 22%, P<0.0001).

The 30‐day readmission rates and the rates of unscheduled urgent or emergent care visits were higher for patients calling the AL compared with those who did not call (46/308, 15% vs 706/18,995, 4%, and 92/308, 30% vs 1303/18,995, 7%, respectively, both P<0.0001). Similar differences were found for patients discharged from medicine or surgery services who called the AL compared with those who did not (data not shown, both P<0.0001). The median number of days between AL call and rehospitalization was 0 (IQR, 02) and 1 (IQR, 08) for medicine and surgery patients, respectively. Ninety‐three percent of rehospitalizations were related to the index hospitalization, and 78% of patients who were readmitted had no outpatient encounter in the interim between discharge and rehospitalization.

DISCUSSION

We investigated the source and nature of patient telephone calls to an AL following a hospitalization or surgery, and our data revealed the following important findings: (1) nearly one‐half of the calls to the AL occurred within the first 48 hours following discharge; (2) the majority of the calls came from surgery patients, and a greater fraction of patients discharged from surgery services called the AL than patients discharged from medicine services; (3) the most common issues were uncontrolled pain, questions about medications, and problems understanding or executing aftercare instructions (particularly pertaining to the care of surgical wounds); and (4) patients calling the AL had higher rates of 30‐day rehospitalization and of unscheduled urgent or emergent care visits.

The utilization of our patient‐initiated call line was only 1.5%, which was on the low end of the 1% to 10% reported in the literature.[7, 12] This can be attributed to a number of issues that are specific to our system. First, the discharge instructions provided to our patients stated that they should call their primary care provider or the AL if they had questions. Accordingly, because approximately 50% of our patients had a primary care provider in our system, some may have preferentially contacted their primary care provider rather than the AL. Second, the instructions stated that the patients should call if they were experiencing the symptoms listed on the instruction sheet, so those with other problems/complaints may not have called. Third, AL personnel identified patients as being in our cohort by asking if they had been discharged or underwent a surgical procedure within 30‐days of their call. This may have resulted in the under‐reporting of patients who were hospitalized or had outpatient surgical procedures. Fourth, there may have been a number of characteristics specific to patients in our system that reduced the frequency with which they utilized the AL (eg, access to telephones or other community providers).

Most previous studies of patient‐initiated call lines have included them as part of multi‐intervention pre‐ and/or postdischarge strategies.[7, 8, 9, 10, 11, 12, 13] One prior small study compared the information reported by 37 patients who called an AL with that elicited by nurse‐initiated patient contact.[12] The most frequently reported problems in this study were medication‐related issues (43%). However, this study only included medicine patients and did not document the proportion of calls occurring at various time intervals.

The problems we identified (in both medicine and surgery patients) have previously been described,[2, 3, 4, 13, 14, 15, 16] but all of the studies reporting these problems utilized calls that were initiated by health care providers to patients at various fixed intervals following discharge (ie, 730 days). Most of these used a scripted approach seeking responses to specific questions or outcomes, and the specific timing at which the problems arose was not addressed. In contrast, we examined unsolicited concerns expressed by patients calling an AL following discharge whenever they felt sufficient urgency to address whatever problems or questions arose. We found that a large fraction of calls occurred on the day of or within the first 48 hours following discharge, much earlier than when provider‐initiated calls in the studies cited above occurred. Accordingly, our results cannot be used to compare the utility of patient‐ versus provider‐initiated calls, or to suggest that other hospitals should create an AL system. Rather, we suggest that our findings might be complementary to those reported in studies of provider‐initiated calls and only propose that by examining calls placed by patients to ALs, problems with hospital discharge processes (some of which may result in increased rates of readmission) may be discovered.

The observation that such a large fraction of calls to our AL occurred within the first 48 hours following discharge, together with the fact that many of the questions asked or concerns raised pertained to issues that should have been discussed during the discharge process (eg, pain control, care of surgical wounds), suggests that suboptimal patient education was occurring prior to discharge as was suggested by Henderson and Zernike.[17] This finding has led us to expand our patient education processes prior to discharge on both medicine and surgery services. Because our hospitalists care for approximately 90% of the patients admitted to medicine services and are increasingly involved in the care of patients on surgery services, they are integrally involved with such quality improvement initiatives.

To our knowledge this is the first study in the literature that describes both medicine and surgery patients who call an AL because of problems or questions following hospital discharge, categorizes these problems, determines when the patients called following their discharge, and identifies those who called as being at increased risk for early rehospitalizations and unscheduled urgent or emergent care visits. Given the financial penalties issued to hospitals with high 30‐day readmission rates, these patients may warrant more attention than is customarily available from telephone call lines or during routine outpatient follow‐up. The majority of patients who called our AL had Medicare, Medicaid, or a commercial insurance, and, accordingly, may have been eligible for additional services such as home visits and/or expedited follow‐up appointments.

Our study has a number of limitations. First, it is a single‐center study, so the results might not generalize to other institutions. Second, because the study was performed in a university‐affiliated, public safety‐net hospital, patient characteristics and the rates and types of postdischarge concerns that we observed might differ from those encountered in different types of hospitals and/or from those in nonteaching institutions. We would suggest, however, that the idea of using concerns raised by patients discharged from any type of hospital in calls to ALs may similarly identify problems with that specific hospital's discharge processes. Third, the information collected from the AL came from summaries provided by nurses answering the calls rather than from actual transcripts. This could have resulted in insufficient or incorrect information pertaining to some of the variables assessed in Table 2. The information presented in Table 1, however, was obtained from our data warehouse after matching medical record numbers. Fourth, we could have underestimated the number of patients who had 30‐day rehospitalizations and/or unplanned for urgent or emergent care visits if patients sought care at other hospitals. Fifth, the number of patients calling the AL was too small to allow us to do any type of robust matching or multivariable analysis. Accordingly, the differences that appeared between patients who called and those who did not (ie, English speakers, being medically indigent, the length of stay for the index hospitalization and the discharging service) could be the result of inadequate matching or interactions among the variables. Although matching or multivariate analysis might have yielded different associations between patients who called the AL versus those who did not, those who called the AL still had an increased risk of readmission and urgent or emergent visits and may still benefit from targeted interventions. Finally, the fact that only 1.5% of unique patients who were discharged called the AL could have biased our results. Because only 55% and 53% of the patients who did or did not call the AL, respectively, saw primary care physicians within our system within the 3 years prior to their index hospitalization (P=0.679), the frequency of calls to the AL that we observed could have underestimated the frequency with which patients had contact with other care providers in the community.

In summary, information collected from patient‐initiated calls to our AL identified several aspects of our discharge processes that needed improvement. We concluded that our predischarge educational processes for both medicine and surgery services needed modification, especially with respect to pain management, which problems to expect after hospitalization or surgery, and how to deal with them. The high rates of 30‐day rehospitalization and of unscheduled urgent or emergent care visits among patients calling the AL identifies them as being at increased risk for these outcomes, although the likelihood of these events may be related to factors other than just calling the AL.