Mohammed Salim Al‐Damluji, MD, MPH

Hospital medicine is a main component of healthcare in the United States and is growing.[1] In 1995, 9% of inpatient care performed by general internists to Medicare patients was provided by hospitalists; by 2006, this had increased to 37%.[2] The estimated 30,000 practicing hospitalists account for 19% of all practicing general internists[2, 3, 4] and have had a major impact on the treatment of inpatients at US hospitals.[5] Other specialties are adopting the hospital‐based physician model.[6, 7] The hospitalist model does have unique challenges. One notable aspect of hospitalist care, which is frequently shift based, is the transfer of care among providers at shift change.

The Society of Hospital Medicine recognizes patient handoffs/sign‐outs as a core competency for hospitalists,[8] but there is little literature evaluating hospitalist sign‐out quality.[9] A systematic review in 2009 found no studies of hospitalist handoffs.[8] Furthermore, early work suggests that hospitalist handoffs are not consistently effective.[10] In a recent survey, 13% of hospitalists reported they had received an incomplete handoff, and 16% of hospitalists reported at least 1 near‐miss attributed to incomplete communication.[11] Last, hospitalists perform no better than housestaff on evaluations of sign‐out quality.[12]

Cross‐coverage situations, in which sign‐out is key, have been shown to place patients at risk.[13, 14] One study showed 7.1 problems related to sign‐out per 100 patient‐days.[15] Failure during sign‐out can ultimately threaten patient safety.[16] Therefore, evaluating the quality of hospitalist sign‐outs by assessing how well the sign‐out prepares the night team for overnight events is necessary to improve hospitalist sign‐outs and ultimately increase patient safety.

METHODS

Data Collection

Hospitalists who consented were provided a single pocket card during their shift. For every inquiry that involved a patient that the hospitalist was covering, the hospitalist recorded who originated the inquiry, the clinical significance, the sufficiency of written sign‐out, which information was used other than the written sign‐out, and information regarding the anticipation of the event by the daytime team (Figure 1).

Figure 1

Data were collected on 6 days and distributed from April 30, 2012 through June 12, 2012. Dates were chosen based on staffing to maximize the number of eligible physicians each night and included both weekdays and weekend days. The written sign‐out for the entire service was printed for each night data collection took place.

RESULTS

Hospitalists recorded 124 inquiries about 96 patients. Altogether, 15 of 19 (79%) eligible hospitalists returned surveys. Of the 96 patients, we obtained the written sign‐out for 68 (71%). The remainder were new patients for whom the sign‐out had not yet been prepared, or patients who had not yet been assigned to the hospitalist service at the time the sign‐out report was printed.

Hospitalists referenced the sign‐out for 89 (74%) inquiries, and the sign‐out was considered sufficient to respond to 27 (30%) of these inquiries (ie, the sign‐out was adequate to answer the inquiry without any supplemental information). Hospitalists physically saw the patient for 14 (12%) inquiries. Nurses were the originator for most inquiries (102 [82%]). The most common inquiry topics were medications (55 [45%]), plan of care (26 [21%]) and clinical changes (26 [21%]). Ninety‐five (77%) inquiries were considered to be somewhat or very clinically important by the hospitalist (Table 1).

No written sign‐outs had a composite score of 0 or 1; 3 (4%) had a composite score of 2; 31 (46%) had a composite score of 3; and 34 (50%) had a composite score of 4. Seventy‐two percent of written sign‐outs included neither anticipatory guidance nor tasks, 21% had 1 anticipatory guidance item or task, and 6% had 2 or more anticipatory guidance items and/or tasks.

The primary team caring for a patient did not predict 102 (86%) inquiries, and hospitalists rated 47 (46%) of those unpredicted events as possible for the primary team to predict. Five responses to this question were incomplete and excluded. Of the 64 events predicted by the primary team or rated as predictable by the night hospitalists, 17 (27%) were predicted by the primary team (Table 1).

Sign‐out was considered sufficient in isolation to answer the majority of order reconciliation inquiries (5 [71%]), but was less effective at helping to answer inquiries about clinical change (7 [29%]), medications (10 [28%]), and plan of care (5 [24%]) (P=0.001). (Table 2) Ninety‐five events were rated as either very or somewhat clinically important, but this did not affect the likelihood of sign‐out being sufficient in isolation relative to the not at all clinically important group. Specifically, 33% of sign‐outs were rated sufficient in the very important group, 19% in the somewhat important group, and 50% in the not at all group (P=0.059).

Sign‐out was considered sufficient in isolation more frequently for inquiries about patients admitted <2 days prior to data collection than for inquiries about patients admitted more than 2 days prior to data collection (21 [40%] vs 6 [16%], respectively) (P=0.015) (Table 2).

Sign‐outs with 2 or more anticipatory guidance items were considered sufficient in isolation more often than sign‐outs with 1 or fewer anticipatory guidance item (60% for 2 or more, 14% for 1, 34% for 0; P=0.006) (Table 2). The composite score was grouped into 2 categoriesscore <4 and score=4with no statistical difference in sign‐out sufficiency between the 2 groups (P=0.22) (Table 2).

In multivariable analysis, no predictor variable was significantly associated with sufficient sign‐out (Table 3).

DISCUSSION

In this study of written sign‐out among hospitalists and physician‐extenders on a hospitalist service, we found that the sign‐out was used to answer three‐quarters of overnight inquiries, despite the advanced level of training (completion of all postgraduate medical education) of the covering clinicians and the presence of a robust EHR. The effectiveness of the written sign‐out, however, was not as consistently high as its use. Overall, the sign‐out was sufficient to answer less than a third of inquiries in which it was referenced. Thus, although most studies of sign‐out quality have focused on trainees, our results make it clear that hospitalists also rely on sign‐out, and its effectiveness can be improved.

There are few studies of attending‐level sign‐outs. Hinami et al. found that nearly 1 in 5 hospitalists was uncertain of the care plan after assuming care of a new set of patients, despite having received a handoff from the departing hospitalist.[11] Handoffs between emergency physicians and hospitalists have repeatedly been noted to have content omissions and to contribute to adverse events.[7, 12, 21, 22] Ilan et al. videotaped attending handoffs in the intensive care unit and found that they did not follow any of 3 commonly recommended structures; however, this study did not assess the effectiveness of the handoffs.[23] Williams et al. found that the transfer of patient information among surgical team members, including attending surgeons, was suboptimal, and these problems were commonly related to decreased surgeon familiarity with a particular patient, a theme common to hospital medicine, and a contributor to adverse events and decreased efficiency.[24]

This study extends the literature in several ways. By studying overnight events, we generate a comprehensive view of the information sources hospitalists use to care for patients overnight. Interestingly, our results were similar to the overnight information‐gathering habits of trainees in a study of pediatric trainees.[25] Furthermore, by linking each inquiry to the accompanying written sign‐out, we are able to analyze which characteristics of a written sign‐out are associated with sign‐out effectiveness, and we are able to describe the utility of written sign‐out to answer different types of clinical scenarios.

Our data show that hospitalists rely heavily on written sign‐out to care for patients overnight, with the physician note being the most‐utilized secondary reference used by covering physicians. The written sign‐out was most useful for order clarification compared to other topics, and the patient was only seen for 12% of inquiries. Most notable, however, was the suggestion that sign‐outs with more anticipatory guidance were more likely to be effective for overnight care, as were sign‐outs created earlier in the hospital course. Future efforts to improve the utility of the written sign‐out might focus on these items, whether through training or audit/feedback.

The use of electronic handoff tools has been shown to increase the ease of use, efficiency, and perceptions of patient safety and quality in several studies.[3, 26, 27] This study relied on an electronic tool as the only means of information transfer during sign‐out. Without the confounding effect of verbal information transfer, we are better able to understand the efficacy of the written component alone. Nonetheless, most expert opinion statements as well as The Joint Commission include a recommendation for verbal and written components to handoff communication.[8, 20, 28, 29, 30] It is possible that sign‐outs would more often have been rated sufficient if the handoff process had reliably included verbal handoff. Future studies are warranted to compare written‐only to written‐plus‐verbal sign‐out, to determine the added benefit of verbal communication. With a robust EHR, it is also an open question whether sign‐out needs to be sufficient to answer overnight inquiries or whether it would be acceptable or even preferable to have overnight staff consistently review the EHR directly, especially as the physician notes are the most common nonsign‐out reference used. Nonetheless, the fact that hospitalists rely heavily on written sign‐out despite the availability of other information sources suggests that hospitalists find specific benefit in written sign‐out.

Limitations of this study include the relatively small sample size, the limited collection time period, and the single‐site nature. The YNHH Hospitalist Service uses only written documents to sign out, so the external validity to programs that use verbal sign‐out is limited. The written‐only nature, however, removes the variable of the discussion at time of sign‐out, improving the purity of the written sign‐out assessment. We did not assess workload, which might have affected sign‐out quality. The interpretation of the composite score is limited, due to little variation in scoring in our sample, as well as lack of validation in other studies. An additional limitation is that sign‐outs are not entirely drafted by the hospitalist attendings. Hospitalists draft the initial sign‐out document, but it is updated on subsequent days by the mid‐level provider under the direction of the hospitalist attending. It is therefore possible that sign‐outs maintained directly by hospitalists would have been of different quality. In this regard it is interesting to note that in a different study of verbal sign‐out we were not able to detect a difference in quality among hospitalists, trainees, and midlevels.[12] Last, hindsight bias may be present, as the covering physician's perspective of the event includes more information than available to the provider creating the sign‐out document.

Overall, we found that attending hospitalists rely heavily on written sign‐out documents to address overnight inquiries, but those sign‐outs are not reliably effective. Future work to better understand the roles of written and verbal components in sign‐out is needed to help improve the safety of overnight care.

Hospital medicine is a main component of healthcare in the United States and is growing.[1] In 1995, 9% of inpatient care performed by general internists to Medicare patients was provided by hospitalists; by 2006, this had increased to 37%.[2] The estimated 30,000 practicing hospitalists account for 19% of all practicing general internists[2, 3, 4] and have had a major impact on the treatment of inpatients at US hospitals.[5] Other specialties are adopting the hospital‐based physician model.[6, 7] The hospitalist model does have unique challenges. One notable aspect of hospitalist care, which is frequently shift based, is the transfer of care among providers at shift change.

The Society of Hospital Medicine recognizes patient handoffs/sign‐outs as a core competency for hospitalists,[8] but there is little literature evaluating hospitalist sign‐out quality.[9] A systematic review in 2009 found no studies of hospitalist handoffs.[8] Furthermore, early work suggests that hospitalist handoffs are not consistently effective.[10] In a recent survey, 13% of hospitalists reported they had received an incomplete handoff, and 16% of hospitalists reported at least 1 near‐miss attributed to incomplete communication.[11] Last, hospitalists perform no better than housestaff on evaluations of sign‐out quality.[12]

Cross‐coverage situations, in which sign‐out is key, have been shown to place patients at risk.[13, 14] One study showed 7.1 problems related to sign‐out per 100 patient‐days.[15] Failure during sign‐out can ultimately threaten patient safety.[16] Therefore, evaluating the quality of hospitalist sign‐outs by assessing how well the sign‐out prepares the night team for overnight events is necessary to improve hospitalist sign‐outs and ultimately increase patient safety.

METHODS

Data Collection

Hospitalists who consented were provided a single pocket card during their shift. For every inquiry that involved a patient that the hospitalist was covering, the hospitalist recorded who originated the inquiry, the clinical significance, the sufficiency of written sign‐out, which information was used other than the written sign‐out, and information regarding the anticipation of the event by the daytime team (Figure 1).

Figure 1

Data were collected on 6 days and distributed from April 30, 2012 through June 12, 2012. Dates were chosen based on staffing to maximize the number of eligible physicians each night and included both weekdays and weekend days. The written sign‐out for the entire service was printed for each night data collection took place.

RESULTS

Hospitalists recorded 124 inquiries about 96 patients. Altogether, 15 of 19 (79%) eligible hospitalists returned surveys. Of the 96 patients, we obtained the written sign‐out for 68 (71%). The remainder were new patients for whom the sign‐out had not yet been prepared, or patients who had not yet been assigned to the hospitalist service at the time the sign‐out report was printed.

Hospitalists referenced the sign‐out for 89 (74%) inquiries, and the sign‐out was considered sufficient to respond to 27 (30%) of these inquiries (ie, the sign‐out was adequate to answer the inquiry without any supplemental information). Hospitalists physically saw the patient for 14 (12%) inquiries. Nurses were the originator for most inquiries (102 [82%]). The most common inquiry topics were medications (55 [45%]), plan of care (26 [21%]) and clinical changes (26 [21%]). Ninety‐five (77%) inquiries were considered to be somewhat or very clinically important by the hospitalist (Table 1).

No written sign‐outs had a composite score of 0 or 1; 3 (4%) had a composite score of 2; 31 (46%) had a composite score of 3; and 34 (50%) had a composite score of 4. Seventy‐two percent of written sign‐outs included neither anticipatory guidance nor tasks, 21% had 1 anticipatory guidance item or task, and 6% had 2 or more anticipatory guidance items and/or tasks.

The primary team caring for a patient did not predict 102 (86%) inquiries, and hospitalists rated 47 (46%) of those unpredicted events as possible for the primary team to predict. Five responses to this question were incomplete and excluded. Of the 64 events predicted by the primary team or rated as predictable by the night hospitalists, 17 (27%) were predicted by the primary team (Table 1).

Sign‐out was considered sufficient in isolation to answer the majority of order reconciliation inquiries (5 [71%]), but was less effective at helping to answer inquiries about clinical change (7 [29%]), medications (10 [28%]), and plan of care (5 [24%]) (P=0.001). (Table 2) Ninety‐five events were rated as either very or somewhat clinically important, but this did not affect the likelihood of sign‐out being sufficient in isolation relative to the not at all clinically important group. Specifically, 33% of sign‐outs were rated sufficient in the very important group, 19% in the somewhat important group, and 50% in the not at all group (P=0.059).

Sign‐out was considered sufficient in isolation more frequently for inquiries about patients admitted <2 days prior to data collection than for inquiries about patients admitted more than 2 days prior to data collection (21 [40%] vs 6 [16%], respectively) (P=0.015) (Table 2).

Sign‐outs with 2 or more anticipatory guidance items were considered sufficient in isolation more often than sign‐outs with 1 or fewer anticipatory guidance item (60% for 2 or more, 14% for 1, 34% for 0; P=0.006) (Table 2). The composite score was grouped into 2 categoriesscore <4 and score=4with no statistical difference in sign‐out sufficiency between the 2 groups (P=0.22) (Table 2).

In multivariable analysis, no predictor variable was significantly associated with sufficient sign‐out (Table 3).

DISCUSSION

In this study of written sign‐out among hospitalists and physician‐extenders on a hospitalist service, we found that the sign‐out was used to answer three‐quarters of overnight inquiries, despite the advanced level of training (completion of all postgraduate medical education) of the covering clinicians and the presence of a robust EHR. The effectiveness of the written sign‐out, however, was not as consistently high as its use. Overall, the sign‐out was sufficient to answer less than a third of inquiries in which it was referenced. Thus, although most studies of sign‐out quality have focused on trainees, our results make it clear that hospitalists also rely on sign‐out, and its effectiveness can be improved.

There are few studies of attending‐level sign‐outs. Hinami et al. found that nearly 1 in 5 hospitalists was uncertain of the care plan after assuming care of a new set of patients, despite having received a handoff from the departing hospitalist.[11] Handoffs between emergency physicians and hospitalists have repeatedly been noted to have content omissions and to contribute to adverse events.[7, 12, 21, 22] Ilan et al. videotaped attending handoffs in the intensive care unit and found that they did not follow any of 3 commonly recommended structures; however, this study did not assess the effectiveness of the handoffs.[23] Williams et al. found that the transfer of patient information among surgical team members, including attending surgeons, was suboptimal, and these problems were commonly related to decreased surgeon familiarity with a particular patient, a theme common to hospital medicine, and a contributor to adverse events and decreased efficiency.[24]

This study extends the literature in several ways. By studying overnight events, we generate a comprehensive view of the information sources hospitalists use to care for patients overnight. Interestingly, our results were similar to the overnight information‐gathering habits of trainees in a study of pediatric trainees.[25] Furthermore, by linking each inquiry to the accompanying written sign‐out, we are able to analyze which characteristics of a written sign‐out are associated with sign‐out effectiveness, and we are able to describe the utility of written sign‐out to answer different types of clinical scenarios.

Our data show that hospitalists rely heavily on written sign‐out to care for patients overnight, with the physician note being the most‐utilized secondary reference used by covering physicians. The written sign‐out was most useful for order clarification compared to other topics, and the patient was only seen for 12% of inquiries. Most notable, however, was the suggestion that sign‐outs with more anticipatory guidance were more likely to be effective for overnight care, as were sign‐outs created earlier in the hospital course. Future efforts to improve the utility of the written sign‐out might focus on these items, whether through training or audit/feedback.

The use of electronic handoff tools has been shown to increase the ease of use, efficiency, and perceptions of patient safety and quality in several studies.[3, 26, 27] This study relied on an electronic tool as the only means of information transfer during sign‐out. Without the confounding effect of verbal information transfer, we are better able to understand the efficacy of the written component alone. Nonetheless, most expert opinion statements as well as The Joint Commission include a recommendation for verbal and written components to handoff communication.[8, 20, 28, 29, 30] It is possible that sign‐outs would more often have been rated sufficient if the handoff process had reliably included verbal handoff. Future studies are warranted to compare written‐only to written‐plus‐verbal sign‐out, to determine the added benefit of verbal communication. With a robust EHR, it is also an open question whether sign‐out needs to be sufficient to answer overnight inquiries or whether it would be acceptable or even preferable to have overnight staff consistently review the EHR directly, especially as the physician notes are the most common nonsign‐out reference used. Nonetheless, the fact that hospitalists rely heavily on written sign‐out despite the availability of other information sources suggests that hospitalists find specific benefit in written sign‐out.

Limitations of this study include the relatively small sample size, the limited collection time period, and the single‐site nature. The YNHH Hospitalist Service uses only written documents to sign out, so the external validity to programs that use verbal sign‐out is limited. The written‐only nature, however, removes the variable of the discussion at time of sign‐out, improving the purity of the written sign‐out assessment. We did not assess workload, which might have affected sign‐out quality. The interpretation of the composite score is limited, due to little variation in scoring in our sample, as well as lack of validation in other studies. An additional limitation is that sign‐outs are not entirely drafted by the hospitalist attendings. Hospitalists draft the initial sign‐out document, but it is updated on subsequent days by the mid‐level provider under the direction of the hospitalist attending. It is therefore possible that sign‐outs maintained directly by hospitalists would have been of different quality. In this regard it is interesting to note that in a different study of verbal sign‐out we were not able to detect a difference in quality among hospitalists, trainees, and midlevels.[12] Last, hindsight bias may be present, as the covering physician's perspective of the event includes more information than available to the provider creating the sign‐out document.

Overall, we found that attending hospitalists rely heavily on written sign‐out documents to address overnight inquiries, but those sign‐outs are not reliably effective. Future work to better understand the roles of written and verbal components in sign‐out is needed to help improve the safety of overnight care.

Hospital medicine is a main component of healthcare in the United States and is growing.[1] In 1995, 9% of inpatient care performed by general internists to Medicare patients was provided by hospitalists; by 2006, this had increased to 37%.[2] The estimated 30,000 practicing hospitalists account for 19% of all practicing general internists[2, 3, 4] and have had a major impact on the treatment of inpatients at US hospitals.[5] Other specialties are adopting the hospital‐based physician model.[6, 7] The hospitalist model does have unique challenges. One notable aspect of hospitalist care, which is frequently shift based, is the transfer of care among providers at shift change.

The Society of Hospital Medicine recognizes patient handoffs/sign‐outs as a core competency for hospitalists,[8] but there is little literature evaluating hospitalist sign‐out quality.[9] A systematic review in 2009 found no studies of hospitalist handoffs.[8] Furthermore, early work suggests that hospitalist handoffs are not consistently effective.[10] In a recent survey, 13% of hospitalists reported they had received an incomplete handoff, and 16% of hospitalists reported at least 1 near‐miss attributed to incomplete communication.[11] Last, hospitalists perform no better than housestaff on evaluations of sign‐out quality.[12]

Cross‐coverage situations, in which sign‐out is key, have been shown to place patients at risk.[13, 14] One study showed 7.1 problems related to sign‐out per 100 patient‐days.[15] Failure during sign‐out can ultimately threaten patient safety.[16] Therefore, evaluating the quality of hospitalist sign‐outs by assessing how well the sign‐out prepares the night team for overnight events is necessary to improve hospitalist sign‐outs and ultimately increase patient safety.

METHODS

Data Collection

Hospitalists who consented were provided a single pocket card during their shift. For every inquiry that involved a patient that the hospitalist was covering, the hospitalist recorded who originated the inquiry, the clinical significance, the sufficiency of written sign‐out, which information was used other than the written sign‐out, and information regarding the anticipation of the event by the daytime team (Figure 1).

Figure 1

Data were collected on 6 days and distributed from April 30, 2012 through June 12, 2012. Dates were chosen based on staffing to maximize the number of eligible physicians each night and included both weekdays and weekend days. The written sign‐out for the entire service was printed for each night data collection took place.

RESULTS

Hospitalists recorded 124 inquiries about 96 patients. Altogether, 15 of 19 (79%) eligible hospitalists returned surveys. Of the 96 patients, we obtained the written sign‐out for 68 (71%). The remainder were new patients for whom the sign‐out had not yet been prepared, or patients who had not yet been assigned to the hospitalist service at the time the sign‐out report was printed.

Hospitalists referenced the sign‐out for 89 (74%) inquiries, and the sign‐out was considered sufficient to respond to 27 (30%) of these inquiries (ie, the sign‐out was adequate to answer the inquiry without any supplemental information). Hospitalists physically saw the patient for 14 (12%) inquiries. Nurses were the originator for most inquiries (102 [82%]). The most common inquiry topics were medications (55 [45%]), plan of care (26 [21%]) and clinical changes (26 [21%]). Ninety‐five (77%) inquiries were considered to be somewhat or very clinically important by the hospitalist (Table 1).

No written sign‐outs had a composite score of 0 or 1; 3 (4%) had a composite score of 2; 31 (46%) had a composite score of 3; and 34 (50%) had a composite score of 4. Seventy‐two percent of written sign‐outs included neither anticipatory guidance nor tasks, 21% had 1 anticipatory guidance item or task, and 6% had 2 or more anticipatory guidance items and/or tasks.

The primary team caring for a patient did not predict 102 (86%) inquiries, and hospitalists rated 47 (46%) of those unpredicted events as possible for the primary team to predict. Five responses to this question were incomplete and excluded. Of the 64 events predicted by the primary team or rated as predictable by the night hospitalists, 17 (27%) were predicted by the primary team (Table 1).

Sign‐out was considered sufficient in isolation to answer the majority of order reconciliation inquiries (5 [71%]), but was less effective at helping to answer inquiries about clinical change (7 [29%]), medications (10 [28%]), and plan of care (5 [24%]) (P=0.001). (Table 2) Ninety‐five events were rated as either very or somewhat clinically important, but this did not affect the likelihood of sign‐out being sufficient in isolation relative to the not at all clinically important group. Specifically, 33% of sign‐outs were rated sufficient in the very important group, 19% in the somewhat important group, and 50% in the not at all group (P=0.059).

Sign‐out was considered sufficient in isolation more frequently for inquiries about patients admitted <2 days prior to data collection than for inquiries about patients admitted more than 2 days prior to data collection (21 [40%] vs 6 [16%], respectively) (P=0.015) (Table 2).

Sign‐outs with 2 or more anticipatory guidance items were considered sufficient in isolation more often than sign‐outs with 1 or fewer anticipatory guidance item (60% for 2 or more, 14% for 1, 34% for 0; P=0.006) (Table 2). The composite score was grouped into 2 categoriesscore <4 and score=4with no statistical difference in sign‐out sufficiency between the 2 groups (P=0.22) (Table 2).

In multivariable analysis, no predictor variable was significantly associated with sufficient sign‐out (Table 3).

DISCUSSION

In this study of written sign‐out among hospitalists and physician‐extenders on a hospitalist service, we found that the sign‐out was used to answer three‐quarters of overnight inquiries, despite the advanced level of training (completion of all postgraduate medical education) of the covering clinicians and the presence of a robust EHR. The effectiveness of the written sign‐out, however, was not as consistently high as its use. Overall, the sign‐out was sufficient to answer less than a third of inquiries in which it was referenced. Thus, although most studies of sign‐out quality have focused on trainees, our results make it clear that hospitalists also rely on sign‐out, and its effectiveness can be improved.

There are few studies of attending‐level sign‐outs. Hinami et al. found that nearly 1 in 5 hospitalists was uncertain of the care plan after assuming care of a new set of patients, despite having received a handoff from the departing hospitalist.[11] Handoffs between emergency physicians and hospitalists have repeatedly been noted to have content omissions and to contribute to adverse events.[7, 12, 21, 22] Ilan et al. videotaped attending handoffs in the intensive care unit and found that they did not follow any of 3 commonly recommended structures; however, this study did not assess the effectiveness of the handoffs.[23] Williams et al. found that the transfer of patient information among surgical team members, including attending surgeons, was suboptimal, and these problems were commonly related to decreased surgeon familiarity with a particular patient, a theme common to hospital medicine, and a contributor to adverse events and decreased efficiency.[24]

This study extends the literature in several ways. By studying overnight events, we generate a comprehensive view of the information sources hospitalists use to care for patients overnight. Interestingly, our results were similar to the overnight information‐gathering habits of trainees in a study of pediatric trainees.[25] Furthermore, by linking each inquiry to the accompanying written sign‐out, we are able to analyze which characteristics of a written sign‐out are associated with sign‐out effectiveness, and we are able to describe the utility of written sign‐out to answer different types of clinical scenarios.

Our data show that hospitalists rely heavily on written sign‐out to care for patients overnight, with the physician note being the most‐utilized secondary reference used by covering physicians. The written sign‐out was most useful for order clarification compared to other topics, and the patient was only seen for 12% of inquiries. Most notable, however, was the suggestion that sign‐outs with more anticipatory guidance were more likely to be effective for overnight care, as were sign‐outs created earlier in the hospital course. Future efforts to improve the utility of the written sign‐out might focus on these items, whether through training or audit/feedback.

The use of electronic handoff tools has been shown to increase the ease of use, efficiency, and perceptions of patient safety and quality in several studies.[3, 26, 27] This study relied on an electronic tool as the only means of information transfer during sign‐out. Without the confounding effect of verbal information transfer, we are better able to understand the efficacy of the written component alone. Nonetheless, most expert opinion statements as well as The Joint Commission include a recommendation for verbal and written components to handoff communication.[8, 20, 28, 29, 30] It is possible that sign‐outs would more often have been rated sufficient if the handoff process had reliably included verbal handoff. Future studies are warranted to compare written‐only to written‐plus‐verbal sign‐out, to determine the added benefit of verbal communication. With a robust EHR, it is also an open question whether sign‐out needs to be sufficient to answer overnight inquiries or whether it would be acceptable or even preferable to have overnight staff consistently review the EHR directly, especially as the physician notes are the most common nonsign‐out reference used. Nonetheless, the fact that hospitalists rely heavily on written sign‐out despite the availability of other information sources suggests that hospitalists find specific benefit in written sign‐out.

Limitations of this study include the relatively small sample size, the limited collection time period, and the single‐site nature. The YNHH Hospitalist Service uses only written documents to sign out, so the external validity to programs that use verbal sign‐out is limited. The written‐only nature, however, removes the variable of the discussion at time of sign‐out, improving the purity of the written sign‐out assessment. We did not assess workload, which might have affected sign‐out quality. The interpretation of the composite score is limited, due to little variation in scoring in our sample, as well as lack of validation in other studies. An additional limitation is that sign‐outs are not entirely drafted by the hospitalist attendings. Hospitalists draft the initial sign‐out document, but it is updated on subsequent days by the mid‐level provider under the direction of the hospitalist attending. It is therefore possible that sign‐outs maintained directly by hospitalists would have been of different quality. In this regard it is interesting to note that in a different study of verbal sign‐out we were not able to detect a difference in quality among hospitalists, trainees, and midlevels.[12] Last, hindsight bias may be present, as the covering physician's perspective of the event includes more information than available to the provider creating the sign‐out document.

Overall, we found that attending hospitalists rely heavily on written sign‐out documents to address overnight inquiries, but those sign‐outs are not reliably effective. Future work to better understand the roles of written and verbal components in sign‐out is needed to help improve the safety of overnight care.