Article Text

Download PDFPDF

Controlled trial to improve resident sign-out in a medical intensive care unit
  1. Rahul Nanchal1,
  2. Brian Aebly1,
  3. Gabrielle Graves1,
  4. Jonathon Truwit1,2,
  5. Gagan Kumar3,
  6. Amit Taneja1,
  7. Gaurav Dagar1,
  8. Jeanette Graf1,
  9. Erin Hubertz1,
  10. Vijaya Ramalingam1,
  11. Kathlyn E Fletcher1,4
  1. 1 Department of Medicine, Medical College of Wisconsin, Milwaukee, Wisconsin, USA
  2. 2 Froedtert Health, Milwaukee, Wisconsin, USA
  3. 3 Department of Critical Care, Phoebe Putney Health System, Northeast Georgia Health System Inc, Albany, Georgia, USA
  4. 4 Department of Internal Medicine, Clement J Zablocki VAMC, Milwaukee, Wisconsin, USA
  1. Correspondence to Dr Rahul Nanchal, Medical College of Wisconsin, Department of medicine, 9200 West Wisconsin Ave, Suite E 5200, Milwaukee, Wisconsin, USA; rnanchal{at}


Objective Poor sign-out or handover of care may lead to preventable patient harm. Critically ill patients in intensive care units (ICU) are complex and prone to rapid clinical deterioration. If clinical deterioration occurs, timeliness of appropriate interventions is essential to prevent or reduce adverse outcomes. Therefore sign-outs need to efficiently transmit key information and provide anticipatory guidance. Interventions to improve resident-to-resident ICU sign-outs have not been well described. We conducted a controlled trial to test the effectiveness of a standardised ICU sign-out process to the usual ICU sign-out.

Design Prospective controlled trial.

Setting A 26-bed medical intensive care unit (MICU) in an urban tertiary academic medical centre.

Subjects Residents rotating through the MICU.

Interventions ICU-specific written sign-out template.

Methods Residents completed postcall surveys assessing satisfaction with verbal and written sign-outs and incidence of non-routine events. Our main outcome of interest was the occurrence of non-routine events.

Main results Compared with the intervention group, on significantly more nights, night float residents in the control group encountered patients who were sicker than sign-out would have suggested (15.94% vs 43.75%; p<0.0001). On significantly fewer nights, night float residents in the intervention group indicated that either something happened to patients that was unexpected (18.84% vs 36.51%; p=0.023) or they were insufficiently prepared for (4.35% vs 35.94%; p<0.0001). Similarly, on fewer nights, residents in the intervention group indicated that they had to perform interventions that were unplanned or unanticipated (15.9% vs 37.7%; p=0.005).

Conclusion A structured sign-out process compared with usual sign-out significantly reduced the occurrence of non-routine events in an academic MICU.

  • handoffs
  • non-routine events
  • critical care
  • communication
  • patient safety
  • medical errors
  • preventable adverse events

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Sign-out (handover of care) is the process of transferring information, responsibility and authority between health professionals, either between services or within a service.1–3 This process leaves the patient vulnerable to the occurrence of adverse events and preventable harm. Following the institution of duty-hour restrictions, resident handoffs have been well studied; however, there is little scientific literature in the critical care setting within a given service.4–9 Miscommunications during handoffs are a leading cause of medical error, and miscommunications in general are associated with two-thirds of all sentinel events — the most serious type of adverse events reported to the Joint Commission.10–17 Restricted resident duty hours make shift work unavoidable and increase the number of handoffs between residents. One study demonstrated that these restrictions resulted in an average of 75 handoffs over a 5-day hospitalisation. During a month-long rotation, there were approximately 300 handoffs per first-year resident, a 40% increase since the duty hours were instituted.12

In university or community academic settings, the residents are ubiquitous in intensive care units (ICUs). Many of these ICUs are not staffed by night-time intensivists and cross coverage is provided by residents. One randomised trial found that night-time in-hospital intensivist staffing did not result in better outcomes; however, medical errors were not reported.18 The resident handoff process in ICUs is particularly challenging given the complexity of ICU patients, the plethora of available data, time constraints in which to transmit these data to the next provider and heightened time sensitivity of intervening for patient deterioration. It is therefore imperative that handoffs in ICUs are efficient and succinctly maximise the quality of transmissible information with the goal of reducing preventable adverse events.

The Joint Commission in 2006 made standardisation of handoff procedures a National Patient Safety Goal, emphasising the importance of this transition process.19 Since then, multiple studies evaluating the handoff process show extensive variability in the content and quality of written sign-outs. Although data on standardisation of the handoff process and assessing its effectiveness are emerging, there are no published reports describing interventions to improve ICU resident-to-resident handoffs.20–23

Local problem

In the medical intensive care unit (MICU) at our institution, there are at least three handoffs for each patient in a 24-hour period. Our leadership identified that the handoff document was lengthy, redundant not standardised, and in general lacked a synopsis of key information that could be transmitted efficiently. Examples of gaps in key information included vital information for patients in an ICU setting, such as mental status prior to and on admission, volume status and trajectory of illness.

We therefore conducted a controlled trial to test the effectiveness of a standardised ICU handoff process to the usual MICU handoff. We hypothesised that the standardised process as compared with the usual process would reduce the occurrence of non-routine events (NREs) (please see Methods section for definition). Our goal was to improve the quality of the current handoff process and therefore to enhance patient safety.


Scope and study population

The institutional review board and the graduate medical education office of our institution approved this study and required resident anonymity. All participants gave informed consent.

Setting and participants

The MICU is a single 26-bed unit within a 500-bed hospital. MICU service comprises two teams (MICU 1 and MICU 2 without geographical localisation within the unit), each staffed by one faculty physician, one critical care fellow and five residents. Fellows provide 24/7 coverage working in 12-hour shifts — there is one fellow per team during the day and an independent fellow rotates in a 2-week block rotation for night-time coverage. Faculty physicians take home call. Once assigned, faculty physicians and daytime fellows stay on the same team for a block rotation. Night-time resident coverage is provided via an internal night float system — on a rotating basis one resident from each team takes night call (20:00 to 08:00) for 6 days. In addition, one resident from each team is designated as the late-call resident and stays from 16:00 to 20:00 every fourth day. Therefore, there are three resident handoffs in a 24-hour period — first from the primary resident to the late-call resident, then from the late-call resident to the night call, and then from the night call to the primary resident. The residents are the primary responders to all emergencies, acute deterioration episodes and nursing queries. The resident composition of each team is the same (two internal medicine interns and three internal medicine upper-level residents). Once assigned to a particular team, the residents stay on the team for the entire rotation and do not cross over. Additionally, residents rotate through the ICU only once in an academic year; the duration of our study did not cross academic years and thus there was no crossover of residents to the other team during the later months of the study. Admissions are alternated between the two teams, and thus on average the acuity of illness, demographics and volume on each team tend to be similarly distributed. The residents are the people for all patient-related issues and therefore we only included them in our study.

Usual handoff process

The MICU uses an electronic health record (EHR)-based form for handoffs. The form is autopopulated with patient name, unit/room, medical record number, age, gender, allergies, chief complaint and code status. Additionally, there is a column for free text in which the residents are encouraged to type information in the SBAR (situation, background, assessment, recommendations) format; however, there is wide variation in the quantity and context of entered information. The residents update this information on a daily basis. Handoffs include both face-to-face conversation and written documentation. Handoffs occur in the MICU resident team room.


Based on our previous work,24 we created an ICU-specific written handoff template that included the following fields: (1) patient acuity, (2) mental status, (3) volume status, (4) active problems, (5) plan of care, (6) follow-up tasks and (7) anticipated problems with guidance. This template was placed in our EHR. When used, the resident was prompted to complete all seven fields.


We randomised MICU 1 to the new template and MICU 2 to the standard method of handoff. On the first day of each rotation, the residents were introduced to the study and educated about the handoff template. After a period of open dialogue and questions and answers, we obtained informed consent from the residents.

Postcall survey

The night float residents were asked to fill out two surveys at the end of each overnight shift. One survey addressed the issue of NRE and the other assessed the quality of the sign-outs using a Likert scale as well as some questions with yes/no answers (see online supplementary file 1).

Supplementary file 1

supp data

The questionnaire/survey was developed by the authors and was based on a conceptual method of handoffs of care developed through an institutional quality improvement project.24 The NRE questionnaire has been validated in a previous study as a tool to measure patient safety.25

Study details were posted in the team room along with instructions on completing the surveys. Research coordinators would visit the teams daily, answer questions and remind residents to use the assigned sign-out document and complete surveys. Completed surveys were placed in a lock box by the residents and collected by the research coordinators at the end of the rotation.


Our main outcome of interest was the occurrence of NREs. An NRE is any event that deviated from optimal or expected care for a specific patient in a specific clinical situation.26 27 We used the NRE questionnaire to compare the frequency with which they occurred in both groups. Our secondary outcome measure was improvement in the quality of sign-out in the intervention group.

Data analysis

We compared dichotomous outcomes using Pearson’s χ2 test. For Likert-type questions (ordinal data) that measured the degree of agreement or disagreement with a particular statement, we describe our data using percentage responses in each category and medians because of the non-normal distribution of data. We used the Mann-Whitney U test to compare differences between groups for each Likert-type question.


From 1 October 2014 to 31 May 2015, postcall residents from both the teams completed a survey from Tuesday morning to Saturday morning. We collected our data starting on the first day of the month until the morning of the 28th due to the staggered end dates of resident rotations relative to their academic track. One hundred and forty-two of the 306 (46.4%) potential surveys were collected — the response rate of team 1 was 45%, while of team 2 was 48%. Some of these surveys had incomplete information but included surveys had at least one answered question.

Non-routine events

Table 1 shows the frequency of events in the NRE survey. The intervention night float residents reported fewer nights with events as compared with the control group. As compared with the intervention group, night float residents in the control group reported significantly more nights when they encountered patients who were sicker than expected (15.94% vs 43.75%; p<0.0001). Night float residents in the intervention group also indicated on fewer nights that either something happened to patients that was unexpected (18.84% vs 36.51%; p=0.023) or they were insufficiently prepared for (4.35% vs 35.94%; p<0.0001). Similarly, on fewer nights in the intervention group, residents had to perform interventions that were unplanned or unanticipated (15.9% vs 37.7%; p=0.005). Responses on other questions on the NRE survey were similar between the two groups. Although the frequency of unanticipated worsening of cardiac, pulmonary or other organ system status was numerically higher in the control group, statistical significance was not reached on any of the three questions. 

Table 1

Frequency of non-routine events

Quality of sign-out

Table 2 shows differences in quality ratings of sign-out between the two groups for binary outcome measures. In the intervention group night float residents in the control group indicated on more nights that patient acuity ratings were accurate (95.15% vs 62.07%; p<0.0001) and assessment of the mental status was clear (95.16% vs 37.5%; p<0.0001). In the control group, residents on more nights indicated that the sign-out sheet contained too much information (1.64% vs 24.62%, p<0.0001). Although other outcome measures were not significantly different, over 60% of night float residents in both groups indicated that the fluid status of patients was unclear. For Likert-type items (scale of 1–5, with 1 indicating strong disagreement and 5 strong agreement), significantly more residents in the intervention group agreed or strongly agreed (scores 4 and 5) that the sign-out sheet more accurately anticipated problems (83.6% for team 1 vs 53.13% for team 2; p<0.001; see online supplementary figures 1E). Responses to other items were no different between the two groups (see online supplementary figures 1A–D). The overall rating of the quality of sign-out (scale of 1–7, with 1 indicating unacceptable and 7 excellent) was significantly better in the intervention group than controls (rating of 6 or 7 by 43.5% residents in team 1 vs 18.47% residents for team 2; p=0.02; see online supplementary figures 1F).

Supplementary file 6

fig e

Supplementary file 7

fig f
Table 2

Quality of sign-out (dichotomous outcomes)


We found that using a structured sign-out that incorporated discrete elements highly pertinent to critically ill patients such as assessment of mental status and fluid status, as well as intentional focus on anticipated problems, led to significant decrements in the occurrence of NREs and significantly improved the quality of sign-out. Specifically, night float residents exposed to the intervention in comparison to controls less frequently reported occurrence of unexpected or unusual events or being unprepared for such occurrences. Further, interventions that were unplanned or unanticipated and discrepancies in the degree of sickness of patients were less frequent in the intervention arm. Moreover, accurate patient acuities and clarity surrounding mental status were reported more frequently in the intervention group. Remarkably, despite these findings, only 2% of residents in the intervention group as compared with 25% in the control group reported that the sign-out contained excess information. Not unexpectedly, the overall quality of the sign-out was rated better by the intervention group, with significantly more residents rating it in the ‘very good’ category as compared with controls.

Our findings are likely explained by a template prompting for and resident completion of specific documentation about essential elements like patients’ mental status, fluid status, degree of acuity and anticipatory guidance for potential overnight events, as well as the elimination of unimportant information in the sign-out document. In a study of critical incidents, one of the major contributing factors to communication failure was found to be ‘content omission’, in which critical information needed to care for a patient was not communicated, either verbally or in writing, during the handoff process. Failure to report an active medical problem and failure to report pending or ordered diagnostic tests or consults were other major deficient items.8 13 In another prospective survey of resident physicians rotating on a paediatric acute care ward, many residents commented on the importance of including contingency plans during sign-out.28 Many times, this type of information may not be included in the sign-out, and it is difficult to find in the medical record. We simplified and standardised the handoff instrument by including essential and critical information like patients’ mental status, fluid status, degree of acuity, active problem list, follow-up tasks and anticipatory guidance for potential overnight events. This intervention demonstrated improvements in the incidence of NRE, resident perceptions of accuracy, completeness and quality of sign-out. Similar results have been shown in a study involving surgical residents.29 Moreover in a before and after paediatric study, introduction of a multifaceted handoff programme that included a mnemonic to standardise verbal handoffs resulted in significantly decreased medical errors and preventable adverse events.30 A particular strength of our study over most other handoff studies is the inclusion of NREs as patient outcomes measures as well as quality of sign-out as a process measure. Beyond medical errors and preventable adverse events, NREs capture any deviation from optimal or expected care for a specific patient in a specific clinical situation. NREs therefore provide a framework for an outcome measure that represents clinically meaningful patient safety events that have potential for underappreciated impact on patient outcomes.26

There are several limitations in our study. The biggest limitation was poor compliance and completion of the surveys. We tried to address this by posting reminders in the resident team room. Fellow physicians and research coordinators also gave verbal reminders to the night float residents at the end of each shift to complete surveys. Our total response rate was approximately 50%; residents who responded may have differed from those who did not, as may have their level of training. Because our local Institutional Review Board and Graduate Medical Education office required complete anonymity, we were unable to use techniques to mitigate these biases. Anonymity requirements also precluded us from accounting for clustering of responses at the resident level; we treated each survey response as an independent observation. Our results may also partly be explained by social desirability bias, a phenomenon where residents assigned to the intervention may report fewer problems because they perceive the structured sign-out as a superior method.

The results of this study may not be generalisable across all inpatient care settings as it was performed at a single institution in a single ward service. Although the study was performed prospectively, data were collected through postcall surveys in which residents were asked if problems could and should have been anticipated and discussed during sign-out. This type of assessment can introduce significant hindsight bias, although this is less of a risk because they filled out the survey at the end of the shift, rather than at the end of a week of shifts.

Another limitation is that patients are unstable and critically ill when they are in the ICU. Therefore, vitals, fluid status and organ function can change quickly in the ICU, making it difficult to completely eliminate NREs. However, despite these limitations, the intervention led to a decrease in NREs. We did not measure patient outcomes.


The results of our study show that improving the quality of sign-out may lead to mitigation of events that could adversely affect patient safety. The new sign-out document included acuity of the patient, fluid status, mental status, problem list, plan of care, follow-up tasks and anticipatory guidance for any events that could occur. Including this information led to a decrease in NREs, helped anticipate problems more accurately and improved the quality of the sign-out document. Studies are needed to identify if the new sign-out document changes patient outcome measures like preventable adverse events, mortality, length of ICU stay and costs in relation to NREs.

Supplementary file 2

fig a

Supplementary file 3

fig b

Supplementary file 4

fig c

Supplementary file 5

fig d



  • Competing interests None declared.

  • Ethics approval Medical College of Wisconsin Institutional Review Board (IRB).

  • Provenance and peer review Not commissioned; externally peer reviewed.