eTable 1. Type and frequency of distractions observed during morning handover
eTable 2. Narrative examples of clinically important and benign issues by category
Devlin MK, Kozij NK, Kiss A, Richardson L, Wong BM. Morning Handover of On-Call IssuesOpportunities for Improvement. JAMA Intern Med. 2014;174(9):1479-1485. doi:10.1001/jamainternmed.2014.3033
Handover is the process of transferring pertinent patient information and clinical responsibility between health care practitioners. Few studies have examined morning handover from the overnight trainee to the daytime team.
To characterize current morning handover practices in 2 academic medical centers by assessing the frequency of omissions of clinically important overnight issues during morning handover and identifying factors that influence the occurrence of such omissions.
Design, Setting, and Participants
A prospective, point-prevalence study was conducted in the general internal medicine wards of 2 tertiary care academic medical centers in Toronto, Ontario, Canada, in 2012 and 2013. Participants included on-call third-year medical students and first- and second-year residents.
Main Outcomes and Measures
Completeness of morning handover of clinically important overnight issues identified using a targeted medical records review and processes of morning handover characterized by direct observation.
We identified 141 clinically important overnight issues during 26 days of observation. The on-call trainee omitted 40.4% (95% CI, 32.3%-48.5%) of clinically important issues during morning handover and did not document any information in the patient’s medical record for 85.8% (95% CI 80.1%-91.6%) of these issues. By univariate analysis, running the list patient-by-patient (ie, the entire team discusses each patient) (OR, 4.32; 95% CI, 1.94-9.60; P < .001) and using a dedicated handover location (OR, 2.61; 95% CI, 1.30-5.22; P = .007) positively correlated with handover of an issue taking place, whereas distractions in the meeting area inversely correlated with the likelihood of handover of an issue taking place (OR, 0.96 for every increase in 1 distraction; 95% CI, 0.93-0.98; P = .002). Using a multivariate mixed-effects model, only running the list remained as an independent predictor of the handover of an issue (OR, 3.80; 95% CI, 1.25-11.49; P = .02).
Conclusions and Relevance
On-call trainees omit numerous clinically important issues when handing over to the daytime team. Training programs should introduce educational activities and workflow changes, and provide dedicated time and a distraction-free environment, to improve handover of on-call issues.
Handover is the process of communicating pertinent patient information and transferring clinical responsibility between health care practitioners.1 The Accreditation Council for Graduate Medical Education Task Force on Quality Care and Professionalism (2011)2 and the Canadian National Steering Committee on Resident Duty Hours (2013)3 have recently published recommendations that discourage 24-hour shifts. These restrictions ultimately increase the number of resident-led handovers that occur in academic medical centers. Patient care transitions, including handovers, represent a vulnerable time for patient safety.4- 7 Studies6- 8 have attributed adverse events to suboptimal handover between residents and their colleagues, staff physicians, and other members of the health care team. As a result, the Joint Commission9 made the standardization of patient handover communication a National Patient Safety Goal in 2006, and the Accreditation Council for Graduate Medical Education10 lists transitions in care as 1 of 6 key areas of focus in its new Clinical Learning Environment Review program.
Most studies11- 15 evaluated the adequacy of evening handover to a resident on call. Major inconsistencies and failures in evening handover have led to interventions to improve the handover process.5,7,11,13,15 Strategies such as the use of mnemonics, standardized written sign-out, direct observation and feedback, and dedicated curricula that teach handover communication skills have improved residents’ satisfaction with handover, their perceptions of preparedness, and increased accuracy and completeness of their handover.11,16- 22 More recently, the implementation of a multifaceted handover program significantly reduced medical errors and preventable adverse events among hospitalized children.23 Thus, strategies to improve the handover process may lead to improvements in patient safety.
To date, there has been a paucity of research describing morning handover from the overnight resident to the daytime team. One study11 highlighted the potential for communication failures to occur at the time of morning handover, and another study24 surveyed physicians to assess their perceptions of formal “medical morning handover reporting” rounds, but neither addressed the effect of morning handover processes on the quality or completeness of morning handover. Another study25 compared residents’ adherence to 3 different approaches to morning handover. The investigators found that face-to-face morning handover occurred more reliably than handover via written or electronic notes, but their study did not measure which approach resulted in fewer handover omissions of overnight issues. Given the lack of literature examining the completeness of morning handover, we sought to better characterize this process and assess how frequently the on-call trainee hands over clinically relevant issues to the daytime team. We also sought to identify factors that influence the completeness of morning handover.
We observed morning handover on the inpatient general internal medicine (GIM) service at 2 large, tertiary care academic medical centers in Toronto, Ontario, Canada. One center has a clinical teaching unit with 5 GIM teams, and the other has a clinical teaching unit with 4 GIM teams. The research ethics boards of both institutions approved this study, and the participants provided written informed consent. Each team consists of 1 attending physician, 1 senior resident (second- or third-year resident), at least 2 interns, and 2 medical students. On a typical day, each team is responsible for 15 to 30 admitted patients. Although both sites have an electronic medical record that displays laboratory test results, diagnostic imaging results, medications, and some dictated clinic notes and prior discharge summaries, all physician and nurse documentation occurs on paper.
During weekday daytime hours, most team members are present and provide direct patient care. At the end of the day, the physician team hands over care of the patients to the on-call trainee. Most times, the on-call trainee is a team member familiar with the admitted patients. Occasionally, because of vacation or illness, we assign a trainee from another team (cross-coverage) or clinical service (floater), with no prior affiliation with the team and limited familiarity with the admitted patients, to be on call. To support the verbal handover process, trainees use an Internet-based written sign-out tool that summarizes key patient information. This tool automatically imports demographic (eg, age and sex) and administrative (eg, room number and date of admission) information into the tool, but physicians manually enter all other clinical information. Residents receive evening handover training as part of an academic half-day session, but this training does not focus on developing communication skills to improve morning handover.
During the overnight period (5 pm to 8 am), the on-call trainee admits patients through the emergency department and attends to issues that arise among patients who already have been admitted. The on-call trainee is typically responsible for 1 or sometimes 2 clinical teaching unit teams. The next morning the on-call trainee meets with the daytime team for rounds to discuss new patients and hand over overnight issues for previously admitted patients. The local duty hour regulations limit the maximum number of consecutive work hours to 24 with a crossover period of 2 hours to allow for adequate patient care handover.
We included 5 teams from 1 site and 3 teams from the second site. We excluded 1 team from the second site because a study investigator (M.K.D.) had been a member of that team immediately before the data collection period. At the beginning of the month, we randomly selected 1 team for data collection each Tuesday to Friday morning. We excluded observations on Saturday, Sunday, and Monday because discussions on these days included issues that occurred over a 2- to 3-day period. We carried out observation and data collection at 1 site in October 2012 and at the other site in April 2013.
One researcher (M.K.D. or N.K.K.) observed the selected team during morning rounds. During the observation period, the researcher took field notes to document predefined elements of the morning handover process, paying particular attention to the handover method and the presence of distractions (potentially leading to discontinuity in the process) and interruptions (distractions that interrupt handover). The researcher also noted each overnight issue that the on-call trainee handed over to the daytime team. The researcher (M.K.D. or N.K.K.) conducted a targeted medical records review of patients already admitted to the hospital covered by the on-call trainee (as opposed to patients admitted overnight). We reviewed the progress notes, orders, nursing notes, and laboratory test results. This allowed us to capture clinical issues that occurred during the on-call period.
We defined clinically important issues as factors that are likely to affect a patient’s clinical course. Examples of clinically important issues include assessing a patient with acute abdominal pain or prescribing acetaminophen for a patient with a new fever. Examples of clinically benign issues include prescribing a laxative for an otherwise stable patient with constipation or acetaminophen for a patient with chronic arthritis. Two pairs of the research team members (N.K.K. and B.M.W. or M.K.D. and L.R.) reviewed each issue to determine its significance. To assess the reliability of our judgment of clinical significance, we carried out a κ analysis of interrater reliability using a retrospectively selected random sample of 30 events that 2 study investigators (M.K.D and N.K.K.) independently rated.
We included clinically important issues only when it was clear that the on-call trainee was aware that such an issue existed. These were instances in which the trainee responded by writing a new order or the physician or nursing documentation clearly indicated that the trainee was aware of the issue. We excluded clinically benign issues or when we could not determine the clinical significance of an issue based on the information available in the patient’s medical record.
For the study’s primary outcome, we correlated each clinically relevant issue identified through medical record review with issues communicated during the morning handover to determine the proportion of clinically relevant issues handed over to the team by the on-call trainee. We also recorded whether the on-call trainee documented the issue in the progress notes or the Internet-based written sign-out tool.
We used descriptive statistics to summarize continuous variables using means (SDs) and categorical variables using counts, percentages, and 95% CIs. The primary analysis involved both univariate and multivariate mixed-effects logistic regression models to examine the relationship between predictors of interest and morning handover. We selected predictors a priori, which included the handover method, number of distractions, handover location, caseload of the trainee, training level of the trainee, and training program. We ran the mixed-effects logistic regression model, placing 5 predictors into the model with the objective to produce a parsimonious model, to assess the outcome of interest, treating physician team as a random effect. Before the analysis, we assessed the predictor variables for the presence of multicollinearity. We considered a tolerance coefficient of less than 0.4 to indicate the presence of multicollinearity and in such cases we retained only 1 member of a correlated set of variables for the multivariate model. We reported both the univariate and multivariate model results using odds ratios (ORs) and their associated 95% CIs. We carried out all analyses using SAS, version 9.2 (SAS Institute Inc).
We observed each of the GIM teams 2 to 4 times during 26 mornings (13 observations at each site). Table 1 summarizes key study characteristics. The on-call trainee was usually a first-year resident (19 [73%]) in the internal medicine training program (14 [54%]). The on-call trainee was cross-covering and providing coverage for a second team on 11 occasions (42%). The on-call trainee was responsible for a mean of 17.4 (3.0) patients per team and admitted 2.3 (0.9) new patients per night on call.
Handover took place in many different locations in the hospital (Table 2) and occurred in a dedicated team room only 41% of the time. Teams divided the handover process into more than 1 encounter 68% of the time (eg, handover of overnight issues occurred before and after morning teaching rounds). Teams met for a mean total of 71 (26) minutes to review new cases and hand over overnight issues. Teams spent most of their time reviewing new information on patients admitted during the previous night, with a mean of only 11 (10) minutes dedicated to handing over on-call overnight issues. During these interactions, teams experienced 6.1 (7.1) distractions per hour resulting in 2.6 (2.9) interruptions per hour; eTable 1 in the Supplement summarizes the nature and frequency of these distractions.
Teams used 3 main approaches to hand over overnight issues. The first approach, running the list, involved the entire team discussing each patient one by one using the written sign-out list as a guide. The on-call trainee provides updates as the team discusses each patient. The second approach, single opportunity, involved the team giving the on-call trainee one opportunity to provide updates related to the overnight period for all admitted patients at one point during the encounter. The third approach, handover via intermediary, involved the on-call trainee handing over to an intermediary, usually a senior-level team member (ie, staff physician or senior resident) and then leaving the hospital. The intermediary then relayed overnight issues to the entire team later in the morning. Running the list patient by patient was used 63% of the time (n = 20), providing a single opportunity to hand over issues occurred 22% of the time (n = 7) , and handing over to an intermediary was used in 15% of handover interactions (n = 5).
We reviewed 453 individual medical records and identified a total of 222 overnight issues. Of these, we excluded 81 issues (36%) from the analysis for the following reasons: nursing notes documented 49 issues where it was unclear whether the on-call trainee was aware that the issues had occurred, 20 were clinically benign issues, we could not determine the clinical significance of 7 issues, and in 5 cases it was not clear whether the issue occurred during the overnight period (ie, an order was not time-stamped).
We analyzed the remaining 141 clinically important overnight issues (κ = 0.49). The most common overnight issues were a change in a patient’s clinical status (47.5%), an abnormal laboratory test result (15.6%), and review and response to consultant recommendations (12.8%). Narrative examples are presented in eTable 2 in the Supplement. The on-call trainee did not verbally hand over 40.4% (95% CI, 32.3%-48.5%) of the clinically important overnight issues and did not document a progress note for 85.8% (95% CI, 80.1%-91.6%) of these issues. Trainees documented 7.8% (95% CI, 3.4%-12.2%) of clinically important issues in the Internet-based written sign-out tool. There were 52 (36.9%; 95% CI, 28.9%-44.8%) clinically important issues that were neither handed over nor documented by the on-call trainee. Table 3 provides a description of the clinically important issues that were not handed over.
The univariate analysis identified 3 factors that correlated significantly with an increased likelihood of a clinically important issue being handed over (Table 4). Running the list (OR, 4.32; 95% CI, 1.94-9.60; P < .001) and dedicated handover location (OR, 2.61; 95% CI, 1.30-5.22; P = .007) correlated positively, whereas distractions (OR, 0.96 for every increase in 1 distraction; 95% CI, 0.93-0.98; P = .002) correlated inversely with the likelihood of morning handover occurring. The multivariate analysis revealed a single independently correlated predictor: running the list (OR, 3.80; 95% CI, 1.25-11.49; P = .02).
Our study provides what we believe is the first detailed description of the morning handover process at 2 large academic medical centers. We discovered that the on-call trainee omitted numerous clinically important on-call issues when handing over to the daytime team at the end of an on-call shift. If we extrapolate our findings to all GIM teams at both training sites, approximately 1500 clinically important issues arise in our patients during the on-call period every month. This means that on-call trainees fail to hand over hundreds of clinically important issues to the daytime team every month. These omissions have the potential to lead to delays in care and the development of adverse events for our patients.6
Another major concern that arises from our findings is the fact that only 14% of clinically important issues from the overnight period had an accompanying progress note from the on-call trainee in the patient’s medical record. This strikingly low documentation rate has important continuity of care and medicolegal implications. More than 1 in 3 issues that took place during the on-call period were neither handed over nor documented in the medical record. These omitted and undocumented issues are particularly worrisome because they pose the greatest risk for delays in expeditious follow-up on clinically important overnight issues.
Our local morning rounds did not incorporate standardized morning handover practices, with individual teams using multiple handover methods that were unpredictable and inconsistent. We discovered that variations in handover methods and training environment affected the likelihood of significant issues being handed over. In one case, handover of overnight issues was as brief as 1 minute. On many occasions, there was not enough time to devote to morning handover because of the competing demand of reviewing patients admitted during the overnight period and attending to urgent clinical issues that arose during rounds. The brevity of communication about these issues may have given trainees the impression that they are not a priority for discussion and thus perpetuated the incompleteness and brevity of morning handover. The on-call resident’s training program and seniority did not significantly alter the odds of handing over issues, which suggests that there may be a lack of formal training across all residency programs and training levels.
Systematically running the list patient-by-patient correlated independently with more reliable handover of overnight issues. In a study by Suozzo and colleagues,26 residents who underwent neuropsychological testing after being on call demonstrated impaired verbal immediate memory, delayed recall, increased ease of distraction, and decreased time to forget new information. Running the list may help residents to focus and act as a memory trigger for each patient on the team. Similar to concerns identified with the double-handover that can occur with night float-call systems,27,28 morning handover that occurred through an intermediary also increased the likelihood for trainees to omit communication of important overnight issues after being on call. In our setting, handover via an intermediary occurred primarily because the on-call trainee needed to leave the hospital to adhere to local resident duty hour requirements, yet team members were not present during morning rounds because of competing obligations such as academic half day. Further restrictions to resident duty hours and the associated risk of reduced resident presence during daytime working hours29 serve to promote suboptimal handover processes.
One of the key applications of our findings is to provide guidance for training programs about how to optimize the morning handover process. Opportunities for improvement exist at the trainee, team, training program, and training environment levels. At an individual level, trainees should routinely review physician orders and nursing notes on their daily rounds. This action may reveal additional overnight issues omitted during morning handover. The on-call trainee must also clearly document assessments and actions taken for all clinically relevant issues that occur during the on-call period. At the team level, teams should set aside dedicated time and routinely run the list in the presence of the on-call trainee to facilitate recall of overnight issues at the time of morning handover.
Training programs should introduce new or expand existing handover curricula to raise awareness about the distinct entity of morning handover and to improve the communication process during this period. Although general principles of structured team communication clearly apply, some existing tools or mnemonics commonly used for evening handover may not necessarily be relevant for the morning handover communication.19,30 For example, the first “S” in I-PASS (illness severity, patient summary, action list, situation awareness and contingency planning, and synthesis by receiver) indicates that anticipatory guidance should be provided to the on-call trainee and is not relevant for the on-call trainee to provide to the daytime team. Instead, training programs might invoke the SBAR (situation, background, assessment, and response) mnemonic. The SBAR mnemonic may prove to be the most appropriate structured communication tool for morning handover because it encourages discussion of specific situations and emphasizes the on-call resident’s assessment and response.18
At the training environment level, efforts should be made to minimize distractions, limit nonurgent pages (eg, ask nursing and allied health staff to defer nonurgent pages), and provide a dedicated space where handover can occur. Some institutions use web-based paging systems that triage and store nonurgent messages as a way to limit paging interruptions.31 Recognizing that some interruptions are necessary and relate to urgent issues, residents ultimately need to use interruption-handling strategies such as returning to a previously interrupted handover or pausing during the handover process at opportune moments so that they are more likely to resume after the interruption.32 In fact, running the list patient-by-patient might promote the return to the handover processes after an interruption.
This study has several limitations. First, we did not capture conversations or interactions that took place outside of morning rounds (eg, one-on-one conversations and text messages), which may have overestimated our reporting of handover omissions. On the other hand, we included only issues that were documented in the medical records, nursing notes, or physician orders. There may have been other issues not captured in these documentation sources, the absence of which would have underestimated our reporting of handover omissions. It is possible that the presence of an observer affected the manner in which handover occurred (ie, the Hawthorne effect). This may have led to increased vigilance and improved performance resulting in an underestimation of missed handovers. Ascertainment bias may have affected the number of clinically relevant overnight issues identified during medical record review. However, some days we performed the review before observing handover interactions, which would have mitigated this bias. Differences in the workflow and resources at our academic centers may affect the generalizability of our findings. For example, our trainees cover ward patients and admit patients from the emergency department when they are on call and do not use an electronic medical record for documentation.
Finally, this study did not examine whether morning handover omissions affected patient outcomes. However, unlike many of the prior studies evaluating the quality of handover, we did not rely on trainees’ recall to determine the potential clinical effect. Instead, our methods relied on direct observation and medical record reviews, which we believe is a strength of our study design and provides a more robust description of the potential clinical effect of morning handover omissions.
The morning handover process is highly variable and unreliable and often occurs in a chaotic clinical care environment. On-call trainees fail to hand over numerous clinically important issues to the daytime team and frequently do not document their assessments and responses to the on-call issue in the medical records. These omissions have the potential to cause unnecessary delays and may result in a lack of follow-up for important patient issues; consequently, they can have a negative effect on patient care. Although we recommend potential interventions that training programs could introduce to address the issue of suboptimal morning handover that are relatively feasible and inexpensive, studies are needed to evaluate whether such interventions would increase the reliability of morning handover processes and result in safer patient care.
Accepted for Publication: May 8, 2014.
Corresponding Author: Brian M. Wong, MD, Department of Medicine, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Room H466, Toronto, ON M4N 3M5, Canada (email@example.com).
Published Online: July 21, 2014. doi:10.1001/jamainternmed.2014.3033.
Author Contributions: Drs Devlin and Kozij contributed equally to the study. Dr Wong had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Devlin, Kozij, Richardson, Wong.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Devlin, Kozij, Kiss.
Critical revision of the manuscript for important intellectual content: Devlin, Kozij, Richardson, Wong.
Statistical analysis: Kozij, Kiss.
Administrative, technical, or material support: Kozij, Richardson.
Study supervision: Richardson, Wong.
Conflict of Interest Disclosures: None reported.