By Topic

IEEE Quick Preview
  • Abstract



The healthcare system is moving away from the model where one primary physician assumes responsibility for each patient to a more team-based approach. Thus, assessing team communication is critical for patient care as well as to enhance the design of electronic medical record systems and training.

Physician handovers at shift change are performed to ensure that relevant information is transferred to the oncoming provider so that patient care is neither interrupted nor diminished [1]. Physician handover communication has received significant attention as changes in practice patterns, including the use of hospitalists [2] and resident duty hour restrictions [3], have led to an increased number of handovers. Additionally, communication errors that occur during handovers can contribute to adverse events [4], [5].

At a minimum, a patient handover should contain the following elements: patient identifier information, a summative statement about active issues, patients current status and anticipated clinical course, care plans, tasks requiring attention in the near term, anticipated events, and contingency plans [1], [4]. In addition, information should be supplemented with explanatory statements, rationales [4], [6], and directives [7], [8].

In clinical practice, there is a range of ways that the handover process is accomplished with variation in the structure, content, and method of the handover as well as communication style of the participants [9], [10].

Uncertainty about a patient is one of the factors that can lead to variability in the handover process [11]. This uncertainty can arise because of inherent complexities in the management of the acutely ill or because the participants in the handover process lack familiarity with the patient. Factors such as patient acuity level and the development of a new problem can increase complexity that leads to longer handovers [12], [13]. Uncertainty can also surround newly admitted patients as diagnosis and treatments are being formulated. Additionally, the level of familiarity the physician receiving sign out has about the patient can impact the handover process. For example, Perez [12] showed that when a physician receiving the handover did not have prior knowledge about a patient, such as through a previous cross-cover interaction during the patients hospital stay, there was more discussion of the patients problems during shift change.

The majority of previous research on physician shift change handovers has focused on residents [5], [9], [24], [25]. There is a limited research focusing on hospitalist change of service handovers [8], [14] and little published research on hospitalist handovers at shift change. This type of research is important because electronic medical record systems, training, and related interventions may benefit from the study of more experienced personnel.

The purpose of this mixed methods study was to characterize hospitalist handovers at the change of shift via interviews of hospitalists as well as observation, recording, and conversation analysis of hospitalist handover communications. Specifically, this research examines whether hospitalist handovers contain the content [15], [16] elements of a quality handover: patient identifiers, statements about active issues, and statements about care plans. This research also explores whether hospitalist handovers contain language form [15], [16] quality elements: explanations, rationales, and directives. Additionally, this research investigates the impact of patient factors (new admission, new problem, acuity level) and handover receiver knowledge on the inclusion of the six quality content and language form elements in handover communications.



This study was conducted at the University of Virginia Medical Center. Prior to initiating data collection, study protocol approval was obtained from the institutional review board.

A. Setting and Participants

The University of Virginia Medical Center is a 500–600 bed teaching hospital. At the time of this study, six hospitalists supported one of several areas for one week at a time: the General Medicine Hospitalist Service, the General Medicine Consult Service, or the Preoperative Consult Clinic. Interspersed with these duties, the hospitalists take care of administrative tasks. In addition, the hospitalists directly supervise the internal medicine residents for two weeks at a time by serving as a general medicine service attending.

The Hospitalist Service is staffed by one hospitalist at a time. Every Monday, a new hospitalist starts on-service and works seven consecutive days. The hospitalist hands over his patients to a resident team when going off shift at night. When going off service, the hospitalist hands over his patients to another hospitalist.

B. Independent Variables

The first three independent variables listed in Table I assess whether handover utterances are influenced by patient factors. The last independent variable assesses whether handover utterances are influenced by the handover receivers prior knowledge about the patient.

Table 1

C. Dependent Measures

Descriptive statistics at the individual patient handover level were calculated for duration. Every conversation utterance made during the hospitalist handovers was assigned both a content code and a language form code. The proportion of element verbalization across all patient handovers was calculated for the content and language form codes. The content and language form codes were adapted from a previously published communication assessment instrument [15], [16] and a review of the literature [4], [6], [12], [13], [17]. Additional modifications were made after the codes were used in a pilot study of two hospitalist handover sessions.

Tables II and III present the subset of codes used in the quality analysis of hospitalist handovers. The first column contains the coding category name. The second column provides a definition of the coding category, the third column contains a conversation utterance example for each code, and the fourth column maps the coding categories to the literature recommendations about quality elements of a handover.

Table 2
Table 3

D. Data Collection

Each hospitalist was interviewed separately when not on service. The hospitalists provided the number of years they have been in practice following training and whether they have ever received any formal training in the handover process. They discussed their individual approaches to the verbal and written handovers and issues or concerns about the handover process. Interviews were recorded and the responses were analyzed for common themes.

For the patient handover data, digital recordings occurred during a four week period, although Friday through Sunday handovers were excluded in the analysis due to differences in the weekend schedule. The written handover reports, with hospitalist annotations, were collected. These annotations contained per patient information with regard to new admission status, problem status, acuity level, and handover receiver prior knowledge about the patient.

E. Data processing

The adapted coding scheme was incorporated into a modified version of remote analysis of a team environment (RATE), a discourse analysis software application [10], [12], [18]. RATE was augmented to include the content and language form codes. The output from RATE is a Microsoft Access database.

F. Data analysis

With the exception of the acuity level comparison where a two-sided t -test was employed, the underlying distributions of the populations required the use of a two-sided Mann–Whitney test to compare sign-out durations. Two-sided test of proportions were conducted for the six coding categories for the independent variable subgroup analyses.

G. Interrater Reliability

To assess interrater reliability, 30 individual patient handovers were coded by the study analyst (SM) and a senior hospitalist (GH) who was not a participant in the study. Each rater used a checklist with the coding categories. The raters independently listened to the recordings, and for each patient handover, marked if the category was mentioned. After scoring was completed, the Fleiss kappa statistic was calculated [19]. The initial kappa statistic was 0.59. The two raters met to discuss all coding disagreements. After a consensus was reached, specific changes and clarifications were made to the coding categories. The coders then independently relistened to the same 30 patient handovers and a second Fleiss kappa statistic was calculated. The repeated kappa statistic was 0.74.



Results are reported as significant at the 0.05 level.

A. Hospitalist Interviews

The average number of years in practice as hospitalists following postgraduate training for the four hospitalists observed in the study was 6.6 (s.d.= 5.6). The hospitalists were in practice for a range of 3–15 years. None of the hospitalists have received formal training about the handover process either during residency or as a part of continuing medical education.

In answer to the question “do you alter your handover based on the fact that you are signing out to residents, and, if so, how?” all four hospitalists confirmed that they did alter their handover communications. These hospitalists felt that their handovers were generally more explicit because they were signing out to physicians with less experience. One of the four hospitalists specifically stated that he tried to be more detailed about contingency plans in particular when signing out to residents. All of the hospitalists also indicated that they try to take care of any outstanding patient issues prior to leaving and limit any tasks assigned to the covering resident teams.

B. Hospitalist Handover Sessions

All 16 Monday through Thursday hospitalist handover sessions, by four different hospitalists, were recorded. There was a total of 106 individual patient handovers across the 16 sessions. The average number of patients handed over by a hospitalist at end of shift was 6.6 (s.d.=1.5). The minimum number of patients handed over during a hospitalist handover session was three and the maximum number was nine. The average duration of a patient handover was 01:28 (minutes: seconds) (s.d.=00:56). The range of handover durations was 00:07–05:15.

C. Hospitalist Handover Communications-Quality Analysis

Forty-two (39%) patient handovers contained all six quality elements. Ninety-seven handovers (92%) contained all three content quality elements and 42 handovers (39%) contained all three language form quality elements. The final column in Tables II and III presents the number and percentage of handovers by individual quality element. While the majority of the handovers contained five out of the six quality elements, less than half of the hospitalist handovers included directive utterances.

Fig. 1 contains selected excerpts from two different patient handovers given by the same hospitalist in the same handover session where the first handover did not contain any directive utterances and the second handover did. Conversation codes used for analysis appear in parenthesis.

Figure 1
Fig. 1. Selected excerpts from two different handovers, one without directives and one with directives.

D. Impact of Patient and Handover Participant Independent Factors on Handover Quality

The duration of patient handovers where the patient was a new admission were longer than those where the patient was not a new admission (mean=01:53, s.d.=00:59 versus mean=01:19, s.d.=00:52, W=3686.5, p=0.002). A test of proportion evaluating the effect of new admission status (see Table I) on handovers containing all six quality elements was not significant. Individual tests evaluating each element were also not significant.

There was no difference in handover duration for patients who developed new problems compared with those where the patients did not (mean=01:33, s.d.=00:52 versus mean=01:27, s.d.=00:57). A test of proportion evaluating the effect of new problem status on handovers containing all six quality elements was not significant. Tests evaluating each individual element were also not significant.

Patient handovers for the highest acuity patients were longer than patient handovers for the lowest acuity patients (mean=01:46, s.d.=00:51 versus mean=01:13, s.d.=00:35, Formula$t_{28}=\hbox{2.17}, p=\hbox{0.038}$). However, a test of proportion evaluating the effect of new problem status on handovers containing all six quality elements was not significant. Tests evaluating each individual element were also not significant.

There was no difference in handover duration when the receiver had prior knowledge about the patient compared with when the receiver did not (mean=01:40, s.d.=01:12 versus mean= 01:25, s.d.=00:52). Similar to the aforementioned, a test of proportion evaluating the effect of new problem status on handovers containing all six quality elements was not significant. Tests evaluating each individual element were also not significant.



This study characterizes hospitalist handovers at change of shift and shows that overall hospitalists consistently include quality content such as patient identifiers, statements about active issues and care plans. Additionally, the majority of handovers include explanations and rationales. This is crucial because a better understanding by the physician coming on duty about the reasoning behind diagnoses made and treatments planned can lead to improved medical decision making.

An important finding is that only 48% of the handovers contained anticipatory guidance and contingency plans. Previously reported analysis of resident to resident handovers also showed a low number of handovers containing oral anticipatory guidance [20]. Directives are an essential part of the handover process. One of the main goals of a handover is to ensure the physician coming on shift has the knowledge to anticipate and manage events [21]. Without directives, the on-coming provider may, at a minimum, have to duplicate work or spend an inappropriate amount of time searching other information sources to gain needed insight into a problem which can adversely impact patient care [4], [6].

Handovers may be lacking directives for a variety of reasons. While there is an increasing emphasis on developing a handover training curriculum for residents [22], [23], most hospitalists in practice today have not received any formal training. In addition, electronic medical records do not emphasize the need to enter such information. Instead, their handover skills have been learned on the job. Hospitalists interviewed for this study stated that they spent time each day updating their written sign out. While they used their written sign out to scaffold their verbal handover, the hospitalists did not report that they spent extra time preparing what they wanted to say.

As shown in Fig. 1, the difference between a handover that contains a directive and one that does not can be subtle but important. In the patient handover that did not contain a directive, the hospitalist describes the patients baseline mental status and alerts the physicians coming on shift about what might constitute a change. However, there is no direction given to the oncoming team as to how to proceed if a change in a mental status occurs. Crucial time might be wasted if the oncoming team has to start a thorough chart review prior to initiating evaluative workups or treatments.

Interviews revealed that the hospitalists, because they are signing out to residents, feel that they are being more explicit than if they were signing out to other hospitalists. The average duration of the hospitalist handovers in this study was longer than the duration of internal medicine resident to resident handovers reported in other studies [12], [20]. It may be the case that hospitalists think they are communicating directives when, in fact, they are not. The hospitalists interviewed stated that they made attempts to limit tasks assigned to the covering resident teams. The imperative of including directives in patient handovers when there is “nothing to do” may be missing. In the future, it will be important to study how handovers should be tailored to the experience level of the handover receiver while still containing the basic quality elements. In addition, there may be opportunity for a better electronic support for the task management and handing over tasks as well as information.

Another finding from this study is that inclusion of quality elements in handover communications is not always impacted by patient or handover participant factors. The duration of handovers was longer for new patients and higher acuity patients but not for patients with new problems or when the receiver did not have prior knowledge. A more nuanced analysis would need to be done to have a better understanding about the relationship between handover duration and content and how these components contribute to the total value of the handover.

In addition to characterizing hospitalist handovers, another contribution of this study is the adaptation of an existing communication assessment instrument [15], [16]. The modifications were implemented in order to fully capture and encode hospitalist handovers. Further work needs to be done to develop generalizable and service specific communication analysis tools.

This research has several limitations. The recorded handovers were a convenience sample obtained during a one month observation period. This study was done at a single, academic medical center. While a majority of the hospitalists at the University of Virginia were recorded, the total number of hospitalists studied was small and the impact of different individual hospitalist communication styles on the handovers was difficult to quantify. This particular hospitalist program is unusual in that the hospitalists handover their patients to residents, rather than other hospitalists, at the shift change. Additionally, handover communications may have been altered because the participants were aware they were being recorded.

Another limitation is that new problem status and the rating of patient acuity level were assigned by the hospitalist and not independently confirmed via a medical chart review. A final issue is that neither the hospitalist written sign-out document nor handover receiver communications were included in this analysis. A future goal is to examine the verbal exchange between handover participants in detail in conjunction with the written handover in order to yield a richer characterization of hospitalist handovers.

In 2009, the Society of Hospital Medicine (SHM) published a set of Hospitalist Handoff Task Force recommendations and included anticipatory guidance along with contingency plans as one of the main components of a verbal handover [8]. Use of directives was also considered a key skill put forth in the core competencies curricular plan developed by the SHM [7]. However, this study shows that consistent communication of directives is lacking in hospitalist verbal handovers. Findings from this study highlight that it may be invalid to assume that hospitalists and other attending level physicians have expertise when it comes to the handover process simply because they have clinical experience. Hospitalists are ubiquitous in clinical practice and often serve as mentors in the academic setting. While there is still a debate about what constitutes best handover practices and how to implement these practices into a teaching curriculum [9], [22], [23], results of this study can be used to inform training at all levels as well as the design of electronic support for the handover. Both content and language form elements contribute to a quality handover. Simple interventions to make physicians more aware of this issue may be enough to improve inclusion of directives in handover communications. Therefore, as training and electronic medical record systems are redesigned, it will be important going forward to focus on improving hospitalist and attending handover communications in addition to other healthcare providers.


This work was supported in part by National Library of Medicine (NLM) under Grant T15LM009462. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NLM or NIH.

S. Meth is with the Department of Systems and Information Engineering, University of Virginia, Charlottesville, VA 22903 USA (e-mail:

E. J. Bass is with the College of Information Science and Technology, College of Nursing and Health Professions, Drexel University, Philadelphia, PA 19104 USA (e-mail:

G. Hoke is with the Department of Medicine, University of Virginia, Charlottesville, VA 22903 USA (e-mail:


No Data Available


No Photo Available

Sharon Meth

No Bio Available
No Photo Available

Ellen J. Bass

No Bio Available
No Photo Available

George Hoke

No Bio Available

Cited By

No Data Available





No Data Available
This paper appears in:
No Data Available
Issue Date:
No Data Available
On page(s):
No Data Available
INSPEC Accession Number:
Digital Object Identifier:
Date of Current Version:
No Data Available
Date of Original Publication:
No Data Available

Text Size