Original Contribution

ED handoffs: observed practices and communication errors^?,??

Brandon C. Maughan MD, MHS ^a,?,1, Lei Lei MD ^b,1, Rita K. Cydulka MD, MS ^c

^aDepartment of Emergency Medicine, Rhode Island Hospital & Warren Alpert School of Medicine of Brown University,

Providence, RI, USA

^bDepartment of Emergency Medicine, University of Rochester Medical Center, Rochester, NY, USA ^cDepartment of Emergency Medicine, MetroHealth Medical Center and Case Western Reserve University, Cleveland, OH, USA

Received 18 November 2009; revised 2 December 2009; accepted 12 December 2009

Abstract

Objective: The Study objectives were to identify emergency department (ED) handoff practices and describe handoff communication errors among emergency physicians.

Methods: Two investigators observed patient handoffs among emergency physicians in a major metropolitan teaching hospital for 8 weeks. A data collection form was designed to assess handoff characteristics including duration, location, interruptions, and topics including examination, laboratory examinations, diagnosis, and disposition. Handoff errors were defined as clinically significant examination or laboratory findings in physician documentation that were reported significantly differently during or omitted from verbal handoff. Multivariate negative binomial regression models assessed variables associated with these errors. The study was approved by the institutional review board.

Results: One hundred ten handoff sessions encompassing 992 patients were observed. Examination handoff errors and omissions were noted in 130 (13.1%) and 447 (45.1%) handoffs, respectively. More examination errors were associated with longer handoff time per patient, whereas fewer examination omissions were associated with use of written or electronic support materials. Laboratory handoff errors and omissions were noted in 37 (3.7%) and 290 (29.2%) handoffs, respectively. Fewer Laboratory errors were associated with use of electronic support tools, whereas more laboratory handoff omissions were associated with longer ED lengths of stay.

Conclusions: Clinically pertinent findings reported in ED physician handoff often differ from findings reported in physician documentation. These errors and omissions are associated with handoff time per patient, ED length of stay, and use of support materials. Future research should focus on ED handoff standardization protocols, handoff error reduction techniques, and the impact of handoff on patient outcomes.

(C) 2011

^? Research site: MetroHealth Medical Center, Department of Emergency Medicine, 2500 MetroHealth Drive, Cleveland, Ohio 44109.

^?? Prior presentation poster at SAEM 2008 Annual Meeting, May 29-June 1, 2008, Washington, DC.

* Corresponding author. Department of Emergency Medicine, 593 Eddy Street, Claverick 274, Providence, RI 02903, USA.

E-mail address: [email protected] (B.C. Maughan).

¹ Previous affiliation: Case Western Reserve University School of Medicine, Cleveland, Ohio.

0735-6757/$ - see front matter (C) 2011 doi:10.1016/j.ajem.2009.12.004

ED handoffs

Introduction

Background and importance

Poor communication is recognized as a major factor contributing to the estimated 44 000 to 195 000 patient deaths that occur each year due to Medical error [1,2]. Patient handoffs between physicians are recognized as a time of potential communication lapses that leads to errors in patient care. Fifty-nine percent of medical and surgical residents in a recent study reported that one or more of their patients were harmed as a result of inadequate handoffs [3]. In an effort to reduce handoff errors and resultant Medical errors, the Joint Commission has made implementation of a standardized approach to handoff communications a Hospital National Patient Safety Goal [4].

There is perhaps no area in the hospital with a greater number of handoffs than the emergency department (ED). Multiple handoffs occur each shift between nurses, physi- cians, and ancillary staff, as well as between ED personnel and other hospital personnel during patient testing, imaging, and admission processing. In addition, shift work necessi- tates patient handoffs at the end of every shift. These handoffs often involve unstable patients with high-acuity illness. In spite of the frequency of handoffs and despite the use of established handoff strategies in other high-risk industries, standardized protocols or guidelines for transfer of care of patients in the ED are not yet in place [5,6]. Almost 90% of ED residency directors answering a recent survey reported no existing uniform patient handoff policy [7].

Goals of this investigation

The objectives of this study are to describe ED physician handoff practices and quantify handoff commu- nication errors.

Methods

Study design

This was a prospective observational study of ED handoffs conducted at an urban, public teaching hospital with an annual ED census of approximately 91 000 patients. The ED is a 72-bed unit with an additional 14 beds devoted to an observation unit. The study protocol was approved by the institutional review board through expedited review, with a waiver of the requirement of written informed consent.

Data collection tool

To guide the development of our data collection form, we developed a short questionnaire regarding components of patient handoffs and their perceived importance. The survey questionnaire was based on handoff literature and the

503

guidance of a senior emergency medicine (EM) faculty member [5,6]. It consisted of 26 multiple choice questions detailing physician opinion on handoff duration, location, ED personnel who should be included in handoffs, the frequency with which incorrect information is communicat- ed at handoff, and the importance of listing medications, medical history, and results of laboratories and imaging studies during handoffs. Respondents were also encouraged to use additional space for comments. The survey was administered to resident and attending physicians affiliated with an urban EM training program. Surveys were distributed to residents during weekly conference and to attendings’ departmental mailboxes. No identifying infor- mation except for Experience level was collected.

On the basis of a questionnaire results, discussion with content experts, literature review, and direct observation of ED patient handoffs, we organized a set of topics that encompassed most information transfer during handoff. Repeated pilot trials with handoff observation were used to refine the content and layout of a data collection form that would permit rapid, real-time data collection on these topics during handoff (Table 1). Each data collection point was defined a priori. The form was divided into major content areas: basic characteristics (time, patient number, patient identifiers used), ED visit summary (diagnosis, treatment, disposition), interruptions, physical examination, laboratory examinations, and imaging. The physical examination was divided into 12 components including vital signs, constitu- tional signs, and organ systems. The section for laboratory results included 8 specific laboratory examinations that are commonly ordered in the ED as well as a space for one additional laboratory examination that could be specified by the observer if necessary. A section for imaging studies was initially included; however, based on logistic concerns, the decision was made before the start of data collection to not collect imaging data.

Examination findings reported during handoff were classified as normal or abnormal; otherwise, the corresponding examination section was left blank if the topic was not discussed. A specific option for “none” was not included because examination results were presented in variable order and a premature mark in this box would complicate data collection if the physician later reported pertinent findings. A high threshold was set for classifying results as abnormal. Handoff results were only marked as abnormal if the outgoing physician noted them to be clinically significant or pertinent to the patient’s chief complaint. Mildly abnormal, incidental, or clinically insignificant examination findings were not recorded as abnormal. The same threshold of clinical significance was applied to laboratory results reported during handoff.

Study setting and population

Two medical students with clinical clerkship experience were trained for 5 weeks to observe and assess patient

handoffs using the structured data collection form. During this time, the observers recorded data from more than 60 individual patient handoffs and reviewed the results of each session with a senior EM faculty member. To maintain a high degree of objectivity, the observers defined criteria for each component of the observed handoffs and revised these definitions with a faculty member on multiple occasions before the start of data collection. The EM residents and attending physicians were informed before the study that observers would be conducting “workflow” research in the ED, but they were not informed of the study’s emphasis on handoffs or of the specific data being collected. The handoff observations were conducted in a way to minimize observer bias. Observers did not allow ED physicians to view the content of the data collection forms, and the observers remained in the ED between handoff sessions and feigned data collection during these nonhandoff times.

Observers collected handoff data during an 8-week period in September and October 2007. Before each handoff, the observer recorded identifiers (eg, name, room, medical record number) on all patients being treated by the outgoing physician. This information was used after the handoff to retrieve patient data from the electronic medical record (EMR) so that verbal handoff results could be compared with chart notes. After the review of the patient’s EMR was complete, the patient-specific identifiers were shredded and discarded. At this institution, EM physicians are required to complete end-of-shift notes in the EMR on all patients seen and handed off. These notes are typed directly into the EMR, with or without note templates, and do not require delays (eg, for transcription) before becoming available for review. For patients seen primarily by the EM residents, both the resident and attending type separate notes. Attending physicians type a single note for patients they see primarily (eg, without residents). Of note, attending physicians in this ED treat a relatively large proportion of patients as their primary patients compared to many other teaching hospital EDs. Physicians will often complete their end-of-shift notes before the handoff, thereby, making the notes available for our investigators to review immediately following the end of handoff discussion. If these notes were not complete by the end of handoff discussion, the observers would review the EMR for completed notes at the end of their observational shifts, usually 2 to 3 hours later. Only in rare instances were the EMR handoff notes still unavailable for review at this time. In all such cases, the observers returned to the ED within 24 hours and reviewed the completed handoff notes in the EMR at that time.

At the study site, there are multiple EM physician shifts each day, leading to complex handoff arrangements. All shifts are staggered between resident and attending (except for the 7:00 AM handoff) to avoid complete turnover in patient care all at once. Handoffs were observed for 8-hour attending shifts ending at 3:00 PM, 4:30 PM, 6:00 PM, 11:00 PM, and midnight as well as for 10-hour resident shifts ending at 5:00 PM, 7:00 PM, and 9:00 PM. A “team handoff”

involving all incoming and outgoing ED physicians, both residents and attendings, was observed for the shifts ending at 7:00 AM. Attending handoffs included patients seen primarily by the attending as well as patients seen with residents. The attending handoff at 1:00 AM and the resident handoff at 3:00 AM were not observed for data collection based on the very low number of handoffs anticipated at these times. The resident handoff at midnight was rarely observed as it usually conflicted with an attending handoff scheduled at the same time. Observed resident handoffs included only EM residents not off-service residents from other departments. Data collection dates were assigned using a convenience sampling method. On dates during which data collection occurred, all handoffs in a given period were observed and documented. Twice monthly, both observers collected data on the same handoffs to measure Interobserver variability. An 8% sample of handoffs was observed by both investigators.

Outcome measures

The primary outcome measures in this study were handoff communication errors and omissions. Handoff errors were defined as clinically significant discrepancies between verbal handoff and the EMR physician documentation. For example, an error could include an examination finding or laboratory value that was noted as normal in the EMR documentation but reported as abnormal during handoff discussion. Handoff omissions were defined as clinically pertinent findings in EMR documentation that were excluded from handoff. The expectation was that handoff should include abnormal examination findings or laboratory results that the outgoing physician regarded as important. We expected physicians to report normal findings only if they believed the results were pertinent.

The EMR data on physical examination were found in the physician’s typed end-of-shift note, whereas EMR laboratory data were automatically updated once results were available. A section for imaging was initially included on the form but was not used for data collection because investigators learned that official EMR results for this section would not be available in a timely manner.

To measure handoff communication errors, the physical examination and laboratory results discussed during handoff were compared with the physician’s note in the EMR. Physical examination findings reported during observed handoff were coded as normal, abnormal, or blank (not discussed). All examination subcomponents were coded as global normal if the outgoing physician stated that the examination was normal but offered no specific details. Laboratory results were coded as normal, abnormal, global normal, pending, or global pending. Of note, the EMR at the study institution reports all laboratory examinations that have been ordered for a patient, even if a result is not yet available. In addition, the EMR reports the time at which laboratory results became available. Our expectation was that physi-

Table 1 Handoff data collection form

ED handoffs

505

MRN, medical record number; HEENT, head/eyes/ears/nose/throat exam; CBC, complete blood count; CT, computed tomography; MRI, magnetic resonance imaging, EKG, electrocardiogram.

cians should check laboratory results within 30 minutes of handoff and report them if pertinent to the patient’s care. If clinically significant laboratory results were available for more than 30 minutes before handoff and the outgoing physician did not include them in handoff, it was considered a handoff omission based on the failure to report clinically pertinent data.

Our expectation was that the outgoing physician should report all significantly abnormal examination and laboratory findings during handoff and that all findings reported in verbal handoff should be correct. We considered it acceptable for a physician to forgo reporting normal results, but it would be an error to report normal findings as abnormal. Similarly, pending laboratory results could be reported as pending or could be omitted from handoff, but it would be an error to report pending laboratory results as normal. See Table 2 for the criteria defining all communi- cation errors. Handoff errors and omissions were summed for both physical examination and laboratory studies as the 4 outcomes measures of every patient handoff.

Data analysis

Handoff data were recorded in Microsoft Excel (Microsoft Corporation, Redmond, WA) and transferred to Stata 9 (Stata Corporation, College Station, TX) for analysis. Handoff observation data are presented as frequencies or as means with SD. Multivariate negative binominal regression models were created to determine whether handoff variables were associated with communication errors. Model covariates were selected based on data types collected during handoff and included physician training level, handoff duration, number of patients handed off during a session, patient triage acuity, ED length of stay , interruptions during handoff, degree of verbal interaction during handoff, and use of electronic or written support materials. The degree of verbal interaction was measured on a numeric scale of 0 to 4, with the total representing the number of “interactive techniques” (Table 1) measured by the observer for each handoff. A single continuous variable with a nonlinear relationship to outcomes was converted to a set of categorical variables. All other continuous data were included in the regression model as linear variables. Regression model results are presented as relative risk with a 95% confidence interval (CI).

Results

Handoff observation

One hundred ten physician-to-physician handoff sessions were observed for a total of 992 patient handoffs. Seventy- three (7%) handoffs were between resident physicians only, 779 (79%) were between attending physicians only, and 140 (14%) included both in group handoffs. Most resident

handoffs were among senior residents. All observed handoffs were face-to-face exchanges among physicians. Nurses did not participate in any observed handoffs. The average attending handoff session included 10.4 patients and lasted

14.4 minutes. Resident handoffs were shorter with an average of 7.5 patients in 10.4 minutes, whereas team handoffs (including multiple attending and residents) were longer with 15.5 patients in 17.6 minutes. Nearly 4 (78.7%) in 5 handoffs occurred in a central staff work area in the ED but out of patient view, whereas the remaining handoffs occurred at 1 of 2 large physician workstations centered among patient rooms. Physical examination findings were reported in 405 handoffs (40.8%). Details of handoff characteristics are noted in Table 3. Interobserver agreement was 0.75 (95% CI, 0.72-0.78) as measured by Cohen’s ?.

Physician use of error-reduction techniques was variable. Use of support materials to record or recall handoff data was common, with incoming physicians referencing the EMR and written notes in 716 (72.1%) and 429 (43.4%) handoffs, respectively. Outgoing physicians reviewed the EMR and written notes in 573 (57.8%) and 614 (61.9%) handoffs, respectively. The incoming physician asked questions in 406 handoffs (40.5%), often regarding data omitted during the initial verbal presentation. The incoming physician repeated details for verification in 60 handoffs (6.1%). Seventy-two handoffs (7.3%) were “disposition only” handoffs in which the outgoing physician stated only that the patient would be discharged or transferred soon. The outgoing physician reported information on the wrong patient during 7 handoffs (0.7%), all instances of which were recognized and corrected before the end of the handoff session. During 3 handoffs (0.3%), the incoming physician corrected the report given by the outgoing physician, based on information available in the EMR.

Handoff communication errors

Physical examination handoff errors and omissions were observed in 130 (13.1%) and 447 (45.1%) handoffs, respectively. Laboratory errors and omissions were noted in 37 (3.7%) and 290 (29.2%) handoffs, respectively. Twenty-one handoff sessions including 195 patient handoffs lacked sufficient EMR data describing laboratory results or LOS for inclusion in regression analysis. Based on Student t test, these handoffs did not differ significantly from the included handoffs about location, patient acuity, or propor- tion of attending handoffs. However, analysis did reveal a higher proportion of resident handoffs and lower proportion of team handoffs with a significantly lower mean patient number and shorter handoff duration.

Among the 797 patient handoffs included in analysis, several covariates demonstrated significant associations with communication errors and omissions. Regression model details are in Table 4. More examination errors were associated with longer average handoff duration per patient (relative risk = 1.45; CI, 1.03-2.05). Fewer examination

handoff errors were associated with LOS longer than 6 hours (RR = 0.56; CI, 0.32-0.97), whereas this LOS was associated with a greater number of examination omissions (RR = 1.47; CI, 1.08-2). Examination findings were omitted less frequently during handoffs in which the incoming physician used the EMR (RR = 0.72; CI, 0.54-0.96) or wrote handoff notes (RR = 0.65; CI, 0.5-0.85).

Laboratory errors were significantly more frequent during team handoffs (RR = 11.55; CI, 1.77-75.4) but were less common when the incoming physician used the EMR (RR = 0.93; CI, 0.04-0.65). Laboratory handoff omissions were associated with LOS in a dose-dependent fashion, with risk increasing more than 3 times (RR = 3.52; CI, 1.95-6.36) for LOS between 2 to 6 hours and more than 7 times (RR = 7.42; CI, 4.11-13.4) if LOS exceeded 6 hours. Laboratory omissions were also associated with longer handoff duration per patient (RR = 1.41; CI, 1.11-1.81).

Discussion

This study identified emergency physician handoff practices and communication errors through direct observa- tion. During the handoff observation period, we were able to quantify handoff communication errors and to identify variables independently associated with handoff communi- cation errors. Measuring patient outcomes associated with handoff errors was beyond the scope of this project.

Omission of physical examination findings was common, as 40% of handoffs included no physical examination information. This omission could lead to duplication of provider effort, slower rates of Patient throughput, and reduced comprehension of the patient’s clinical status by the incoming physician. Poor communication regarding the

patient’s active medical problems, baseline status, and the rationale of the previous physician has been associated with critical incidents after inpatient handoff [8]. We do not suggest that all of these communication errors significantly compromisED patient care. However, this level of miscom- munication, along with the degree of variation observed in handoff practices, suggests that EM physician handoffs could be improved by the implementation of a standardized handoff protocol.

We observed variable use of error-reduction techniques during handoff. The timely and unambiguous transfer of responsibility is a critical handoff task in high-risk setting such as railroad dispatch, nuclear power, and NASA Johnson Space Center (Houston, TX) [5]. The ED handoffs we observed were generally timely, as three quarters of handoffs (77.9%) began within 10 minutes of their scheduled time. In two thirds of handoffs (67.9%), electronic records were updated to reflect a new responsible physician within 10 minutes of handoff completion. The readback, or restating information for verification, is another established error-reduction technique implemented in settings as diverse as coffee shops and air traffic control [9,10]. However, this practice was observed in only 60 handoffs (6.1%) in this study.

Outpatient medications were discussed in only a small percentage of handoffs. This finding raises the question of whether a more careful review of medications during handoff might unveil more adverse drug events. Although we have no data to suggest that such events were overlooked in our study, recent literature suggests that medication handoff errors are common and that one third of percentage of ED visits may be related to medication complications [9,11].

Use of the EMR and handwritten support materials were associated with fewer handoff errors and omissions. These visual reminders may provide structure to the handoff, helping physicians focus on key details of the patient

Table 2 Criteria for handoff errors and omissions Physical examination communication errors

Error Acceptable Error

Error Acceptable Acceptable

Error Error Error

Acceptable Acceptable Acceptable

Error Error

Acceptable Error

Error Omission

Acceptable Acceptable Error

Error Error

Acceptable

Normal Global normal Abnormal Pending

Global pending Not reported

		EMR
		Normal	Abnormal	No findings recorded
Verbal handoff	Normal Global normal Abnormal Not reported	Acceptable Acceptable Error Acceptable	Error Error Acceptable Omission	Acceptable Acceptable Error Acceptable

Laboratory communication errors

Verbal handoff

EMR

Normal Abnormal Pending No results

Table 3 Characteristics of observed handoffs General patient characteristics

No. of patients 992

Patient acuity (per ED nursing triage)

1 (most acute) 2 (0.2%)

2 299 (30.1%)

3 572 (57.7%)

4 36 (3.6%)

5 (least acute) 3 (0.3%)

No triage level assigned by time of handoff 80 (8.1%)

Patient identifiers used

Name 756 (76.2%)

Age 409 (41.2%)

Sex 336 (33.9%)

Room number 373 (37.6%)

Handoff topics discussed

Diagnosis 889 (89.6%)

Treatment in ED 392 (39.5%)

Disposition 858 (86.5%)

Home medications 114 (11.5%)

Allergies 6 (0.6%)

Use of support materials during handoff

Outgoing physician used EMR 716 (72.1%)

Outgoing physician used handwritten notes 429 (43.3%)

Incoming physician used electronic medical board 573 (57.8%)

Incoming physician took notes during handoff 614 (61.9%)

Location

Staff work area (out of patient view) 781 (78.7%)

Physician charting station (in patient view) 211 (21.3%)

Bedside 0

Interruptions

Interruptions per handoff session (mean +- SD) 0.8 +- 1.4

Handoffs with at least 1 interruption 101 (10.2%)

Handoffs with interruption by staff 90 (9.1%)

Handoffs with interruption by telephone/pager 17 (1.7%)

Handoffs with interruption for patient care 0

Timing

LOS at time of handoff (h) (mean +- SD) (n = 797) 5.4 +- 4.1

Total handoff duration (min) (mean +- SD) 14.2 +- 6.3

Average handoff duration per patient (min) (mean +- SD) 1.4 +- 0.5

Started within 10 min of scheduled time (% yes) 773 (77.9%)

Responsible physician updated within 10 min of handoff (% yes) 674 (67.9%) Interactive techniques

Incoming physician asked questions to clarify handoff 406 (40.9%)

Incoming physician repeated details to confirm accuracy (“readback”) 60 (6.1%)

Outgoing physician asked for questions 2 (0.2%)

Interactive discussion of medical decision making 163 (16.4%)

Communication errors

Handoffs with at least one examination handoff error 130 (13.1%)

Handoffs with at least one examination handoff omission 447 (45.1%)

Handoffs with at least one laboratory handoff error 37 (3.7%)

Handoffs with at least one laboratory handoff omission 290 (29.2%)

Examination handoff errors (mean +- SD) 0.2 +- 0.4

Examination handoff omissions (mean +- SD) 0.7 +- 1

Laboratory handoff errors (mean +- SD) 0.1 +- 0.3

Laboratory handoff omissions (mean +- SD) 0.4 +- 0.8

Notable events

Outgoing physician reported information on wrong patient (%) 7 (0.7%)

Incoming physician corrected outgoing handoff based on EMR data 3 (0.3%)

“Disposition only” handoff 72 (7.3%)

encounter and removing the burden of relying entirely on memory. Incoming physicians who use these support materials may be reminded to ask questions about missing data, thereby, improving handoff completeness. Standard- ized handoff forms have been associated with improvements in handoff accuracy and completeness [12,13]. Electronic handoff systems may facilitate communication by displaying a standardized set of critical data for physicians to review during signout [14]. Patient safety may also benefit from the increased use of error-reporting tools integrated with an electronic handoff system [15].

The average handoff duration per patient was associated with increased examination errors and laboratory omissions. This result is surprising at first because many people assume that errors will decrease as physicians spend more time discussing each patient. However, as this variable reflects the mean handoff time per patient and not the actual handoff time per patient, the result may instead reflect a situation in which handoff has lasted longer than intended and physicians commit errors or omit findings as they hurry to discuss the remaining patients. Future research could examine the impact of each handoff’s duration on commu- nication errors.

Emergency department LOS was associated with omis- sion of both examination and laboratory results. Similar findings have been noted in patients boarding in the ED [16-18]. Although this was not a study of ED boarding, the

Table 4 Regression models for handoffs

RR

95% CI

RR

95% CI

RR

95% CI

RR

95% CI

issue of delayed care and poor handoff communication for patients staying in the ED for prolonged periods is similar. The increase in errors among such patients suggests that referring to the EMR and handwritten notes during handoffs may be helpful in decreasing communication errors. Of note, however, LOS was also associated with a reduction in examination handoff errors. This result suggests that as a patient’s LOS increases, physicians may continue to accurately describe a small set of examination findings while omitting other examination or laboratory findings from earlier in the patient’s stay.

Previous investigators have noted a higher rate of handoff errors among residents compared to attending physicians [19]. Our findings fail to reproduce this result, as our observation reveals no significant difference in communica- tion errors or omissions between attending and residents. We did observe an elevated risk of laboratory communication errors during team handoffs. Compared to one-on-one handoffs, the many participants in a team or group handoff can make communication more difficult. However, as the only team handoffs at this institution were in the early morning, we cannot exclude an additional effect from physicians finishing an overnight shift. In addition, team handoffs may be affected by the inclusion of teaching time during this group session.

Numerous studies have noted that communication errors, handoff content errors, and interruptions during

Variable	Physical examination			Laboratory results
	Errors	Omissions		Errors	Omissions

Handoff characteristics
No. of Patients 0.99 0.93-1.05	1.02	0.99-1.06	0.91	0.80-1.04	1.02	0.98-1.06
Average handoff time/patient (min) 1.45 1.03-2.05	0.83	0.67-1.02	1.34	0.57-3.13	1.41	1.11-1.81
Patient acuity 1.11 0.88-1.39	0.96	0.87-1.07	1.24	0.73-2.08	0.94	0.82-1.08
No. of interruptions 0.92 0.60-1.42	1.03	0.80-1.33	0.52	0.15-1.81	1.09	0.83-1.45
Degree of verbal interaction 1.20 0.94-1.52	1.00	0.87-1.14	1.32	0.77-2.27	0.94	0.79-1.11
ED LOS at time of handoff
b2 h 1	1		1		1
2-6 h 0.73 0.45-1.18	1.17	0.87-1.57	2.87	0.70-11.8	3.52	1.95-6.36
N6 h 0.56 0.32-0.97	1.47	1.08-2.00	3.25	0.76-13.9	7.42	4.11-13.4
Physician training level
Attending physicians only 1	1		1		1
Residents physicians only 0.97 0.40-2.32	1.16	0.74-1.81	0.48	0.04-5.52	1.31	0.77-2.21
Group handoff 0.45 0.16-1.27	1.02	0.67-1.56	11.55	1.77-75.4	0.99	0.60-1.64
Outgoing physician: means of recalling handoff data
No support materials used 1	1		1		1
EMR 1.27 0.68-2.36	1.19	0.85-1.66	0.95	0.17-5.46	0.96	0.64-1.42
Reviewed written notes 1.01 0.67-1.51	0.92	0.74-1.16	0.70	0.25-1.93	0.86	0.65-1.14
Incoming physician: means of recording handoff data
No support materials used 1	1		1		1
Typed notes or reviewed EMR 0.93 0.56-1.56	0.72	0.54-0.96	0.15	0.04-0.65	0.97	0.69-1.35
Wrote notes 1.09 0.64-1.84	0.65	0.50-0.85	0.39	0.11-1.38	1.16	0.84-1.59
Bolded values indicate regression model variables with P b .05.

handoff contribute to errors in patient care [17,20-25]. Our data confirm that handoffs in the ED are inconsistent and error prone. Multiple organizations have proposed handoff tools to improve transfer of care [26-29]. In addition, communication development programs, such as discussion- based clinical workshops, teamwork training, “handoff clinic,” and simulation-based training, have been devel- oped to improve clinical handoff by nurses and residents [30-33]. Early evidence suggests these programs may improve handoff effectiveness, but additional research is required. Further study using these programs and tools is needed to determine their feasibility and effectiveness for ED handoffs.

Limitations

This study has several limitations. First, we observed physicians in a single institution without a standardized handoff protocol, and thus, our findings may not be generalizable to all institutions. Second, physicians were classified as residents or attendings but were not further stratified within those groups by their level of experience. Recent literature suggests that attending EM physicians with limited experience may commit errors more often than those with more experience [34]. Attending handoffs also comprised a much larger fraction of the total study than did resident handoffs. Third, the handoffs excluded from analysis had a significantly higher fraction of resident handoffs with a corresponding lower mean patient number and shorter handoff duration. However, patient acuity and handoff location did not differ significantly between the 2 groups. Fourth, data collection used a convenience sample that did not include handoffs at 1:00 AM or 3:00 AM, times that are potentially more prone to error. Our data collection also failed to obtain Imaging results for comparison with handoff, although this omission did not affect collection of other handoff data. Fifth, it is possible that ED physicians were able to discern the underlying purpose of this study, thereby, influencing their handoff behavior. However, handoff observers did not allow others to view the data collection forms, and they feigned data collection in the ED during nonhandoff times to minimize any perceived emphasis on handoff observation.

Finally, because we did not take audio recordings of the handoffs, comments that were missed by observers during a handoff were permanently lost. We used this approach to data collection because it represents the same means by which incoming physicians receive information. Recorded handoffs may allow for more complete data collection, but they may not accurately reflect the information received or understood by the incoming physician. If a verbal handoff is sufficiently rapid or quiet as to require playback on a recording device, we considered it unlikely that the incoming physician would

comprehend it. Furthermore, the observers watched many handoffs while developing the data collection forms and were well practiced in accurately recording information presented during verbal reports.

Conclusions

Clinically pertinent findings reported in ED physician handoff often differ from findings reported in chart notes. These communication errors are associated with handoff time per patient, ED length of stay, and use of support materials. Future research should focus on ED handoff standardization protocols, error-reduction techniques in handoffs, and the impact of ED handoff errors on patient outcomes.

References

1. Kohn L, Corrigan J, Donaldson M, editors. To err is human: building a safer health system. Institute of Medicine. Washington, DC: National Academy Press; 2000.
2. HealthGrades. Patient safety in American hospitals, July 2004. Available at: http://www.healthgrades.com/media/english/pdf/HG_ Patient_Safety_Study_Final.pdf. [Accessed November 18, 2009].
3. Kitch B, Cooper J, Zapol W, et al. Handoffs causing Patient harm: a survey of medical and surgical housestaff. Jt Comm J Qual Patient Saf 2008;34:563-70.
4. The Joint Comission. National patient safety goals: hospital. Available at: http://www.jointcommission.org/PatientSafety/National PatientSafetyGoals/09_hap_npsgs.htm. 2009 [Accessed November 18, 2009].
5. Patterson ES, Roth EM, Woods DD, et al. Handoff strategies in seetings with high consequences for failure: lessons for health care operations. Int J Qual Health Care 2004;16:125-32.
6. Solet DJ, Norvell JM, Rutan GH, et al. Lost in translation: challenges and opportunities in physician-to-physician communication during patient handoffs. Acad Med 2005;80:1094-9.
7. Sinha M, Shriki J, Salness R, et al. Need for standardized sign-out in the emergency department: a survey of emergency medicine residency and pediatric emergency medicine fellowship program directors. Acad Emerg Med 2007;2:192-6.
8. Arora V, Johnson J, Lovinger D, et al. Communication failures in patient sign-out and suggestions for improvement: a critical incident analysis. Qual Saf Health Care 2005;14:401-7.
9. Berthold J. Improving handoffs through better communication. ACP Hospitalist, July 2009. Available at http://www.acphospitalist.org/ archives/2009/07/handoffs.htm. [Accessed November 18, 2009].
10. Monan B. Readback hearback. Aviation safety reporting system directline, March 1991. Available at http://asrs.arc.nasa.gov/publica- tions/directline/dl1_read.htm. [Accessed November 18, 2009].
11. Baena MI, Faus MJ, Fajardo PC, et al. Medicine-related problems resulting in emergency department visits. Eur J Clin Pharmocol 2006; 62:387-93.
12. Wayne JD, Tyagi R, Reinhardt G, et al. Simple standardized patient handoff system that increases accuracy and completeness. J Surg Educ 2008;65:476-85.
13. Salerno SM, Arnett MV, Domanski JP. Standardized sign-out reduces intern perception of medical errors on the general internal medicine ward. Teach Learn Med 2009;21:121-6.
14. Sarkar U, Carter JT, Omachi TA, et al. SynopSIS: integrating physician sign-out with the electronic medical record. J Hosp Med 2007;2:336-42.
15. Foster PN, Sidhu R, Gadhia DA, et al. Leveraging computerized sign- out to increase error reporting and addressing patient safety in graduate medical education. J Gen Intern Med 2008;23:481-4.
16. Apker J, Mallak L, Gibson S. Communicating in the “gray zone“: perceptions about emergency physician hospitalist handoffs and patient safety. Acad Emerg Med 2007;14:884-94.
17. Horwitz LI, Meredith T, Schuur JD, et al. Dropping the baton: a qualitative analysis of failures during the transition from emergency department to inpatient care. Ann Emerg Med 2009;53:701-10.
18. Moore C, Wisnivesky J, Williams S, et al. Medical errors related to discontinuity of care from an inpatient to an Outpatient setting. J Gen Intern Med 2003;18:646-51.
19. Singh H, Thomas EJ, Petersen LA, et al. Medical errors involving trainees: a study of closed malpractice claims from 5 insurers. Arch Intern Med 2007;167:2030-6.
20. Borowitz SM, Waggoner-Fountain LA, Bass EJ, et al. Adequacy of information transferred at resident sign-out (in-hospital handover of care): a prospective survey. Qual Saf Health Care 2008;17:6-10.
21. Horwitz LI, Moin T, Krumholz HM, et al. Consequences of inadequate sign-out for patient care. Arch Intern Med 2008;168:1755-60.
22. Jagsi R, Kitch BT, Weinstein DF, et al. Residents report on adverse events and their causes. Arch Intern Med 2005;165:2607-13.
23. Sutcliffe K, Lewton E, Rosenthal M. Communication failures: an insidious contributor to medical mishaps. Acad Med 2004;2:186-94.
24. Williams RG, Silverman R, Schwind C, et al. Surgeon information transfer and communication: factors affecting quality and efficiency of inpatient care. Ann Surg 2007;245:159-69.
25. Laxmisan A, Hakimzada F, Sayan OR, et al. The multitasking clinician: decision-making and cognitive demand during and after team handoffs in emergency care. Int J Med Inform 2007;76:801-11.
26. Coleman E. The care transitions program. Available at http://www. caretransitions.org/. 2007. [Accessed November 18, 2009].
27. Department of Defense Patient Safety Program. Available at: http:// www.dodpatientsafety.net/. [Accessed November 18, 2009].
28. Institute for Healthcare Improvement. Patient safety. Available at http:

//www.ihi.org/IHI/Topics/PatientSafety/. [Accessed November 18, 2009].

1. WHO Collaborating Centre for Patient Safety Solutions. Patient safety solutions. Available at: http://www.ccforpatientsafety.org/. [Accessed November 18, 2009].
2. Berkenstadt H, Haviv Y, Tuval A, et al. Improving handoff communications in critical care: utilizing simulation-based training toward process improvement in managing patient risk. Chest 2008; 134:158-62.
3. Horwitz L, Moin T, Green M. Development and implementation of an oral sign-out skills curriculum. J Gen Intern Med 2007;22:1470-4.
4. Shendell-Falik N, Feinson M, Mohr B. Enhancing patient safety: improving the patient handoff process through appreciative inquiry. J Nurs Adm 2007;37:95-104.
5. Arora V, Johnson J. A model for building a standardized hand-off protocol. Jt Comm J Qual Patient Saf 2006;32:646-55.
6. Berk WA, Welch RD, Levy PD, et al. The effect of clinical experience on the error rate of emergency physicians. Ann Emerg Med 2008;52: 497-501.