Chu, Eugene S. MD; Reid, Mark MD; Schulz, Tara MD; Burden, Marisha MD; Mancini, Diana MD; Ambardekar, Amrut V. MD; Keniston, Angela MPH; Albert, Richard K. MD
Dr. Chu is chief, Division of Hospital Medicine, Denver Health Medical Center, Denver, Colorado, and associate professor of medicine, University of Colorado School of Medicine, Denver, Colorado.
Dr. Reid is assistant professor of medicine, University of Colorado School of Medicine, Denver, Colorado.
Dr. Schulz is a staff hospitalist at Community Memorial Hospital, Ventura, California. At the time this report was written, she was assistant professor of medicine, University of Colorado School of Medicine, Denver, Colorado.
Dr. Burden is assistant professor of medicine, University of Colorado School of Medicine, Denver, Colorado.
Dr. Mancini is assistant professor of medicine, University of Colorado School of Medicine, Denver, Colorado.
Dr. Ambardekar is a fellow in cardiology, University of Colorado School of Medicine, Denver, Colorado. At the time this report was written, he was chief medical resident, Denver Health Medical Center, University of Colorado School of Medicine Internal Medicine Training Program, Denver, Colorado.
Ms. Keniston is professional data analyst, Department of Medicine, Denver Health Medical Center, Denver, Colorado.
Dr. Albert is chief, Department of Medicine, Denver Health Medical Center, Denver, Colorado, and professor of medicine, University of Colorado School of Medicine, Denver, Colorado.
Please see the end of this article for information about the authors.
Correspondence should be addressed to Dr. Chu, Denver Health Medical Center, 777 Bannock, MC 4000; Denver, CO 80204-4507; telephone: (303) 436-6992; fax: (303) 436-7249; e-mail: (firstname.lastname@example.org).
Communication failures are the most common root cause of sentinel events in U.S. hospitals.1 In July 2003, the Accreditation Council for Graduate Medical Education (ACGME) mandated that residency programs decrease resident work hours in an effort to improve patient care and safety. Reduced duty hours have the effect of increasing the number of handoffs of patient responsibility, which can also increase discontinuity of care and worsen patient safety.2–4 Because poor sign-out practices have been directly linked to adverse events,5 the Joint Commission on Accreditation of Health Care Organizations requires that hospitals implement its new National Patient Safety Goal to improve communication between caregivers.6
Although transfers of care are now increasingly frequent, as recently as three years ago, 60% of training programs had no formal process for teaching interns how to hand off patients.5 We suspect that, even now, few use comprehensive systems for conducting handoffs, generally leaving this task to unsupervised interns.5 With only 8% of U.S. medical schools formally teaching handoffs in a lecture/small-group session as of 2005,7 lack of adequate training in this increasingly important aspect of patient care is undoubtedly still prevalent.
In settings in which communication errors have high consequences (e.g., space shuttle mission control, nuclear power plants, railroad and ambulance dispatch centers), Patterson and colleagues8 found that handoff processes used verbal, face-to-face discussions that took place during periods when interruptions were limited. Handoffs also included the transmission of contingency plans with oncoming personnel actively participating in the process through interactive questioning. Health care systems have been slow to implement similar standardized, structured processes.7,9
In July 2005, in response to ACGME duty hours regulations, the resident work schedule on the medicine teaching service at our teaching hospital was changed from an every-fifth-night call system with day floats to a shift system in which all housestaff worked defined shifts with fixed starting and stopping times. We recognized that this new system would result in more handoffs of care and that some of these handoffs would occur at times when patients still had numerous active issues. Accordingly, we also decided to improve handoffs by developing formal written and verbal (i.e., oral) templates as part of a highly structured process. By teaching how it should be done and by providing supervision, we attempted to ensure that patient-care handoffs were being conducted as designed. In the remainder of this report, we explain our new approach and report our evaluation of it to date.
We conducted this study at Denver Health Medical Center, a public safety net hospital that is part of the University of Colorado School of Medicine’s internal medicine residency program. The study was approved by the local institutional review board.
The handoff structure and process
A focus group of 3 internal medicine attendings and 12 residents was assembled in the spring of 2005 with the charge of creating a new process for conducting and teaching handoffs. Members of the work group observed the verbal and written handoff processes used by nurses on several hospital units and by physicians working in the emergency department and on the labor deck. They also reviewed literature pertaining to handoffs of responsibility both inside and outside the heath care industry. The group’s concluding consensus was that handoffs should be conducted face-to-face, occur at a fixed time and place each day, use a standard verbal template, and include discussion of information that is displayed on a standard written template that would be available to all through the hospital-wide computer system. Accordingly, verbal and written templates were developed, and evening handoffs, which in the past had been conducted at varying times of day with no formal process, were restructured to occur at a consistent, specified time and place.
In June 2006, a second focus group of three attendings and eight residents (about an 80% overlap with the members of the 2005 group) met to develop what they perceived to be the best way to teach and supervise the handoff process. Based on their recommendations, the verbal portion of the handoff was restructured so that the departing intern provided a short background statement, an accounting of all active problems, a description of contingency plans, and a list of all pending tests and issues. The oncoming intern was encouraged to engage in interactive questioning and use read-backs (i.e., repeating what he or she perceived as critical information). At this time, the written template underwent minor modifications. The results of the second workgroup are the basis of the current standards and structures of our handoff program, which includes geographic, temporal, written, verbal, education, and supervision elements and standards as described in the list below. The structure and processes developed for our program paralleled handoffs in systems such as nuclear power plants and NASA’s mission control as described by Patterson and colleagues.8
* The geographic element of the program locates handoffs in a quiet, resident work room with computer, and therefore hospital server access.
* The temporal element of the program requires that handoffs occur between 18:00 and 19:00 on weekdays and between 16:00 and 17:00 on weekends.
* The written element consists of a Microsoft Word table that is templated with standard content and is accessed on a hospital server.
* The verbal element is the mnemonic SAIF-IR, which is described later in the text.
* The educational component requires hospital medicine attendings to teach and give feedback to interns receiving patient handoffs on their first night of call. There is also a lecture on handoffs during first-week-of-ward month.
* The supervision component requires that hospital medicine attendings be present on each intern’s first night of call for the first four months of the academic year.
Before starting the supervision and education process, the six handoff teaching attendings conducted a series of meetings. During each meeting, some time was spent practicing handoffs using the recommended written and verbal templates. Attendings would receive peer feedback to standardize their own handoffs.
During the first six nights of each month long rotation, attendings supervised the handoff process between interns who were starting their night shifts and interns who were finishing their day shifts. Each intern was therefore exposed twice to attending supervision and teaching of the handoff process when finishing a shift and transferring care and once when starting a shift and assuming care of patients. This process was continued for the first four months of the academic year. Because multiple attendings were involved, all met on a monthly basis to standardize and refine their teaching approach.
We developed a two-part survey to evaluate the effectiveness of the process and obtain feedback on its different components. The survey was administered to each of the 72 interns on the first day and last week (to decrease expectation bias but preserve response rate) of their month long rotations for a total of 144 surveys. All responses were anonymous. The first part of the survey was designed to assess the interns’ perceptions of their knowledge of, and their attitudes toward, the handoff process, along with their perceived ability to transfer patient care. In this section, we asked them to respond to a series of statements using a 10-point Likert scale, with 1 indicating that they strongly disagreed with the statement and 10 indicating that they strongly agreed. The second section of the survey was designed to assess the interns’ perceptions of the effectiveness of each of the components of the handoff process, as well as how long they thought that supervision and teaching should occur. For this, we used a 5-point Likert scale, with 1 indicating that the component was not at all useful and 5 indicating that it was extremely useful. To address the possibility of “leading” questions, our survey consisted of neutral statements such as “I know how to handoff patients when I leave work at the end of the day” and “Handoffs were well taught in my medical school.” The same statements were used in pre and post surveys. Unstructured comments were also solicited.
All analyses were performed using SAS, version 8.1 (SAS Institute, Inc., Cary, North Carolina). A Student t test was used to compare continuous variables, and a chi-square test was used to compare categorical variables recorded at the beginning of each rotation with those recorded at the end. To assess the effect of training, responses recorded during the first three months of the academic year were compared with those recorded during months four though six. Bonferroni corrections were used for multiple comparisons such that P values of <.001 and <.025 were considered significant for continuous and categorical variables, respectively. All data are reported as means with standard deviations.
Of the 144 surveys distributed to 72 interns, 137 were returned for a response rate of 95%. Ten of these were received from interns rotating through Denver Health a second time during the six-month period of study.
During the first three months of the academic year, the interns’ perceptions of their knowledge about how to hand off patients, assume care of patients from other services, perform read-backs, make contingency plans, and efficiently communicate information on the first day of a rotation ranged from 4.1 (SD = 2.5) to 5.5 (SD = 2.2) on the 10-point scale (Table 1). After having been taught the structured process, using the process for one month, and having the process supervised by attendings during their first night of assuming care while on call and their first two nights of transferring care, these same perceptions all increased significantly (Table 1).
Interns’ responses at the beginning of the rotation were higher in months four through six than they were in the beginning of months one through three, consistent with the learning process that comes with being an intern (Table 1).
By comparing the responses of interns finishing their first three months of Denver Health wards after being supervised and taught handoffs with the responses of interns starting Denver Health wards in months four to six of the academic year, the post survey scores for months one to three were higher than the pre survey scores for months four to six (Table 2). And, by the end of the interns’ rotations during months four through six, scores had increased even further.
Eighty-four percent of all the responses to queries about the usefulness of the different elements of the handoff program indicated that the interns judged the overall program to be “useful” or ”extremely useful,” including the component of attending supervision (Table 3). The aspect of the program thought to be most useful was conducting the handoffs at the same time and in the same place on a daily basis. The aspect thought to be least useful was the lecture/small-group session to explain the process (Table 3). In addition (not in the table), the clear majority of interns preferred that attendings should teach the handoff process one time (23 interns; 36%) or two times (24; 38%) per month during the first three months of their training and should supervise handoffs three months (33 interns; 52%) or six months (14; 22%) per year.
Regardless of when they were surveyed, interns felt strongly that handoffs were important for patient safety and that standardizing the process was important. Scores in response to whether they felt handoffs were well taught in medical school were low (Table 1). None of the preceding attitudes changed during the rotation or over time.
Two themes were generated from the unstructured comment section of the survey:
1. Attending supervision of handoffs should be a standard part of teaching the handoff process. Representative comments included “It would be great to have this . . . even if not at Denver Health” and the attending “handoff schedule is a good idea, wish all hospitals were like this.”
2. The accuracy of the data summarized on the written handoff template needed to improve. Representative comments included “[Assembling handoff] information should be computerized [to] decrease errors” and “An excellent, up to date sign-out sheet is most important.”
The important findings of this study were that interns at Denver Health felt unprepared to hand off patient care adequately, particularly in the first three months of the academic year, and that creating geographic and temporal handoff structures that used written and verbal templates, together with teaching and providing attending supervision of the process, were all well received and thought to be important components of learning how to transfer patient-care responsibilities.
Our written template evolved from written three- by five-inch cards to a computerized spreadsheet that was designed to include standard information presented in a standard format. Finding that 93% of our interns felt that our computerized template was a useful or very useful part of the handoff process is consistent with the observations of Lee and colleagues9 and Petersen and colleagues,10 who noted that standardized sign-outs and computerized systems resulted in the transfer of more complete data and fewer adverse events. The interns’ open-ended comments did note, however, that the accuracy of the written handoff data was still suboptimal and that ways to improve this would further improve the quality of the handoff process. We recognize that computerized processes that electronically transfer data from one field to another can improve the handoff process11; however, the technology that was available at Denver Health when this process was implemented precluded this approach. Our experience suggests, however, that interns might not scrutinize electronically transferred data on a day-by-day basis, so this “improvement” might also have some limitations.
Vidyartha and colleagues12 have proposed that verbal handoffs should be structured using the SBAR acronym (situation, background, assessment, recommendation), but this tool was developed by Leonard and colleagues13 for a specific purpose (i.e., for use by nurses when they are reporting changes in patient status to physicians). Similarly, the SOAP system (subjective, objective, assessment, plan) was developed specifically to provide a structure for daily patient progress. We considered using the SBAR approach (particularly since it has recently been adopted by our nursing staff as their standard format for communicating with physicians) and the SOAP system, but we felt that neither provided a suitable verbal structure for patient-care handoffs. Handoffs are not generally situation based, and they frequently include instructions for oncoming interns to follow up on pending tests or procedures. Although we did not employ a mnemonic to help the interns learn our structured handoff process, our experience has led us to develop the SAIF-IR system as follows:
Offgoing provider performs an SAIF handoff
* Summary statement(s). These are one to three sentences summarizing a patient’s hospital stay. These are not a repeat of the history of present illness.
* Active issues. Although the written template lists all issues including chronic conditions, we encourage our housestaff to only verbalize active medical issues.
* If–then contingency planning. These are clues to the oncoming provider of potential issues arising and what the offgoing provider would suggest on the basis of his or her clinical knowledge of the patient.
* Follow-up activities. These are the tests, procedures, or therapeutics which need to be reevaluated by the oncoming provider.
Oncoming provider makes the handoff SAIF-IR
* Interactive questioning. These are questions to clarify or correct information presented by the offgoing provider.
* Read-backs. These are repetitions of important information to ensure understanding.
The goal of the above system is to facilitate the handoff process using a standardized approach that emphasizes the importance of handoffs and that links the process to safer patient care.
Standard processes are recognized as key elements in many quality-improvement activities (e.g., the Toyota Production system14). The literature suggests, however, that unstructured handoffs are still prevalent in medicine.5,12 Of the five hospitals in the University of Colorado internal medicine training program, only Denver Health mandated a designated time and place for handoffs to occur 365 days per year and developed a formal process by which the transfer of care responsibility should occur. In response to the results of this study, however, the SAIF-IR handoff structure has been included in the curriculum at all the hospitals in our system since July 2007.
The evaluation of our program has a number of limitations. The outcomes we obtained were subjective assessments of our interns’ understanding, perceptions, and opinions. Although determining the effect of the handoff program on more objective outcomes such as mortality or length of stay would provide more convincing evidence that our handoff structure and teaching process were effective, doing so was well beyond the scope of this project. In addition, if subsequent studies found that more objective outcomes did not change as a result of this program, we would not have abandoned our approach, because the interns thought the program improved communication, reduced the discontinuity of care, and added to their education.
Our study design did not include a concurrent control group of residents who did not use the handoff process and were not exposed to the teaching process. Although having such a control would have allowed us to more specifically assess the effect of the program, it would not have been possible to keep the two groups of interns separate at one hospital, and it would have been contrary to the underlying educational objectives of our efforts. However, our results suggest that our program has an effect beyond practice alone. As stated earlier, by comparing the responses of interns finishing their first three months of Denver Health wards after being supervised and taught handoffs with the responses of interns starting Denver Health wards in months four to six of the academic year, the post survey scores for months one to three were higher than the pre survey scores for months four to six (Table 2). This finding suggests that the teaching and supervision of handoffs by our attendings improved the self-perceived knowledge and skills of our interns more than did unstructured practice alone.
Most of the interns indicated that having attendings supervise the process the first two nights they were on call at the beginning of the academic year would provide the best teaching of the process. Not having the resources to supervise more than their first night on call may have limited the effectiveness of the process. Its effectiveness may also have been limited by the fact that, at the time we implemented our handoff process, Denver Health was the only one of five hospitals in our system that was focusing on the handoff process. Finally, it was difficult to distinguish what effect interns coming from shift-based training programs such as emergency medicine and obstetrics and gynecology may have had on our results. To preserve anonymity, we did not ask our interns to identify their primary training programs.
In summary, the results of our study indicate that formal teaching and supervision of a highly structured process for transferring care was well received by the participating interns and improved their self-perceived knowledge of handoffs and their comfort in assuming responsibility for patients transferred to their care. We hope that our report will be a stimulus for other residency programs to implement similar instruction, which we suggest should be added to the internship curriculum early in the academic year.
The authors thank Dr. Carol Hodgson and Dr. Gretchen Guiton for their input into the survey design and overall development of this project. This project was done in conjunction with the Teaching Scholars Program at the University of Colorado at Denver School of Medicine.