|Home | About | Journals | Submit | Contact Us | Français|
Communication failures contribute to errors in the transfer of patients from the emergency department (ED) to inpatient medicine units. Oral (synchronous) communication has numerous benefits but is costly and time-consuming. Taped (asynchronous) communication may be more reliable and efficient, but lacks interaction. We evaluated a new asynchronous physician-physician sign-out compared to the traditional synchronous sign-out.
A voicemail-based, semi-structured sign-out for routine ED admissions to internal medicine was implemented in October, 2007 at an urban, academic medical center. Outcomes were obtained by pre- and post-intervention surveys of ED and IM housestaff, physician assistants and hospitalist attendings; and by examination of access logs and administrative data. Outcome measures included utilization; physician perceptions of ease, accuracy, content, interaction and errors; and rate of transfers to intensive care from the floor within 24 hours of ED admission. Results were analyzed both quantitatively, and qualitatively using standard qualitative analytic techniques.
During September-October, 2008 (one year post-intervention), voicemails were recorded regarding 90.5% of medicine admissions; 69.7% of these were accessed at least once by admitting physicians. The median length of each sign-out was 2.6 minutes (IQR 1.9 to 3.5). We received 117/197 responses (59%) to the pre-intervention survey and 113/206 responses (55%) to the post-intervention survey. A total of 73/101 (72%) respondents reported dictated sign-out was easier than oral sign-out and 43/101 (43%) reported it was more accurate. However, 70/101 (69%) reported interaction among participants was worse. There was no change in the rate of ICU transfer within 24 hours of admission from the ED in April-June, 2007 (65/6,147; 1.1%) versus April-June 2008 (70/6,263; 1.1%); difference of 0%, 95% CI, −0.4% to 0.3%. The proportion of internists reporting at least one perceived adverse event relating to transfer from the ED decreased a non-significant 10% after the intervention (95% CI, −27% to 6%), from 44% pre-intervention (32/72) to 34% post-intervention (23/67).
Voicemail sign-out for ED-internal medicine communication was easier than oral sign-out without any change in early ICU transfers or the perception of major adverse events. However, interaction among participants was reduced. Voicemail sign-out may be an efficient means of improving sign-out communication for stable ED admissions.
The transfer of a patient from the emergency department (ED) to the inpatient floors is a complex process fraught with potential for error.1–3 Although errors at this transition of care are multifactorial,1 sign-out communication plays an important role.1–3
Sign-out between emergency medicine (EM) and internal medicine (IM) physicians is typically oral (or “synchronous”), involving a face-to-face or telephone conversation about the patient. Synchronous communication has numerous advantages, including opportunities for confirmation, clarification, error detection, relationship-building, negotiation about management and disposition, and multidirectional information flow.4–7 Studies of healthcare workers show they are overwhelmingly inclined towards synchronous communication.8 The emergency department is no exception: 80% of ED communication is typically synchronous.9–11
However, synchronous communication between EM and IM physicians can be challenging to conduct because of the physical distance between the ED and the inpatient unit, multiple competing demands on physicians’ time, rapidly changing information, professional differences in expectations, the chaotic nature of the ED, and frequent physician turnover.1, 2, 6, 12 At times, these challenges inhibit the quality of communication or prevent it from occurring at all. In addition, the interruptions in work required to initiate the conversation, as well as the interruptions that occur during the conversation, may themselves pose a threat to patient safety.8, 13, 14
Written or recorded (“asynchronous”) communication is an alternate format that has several advantages. It is efficient, allowing information to be provided and received at the most convenient time for workflow.15, 16 It is durable, eliminating the “game of telephone” that occurs with sequential oral communication and reducing the risk of missing sign-out altogether.16–18 In some cases, it has been shown to promote more complex and sophisticated interactions, as participants have time to reflect before providing information or asking questions.16, 19, 20 Finally, it can be accessed repeatedly, simultaneously and by a variety of people with different needs.21, 22 Nurses have long used asynchronous communication methods for handoffs.23–27
Nonetheless, asynchronous communication has important limitations. It can not be adjusted in response to the experience or understanding of the recipient, may reduce opportunities for error detection and feedback, limits relationship-building and restricts the ability of the recipient to influence either the discussion or future action.16, 20 If the team is thus prevented from coming to a shared mental model, performance can suffer.28 Perhaps for these reasons, asynchronous handoff among physicians is relatively uncommon.29 One exception in the ED is the “white board” on which key information can be seen at a glance, and which can also serve as a means of two-way communication.22, 30
Qualitative analysis revealed that the existing oral sign-out between EM and IM physicians at our institution suffered from the typical challenges of synchronous communication, leading to dissatisfaction among both EM and IM physicians.1 As one part of a multifaceted initiative to improve handoffs and workflow, internal medicine and emergency medicine clinical leaders implemented a new, asynchronous sign-out system in which EM physicians dictated a semi-structured sign-out to a voicemail system. IM physicians listened to the voicemail and called with follow-up questions as needed.
Asynchronous and synchronous communication has not been directly compared to our knowledge in interspecialty handoffs. This initiative was expected to improve efficiency, reduce the frequency of missed sign-out and increase the participation of ED physicians with direct knowledge of the patient. However, because of the drawbacks of asynchronous communication, its potential overall effect on patient care was uncertain. We thus describe and evaluate this quality improvement initiative with particular attention to the quality and safety of handoff communication.
The intervention was structured as a quality improvement initiative, featuring rapid evaluation, feedback and modification as needed. An important feature of such “action research” is that the intervention often changes mid-cycle in response to continuous evaluation and feedback.31 This approach is akin to the Plan-Do-Study-Act method pioneered by Shewart, in which the intervention is studied and modified in real time.32 We describe both the initial intervention and modifications made in response to feedback.33 The study was designed as a prospective, pre-post analysis involving both quantitative and qualitative assessment.
The study was conducted at a 944-bed urban, academic medical center with both EM and IM residency training programs. The hospital uses a computerized physician order entry system, but at the time of the study ED orders, notes and vital signs were not always available in the electronic medical record. Medicine patients are divided approximately equally between a teaching service and a non-teaching service. Sign-out is conducted by ED residents or physician assistants. Sign-out is received by house staff for patients admitted to the teaching service and by attending hospitalists for patients admitted to the non-teaching service.
Pre-intervention, when emergency physicians determined that a patient required admission to internal medicine they alerted the admission office to find a bed. When a bed and team was identified, the admitting office notified the ED. The EM physician then paged the admitting resident. EM physicians sometimes had to repeat the page or contact several physicians before reaching the correct provider. Likewise, admitting residents sometimes had to call the ED repeatedly to reach the ED physician. When communication with the admitting physician was established, the ED physician provided sign-out, answered any questions, and sometimes altered patient management or disposition based on comments from the admitting team. Transport was then called to bring the patient to the floor. Patients admitted to the hospitalist service were signed out to one hospitalist who held the admitting pager. At times, this hospitalist then repeated the sign-out to a different attending hospitalist who would be the attending of record.
In October, 2007 we implemented a commercial dictation program, CBaySystems (Annapolis, MD). In the new system, when ED physicians determined that a patient required admission to internal medicine they alerted the admission office to find a bed and then dictated a sign-out if the patient was stable. ED physicians were also asked to follow a new, semi-structured format for sign-out that included provisional diagnosis, ED course, pending data and the name and phone number of the ED physician to call in case of additional questions. The dictation was retained as a voicemail but was not transcribed.
When the patient was assigned a team, the ED business associate sent a text page to the admitting team with the patient’s name and medical record number. Simultaneously, the transfer process was set in motion to bring the patient up to the floor. It was the responsibility of the admitting team to listen to the voicemail upon receiving the page. ED staff continued to page receiving physicians directly for real-time discussions about patients admitted to the medical intensive care units or step-down units. Most ED physicians had portable telephones, allowing the admitting team to reach the physician directly without paging or calling the main ED desk. Only internal medicine patients were included in the intervention since patients admitted to other specialties had to be seen in the ED by the accepting physician before they could be admitted.
We measured uptake of the intervention 5–6 months after implementation by obtaining records of all voicemails left on the system and all accesses to the system between February 1, 2008 and March 31, 2008. We measured maintenance of the intervention at one year by reviewing all accesses between September 1, 2008 and October 31, 2008.
Intensive care unit transfer has been identified as a potential adverse event following admission from the ED to the floor.1 Consequently, we identified all instances of patients admitted from the ED to an inpatient unit and then transferred to the intensive care unit within 24 hours from April 1-June 30, 2007 (prior to the intervention) and April 1-June 30, 2008 (after the intervention).
We assessed perceptions of sign-out quality, accuracy, efficacy, ease of use, interaction, content and adverse events through a survey, which included 22 5-point scale questions and 3 open-ended questions about adverse events, use of the voicemail system and general comments. The questionnaire was distributed to all EM house staff and physician assistants, IM house staff, and IM hospitalist attendings in March-April, 2007 and then again in April-May, 2008 (six months after the intervention). The post-intervention survey included additional questions about experiences with the voicemail system. For technical reasons the survey was distributed to all IM house staff, but we excluded all intern responses because they did not routinely receive sign-out before or after the intervention. These surveys were pilot tested for clarity and content by chief residents in EM and IM. The questionnaires were self-administered and anonymous and were sent via e-mail three times and distributed at conferences. A lottery for one of three $50 Amazon.com gift certificates was a financial incentive for participation in each round.
The Yale Human Investigation Committee approved the study, granted a waiver of signed informed consent to preserve anonymity of respondents, and granted a Health Insurance Portability Act and Accountability waiver. Return of the survey was considered consent to participate.
We sent the survey to all medical staff who were routinely involved in providing or receiving sign-out about medicine admissions from the ED. This included house staff, physician assistants and attending hospitalists.
We used qualitative data analysis methods34 to analyze open-ended comments about the voicemail system. The study team included EM physicians (R.K., T.M., V.P., J.S.), and IM physicians (L.I.H., G.J., N.S.). Several participants were experienced in qualitative analysis (L.I.H., T.M., J.S.). We began with a start list of codes based on a separate analysis we had conducted of flaws in the ED to inpatient transfer.1 Three study investigators (L.I.H., R.K., V. P.) independently reviewed each comment to assign codes. At subsequent coding meetings, we added, subtracted and revised codes as needed using the constant comparative method.35 Disagreements were resolved by negotiated consensus. We continued this process until thematic saturation was achieved, i.e., no new concepts were being generated. The full study group periodically reviewed the code structure. We used Atlas.ti 5.0 (GmbH, Berlin, Germany) to facilitate qualitative analysis.
We used descriptive statistics to characterize the data. We compared categorical results among groups using the chi square test. We used SAS 9.2 (SAS Institute, Cary, NC) to facilitate quantitative analysis.
We received a total of 117/197 responses (59%) to the pre-intervention survey, excluding IM interns. These included responses from 39/60 ED house staff and physician assistants (65%), 21/37 hospitalists (57%), and 57/99 internal medicine house staff (58%). We received a total of 113/206 responses (55%) to the post-intervention survey. These included responses from 39/63 ED house staff and physician assistants (62%), 34/44 hospitalists (77%), and 40/99 internal medicine house staff (40%). There was no difference between respondents and non-respondents by specialty. Internal medicine respondents to the post-intervention survey were more likely to be hospitalists and less likely to be senior residents than non-responders (Table 1).
Respondents wrote free-text comments about the voicemail system in 13/39 (31%) EM pre-intervention surveys, 50/100 (50%) IM pre-intervention surveys, 14/39 (36%) EM post-intervention surveys and 34/79 (43%) IM post-intervention surveys.
From February 1, 2008 to March 31, 2008, sign-outs were recorded regarding 1,643 of the 1,836 patients admitted from the ED to medicine units (89.5%). A total of 1,074 were accessed at least once (65.4% of voicemails; 58.5% of admissions). The median length of each sign-out was 2.9 minutes (IQR 2.2 to 3.8 minutes). Those that were accessed were listened to a mean of 1.2 times each (SD 0.5).
Utilization was sustained one year after the intervention. Between September 1, 2008 and October 31, 2008, sign-outs were recorded regarding 1,724 of the 1,910 patients admitted from the ED to medicine units (90.3%). A total of 1,201 were accessed at least once (69.7% of voicemails; 62.9% of admissions). The median length of each sign-out was 2.6 minutes (IQR 1.9 to 3.5 minutes). Those that were accessed were listened to a mean of 1.2 times each (SD 0.5).
There was no change in the rate of ICU transfer within 24 hours of admission from the ED in April-June, 2007 (65/6,147; 1.1%) versus April-June 2008 (70/6,263; 1.1%); difference of 0%, 95% CI, −0.4% to 0.3%.
The proportion of internists reporting at least one perceived adverse event relating to transfer from the ED decreased a non-significant 10% after the intervention (95% CI, −27% to 6%), from 44% pre-intervention (32/72) to 34% post-intervention (23/67). The change in EM perceptions was similar: a non-significant decrease of 11% (95% CI, −23% to 2%), from 5/37 (14%) to 1/37 (3%).
Prior to the intervention, the majority of respondents of both specialties had been skeptical that a voicemail system would result in improvements in the ease of providing or receiving sign-out, the accuracy of content, or the likelihood of errors (Table 2). Open-ended comments from both specialties about the anticipated effects of a voicemail system were overwhelmingly negative (Table 3). Respondents were most concerned that voicemail would lead to a reduction in real-time interaction, commenting that interaction improved retention of information; increased mutual understanding; altered diagnostic, therapeutic and disposition plans; reduced errors; and provided opportunities for real-time feedback. IM physicians were also concerned about the possibility of waning influence on ED management and disposition. Most respondents were skeptical about potential improvements in workflow, feeling that both parties valued interaction so highly that it would continue in addition to the voicemail. Respondents were similarly pessimistic about the effect on the quality, comprehensiveness and accuracy of sign-out content, and felt the system would have a negative impact on patient care.
Six months after the intervention, a total of 52/107 (49%) respondents felt communication with the other specialty was good or excellent compared with 58/113 (51%) in the baseline period (difference −3%; 95% CI, −16% to 11%). A majority of internists (56%) and all EM physicians felt sign-out was easier. The ease of contacting counterparts in the other specialty, the helpfulness of sign-out for patient care, sign-out accuracy, the likelihood of errors, and nearly all the content of sign-out were felt to be unchanged (Tables 4 and and5).5). Internal medicine respondents felt the clinical condition of the patient was less likely to be conveyed in the voicemail system (“always or usually stated”: 45% vs. 26%; difference −20%; 95% CI, −35% to −4%). Both EM and IM respondents agreed that the frequency of questions asked about patients decreased after the intervention (Table 5), and that the amount of interaction declined (Table 4).
Qualitatively, emergency physicians were generally strongly in favor of the voicemail system, reporting substantially improved efficiency. Internists also reported improved efficiency in open-ended comments. However, many were concerned by reduced interaction with EM physicians. Internists also commented that although sign-outs were longer and more thorough, they were not correspondingly more useful. In fact, internists felt recorded sign-outs contained too much factual information about labs and studies that were readily obtained elsewhere and too little information about difficult-to-obtain details, such as what had been done for the patient and how the patient responded. (Table 6)
EM physicians reported several situations in which they did not dictate sign-out: when there were technical barriers such as a busy phone line or a receiving physician who did not have access, when the IM physician was already in the ED and could receive sign-out in person, when the patient was particularly complex, or when the ED provider was shortly leaving the hospital and wanted to have an in-person discussion. IM physicians reported several additional reasons, chiefly involving workflow mismatch; perceived low utility of sign-out; and reduced influence on management and disposition. (Table 6)
Respondents from both specialties commented that some of the problems of the previous system had been remedied by the voicemail. They were particularly pleased that sign-out was generally recorded by the EM physician who knew the patient best, and that it remained accessible to new internal medicine teams or providers. However, they reported several new problems. The access number for the voicemail system was frequently busy. If the text page notification about the new admission went awry, new patients could arrive to the floor without an admitting physician being aware of them. Finally, the new parallel rather than sequential admission process meant that patients could arrive on the unit before or while the receiving physician was listening to the voicemail, leaving little opportunity to affect ED disposition or treatment plans. Indeed, 54/69 (78%) internists said they always or usually accessed the voicemails, but only 10/66 (15%) reported they always or usually had time to do so before the patient arrived on the floor. Of note, while 100% of EM physicians reported they always or usually recorded a sign-out, only 11/37 (30%) reported always or usually doing so prior to booking the patient for admission.
Based on this feedback, leadership made several immediate modifications to the system. They obtained a new voicemail access number that was rarely busy. They required the receiving team to call the ED to confirm receipt of the text page notification before transfer procedures were initiated. They altered the instruction card so that it emphasized information requested by IM physicians and eliminated information that was unnecessary or readily available elsewhere. They asked IM physicians to identify high and low quality sign-outs, and planned to implement sign-out training for ED residents that included playbacks of those sign-outs.
This study has several limitations. First, the majority of our results are subjective, although we included one objective measure of adverse events, ICU transfers. We attempted to mitigate recall bias by prospectively collecting pre-intervention data. Nonetheless, we did not evaluate the sign-outs themselves, nor did we examine efficiency measures such as ED length of visit, which was confounded by simultaneous initiatives at our institution. Second, the study was conducted at a single, academic institution. Results might differ at other institutions, particularly those without teaching programs, for which recorded sign-out quality might be higher. Third, overall response rates in both periods were less than 60% and the post-intervention survey included few third year IM residents. However, house staff and hospitalist results did not differ significantly, and the post-intervention hospitalist response rate was 77%, reducing the likelihood of bias. Fourth, some respondents participated in both the pre- and post-intervention surveys. As we could not identify which these were, we were unable to adjust our statistical testing to account for some paired results. Fifth, we could not identify how many sign-outs were not accessed because they occurred in person instead, although we know from survey comments that this occurred with some regularity. Thus, we overstate the number of missed sign-outs. In addition, since we were not able to identify how many sign-outs were missed in the baseline period, we were unable to compare this outcome. Finally, we did not resurvey physicians after modifying the intervention.
This report describes the effects of a switch from synchronous to asynchronous communication between internal medicine and ED physicians about newly-admitted patients. Overall, physicians reported improvements to workflow, ease and efficiency without adverse effects on patient safety, despite having been deeply skeptical about the intervention beforehand. ED utilization was high and was sustained at one year. As predicted by participants, however, the new system substantially reduced interaction among physicians, causing some dissatisfaction among internists. In addition, 30% of voicemails were not accessed by internists.
All interventions require tradeoffs. In this case, we traded increased efficiency for decreased interaction. We were unable to detect any effect of this tradeoff on perceptions of patient safety or on early ICU transfers, perhaps because both forms of communication have positive and negative effects on safety. Studies of nurses and physicians have found that asynchronous communication can lead to more time spent on patient care and less on failed communication attempts, potentially improving care.15–17, 30 However, reduced two-way conversation may worsen care by diminishing opportunities to correct misunderstandings, identify errors, discuss care, change management or alter disposition.7, 28, 36 It may also impede team-building and feedback.4 We did not directly examine effects on education, interpersonal relationships and teamwork in this study.
On the other hand, while synchronous communication when it does occur may have numerous benefits both for safety and team-building,4–7 it is highly interruptive.13, 37 Interruptions are a particular burden for emergency department personnel.30 Emergency physicians are interrupted 5–20 times an hour10, 30, 38–40 – more than twice as often as primary care physicians.39 Not only do interruptions make work less efficient, they can lead to errors.13, 30, 37, 41, 42 For example, one half of aviation accidents attributed to crew error are precipitated by interruptions, distractions or preoccupations.14 In addition, synchronous communication is vulnerable to conflict, competition, hierarchy, peer pressure, deference, role expectations and other factors6, 43–45 that might be mitigated by less confrontational asynchronous communication methods.21
Although information was received in sufficient time to assist the admitting team’s care, it was often only accessed after the patient had reached the floor. This reduced opportunities for internists to affect care in the ED. Such diminished influence may have prompted some internists to refuse to participate in the process at all. In fact, a striking finding of this study was that – despite the importance both specialties placed on sign-out in surveys – 30% of the voicemails were not accessed.
Respondents gave several reasons for not accessing the voicemails, including receiving sign-out in person instead, not having time to listen before the patient arrived on the floor, or being hindered by technical barriers which were later solved. A few internists, however, reported the sign-out did not add enough value to justify the access time. Three factors may have contributed to this perception. First, internists perceived some sign-outs to be a litany of sometimes irrelevant facts instead of a synthesis that allowed participants to come to a shared mental model of the patient.28, 46 Such communication was not useful and therefore felt not to be worth accessing. Second, the ED-to-floor handoff is unique in that the receiving physician must conduct a thorough history and examination of every new patient no matter how effective the sign-out. It may therefore be rational for a busy receiving physician not to listen to information s/he will feel obligated to re-obtain. Again, encouraging EM physicians to provide interpretation during sign-out (“she responded best to 20mg of furosemide”) rather than only facts (“her creatinine was 0.9”) might improve its perceived value. Third, some internists declined to participate because they had less opportunity to influence ED management and disposition – “The patient is sent to the floor regardless, so I don’t waste my time listening.” This suggests that successful asynchronous models must either deliberately encourage bidirectional influence (perhaps by altering timing or including a mandatory follow up call), or provide sufficient value for recipients by other means.
Another notable finding was that EM physicians gave the quality, content and accuracy of their sign-out communication much higher ratings than IM physicians did. On the one hand, self-assessment correlates poorly with external assessment.47, 48 On the other hand, IM perceptions might have been distorted by negative stereotypes of EM physicians1, 2, 30 or by conflicting expectations about the purpose and content of sign-out.1, 2 Objective external assessment of sign-outs in future studies would be helpful to resolve these discrepancies and might point to areas in which sign-outs could be improved.
The regulatory and legal implications of communication models are always important to consider. The Joint Commission mandates that handoff communication include “an opportunity to ask and respond to questions,”49 and allows taped reports if there is a mechanism to facilitate such questions in a timely fashion.50 To fulfill this mandate, we required every sign-out to conclude with a direct phone number to the ED physician, and to be completed prior to the patient leaving the ED. The durability of asynchronous communication may raise concern about future legal liability. We encourage our staff to assume that this communication – like the written sign-out sheets used for shift-to-shift handoffs – is potentially discoverable, although it is not officially part of the medical record and is only retained for 30 days.
Our intervention initially included two asynchronous communications (the sign-out itself as well as the text page notification about the new admission). Respondents identified this as a major safety concern, since if the text page notification was not received, the patient could arrive on the floor without anyone’s knowledge. We rapidly solved this problem, while still retaining the asynchronous sign-out, by asking internists to call the ED to verify receipt of the page. This was one of several problems we were able address as they were identified, following the PDSA model.32 Improving the quality and utility of the sign-out itself, however, will require curricular and feedback efforts. We plan to use the recorded sign-outs to anchor a curriculum to improve sign-out skills. Other future modifications might include making the sign-outs accessible to nurses and administrators.
In summary, we devised, implemented and evaluated a new system of sign-out communication among physicians to facilitate the ED-internal medicine transfer. The new system was perceived to be easier for participants while not appearing to increase major adverse events. However, it did reduce interaction among physicians, and we therefore continued to mandate in-person discussion of unstable or critically ill patients. Busy emergency departments might consider this method as an efficient means of conducting some types of physician-physician communication.
Funding: Dr. Horwitz is supported by Yale-New Haven Hospital and by the National Center for Research Resources (NCRR). Neither Yale-New Haven Hospital nor the NCRR had any role in the design and conduct of the study; collection, management, analysis and interpretation of the data; or preparation, review and approval of the manuscript.
This publication was made possible by the CTSA Grant UL1 RR024139 and KL2 RR024138 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and NIH roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of the NCRR or NIH.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.