Transition of Care from the Emergency Department to the Outpatient Setting: A Mixed-Methods Analysis

Introduction The goal of this study was to characterize current practices in the transition of care between the emergency department and primary care setting, with an emphasis on the use of the electronic medical record (EMR). Methods Using literature review and modified Delphi technique, we created and tested a pilot survey to evaluate for face and content validity. The final survey was then administered face-to-face at eight different clinical sites across the country. A total of 52 emergency physicians (EP) and 49 primary care physicians (PCP) were surveyed and analyzed. We performed quantitative analysis using chi-square test. Two independent coders performed a qualitative analysis, classifying answers by pre-defined themes (inter-rater reliability > 80%). Participants’ answers could cross several pre-defined themes within a given question. Results EPs were more likely to prefer telephone communication compared with PCPs (30/52 [57.7%] vs. 3/49 [6.1%] P < 0.0001), whereas PCPs were more likely to prefer using the EMR for discharge communication compared with EPs (33/49 [67.4%] vs. 13/52 [25%] p < 0.0001). EPs were more likely to report not needing to communicate with a PCP when a patient had a benign condition (23/52 [44.2%] vs. 2/49 [4.1%] p < 0.0001), but were more likely to communicate if the patient required urgent follow-up prior to discharge from the ED (33/52 [63.5%] vs. 20/49 [40.8%] p = 0.029). When discussing barriers to effective communication, 51/98 (52%) stated communication logistics, followed by 49/98 (50%) who reported setting/environmental constraints and 32/98 (32%) who stated EMR access was a significant barrier. Conclusion Significant differences exist between EPs and PCPs in the transition of care process. EPs preferred telephone contact synchronous to the encounter whereas PCPs preferred using the EMR asynchronous to the encounter. Providers believe EP-to-PCP contact is important for improving patient care, but report varied expectations and multiple barriers to effective communication. This study highlights the need to optimize technology for an effective transition of care from the ED to the outpatient setting.


INTRODUCTION
The vast majority of patients presenting to the emergency department (ED) are evaluated and subsequently discharged. 1 Many of these visits will require a follow-up plan of care with a primary care physician (PCP). An appropriate transfer of information, within a reasonable time frame, must occur during this hand-off to ensure high-quality patient care and continuity of disease management.

Background
Patient care transitions directly impact quality and patient safety. The discharge of a patient from the ED is a time of high vulnerability. Given the increasing complexity of medical care and the limitations that some patients may have due to language fluency or health literacy, the expectation of high-fidelity information transfer through discharge instructions alone is unrealistic in many cases. 2 Prior studies have looked at handoffs between emergency physicians (EP) at shift change, between EPs and hospitalists, and between EPs and nursing homes. [3][4][5][6] Common themes that arise from the literature regarding transitions of patient care from one healthcare provider to another are the need for bidirectional communication and a balance between standardization and flexibility. 2 Both EPs and PCPs believe that coordination of care between the two settings is an important transition in healthcare. 7 Communication between the EPs and PCPs has been regarded as unsatisfactory, if performed at all. 8 Poor communication results in provider confusion regarding followup needs, which may predispose patients to error or adverse events. 9-10 The process of hospital discharge to outpatient care currently has multiple barriers that contribute to poor transitions of care. These include unstructured communication systems, such as the electronic medical record (EMR), lack of longitudinal care and absence of follow-up standards. 11 To begin to address barriers to effective transitions of care, it is necessary to investigate current systems of communication, with a focus on provider expectations and use of the EMR.

Goals of this Investigation
The goal of this study was twofold. First, we aimed to characterize the current practices in the transition of care from the ED to the outpatient setting. Second, we sought to clarify providers' preferences and use of EMR technology in managing that transition.

Study Design
This was a prospective study using semi-structured interviews. We developed a mixed-methods survey based on literature review and modified Delphi technique. 12 The survey was designed in two phases ( Figure 1). In the first phase, the authors created the survey tool, which consisted of a set of general questions in regard to professional setting. The remainder of the survey was divided into questions specific for either EPs or PCPs. We included both multiple-choice questions and freeresponse questions to ensure capture of individual practices (Appendix). The survey was piloted via email to a group of 16 EPs and PCPs.
After the initial pilot test, we reviewed and modified the survey for usability and redundancy. Additionally, we changed the format to an in-person interview. Four external reviewers were consulted and the survey was further refined for the inperson interview. Content validity and face validity were assessed by a process of multiple revisions based on pilot-test results, expert analysis, and triangulation. 13 The survey was re-piloted in an in-person interview format with three EPs and three PCPs to obtain feedback on structure. We surveyed participants at eight different institutions. Participant anonymity was maintained by collection of data without identifying information. 14, 15

Setting
The survey tool was developed by members of the American College of Emergency Physicians (ACEP) Academic Affairs Committee. Academic and community physicians at eight different sites across the country participated in the study. We selected institutions based on author affiliation; these

Selection of Participants
We obtained institutional review board approval at each site. These institutions were primarily urban-based academic centers. Each author selected a convenience sample of EPs and PCPs. Only attending-level physicians were enrolled.

Methods and Measurement
All participants underwent a verbal, informed consent prior to completing the survey. Interviews took place in person and lasted 15-20 minutes. We de-identified all data upon response submission. Data responses were collected electronically into a single electronic database at each site, and subsequently combined into a master database.

Analyses
Quantitative data was extracted and entered into a processing program. We examined five demographic questions for all participants. The PCP survey contained 12 additional multiple-choice questions about transitions of care, and the EP survey contained eight additional questions. Data analysis was performed in SAS (version 9.2; SAS Institute, Inc. Cary, NC) using Fisher's exact test. We performed qualitative data analysis for responses to the open-ended questions.
Themes were developed based on a grounded theory approach. 16,17 Two independent coders used the constant comparative method to identify themes in the data collected from the interviews. If a discrepancy occurred, the reviewers discussed to achieve consensus. Themes were standardized for each question, and data were independently coded according to established themes. We assigned responses to one or more themes for each question. In an effort to minimize rater subjectivity regarding identified themes, we measured inter-rater reliability for each question with the goal of > 80% agreement. 18 Percent agreement was based on the alignment of all selected themes for each question. We analyzed qualitative data for percent representation of individual themes. The number of responses for each theme was calculated and averaged between the coders for a percent representation of each theme.

Characteristics of Study Subjects
Between November 2014 and February 2015, 102 interviews were attempted. Forty-nine respondents were PCPs and 53 were EPs. Of these respondents, one EP participant did not complete the survey in its entirety. This respondent was omitted from the analysis, resulting in 101 responses analyzed.
PCPs were divided between family medicine, internal medicine, and pediatrics. All of the EPs were trained in emergency medicine, with one individual also trained in pediatrics. Both EPs and PCPs share similar demographic trends with the majority of the sites categorized as urban academic settings. The remainder of the demographic information is shown in Table 1.

Quantitative Analysis
PCPs reported receiving communication about an ED visit much more frequently than EPs reported communicating back to the PCPs. Forty percent of PCPs reported actually receiving follow-up on their patients "most" of the time (81-100%), but a greater proportion (61.2%) felt that they should be contacted at this frequency. EPs preferred telephone communication to EMR and reported greater use of this modality, whereas PCPs preferred discharge communication through the EMR.
EPs were more likely to report no PCP communication for a patient with a benign and stable condition.
For patients requiring urgent follow-up, EPs were more likely to report the need for verbal communication prior to the patient's discharge from the ED than PCPs. Both groups thought that a patient with an urgent condition required direct discharge communication with the PCP (Table 2).
Regarding perceptions of EMR use, EPs believe that the majority of PCPs (53.8%) use the same EMR and view the patient's records directly. A minority of EPs believed that PCPs receive EMR notifications of a patient visit, whereas many PCPs reported receiving a notification.

Qualitative analysis
Analysis of the 101 responses to qualitative-response questions identified thematic concepts for each question. The first question, "Under what circumstance is it important that the emergency physician communicate with a patient's primary care physician?" demonstrated that follow-up needs were the most important reason to communicate with PCPs. Figure 2 demonstrates physician response.
Physicians also responded to the question, "How should EMR be used as a tool in the transition of care?" More PCPs reported EMR notification/alert systems as a valuable use of EMR, compared to EPs who cited the EMR's ability to aid in follow-up and continuity of care. Figure 3 demonstrates physician response.
Finally, providers were asked, "What are major barriers to efficient communication with EP/PCPs?" Themes included setting/environmental constraints, communication logistics, and EMR barriers. Patient constraints were reported to a lesser extent by both groups. Poor documentation was mentioned among the PCP group. Figure 4 demonstrates physician response.

DISCUSSION
Our multi-center prospective study examines and highlights the current practices and preferences for handoffs between EPs and PCPs and highlights quality gaps in the There are also existing, under-used tools for communicating discharge information that are highly regarded and improve provider satisfaction. 19 Limpahan et al. developed a set of best practices for patient discharge, including sending a summary to the PCP, performing medication reconciliation, and providing patient education. 20 The authors suggest using the EMR as a potential avenue for automated inclusion of the described practices. Separate EMR systems have been identified as a challenge in the transition of care, while an interface for a shared EMR has been cited as a way to minimize transitions-of-care losses. 7 Furthermore, the ability of EPs to provide an alert to the PCPs through flagging or email notification has been described as a potential tool for communication.
In the present study providers reported setting and environmental constraints including a high patient volume, coordinating time to call, and communication during non-business hours. A standard EMR notification system may alleviate some of these constraints; however, EMR barriers to effective transitions were also noted, including lack of EMR access or shared EMR, uncertain receipt of information, and limited EMR literacy. Other logistical barriers to communication included inability to identify the PCP, difficulty getting in touch with the appropriate provider, and lack of resources. These are systems issues that could be addressed with increased emphasis on the ED-to-outpatient communication. Specific strategies might include readily available electronic documentation of a patient's PCP, shared EMR access among hospitals and clinics, and professional coordinators to relay information during the discharge process.
Healthcare providers believe that both technology and standardization should be the focus for future improvements in the transition of care. 8,21 Shared EMR access and EMR notifications are potential areas for development. There are also new tools of clinical communication that may bridge the gap between the synchronous phone communication preferred by EPs and

LIMITATIONS
There are several limitations to this study. Most notably, the participants comprised a convenience sample of physicians from eight academic institutions. All community physicians worked at a community site affiliated with one of the primary academic sites. The present study lacks representation from community sites without academic affiliation, military, and rural institutions. Our responses may not reflect practice patterns in these settings. This study also lacks input from mid-level providers and residents who are also involved in the hand-off process. Interviews were performed in-person and therefore may have led to reporting bias on the part of the participant.
Furthermore, this data is based on perception rather than objective measure of phone calls and EMR notifications, which is subject to recall bias. The qualitative questions served as a strategy to recruit more diverse responses. The process of coding synthesizes information, thus losing the context of specific statements in favor of categorizing data into themes. Finally, the majority of subjects in this study reported using EPIC EMR software. Other interfaces may allow for varying degrees of electronic communication between the ED and PCP, thus altering one's perception of EMR utility.  Conflicts of Interest: By the WestJEM article submission agreement, all authors are required to disclose all affiliations, funding sources and financial or management relationships that could be perceived as potential sources of bias. No author has professional or financial relationships with any companies that are relevant to this study. There are no conflicts of interest or sources of funding to declare.