Consultants’ and referrers’ perceived barriers to closing the cross-institutional referral loop, and perceived impact on clinical care

Participants, Settings, and Recruitment

Physicians from primary care clinics and specialized consulting clinics at academic medical centers were targeted for recruitment, primarily from two healthcare networks in the Midwest. As a tax-supported, federally qualified health center, one of these networks serves vulnerable populations and uses EPIC’s electronic health record (EHR) system. The other healthcare network has the largest network of physicians in the state and uses Cerner’s EHR system. Both provide physicians access to HIE technology, such as CareWeb³¹ and Docs4Docs³². We sought to recruit at least 12 PCP referrers and 12 consultants to reach data saturation, which is typically reached with approximately 10–15 participants.³³ We defined data saturation as the point at which no new data topics from participant interviews. We used convenience sampling for both groups, due to recruiting difficulties posed by the COVID-19 pandemic. PCPs from primary care and internal medicine were eligible for the referrer interviews. We targeted specialties with high numbers of referrals.^{34, 35} Consultants from neurology, gastroenterology, cardiology, and oncology were eligible. Our intent was to recruit at least three participants from each specialty. Contact information for potential participants was identified by key administrators and online directories. Eligible clinicians were contacted via email and phone calls.

Data Collection

We conducted semi-structured interviews with participating PCPs and specialty consultants. Interviews lasted up to 45 minutes and were conducted one-on-one virtually, using Zoom for Healthcare. Interview guides (see Appendix) were developed using the SEIPS 2.0 framework with questions emphasizing work system components related to internal and external referrals (Table 1). Questions focused on physicians’ needs and experiences in relation to their tasks, existing tools and technology, environment, and related policies. Interviews began with questions related to demographics, including each participant’s role, years of experience, and specialty. Two members of the research team (AS1, AS2, AM, EC, and AMG) were present for each interview; one served as the primary interviewer, and the other as a notetaker. All interviews were audio-recorded with participants’ consent and were transcribed for analysis.

Data Analysis

Rapid qualitative data analysis ³⁶ occurred with an iterative inductive approach, with codes being defined a priori from the interview guide, and revised during initial analysis to include emerging themes. Three researchers (AM, AMG, SK) independently read all transcripts prior to analysis to become familiar with the data and to assess the utility of the pre-established template. The team met to discuss assessments and refined a common understanding of the template. Each interview was summarized by four researchers (AM, AMG, SK, MH) into an episode profile template. Themes were identified iteratively through the data reduction process until the team reached a consensus. Episode profiles and themes were then reviewed and validated by the larger research team. We illustrated themes with respondent quotations. Two members of the team (SK, AM) created a matrix of barriers, system factors, and quotations.

Participants

We interviewed six referring PCPs and 12 specialty consultants (Table 2). Referrers were predominantly female (83%) and White (66%), with work experience ranging from 10 to 32 years. Referrers practiced internal medicine (N=3), geriatrics (N=2), or primary care (N=1). Consultants were predominantly male (58%), White (58%) or Asian (33%), with experience ranging from 11 to 46 years. Consultants practiced in gastroenterology (N=4); neurology (N=5); cardiology (N=2); or oncology (N=1).

Overview: Perceived work system barriers and negative outcomes for physicians and patients

Based on our analysis using the SEIPS 2.0 model, technology and organizational factors were associated with the largest number of barriers. Person, task, and environment factors were not as prominent in the analysis of barriers. HIE technologies were described as “extremely difficult to navigate,” not user-friendly, and often containing errors and outdated or missing information. None of the participants reported integrating HIE technologies into their workflows. Overall, consultants perceived external referrals as being “a constant struggle for communication.” A consultant said, “I’m sure there are other avenues by which information should be shared, but you’ve got to deal with privacy and institutional policy and so forth. Bottom line, the better the information sharing is, the better the coordination of care and the smoothness of the delivery of care for the patient.” [Consultant 11] With current workflow and information flow challenges, referrers shared that they avoid making external referrals unless they are necessary (i.e., a required specialty not available in the referrer’s hospital system). The perceived barriers to closing the external referral loop were associated with perceived increases in patient safety risks, patient burden and frustration, and physician workload (Figure 2).

Technological barriers

When discussing barriers related to technology, information sharing and communication were hindered due to the use of rudimentary technology (see Table 3). Since communication and sharing of information occurred via fax and phone, consultants reported that it is “much harder” to communicate with referrers from other institutions. Consistent among consultants were descriptions of the “100-page fax,” detailing the frequent receipt of clinical information that was both excessive and unnecessary (i.e., included information not needed). Consultants also reported time wasted by trying to access information, only to discover that the desired information (labs, medications, etc.) was missing. Furthermore, consultants noted that they have no way of knowing whether the referrers ever receive their notes due to issues with outdated or missing contact information for referrers.

Organizational barriers

Organizational policies reportedly prevented information sharing and interpersonal communication between referrers and consultants from different institutions (Table 3). For example, referrers mentioned that external consulting institutions in need of information might have different policies and procedures about requirements for sharing or releasing their own information. Additional or unique requirements for consent forms were said to lead to communication breakdowns or delayed information sharing. According to consultants and referrers, some physicians’ privileges at multiple institutions helped to circumvent technical limitations for information sharing because they had access to multiple EHRs. However, this practice was not necessarily approved or considered to be an aspect of standard workflow.

Perceived increase in physician workload, patient burden, and patient safety risks

Attempting to access patient information, consultants perceived an increase in workload for both them and their teams. Consultants described repetitive phone calls to referrers’ clinics and testing sites to request information. Often, the information received was considered to be of poor quality regarding timeliness, relevance, and completeness. As a result, consultants noted that their workload increased after patient encounters. For instance, consultants described multiple phone calls to patients after a visit to discuss lab results that were not received prior to the consultation. Consultants said these phone calls were frequent and could last up to 20 or 30 minutes, sometimes resulting in a need to schedule another appointment. While these phone calls and additional appointments were perceived as an increase in workload, consultants mentioned that the extra time spent on phone calls is not recognized as a standard work metric (i.e., billable, reimbursable) by their organizations.

Consultants noted that communication breakdowns resulted in delayed patient care, and less time for the patient to receive direct care. Across specialties, consultants described scenarios in which patients were visibly burdened and frustrated (see Table 4). Providers perceived patients as disheartened to hear that not all of the necessary information was available to their physicians, delaying diagnosis or treatment. Without essential patient information, including all relevant labs, consultants relied on patients to recall past tests, medications, or clinical encounters. Consultants noted patients’ frustration as they attempted to find necessary information in folders of printouts or on mobile phones, which could dominate most of the encounter time. Additionally, patients would need to schedule yet another appointment, adding further stress and potentially redoing uncomfortable tests. Consultants noted that some patients wait a long time for an appointment or travel long distances, which is taxing on them and their caregivers. Furthermore, added stress could impact their medical conditions or symptoms.

Although most consultants stated that external referrals present additional challenges, they also reported that aforementioned barriers did not necessarily affect patient safety (see Table 4). Given these barriers and challenges, referrers shared that they would encourage patients to stay “in the system” for all of their care, or to venture outside at their own risk. Both referrers and consultants indicated that the issue of communication and information sharing barriers did not contribute to medical errors or poor treatment. To the best of their knowledge, consultants stated that they were able to provide the same level of care to patients regardless of whether referral and consultation occurred at the same institution. However, some consultants reported they did not often know how patients’ health or safety was affected when the patient returned to the external referrer. Without clear communication or information sharing with the referrer, consultants explained that they would only become aware of patient issues, recommendation adherence, or need for additional treatment if the patient returned for another consultation.

Physician workload

Physicians said they felt a sense of “helplessness” and overload both professionally and personally, especially with respect to competing demands of clinical care and organizationally emphasized metrics (e.g., reimbursement). Our study described how misaligned organizational metrics and policies negatively affect physicians. Consultants described their institutions’ stance as closer to a business-centric approach to care, where the aim is to maximize patients and profits rather than emphasize meaningful experiences for patients and physicians. This phenomenon closely resembles emerging work³⁷ characterizing physician burnout linked to EHR usage logs. In contrast, the use of rudimentary or inefficient technology such as telephones, fax machines, and printers does not generate the same types of usage data (e.g., time of logins, number of clicks, or page view duration) or workflow efficiency indicators. For example, time reviewing the “100-page fax” in addition to the EHR is not captured. The relative “invisibility” of such technologies in terms of logging, auditing, quality assessment, and research contributes to overlooked barriers. Our results indicate that this type of workload increase is experienced by the entire care team and is not limited to physicians.

Patient burden and frustration

When discussing their perceived barriers to closing external referral loops, physicians expressed concerns regarding patient burden, including increased information management, duplicate testing, delays in diagnosis, and patient frustration. Aligned with previous studies,^{15, 20} neither referrers nor consultants interviewed in this study trusted the successful transmission of clinical information via EHR or HIE technologies because of their firsthand experience with communication breakdowns. Our study described how physicians were forced to be over-reliant on patients to fill health information gaps during clinical encounters. Although physicians wanted to discuss undocumented information (e.g., goals, resources, new health concerns), the time required to search for missing medical data limited interpersonal discussions. The alternative option was duplicate testing, which is burdensome for patients due to costs, travel, discomfort, time required, and prolonged anxiety over results. Beyond delays to care access, our findings highlight the negative impacts of delayed access to patient information. Existing studies denote that reduced patient satisfaction or increased patient frustration affects health outcomes.³⁸ Our findings indicate that HIE technology has fallen short of potential benefits, including duplicate testing regarding cross-institutional referrals. Across patient populations from varying backgrounds and with a variety of medical conditions, relying on patients’ access to and management of medical records to support referrals is not a robust solution and could introduce or exacerbate health disparities.

Patient Safety Risks

Although neither consultants nor referrers perceived great risks to patient safety or witnessed adverse outcomes from existing studies^{15, 23}, both groups noted that communication barriers often limited or excluded knowledge or confirmation of patients’ health outcomes after external referrals. According to our interviewees, physicians’ privileges at multiple institutions helped to circumvent technical limitations for information sharing because they had access to multiple EHRs. However, usability and data quality issues are not limited to a specific EHR system.³⁹ Therefore, access to multiple EHRs may not increase physicians’ awareness of patient safety risks.⁴⁰ Regarding consent for release and electronic communication of medical records, both referrers and consultants reported inconsistencies among communication policies at different institutions limited interpersonal communication between the two groups. According to Singh et al.,⁴¹ limited follow-up and communication predisposes patients to diagnostic errors, degrading the evolution of diagnostic impressions.

Early recommendations and future research

Based on our results, the technology and organizational barriers perceived by physicians illustrated the necessity of communication and information sharing for patient care coordination across institutions. Consequently, potential for HIE technological solutions for cross-institutional referrals exists, but effective designs, implementation, and adoption are essential for future success. Applying a human factors approach, previous studies of newly designed, adopted, and implemented health information technology reported improvements in the referral process.^{17, 42, 43} This study provides evidence that informs future human factors research of HIE technology for closing external referral loops, emphasizing the human-centered design process (Box 1). Future research should investigate how emerging health information communication technology and applications (apps) can foster better interpersonal and medical record communication. For example, assessment of clinicians’ interactions via telephone and fax machines (e.g., time spent, number of attempts) can provide insights on extraneous efforts, essential information gaps, and reliable information sources. Existing literature describes design guidelines and use of electronic templates to improve referral orders.^{3, 44} For consultants, our evidence recommends applying HIE electronic templates or user interfaces accessible via EHRs, which can assist in transferring information from consultants to referrers. Extending the guidelines for data entry, we recommend customizable information displays that support filtering, sorting, and queries for additional information. This recommendation would address the perceived increased physician workload and patient burden invoked by relying on phone calls and faxes. Furthermore, we recommend salient feedback on these displays to confirm information receipt and post-consultation health outcomes. Future evaluation efforts to analyze health outcomes among internal and external referrals are needed to improve understanding of positive and negative associations, and implications for patient-safety risks.

Limitations

Due to the COVID-19 pandemic, this study was limited by a small convenient sample of interviewees in two healthcare facilities. However, one healthcare facility has the largest network of physicians in the state, and the other facility primarily cares for vulnerable populations. Although a small proportion of clinics within each facility were represented in this analysis, these specialties reported having large numbers of specialty referrals. Participants and non-participants may differ in their characteristics, practices, experiences, or perspectives. Further research in a representative sample is needed to confirm these results.