Behavioral health conditions adversely affect an individual’s well-being, quality of life, and life expectancy. Behavioral health conditions are defined as mental illnesses (eg, depression, anxiety, schizophrenia, and bipolar disorder) and substance use disorders that impair the functioning of an individual [1]. Patients with these conditions require the delivery of team-based care to achieve optimal health outcomes [2,3]. Nearly half of the adult population will meet the criteria for a mental illness during their lifetime [4]. Recent projections indicate that in the United States, approximately 19% of the adult population may have a mental illness at any point in time [5-7]. Behavioral health conditions are expected to surpass chronic medical conditions to become the leading cause of disability worldwide [8], with depression alone projected to become the second most prevalent cause of disability [7].

The relationship between physical health and mental health is complex [9-11]. Patients diagnosed with a behavioral health condition frequently have chronic medical comorbidities [4,5,12-15]. The association of mental illness with chronic disease contributes to the inefficient use of health care services (eg, increased inpatient hospital utilization) [5,16-18]. Other factors such as infrequent use of primary care and preventive services, poor chronic disease management, and the use of antipsychotic medications [19-23] also contribute to the overutilization of hospital services. Patients with behavioral health conditions, coupled with chronic medical comorbidities, contribute substantially to health care spending. A population-based cohort study in Alberta, Canada, revealed that patients with a chronic medical illness and a co-occurring mental health condition had a higher mean of 3-year adjusted health care costs (Can $38,250 [US $31,700]) than chronically ill patients without a mental health condition (Can $22,250 [US $18,440]) [24]. A similar pattern is expected in the United States. In fact, the costs are likely to be higher [25]. The costs alone indicate that it is both financially and clinically prudent to consider new ways to manage behavioral health conditions in chronically ill patients [25]. A novel approach to control costs is a coordinated care delivery system.

Behavioral health patients require coordinated care but often fail to receive it [17,26]. This level of care requires better communication and information sharing between general medical and behavioral health care providers. Therefore, it is necessary to develop new approaches that facilitate the effective treatment of both medical and behavioral health conditions [27]. Health information exchanges (HIEs), a component of health information technology (IT), can facilitate the integration of different health care services provided to behavioral health patients with chronic conditions.

HIEs are organizations that enable the digital exchange of health-related data [28], and they are one type of health IT thought to facilitate such integration. The exchange of health information through HIEs has many benefits, including safer, more efficient care that effectively manages both the behavioral and physical health needs of individual patients [13,29-31]. HIEs lead to efficient information sharing between providers in emergencies that require prompt, accurate diagnosis and treatment [29]. The effectiveness of HIEs has been studied in care coordination in general, especially in chronic disease management, but there are few studies that examine HIEs in the context of behavioral health [6,32-35]. The dearth of studies in behavioral health and HIE is a result of rising challenges with behavioral health data. Specifically, there is a disjoint between the clinical language, codes, and data reporting between behavioral health and general medicine among providers [6]. In addition, federal regulations complicate behavioral health information sharing. One of these regulations, that is, Section 42 Part 2, was adopted in the Code of Federal Regulations (42 CFR Part 2) in 1972 to ensure privacy protection for individuals receiving treatment for substance use disorders. It prohibited the disclosure of records related to substance use treatment without express authorization from the patient (with certain exceptions in emergency situations) [36]. Revisions of 42 CFR Part 2 in 2017 and 2018 have aimed to modernize the regulation and to facilitate the integration of behavioral health in general medical settings. Despite these efforts, challenges remain [6,36]. Taken together, these factors create barriers to exchanging behavioral health information and delivering integrated care for this patient population.

Therefore, the aim of this study is to identify and describe various factors that may influence health care providers’ intention to use and actual use of behavioral health information obtained from an HIE. This is accomplished by using a mixed methods approach guided by the theoretical backdrop of the unified theory of acceptance and use of technology (UTAUT) and diffusion of innovation (DOI). This study addresses the following research questions:

Integrated behavioral health care has demonstrated effectiveness in treating patients with behavioral health disorders and co-occurring medical conditions [37,38]. Such models of care delivery can improve the quality of care by delivering treatments that align with the clinical needs of patients [38]. A systematic review revealed that among 4 randomized controlled trials, 2 of the studies showed minor improvements in physical health, whereas the other 2 studies showed no significant changes in physical health [39]. Furthermore, improvements in seeking preventive care were also observed. However, none of these studies examined the effect of these interventions on functional or clinical outcomes [39]. Improvements in both access and quality of care have been attributed to communication and information sharing between clinicians, particularly when treating patients with complex medical and behavioral needs [40]. This finding suggests that establishing effective communication channels is crucial for the success of integrated care delivery models. However, there are barriers to communication between medical and behavioral health care providers.

A 2011 study examined the perspectives of behavioral health care providers regarding the benefits and barriers of HIE utilization [41]. According to the study, behavioral health care providers stated that the use of HIEs would (1) improve the quality of care and communication between different types of providers, (2) increase cost and time burdens, (3) raise concerns related to access to and vulnerability of patient and client information, and (4) impact workflow and control. Similarly, a 2019 qualitative study explored the perspectives of patients with mental health conditions to gain an understanding of privacy in the context of exchanging personal health information [42]. All 14 participants acknowledged the importance of privacy in health care, particularly concerning information related to sensitive diagnoses (eg, mental health conditions). However, the degree of concern varied and was most frequently related to negative experiences and perceptions related to privacy. The interviewees expressed that, overall, they trusted that health care professionals and organizations would take appropriate measures to protect the privacy of information related to mental health diagnoses. The participants were largely uninformed about existing policies and regulations designed to protect the privacy of mental health information, and those who had negative experiences related to privacy expressed doubt about the effectiveness of these regulations. However, the participants acknowledged that the electronic exchange of health information was a practical approach to ensuring high-quality patient care. It is important to note that this study was conducted in Canada, where a single-payer system exists.

The 2 theories that are deemed appropriate to explore our overall research questions are the UTAUT and the DOI theory.

UTAUT integrates constructs from earlier IT adoption theories to examine their influence on individuals’ intentions to use technology and the subsequent use of the technology [43]. UTAUT explains nearly 70% of the variance related to the intention to use technology [44]. The original UTAUT model attributes behavioral intention and subsequent technology use to 4 main constructs: (1) performance expectancy (perceptions of the ability of the technology to assist in meeting job-related goals), (2) effort expectancy (ease of use of technology), (3) social influence (perception that important figures support the use of technology), and (4) facilitating conditions (infrastructure that supports the use of technology) [43,45]. Extensions of the UTAUT model have considered the influence of additional factors on behavioral intention to use technology, such as trust and perceived risk [46].

DOI examines 5 factors that determine the rate of adoption of a new technology (innovation) within an organization: (1) relative advantage (the innovation is perceived as a better solution than its predecessor), (2) compatibility (the innovation fits with the organization’s culture and norms), (3) complexity (the innovation is easy to understand and use), (4) trialability (individuals can experiment with innovation before adoption), and (5) observability (the results of using the innovation are visible to others) [47,48]. The trialability construct was extracted from this theory because it was not captured in the UTAUT framework. The exchange of behavioral health data via an HIE is a relatively novel concept, and it is possible that pilot testing this type of data exchange could influence behavioral intention. The conceptual model for this study, which is derived from these 2 theories, is shown in Figure 1. The research questions were identified using this model.

A sequential explanatory mixed methods design was used in this study. This mixed methods design enables researchers to provide more thorough explanations of the trends that emerge from the quantitative phase of the study by further exploring participants’ perspectives [49,50]. The quantitative and qualitative study phases are described in the following sections.

The first author (RC) conducted the interviews with the survey respondents. An initial request for interview email was sent to the participants, and follow-up emails were sent 2 and 4 weeks after the initial request. Interviews were conducted either in person or via teleconference (Zoom), based on interviewee preference and timing. All interviews were recorded and transcribed using the Otter Voice Notes mobile app. The interviews were conducted from March 19 to May 2, 2019.

Verbatim interview transcripts were analyzed using inductive and deductive thematic analysis with NVivo 12. The development of the interview questions was partially guided by the constructs in the conceptual model; additional themes were derived from the content within the interview transcripts [60]. The following 4-stage analytical process was adopted for the study: (1) becoming familiar with data, (2) generating themes by aggregating coded segments of data, (3) creating comparative categories, and (4) revising and refining the themes [61]. Two of the authors (RC and SF) engaged in peer debriefing to discuss the interview process and the resulting themes, thus generating additional insights into the interview findings [62]. To increase the trustworthiness of data, we used member checking: participants received a 1- to 2-page summary of the interview transcript from the interviewee (RC) by email and were asked to provide clarification or ask additional questions. Of the 5 participants, 2 responded to the follow-up email. One participant provided additional clarification for one of the discussion points; the other participant found no errors in the summary document. Quantitative and qualitative findings were further integrated to create meta-inferences using a joint display [63].

We further integrated the quantitative and qualitative findings using a joint display. The joint display enabled us to compare the quantitative and qualitative results, explain similarities and differences between the findings, and develop meta-inferences regarding the exchange of behavioral health information via an HIE (Multimedia Appendix 5). In summary, the participants stated that having access to a patient’s full medical record, including behavioral health diagnoses, would be useful in delivering appropriate treatment. Despite the potential benefits of having access to this information, there are several barriers that prohibit the exchange of behavioral health information. These barriers are primarily related to technical capabilities, policies and regulations, stigma, and training. Even with these challenges, the interviewees acknowledged the value of having complete information and expressed some hope that behavioral HIE will become standard practice in the future.

The purpose of this mixed methods study is to identify and describe factors that may influence the intention to exchange and use behavioral health information obtained from an HIE. The preliminary findings from the quantitative phase of the study indicate that performance expectancy and trust are significant predictors of behavioral intention to exchange behavioral health information via HIEs. In other words, participants believed that exchanging behavioral health information via HIEs would improve their job performance. However, 90 % (56/62) of survey participants reported that they did not use HIEs to exchange behavioral health information, which likely contributed to the insignificant association between behavioral intention and actual use. The misalignment between performance expectancy and behavioral intention is supported in the literature [66,67]. One explanation for this finding is that understanding the potential job-related benefits of using a technology might not motivate health care professionals to adopt it if there are other contextual factors to consider [67].

Within health care, trust is a predictor of the intention to use technology among both patients and physicians [68,69]. Our findings were consistent with the literature in that participants suggested that the lack of full disclosure of behavioral health conditions, possibly due to stigma, may produce an incomplete record and not provide practitioners with valuable clinical information at the point of care.

In summary, 2 constructs (performance expectancy and trust) emerged as significant predictors in the initial quantitative phase of the study. One theme (usefulness of behavioral health information in care delivery) emerged from qualitative interviews that provided further insights into the significance of the performance expectancy construct. However, there were several divergent findings. Perceived risk did not emerge as a significant predictor in the quantitative phase; however, the interviewees stated that regulations and stigma are barriers to behavioral HIE. The interviewees suggested some connection between trust and stigma of behavioral health diagnoses. One interviewee stated that people were hesitant to disclose behavioral health information; therefore, the information that could truly be helpful at the point of care is often missing in terms of behavioral health (interviewee 1). Trialability did not emerge as a significant predictor of behavioral intention in the quantitative phase; however, the interviewees discussed a lack of in-depth training in the subsequent qualitative phase. Finally, behavioral intention was not a significant predictor of the actual use of HIEs for the purpose of behavioral HIE. This is likely due to the fact that participation in behavioral HIE efforts at these 2 study sites is low. However, interviewees expressed hope that in the future, HIEs will be used to exchange the full patient record, including any behavioral health diagnoses or treatment.

As behavioral health becomes a growing public health concern, health care practitioners have largely accepted the need for integrated care [70,71]. Early integrated care models have shown improvements in patients’ mental and physical health [71-74]. In addition, health care providers have expressed the belief that omitting behavioral health data from patient records hinders their ability to make appropriate treatment recommendations [71]. The findings of this study provide additional support to these notions.

The successful delivery of integrated care requires a reconsideration of existing policies and regulations as well as the establishment of clear guidelines and best practices for handling sensitive health-related information [75]. In the wake of the COVID-19 pandemic, restrictions on the sharing of personal health information have been relaxed. The requirement to obtain patient consent may be waived if the provider determines that there is a bona fide medical emergency [76]. It is crucial for stakeholders to consider whether this regulatory change should be retained following the pandemic.

Although the findings are valid in the context of this study, the small sample size may limit its generalizability. This is a common limitation of studies that are conducted during early uses of IT, thus necessitating the use of nonparametric methods such as PLS-SEM [77]. With the exchange of behavioral health data via HIE being relatively nascent (at the time of this study), this limitation is expected. Furthermore, the relationships between these constructs cannot be deemed causal but rather provide some insights into perceptions and reality. Similarly, because there were few participants who exchanged behavioral health information via HIEs, there was not enough variation in the responses to determine whether there were different sets of predictors for those who used the HIEs to exchange behavioral health information and those who did not. It is possible that the use of convenience sampling contributed to the homogeneity in the responses. However, the lack of active participants in behavioral HIE efforts could also be partially explained by the novelty of both the Alabama and Oklahoma HIEs.

Even in light of these limitations, the findings provide early positive suggestions for the use of behavioral health data shared by HIEs. This study is the first to adopt a mixed methods approach to examine the use of HIEs to share patients’ behavioral health information. The use of mixed methods allowed the researchers to identify and explore factors that may influence the adoption and use of HIEs to facilitate behavioral HIE. This study was conducted across multiple states. Finally, the study is applicable to both the health services and health IT arenas. It is relevant to the health services literature because it acknowledges the importance of a whole-person approach to care. Likewise, it is relevant to health IT because it considers the role of technology in delivering integrated care to patients with complex needs.

Patients with behavioral health conditions are readmitted to the hospital 30 days after discharge at a higher rate than patients without behavioral health disorders. Several factors contribute to this pattern of overutilization. Existing regulations have traditionally restricted the exchange of behavioral health information unless the patient has given authorization at the person level, meaning that the specific person must be identified. The prevalence of suboptimal treatment outcomes for behavioral health patients with chronic illnesses indicates that it is necessary to commit more efforts to providing higher quality care to some of the most vulnerable patients.

Health care providers acknowledge that it is necessary to deliver holistic care to improve the quality of care and health-related outcomes for this subset of difficult-to-treat patients. This study contributes to our understanding of the potential role of HIEs in integrating the traditionally fragmented behavioral and general health service arenas.