Perspectives on Challenges and Opportunities for Interoperability: Findings From Key Informant Interviews With Stakeholders in Ohio

Introduction

Background

Starting with the Health Information Technology for Economic and Clinical Health (HITECH) Act in 2009, the United States has invested over US$ 36 billion to promote interoperability—the ability of 2 or more systems to exchange and use information [1-3]—through health information exchange (HIE) networks and electronic health records (EHRs). HITECH promoted the adoption and implementation of certified EHRs by providing financial incentives through its “meaningful use” programs and funded grants that helped establish regional HIE networks [4]. Some of these financial incentive programs, such as the state Medicaid Provider Incentive Program (MPIP) were phased out in 2021. However, subsequent legislation such as the Patient Protection and Affordable Care Act reinforced this financing by advancing payment and care delivery models that use risk-based contracts to incentivize quality of care and patient outcomes, such as accountable care organizations, value-based care [5], and patient-centered medical homes [6], which stand to benefit from enhanced electronic data exchange. Other population health policy programs, including Comprehensive Primary Care (CPC) [7], CPC-Plus [8], and Primary Care First [9], further rely on robust data exchange among regional providers. Ongoing federal policy development has continued and enhanced support for meaningful use programs through the Medicare Access and Children’s Health Insurance Program Reauthorization Act of 2015, leading to the Promoting Interoperability Program in 2018.

Recent reports show that 96% of acute care hospitals and 80% of primary care providers (PCPs) have implemented certified EHRs with interoperability capabilities [10,11]. However, actual use of data from an HIE network in clinical care encounters remains low [12]. Researchers have identified several barriers to the use of HIE networks, including poor user interfaces and lack of leadership support [13-15]. Although interoperable health information technology (HIT) is theorized to address these barriers, it continues to elude the health care system [16,17]. For instance, a recent study found that only 45% of all US hospitals engaged in the 4 core elements of interoperability, such as the capability of different EHR systems to find, send, receive, and use or integrate clinical information with one another [16]. Further, rural and smaller hospitals, ambulatory practices, as well as those ineligible for meaningful use incentives (eg, rehabilitation, long-term care, or behavioral health providers [BHPs]) lag behind large, integrated systems in adopting interoperability [18,19].

To promote interoperability, the 21st Century Cures Act (21CCA; 21CCA 2016) mandated the sharing of certain data elements, placed restrictions on information blocking, and promoted the use of application programming interfaces (APIs; eg, Fast Healthcare Interoperability Resources [FHIR]). The 21CCA also established the Trusted Exchange Framework Common Agreement (TEFCA) that provides an infrastructure model and governing approach for HIE networks [20]. However, in the sixth national survey of HIEs, 56% (84/151) of the regional HIEs planned to participate in TEFCA [21], which points to ongoing challenges that continue to limit data exchange. HIEs that intended to participate in TEFCA already had connections with HIEs in other states and participated in at least one national network. Therefore, the decision to participate in TEFCA may be determined by the alignment of HIE processes with existing data sharing rather than convincing HIEs of the benefits of participation [21].

Purpose

In alignment with policy efforts geared toward the promotion of interoperability, this study aimed to explore provider perspectives on the current state of interoperability challenges. Given the confluence of timing around the conclusion of state MPIPs and the ramp up of 21CCA and TEFCA, questions remain about the progress and remaining challenges related to achieving an interoperable health system. Moreover, previous research on interoperability typically focuses on a single perspective rather than those of multiple stakeholders. This multistakeholder lens is particularly relevant to consider when examining interoperability, as a key goal of this innovation is to connect disparate parts of the health system. Specifically, we conducted semistructured interviews with a stratified sample of key informants, including providers and individuals in leadership positions representing diverse organizations in Ohio, to identify barriers and facilitators to interoperability. The study’s findings add to the body of knowledge about interoperability and may contribute to the efforts of state agencies and federal policy makers, such as the Office of the National Coordinator for Health Information Technology and the Centers for Medicare and Medicaid Services, to advance interoperability. For instance, our findings may provide evidence that supports alignment between the Health Insurance Portability and Accountability Act (HIPAA) with HITECH and 21CCA. Finally, technology vendors may benefit from increased understanding of the end user perspective of their applications to develop user-friendly software.

Methods

Study Setting and Design

We used a cross-sectional qualitative design to solicit multistakeholder perspectives on the state of interoperability in Ohio. Ohio provided financial incentives for the adoption of interoperable EHRs for eligible professionals and hospitals through the MPIP. Eligible professionals included those with an active Ohio Medicaid Provider Agreement, such as physicians, optometrists, dentists, certified nurse-midwives, nurse practitioners, and physician assistants practicing in a federally qualified health center or a rural health center led by a physician assistant. Eligible hospitals were also required to have an active Ohio Medicaid Provider Agreement and include acute care hospitals, critical access hospitals, cancer hospitals, and children’s hospitals. Ineligible providers included most behavioral and mental health, long-term care, and home health providers. MPIP operated through 7 incentive cycles, with the final cycle occurring in 2021. On the basis of the 2021 Ohio Medicaid Electronic Health Records Survey for Practices and Hospitals [22], 23.87% (5593/23,435) of the eligible providers and hospitals had received at least one MPIP payment, 96.06% (5280/5593) of the MPIP recipient providers and hospitals had adopted and used an EHR, and 90.73% (16,188/17,842) of those who did not receive an MPIP payment had adopted and used an EHR. Among the 735 ineligible providers, 72.1% (530/735) reported adopting and using an EHR. Epic Systems Corporation is the most prevalent EHR vendor in the state, with 36.80% (2058/5593) of MPIP recipient providers and hospitals and 56.56% (10,092/17,842) of non-MPIP recipient providers and hospitals using an Epic EHR. However, though Epic remained the EHR of choice for group practices (across multiple or single sites) and hospitals, individual practices were more likely to choose other EHR vendors (eg, NextGen and eClinicalWorks). Among the ineligible providers, there was substantial variation in EHR vendors, and CareLogic was the single most prevalent EHR vendor, with 7.8% (57/735) of providers adopting it. Presently, Ohio has 2 large-scale regional HIEs that facilitate electronic exchange of patient data. Almost 31.64% (7416/23,435) of MPIP eligible providers and hospitals had existing arrangements with the regional HIEs to share electronic patient-level clinical data, whereas 21.5% (158/735) of MPIP ineligible providers participated in the regional HIEs.

Data collection and analysis were guided by the technology-organization-environment (TOE) framework [23]. TOE is an organization-level theory that has been applied to explain how the 3 interacting contextual domains (ie, technology-organization-environment) influence a health care organization’s technology-related decision-making [24].

Ethics Approval

This study was considered to have minimal risk and was approved by the Ohio State University institutional review board (2021B0378).

Sample Selection

We used a stratified purposive sampling approach to gather diverse perspectives on interoperability based on four representative groups: (1) acute care and children’s hospital leaders, (2) PCPs, (3) BHPs (ie, providers or organizations that provide care for mental health, substance use disorders, stress-related physical symptoms, and life stressors and crises), and (4) regional HIEs (ie, organizations that facilitate information exchange within a network of facilities within a geographic boundary). In addition, both rural and urban subsamples within each of the 3 provider groups were interviewed to ensure geographic representation. We focused on key informants with first-hand perspectives on HIT adoption and its future directions. These key informants were administrative leaders with clinical and nonclinical backgrounds within organizations with decision-making capacity regarding HIT (eg, chief medical information officers, executive directors, chief executive officers, and strategy officers). The study leads received a list of potential key informants from the Ohio Department of Medicaid. Emails were sent to gauge interest in participation; key informants who agreed to participate (20 of 38 organizations invited) were then interviewed. To eliminate any potential conflict of interest, the Ohio Department of Medicaid was not notified of who agreed or refused to participate, and did not participate in any interviews.

Data Collection

We conducted semistructured interviews with key informants from July to October 2021. The interview guide was developed to ask about EHR implementation in general, the alignment of public policy with organizational strategy, interoperability implementation challenges, and opportunities to improve HIT effectiveness. The interview guide was piloted with administrative leaders (n=2; eg, chief information officer) at an urban hospital in Ohio. This process yielded two versions of the semistructured interview guide for (1) providers (ie, those from hospitals, primary care, and behavioral health) and (2) representatives of HIEs (Multimedia Appendix 1). All interviews were conducted remotely via Zoom and audio recorded.

Data Analysis

Interviews were transcribed verbatim and analyzed using directed content analysis—an approach that begins with the a priori codes from the TOE framework yet is permissive of emergent themes [25]. The coding team (DMW, WLT, and LP) met weekly to discuss the interviews and preliminary findings throughout the data collection phase. A preliminary codebook was developed by reviewing transcripts and identifying broad themes that emerged from the interview transcripts to organize the data into the 3 TOE domains. Next, to build consensus on the coding guide, the analysis team collectively reviewed common transcripts (n=3) to compare results, refine the codebook, and reconcile any coding discrepancies. The remaining transcripts were divided by organization type for coding. The team continued to compare findings throughout the analysis phase to achieve thematic saturation. The trustworthiness of our findings is ensured by our rigorous and iterative approach to analysis [26]. NVivo software (version 12) was used to support data coding and analysis.

Results

Overview

Overall, 24 key informants were interviewed representing 20 distinct organizations: 5 hospitals, 7 PCPs, 6 BHPs, and 2 HIEs. Of the 20 organizations, 15 (75%) were located in urban areas (Table 1).

Below, we report on the key themes and subthemes of the analysis organized by the domains of the TOE framework, including comparing and contrasting those that cut across organizational types as well as those specific to each individual organization.

Technology Domain

Overview

Three themes within the technology domain that related to the usability and technological aspects of interoperability were identified: (1) implementation challenges, (2) interoperability capabilities, and (3) opportunities (Table 2).

Implementation Challenges

For hospitals, a primary concern focused on the growing number and proliferation of application types (eg, EHR, personal health records, HIE, and population health platforms) that contain health information and can be used in clinical encounters. These applications were viewed as difficult and costly to maintain. Similarly, these applications do not use a common data structure and storage format, resulting in too many places for data to be located and for clinicians to search for useful information. The lack of interoperability necessitates that providers leave the EHR to access other applications. As one hospital representative stated as follows:

...It’s one of those things where now you’ve got to subscribe to it [eg, a regional HIE], and it’s another place [eg, application] for you to go to look for more information [eg, patient clinical data]...There’s too many places for data to land and get sent. People just stop looking.
[Hospital representative #10]

HIE leadership echoed this perspective, adding that a lack of interoperability was an impediment to creating a complete medical record. PCPs noted similar issues related to the number and type of applications needed to overcome gaps in interoperability, such as population health tools (eg, Innovaccer). In contrast, for BHPs, interoperability across applications was not discussed; instead, remarks focused on off-the-shelf EHR systems being misaligned to the specific requirements for BHPs, such as lacking additional protections for substance abuse data.

Interoperability Capabilities

Hospitals described advanced interoperability functionality and attributed their advanced capabilities to their EHR vendor rather than the regional HIEs. The regional HIEs were helpful for exchanging continuity of care documents [27] but did not facilitate the integration of information across EHRs. Typically, hospitals only expressed limitations with interoperability functionality as being a function of data recipients’ capacities. For instance, a major concern for hospitals was connecting and being able to integrate EHR data with state-hosted information systems such as the state immunization registry, vital statistics, death certificates, or birth records.

Hospitals and PCPs both noted limitations of exchanging data across the continuum of care, such as with long-term care and BHPs. PCPs additionally noted interoperability challenges with hospitals in their own health systems even when they all used the same EHR vendor, as some vendors are not capable of data exchange in different instances of the same EHR.

BHPs described much more basic data exchange capabilities relative to hospitals and PCPs. For instance, they mentioned that their data exchange is primarily focused on billing. In contrast to hospitals, BHPs mentioned a greater reliance on the regional HIEs for access to health records from other providers and event notifications. The regional HIEs echoed this relationship and discussed how BHPs lag in their interoperability capabilities.

Opportunities

Hospitals described opportunities related to new population health platforms that may be able to improve interoperability without the costs associated with interfacing with different EHR systems or HIEs. Both hospitals and PCPs felt that increasing analytic rigor and predictive modeling with artificial intelligence and machine learning would support their population health efforts. API-based data exchange was generally identified as an opportunity to promote interoperability. Hospitals mentioned that the use of applications with this technology will benefit remote patient monitoring and chronic care management. API-based data sharing was also identified as an opportunity to improve interoperability in behavioral health despite the focus on more basic technological opportunities, such as increasing EHR functionality.

The regional HIEs also shared these perspectives on API-based data sharing and were particularly attentive to FHIR APIs for the development and implementation of EHR-integrated applications to advance interoperability. FHIR-based applications were also expected to improve patient access to their medical records. Although most hospitals and PCPs offered patient access to their medical records through patient portals at the time of our study, these were not unified across providers. FHIR-based applications could potentially enable broader patient access through a unified patient portal that collects information from disparate providers. However, the regional HIEs tempered this enthusiasm, recognizing that FHIR-based applications are not currently broadly deployed in existing technology builds.

Hospitals, PCPs, and HIEs also discussed using care coordination programs (eg, Aunt Bertha) that can track referrals to social service agencies that address social determinants of health (SDoH). These providers described collecting SDoH data in discrete data fields but also noted that the lack of standardization of SDoH data fields remains problematic and results in questionable data quality and limited ability to combine data across sources. Efforts to develop SDoH data standards, such as the Gravity Project [28], were raised as opportunities to improve the interoperability of SDoH data.

Organization Domain

Overview

We identified two themes within the organizational domain that affected interoperability: (1) facilitators and (2) strategic alignment (Table 3).

Facilitators

A consistent subtheme related to the importance of relationships with the EHR vendor to support interoperability emerged. Hospitals placed considerable value on their relationship with Epic and perceived the high concentration of state-wide Epic institutions and the integration between HIEs and Epic as a benefit. Hospitals also mentioned the push for data standards through meaningful use as facilitating interoperability.

PCPs described the important role of senior leadership, who can designate sufficient human resources for tasks that typically increase clinical workload, such as data gathering before appointments. These staff resources can help physicians access and use information from other sources.

Strategic Alignment

A critical driver for interoperability among hospitals was their participation in value-based payment programs and population health initiatives. To facilitate the data exchange and integration for care coordination, billing, and reporting required to support these programs, hospitals reported purchasing EHRs for network partners that lacked these advanced capabilities, thus promoting interoperability among their clinical partners. Likewise, hospitals preferred to develop their own in-house population health analytics platforms rather than outsource this function to HIEs.

PCPs also felt that interoperability is central to achieving strategic goals such as population health management, which is integral to participation in alternative payment models (ie, accountable care organizations, CPC-Ohio, CPC-plus, and Primary Care First). They noted the benefits of event notifications made possible by admission, discharge, and transfer feeds that allow PCPs to be notified when their patients have received care in other settings and follow-up accordingly with them to meet their needs.

Conversely, BHPs indicated that they were not participating in value-based purchasing programs to the same degree. However, similar to hospitals and PCPs, their technology investment decisions were guided by their participation in payment programs with sponsors (eg, Health Resources and Services Administration and Substance Abuse and Mental Health Services Administration) that require specific reports, although they may not necessarily aid in interoperability.

The regional HIEs viewed the development of their exchange networks as an opportunity to advocate for cross-sector alignment, particularly as it pertains to streamlining duplicate efforts toward population health management. They remarked that regional HIEs are in a unique position to negotiate partnerships that address the concerns of different stakeholders. Finally, the regional HIEs noted challenges related to some organizations, particularly BHPs, not viewing interoperability as valuable to their organization or aligning with their strategy.

Environment Domain

Within the environmental domain, a policy theme was identified that focused on how policy may hinder or facilitate interoperability (Table 4).

Hospitals felt that there was room for a greater policy push for managed care plans to initiate value-based payment contracts and distribute incentives to providers. Hospitals also felt the costs of interoperability, such as establishing admission, discharge, and transfer feeds or registry reporting, fall predominantly on hospitals, yet there are no corresponding changes to reimbursement. Similarly, BHPs also desired additional funding to support their adoption of advanced EHR systems.

All providers, including the regional HIEs, also noted the considerable impact of 42 Code of Federal Regulations (CFR) Part 2 (ie, Substance Abuse Confidentiality Regulation) as limiting interoperability and connectivity with BHPs. Despite the general recognition of the policy’s good intentions, it was viewed as negatively impacting clinical care. Some key informants felt that this policy was incongruent with new information blocking rules as part of the 21CCA. The regional HIEs also saw consequences of this policy with respect to responding to public health emergencies such as the opioid epidemic. BHPs, PCPs, and the regional HIEs all advocated for aligning 42 CFR Part 2 with the HIPAA to clarify what is protected and not shareable versus what can be shared for continuity of care.

The regional HIEs noted some specific concerns, given their experience in the COVID-19 response. First, they described a need for greater enforcement of information-blocking rules. The implications of information blocking were particularly notable during the pandemic when withheld information about addresses prevented HIE from tracking COVID-19 at the local level. Second, the regional HIEs felt that HIPAA guidance on reporting on geographic areas with less than 20,000 patients is both vague and confusing, limiting population health efforts.

Finally, the regional HIEs also mentioned that TEFCA helps to establish standards but did not include any information about FHIR-based standards. They noted that this update would be helpful to promote adoption of FHIR-based applications [30].

Discussion

Principal Findings

The near-ubiquitous adoption of certified EHRs over the past decade has resulted in the capture of vast amounts of data across the care continuum. Recent policy efforts such as the 21CCA aim to promote vendor-agnostic integration of external data into the EHRs of all provider types and address information blocking to mitigate challenges. However, questions persist about how well these efforts assist health care organizations in achieving interoperability. Our study examined perspectives from a variety of provider types in Ohio on the state of interoperability. We identified important barriers and facilitators to interoperability among hospitals, PCPs, BHPs, and regional HIEs.

Our findings related to implementation issues within the technology domain suggest that the proliferation of applications that address various use cases promotes capture of rich data.

However, from an end user perspective, this approach can create inefficiencies because of excess information. User interfaces that do not embed multiple discrete applications within the EHR may ultimately create a fragmented medical record, which makes it harder to find relevant information. These types of information silos of patient data can potentially jeopardize patient safety, care quality, and organizational efficiency [31,32]. Fragmented or siloed information may also contribute to provider burnout [33]. Thus, our findings highlight a need for user-centric approaches in technology design and implementation to translate increased information access to use.

Providers did note that technological advances such as computer vision-based population health software (eg, Innovaccer) can help overcome these barriers by pulling data directly into the EHR without requiring back-end integration. Likewise, the growing support for API-based data sharing can further support interoperable information exchange between dissimilar EHR vendors. APIs can extend EHR capabilities [34], although this potential remains unrealized to date. Recent initiatives such as TEFCA that mandate sharing standardized sets of data and promote the use of FHIR APIs are expected to facilitate interoperability in the coming years [30]. Moreover, the API approach has the potential to facilitate population health analytics using machine learning techniques [35].

Within the technology domain, our findings reiterated a gap between hospitals and PCPs at one end of the spectrum and BHPs at the other with respect to their interoperability capabilities. This gap may be a result of most BHPs being ineligible for federal incentive programs [18]. This omission likely not only disincentivizes the adoption of advanced EHRs capable of interoperability among BHPs but also discourages investment and development by EHR vendors of tools designed to meet the needs of BHPs [36].

In addition, across the technology, organization, and environment domains and across provider types, a notable issue that emerged from the key informant interviews was the limitation on interoperability imposed by 42 CFR Part 2, a rule that restricts sharing substance use and behavioral health data. Our findings suggest that despite its intentions, 42 CFR Part 2 effectively operates to prevent, or severely limit, BHPs from participating in exchange. In addition to potential medical errors, this restriction may result in missing data from analytic data sets used by providers, insurers, or researchers [37]. Further, providers expressed concern that this rule may no longer be in alignment with 21CCA information-blocking rules. Moving forward, modifications to 42 CFR Part 2 may be necessary to support further interoperability. Consideration of end user needs and incorporating perspectives of BHPs are essential in any policy changes to carefully consider the need for privacy related to substance use and behavioral health data balanced against the benefits of interoperability [38]. To this effect, the United States Department of Health and Human Services issued a Notice of Proposed Rulemaking in November 2022 [39]. The Notice of Proposed Rulemaking proposes to permit the use and disclosure of Part 2 records based on a single prior signed consent, to expand prohibitions on the use and disclosure of Part 2 records in legal proceedings and to expand patient rights that align with the HIPAA Privacy Rule. The importance of this issue has recently become more visible because of concerns around the protections of reproductive health information following increased abortion restrictions because of the overturning of the Roe versus Wade supreme court case [40,41]. The revisions to Part 2, if enacted, would not only help BHPs engage in interoperability but would also provide greater protection for how sensitive health data can be used in legal proceedings.

Interestingly, hospitals attributed their interoperability capabilities more to their relationship with EHR vendors as opposed to the regional HIEs. This finding is likely a consequence of the strong foothold of Epic in Ohio. From an operational and governance standpoint, providers may face fewer barriers to participating in Epic’s Care Everywhere vendor-mediated HIE network. Conversely, participation with a regional HIE may require further effort to establish data exchange policies that require buy-in from multiple stakeholders. Vendor-mediated HIE may address barriers to use, such as the need to leave the EHR to access information. Nonetheless, these vendor-mediated HIEs may create a divide in HIE participation between providers with different vendors [42]. Indeed, as the HITECH funding period drew to a close, the number of state and regional HIEs declined, partly because of the mergers of regional HIEs, funding challenges, and competition from vendor-mediated HIEs [43]. Vendor-mediated and regional HIEs are not necessarily mutually exclusive, although regional HIEs may be more inclusive of a range of provider types. Policies targeted at supporting regional HIEs may be necessary to counteract the market forces of vendor-mediated HIEs and keep interoperability obtainable for nonhospital or hospital-affiliated practices.

Our findings related to strategic alignment may offer useful policy recommendations; provider participation in value-based payment programs plays a critical role in how providers are considering (or not considering) investments in interoperability. To the extent that most of a provider’s patients are beneficiaries of these programs, providers may expand the breadth of their interoperability functionality, such as participation in regional HIEs or use of population health platforms to meet the needs of that particular patient population [44]. Key informants also described the use of social needs referral platforms; however, developing exchange with non-HIPAA–covered social service or community-based organizations can be challenging without properly aligned incentives [45]. In addition to clarifying HIPAA rules around exchange with noncovered entities, the expansion of value-based payment programs may leverage HIT investment. Further, regional HIEs may be well positioned to advocate for cross-sector strategic alignment.

The providers in our study reported capacity-related challenges in interoperability with public health agencies. Chronically underfunded public health systems impeded efficient and timely electronic information exchange during the COVID-19 pandemic [46]. In response, the Centers for Disease Control and Prevention launched the Data Modernization Initiative, resulting in changes to core data sources and facilitating access to electronic case reports and the COVID electronic laboratory reporting that makes test results available. Other initiatives such as TEFCA, through their emphasis on interoperability, are also expected to mitigate challenges, particularly from differing vocabulary standards. Other barriers to interoperability in public health arise from the complex legal and regulatory environment [47]. Even though 21CCA established a legal framework to address information blocking [48], the HIEs participating in our study reported information blocking that prevented tracking cases of COVID-19. Moving forward, it will be critical to monitor the impact of the Data Modernization Initiative and TEFCA on the interoperability of public health data.

Limitations

We purposely sampled from multiple stakeholders to gain a representative perspective on interoperability. The design focused on breadth across providers rather than depth within a specific provider type or within a single health care organization. Similarly, the study only included Ohio stakeholders, which minimizes variation in the policy environment and may limit generalizability.

The sampling approach was designed to include individuals with decision-making authority with respect to HIT. These perspectives may differ from those of other end users. Finally, owing to time and resource constraints, the interview guide may not have probed all issues relevant to interoperability but was purposefully open-ended to allow participants to discuss topics they deemed important.

Conclusions

Our findings suggest that despite the ubiquity of data and applications, seamless interoperability into a comprehensive medical record, both within and across providers, remains out of reach. Technological solutions offer promise to overcome these challenges. Likewise, the expansion of value-based payment programs can further incentivize interoperability. Although policy initiatives to expand interoperability existed, they were often misaligned to operational needs and may not be sufficient to overcome market forces. A policy focus toward embracing user-centric design to incorporate end user experience into HIT development may overcome barriers associated with achieving interoperability.