Adoption of Clinical Data Exchange in Community Settings: A Comparison of Two Approaches

Abstract

Adoption of electronic clinical data exchange (CDE) across disparate healthcare organizations remains low in community settings despite demonstrated benefits. To expand CDE in communities, New York State funded sixteen community-based organizations to implement point-to-point directed exchange (n=8) and multi-site query-based health information exchange (HIE) (n=8). We conducted a cross-sectional study to compare adoption of directed exchange versus query-based HIE. From 2008 to 2011, 66% (n=1,747) of providers targeted for directed exchange and 21% (n=5,427) of providers targeted for query-based HIE adopted CDE. Funding per provider adoptee was almost two times greater for directed exchange (median (interquartile range): $25,535 ($17,391–$42,240)) than query-based HIE ($14,649 ($9,897–$28,078)), although the difference was not statistically significant. Because its infrastructure can cover larger populations using similar levels of public funding, query-based HIE may scale more broadly than directed exchange. To our knowledge, this is among the first studies to compare directed exchange versus query-based HIE.

Introduction

Electronic clinical data exchange (CDE) across disparate healthcare organizations has the potential to improve care delivery, strengthen public health efforts, and reduce costs (1,2). For large academic medical centers, sharing clinical data with other institutions has been associated with decreased laboratory test ordering (3) and emergency department charges (4). Additionally, regional efforts promoting CDE usage in emergency departments have been associated with reduced costs (5,6). Despite the demonstrated benefits of CDE, widespread adoption remains low (7), particularly in smaller community settings (8).

To increase adoption of CDE, multiple organizational and technological options exist. Non-profit health information organizations (5), accountable care organizations (9), market-oriented health information technology vendors (10), incentives-driven independent healthcare practitioners (11), and health-minded consumers are pursuing different clinical goals, financial motivations, and administrative structures for CDE. Technological approaches for CDE include point-to-point directed exchange of clinical data such as laboratory results and patient referrals, query-based health information exchange (HIE) for aggregating patient data from multiple sites, and consumer-mediated exchange using personal health records (12). The multiple forms of exchange address different but complementary clinical information needs (11).

Despite the diversity of available CDE approaches, studies comparing organizational and technological configurations for CDE are limited, and optimal approaches to CDE are unknown. In New York State, two types of community-based organizations received funding to implement CDE as part of an effort to create a statewide health information network; clinically affiliated groups of providers implemented directed exchange, and regional organizations implemented query-based HIE. The objective of this study was to compare adoption of and barriers to the two CDE approaches.

Background

New York State has 19 million residents and 91,918 healthcare providers, of which 75% (n=68,898) are physicians (medical doctors (MDs) and doctors of osteopathic medicine (DOs)) (75%) and 25% (n=23,020) are physician extenders (nurse practitioners (NPs) and physician assistants (PAs)) (13). Defined policies for CDE exist in New York State. Directed exchange of patient data does not require affirmative patient consent, as point-to-point data transmission mimics existing fax and telephone-based clinical information sharing. However, a patient must opt in to query-based HIE participation by providing affirmative consent to each healthcare organization that wishes to access his or her data. Clinicians also have the ability to “break the glass” and access query-based HIE during emergencies.

HEAL NY

New York State has invested more than $800 million in health information technology through four phases of the Healthcare Efficiency and Affordability Law for New Yorkers (HEAL NY). Launched in 2006, HEAL 1 invested in community CDE capacity, electronic health record (EHR) adoption, and electronic prescribing. In 2008, HEAL 5 aimed to increase adoption of interoperable EHRs in communities as well as expand and connect multiple community-level CDE efforts to form the Statewide Health Information Network of New York (SHIN-NY). To develop and govern the SHIN-NY, New York State established the New York eHealth Collaborative (NYeC), a public-private partnership with oversight from the New York State Department of Health. Underway at the time of this writing are HEAL 10, which addresses EHR-based patient centered medical home implementation, and HEAL 17, which focuses on EHR-based care coordination; these phases also expand and facilitate statewide CDE. The current study focused on activities funded through HEAL 5.

HEAL 5

To expand CDE in communities, HEAL 5 awarded funding to sixteen organizations of two types: community health information technology adoption collaborations (CHITAs) comprised of community-based providers affiliated for care purposes but not under the same corporate umbrella and non-profit, non-governmental regional health information organizations (RHIOs) that convened and governed multiple community stakeholders in a defined geographic area. Whereas RHIOs were legally established entities, CHITAs were less formal alliances of providers with leadership from an organization such as a community hospital. CHITAs promoted adoption of directed exchange via software license purchase and implementation assistance of health information technology, particularly certified EHRs, that featured point-to-point interfaces for laboratory result delivery, electronic prescribing, quality reporting, and/or clinical transfer forms. RHIOs promoted adoption of query-based HIE via purchase of software licenses and implementation assistance for standalone web portals and/or EHR interfaces connected to federated or centralized repositories with master patient index, record locator, provider directory, user authentication, and patient consent management services.

CHITAs and RHIOs agreed to participate in a statewide collaboration process and adhere to state technology and policy standards. CDE activities of HEAL 5-funded organizations addressed up to three use cases aligned with state clinical and public health priorities that were based on Office of the National Coordinator for Health Information Technology (14) and Centers for Disease Control use cases (15). Organizations that received HEAL 5 funding also agreed to participate in evaluation activities conducted by the Health Information Technology Evaluation Collaborative (HITEC), the multi-institutional academic consortium charged with evaluating HEAL NY. Members of HITEC conducted this study.

Methods

Using data collected in 2010 and 2011, we conducted a cross-sectional study of organizations awarded funding through HEAL 5 to implement CDE in communities in New York State. Measures included adoption of CDE and barriers to implementation. We compared measures between CHITAs that implemented directed exchange and RHIOs that implemented query-based HIE. Understanding adoption of CDE by providers (16,17) and patients (18) as well as barriers to implementation (19) can inform best practices for CDE. The Institutional Review Board of Weill Cornell Medical College approved this study.

Data collection

To measure adoption of CDE, we obtained data in March 2011 from sources available to the research team. From NYeC, we obtained the number of providers (e.g. MD, DO, NP, PA) who adopted query-based HIE and the number of patients who affirmatively consented to query-based HIE as reported by each RHIO per state requirement. From unpublished survey data collected by the research team as part of general HEAL 5 evaluation activities (20), we obtained the number of providers who adopted EHRs or other health information technology with directed exchange as reported by each CHITA. From HEAL 5 grant applications, we obtained the number of providers and patients targeted for adoption by each organization as well as details about technology implementations. From the New York State Department of Health website (21), we obtained funding amounts for each organization’s HEAL 5 activities.

To measure barriers to implementation, we conducted a survey in November 2010. For each use case that an organization implemented, an organization’s executive director (or executive director’s designee) rated each of thirteen items as having served as a barrier, served as a facilitator, or had no effect. The thirteen items were based on a previous survey instrument developed and administered by members of the research team (22). If a survey respondent rated an item as a barrier for at least one use case, we considered the item as having served as a barrier for the respondent’s organization. The survey contained four other sections not used in the current study addressing organization and governance, financial sustainability, statewide collaboration, and federal context.

Data analysis

We compared adoption of CDE and barriers to implementation between CHITAs and RHIOs. We determined descriptive statistics including mean (±standard deviation (SD)) for normally distributed data and median (interquartile range (IQR)) for non-normally distributed data. To examine differences between CHITAs and RHIOs, we used Fisher’s exact tests for proportions, t-tests for normally distributed data, and Wilcoxcon rank-sum tests for non-normally distributed data. As in a prior study (23), the threshold for statistical significance was p < 0.2. To perform calculations, we used the R statistical software package.

Results

100% (n=16) of organizations awarded HEAL 5 funding in 2008 to implement CDE in communities participated in the survey (Table 1). Multiple types of entities led the directed exchange efforts of CHITAs while health information organizations exclusively implemented query-based HIE. Of CHITAs, seven implemented certified EHRs with interfaces, and one implemented a secure web application for facility-to-facility care transitions.

Table 1.

Characteristics of organizations that implemented CDE.

	Total (n=16)	CHITAs (Directed) (n=8)	RHIOs (Query-Based HIE) (n=8)
Leadership entity
Health information organization, n (%)	9 (56)	1 (13)	8 (100)
Community hospital, n (%)	3 (19)	3 (38)	NA
Federally qualified health center, n (%)	2 (13)	2 (25)	NA
Long-term care facility, n (%)	1 (6)	1 (13)	NA
Municipal health department, n (%)	1 (6)	1 (13)	NA
Total funding in USD millions
Total, n (%)	105.1	30.4 (29)	74.7 (71)
Organization, median (IQR)	(3.5–9.4)	2.9 (1.7–5.5)*	9.6 (7.3–12.3)*
Providers targeted for adoption
Total, n (%)	28,094	2,631 (9)	25,463 (91)
Organization, median (IQR)	738.5 (327.5–2,775)	255 (71–450)*	2,850 (1,000–3,516)*
Patients targeted for adoption
Total, n (%)	13,419,219	1,054,708 (8)	12,364,511 (92)
Organization, median (IQR)	385,000 (103,628–1,290,640)	95,752 (52,176–207,500)*	1,381,281 (986,500–2,275,712)*
Funding per target for adoption in USD
Provider, median (IQR)	9,224 (4,309–15,798)	12,292 (10,682–25,836)*	3,914 (2,592–5,917)*
Patient, median (IQR)	13 (8–27)	27 (22–37)*	7 (5–9)*

^*Denotes p < 0.2

Compared to RHIOs, CHITAs received 59% less total funding (p=0.003) and targeted 90% fewer providers (p=0.029) and 91% fewer patients (p=0.007) for adoption. However, CHITAs received more than twice as much funding per provider targeted for adoption (p=0.003) and almost three times as much funding per patient targeted for adoption (p=0.002) than RHIOs.

Adoption of CDE

Of all providers targeted by organizations for CDE adoption, 24% (n=7,174) adopted CDE. This included 66% (n=1,747) of providers targeted by CHITAs for directed exchange and 21% (n=5,427) of providers targeted by RHIOs for query-based HIE. As shown in Figure 1, the percentage of providers who adopted directed exchange through CHITAs (mean (±SD): 70 (±42)) was twice as great as adopted query-based HIE through RHIOs (35 (±26)) (p=0.067). Of note, one CHITA achieved 140% CDE adoption after 25 more providers than the 62 targeted adopted directed exchange. Funding per provider adoptee was almost two times greater for CHITAs (median (IQR): $25,535 ($17,391–$42,240)) than RHIOs ($14,649 ($9,897–$28,078)), although the difference was not statistically significant (p=0.234).

Figure 1.

Percentage of providers who adopted directed exchange through CHITAs (n=8) and query-based HIE through RHIOs (n=8).

In total, RHIOs targeted more than 12 million patients to opt-in for query-based HIE adoption per New York State requirement. Of the 1,985,841 patients that RHIOs approached regarding query-based HIE participation during the study period, 1,787,257 (90%) provided affirmative consent, representing 14% of all patients targeted by RHIOs. On average, each RHIO obtained affirmative consent for query-based HIE participation from 16 (±11) percent of patients targeted for adoption (Figure 2).

Figure 2.

Percentage of patients who adopted query-based HIE through RHIOs (n=8).

Barriers to implementation

Organizations most frequently identified technology maturity (88%, n=14) and vendor participation (63%, n=10) as barriers to implementation (Table 2). Although CHITAs and RHIOs identified barriers similarly in most cases, a greater proportion of RHIOs (88%, n=7) than CHITAs (38%, n=3) identified vendor participation as a barrier to implementation (p=0.118). Vendors for RHIOs included Axolotl, dbMotion, Lawson, InterSystems, MedAllies, and MedPlus while vendors for CHITAs included Allscripts, eClinicalWorks, General Electric, MDLand, Medent, Meditech, NextGen, and Orion Health.

Table 2.

Barriers to implementation of CDE.

	Total (n=16)	CHITAs (Directed) (n=8)	RHIOs (Query-Based HIE) (n=8)
Technology maturity, n (%)	14 (88)	7 (88)	7 (88)
Vendor participation	10 (63)	3 (38)*	7 (88)*
Privacy and security policies	9 (56)	4 (50)	5 (63)
Regulatory requirements	9 (56)	3 (38)	6 (75)
Technical support	7 (44)	4 (50)	3 (38)
Organizational structure and culture	7 (44)	3 (38)	4 (50)
Workflow integration	7 (44)	2 (25)	5 (63)
Existing standards	7 (44)	2 (25)	5 (63)
Provider attitudes	4 (25)	2 (25)	2 (25)
Financial resources	4 (25)	2 (25)	2 (25)
Certification rules	4 (25)	2 (25)	2 (25)
Health plan participation	1 (6)	0 (0)	1 (13)

^*Denotes p < 0.2

Discussion

This study is among the first to compare adoption of and barriers to directed exchange and query-based health information exchange. Through two types of community organizations, more than 7% of providers and about 9% of patients in New York State adopted CDE between 2008 and 2011. Although a greater percentage of providers adopted directed exchange per CHITA than query-based HIE per RHIO, more than three times as many providers adopted query-based HIE from RHIOs than directed exchange from CHITAs and with New York State funding per provider that was about 50% lower, although not statistically different. Barriers to implementation most frequently identified by organizations were technology maturity and vendor participation, with RHIOs more frequently identifying vendor participation as a barrier than CHITAs.

Federal and state investments have aimed to correct market failure in CDE adoption, but public funding for directed exchange may be to the detriment of existing activities and the public good provided by RHIOs (10). Although financial sustainability of RHIOs is unknown (7) and policymakers expect directed exchange to require less public funding than query-based HIE (10), findings from this study suggest that public funding per provider for query-based HIE may be similar to or lower than public funding per provider for directed exchange. Current federal policy encourages directed exchange that relies on technologies such as provider directories, patient matching, and secure data transport (12), which RHIOs typically already provide for query-based HIE. However, current federal policy does not incentivize use of RHIOs and instead promotes states and commercial entities to build and deliver such services. RHIOs may enable the benefits of CDE to scale more broadly than directed exchange because query-based HIE infrastructure can cover larger populations of providers and patients with similar levels of public funding and more complete data.

As members of Congress question current federal health information technology adoption programs (24) and multiple professional societies request delays to meaningful use deadlines (25), policymakers may find value in revisiting the role of query-based HIE and RHIOs. Existing technology infrastructure for query-based HIE provides a platform for quality improvement, population health management, clinical research, and other rich uses (26). For at least one RHIO, query-based HIE has facilitated care transitions in support of stage two of meaningful use (27), a service other RHIOs can offer to align with provider incentives. Additionally, RHIOs provide a forum for stakeholders to develop trust and govern exchange efforts. For example, health information organizations serving rural areas in the Rocky Mountains and Great Plains recently aligned to support the use of query-based HIE for meeting regional and inter-state patient data needs (28). Query-based HIE and RHIOs add technological and organizational value for CDE absent from current approaches to directed exchange.

Critical attitudes toward vendor-based CDE technology observed in this investigation are consistent with previous reports. In a recent national survey, a quarter of RHIOs expressed concern about “poor customer service, configuration and implementation issues, and cost” of vendor systems for CDE (29). Additionally, small practices and small hospitals have lagged in EHR and CDE adoption (30,31), indicating a need for improved vendor technologies and services in community settings (30,32). Accelerated adoption of CDE, and query-based HIE in particular, will likely require improved relationships between vendors and community organizations.

This study has limitations. First, our sample size was small and statistical comparisons were of borderline significance. Second, self-reported data from community organizations may be subject to respondent bias. Unlike our previous investigations that analyzed HIE system log files (27,33), the current study does not illustrate actual sharing of patient data. Third, unmeasured barriers and facilitators may have affected CDE implementation. Specifically, funding from non-HEAL NY sources may have contributed to adoption of CDE. Additionally, respondents may have interpreted the meaning of barriers differently due to potentially ambiguous definitions. Regardless, results of this study suggest a need to test assumptions about the cost of CDE approaches, including directed exchange (12). Finally, findings may not generalize beyond New York State, as privacy, technology, financial, and other factors may vary.

Conclusion

To our knowledge, this study is the first to quantify differences in adoption of and barriers to directed exchange and query-based HIE. Using similar levels of public funding, query-based HIE may enable broader adoption of clinical data exchange than directed exchange.