Usage and Effect of Health Information Exchange

Key Summary Points Relatively few of the more than 100 operational U.S. health information exchanges (HIEs) have been the subject of published evaluations. Of the HIEs that have been evaluated, low-quality evidence from 12 hypothesis-testing studies supports an effect of HIE use on reduced utilization or costs in the emergency department. The effects of HIE on other health care outcomes are uncertain. The use of HIE is low relative to the estimated potential need; most studies reported use in 2% to 10% of encounters. However, some sites have reported much greater HIE use, and specifics of the context and implementation may be responsible for these differences. All stakeholders claim to value HIE, but many barriers to acceptance and sustainability exist, including workflow and interface issues, privacy and security of patient health information, and the lack of a compelling business case for sustainability. Health care providers are increasingly sharing clinical data with other providers who care for the same patient by using electronic health information exchange (HIE). In the United States, more than 100 organizations facilitate HIEs among provider organizations, and 30% of hospitals and 10% of ambulatory clinics participate (1). This development has been driven by the belief that information exchange will address the current fragmentation of patient care across providers, thereby improving care, enhancing patient satisfaction, and making care more efficient by reducing the use of redundant health care services (2, 3). Use of HIE is also spurred by financial incentives created by the Health Information Technology for Economic and Clinical Health Act of 2009 and outlined in the Centers for Medicare & Medicaid Services meaningful use rules. Stage 2 meaningful use objectives include exchanging clinical data with other providers and may be expanded in stage 3. Nearly $600 million in federal funding was designated to support statewide HIE organizations. Some states have invested substantial additional funding (4). Although policy already creates incentives for providers and hospitals to achieve these objectives, little is known about how HIE capabilities are used in clinical care and their observed effects. This information could inform better design and implementation of HIE programs to improve patient outcomes and achieve efficiencies. We conducted a systematic review to evaluate the HIE literature, specifically investigating the evidence of effect on health outcomes, health care utilization and efficiency, evidence of clinicians' usage of HIE, and financial sustainability of HIE organizations. We also evaluated evidence about patient and providers attitudes toward HIE as well as barriers or facilitators to HIE use. Methods This systematic review is reported according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (5). A formal protocol was developed and submitted to PROSPERO, which included the key questions, search strategy, and inclusion and exclusion criteria (CRD42014007469). A technical expert panel provided input on the protocol and preliminary results. Data Sources and Searches We conducted a review of key terms related to HIE from PubMed, Web of Science, the Cochrane Databases, and gray literature from 1 January 2003 to 31 May 2014 (Supplement 1). We obtained additional references from key articles (reference mining) and content experts, and we used results from previous health information technology (IT) literature reviews conducted between 1 January 1990 and 13 August 2013, which included HIE (69). Because approaches to HIE are embedded within the idiosyncrasies of health systems and our focus was the U.S. health care system, we limited the search to articles published in English and within the U.S. context. Supplement 1. Search Strategy Study Selection We focused on the potential to exchange several types of clinical information on specific patients across institutions in different health systems, which is consistent with the definition described in the U.S. national HIE strategy and with the definition of a basic HIE according to Adler-Milstein and colleagues (10). We excluded administration and financial information exchange; studies that involved the transmission of data for public health, such as those on reportable diseases; single-focus data, such as those about antimicrobial resistance or HIV laboratory tests; data exchange across facilities but within systems, such as across Veterans Affairs or Kaiser Permanente facilities; and 1 study about exchange of data between the U.S. Department of Defense and Veterans Affairs because of limited generalizability. Articles were eligible if they met this definition of HIE and if the study design was descriptive, qualitative; descriptive, quantitative; or hypothesis-testing (including randomized controlled trials) following classifications first used by Chaudhry and colleagues [8]. Articles were also eligible if they were systematic reviews. Descriptive, qualitative studies that provided only a description of an HIE were not included, but those that used qualitative methods to evaluate barriers or facilitators were included. Descriptive, quantitative or hypothesis-testing papers were included if they focused on clinical or health outcomes, efficiency, health care utilization, patient or provider satisfaction, attitudes, HIE usage, and financial sustainability of HIE organizations (11). We excluded a class of papers we called case studies, which were narrative descriptions about particular HIEs (which included their history, architecture, and challenges) but were distinguished from included studies in that they did not include Methods sections or a priori study questions about outcomes, use, costs, sustainability, or attitudes or barriers. Data Extraction and Quality Assessment For hypothesis-testing studies of outcomes and efficiency, we extracted data about the type of HIE, settings, outcomes reported, and results. We classified the type of HIE evaluated in these studies into 3 categories designated by the Office of the National Coordinator for Health Information Technology: query-based, directed, and consumer-mediated (11). We created a separate category for more advanced data exchange functionality, such as HIE-generated alerts (Table 1). Table 1. Classifications of HIE For studies of sustainability, we extracted information about the HIEs under study, study period, method of data collection, and results. For studies of usage, we extracted the same information as for sustainability as well as data on HIE functionality. For studies of attitudes and barriers, we extracted data about the sample and its generalizability, the stakeholders assessed, and the findings. All data extraction was done by 1 reviewer and then independently evaluated by the other. Data for hypothesis-testing studies and descriptive studies of HIE usage were independently extracted by a third reviewer with consensus resolution of discrepancies. We assessed the quality of hypothesis-testing studies using study design, specification of the HIE, and data about context and implementation. For the latter, we used an existing framework for implementation science (13), context and implementation recommendations specific to health IT (14), and input from HIE experts to select 2 criteria for context (a description of the existing health IT capabilities of the participating providers and previous history of cooperation or engagement at the organizational or community level) and 4 criteria for implementation (local tailoring or iterative process of development or implementation that considered local workflow; the use of an internal or external designated person or persons responsible for implementation, use of clinical champions, or super users; a description of efforts to promote awareness and engagement; and other barriers or facilitators not already listed). For descriptive studies, we assessed quality by the representatives of the sample in terms of geographic area and response rate. Data Synthesis and Analysis Because of the variability in HIE characteristics and clinical setting in which they were used, we determined that a meta-analysis was inappropriate. Instead, we constructed evidence tables showing the study characteristics and results for all included studies, organized by outcome measure, and critically analyzed studies to compare their characteristics, methods, and findings. We compiled a summary of findings for each general class of outcome and used them to draw conclusions. We assessed the strength of evidence for hypothesis-testing studies with the Grading of Recommendations Assessment, Development and Evaluation criteria (15). Role of the Funding Source Funding was provided in part by the U.S. Department of Veterans Affairs. The funding source had no role in the design, analysis, or interpretation of the data or the decision to submit the manuscript for publication. Results Description of the Evidence We reviewed 1149 titles and abstracts from the electronic search (including gray literature), 2 articles from content experts, 1 article from a peer reviewer, 148 articles categorized as health information exchange or clinical data exchange from the related systematic review, and 14 articles from reference mining, for a total of 1314 references. From this, we identified 85 papers on HIE that addressed our study questions. Studies could address more than 1 category. The Appendix Figure details the inclusion of studies and the number of studies related to each key question. The abstracted data for the included articles can be found in Supplements 2, 3, 4, 5 and 6. All studies that specified the type of HIE functionality involved query-based HIE. We did not find any evaluations of directed exchange or consumer-mediated HIE except 1 study that combined query-based and directed exchange and 1 other study that evaluated a nonsustainable, consu