Engagement In Hospital Health Information Exchange Is Associated With Vendor Marketplace Dominance

There is broad agreement that health information exchange (HIE)—through which patients’ health information, such as laboratory results, medication lists, and clinical summaries, is transferred electronically across provider organizations—should enable better, more efficient care. ^1,2 Accordingly, policy makers have pursued an array of approaches to increase HIE adoption. ³ While some progress has been made, technical, organizational, financial, and regulatory barriers have limited the diffusion of HIE. ^1,4–6 And while HIE remains a leading policy priority, ⁷ there is little consensus about how to achieve broader uptake. ^3,8,9

Recent attention has focused on the efforts of electronic health record (EHR) vendors to build HIE capabilities into their products. Such efforts have been supported by 2014 EHR certification requirements (which are intended to ensure that EHRs can support meaningful-use criteria) issued by the Office of the National Coordinator for Health Information Technology (ONC) of the Department of Health and Human Services. The requirements specify that all EHR vendors support a basic HIE capability akin to secure e-mail. ^8,10

Providers seeking to establish robust HIE connections with other providers will find doing so easier if they use the same EHR vendor, because their systems will store and treat patient data in similar ways. Indeed, some vendors have made it easier for hospitals and other providers that use their products to engage in HIE with each other than to engage in HIE with other vendors’ competing systems. ^9,11 In addition, there have been concerns that vendors may be strategically putting up barriers to HIE across different vendors’ systems. Specifically, ONC has cited concerns that EHR vendors are charging fees “that make it cost-prohibitive for most customers to send, receive, or export electronic health information stored in EHRs” and may be refusing to allow information exchange with other providers or EHR vendors for competitive reasons. ¹⁰

If these actions are occurring, they could create “islands” of HIE in which providers have limited ability to share clinical data about patients who see providers that use different EHR systems. ^10–12 Such islands would have negative effects on the cost and quality of care. ¹³ As a result, policy makers have expressed concern that the concentration of EHR vendors may influence the incentives and resulting business practices through which vendors facilitate or impede HIE. In other words, because of either the ease of connecting two systems from the same vendor or the specific business practices of vendors to make cross-vendor exchange difficult, a hospital using the same vendor as many other hospitals in its market may be more able to engage in HIE compared to hospitals using different vendors. ^12,14

However, these concerns may fail to recognize the potential for a positive spillover effect. Greater vendor dominance in the market could promote engagement in HIE by hospitals that do not use the dominant vendor, because the cost and complexity of interfacing with a single other vendor’s system are less than those of interfacing with multiple different vendors’ systems. ¹⁵ In contrast, in markets with low vendor dominance, the cost of hospital engagement in HIE with multiple vendors’ systems may be prohibitive, and vendors may be less interested in collaborating because competition is stronger.

Before policy makers devise actions to take in response to allegations of interference on the part of EHR vendors or to a more benign scenario in which cross-vendor HIE is simply too complex for providers to pursue, it is important to empirically assess the extent to which patterns of market dominance by EHR vendors are associated with more or less hospital engagement in HIE. In this study, we used data for 2012 and 2013 from national surveys of hospitals to answer three research questions. First, do hospitals that use the dominant vendor in a market engage in more HIE than hospitals that do not? Second, does the difference in level of HIE engagement between these two groups of hospitals vary based on the level of vendor dominance in the market? And finally, does the relationship between market dominance and hospital engagement in HIE vary by specific EHR vendor?

Study Data And Methods

Data And Sample

Our national data consisted of information about all US hospitals in the fifty states and the District of Columbia. We merged data from three sources. Information about our key EHR- and HIE-related variables came from the 2012 and 2013 American Hospital Association (AHA) Annual Survey Health Information Technology (IT) Supplements, which have been used in previous research to measure hospital engagement in HIE. ^12,16 We combined these with control variables from the 2012 AHA Annual Survey and the 2013–14 Area Health Resources File (AHRF).

To create our analytic data set, we first matched hospitals to their respective hospital referral region (HRR), a commonly used geographic definition of hospital markets. ¹⁷ Next, we ensured that we had a sufficient number of respondents to the AHA IT Supplement within each HRR to create accurate measures of EHR vendors’ market-level penetration. We therefore calculated HRR-level response rates by dividing the number of hospitals that responded to the AHA IT Supplement by the total number of hospitals in the HRR. We excluded HRRs in which less than one-third of hospitals responded, and for HRRs with fewer than ten hospitals, we excluded those in which fewer than three hospitals responded. This resulted in our including 284 of the 306 HRRs. In the included HRRs, on average, 68.5 percent of hospitals responded. The characteristics of hospitals in our sample are reported in online Appendix Exhibit 1. ¹⁸

Key Measures

Vendor Concentration:

In the 2012 and 2013 AHA IT Supplements, responding hospitals indicated which of sixteen EHR vendors (including “self-developed” and “other”) they used for inpatient services. We assigned each hospital to its 2013 vendor or, for hospitals that did not respond in 2013, to its 2012 vendor (if that information was available). If hospitals responded to the IT Supplement but did not report an inpatient vendor, we treated them as having no vendor, instead of as having missing data.

Although EHR markets are national, we developed measures of EHR market share at the level of the local hospital market (HRR) because a hospital’s decision to adopt a specific EHR system or invest in HIE is likely to be influenced by the decisions of the hospitals near it. Similarly, a vendor’s interest in facilitating HIE in a given market may depend on the vendor’s local market share and prospects.

To measure the market share of each EHR vendor in each hospital referral region, we divided the number of hospitals that reported using each vendor by the total number of respondent hospitals in that HRR (including those with no vendor). We weighted this proportion by hospital size, measured by number of hospital beds, to improve our ability to capture the volume of information held by each vendor. This allowed us to identify the most dominant vendor in the HRR and the proportion of the market it controlled. Next, to capture the overall level of vendor concentration in each market (for example, to differentiate between a market with a 60 percent–40 percent split and one with four vendors that have market shares of 60 percent, 20 percent, 10 percent, and 10 percent, respectively), we created a marketwide measure of vendor concentration using a Herfindahl-Hirschman Index based on the number of beds in hospitals that used each EHR vendor.

We then used these market-level measures of vendor dominance and vendor concentration to create a set of hospital-level measures. To address our first research question (whether hospitals that used the dominant vendor engaged in more HIE than hospitals that did not), we created an indicator variable that captured whether or not the hospital used the dominant vendor in its market.

To address our second research question (about the varied relationship between the level of vendor dominance and HIE engagement for the group of hospitals that use and do not use the dominant vendor), we created two variables. In one variable, we assigned every hospital in a given HRR the market share of the most dominant vendor in its HRR (that is, the vendor with the largest market share), regardless of whether or not the hospital used that vendor. In the second variable, we assigned the marketwide measure of vendor concentration in a hospital’s HRR to that hospital.

To answer our third research question (whether the relationship between market dominance and hospitals’ engagement in HIE differs according to EHR vendor), we first split markets into those in which Epic Systems Corporation was the dominant vendor and those in which another vendor was dominant. We did this because we suspected that Epic’s market dominance would be more influential in increasing hospitals’ engagement in HIE than other vendors’ dominance, given Epic’s large Care Everywhere Network, ¹⁹ which is designed to easily enable HIE between its participants but not between the participants and providers who use other vendors.

To measure the effect of Epic’s market share on HIE, we assigned Epic’s market share to all hospitals in HRRs in which Epic was the dominant vendor, and we coded hospitals in all other HRRs as having missing data. To measure the effect of other vendors’ market share on HIE, we assigned all hospitals the market share of the dominant vendor in HRRs in which Epic was not the dominant vendor, and we coded hospitals in Epic-dominated HRRs as having missing data.

We then further split markets in which Epic was not the dominant vendor into those in which each of the next three largest vendors—Cerner Corporation, Medical Information Technology Inc. (Meditech), and McKesson Corporation—were dominant. We created three measures (one for each vendor) akin to the Epic measure described above, in which all hospitals were assigned the market share of the dominant vendor in their HRR and all others were coded as having missing data. For example, the Cerner measure was populated only for hospitals in HRRs in which Cerner was dominant (60 of the 284 HRRs), and all hospitals in those HRRs were assigned Cerner’s market share in their market (ranging from 15.1 percent to 97.8 percent).

Hospitals’ Engagement In Health Information Exchange:

In the AHA IT Supplement, hospitals responded to ten questions related to their HIE activity. For each of five different types of data (patient demographics, laboratory results, medication history, radiology reports, and clinical or summary care records), hospitals were asked whether or not they exchanged data with hospitals outside of the respondent’s system and with ambulatory care providers outside of the respondent’s system. Responses across the five types of data within a given exchange partner were highly correlated, as were responses to items across types of exchange partners. We therefore summed these measures to create the following three scales: one ten-item scale, which included all responses to both outside hospital and outside ambulatory care provider questions; and two five-item scales, one for each type of exchange partner. This allowed us to assess whether there was a different relationship between vendor dominance and hospital HIE with outside hospitals versus HIE with outside ambulatory providers.

Control Variables

We included several variables from the AHA Annual Survey and IT Supplement that could confound the relationship between vendor dominance and hospital HIE. These were whether the hospital had at least a basic EHR, hospital size (0–99 beds, 100–399 beds, or 400 or more beds), hospital (or multihospital system) market share, teaching status (none, minor, or major), urban or rural location, system membership, network membership, ownership (government, nonprofit, or for profit), market-level Herfindahl-Hirschman Index based on bed size, percentage of Medicare patient days, and percentage of Medicaid patient days.

We also included the following control variables that captured patient demographics for the county in which the hospital was located: population density, percentage of the county’s residents older than age sixty-five, percentage of the county’s residents older than age twenty-five with less than a high school diploma, unemployment rate, and population density. These characteristics may be related to hospitals’ available resources to pursue HIE in the market.

Finally, we included variables related to the availability of health care services, such as physician density per thousand people and hospital beds per thousand people. These may relate to the level of competition between providers in attracting and keeping patients, which is believed to influence HIE participation. ²⁰

Analytic Approach

To address our first research question, we modeled the relationship between the indicator variable for whether or not the hospital used the dominant vendor and hospital HIE, controlling for hospital and market characteristics. To address our second research question, we ran a model that interacted the binary indicator of whether or not the hospital used the dominant vendor in its market with the proportion of the market covered by that vendor. We then ran a second model that substituted the dominant vendor’s market share for overall EHR vendor market concentration, to observe the effect of increasing concentration (that is, decreasing competition) on hospital HIE engagement.

Finally, we investigated our third research question by rerunning our model that interacted vendor dominance and the indicator for whether or not the hospital used the dominant vendor for five subgroups. First, we examined the relationship between vendor dominance and hospital engagement in HIE in markets where Epic was the dominant vendor and in the remaining markets where a different vendor was dominant. We then examined the relationship between greater dominance and hospital HIE in markets in which each of the next three largest vendors had the greatest market share.

We ran all models at the hospital level so that we could control for hospital-level factors that might influence the relationship between vendor dominance and hospital-level HIE. Because our outcome was count data that ranged from 0 to 10 and exhibited overdispersion, we used negative binomial regression models. We clustered standard errors by HRR to account for grouping of hospitals within markets.

Robustness Tests

We performed several robustness tests to lend support to our results. First, we used an alternative definition of vendor dominance based on the percentage of hospitals instead of hospital beds, because the number of hospitals may be more influential in determining hospitals’ engagement in HIE than our primary measure, which was weighted by bed size.

Second, to address concerns related to incomplete measurement of EHR dominance in HRRs, we limited our sample to the 246 HRRs in which more than 50 percent of hospitals responded to the AHA IT Supplement. These HRRs had an average response rate of 72.2 percent, versus the response rate of 68.5 percent for the 284 HRRs in our primary analyses.

Third, we separately examined the relationship between vendor dominance and HIE with outside hospitals versus HIE with outside ambulatory providers, to ensure that neither category was driving our main results.

Limitations

Our study had several limitations. First, our findings focused on the influence of EHR vendor dominance on hospitals’ engagement in HIE, and we measured HIE engagement based on whether or not hospitals reported exchanging various types of information with outside hospitals and ambulatory providers—not the extent to which they reported doing so within any given category or the extent to which they reported doing so within versus across various vendor systems. As a result, we could not know whether a hospital that reported, for example, exchanging test results with hospitals not owned by the same parent organization was doing so with only a single hospital or with multiple hospitals, or whether hospitals were strategically selecting exchange partners. We were less concerned about the latter point than the former because we could not think of a plausible mechanism through which increased dominance of one vendor would influence the HIE engagement of hospitals using a nondominant vendor by affecting only the extent of exchange between hospitals using that nondominant vendor.

Second, our study was cross-sectional, and the relationships we observed were associational. The findings should not be taken to suggest or imply that any particular EHR vendor has engaged in any improper or questionable business practices. Other factors may be involved in hospitals’ and providers’ engagement in HIE, including the availability of incentives. It is also possible that some hospitals and providers engage in information-blocking practices to maintain and protect their patient and practice bases.

Third, we recognize that EHR vendors’ market share and hospitals’ engagement in HIE are rapidly changing. Our results, which relied on data from 2012 and 2013, therefore represent a baseline against which future assessments can be compared.

Finally, our results address only hospital dynamics. It will be important to study these relationships in the ambulatory care setting before pursuing a broad policy response.

Study Results

Descriptive Statistics On Vendor Dominance

The analytic sample included 2,924 hospitals. Among these, Epic was the largest EHR vendor, accounting for 23.5 percent of all beds, and it was used in 485 (16.6 percent) hospitals (Appendix Exhibit 2). ¹⁸ Cerner was second, with 19.9 percent of beds, and it was used in 441 (15.1 percent) hospitals. It was followed by Meditech, with 15.9 percent of beds in 563 (19.2 percent) hospitals; McKesson, with 9.3 percent of beds in 285 (9.7 percent) hospitals; Allscripts Healthcare Solutions, with 6.9 percent of beds in 106 (3.6 percent) hospitals; and Siemens, with 6.4 percent of beds and 166 (5.7 percent) hospitals.

Across the 284 HRRs in the study, the dominant vendor accounted for 50.5 percent of the market, on average ( Exhibit 1 ). Epic had the largest market share (51.6 percent, on average) in ninety-three HRRs (32.7 percent of all HRRs).

Relationship Between Health Information Exchange And Using The Dominant Vendor

Hospitals that used the dominant vendor engaged in significantly more HIE activities than hospitals that used a different vendor (for details, see Appendix Exhibit 3, column 1). ¹⁸ Marginal effects estimates showed that hospitals that did not use the dominant vendor engaged in 3.1 out of the 10.0 HIE activities described above, while hospitals that did use the dominant vendor engaged in 4.6 out of 10.0—an increase of 45 percent (data not shown).

Relationship Between Vendor Dominance And Hospital Health Information Exchange

Hospital HIE engagement did not vary based on the level of EHR vendor dominance for hospitals that used the dominant vendor. In contrast, as EHR vendor dominance increased, hospitals that did not use the dominant vendor engaged in more HIE (Appendix Exhibit 3, column 2). ¹⁸ This pattern of results was replicated when we used market concentration in place of market share of the most dominant vendor (Appendix Exhibit 3, column 3). ¹⁸

Marginal effects estimates showed that, with all other variables at the mean, at a low level of market dominance (that is, when the largest vendor had a 20 percent market share), hospitals using the dominant vendor engaged in 4.3 of 10.0 HIE activities, while hospitals not using that vendor engaged in 2.7 of them—a 59 percent difference ( Exhibit 2 ). At 80 percent market dominance, hospitals using the dominant vendor engaged in 5.0 of 10.0 HIE activities, while hospitals not using that vendor engaged in 4.0 of them—a 25 percent difference.

SOURCE Authors’ analysis of data for 2012 from the American Hospital Association Annual Survey and for 2012–13 from the survey’s Health Information Technology Supplement. NOTES This figure was created using marginal effects estimates derived from model 2 in Appendix Exhibit 3 (see Note  20 in text). The ten HIE activities analyzed are described in the text.

Relationship Between Vendor Dominance And Health Information Exchange By Vendor

We observed different relationships between the level of vendor dominance and hospital HIE, depending on which vendor was dominant. In markets where Epic was dominant, hospitals that used Epic engaged in significantly more HIE activities than those that did not use Epic ( Exhibit 3 ). However, higher market shares for Epic produced no significant benefit (in terms of more HIE) for hospitals that used Epic or hospitals that used a different vendor.

In markets in which a vendor other than Epic was dominant, hospitals that used the dominant vendor also engaged in significantly more HIE than hospitals that used a different vendor ( Exhibit 3 ). Higher levels of vendor dominance were not associated with higher levels of HIE for hospitals using the dominant vendor. However, in these markets, higher levels of vendor dominance were associated with higher levels of HIE for hospitals that did not use the dominant vendor. When we examined specific vendors instead of all vendors other than Epic, the results held for Cerner and McKesson but not for Meditech—for which none of the relationships was significant ( Exhibit 3 ).

In markets in which the dominant vendor—Epic or another vendor—had a low market share, hospitals using the dominant vendor engaged in more HIE activities than those that used a different vendor ( Exhibit 4 ). However, in markets in which Epic was the dominant vendor and had a high market share, the gap in HIE engagement between hospitals that did and did not use Epic persisted. In contrast, when another vendor was dominant, hospitals that used the dominant vendor engaged in HIE at indistinguishable levels from hospitals that used a different vendor.

SOURCE Authors’ analysis of data for 2012 from the American Hospital Association Annual Survey and for 2012–13 from the survey’s Health Information Technology Supplement. NOTE This figure was created using marginal effects estimates derived from the results presented in the first two columns of Exhibit 3 . The ten HIE activities analyzed are described in the text.

Robustness Tests

When we used an alternative definition of dominance —based on the percentage of hospitals that used the dominant vendor instead of the percentage of hospital beds covered by the dominant vendor—the results were the same as those in our main analysis (Appendix Exhibit 4, column 1). ¹⁸ Our results were also robust when we limited the data to HRRs in which more than 50 percent of hospitals responded to the AHA IT Supplement, although larger standard errors in that case than in our main analysis reduced the significance of vendor dominance (Appendix Exhibit 4, column 2). ¹⁸

We found that higher vendor dominance was associated with both more HIE with outside hospitals and more HIE with outside ambulatory care providers (Appendix Exhibit 5). ¹⁸ This suggests that the effect of vendor dominance on hospital HIE is consistent across provider exchange partners.

Hospital Competitiveness

Previous work ¹⁶ that examined the relationship between hospital competitiveness and HIE led us to include the percentage of the market controlled by each hospital or hospital system and the hospital market Herfindahl-Hirschman Index. We did not find a relationship between hospital market share and level of HIE activity. In our model that included all HRRs, we found a weakly significant negative association between market concentration and HIE activity. This appeared to be driven by markets in which a vendor other than Epic was dominant—markets in which the relationship was negative and significant ( $p=0.05$ ) (Appendix Exhibit 6). ¹⁸

Discussion

In this study we examined the relationship between EHR vendor dominance and hospitals’ engagement in HIE. Our results come as policy makers express concern over the relationship between EHR vendor concentration and vendors’ interest in and ability to facilitate or impede HIE among providers that use their systems and across providers using different vendors’ products. We found that, overall, hospitals that used the dominant vendor in their market engaged in more HIE than hospitals that used a different vendor. This was counterbalanced by our finding that greater vendor dominance was associated with greater hospital engagement in HIE for hospitals that did not use the dominant vendor. However, that positive spillover effect held for only a subset of vendors, and not the largest vendor (Epic). These results suggest that concerns about EHR vendors’ dominance impeding HIE may be merited, and that dominant vendors in competitive markets may be least likely to facilitate HIE with other vendors.

We suspect that the reason why hospitals that use the dominant vendor engage in more HIE than hospitals that do not is because the former institutions likely face fewer technical obstacles in implementing HIE, and a given vendor may see strategic advantages to facilitating HIE among multiple hospitals that all use its systems. In addition, dominant vendors might facilitate the collaboration needed for successful HIE across hospitals by providing a technical strategy to create interoperability between systems and by mediating a collaborative solution between competitive health care providers.

Greater EHR vendor market dominance may encourage greater HIE among hospitals that do not use the dominant vendor because the presence of a dominant vendor may make a hospital’s investment in HIE more attractive. Specifically, higher market share for a single vendor likely reduces the cost for other hospitals to invest in HIE capabilities because the cost and complexity of interfacing with that single vendor’s system are less than those of interfacing with multiple different vendor systems. The fact that when Epic was the dominant vendor we saw a different pattern—in which increasing dominance was not associated with greater HIE among hospitals that did not use Epic—may reflect the fact that Epic clients can readily exchange data with other Epic clients, and as a result, doing so with providers that are not Epic clients may appear prohibitively costly and complex.

Even without systematic empirical evidence, policy makers’ concerns about vendors’ and providers’ HIE business practices were sufficiently salient that members of Congress requested a report from the ONC in late 2014 on information blocking and how it could be addressed. ¹⁰ Our results are consistent with a scenario in which vendors behave differently based on market conditions, with the least HIE occurring in markets in which vendors are the most competitive. It may be that dominant vendors in competitive markets have little interest in facilitating HIE with hospitals that use competitors’ EHR systems because they want to maintain or even increase their market share by encouraging hospitals to switch to their systems.

Before action can be taken, it will be necessary to go beyond our associational analyses. Indeed, other mechanisms may explain our findings. For example, there may be different levels of need for HIE based on variations across markets in patient care patterns, or hospitals may be selecting vendors and levels of HIE based on their own competitive interests. To guide an effective policy response, further research is therefore needed on the causes of lower levels of HIE in certain markets than in others. If the dynamics we postulate are real, policy makers should consider pursuing targeted efforts to combat information-blocking practices and may be well served by creating stronger incentives for providers and vendors to pursue cross-vendor HIE, which has been successful in other nations. ²¹

Conclusion

Using national hospital survey data, we found that hospitals using the dominant vendor engaged in substantially more HIE, compared to hospitals using nondominant vendors. Furthermore, we found that the gap in HIE activity between hospitals using the dominant vendor and those using a nondominant vendor was widest in markets where vendor competition was greatest. Our findings therefore suggest that a targeted policy response, focused on markets in which EHR vendors are competitive, may help facilitate broader-based electronic health information sharing than has been achieved through current HIE activity.

ACKNOWLEDGMENTS

Julia Adler-Milstein is on the scientific advisory board of QPID Health, Inc.

NOTES

• 1 Rudin RS , Motala A , Goldzweig CL , Shekelle PG . Usage and effect of health information exchange: a systematic review . Ann Intern Med . 2014 ; 161 ( 11 ): 803 – -11 . Crossref, Medline, Google Scholar
• 2 Walker J , Pan E , Johnston D , Adler-Milstein J , Bates DW , Middleton B . The value of health care information exchange and interoperability . Health Aff (Millwood) . 2005 ; 24 : w5-10 – 18 . DOI: 10.1377/hlthaff.w5.10 . Go to the article, Google Scholar
• 3 Williams C , Mostashari F , Mertz K , Hogin E , Atwal P . From the Office of the National Coordinator: the strategy for advancing the exchange of health information . Health Aff (Millwood) . 2012 ; 31 ( 3 ): 527 – 36 . Go to the article, Google Scholar
• 4 Rubin RD . The community health information movement: where it’s been, where it’s going . In: O’Carroll PWRipp LHYasnoff WAWard MEMartin EL , editors. Public health informatics and information systems . New York (NY) : Springer-Verlag ; 2003 . p.  595 – 616 . Crossref, Google Scholar
• 5 Grossman JM , Kushner KL , November EA . Creating sustainable local health information exchanges: can barriers to stakeholder participation be overcome? [Internet]. Washington (DC) : Center for Studying Health System Change ; 2008 Feb [cited 2016 May 11 ]. (HSC Research Brief No. 2). Available from: http://www.hschange.org/CONTENT/970/ Google Scholar
• 6 Adler-Milstein J , Bates DW , Jha AK . Operational health information exchanges show substantial growth, but long-term funding remains a concern . Health Aff (Millwood) . 2013 ; 32 ( 8 ): 1486 – 92 . Go to the article, Google Scholar
• 7 Department of Health and Human Services, Office of the National Coordinator for Health Information Technology . Connecting health and care for the nation: a shared nationwide interoperability roadmap: final version 1.0 [Internet]. Washington (DC) : ONC ; [cited 2016 May 11 ]. Available from: https://www.healthit.gov/sites/default/files/hie-interoperability/nationwide-interoperability-roadmap-final-version-1.0.pdf Google Scholar
• 8 Kuperman GJ . Health-information exchange: why are we doing it, and what are we doing? J Am Med Inform Assoc . 2011 ; 18 ( 5 ): 678 – 82 . Crossref, Medline, Google Scholar
• 9 Vest JR , Campion TR , Kaushal R . Challenges, alternatives, and paths to sustainability for health information exchange efforts . J Med Syst . 2013 ; 37 ( 6 ): 9987 . Crossref, Medline, Google Scholar
• 10 Department of Health and Human Services, Office of the National Coordinator for Health Information Technology . Report to Congress: report on health information blocking [Internet]. Washington (DC) : ONC ; 2015 Apr [cited 2016 May 11 ]. Available from: https://www.healthit.gov/sites/default/files/reports/info_blocking_040915.pdf Google Scholar
• 11 Winden TJ , Boland LL , Frey NG , Satterlee PA , Hokanson JS . Care everywhere, a point-to-point HIE tool: utilization and impact on patient care in the ED . Appl Clin Inform . 2014 ; 5 ( 2 ): 388 – 401 . Crossref, Medline, Google Scholar
• 12 Miller AR , Tucker C . Health information exchange, system size, and information silos . J Health Econ . 2014 ; 33 : 28 – 42 . Crossref, Medline, Google Scholar
• 13 Frandsen BR , Joynt KE , Rebitzer JB , Jha AK . Care fragmentation, quality, and costs among chronically ill patients . Am J Manag Care . 2015 ; 21 ( 5 ): 355 – 62 . Medline, Google Scholar
• 14 Tsai TC , Jha AK . Hospital consolidation, competition, and quality: is bigger necessarily better? JAMA . 2014 ; 312 ( 1 ): 29 – 30 . Crossref, Medline, Google Scholar
• 15 Yaraghi N , Du AY , Sharman R , Gopal RD , Ramesh R . Health information exchange as a multisided platform: adoption, usage, and practice involvement in service co-production . Information Systems Research . 2014 ; 26 ( 1 ): 1 – 18 . Crossref, Google Scholar
• 16 Adler-Milstein J , Jha AK . Health information exchange among US hospitals: who’s in, who’s out, and why? Healthc (Amst) . 2014 ; 2 ( 1 ): 26 – 32 . Crossref, Medline, Google Scholar
• 17 Wennberg JECooper MM , editors. The Dartmouth atlas of health care [Internet]. Chicago (IL) : American Hospital Publishing ; 1996 [cited 2016 May 11 ]. Available from: http://www.dartmouthatlas.org/downloads/atlases/96Atlas.pdf Google Scholar
• 18 To access the Appendix, click on the Appendix link in the box to the right of the article online.
• 19 Epic . Organizations on the Care Everywhere Network [Internet]. Verona (WI) : Epic ; c 2016 [cited 2016 May 11 ]. Available from: http://www.epic.com/CareEverywhere/ Google Scholar
• 20 Kruse CS , Regier V , Rheinboldt KT . Barriers over time to full implementation of health information exchange in the United States . JMIR Med Inform . 2014 ; 2 ( 2 ): 26 . Crossref, Google Scholar
• 21 Payne TH , Detmer DE , Wyatt JC , Buchan IE . National-scale clinical information exchange in the United Kingdom: lessons for the United States . J Am Med Inform Assoc . 2011 ; 18 ( 1 ): 91 – 8 . Crossref, Medline, Google Scholar