- Email: [email protected]
Health information exchange among U.S. hospitals: who's in, who's out, and why?
Healthcare 2 (2014) 26–32
Contents lists available at ScienceDirect
Healthcare journal homepage: www.elsevier.com/locate/hjdsi
Health information exchange among U.S. hospitals: who0 s in, who0 s out, and why? Julia Adler-Milstein a,b,n, Ashish K. Jha c,d,e a
School of Information, University of Michigan (JAM), 4376 North Quad, 105 South State Street, Ann Arbor, MI 48109, United States School of Public Health, University of Michigan (JAM), Ann Arbor, MI, United States c Department of Health Policy and Management, Harvard School of Public Health (AKJ), Boston, MA, United States d VA Boston Healthcare System (AKJ), Boston, MA, United States e Division of General Medicine, Brigham and Women's Hospital, Boston, MA, United States b
art ic l e i nf o
a b s t r a c t
Article history: Received 14 August 2013 Received in revised form 15 November 2013 Accepted 13 December 2013 Available online 5 February 2014
Background: A key goal of the 2009 HITECH Act is to ensure broad electronic exchange of clinical data among providers. We sought to assess whether current policy efforts, many of which are being developed by states, appear to be tackling key barriers to hospital participation in health information exchange (HIE). Methods: We used the most recent national data from the American Hospital Association0 s IT Supplement to assess U.S. hospital participation in HIE and how participation varies by state. We then examined whether HIE is being pursued by all types of hospitals, or whether specific types of hospitals are not yet engaged. We focused on for-profit hospitals, those with smaller market share, and those in more competitive markets. Results: We found that 30% of U.S. hospitals engaged in health information exchange with unaffiliated providers. There was large variation in state-level participation, with some states achieving more than 70% participation (Rhode Island, Delaware and Vermont) and others with minimal participation. In markets where exchange occurred, for-profit hospitals were far less likely to engage in HIE than non-profit hospitals (OR¼0.17; po0.001). Hospitals with a larger market share were more likely to engage in exchange (OR¼2.05 for hospitals in the highest relative to the lowest quartile of market share; po0.001), as were hospitals in less competitive markets (OR¼ 2.15 for hospitals in the most relative to least concentrated market quartile; p¼0.04). Conclusions: Despite an uptick in hospital HIE participation since the start of HITECH, the majority of hospitals still do not engage in HIE and there is large state-to-state variation. Specific types of hospitals appear to feel that they are better off not engaging in HIE. Implications: Stronger policies and incentives may be needed to convince organizations to share their data electronically. Pursuing these is critical to ensuring that the highly anticipated quality and efficiency gains from our large national investment in health information technology are realized. & 2014 Elsevier Inc. All rights reserved.
Keywords: Health information technology Health information exchange Hospitals State policy Competition
1. Introduction The 2009 Health Information Technology for Economic and Clinical Health (HITECH) Act put in place a set of policies that sought to achieve widespread use of electronic health records (EHRs) and the ability to exchange data between them. While neither goal is easy, there is growing consensus that enabling data to electronically follow patients between delivery settings requires addressing a formidable set of obstacles.1,2 Part of the challenge is technical: how to interconnect EHRs from hundreds of different
n Corresponding author at: School of Information, University of Michigan (JAM), 4376 North Quad, 105 South State Street, Ann Arbor, MI 48109, United States. E-mail address: [email protected] (J. Adler-Milstein).
2213-0764/$ - see front matter & 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.hjdsi.2013.12.005
vendors. The far more difficult set of challenges, however, relate to the financial, cultural, legal and organizational issues that arise when unaffiliated organizations share sensitive patient data with one another. Despite the growing recognition that health information exchange (HIE) is an essential component of a highperforming healthcare system,3 it is not clear whether our national approach to foster HIE is successfully tackling the key obstacles. The approach to HIE under the HITECH Act chose to heavily rely on states to design approaches to ensure that key clinical data can flow across a broad swath of providers. Through the State HIE Cooperative Agreement Program, states received grants to develop and implement strategies in which every provider would have at least one option to engage in HIE.4 Some states are building on prior HIE efforts and are funding expansions of existing exchange options. Other states started de novo and are working to create
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
options for providers to engage in HIE. Data from just prior to HITECH reveals the enormity of the task: less than 15% of hospitals and 1% of ambulatory practices were participating in efforts to enable HIE between unaffiliated organizations.5 Further, barriers to HIE at the start of HITECH included not only a lack of funding, but also concerns about data privacy and security, legal and regulatory challenges, stakeholder concerns about the competitive implications of sharing data, and technical challenges.6 Recent studies point to an increase in hospital engagement in HIE following the passage of HITECH.7 While encouraging, the evidence does not speak to state-level variation in hospital HIE participation, and in particular, whether some states are further ahead while others are lagging. We also do not yet know whether current approaches to HIE are successfully engaging a broad range of hospitals, or whether specific types of hospitals are choosing to sit on the sidelines and not share their data.8 We therefore used nationally-representative data to assess hospital HIE participation rates by state as well as examine the relationship between key hospital characteristics and participation in HIE. We focused on specific characteristics that, prior to HITECH, were associated with hospital decisions not to engage in clinical data exchange.8 These included for-profit hospitals, hospitals with a small market share, and hospitals in more competitive markets. Taken together, these analyses speak to the early progress of state and federal policies to promote HIE under HITECH.
2. Material and methods We used national data from the Information Technology supplement to the annual American Hospital Association (AHA) survey, which was administered at the end of 2012 to all acutecare hospitals.9 Our analytic sample was limited to the 2849 medical/surgical, non-federal hospitals located in the 50 states
27
and the District of Columbia who responded to the IT supplement survey. We merged the IT supplement data with results from the 2010 AHA survey to develop measures at the market-level as well as capture additional hospital characteristics. The Area Resource File (2010), Medicare Inpatient Claims (2011), and Dartmouth Atlas (2010) were used for additional market-level measures. We defined a market as a hospital referral region (HRR), a designation developed by the Dartmouth Atlas to identify healthcare delivery markets. HRR boundaries are based on Medicare beneficiary travel patterns for tertiary hospital care, and should therefore also capture information exchange needs as patients move between different institutions.
2.1. Measures 2.1.1. HIE participation The AHA IT supplement captured two different dimensions of hospital participation in HIE: (1) whether the hospital participates in a regional HIE effort to share electronic patient-level clinical data and (2) whether the hospital electronically exchanges specific types of patient data (e.g., lab reports and clinical care records) with unaffiliated hospitals or ambulatory providers. Our primary dependent variable considered hospitals to participate in HIE as of late 2012 if they (1) reported participating and actively exchanging data through a regional HIE effort and (2) reported that they exchanged at least one type of patient data (excluding patient demographics) with either unaffiliated hospitals or ambulatory providers. We created alternative measures that looked at each of these dimensions separately in order to confirm the robustness of our results to different approaches to engaging in HIE.7 We then identified markets with HIE activity as those in which at least one hospital in the HRR was classified as participating based on our primary measure of HIE participation.
Fig. 1. Hospital HIE Participation Rates by State (2012). Note: Numbers represent total hospitals in the state in our analytic sample.
28
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
2.2. Hospital characteristics Hospital ownership (for-profit, non-profit private, or public) and hospital bed-share in the market were our two focal characteristics that had been shown to predict HIE participation prior to HITECH.8 We also examined an array of additional characteristics that we thought might be directly related to participating or might confound the relationship between the two focal variables and HIE participation: teaching status, size, proportion of Medicaid admissions, whether the hospital was affiliated with a system, and whether the hospital had significant technological capability (a cardiac intensive care unit) and IT resources (at least a basic EHR). 2.3. Market characteristics At the market level we focused on two dimensions of the competitive environment that could dissuade hospitals from engaging in HIE: market concentration and market fragmentation. The former was measured using the Herfindahl–Hirschman Index (HHI). Fragmentation was measured by calculating the proportion of patients in the HRR with an index hospitalization during the first 6 months of 2011 who had a subsequent readmission within the next 6 months to at least one other, unaffiliated hospital in the same HRR. Additional market-level measures included mean annual Medicare inpatient expenditures, mean proportion of hospitalizations from Medicare patients, urban/rural location, geographic region, population density, percent of population uninsured, and per capita income. These variables was chosen because we believed that each might be directly related to promoting HIE or confound the relationship between the market variables of interest (market concentration and market fragmentation) and HIE. 2.4. Analysis We first calculated a national HIE participation rate among all hospitals in our sample and then calculated participation rates by state. We next examined whether our hypothesized hospital and market characteristics predicted hospital participation in HIE. We started with bivariate relationships and then built multivariable logistic regression models with robust standard errors adjusted for clustering at the market level. Our models include hospital-level sampling weights to adjust for potential nonresponse bias after finding differences between respondents and non-respondents to the AHA IT supplement. (Appendix Table A1) Our bivariate and multivariate analyses compared hospitals that participated in HIE to those that did not within HRRs that had at least one hospital that did participate. Looking within the 201 HRRs (66%) with at least one hospital engaged in HIE helped to ensure that hospitals had an opportunity to participate. Finally, we assessed the robustness of our findings to the two alternative specifications of the dependent variable described above.
3. Results Overall, we found that 30% of hospitals across the country participated in HIE as of late 2012. However, behind this number is substantial variation in hospital HIE participation rates across states (Fig. 1). This ranged from widespread participation in states like Vermont (86%), Delaware (75%) and Rhode Island (71%) to minimal participation in many others. While several of the states with high participation rates were small (i.e., few hospitals), some large states like New York and Indiana had among the highest rates; by the same token, states with very low rates included both
large states like Minnesota as well as some very small states such as Alaska, North Dakota, New Hampshire and Wyoming (Fig. 1). When we examined HIE participation based on key hospital and market characteristics, our bivariate analysis revealed large differences. For-profit hospitals were less likely than their non-profit counterparts to participate (8% versus 37%; p o0.001, Table 1). Hospitals with a more dominant market share were more likely to participate: while only 14% of hospitals in the lowest quartile of market share participated, 49% of hospitals in the highest quartile did so (p-value across the four quartiles o0.001). Multivariate results were similar: for-profit hospitals were far less likely to participate in HIE compared to non-profit hospitals (OR: 0.17, 95% CI 0.10–0.29, Table 2). Hospitals in the highest quartile of market share were more than twice as likely to participate compared to hospitals in the lowest quartile (OR: 2.05, 95% CI 1.15–3.63, Table 2). One additional hospital characteristic was associated with HIE participation: hospitals that were part of a centralized system were more likely to engage in HIE (OR: 1.73, 95% CI 1.33–2.25, Table 2). Hospitals in the most concentrated markets (those in the top quartile of HHI) had 2.15 greater odds of participating (95% CI 1.22–3.80) compared to the least concentrated (and most competitive) markets (p-value across the four quartiles ¼0.04, Table 3). We did not find a significant difference in hospital participation
Table 1 Hospital participation in health information exchange – hospital characteristics. Source: Authors0 analysis. Participates in HIE (2012) N ¼689 (30%) N
%
p-Value
1515 272 544
37 8 21
o 0.001
Market position (percent of bed share in market) First quartile – Lowest % 645 Second quartile 581 Third quartile 573 Fourth quartile – highest % 532
14 24 35 49
o 0.001
Other characteristics examined Hospital size Small Medium Large
1107 943 281
20 35 49
o 0.001
Teaching status Non-teaching Minor teaching Major teaching
1707 430 194
25 40 47
o 0.001
Presence of cardiac ICU (CICU) Yes No
729 1602
40 25
o 0.001
System affiliation Yes No
1154 1177
36 23
o 0.001
584 571 578 598
22 29 33 34
o 0.001
Electronic health record (EHR) system Does not have at least basic EHR 1288 Has at least basic EHR 1043
31 46
0.225
Primary characteristics of interest Ownership Private, non-profit For-profit Public
Percentage of medicaid admissions First quartile – lowest % Second quartile Third quartile Fourth quartile – highest %
Notes: Limited to hospitals in markets with HIE presence.
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
29
Table 2 Adjusted likelihood of participation in health information exchange – hospital characteristics. Source: Authors0 analysis. Participates in HIE (2012)
p-Value
Odds ratio
95% Confidence interval
Reference 0.17 0.72
[0.10,0.29] [0.53,0.98]
o 0.001
Reference 1.16 1.73 2.05
[0.82,1.66] [1.12,2.69] [1.15,3.63]
0.03
Small Medium Large
Reference 1.01 1.09
[0.71,1.45] [0.64,1.88]
0.93
Presence of CICU
Presence of cardiac ICU
1.14
[0.87,1.50]
0.34
System affiliation
System affiliated
1.73
[1.33,2.25]
o 0.001
Percentage of medicaid admissions
Quartile Quartile Quartile Quartile
Reference 0.96 1.07 1.18
[0.70,1.31] [0.79,1.45] [0.82,1.70]
0.61
Primary characteristics of interest Ownership
Market Position
Other characteristics examined Hospital size
Private, non-profit For-profit Public Quartile Quartile Quartile Quartile
1 – lowest bed share 2 3 4 – highest bed share
1 – lowest percent 2 3 4 – highest percent
EHR system
Has at least basic EHR
0.91
[0.70,1.20]
0.52
Teaching status
Non-teaching Minor teaching Major teaching
Reference 1.31 1.28
[0.999,1.73] [0.79,2.08]
0.16
Notes: Limited to hospitals in markets with HIE presence.
based on the degree of market fragmentation. The only other significant market-level characteristic was region: hospitals in the Midwest were less likely to participate compared to other regions (OR ¼0.42, 95% CI 0.25–0.71, Table 3). Our results were largely unchanged when we only required hospitals to report participating in a regional HIE effort (Appendix Table A2) as well as when we only required hospitals to exchange specific types of data with unaffiliated hospitals or ambulatory providers (Appendix Table A3). In both models, the relationships were of similar magnitudes, although some relationships were no longer statistically significant at conventional levels.
4. Discussion More than 2 years into HITECH, one out of every three hospitals is engaged in health information exchange, a notable increase from just a few years prior.7,8 However, HIE participation is not equally spread across states or across types of hospitals. Some states have the vast majority of hospitals engaged while other states have substantial work ahead of them. Similarly, our study suggests that some types of hospitals have bought in and are actively pursuing HIE while others are still not convinced that it is worthwhile. Even with explicit policy efforts to create options for providers to engage in HIE, for-profit hospitals, those with small market share, and those in more competitive markets are continuing to opt out. This suggests that stronger policy efforts may be required to realize the vision of nationwide data exchange. While we can only speculate about the variety of factors that explain which hospitals are participating in HIE and which are not, we suspect that it is a combination of national, state, market, and hospital-level factors. At the national level, meaningful use criteria are expected to increasingly ask more of hospitals with respect to HIE, which has likely pushed some hospitals to begin engaging in
HIE. However, we expect that a more potent force, and one that explains the state-to-state variation in hospital HIE engagement that we observed, is state-level policy efforts. Some states have moved quickly and expanded options for how hospitals can engage in HIE as well as implemented other policies supporting HIE. For example, legislation in Maryland created a statewide exchange and allocated $10 million through Maryland0 s all-payer hospital payment system to establish the technical infrastructure and offset participant costs in the initial years.10 Maryland's State HIE Cooperative Agreement Program funding has been devoted to garnering broad participation of hospitals (and national labs),11 likely explaining their high participation rates. Other states have worked more slowly, with less focus on hospitals. The striking differences in state-level HIE participation rates suggest that many states may have to substantially ramp up their efforts in order to achieve broad hospital participation in the remaining 2 years of the State HIE Cooperative Agreement Program. This may be particularly challenging for large states in light of the fact that those states that are furthest ahead are predominantly small or heavily invested in health information exchange over a long period of time. Understanding the specific barriers in lagging states and identifying successful strategies from other states will likely help policymakers enhance the effectiveness of state strategies to foster broad-based HIE. Beyond national and state-level factors, our results suggest that market and hospital-level characteristics also shape hospital decisions about whether to engage in HIE. The fact that for-profit hospitals and hospitals in more competitive markets as well as those with a smaller market share continue to be less engaged in HIE, which was true prior to HITECH,8 poses a second challenge for policymakers. These organizations appear to be making a strategic decision not to participate, perhaps concluding that the potential loss of patients that may be facilitated by HIE outweigh the potential gains. This is supported by a recent paper that finds that the ease with which patients can leave a hospital0 s system predicts lack of HIE with
30
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
Table 3 Adjusted likelihood of participation in regional health information organization –market characteristics. Source: Authors0 analysis. Participates in HIE (2012)
Primary characteristics of interest – market Market concentration
Odds ratio
95% Confidence interval
p-Value
Quartile Quartile Quartile Quartile
1 – least concentrated 2 3 4 – most concentrated
Reference 1.42 1.70 2.15
[0.96,2.11] [1.12,2.58] [1.22,3.80]
0.042
Quartile Quartile Quartile Quartile
1 – least fragmentation 2 3 4 – most fragmentation
Reference 0.84 0.81 0.87
[0.64,1.09] [0.61,1.07] [0.62,1.22]
0.47
Quartile Quartile Quartile Quartile
1 – lowest spending 2 3 4 – highest spending
Reference 1.16 1.56 1.40
[0.75,1.80] [0.91,2.66] [0.75,2.63]
0.46
Quartile Quartile Quartile Quartile
1 – lowest percent 2 3 4 – highest percent
Reference 1.42 0.97 0.74
[0.92,2.19] [0.60,1.58] [0.42,1.31]
0.19
Quartile Quartile Quartile Quartile
1 – lowest penetration 2 3 4 – highest penetration
0.89 0.83 0.80
[0.60,1.32] [0.51,1.34] [0.47,1.35]
0.83
Northeast Midwest South West
Reference 0.42 0.67 1.01
[0.25,0.71] [0.38,1.19] [0.54,1.88]
0.002
Location type
Non-urban hospital Urban hospital
Reference 1.07
[0.72,1.58]
0.75
Market population density
Quartile Quartile Quartile Quartile
1 – lowest density 2 3 4 – highest density
Reference 0.81 1.32 1.04
[0.52,1.27] [0.75,2.32] [0.56,1.96]
0.34
Quartile Quartile Quartile Quartile
1 – lowest percent uninsured 2 3 4 – highest percent uninsured
Reference 1.20 1.02 0.60
[0.70,2.07] [0.54,1.93] [0.27,1.35]
0.09
Quartile Quartile Quartile Quartile
1 – lowest income 2 3 4 – highest income
1.42 1.10 0.94
[0.85,2.37] [0.64,1.89] [0.50,1.77]
0.35
Market fragmentation
Other characteristics examined Market medicare spending
Market medicare admissions
Market teaching hospital penetration
Region
Market population uninsured
Market per capita income
Notes: Limited to hospitals in markets with HIE presence.
hospitals that are part of other systems.12 Alternatively, it may be the case that hospitals in more competitive markets, those with smaller market share, and those that are for-profit are more acutely sensitive to the cost of engaging in HIE and feel that the potential gains in quality and efficiency from HIE are uncertain and may not outweigh the costs. Regardless of the cause, given the growing evidence of patient and societal benefits from HIE 13–18, it is critical for policymakers to determine how to increase hospital engagement in HIE. While some organizations facing certain market pressures may be better off not engaging in health information exchange, the patients in those communities will likely lose. This may suggest a market failure, indicating that stronger policies that address these underlying concerns and push organizations to share clinical data may be necessary. While future stages of meaningful use that directly require HIE may be enough to begin to shift behavior, hospitals will continue to have latitude to choose with whom they share data. Particularly when coupled with broader pressures to more tightly align specific providers (in particular, Accountable Care
Organizations), these forces could result in islands of HIE with little connectivity between them, causing us to fall short of the vision of true nationwide health information exchange. There are important limitations to our work. First, the AHA IT supplement is self-reported and we were unable to verify accuracy of responses. However, we found approximately the same percentage of hospitals engaging in HIE as has been reported in other data collection efforts.1 Second, while we attempted to include a comprehensive set of variables, there are some that we were unable to measure that could have affected our key results. For example, we were unable to capture the financial position of hospitals. More broadly, there is a multitude of factors that likely affects HIE participation, and we were not able to examine all of them. Finally, we were not able to assess causality and could only assess associations. However, other studies have identified the role of competition in inhibiting sharing of clinical data across hospitals, which reinforces our choice of focal characteristics as well suggests that they reflect real factors that hinder hospital engagement in HIE.19
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
31
a hospital0 s ownership status, its market share, and the competitiveness of its market are associated with its decision to participate in HIE. It is unclear whether current federal and state policies will be sufficiently potent to convince these hospitals to begin sharing their data and to do so in ways that facilitate optimal patient care.
5. Conclusions A key goal of HITECH is to foster broad-based health information exchange and enable clinical data to follow patients between any care delivery setting. Using national data, we report the first set of results examining progress towards this goal along two key dimensions. States were chosen to lead the HITECH effort to create options for providers to engage in HIE. Our data reveals substantial variation in hospital HIE participation across states, and suggests that many states still have the majority of their work ahead of them, particularly those that began their work in response to HITECH. We also found key differences in the types of hospitals that are engaged in HIE, and in particular, that
Acknowledgements This work was supported by funding from the Robert Wood Johnson Foundation.
Table A1 Characteristics of respondent and non-respondent hospitals. Source: Authors0 analysis. All hospitals, N¼ 4565 (%)
Responding hospitals, N¼ 2849 (%)
Non-responding hospitals, N¼ 1716 (%)
p-Value
Small ( o 100 beds) Medium (100–399 beds) Large ( Z400 beds)
50 40 10
48 40 12
54 39 6
o 0.001
Percentage of medicaid admissions
First quartile – lowest % Second quartile Third quartile Fourth quartile – highest %
25 25 25 25
25 25 24 26
25 25 26 24
0.23
Ownership
For-profit Non-profit Public
17 60 23
12 64 24
26 53 21
o 0.001
Market position
Quartile Quartile Quartile Quartile
25 25 25 25
24 24 25 27
27 27 25 22
o 0.001
6 28 50 45
8 31 49 45
3 22 51 unknown
o 0.001 o 0.001 0.15
Hospital characteristic Hospital size
1 – lowest bed share 2 3 4 – highest bed share
Major teaching hospital Presence of CICU System affiliation EHR system
Table A2 Adjusted Likelihood of participation in health information exchange: alternative dependent variable – exchange clinical data with outside hospitals or ambulatory providers only. Source: Authors0 analysis. p-Value
Participates in HIE (2012)
Primary characteristics of interest: hospital Ownership Private, non-profit For-profit Public
Odds ratio
95% Confidence interval
Reference 0.35 0.95
[0.25,0.49] [0.74,1.22]
o 0.001
Quartile Quartile Quartile Quartile
1 – lowest bed share 2 3 4 – Highest bed share
Reference 1.21 1.52 1.92
[0.88,1.65] [1.03,2.26] [1.11,3.31]
0.08
Primary characteristics of interest: market Market concentration Quartile Quartile Quartile Quartile
1 – least concentrated 2 3 4 – most concentrated
Reference 1.43 1.15 1.77
[1.04,1.97] [0.75,1.77] [1.00,3.11]
0.09
1 – least fragmentation 2 3 4 – most fragmentation
Reference 0.79 0.80 0.75
[0.61,1.02] [0.67,1.39] [0.51,1.15]
0.36
Market position
Market fragmentation
Quartile Quartile Quartile Quartile
Notes: Limited to hospitals in markets with HIE presence.
32
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
Table A3 Adjusted likelihood of participation in health information exchange: alternative dependent variable – regional HIE effort participation only. Source: Authors0 analysis. p-Value
Participates in HIE (2012) Odds ratio
95% Confidence interval
Reference 0.21 0.69
[0.13,0.34] [0.50,0.95]
o 0.001
1 – lowest bed share 2 3 4 – highest bed share
Reference 1.13 1.55 1.70
[0.82,1.56] [1.02,2.37] [0.97,3.00]
0.16
1 – least concentrated 2 3 4 – most concentrated
Reference 1.30 1.66 1.80
[0.86,1.96] [1.04,2.64] [1.00,3.24]
0.15
1 – least fragmentation 2 3 4 – most fragmentation
Reference 0.81 0.80 0.95
[0.63,1.05] [0.60,1.07] [0.67,1.34]
0.22
Primary characteristics of interest: hospital Ownership Private, non-profit For-profit Public Market position
Quartile Quartile Quartile Quartile
Primary characteristics of interest: market Market concentration Quartile Quartile Quartile Quartile Market Fragmentation
Quartile Quartile Quartile Quartile
Notes: Limited to hospitals in markets with HIE presence.
Appendix A See Appendix Tables A1–A3. Appendix B. Supplementary materials Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.hjdsi.2013.12.005. References 1. Adler-Milstein J, Bates DW, Jha AK. Operational health information exchanges show substantial growth, but long-term funding remains a concern. Health Aff. 2013;32 (8):1486–1492. 2. Blumenthal D. Implementation of the federal health information technology initiative. N Engl J Med. 2011;365:2426–2431. 3. Buntin MB, Jain SH, Blumenthal D. Health information technology: laying the infrastructure for national health reform. Health Aff. 2010;29:1214–1219. 4. 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. 2012;31:527–536. 5. Adler-Milstein J, Bates DW, Jha AK. A survey of health information exchange organizations in the United States: implications for meaningful use. Ann Int Med. 2011;154:666–671. 6. Adler-Milstein J, Bates DW, Jha AK. U.S. Regional Health Information Organizations: progress and challenges. Health Aff, 28; 2009; 483–492. 7. Furukawa MF, Patel V, Charles D, Swain M, Mostashari F. Hospital electronic health information exchange grew substantially in 2008–12. Health Aff. 2013;32:1346–1354.
8. Adler-Milstein J, Desroches C, Jha A. Health information exchange among US hospitals. Am J Manag Care. 2011;17:761. 9. DesRoches C.M., Charles D., Furukawa M.F., et al. Adoption of electronic health records grows rapidly, but fewer than half Of US hospitals had at least a basic system in 2012. Health Aff 2013;31(5):1478–1485. 10. An Act Concerning Medical Records – Health Information Exchanges. H.B. 784, Pr No. 1lr1901 The General Assembly of Maryland Session of 2011, Regular Session. 11. ONC. Health Information Exchange Strategic and Operational Plan Profile. Maryland. 〈http://statehieresources.org/wp-content/uploads/2011/07/MD_Sta te_HIE_profile.pdf〉. 12. Miller AR, Tucker C. Health information exchange, system size and information silos. J. Health Econ. 2014;33:28–42. 13. Walker J, Pan E, Johnston D, Adler-Milstein J, Bates DW, Middleton B. The value of health care information exchange and interoperability. Health Aff. 2005; w5:10–18. 14. Tripathi M, Delano D, Lund B, Rudolph L. Engaging patients for health information exchange. Health Aff. 2009;28:435–443. 15. Ross S.E., Radcliff T.A., LeBlanc W.G., Dickinson L.M., Libby A.M., Nease D.E. Effects of health information exchange adoption on ambulatory testing rates. J Am Med Inform Assoc 2013:1–6. 16. Bailey J.E., Pope R.A., Elliott E.C., Wan J.Y., Waters T.M., Frisse M.E. Health information exchange reduces repeated diagnostic imaging for back pain. Ann Emerg Med 2013;62(1):16–24. 17. Overhage JM, Dexter PR, Perkins SM, et al. A randomized, controlled trial of clinical information shared from another institution. Ann Emerg Med. 2002;39:14–23. 18. Frisse ME, Johnson KB, Nian H, et al. The financial impact of health information exchange on emergency department care. J Am Med Inform Assoc. 2012;19:328–333. 19. Grossman JM, Kushner KL, November EA. Creating sustainable local health information exchanges: can barriers to stakeholder participation be overcome? Cent Stud Health Syst Change. 2008.
Contents lists available at ScienceDirect
Healthcare journal homepage: www.elsevier.com/locate/hjdsi
Health information exchange among U.S. hospitals: who0 s in, who0 s out, and why? Julia Adler-Milstein a,b,n, Ashish K. Jha c,d,e a
School of Information, University of Michigan (JAM), 4376 North Quad, 105 South State Street, Ann Arbor, MI 48109, United States School of Public Health, University of Michigan (JAM), Ann Arbor, MI, United States c Department of Health Policy and Management, Harvard School of Public Health (AKJ), Boston, MA, United States d VA Boston Healthcare System (AKJ), Boston, MA, United States e Division of General Medicine, Brigham and Women's Hospital, Boston, MA, United States b
art ic l e i nf o
a b s t r a c t
Article history: Received 14 August 2013 Received in revised form 15 November 2013 Accepted 13 December 2013 Available online 5 February 2014
Background: A key goal of the 2009 HITECH Act is to ensure broad electronic exchange of clinical data among providers. We sought to assess whether current policy efforts, many of which are being developed by states, appear to be tackling key barriers to hospital participation in health information exchange (HIE). Methods: We used the most recent national data from the American Hospital Association0 s IT Supplement to assess U.S. hospital participation in HIE and how participation varies by state. We then examined whether HIE is being pursued by all types of hospitals, or whether specific types of hospitals are not yet engaged. We focused on for-profit hospitals, those with smaller market share, and those in more competitive markets. Results: We found that 30% of U.S. hospitals engaged in health information exchange with unaffiliated providers. There was large variation in state-level participation, with some states achieving more than 70% participation (Rhode Island, Delaware and Vermont) and others with minimal participation. In markets where exchange occurred, for-profit hospitals were far less likely to engage in HIE than non-profit hospitals (OR¼0.17; po0.001). Hospitals with a larger market share were more likely to engage in exchange (OR¼2.05 for hospitals in the highest relative to the lowest quartile of market share; po0.001), as were hospitals in less competitive markets (OR¼ 2.15 for hospitals in the most relative to least concentrated market quartile; p¼0.04). Conclusions: Despite an uptick in hospital HIE participation since the start of HITECH, the majority of hospitals still do not engage in HIE and there is large state-to-state variation. Specific types of hospitals appear to feel that they are better off not engaging in HIE. Implications: Stronger policies and incentives may be needed to convince organizations to share their data electronically. Pursuing these is critical to ensuring that the highly anticipated quality and efficiency gains from our large national investment in health information technology are realized. & 2014 Elsevier Inc. All rights reserved.
Keywords: Health information technology Health information exchange Hospitals State policy Competition
1. Introduction The 2009 Health Information Technology for Economic and Clinical Health (HITECH) Act put in place a set of policies that sought to achieve widespread use of electronic health records (EHRs) and the ability to exchange data between them. While neither goal is easy, there is growing consensus that enabling data to electronically follow patients between delivery settings requires addressing a formidable set of obstacles.1,2 Part of the challenge is technical: how to interconnect EHRs from hundreds of different
n Corresponding author at: School of Information, University of Michigan (JAM), 4376 North Quad, 105 South State Street, Ann Arbor, MI 48109, United States. E-mail address: [email protected] (J. Adler-Milstein).
2213-0764/$ - see front matter & 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.hjdsi.2013.12.005
vendors. The far more difficult set of challenges, however, relate to the financial, cultural, legal and organizational issues that arise when unaffiliated organizations share sensitive patient data with one another. Despite the growing recognition that health information exchange (HIE) is an essential component of a highperforming healthcare system,3 it is not clear whether our national approach to foster HIE is successfully tackling the key obstacles. The approach to HIE under the HITECH Act chose to heavily rely on states to design approaches to ensure that key clinical data can flow across a broad swath of providers. Through the State HIE Cooperative Agreement Program, states received grants to develop and implement strategies in which every provider would have at least one option to engage in HIE.4 Some states are building on prior HIE efforts and are funding expansions of existing exchange options. Other states started de novo and are working to create
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
options for providers to engage in HIE. Data from just prior to HITECH reveals the enormity of the task: less than 15% of hospitals and 1% of ambulatory practices were participating in efforts to enable HIE between unaffiliated organizations.5 Further, barriers to HIE at the start of HITECH included not only a lack of funding, but also concerns about data privacy and security, legal and regulatory challenges, stakeholder concerns about the competitive implications of sharing data, and technical challenges.6 Recent studies point to an increase in hospital engagement in HIE following the passage of HITECH.7 While encouraging, the evidence does not speak to state-level variation in hospital HIE participation, and in particular, whether some states are further ahead while others are lagging. We also do not yet know whether current approaches to HIE are successfully engaging a broad range of hospitals, or whether specific types of hospitals are choosing to sit on the sidelines and not share their data.8 We therefore used nationally-representative data to assess hospital HIE participation rates by state as well as examine the relationship between key hospital characteristics and participation in HIE. We focused on specific characteristics that, prior to HITECH, were associated with hospital decisions not to engage in clinical data exchange.8 These included for-profit hospitals, hospitals with a small market share, and hospitals in more competitive markets. Taken together, these analyses speak to the early progress of state and federal policies to promote HIE under HITECH.
2. Material and methods We used national data from the Information Technology supplement to the annual American Hospital Association (AHA) survey, which was administered at the end of 2012 to all acutecare hospitals.9 Our analytic sample was limited to the 2849 medical/surgical, non-federal hospitals located in the 50 states
27
and the District of Columbia who responded to the IT supplement survey. We merged the IT supplement data with results from the 2010 AHA survey to develop measures at the market-level as well as capture additional hospital characteristics. The Area Resource File (2010), Medicare Inpatient Claims (2011), and Dartmouth Atlas (2010) were used for additional market-level measures. We defined a market as a hospital referral region (HRR), a designation developed by the Dartmouth Atlas to identify healthcare delivery markets. HRR boundaries are based on Medicare beneficiary travel patterns for tertiary hospital care, and should therefore also capture information exchange needs as patients move between different institutions.
2.1. Measures 2.1.1. HIE participation The AHA IT supplement captured two different dimensions of hospital participation in HIE: (1) whether the hospital participates in a regional HIE effort to share electronic patient-level clinical data and (2) whether the hospital electronically exchanges specific types of patient data (e.g., lab reports and clinical care records) with unaffiliated hospitals or ambulatory providers. Our primary dependent variable considered hospitals to participate in HIE as of late 2012 if they (1) reported participating and actively exchanging data through a regional HIE effort and (2) reported that they exchanged at least one type of patient data (excluding patient demographics) with either unaffiliated hospitals or ambulatory providers. We created alternative measures that looked at each of these dimensions separately in order to confirm the robustness of our results to different approaches to engaging in HIE.7 We then identified markets with HIE activity as those in which at least one hospital in the HRR was classified as participating based on our primary measure of HIE participation.
Fig. 1. Hospital HIE Participation Rates by State (2012). Note: Numbers represent total hospitals in the state in our analytic sample.
28
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
2.2. Hospital characteristics Hospital ownership (for-profit, non-profit private, or public) and hospital bed-share in the market were our two focal characteristics that had been shown to predict HIE participation prior to HITECH.8 We also examined an array of additional characteristics that we thought might be directly related to participating or might confound the relationship between the two focal variables and HIE participation: teaching status, size, proportion of Medicaid admissions, whether the hospital was affiliated with a system, and whether the hospital had significant technological capability (a cardiac intensive care unit) and IT resources (at least a basic EHR). 2.3. Market characteristics At the market level we focused on two dimensions of the competitive environment that could dissuade hospitals from engaging in HIE: market concentration and market fragmentation. The former was measured using the Herfindahl–Hirschman Index (HHI). Fragmentation was measured by calculating the proportion of patients in the HRR with an index hospitalization during the first 6 months of 2011 who had a subsequent readmission within the next 6 months to at least one other, unaffiliated hospital in the same HRR. Additional market-level measures included mean annual Medicare inpatient expenditures, mean proportion of hospitalizations from Medicare patients, urban/rural location, geographic region, population density, percent of population uninsured, and per capita income. These variables was chosen because we believed that each might be directly related to promoting HIE or confound the relationship between the market variables of interest (market concentration and market fragmentation) and HIE. 2.4. Analysis We first calculated a national HIE participation rate among all hospitals in our sample and then calculated participation rates by state. We next examined whether our hypothesized hospital and market characteristics predicted hospital participation in HIE. We started with bivariate relationships and then built multivariable logistic regression models with robust standard errors adjusted for clustering at the market level. Our models include hospital-level sampling weights to adjust for potential nonresponse bias after finding differences between respondents and non-respondents to the AHA IT supplement. (Appendix Table A1) Our bivariate and multivariate analyses compared hospitals that participated in HIE to those that did not within HRRs that had at least one hospital that did participate. Looking within the 201 HRRs (66%) with at least one hospital engaged in HIE helped to ensure that hospitals had an opportunity to participate. Finally, we assessed the robustness of our findings to the two alternative specifications of the dependent variable described above.
3. Results Overall, we found that 30% of hospitals across the country participated in HIE as of late 2012. However, behind this number is substantial variation in hospital HIE participation rates across states (Fig. 1). This ranged from widespread participation in states like Vermont (86%), Delaware (75%) and Rhode Island (71%) to minimal participation in many others. While several of the states with high participation rates were small (i.e., few hospitals), some large states like New York and Indiana had among the highest rates; by the same token, states with very low rates included both
large states like Minnesota as well as some very small states such as Alaska, North Dakota, New Hampshire and Wyoming (Fig. 1). When we examined HIE participation based on key hospital and market characteristics, our bivariate analysis revealed large differences. For-profit hospitals were less likely than their non-profit counterparts to participate (8% versus 37%; p o0.001, Table 1). Hospitals with a more dominant market share were more likely to participate: while only 14% of hospitals in the lowest quartile of market share participated, 49% of hospitals in the highest quartile did so (p-value across the four quartiles o0.001). Multivariate results were similar: for-profit hospitals were far less likely to participate in HIE compared to non-profit hospitals (OR: 0.17, 95% CI 0.10–0.29, Table 2). Hospitals in the highest quartile of market share were more than twice as likely to participate compared to hospitals in the lowest quartile (OR: 2.05, 95% CI 1.15–3.63, Table 2). One additional hospital characteristic was associated with HIE participation: hospitals that were part of a centralized system were more likely to engage in HIE (OR: 1.73, 95% CI 1.33–2.25, Table 2). Hospitals in the most concentrated markets (those in the top quartile of HHI) had 2.15 greater odds of participating (95% CI 1.22–3.80) compared to the least concentrated (and most competitive) markets (p-value across the four quartiles ¼0.04, Table 3). We did not find a significant difference in hospital participation
Table 1 Hospital participation in health information exchange – hospital characteristics. Source: Authors0 analysis. Participates in HIE (2012) N ¼689 (30%) N
%
p-Value
1515 272 544
37 8 21
o 0.001
Market position (percent of bed share in market) First quartile – Lowest % 645 Second quartile 581 Third quartile 573 Fourth quartile – highest % 532
14 24 35 49
o 0.001
Other characteristics examined Hospital size Small Medium Large
1107 943 281
20 35 49
o 0.001
Teaching status Non-teaching Minor teaching Major teaching
1707 430 194
25 40 47
o 0.001
Presence of cardiac ICU (CICU) Yes No
729 1602
40 25
o 0.001
System affiliation Yes No
1154 1177
36 23
o 0.001
584 571 578 598
22 29 33 34
o 0.001
Electronic health record (EHR) system Does not have at least basic EHR 1288 Has at least basic EHR 1043
31 46
0.225
Primary characteristics of interest Ownership Private, non-profit For-profit Public
Percentage of medicaid admissions First quartile – lowest % Second quartile Third quartile Fourth quartile – highest %
Notes: Limited to hospitals in markets with HIE presence.
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
29
Table 2 Adjusted likelihood of participation in health information exchange – hospital characteristics. Source: Authors0 analysis. Participates in HIE (2012)
p-Value
Odds ratio
95% Confidence interval
Reference 0.17 0.72
[0.10,0.29] [0.53,0.98]
o 0.001
Reference 1.16 1.73 2.05
[0.82,1.66] [1.12,2.69] [1.15,3.63]
0.03
Small Medium Large
Reference 1.01 1.09
[0.71,1.45] [0.64,1.88]
0.93
Presence of CICU
Presence of cardiac ICU
1.14
[0.87,1.50]
0.34
System affiliation
System affiliated
1.73
[1.33,2.25]
o 0.001
Percentage of medicaid admissions
Quartile Quartile Quartile Quartile
Reference 0.96 1.07 1.18
[0.70,1.31] [0.79,1.45] [0.82,1.70]
0.61
Primary characteristics of interest Ownership
Market Position
Other characteristics examined Hospital size
Private, non-profit For-profit Public Quartile Quartile Quartile Quartile
1 – lowest bed share 2 3 4 – highest bed share
1 – lowest percent 2 3 4 – highest percent
EHR system
Has at least basic EHR
0.91
[0.70,1.20]
0.52
Teaching status
Non-teaching Minor teaching Major teaching
Reference 1.31 1.28
[0.999,1.73] [0.79,2.08]
0.16
Notes: Limited to hospitals in markets with HIE presence.
based on the degree of market fragmentation. The only other significant market-level characteristic was region: hospitals in the Midwest were less likely to participate compared to other regions (OR ¼0.42, 95% CI 0.25–0.71, Table 3). Our results were largely unchanged when we only required hospitals to report participating in a regional HIE effort (Appendix Table A2) as well as when we only required hospitals to exchange specific types of data with unaffiliated hospitals or ambulatory providers (Appendix Table A3). In both models, the relationships were of similar magnitudes, although some relationships were no longer statistically significant at conventional levels.
4. Discussion More than 2 years into HITECH, one out of every three hospitals is engaged in health information exchange, a notable increase from just a few years prior.7,8 However, HIE participation is not equally spread across states or across types of hospitals. Some states have the vast majority of hospitals engaged while other states have substantial work ahead of them. Similarly, our study suggests that some types of hospitals have bought in and are actively pursuing HIE while others are still not convinced that it is worthwhile. Even with explicit policy efforts to create options for providers to engage in HIE, for-profit hospitals, those with small market share, and those in more competitive markets are continuing to opt out. This suggests that stronger policy efforts may be required to realize the vision of nationwide data exchange. While we can only speculate about the variety of factors that explain which hospitals are participating in HIE and which are not, we suspect that it is a combination of national, state, market, and hospital-level factors. At the national level, meaningful use criteria are expected to increasingly ask more of hospitals with respect to HIE, which has likely pushed some hospitals to begin engaging in
HIE. However, we expect that a more potent force, and one that explains the state-to-state variation in hospital HIE engagement that we observed, is state-level policy efforts. Some states have moved quickly and expanded options for how hospitals can engage in HIE as well as implemented other policies supporting HIE. For example, legislation in Maryland created a statewide exchange and allocated $10 million through Maryland0 s all-payer hospital payment system to establish the technical infrastructure and offset participant costs in the initial years.10 Maryland's State HIE Cooperative Agreement Program funding has been devoted to garnering broad participation of hospitals (and national labs),11 likely explaining their high participation rates. Other states have worked more slowly, with less focus on hospitals. The striking differences in state-level HIE participation rates suggest that many states may have to substantially ramp up their efforts in order to achieve broad hospital participation in the remaining 2 years of the State HIE Cooperative Agreement Program. This may be particularly challenging for large states in light of the fact that those states that are furthest ahead are predominantly small or heavily invested in health information exchange over a long period of time. Understanding the specific barriers in lagging states and identifying successful strategies from other states will likely help policymakers enhance the effectiveness of state strategies to foster broad-based HIE. Beyond national and state-level factors, our results suggest that market and hospital-level characteristics also shape hospital decisions about whether to engage in HIE. The fact that for-profit hospitals and hospitals in more competitive markets as well as those with a smaller market share continue to be less engaged in HIE, which was true prior to HITECH,8 poses a second challenge for policymakers. These organizations appear to be making a strategic decision not to participate, perhaps concluding that the potential loss of patients that may be facilitated by HIE outweigh the potential gains. This is supported by a recent paper that finds that the ease with which patients can leave a hospital0 s system predicts lack of HIE with
30
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
Table 3 Adjusted likelihood of participation in regional health information organization –market characteristics. Source: Authors0 analysis. Participates in HIE (2012)
Primary characteristics of interest – market Market concentration
Odds ratio
95% Confidence interval
p-Value
Quartile Quartile Quartile Quartile
1 – least concentrated 2 3 4 – most concentrated
Reference 1.42 1.70 2.15
[0.96,2.11] [1.12,2.58] [1.22,3.80]
0.042
Quartile Quartile Quartile Quartile
1 – least fragmentation 2 3 4 – most fragmentation
Reference 0.84 0.81 0.87
[0.64,1.09] [0.61,1.07] [0.62,1.22]
0.47
Quartile Quartile Quartile Quartile
1 – lowest spending 2 3 4 – highest spending
Reference 1.16 1.56 1.40
[0.75,1.80] [0.91,2.66] [0.75,2.63]
0.46
Quartile Quartile Quartile Quartile
1 – lowest percent 2 3 4 – highest percent
Reference 1.42 0.97 0.74
[0.92,2.19] [0.60,1.58] [0.42,1.31]
0.19
Quartile Quartile Quartile Quartile
1 – lowest penetration 2 3 4 – highest penetration
0.89 0.83 0.80
[0.60,1.32] [0.51,1.34] [0.47,1.35]
0.83
Northeast Midwest South West
Reference 0.42 0.67 1.01
[0.25,0.71] [0.38,1.19] [0.54,1.88]
0.002
Location type
Non-urban hospital Urban hospital
Reference 1.07
[0.72,1.58]
0.75
Market population density
Quartile Quartile Quartile Quartile
1 – lowest density 2 3 4 – highest density
Reference 0.81 1.32 1.04
[0.52,1.27] [0.75,2.32] [0.56,1.96]
0.34
Quartile Quartile Quartile Quartile
1 – lowest percent uninsured 2 3 4 – highest percent uninsured
Reference 1.20 1.02 0.60
[0.70,2.07] [0.54,1.93] [0.27,1.35]
0.09
Quartile Quartile Quartile Quartile
1 – lowest income 2 3 4 – highest income
1.42 1.10 0.94
[0.85,2.37] [0.64,1.89] [0.50,1.77]
0.35
Market fragmentation
Other characteristics examined Market medicare spending
Market medicare admissions
Market teaching hospital penetration
Region
Market population uninsured
Market per capita income
Notes: Limited to hospitals in markets with HIE presence.
hospitals that are part of other systems.12 Alternatively, it may be the case that hospitals in more competitive markets, those with smaller market share, and those that are for-profit are more acutely sensitive to the cost of engaging in HIE and feel that the potential gains in quality and efficiency from HIE are uncertain and may not outweigh the costs. Regardless of the cause, given the growing evidence of patient and societal benefits from HIE 13–18, it is critical for policymakers to determine how to increase hospital engagement in HIE. While some organizations facing certain market pressures may be better off not engaging in health information exchange, the patients in those communities will likely lose. This may suggest a market failure, indicating that stronger policies that address these underlying concerns and push organizations to share clinical data may be necessary. While future stages of meaningful use that directly require HIE may be enough to begin to shift behavior, hospitals will continue to have latitude to choose with whom they share data. Particularly when coupled with broader pressures to more tightly align specific providers (in particular, Accountable Care
Organizations), these forces could result in islands of HIE with little connectivity between them, causing us to fall short of the vision of true nationwide health information exchange. There are important limitations to our work. First, the AHA IT supplement is self-reported and we were unable to verify accuracy of responses. However, we found approximately the same percentage of hospitals engaging in HIE as has been reported in other data collection efforts.1 Second, while we attempted to include a comprehensive set of variables, there are some that we were unable to measure that could have affected our key results. For example, we were unable to capture the financial position of hospitals. More broadly, there is a multitude of factors that likely affects HIE participation, and we were not able to examine all of them. Finally, we were not able to assess causality and could only assess associations. However, other studies have identified the role of competition in inhibiting sharing of clinical data across hospitals, which reinforces our choice of focal characteristics as well suggests that they reflect real factors that hinder hospital engagement in HIE.19
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
31
a hospital0 s ownership status, its market share, and the competitiveness of its market are associated with its decision to participate in HIE. It is unclear whether current federal and state policies will be sufficiently potent to convince these hospitals to begin sharing their data and to do so in ways that facilitate optimal patient care.
5. Conclusions A key goal of HITECH is to foster broad-based health information exchange and enable clinical data to follow patients between any care delivery setting. Using national data, we report the first set of results examining progress towards this goal along two key dimensions. States were chosen to lead the HITECH effort to create options for providers to engage in HIE. Our data reveals substantial variation in hospital HIE participation across states, and suggests that many states still have the majority of their work ahead of them, particularly those that began their work in response to HITECH. We also found key differences in the types of hospitals that are engaged in HIE, and in particular, that
Acknowledgements This work was supported by funding from the Robert Wood Johnson Foundation.
Table A1 Characteristics of respondent and non-respondent hospitals. Source: Authors0 analysis. All hospitals, N¼ 4565 (%)
Responding hospitals, N¼ 2849 (%)
Non-responding hospitals, N¼ 1716 (%)
p-Value
Small ( o 100 beds) Medium (100–399 beds) Large ( Z400 beds)
50 40 10
48 40 12
54 39 6
o 0.001
Percentage of medicaid admissions
First quartile – lowest % Second quartile Third quartile Fourth quartile – highest %
25 25 25 25
25 25 24 26
25 25 26 24
0.23
Ownership
For-profit Non-profit Public
17 60 23
12 64 24
26 53 21
o 0.001
Market position
Quartile Quartile Quartile Quartile
25 25 25 25
24 24 25 27
27 27 25 22
o 0.001
6 28 50 45
8 31 49 45
3 22 51 unknown
o 0.001 o 0.001 0.15
Hospital characteristic Hospital size
1 – lowest bed share 2 3 4 – highest bed share
Major teaching hospital Presence of CICU System affiliation EHR system
Table A2 Adjusted Likelihood of participation in health information exchange: alternative dependent variable – exchange clinical data with outside hospitals or ambulatory providers only. Source: Authors0 analysis. p-Value
Participates in HIE (2012)
Primary characteristics of interest: hospital Ownership Private, non-profit For-profit Public
Odds ratio
95% Confidence interval
Reference 0.35 0.95
[0.25,0.49] [0.74,1.22]
o 0.001
Quartile Quartile Quartile Quartile
1 – lowest bed share 2 3 4 – Highest bed share
Reference 1.21 1.52 1.92
[0.88,1.65] [1.03,2.26] [1.11,3.31]
0.08
Primary characteristics of interest: market Market concentration Quartile Quartile Quartile Quartile
1 – least concentrated 2 3 4 – most concentrated
Reference 1.43 1.15 1.77
[1.04,1.97] [0.75,1.77] [1.00,3.11]
0.09
1 – least fragmentation 2 3 4 – most fragmentation
Reference 0.79 0.80 0.75
[0.61,1.02] [0.67,1.39] [0.51,1.15]
0.36
Market position
Market fragmentation
Quartile Quartile Quartile Quartile
Notes: Limited to hospitals in markets with HIE presence.
32
J. Adler-Milstein, A.K. Jha / Healthcare 2 (2014) 26–32
Table A3 Adjusted likelihood of participation in health information exchange: alternative dependent variable – regional HIE effort participation only. Source: Authors0 analysis. p-Value
Participates in HIE (2012) Odds ratio
95% Confidence interval
Reference 0.21 0.69
[0.13,0.34] [0.50,0.95]
o 0.001
1 – lowest bed share 2 3 4 – highest bed share
Reference 1.13 1.55 1.70
[0.82,1.56] [1.02,2.37] [0.97,3.00]
0.16
1 – least concentrated 2 3 4 – most concentrated
Reference 1.30 1.66 1.80
[0.86,1.96] [1.04,2.64] [1.00,3.24]
0.15
1 – least fragmentation 2 3 4 – most fragmentation
Reference 0.81 0.80 0.95
[0.63,1.05] [0.60,1.07] [0.67,1.34]
0.22
Primary characteristics of interest: hospital Ownership Private, non-profit For-profit Public Market position
Quartile Quartile Quartile Quartile
Primary characteristics of interest: market Market concentration Quartile Quartile Quartile Quartile Market Fragmentation
Quartile Quartile Quartile Quartile
Notes: Limited to hospitals in markets with HIE presence.
Appendix A See Appendix Tables A1–A3. Appendix B. Supplementary materials Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.hjdsi.2013.12.005. References 1. Adler-Milstein J, Bates DW, Jha AK. Operational health information exchanges show substantial growth, but long-term funding remains a concern. Health Aff. 2013;32 (8):1486–1492. 2. Blumenthal D. Implementation of the federal health information technology initiative. N Engl J Med. 2011;365:2426–2431. 3. Buntin MB, Jain SH, Blumenthal D. Health information technology: laying the infrastructure for national health reform. Health Aff. 2010;29:1214–1219. 4. 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. 2012;31:527–536. 5. Adler-Milstein J, Bates DW, Jha AK. A survey of health information exchange organizations in the United States: implications for meaningful use. Ann Int Med. 2011;154:666–671. 6. Adler-Milstein J, Bates DW, Jha AK. U.S. Regional Health Information Organizations: progress and challenges. Health Aff, 28; 2009; 483–492. 7. Furukawa MF, Patel V, Charles D, Swain M, Mostashari F. Hospital electronic health information exchange grew substantially in 2008–12. Health Aff. 2013;32:1346–1354.
8. Adler-Milstein J, Desroches C, Jha A. Health information exchange among US hospitals. Am J Manag Care. 2011;17:761. 9. DesRoches C.M., Charles D., Furukawa M.F., et al. Adoption of electronic health records grows rapidly, but fewer than half Of US hospitals had at least a basic system in 2012. Health Aff 2013;31(5):1478–1485. 10. An Act Concerning Medical Records – Health Information Exchanges. H.B. 784, Pr No. 1lr1901 The General Assembly of Maryland Session of 2011, Regular Session. 11. ONC. Health Information Exchange Strategic and Operational Plan Profile. Maryland. 〈http://statehieresources.org/wp-content/uploads/2011/07/MD_Sta te_HIE_profile.pdf〉. 12. Miller AR, Tucker C. Health information exchange, system size and information silos. J. Health Econ. 2014;33:28–42. 13. Walker J, Pan E, Johnston D, Adler-Milstein J, Bates DW, Middleton B. The value of health care information exchange and interoperability. Health Aff. 2005; w5:10–18. 14. Tripathi M, Delano D, Lund B, Rudolph L. Engaging patients for health information exchange. Health Aff. 2009;28:435–443. 15. Ross S.E., Radcliff T.A., LeBlanc W.G., Dickinson L.M., Libby A.M., Nease D.E. Effects of health information exchange adoption on ambulatory testing rates. J Am Med Inform Assoc 2013:1–6. 16. Bailey J.E., Pope R.A., Elliott E.C., Wan J.Y., Waters T.M., Frisse M.E. Health information exchange reduces repeated diagnostic imaging for back pain. Ann Emerg Med 2013;62(1):16–24. 17. Overhage JM, Dexter PR, Perkins SM, et al. A randomized, controlled trial of clinical information shared from another institution. Ann Emerg Med. 2002;39:14–23. 18. Frisse ME, Johnson KB, Nian H, et al. The financial impact of health information exchange on emergency department care. J Am Med Inform Assoc. 2012;19:328–333. 19. Grossman JM, Kushner KL, November EA. Creating sustainable local health information exchanges: can barriers to stakeholder participation be overcome? Cent Stud Health Syst Change. 2008.