Hospitals that participate in accountable care organizations (ACOs) expand their health information exchange networks as a result, but hospitals in markets with existing ACO infrastructure can expand more quickly.
Objectives: First, to assess whether hospitals expand the network breadth of their health information exchange (HIE) partners after joining an accountable care organization (ACO). Second, to analyze whether this HIE network expansion effect varies across markets with differing levels of ACO penetration.
Study Design: Difference-in-differences analyses of US nonfederal acute care hospitals, 2014-2017.
Methods: We used data from the American Hospital Association Annual Survey and Information Technology Supplement to measure hospital ACO participation, HIE network breadth (defined as number of different partner types), and ACO market penetration at the hospital referral region level. We implemented a difference-in-differences model to estimate changes in hospitals’ HIE network breadth with ACO participation in different years. We estimate these effects combined across all markets and stratified by markets with high and low ACO market penetration.
Results: In combined analyses, HIE breadth increased by 0.35 partner types with ACO participation, a 30.7% increase (P < .001). In stratified analyses, this effect was larger for hospitals in high–ACO penetration markets (0.41 partner types, a 32.0% increase; P < .001) and smaller for hospitals in low–ACO penetration markets (0.25 partner types, a 24.8% increase; P < .05). We found dynamic effects of ACO adoption illustrating an immediate effect in high–ACO penetration markets and a 2-year delayed effect in low–ACO penetration markets.
Conclusions: Hospitals that joined ACOs increased their HIE breadth, but this effect was heterogenous across markets and across time. Our findings illustrate a “network effect,” with large, immediate effects in HIE breadth following ACO participation in high–ACO penetration markets and smaller, delayed effects in low–ACO penetration markets.
Am J Manag Care. 2022;28(1):e7-e13. https://doi.org/10.37765/ajmc.2022.88815
Accountable care organization (ACO) success depends in part on a broad network of health information exchange (HIE) partners to effectively coordinate care. Our study illustrates that although ACO participation broadens HIE networks, the effects accrue differently across markets with varying levels of ACO penetration.
Accountable care organizations (ACOs) are a popular form of alternative payment model (APM), and the number of individuals covered by ACOs continues to rise. Now covering more than 45 million US lives,1 ACOs incentivize efficient and coordinated care by holding participating provider organizations accountable for the total cost of care for covered individuals.2 This model aims to reduce costs and improve quality, and ACOs have thus far had modest success in these domains.3-8
To achieve these goals, ACO participants must coordinate care for covered individuals across care settings including inpatient care, long-term and postacute care, outpatient providers, and others.2 This coordination can be facilitated by effective use of health information technology (IT)—in particular, health information exchange (HIE)—across providers who share ACO-attributed patients.9,10 The extent to which a hospital’s HIE network includes all partner organizations with which the hospital shares patients is referred to as HIE breadth.11 Encouraging HIE participation to increase HIE breadth and support a nationally interoperable health care delivery system remains an important policy goal in the United States.12-15 However, early evidence on the impact of ACOs in driving HIE across health care providers is mixed. One study illustrated broader HIE networks among APM provider participants.11 Others showed that ACO participation encouraged intraorganizational HIE and strengthened existing patient-sharing relationships16,17 but did not increase interorganizational HIE.17 Furthermore, interorganizational and intraorganizational HIE are known to be inversely related,18 raising questions with respect to whether ACOs simply provide financial buttressing for existing partnerships or actively expand care coordination networks. Although value-based payment models like ACOs could increase HIE adoption and use,19,20 their effect could be limited if ACO membership does not also increase the types of organizational partners with which participants routinely exchange patient data to include more of the full continuum of care.
Among hospitals, ACO participation is associated with greater adoption of health IT capacity,21-26 and recent work has shown that hospitals participating in APMs have broader networks of exchange partners but lower overall volume of exchange.11 However, there also are concerns that ACO participation may occur disproportionately among technologically advanced hospitals and may provide financial incentives for HIE only among existing or intrasystem exchange partners.16,18 Moreover, ACO success is more common in markets with a critical mass of ACO participants.11,27-30 This may be because ACO-participating hospitals in markets with low ACO participation have fewer nonhospital ACO-participating partner organizations with which to share data and therefore have a lower incentive to invest in data exchange. Conversely, in high–ACO penetration markets, ACO-participating hospitals may have a bevy of technologically equipped and financially aligned partner organizations with which to establish new HIE relationships. Ultimately, these market dynamics may limit the potential of ACOs as a driver of nationwide HIE adoption, as demand might only increase where supply already exists.
Understanding whether and to what extent hospital ACO participation drives HIE breadth—or broader HIE networks with increased coverage of the care continuum—is critical to understanding whether APMs help improve nationwide health IT infrastructure and breadth. Although ACOs that include nonhospital provider organizations such as long-term care (LTC) and behavioral health providers are associated with improved care quality31,32 and savings,33,34 incentives for hospitals to establish exchange with these nonhospital providers were not included in the initial Meaningful Use electronic health record (EHR) program. Implementing net new HIE capabilities with these nonhospital providers might therefore particularly benefit from hospital ACO participation as a means to improve care coordination, even if business relationships already exist. Furthermore, if the impact of ACOs on HIE network breadth varies across geographies due to variation in ACO market penetration,1 we can gain insight into potential mechanisms (ie, HIE-enabled care coordination with nonacute providers) by which higher ACO penetration supports ACO success and pinpoint areas for policy intervention in low–ACO penetration markets.
Despite the importance of understanding these dynamics, we know of only 1 descriptive study examining the more general association between APM participation and HIE breadth,11 but no studies that examine whether this relationship was causal. In this study, we focus on 2 research questions. First, do hospitals that join ACOs subsequently increase the breadth of their HIE networks? Second, do these effects vary across markets with differing levels of ACO penetration? We hypothesize that in aggregate, joining an ACO will lead to an expansion of hospital partner networks to include new organizational types, consistent with ACO incentives to ensure coordinated care across the continuum of care. We hypothesize that this increase will be more pronounced in markets with high ACO penetration compared with low-penetration markets, because high–ACO penetration markets are likely to have more provider organizations (of all types) with the capabilities and incentives to build HIE relationships with new ACO members.
Setting and Data Sources
We used the American Hospital Association (AHA) Annual Survey and IT Supplement for 2014-2017 to measure our outcome of HIE breadth and primary independent variable of hospital-reported ACO participation. The AHA survey is sent annually to the chief executive of each US hospital, who either completes the survey or assigns an informed delegate to ensure accuracy of responses35; the mean response rate for the IT Supplement during our study period was 59.5% (range, 56% in 2015 to 64% in 2017). We limited our sample to nonfederal acute care hospitals to construct a sample of hospitals subject to analogous incentives and market forces and comparable with previous literature using AHA survey data.11,13,36,37 We applied nonresponse weights to all analyses to adjust our estimates to be nationally representative and removed 338 hospitals from the sample that reported ACO participation in one year and no participation in subsequent years, as this behavior violates the “irreversibility of treatment” assumption required for our analysis strategy.
Measures: HIE Breadth
Our primary outcome variable was HIE network breadth, measured as the number of different types of exchange partners that a hospital reported; this was drawn from previous work exploring hospital engagement in HIE.11 In the 2014-2017 AHA IT surveys, hospitals were asked to report whether they routinely sent structured electronic information during patient transfers to (1) hospitals outside their system, (2) outside ambulatory care providers, (3) long-term and postacute care providers, and/or (4) behavioral health providers. From these 4 response options we constructed our primary outcome at the hospital-year level: a count measure from 0 to 4 measuring the breadth of a hospital’s HIE network as the number of different partner types with which a hospital regularly exchanged data. We also constructed binary outcomes at the hospital-year level for each of the 4 component partner types for use in a decomposed analysis of partner-type specific effects. Our primary outcome measure is consistent with the definition by Lin et al of HIE breadth and draws on information systems measurement theory of electronic data interchange capturing different dimensions of interorganizational information exchange.11,38
We would expect HIE breadth to increase as hospitals join ACOs, because as a hospital joins or establishes an ACO, a new set of diverse partner organizations with mutual incentives to strengthen internal ties becomes relevant to the hospital.16 Our measure captures the degree to which hospitals broaden their HIE networks to new organizational types via net new HIE capabilities, as ACOs are specifically intended to bring together health care provider organizations to cover the continuum of care.
Measures: ACO Participation and ACO Market Penetration
Hospital ACO participation was determined based on hospital response to the AHA Annual Survey item “Has your hospital or health system established an accountable care organization?” to which hospitals responded yes or no. To distinguish between hospital markets with high ACO penetration and those with low ACO penetration, we calculated regional measures of the proportion of discharges attributable to ACO-participating hospitals. Our market definition was the hospital referral region (HRR), as this geography has direct implications for care coordination and patient referrals. We denoted an HRR as “high ACO penetration” if at least 30% of Medicare discharges came from ACO-participating hospitals in at least 3 of our 4 study years.39 ACO market penetration was fixed throughout the study period.
We included hospital and market characteristics from the AHA survey that have been demonstrated previously as factors associated with hospital HIE engagement and ACO participation.11 We included 2 controls for hospital IT adoption: EHR adoption level (less than basic EHR, basic EHR, and comprehensive EHR) and primary inpatient EHR vendor. The 4 most common vendors, Epic, Cerner, McKesson, and Meditech, captured 75.4% of hospitals in our sample. Those vendors were preserved; all others were categorized as “other vendors.” We also measured the following hospital factors: participation in a regional health information organization, participation in a bundled payments program, establishment of a medical home program, teaching status, critical access hospital status, system membership, ownership (for-profit, nonprofit, or local government ownership), number of beds (< 100, 100-400, or > 400 beds), and hospital market share in terms of beds in the HRR. Market-level variables included urban or rural status and HRR market concentration measured as a Herfindahl-Hirschman Index using number of beds as the measure of market share.
We computed descriptive statistics of hospitals, overall and stratified by ACO market penetration and never-ACO hospitals vs eventual ACO participants. We also calculated average HIE breadth measures relative to ACO participation for hospitals that joined ACOs (the year prior to ACO participation, the year in which the hospital initiated ACO participation, and the year following ACO initiation). Given recent advances in estimating difference-in-differences with variations in intervention timing (as is the case in our study), standard 2-way fixed effects methods may result in biased estimates.40 Our primary analysis estimates a group-time average treatment effect on the treated, abbreviated ATT(g,t), in aggregate and for groups of hospitals that first reported ACO participation in 2015, 2016, and 2017. This estimator from Callaway and Sant’Anna relaxes the traditional unconditional parallel pretrends assumption41 to an assumption of parallel pretrends across treated units and never-treated units conditional on covariates, justified by the observed differences in the characteristics of hospitals that never participate in ACOs compared with those that do.42 This approach applies weighting for hospitals that initiate ACO participation at different times, directly measures dynamic treatment effects (similar to event study regression), and adjusts preintervention estimates for selective treatment timing, a phenomenon that may occur with later–ACO adopting hospitals, especially in high–ACO penetration markets.40 Our results are therefore robust to potential bias from negative weighting issues identified by Goodman-Bacon in 2-way fixed effects with variation in treatment timing.43 We then report an aggregated causal estimate of ACO participation derived from a balanced-panel group-level ATT(g,t). This estimator is preferred in this context to the recently developed deChaisemartin and D’Haultfœuille approach,44 which identifies instantaneous treatment effects in designs without relaxing the unconditional pretrends assumption.
We ran 3 models to address our 2 research questions. For our first question examining the overall effect of ACO participation, we ran a combined model including all ACO markets to estimate the group-average ATT(g,t) of ACO participation. For our second research question analyzing differences in the effect across markets with differing levels of ACO penetration, we stratified the sample and ran separate models for hospitals in markets with high and low ACO penetration. This stratified sample approach both directly addresses our research questions regarding market effects and helps mitigate concerns of selection into treatment driven by ACO market penetration. We performed these 3 models for our primary HIE breadth outcome measure as well as all 4 component measures of our primary outcome. We focus on the results from our primary analysis; component analyses are presented in the eAppendix (available at ajmc.com). We also varied the cutoff value for defining markets with high and low ACO penetration, using 20% and 40% as robustness tests for our chosen value of 30% of Medicare discharges; these results are also presented in the eAppendix. All models controlled for baseline values of the hospital and market characteristics outlined earlier, and the combined model included an indicator variable for our measure of high–ACO penetration HRRs. Standard errors were clustered at the hospital level to account for autocorrelation of errors within hospitals over time. Hospitals reporting ACO membership in 2014 were dropped from these analyses, as it is impossible to compute a treatment effect and they cannot serve as controls because they are “always treated.” Finally, for comparison with more common (but likely biased) 2-way fixed effects approaches, we report results in the eAppendix from stratified models using 2-way fixed effects for comparison with our group-aggregated ATT estimates.
Our final sample included 4186 hospitals over 4 survey-years, totaling 16,336 hospital-years. The mean (SD) HIE network breadth was 1.14 (1.55) partner types. Mean HIE network breadth increased from 0.96 partner types in 2014 to 1.20 in 2015 and remained stable through 2017. In HRRs with high ACO penetration, the mean (SD) HIE breadth was 1.28 (1.59) partner types compared with 1.01 (1.48) in low–ACO penetration markets (eAppendix Table 1). The number of hospitals reporting ACO participation doubled over the study period, from 682 (16.6%) in 2014 to 1399 (34.4%) in 2017. Our measure of ACO market penetration was fixed, but as ACO participation grew, the share of hospitals in high-penetration and low-penetration markets shifted (Figure 1). The proportion of ACO hospitals in high-penetration markets grew from 14.5% in 2014 to 24.6% of hospitals in 2017. Bivariate differences in hospital characteristics across markets with high and low ACO penetration are presented in eAppendix Table 1; differences across ACO participants and hospitals that never participated in ACOs are in eAppendix Table 2.
ACO Participation and HIE Breadth
ACO participants across markets differed in HIE breadth trends. ACO hospitals in high–ACO penetration markets saw increases in average HIE breadth from 1.21 partner types to 1.98 over the years prior to and following ACO participation, whereas ACO hospitals in low-penetration markets saw no change in HIE breadth over time (Figure 2).
In our difference-in-differences models, ACO participation led to a significant increase in HIE breadth; however, these effects were heterogenous across markets and across time. Across all markets, there was an increase in partner types with ACO participation (ATTcombined, 0.35 partner types, a 30.7% increase; 95% CI, 0.22-0.48; P < .001) (Table). ACO participation in high–ACO penetration markets led to an increase of 0.41 partner types, a 32.0% increase (95% CI, 0.23-0.59; P < .001). In low-penetration markets, however, ACO participation had a smaller effect on the number of partner types (ATTlow ACO, 0.25, a 24.8% increase; 95% CI, 0.05-0.44; P < .05). We observed similar patterns of effects in all 4 component outcomes (eAppendix Figure).
Treatment effects averaged at the ACO participation cohort level on HIE breadth demonstrated heterogenous effects by cohort group, the estimation of which is a key strength of the difference-in-differences estimation approach we use. In all models, the cohort of hospitals initiating ACO participation in 2015 illustrated strong average effects of ACO participation on the number of partner types (Table). In contrast, the 2016 cohort ACO participation had no effect on the number of partner types in any of the 3 models. Finally, ACO participation increased the number of partner types for the 2017 cohort of hospitals, but only in high–ACO penetration markets (ATThigh,2017, 0.54; 95% CI, 0.05-1.04; P < .05).
Because of these heterogenous cohort effects, we also examined dynamic effects of ACO participation across different time frames of exposure. This approach averages the treatment effect by time relative to the first year that a hospital reported ACO participation, similar to event study regression. For example, hospital A that began ACO participation in 2015 and hospital B that began in 2016 would have 2016 and 2017, respectively, denoted as t = 1, and those treatment effects would be averaged together to calculate an average treatment effect across all ACO hospitals at 1 year after ACO initiation. Figure 3 presents the dynamic effects of ACO participation on HIE breadth across our 3 models. Consistent with Figure 2, we see a large immediate increase in HIE breadth in the year in which a hospital first reports ACO participation among hospitals in high–ACO penetration markets (ATThigh,t = 0, 0.42; 95% CI, 0.16-0.68; P < .001). We also find that the only statistically significant treatment effect in low–ACO penetration markets occurs at 2 years after the hospital first participates in an ACO (ATTlow,t = 2, 0.95; 95% CI, 0.16-1.74; P < .01). Full dynamic effects results are presented in eAppendix Table 3 and estimates from our robustness tests varying the ACO market penetration cutoff percentages are presented in eAppendix Table 4. We find qualitatively similar aggregate effects when varying this cutoff value. Finally, stratified 2-way fixed effects estimates are presented for comparison with our group-aggregate ATT estimates in eAppendix Table 5.
Our results illustrate that ACO participation expands HIE breadth by sharing data electronically across more partner types, with heterogenous effects across markets with differing levels of ACO penetration. The returns to HIE from ACO participation are larger and immediate for hospitals in high–ACO penetration markets and smaller on average and delayed for hospitals in low–ACO penetration markets. We hypothesized that hospitals in low–ACO penetration markets may have fewer ready partners for exchange upon ACO participation and thus see a limit on the efficacy of ACO participation with respect to expanding HIE networks. Our findings of a smaller, delayed effect size in low–ACO penetration markets paired with null aggregate effects in 3 of the 4 component measures provide moderate support for this proposed mechanism of market “readiness” for HIE expansion.
The financial incentives of ACOs should encourage hospitals to share data electronically with the care delivery organizations that they share patients with. Broad data sharing in the ACO context is especially relevant for LTC and behavioral health providers, which were ineligible to receive Health Information Technology for Economic and Clinical Health Act funding and are less likely to have IT infrastructure in place45,46 but often share patients with acute care hospitals. Despite the potential benefit to patients, these organizational linkages are rare.47,48 Encouragingly, our component analyses illustrate positive effects among behavioral health and LTC providers; in fact, net new HIE with LTC providers demonstrated the largest effect size in high–ACO penetration markets. This suggests that our aggregate effects are at least in part driven by new HIE capabilities with provider organizations that were not included in the original Meaningful Use incentive program. Although our measure cannot capture the total number of unique HIE partners within a given partner type, our measure allows us to observe net new connections such as these, which appear to be driving expanded HIE breadth among ACO-participating hospitals.
Beyond the returns to individual hospital ACO participants and the patients they care for, our findings also have implications for national efforts to expand HIE networks, which lean on APMs such as ACOs to drive some part of that expansion.49 Even with ACO incentives in place and benefits of expanded partnerships, hospitals’ ability to expand their HIE network breadth is dependent at least in part upon the existing ACO market, and these returns are likely to be delayed for hospitals in low-penetration markets. Our findings suggest that even with ACO adoption, HIE breadth may lag in markets without a critical mass of ACO participants. This in turn limits the role that ACO participation can play in building truly national HIE networks. This may be particularly true in low–ACO penetration markets, which tend to be more rural and less well resourced in general.
Taken together, our findings suggest that broader adoption of APMs may not be enough to stimulate rapid and balanced progress in nationwide interoperability, especially if those payment reform models remain unevenly distributed across geographies. One possible explanation is that ACO incentives in isolation may simply be too weak to justify the expense of building connectivity across a range of partners unless they are also organizationally committed to reducing costs through ACO membership themselves. As a result, policy makers may wish to focus on more direct incentives to stimulate engagement in electronic data sharing and more direct regulation and support of data sharing, such as that found in the rulemaking following the 21st Century Cures Act.12
Our results also point to increased HIE breadth as a possible reason for success of ACOs in high–market penetration areas.29 Increasing HIE breadth allows ACO participants to incorporate the use of these networks into their care coordination workflows more quickly, which may in turn facilitate more rapidly realized savings. In contrast, hospitals in low–ACO penetration markets may not see these returns to interoperability for years following their ACO participation. For hospitals that join ACOs, this may hinder their ability to hit program targets in a timely manner; others may opt simply to not join ACOs due to the lack of network benefits. As such, it may be important to consider the local context when designing ACO performance goals, given that some gains may be immediate in ACO-dense markets and delayed in others. Future research into ACO formation, adoption, and evaluation should include explicit treatment of the local ACO market, as aggregate market measures mask important differences across markets with varying levels of ACO participation.
Our results should be interpreted with several limitations in mind. First, we rely on self-reported data for both our outcome variable of HIE breadth and our treatment variable of ACO participation. Although the AHA survey and IT Supplement are widely used and have been validated against other sources, we are unable to independently verify the veracity of hospital responses to the survey. Second, all estimates from observational data necessarily rely on 1 or more assumptions for valid causal interpretation. In our case, causal interpretation of our models relies on an assumption of parallel pretrends. Typically, difference-in-differences designs assume unconditional parallel trends; our empirical tests show no evidence of violation of this assumption in our model examining high-penetration markets, but there is some evidence that this assumption is violated in our combined and low–ACO penetration models. However, as Figure 3 makes clear, we find no evidence of violation of conditional parallel pretrends, which is a more appropriate assumption in our case due to observed differences across treated and control hospitals (eAppendix Table 2) and the likelihood of selective treatment effects.42 As a result, we interpret our ATT estimates as causal effects. Third, our measure of ACO participation does not distinguish among different types of Medicare ACOs or between Medicare and private payer ACOs, which may present hospitals with different incentives for data sharing. We are also unable to determine whether hospitals in the same market joined the same or different ACOs. Rather, our definition broadly captures all ACO participation rather than limiting participation to a particular program such as the Medicare Shared Savings Program. Finally, although we operationalize HIE breadth as the number of sharing partner types and thus can identify net new HIE relationships, a single number is unable to capture the full complexity of HIE, such as differences in volume of exchange, what data elements are exchanged, whether exchanged data are integrated and used in clinical decision-making, and other important factors. Unfortunately, due to changes in response options and item availability over time in the AHA IT Supplement, we were unable to operationalize other dimensions of HIE such as volume, depth, and diversity.
Although hospitals broaden their HIE networks after joining an ACO, this effect is heterogenous across market types. Markets with high ACO penetration see an immediate and pronounced effect, whereas markets with low ACO penetration see only a delayed and smaller effect. Our results show that although ACOs are encouraging hospitals to broaden their data-sharing networks, they do so unevenly, suggesting there is a “network effect” where greater ACO presence in a local market creates stronger incentives for investing in care coordination infrastructure such as HIE. This may limit the impact of ACOs as a mechanism for achieving nationwide HIE in areas with low ACO participation.
Author Affiliations: Perelman School of Medicine, University of Pennsylvania (NCA, RMW), Philadelphia, PA; Leonard Davis Institute of Health Economics, University of Pennsylvania (NCA, RMW), Philadelphia, PA; Regenstrief Institute (NCA), Indianapolis, IN; University of California, San Francisco School of Medicine (AJH), San Francisco, CA.
Source of Funding: Dr Apathy is supported by a training grant from the Agency for Healthcare Research and Quality (T32-HS026116-04). Dr Werner is supported by K24-AG047908 from the National Institute on Aging.
Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (NCA, AJH, RMW); acquisition of data (NCA); analysis and interpretation of data (NCA, AJH); drafting of the manuscript (NCA, AJH, RMW); critical revision of the manuscript for important intellectual content (NCA, AJH, RMW); statistical analysis (NCA); provision of patients or study materials (NCA); obtaining funding (RMW); administrative, technical, or logistic support (NCA, AJH, RMW); and supervision (RMW).
Address Correspondence to: Nate C. Apathy, PhD, University of Pennsylvania, 3641 Locust Walk, Philadelphia, PA 19104. Email: email@example.com.
1. Muhlestein D, Bleser WK, Saunders RS, Richards R, Singletary E, McClellan MB. Spread of ACOs and value-based payment models in 2019: gauging the impact of Pathways to Success. Health Affairs. October 21, 2019. Accessed June 2, 2020. https://www.healthaffairs.org/do/10.1377/hblog20191020.962600/full/
2. Fisher ES, Shortell SM. Accountable care organizations: accountable for what, to whom, and how. JAMA. 2010;304(15):1715-1716. doi:10.1001/jama.2010.1513
3. McWilliams JM, Landon BE, Chernew ME. Changes in health care spending and quality for Medicare beneficiaries associated with a commercial ACO contract. JAMA. 2013;310(8):829-836. doi:10.1001/jama.2013.276302
4. Nyweide DJ, Lee W, Cuerdon TT, et al. Association of Pioneer Accountable Care Organizations vs traditional Medicare fee for service with spending, utilization, and patient experience. JAMA. 2015;313(21):2152-2161. doi:10.1001/jama.2015.4930
5. McWilliams JM, Hatfield LA, Chernew ME, Landon BE, Schwartz AL. Early performance of accountable care organizations in Medicare. N Engl J Med. 2016;374(24):2357-2366. doi:10.1056/NEJMsa1600142
6. McWilliams JM, Hatfield LA, Landon BE, Hamed P, Chernew ME. Medicare spending after 3 years of the Medicare Shared Savings Program. N Engl J Med. 2018;379(12):1139-1149. doi:10.1056/NEJMsa1803388
7. Gonzalez-Smith J, Bleser WK, Muhlestein D, Richards R, McClellan MB, Saunders RS. Medicare ACO results for 2018: more downside risk adoption, more savings, and all ACO types now averaging savings. Health Affairs. October 25, 2019. Accessed June 2, 2020. https://www.healthaffairs.org/do/10.1377/hblog20191024.65681/full/
8. Chernew ME, Conway PH, Frakt AB. Transforming Medicare’s payment systems: progress shaped by the ACA. Health Aff (Millwood). 2020;39(3):413-420. doi:10.1377/hlthaff.2019.01410
9. Bitton A, Flier LA, Jha AK. Health information technology in the era of care delivery reform: to what end? JAMA. 2012;307(24):2593-2594. doi:10.1001/jama.2012.6663
10. King J, Patel V, Jamoom E, DesRoches C. The role of health IT and delivery system reform in facilitating advanced care delivery. Am J Manag Care. 2016;22(4):258-265.
11. Lin SC, Hollingsworth JM, Adler-Milstein J. Alternative payment models and hospital engagement in health information exchange. Am J Manag Care. 2019;25(1):e1-e6.
12. Office of the National Coordinator for Health Information Technology, HHS. 21st Century Cures Act: interoperability, information blocking, and the ONC Health IT certification program. Fed Regist. 2019;84(42):7424-7610.
13. Pylypchuk Y, Barker W, Encinosa W, Searcy T. Impact of the 2015 Health Information Technology Certification Edition on interoperability among hospitals. J Am Med Inform Assoc. 2021;28(9):1866-1873.
14. Apathy NC, Vest JR, Adler-Milstein J, Blackburn J, Dixon BE, Harle CA. Practice and market factors associated with provider volume of health information exchange. J Am Med Inform Assoc. 2021;28(7):1451-1460. doi:10.1093/jamia/ocab024
15. Adler-Milstein J, Garg A, Zhao W, Patel V. A survey of health information exchange organizations in advance of a nationwide connectivity framework. Health Aff (Millwood). 2021;40(5):736-744. doi:10.1377/hlthaff.2020.01497
16. Everson J, Adler-Milstein J, Ryan AM, Hollingsworth JM. Hospitals strengthened relationships with close partners after joining accountable care organizations. Med Care Res Rev. 2020;77(6):549-558. doi:10.1177/1077558718818336
17. Nwafor O, Johnson NA. The effect of participation in accountable care organization on electronic health information exchange practices in U.S. hospitals. Health Care Manage Rev. Published online July 26, 2021. doi:10.1097/HMR.0000000000000319
18. Vest JR, Simon K. Hospitals’ adoption of intra-system information exchange is negatively associated with inter-system information exchange. J Am Med Inform Assoc. 2018;25(9):1189-1196. doi:10.1093/jamia/ocy058
19. CMS, HHS. Medicare program; Merit-Based Incentive Payment System (MIPS) and Alternative Payment Model (APM) incentive under the Physician Fee Schedule, and criteria for physician-focused payment models. final rule with comment period. Fed Regist. 2016;81(214):77008-77831.
20. Rudin RS, Bates DW, MacRae C. Accelerating innovation in health IT. N Engl J Med. 2016;375(9):815-817. doi:10.1056/NEJMp1606884
21. Yeager VA, Zhang Y, Diana ML. Analyzing determinants of hospitals’ accountable care organizations participation: a resource dependency theory perspective. Med Care Res Rev. 2015;72(6):687-706. doi:10.1177/1077558715592295
22. Walker DM, Mora AM, Scheck McAlearney A. Accountable care organization hospitals differ in health IT capabilities. Am J Manag Care. 2016;22(12):802-807.
23. Chukmaitov AS, Harless DW, Bazzoli GJ, Deng Y. Factors associated with hospital participation in Centers for Medicare and Medicaid Services’ accountable care organization programs. Health Care Manage Rev. 2019;44(2):104-114. doi:10.1097/HMR.0000000000000182
24. Cross DA, Adler-Milstein J. Investing in post-acute care transitions: electronic information exchange between hospitals and long-term care facilities. J Am Med Dir Assoc. 2017;18(1):30-34. doi:10.1016/j.jamda.2016.07.024
25. Heisey-Grove D, Patel V. National findings regarding health IT use and participation in health care delivery reform programs among office-based physicians. J Am Med Inform Assoc. 2017;24(1):130-139. doi:10.1093/jamia/ocw065
26. Balio CP, Apathy NC, Danek RL. Health information technology and accountable care organizations: a systematic review and future directions. EGEMS (Wash DC). 2019;7(1):24. doi:10.5334/egems.261
27. Hacker K, Walker DK. Achieving population health in accountable care organizations. Am J Public Health. 2013;103(7):1163-1167. doi:10.2105/AJPH.2013.301254
28. Lewis VA, Colla CH, Carluzzo KL, Kler SE, Fisher ES. Accountable care organizations in the United States: market and demographic factors associated with formation. Health Serv Res. 2013;48(6, pt 1):1840-1858. doi:10.1111/1475-6773.12102
29. Ouayogodé MH, Colla CH, Lewis VA. Determinants of success in Shared Savings Programs: an analysis of ACO and market characteristics. Healthc (Amst). 2017;5(1-2):53-61. doi:10.1016/j.hjdsi.2016.08.002
30. Tanenbaum JE, Votruba M, Einstadter D, Love TE, Cebul RD. Adoption of health system innovations: evidence of urban-rural disparities from the Ohio primary care marketplace. J Gen Intern Med. 2021;36(6):1584-1590. doi:10.1007/s11606-020-06440-7
31. Clarke RMA, Jeffrey J, Grossman M, Strouse T, Gitlin M, Skootsky SA. Delivering on accountable care: lessons from a behavioral health program to improve access and outcomes. Health Aff (Millwood). 2016;35(8):1487-1493. doi:10.1377/hlthaff.2015.1263
32. Chang CH, Mainor A, Colla C, Bynum J. Utilization by long-term nursing home residents under accountable care organizations. J Am Med Dir Assoc. 2021;22(2):406-412. doi:10.1016/j.jamda.2020.05.055
33. Agarwal D, Werner RM. Effect of hospital and post-acute care provider participation in accountable care organizations on patient outcomes and Medicare spending. Health Serv Res. 2018;53(6):5035-5056. doi:10.1111/1475-6773.13023
34. Bleser WK, McClellan MB, Winfield L. Medicare accountable care spending patterns: shifting expenditures associated with savings. Am J Accountable Care. 2018;6(1):11-19.
35. Everson J, Lee SYD, Friedman CP. Reliability and validity of the American Hospital Association’s national longitudinal survey of health information technology adoption. J Am Med Inform Assoc. 2014;21(e2):e257-e263. doi:10.1136/amiajnl-2013-002449
36. Apathy NC, Holmgren AJ. Opt-in consent policies: potential barriers to hospital health information exchange. Am J Manag Care. 2020;26(1):e14-e20. doi:10.37765/ajmc.2020.42148
37. Pylypchuk Y, Johnson C, Patel V. State of interoperability among U.S. non-federal acute care hospitals in 2018. ONC data brief No. 51. March 2020. Accessed July 15, 2021. https://www.healthit.gov/sites/default/files/page/2020-03/State-of-Interoperability-among-US-Non-federal-Acute-Care-Hospitals-in-2018.pdf
38. Massetti B, Zmud RW. Measuring the extent of EDI usage in complex organizations: strategies and illustrative examples. MIS Q. 1996;20(3):331-345. doi:10.2307/249659
39. Kanter GP, Polsky D, Werner RM. Changes in physician consolidation with the spread of accountable care organizations. Health Aff (Millwood). 2019;38(11):1936-1943. doi:10.1377/hlthaff.2018.05415
40. Callaway B, Sant’Anna PHC. Difference-in-differences with multiple time periods. J Econom. 2021;225(2):200-230. doi:10.1016/j.jeconom.2020.12.001
41. Wing C, Simon K, Bello-Gomez RA. Designing difference in difference studies: best practices for public health policy research. Annu Rev Public Health. 2018;39:453-469. doi:10.1146/annurev-publhealth-040617-013507
42. Heckman JJ, Ichimura H, Todd PE. Matching as an econometric evaluation estimator: evidence from evaluating a job training programme. Rev Econ Stud. 1997;64(4):605-654. doi:10.2307/2971733
43. Goodman-Bacon A. Difference-in-differences with variation in treatment timing. National Bureau of Economic Research working paper No. 25018. September 2018. Accessed October 16, 2021. http://www.nber.org/papers/w25018
44. de Chaisemartin C, D’Haultfœuille X. Two-way fixed effects estimators with heterogeneous treatment effects. Am Econ Rev. 2020;110(9):2964-2996. doi:10.1257/aer.20181169
45. McCullough J, Casey M, Moscovice I, Burlew M. Meaningful use of health information technology by rural hospitals. J Rural Health. 2011;27(3):329-337. doi:10.1111/j.1748-0361.2010.00359.x
46. Apathy NC, Holmgren AJ, Adler-Milstein J. A decade post-HITECH: critical access hospitals have electronic health records but struggle to keep up with other advanced functions. J Am Med Inform Assoc. 2021;28(9):1947-1954. doi:10.1093/jamia/ocab102
47. Burke RE, Phelan J, Cross D, Werner RM, Adler-Milstein J. Integration activities between hospitals and skilled nursing facilities: a national survey. J Am Med Dir Assoc. 2021;22(12):2565-2570.e4. doi:10.1016/j.jamda.2021.05.005
48. Adler-Milstein J, Raphael K, O’Malley TA, Cross DA. Information sharing practices between US hospitals and skilled nursing facilities to support care transitions. JAMA Netw Open. 2021;4(1):e2033980. doi:10.1001/jamanetworkopen.2020.33980
49. Connecting health and care for the nation: a shared nationwide interoperability roadmap. Office of the National Coordinator for Health Information Technology. 2016. Accessed October 16, 2021. https://www.healthit.gov/sites/default/files/hie-interoperability/nationwide-interoperability-roadmap-final-version-1.0.pdf