Abstract
Objective
Enterprise health information exchange (HIE) and a single electronic health record (EHR) vendor solution are 2 information exchange approaches to improve performance and increase the quality of care. This study sought to determine the association between adoption of enterprise HIE vs a single vendor environment and changes in unplanned readmissions.
Materials and Methods
The association between unplanned 30-day readmissions among adult patients and adoption of enterprise HIE or a single vendor environment was measured in a panel of 211 system-member hospitals from 2010 through 2014 using fixed-effects regression models. Sample hospitals were members of health systems in 7 states. Enterprise HIE was defined as self-reported ability to exchange information with other members of the same health system who used different EHR vendors. A single EHR vendor environment reported exchanging information with other health system members, but all using the same EHR vendor.
Results
Enterprise HIE adoption was more common among the study sample than EHR (75% vs 24%). However, adoption of a single EHR vendor environment was associated with a 0.8% reduction in the probability of a readmission within 30 days of discharge. The estimated impact of adopting an enterprise HIE strategy on readmissions was smaller and not statically significant.
Conclusion
Reductions in the probability of an unplanned readmission after a hospital adopts a single vendor environment suggests that HIE technologies can better support the aim of higher quality care. Additionally, health systems may benefit more from a single vendor environment approach than attempting to foster exchange across multiple EHR vendors.
Keywords: health information exchange, electronic health records, hospitals, patient readmission policy
INTRODUCTION
A majority of hospitals and ambulatory practices now use electronic health records (EHRs) due in part to the $26 billion Medicare and Medicaid EHR Incentive Program, commonly called Meaningful Use.1,2 Despite national certification criteria in 2012 requiring exchange capabilities and the declaration that “widespread exchange of health information” is a national objective,3 the patient information stored within EHRs is often not easily shared or is difficult to combine across different vendors’ products.4,5 The inability to efficiently and effectively share patient data between EHRs is a particularly acute problem for health systems.4 Health systems are responsible for individual patients seeking care across multiple points of service, but they can have different vendors supplying EHRs for inpatient and outpatient settings or the health system may be comprised of hospitals using different EHR vendors or products.6–8 Health systems need a strategy to improve internal information exchange among their disparate technological investments. Two principal technological strategies—enterprise health information exchange (HIE) and single vendor solutions—exist for health systems to improve information sharing and enhance access to patient information.
Health systems with EHRs from multiple vendors have often adopted enterprise HIE technology.9–11 The presence of different EHR vendors within a single organization may be a product of mergers and acquisitions involving other hospitals and provider groups12 or the explicit choice to use 1 vendor for the inpatient setting and another for the outpatient settings.13 Enterprise HIE is an information-sharing network that connects the EHRs of a health systems’ hospitals and affiliated providers (eg, owned practices or joint venture hospitals). Beyond these internal organizational members, health systems may extend enterprise HIE connectivity to any referring practices and other health care organizations with which the system wants to foster a more exclusive or stronger relationship.10 Enterprise HIE supports organizations’ access to patient information through internal information exchange, which health system leaders anticipate will result in improved care and reductions in potentially avoidable utilization.7
Health systems may choose to pursue a single EHR vendor solution for all member hospitals and ambulatory care practices.14 By placing all providers of a health system’s care settings into a single vendor’s EHR, information accessibility within the organization may be improved. In addition, a single EHR may further support better care coordination, comprehensive patient information, easier scheduling, less complexity, standardized processes, providing an analytics platform for strategic decision-making, and use of a single sign-on environment for end users.15–17 Health systems with a single vendor often have a single EHR instance (ie, 1 environment that includes all patient data and functionality) serving all of the system’s providers. However, even if a health system opts to use different EHR instances from the same vendor at 1 or more member sites, the inherent interoperability within a single vendor product continues to support the exchange of relevant patient information.18 A single EHR vendor environment is effective in fulfilling providers’ information needs for patient history and prior results.19
Both enterprise HIE and a single EHR vendor environment improve access to the patient information created by the health system through care delivery. Better access to needed patient information at the point of care is the underlying mechanism by which HIE may produce beneficial results.20 For example, better information-sharing supports more effective and efficient care delivery during patients’ transitions across different settings.21,22 Likewise, access to comprehensive patient information supports medication safety,23 and better information-sharing facilitates ambulatory care providers’ access to specialists’ information24 or postdischarge reports.25 Moreover, both approaches may better position health systems pursuing the financial benefits of, or avoiding the penalties associated with, current health policy initiatives. For example, the aggregated clinical data resulting from these approaches can support population health monitoring26 as well as risk-based analytics.7 These capabilities directly address many of the challenges posed by the Hospital Readmissions Reduction Program,27 Accountable Care Organizations,28 and Bundled Payments for Care Improvement.29
However, the evidence indicating improvement in the quality of care associated with either enterprise HIE or a single vendor environment is generally limited.11 Some evidence suggests that health systems’ adoption of a single vendor environment creates efficiency gains.30 Likewise, use of vendor-mediated exchange has documented decision-making benefits19 and enterprise HIE systems have been shown to reduce health care utilization.31 However, these examples are based on studies of a single institution or in an international setting, thereby limiting generalizability.
This study sought to determine the association between adoption of enterprise HIE vs a single vendor environment and unplanned readmissions among hospitalized adults in 7 states. We focus on readmissions due to their recognition and use as a measure of quality for numerous policy initiatives.32 In addition, because readmissions frequently result from poor communication during transitions of care,33 they may be amenable to reductions through better information access due to HIE.20
MATERIALS AND METHODS
In a longitudinal analysis covering a 5-year period (2010–2014), we examined unplanned readmissions among patients in health systems in 7 states, before and after adoption of enterprise HIE or a single vendor environment.
Data and sample
The analytic file was constructed from the American Hospital Association’s (AHA) Annual Survey and Health Information Technology (HIT) Supplement Survey and the Healthcare Cost and Utilization Project’s (HCUP) State Inpatient Databases from the Agency for Healthcare Research and Quality (AHRQ).34
The study population included adult (≥ 20 years old) admissions to nonfederal hospitals in Arkansas, California, Florida, Iowa, Maryland, New York, and Washington. These 7 states were selected because each participates in HCUP by providing information on hospitalizations that includes readmission flags. Currently, California only provides information on hospitalizations through 2011, but additional years of data were obtained directly from the state’s Office of Statewide Health Planning and Development. Additionally, these states reflect a wide geographic and socio-economic subset of the nation’s population. Hospitals from these states meeting the following conditions were included in the sample: (1) the hospital had responded to 2 or more years of the AHA’s Annual Survey and HIT Supplement Survey during the study period; (2) at baseline, it had not adopted at least a basic EHR,35 but eventually adopted either enterprise HIE or a single vendor environment; (3) the hospital was a member of a system (as indicated by a system identifier in the AHA survey); and (4) it was not a children’s or psychiatric hospital. The first 2 criteria allow for longitudinal measurement and estimation of the impact of a change in HIE status on readmissions. Also, because all hospitals at baseline did not have an EHR, none had prior experience with interoperable (ie, the ability to exchange information between different systems)36,37 health information technology. System membership was required because enterprise HIE connects affiliated hospitals and practices. We also excluded admissions related to labor and delivery, admissions where the patient died prior to discharge, and admissions with a length of stay of > 50 days to reduce the influence of outliers. To account for mergers, we followed a previously used approach38 of combining a hospital’s premerger characteristics into the single postmerger entity for the entire study period. The final sample included 7 956 515 admissions to 211 hospitals.
Determinants of interest
The primary determinants of interest were 2 binary measures reflective of a hospital’s global adoption of an internal information exchange strategy in a given year based on AHA HIT survey responses: 1 for enterprise HIE and 1 for a single vendor environment. In our sample, all hospitals that adopted EHRs did so with 1 of these 2 strategies. Enterprise HIE adoption was defined based on a hospital meeting 3 criteria: (1) the hospital had an EHR; (2) the hospital reported exchanging or sharing at least 1 of 5 types of electronic information (demographics, laboratory results, medication history, radiology reports, or clinical care records/summaries) with another inpatient or outpatient member of the same health system (ie, intraorganizational exchange); and (3) members of the health system reported different EHR vendors for either inpatient or outpatient services.7,39,40 We aggregated all members of the same health system to establish concordance or differences among the reported inpatient and outpatient EHR vendors and included all types of information sharing (demographics, laboratory results, medication history, radiology reports, or clinical care records/summaries) recorded in the AHA HIT survey. Hospitals that were part of a single EHR vendor environment were identified through a similar process: (1) the hospital had an EHR; (2) the hospital reported at least 1 of 5 types of electronic sharing of information with another inpatient or outpatient member of the same health system; and (3) all other members of the health system reported the same EHR vendor. In order to ensure that adoption of either internal information exchange strategy occurred prior to outcomes experienced by patients, we lagged adoption by 1 time period (eg, using the prior year). By the end of the study period, all hospitals in the sample had adopted 1 of the 2 strategies. See the Supplementary Appendix for more details.
Outcomes
The primary outcome was a measure of an all-cause, unplanned hospital readmission within 30 days of discharge. Using synthetic identifiers for unique patients, these events were identified based on the Centers’ for Medicare & Medicaid Services definition for an unplanned readmission.41 Following Agency for Healthcare Research and Quality’s guidance,42 our readmission measure accounted for patient transfers between facilities and known state-specific issues with following patients across calendar years (see Supplementary Appendix). Admissions without record linkage identifiers (ie, the admission had insufficient information to accurately identify the patient) and records that included negative days until readmission were excluded from the analysis.43
Covariates
Admission-level covariates included patient demographics (eg, age, gender, race/ethnicity), primary payer, Elixhauser comorbidity index,44 length of stay, and whether the admission occurred on a weekend.45 We characterized hospitals by size in terms of bed count, ownership, and rural location. We also used the health care system taxonomy to describe the various dominant combinations of service diversification and centralization of authority observed across health systems.46 Because vertical integration between hospitals and outpatient practices could improve coordination of care and subsequently lead to lower readmission rates,47 we created annual indicators for any contractual or employment affiliations reported at the hospital or system levels.48 The number of hospitals with common health system identifiers in the AHA survey was used as a proxy for overall health system size. External information exchange, the ability to share information with ambulatory care providers or hospitals outside the health system, was measured as self-reported participation in a regional health information organization or health information exchange and lagged by 1 period.
Analyses
We described the adoption of enterprise HIE and single vendor environments over time for the study sample. Next, we characterized the time-invariant and baseline characteristics using frequencies and percentages stratified by the type of internal information exchange strategy eventually adopted. A comparison of the characteristics of hospitals included in the sample with the characteristics of excluded hospitals is available in the Supplementary Appendix.
To estimate the effect of adopting an enterprise HIE or a single EHR vendor environment, we used a hospital fixed-effects regression that compares the change in readmission probabilities among patients before and after a hospital adopted either approach between 2010 and 2014, while controlling for time invariant differences. The following model was used for estimation:
Patients (ie, admissions) are indexed by , hospitals by , health systems by , and time by . is the binary measure of a hospital’s adoption of a single EHR vendor approach to internal information exchange in a given year and is the binary measure of a hospital’s adoption of an enterprise HIE in a given year (ie, hospitals may change approaches from year to year). is a set of time dummies, represents hospital fixed effects, includes time varying, patient-level factors, and is the error term. Time invariant hospital and system level characteristics were controlled through the inclusion of respective fixed effects. provides the estimated effect of a hospital adopting a single EHR vendor environment and the effect of adopting an enterprise HIE approach, compared to no internal information sharing technology adoption. The difference between and represents the relative effect of adopting a single EHR environment compared to enterprise HIE. Estimates from a series of robustness checks for our primary modeling strategy are reported in the Supplementary Appendix. Regression models were fitted using the REGHDFE module in Stata MP, version 15.149 and were estimated with robust standard errors adjusted for clustering at the level of the hospital.
Our primary analysis, described above, provides estimates for the overall effects of enterprise HIE and single EHR vendor adoption. In the following series of supplemental analyses, we explored the context and nature of information exchange in more detail. First, we assessed the overall effect of adopting either interoperability strategy using a binary measure of a hospital’s adoption of either a single EHR vendor or enterprise HIE in a given year. Second, we controlled for annual counts of the types of information exchanged intraorganizationally and interorganizationally between ambulatory providers and hospitals. This provided estimated effects for enterprise HIE and vendor-mediated exchange adoption adjusted for the total types of information exchanged internally and externally. Third, because enterprise HIEs may be used to connect any number of different EHR vendors, we created a categorical measure of: no adoption; enterprise HIE with 2 EHR vendors; and enterprise HIE with 3 or more different EHR vendors. This analysis was limited to hospitals that eventually adopted enterprise HIE. All models controlled for the same factors described for our primary analysis. Estimates were expressed as marginal means and graphed for ease of interpretation.
RESULTS
In our sample of hospitals, both adoption of enterprise HIE and single vendor environments increased over time, with enterprise HIE (75%) being the more common internal information exchange solution by the end of the study period (Figure 1). Hospitals in the sample were most commonly nonprofit and general acute care facilities; however, larger proportions of those adopting single vendor environments were public or specialty hospitals (Table 1). Hospitals adopting enterprise HIE were more likely to be members of larger health systems and to belong to decentralized systems, whereas those adopting single vendor environments tended to be part of a centralized system. In terms of patient admissions, those of single vendor environment hospitals had higher mean Elixhauser scores and longer average lengths of stay. Among hospitals in the first year of inclusion in the sample, readmission rates of those adopting enterprise HIE and of those adopting a single vendor environment were not statistically different. Adoption trends were similar for hospitals in all 7 states, but a larger proportion of those excluded from the sample were specialty and/or for-profit hospitals, and were more likely to be part of decentralized systems (see Supplementary Appendix).
Figure 1.
Adoption of enterprise health information exchange and single electronic health record vendor environments among hospitals initiating health information exchange in Arkansas, California, Florida, Iowa, Maryland, New York, and Washington over the period 2010–2014.
Table 1.
Characteristics of all hospitals and admissions in the study sample, as well as characteristics stratified by the adoption of enterprise health information exchange or single vendor environments
Hospital Characteristics | Total | Adopted enterprise HIE | Adopted single vendor a | P |
---|---|---|---|---|
n = 211 | n = 158 | n = 53 | ||
Control | ||||
Public | 13.3 | 10.1 | 22.6 | .003 |
Non-profit | 73.5 | 72.8 | 75.5 | |
Private | 13.3 | 17.1 | 1.9 | |
Service Category | ||||
General acute care | 97.6 | 99.4 | 92.5 | .004 |
Specialty | 2.4 | 0.6 | 7.5 | |
Bed size | ||||
< 50 | 16.6 | 15.8 | 18.9 | .958 |
50–99 | 12.8 | 12.7 | 13.2 | |
100–299 | 38.9 | 39.2 | 37.7 | |
≥ 300 | 32.8 | 32.3 | 30.2 | |
System type | ||||
Centralized | 24.2 | 17.1 | 45.3 | < .001 |
Moderately centralized | 16.6 | 15.8 | 18.9 | |
Decentralized | 44.5 | 52.3 | 20.8 | |
Independent system | 11.8 | 12.7 | 9.4 | |
Unclassified | 2.8 | 1.9 | 5.7 | |
Vertical integration | 75.0 | 76.0 | 71.7 | < .001 |
Hospitals in system (mean, sd) | 16.69 (18.65) | 19.93 (20.14) | 7.04 (7.29) | < .001 |
Metropolitan location | 75.8 | 75.3 | 77.4 | .764 |
State | ||||
AR | 10.0 | 8.2 | 15.1 | < .001 |
CA | 25.1 | 30.4 | 9.4 | |
FL | 21.3 | 15.2 | 39.6 | |
IA | 15.2 | 13.9 | 18.9 | |
MD | 6.6 | 7.0 | 5.7 | |
NY | 19.0 | 24.5 | 11.3 | |
WA | 2.8 | 3.8 | 0.0 | |
RHIO participant | 28.9 | 29.8 | 26.4 | .643 |
Admission Characteristicsb | ||||
Female gender | 57.7 | 58.1 | 56.4 | < .001 |
Race/ethnicity | ||||
White | 65.3 | 65.9 | 63.1 | < .001 |
African American | 14.4 | 14.0 | 16.0 | |
Hispanic | 12.1 | 10.9 | 16.0 | |
Other/unknown | 8.2 | 9.2 | 4.9 | |
Age | ||||
< 45 | 24.4 | 24.4 | 24.4 | < .001 |
45–64 | 30.6 | 30.5 | 30.9 | |
65–74 | 17.1 | 17.1 | 17.2 | |
74–85 | 17.1 | 17.1 | 17.0 | |
> 85 | 10.8 | 10.9 | 10.5 | |
Payer | ||||
Medicare | 49.2 | 48.9 | 50.1 | < .001 |
Medicaid | 14.6 | 14.7 | 13.9 | |
Private | 27.8 | 28.4 | 25.7 | |
Other | 8.5 | 7.9 | 10.3 | |
Weekend admission | 20.4 | 22.8 | 13.2 | .134 |
Unplanned readmission rate (mean, sd) | 0.158 (0.364) | 0.157 (0.364) | 0.158 (0.365) | .379 |
Elixhauser (mean, sd) | 12.94 (14.30) | 12.85 (14.25) | 13.26 (14.48) | < .001 |
Length of stay (mean, sd) | 4.61 (4.93) | 4.57 (4.91) | 4.76 (4.99) | < .001 |
Abbreviations: RHIO, regional health information organization; sd, standard deviation.
45.3% Epic; 15.1% Cerner; 9.4% Meditech; 30.2% all others.
At hospital’s first study year.
In analyses adjusted for hospital, system, and time fixed effects (Table 2), adoption of a single EHR vendor environment was associated with a lower probability of an unplanned readmission (ß = −0.0077; P = .032). The marginal effect was equivalent to a reduction in the readmission rate from 15.8% to 15.0%. The estimated marginal effect of enterprise HIE was equivalent to a reduction from a readmission rate of 15.8% to 15.5% (ß = −0.0030; P = .095), but the estimate was not statistically significant at the 5% level. There was no statistically significant difference between the 2 information exchange approaches (P = .210).
Table 2.
Association between enterprise health information exchange or single vendor environment adoption and unplanned 30-day readmissionsa
Unplanned readmission |
||
---|---|---|
Unadjusteda | Adjusted | |
n=5 766 952 | n=5 765 210 | |
Adopted single EHR vendor environment | −0.0077 (−0.0148, −0.0007)* | −0.0073 (−0.0144, −0.0002)* |
Adopted enterprise HIE | −0.0030 (−0.0065, 0.0005) | −0.0036 (−0.0081, 0.0008) |
RHIO participant | −0.0029 (−0.0036, 0.0094) | −0.0040 (−0.0029, 0.0111) |
Vertical integration | −0.0013 (−0.0050, 0.0024) | −0.0004 (−0.0071, 0.0023) |
Hospitals in system | 0.0001 (−0.0001, 0.0001) | 0.0001 (−0.0001, 0.0001) |
Patient characteristics | ||
Female | −0.0335 (−0.0376, −0.0294)*** | −0.0200 (−0.0239, −0.0161)*** |
Age | ||
< 45 | Reference | Reference |
45 − 64 | 0.0458 (0.0426, 0.0490)*** | 0.0080 (0.0049, 0.0111)*** |
65 − 74 | 0.0486 (0.0443, 0.0528)*** | −0.0276 (−0.0337, −0.0215)*** |
74 − 85 | 0.0580 (0.0530, 0.0630)*** | −0.0302 (−0.0374, −0.0230)*** |
> 85 | 0.0568 (0.0511, 0.0624)*** | −0.0342 (−0.0420, −0.0264)*** |
Race/ethnicity | ||
White | Reference | Reference |
African American | 0.0276 (0.0238, 0.0314)*** | 0.0139 (0.0101, 0.0177)*** |
Hispanic | −0.0111 (−0.0155, −0.0066)*** | −0.0075 (−0.0104, −0.0045)*** |
Other/unknown | −0.0320 (−0.0370, −0.0270)*** | −0.0224 (−0.0275, −0.0173)*** |
Payera | ||
Private | Reference | Reference |
Medicare | 0.0814 (0.0775, 0.0853)*** | 0.0611 (0.0558, 0.0665)*** |
Medicaid | 0.0725 (0.0638, 0.0812)*** | 0.0566 (0.0491, 0.0642)*** |
Other | 0.0127 (0.0076, 0.0177)*** | 0.0012 (-0.0024, 0.0047) |
Weekend admission | 0.0108 (0.0094, 0.0122)*** | 0.0052 (0.0041, 0.0063)*** |
Elixhauser (mean, sd) | 0.0045 (0.0043, 0.0047)*** | 0.0036 (0.0035, 0.0038)*** |
Length of stay (mean, sd) | 0.0085 (0.0078, 0.0093)*** | 0.0044 (0.0037, 0.0050)*** |
Abbreviations: HER, electronic health record; HIE, health information exchange; RHIO, regional health information organization; sd, standard deviation.
Hospital and time dummies not reported. *** P < .001; *P < .05.
The negative association between a single EHR vendor environment with unplanned readmissions persisted after adjusting for patient-level factors (ß = −0.0073; P = .043), equivalent to a 0.7 percentage point reduction. Adoption of an enterprise HIE strategy was negatively associated with the probability of an unplanned readmission (ß = −0.0036; P = .110), but was not statistically significant. In addition, the effect of enterprise HIE adoption did not differ significantly from the adoption of a single vendor environment (P = .345). The probability of an unplanned readmission was higher when Medicare or Medicaid was the payer, when the index admission was on a weekend, for higher Elixhauser scores, and for longer hospital stays. The results of our robustness checks were consistent with these findings (see Supplementary Appendix).
Our supplemental analyses provided additional insights into the relationships between adoption of enterprise HIE and/or a single vendor environment with readmissions. In our regression specification using a single indicator for adoption of either approach, internal information exchange adoption was associated with a 0.4 percentage point (ß = −0.0039) reduction in the probability of an unplanned readmission (P = .085) after controlling for the same factors used in our primary analysis (Figure 2). After controlling for the count of types of information exchanged intraorganizationally and interorganizationally, the probability of an unplanned readmission was reduced by 0.75 percentage points (ß = −0.0075; P = .044) among hospitals that adopted vendor-mediated exchange and 0.4 percentage points among those that adopted enterprise HIE (ß = −0.0042; P = .062). In our analysis limited to hospitals that eventually adopted enterprise HIE, reductions in the probability of an unplanned readmission were 0.8 percentage points (ß = −0.0079; P = .010) lower when only 2 EHR vendors were connected. When 3 or more EHR vendors were connected via enterprise HIE, no statistically significant difference was observed (P = .173).
Figure 2.
Marginal means of unplanned readmission rates by different approaches and structures to intraorganizational information exchange.11Adjusted for the factors (other than type of health information exchange) reported in Table 1, but omitted for readability (see Supplementary Appendix for full results) Abbreviations: eHIE, enterprise health information exchange; her, electronic health record. * P < .05 compared to other categories within the same panel.
DISCUSSION
Adoption of a single EHR vendor environment or an enterprise HIE are 2 approaches for hospitals to improve internal information exchange. In this 7-state, multiyear sample, decentralized health systems were more likely to pursue an enterprise HIE strategy and a single-vendor approach was more frequent among centralized systems. Hospital adoption of a single EHR vendor strategy for internal information exchange was associated with a lower probability of an unplanned readmission, but adoption of an enterprise HIE strategy was not associated with the likelihood of readmission. These findings suggest that interoperable health information technologies can support improved organizational performance.
Hospitals’ adoption of technologies that foster internal information exchange has been increasing over time.50 In our sample, the predominant strategy for internal information exchange was enterprise HIE. The key characteristics distinguishing hospitals choosing an enterprise HIE approach over a single vendor environment were health system size and a decentralized taxonomy type. The potential influence of health system size is likely a reflection of the challenges of scale and a product of individual history. For health systems with a small number of member hospitals, the installation of a single vendor solution may be easier from an implementation standpoint. For larger systems, an enterprise HIE strategy may be a more pragmatic choice due to inherited legacy systems across member hospitals that joined through mergers and acquisitions.7 Likewise, decentralized health systems could be more prone to multivendor environments that require an enterprise HIE strategy, because the individual hospital members enjoy a greater degree of autonomy for decision-making and also tend to be geographically dispersed.46 The opposite characteristics of centralized health systems, such as decision-making at the system level,51 may support the higher prevalence of single vendor environments observed in this study. Previous research has suggested that centralized systems are associated with greater information system interoperability regardless of strategy.52
Whereas enterprise HIE was the more common strategy, adoption of a single vendor EHR environment was associated with a lower probability of an unplanned readmission. Single vendor approaches can lead to better access to information from multiple settings of care,18 and they have the potential to create a more uniform experience for providers, which are common reasons for switching to this strategy.53 Additionally, a single EHR vendor environment can enable the aggregation of information into data warehouses to apply analytics to identify risk factors for patients.54 Health care organizations rely heavily on their EHRs to support population health initiatives,55 and because the single vendor environment provides a consistent user experience, this may allow organizations to efficiently and broadly introduce standardized workflows and interventions.18,56 In addition, the single vendor approach may represent a lower degree of complexity both in terms of training, customization, and implementation time compared to an enterprise HIE strategy, which must cope with differing needs of users across multiple EHRs. The negative association between readmissions and having 2 connected EHRs via enterprise HIE and the lack of a statistically significant reduction when 3 or more EHRs were connected, provides some evidence for this inference. In general, the fact that an enterprise HIE strategy can provide benefits through aggregated information and analytics11,57 may explain the consistent, though nonsignificant, negative relationship with unplanned readmissions found in our results.
A secondary, but important, finding was the overall negative association between hospitals’ adoption of any internal interoperable EHR and the probability of an unplanned readmission. In the year after adopting an EHR with either strategy, the probability of readmission decreased by 0.4 percentage points, relative to the base of 15.3%. This finding provides among the strongest evidence supporting the potential for interoperable health information technology to reduce readmissions. Although many hospitals use EHRs as part of their strategy to reduce readmissions,58 the results of earlier studies generally have not found a relationship between adoption of these systems and readmissions,59–64 with the exception of analyses based on cross-sectional or descriptive study designs.59,65,66 This study’s focus on the general adult patient population, rather than specifically focusing on Medicare beneficiaries, may explain some of the difference in our findings, as has been the case with studies examining the relationship between EHRs and patient safety among a broader adult population.67 Similarly, the literature on EHR adoption and readmissions in ambulatory care settings is also mixed.68,69
Our findings illustrate the difficulties of ongoing measurement and monitoring of hospitals’ adoption of health information technologies. First, the health information technology landscape has changed dramatically in a short period of time and continues to evolve rapidly.70 As such, our available secondary data sources and surveillance tools, while indispensable, are inevitably behind the curve of market and industry trends. Second, with this evolution, health information technology constructs are becoming more complicated and nuanced. For example, enterprise HIE and single vendor EHR environments are specific constructs defined by the combination of technology and organizational structure.7 Interoperability is multidimensional71 and, in general, reliable measurement of health information technologies as a basic structural feature has proven difficult.72 Problematically, as technologies evolve in complexity and their application becomes ubiquitous, detailed analyses of specific functions and characteristics will become increasingly important but likely more difficult to collect. This study is a case in point of this limitation of structural measurement: 1 that cannot account for differences in functionality or user experience between different products. Moreover, even well-defined constructs face the ongoing challenge of adoption not being equivalent to actual usage.73 Benefits of HIE may be delayed or nonexistent,74 multiple systems may be in use simultaneously,75 and true interoperability may not really be present.5
Limitations
The primary limitation of our study is that we are unable to distinguish the effects of EHR adoption with and without an internal inoperability strategy. For such a comparison, we would need hospitals that adopted an EHR, but did not have any internal information exchange, which did not occur in a sufficient number of hospitals in our sample (less than 2% of all hospitals in our study states reported adopting a noninteroperable EHR). In addition, our focus on the internal information exchange strategies of enterprise HIE and single vendor environments necessarily limits the generalizability of our findings to system member hospitals. Further, since we tend to observe more centralized systems adopting single vendor systems and more decentralized systems adopting enterprise HIE, our results are not necessarily reflective of the experience of the average hospital in each of these system cluster types. Since system taxonomy is a time-invariant factor in our panel, we were not able to sufficiently explore its impact on readmissions. Although our fixed-effects approach controls for permanent differences between hospital systems, selection of an information exchange strategy is not a random choice and may be correlated with unobserved organizational factors that change over time such as the system’s structure, broad information technology capability, or business and clinical goals. In terms of measurement, our identification of single vendor environments was based on self-report and we have no method of distinguishing between systems running single or multiple EHR instances. In addition, during the study period, numerous changes were occurring within the US health care system, such as the implementation of Medicare Accountable Care Organizations, the Hospital Readmissions Reduction Program, Bundled Payments for Care Improvement, among others. However, it is unlikely that hospitals’ adoption of either internal information exchange approach systematically coincided with the timing of the implementation of these initiatives. Given that adoption of either enterprise HIE or a single vendor EHR environment is tied to organizational strategy, adoption decisions may have implications for other areas of organizational performance, such as financial stability and aspects of care delivery. Also, while the Elixhauser comorbidity index is routinely used to risk-adjust readmissions in studies using HCUP data sets,76 use of another type of risk-adjustment index could have influenced our estimates.
CONCLUSION
This is the first study to examine whether the implementation of enterprise HIE or a single vendor environment by a health system reduces hospital readmissions. Reductions in the probability of an unplanned readmission after a hospital adopts a single vendor environment suggests that health information exchange technologies can support the aim of higher quality care.
FUNDING
This work was supported by the Agency for Healthcare Research & Quality (1R21HS024717-01A1; Vest PI).
AUTHOR CONTRIBUTIONS
JV, MU, SF, and KS conceived the research study. JV, MU, and SF led the data analysis. JV, MU, LC, and JS drafted the manuscript and reviewed for critical content.
ETHICS APPROVAL
This research was approved by the Indiana University Institutional Review Board.
SUPPLEMENTARY MATERIAL
Supplementary material is available at Journal of the American Medical Informatics Association online.
Supplementary Material
ACKNOWLEDGMENTS
The authors acknowledge the Indiana University Pervasive Technology Institute (https://pti.iu.edu/) for providing high-performance computing resources that have contributed to the research results reported within this article. This research was supported in part by Lilly Endowment, Inc., through its support for the Indiana University Pervasive Technology Institute. This material is based upon work supported by the National Science Foundation under Grant No. CNS-0521433. The authors thank Nate Apathy for his assistance with the figures.
CONFLICT OF INTEREST STATEMENT
None declared.
REFERENCES
- 1. Furukawa MF, Patel V, Charles D, et al. Hospital electronic health information exchange grew substantially in 2008–12. Health Aff 2013; 328: 1346–54. [DOI] [PubMed] [Google Scholar]
- 2. Furukawa MF, King J, Patel V, et al. Despite substantial progress in EHR adoption, health information exchange and patient engagement remain low in office settings. Health Aff 2014; 339: 1672–9. [DOI] [PubMed] [Google Scholar]
- 3.114th Congress. Medicare Access and CHIP Reauthorization Act of 2015. 2015. https://www.congress.gov/bill/114th-congress/house-bill/2/text. Accessed February 28, 2018
- 4. Sheikh A, Sood HS, Bates DW, et al. Leveraging health information technology to achieve the “triple aim” of healthcare reform. J Am Med Inform Assoc. 2015;22(4):849–856. [DOI] [PMC free article] [PubMed]
- 5. Sittig DF, Wright A. What makes an EHR “open” or interoperable? J Am Med Inform Assoc 2015; 225: 1099–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Unertl KM, Johnson KB, Gadd CS, et al. Bridging organizational divides in health care: an ecological view of health information exchange. JMIR Med Inform 2013; 11: e3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Vest J, Kash B. Differing strategies to meet information sharing needs: the publicly supported community health information exchange versus health systems’ enterprise health information exchanges. Milbank Q 2016; 941: 77–108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Miller AR, Tucker C. Health information exchange, system size and information silos. J Health Econ 2014; 33: 28–42. [DOI] [PubMed] [Google Scholar]
- 9. Harris Healthcare Solutions. Harness the Power of Enterprise HIE. 2012. http://healthcare.harris.com/media/HH_WP_enterprise_HIE_tcm39-21233.pdf. (accessed July 21 2014)
- 10. Miller R. Satisfying patient-consumer principles for health information exchange: evidence from California case studies. Health Aff 2012; 313: 537–47. [DOI] [PubMed] [Google Scholar]
- 11. Everson J. The implications and impact of 3 approaches to health information exchange: community, enterprise, and vendor-mediated health information exchange. Learn Heal Syst 2017; 12: e10021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Becker's Healthcare. When EHRs collide: Dealing with multiple systems after mergers, acquisitions. https://www.beckershospitalreview.com/healthcare-information-technology/when-ehrs-collide-dealing-with-multiple-systems-after-mergers-acquisitions.html. Published 2015. Accessed May 26, 2019.
- 13. Payne T, Fellner J, Dugowson C, et al. Use of more than one electronic medical record system within a single health care organization. Appl Clin Inform 2012; 3: 462–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Conn J. Facilities Turning to Single Vendor for Interoperability. Modern Healthcare. 2013. http://www.modernhealthcare.com/article/20130918/NEWS/309189949. Accessed February 23, 2017.
- 15. Goedert J. Why SSM Health is Moving to One Enterprise EHR. Health Data Management. 2017. https://www.healthdatamanagement.com/news/why-ssm-health-is-moving-to-one-enterprise-ehr. Accessed August 15, 2018.
- 16.Impact Advisors. Newly Merged Presence Health Consolidates Epic Software. https://www.impact-advisors.com/document/newly-merged-presence-health-consolidates-epic-software/. Published 2015. Accessed July 2, 2019. [Google Scholar]
- 17. Ford EW, Menachemi N, Huerta TR, et al. Hospital IT adoption strategies associated with implementation success: implications for achieving meaningful use. J Healthc Manag 2010; 553: 175–88. [PubMed] [Google Scholar]
- 18. Koppel R, Lehmann CU. Implications of an Emerging EHR Monoculture for Hospitals and Healthcare Systems. Vol 22.; 2015. doi:10.1136/amiajnl-2014-003023. [DOI] [PMC free article] [PubMed]
- 19. Winden TJ, Boland LL, Frey NG, et al. Care everywhere, a point-to-point HIE tool. Utilization and impact on patient care in the ED. Appl Clin Inform 2014; 5: 388–401. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Hripcsak G, Kaushal R, Johnson KB, et al. The United Hospital Fund meeting on evaluating health information exchange. J Biomed Inform 2007; 406: S3–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Kash BA, Baek J, Davis E, et al. Review of successful hospital readmission reduction strategies and the role of health information exchange. Int J Med Inform 2017; 104: 97–104. [DOI] [PubMed] [Google Scholar]
- 22. Everson J, Kocher KE, Adler-Milstein J. Health information exchange associated with improved emergency department care through faster accessing of patient information from outside organizations. J Am Med Inform Assoc 2016; 169: 1023–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Boockvar KS, Ho W, Pruskowski J, et al. Effect of health information exchange on recognition of medication discrepancies is interrupted when data charges are introduced: results of a cluster-randomized controlled trial. J Am Med Informatics Assoc 2017; 246: 1095–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Eftekhari S, Gopal R, Ramesh R. Impact of health information exchange adoption on physician’s referral behavior. In: AMCIS 2016 Proceedings Published. Online First: 11 August 2016. https://aisel.aisnet.org/amcis2016/Health/Presentations/13. Accessed January 30, 2019.
- 25. Kierkegaard P, Kaushal R, Vest J. Patient information retrieval in multiple care settings: examining methods of exchange in emergency departments, primary care practices, and public health clinics? Am J Manag Care 2014; 20: SP494–501. [PubMed] [Google Scholar]
- 26. Vest J, Menachemi N. A population ecology perspective on the functioning and future of health information organizations. Health Care Manage Rev 2017; 1 https://insights.ovid.com/crossref?an=00004010-900000000-99727. doi: 10.1097/HMR.0000000000000185. [DOI] [PubMed] [Google Scholar]
- 27. Jackson CT, Trygstad TK, DeWalt DA, et al. Transitional care cut hospital readmissions for North Carolina medicaid patients with complex chronic conditions. Health Aff 2013; 328: 1407–15. [DOI] [PubMed] [Google Scholar]
- 28. DeVore S, Champion RW. Driving population health through accountable care organizations. Health Aff 2011; 301: 41–50. [DOI] [PubMed] [Google Scholar]
- 29. Williams C, Mostashari F, Mertz K, et al. From the office of the national coordinator: the strategy for advancing the exchange of health information. Health Aff 2012; 313: 527–36. [DOI] [PubMed] [Google Scholar]
- 30. Chen C, Garrido T, Chock D, et al. The kaiser permanente electronic health record: transforming and streamlining modalities of care. Health Aff 2009; 282: 323–33. [DOI] [PubMed] [Google Scholar]
- 31. Ben-Assuli O, Shabtai I, Leshno M. The impact of EHR and HIE on reducing avoidable admissions: controlling main differential diagnoses. BMC Med Inform Decis Mak 2013; 13: 49.. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Axon RN, Williams MV. Hospital readmission as an accountability measure. JAMA 2011; 3055: 504.. [DOI] [PubMed] [Google Scholar]
- 33. Bisognano M, Boutwell A. Improving transitions to reduce readmissions. Front Health Serv Manage 2009; 253: 3–10. [PubMed] [Google Scholar]
- 34. Agency for Healthcare Research and Quality. HCUP State Inpatient Databases (SID). Healthcare Cost and Utilization Project. 2019. www.hcup-us.ahrq.gov/sidoverview.jsp. Accessed July 2, 2019.
- 35. Charles D, Gabriel M. Adoption of Electronic Health Record Systems among US Non-Federal Acute Care Hospitals: 2008–2014. ONC Data Brief 35. 2015. https://www.healthit.gov/sites/default/files/briefs/2014HospitalAdoptionDataBrief.pdf. Accessed July 02, 2019
- 36. D’Amore JD, Mandel JC, Kreda DA, et al. Are meaningful use stage 2 certified EHRs ready for interoperability? Findings from the SMART C-CDA collaborative. J Am Med Inform Assoc 2014; 21: 1060–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Fickenscher KM. President’s column: interoperability—the 30% solution: from dialog and rhetoric to reality. J Am Med Informatics Assoc 2013; 203: 593–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Harrison TD. Do mergers really reduce costs? Evidence from hospitals. Econ Inq 2011; 494: 1054–69. [DOI] [PubMed] [Google Scholar]
- 39. Dimick C. Open for business: private networks create a marketplace for health information exchange. J AHIMA 2012; 83: 22–6. [PubMed] [Google Scholar]
- 40. Dennard J. Taking HIE beyond the Enterprise and into the Community. Health IT Outcomes 2014. http://www.healthitoutcomes.com/doc/taking-hie-beyond-the-enterprise-and-into-the-community-0001. Accessed July 2 2019.
- 41. Centers for Medicare & Medicaid Services. 2013 Measures Updates and Specifications: Acute Myocardial Infarction, Heart Failure, and Pneumonia 30-Day Risk-Standardized Mortality Measure. 2013. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/Mortality_AMI-HF-PN_Measures_Updates_Report_FINAL_06-13-2013.pdf. Accessed July 2 2019.
- 42. Healthcare Cost and Utilization Project. User Guide: HCUP Supplmental Variables for Revisit Analysis. 2018. https://www.hcup-us.ahrq.gov/toolssoftware/revisit/UserGuide_SuppRevisitFilesCD.pdf. Accessed July 2 2019.
- 43. Rising KL, Victor TW, Hollander JE, et al. Patient returns to the emergency department: the time-to-return curve. Acad Emerg Med 2014; 218: 864–71. [DOI] [PubMed] [Google Scholar]
- 44. Moore BJ, White S, Washington R, et al. Identifying increased risk of readmission and in-hospital mortality using hospital administrative data. Med Care 2017; 557: 698–705. [DOI] [PubMed] [Google Scholar]
- 45. Unruh MA, Trivedi AN, Grabowski DC, et al. Does reducing length of stay increase rehospitalization of medicare fee-for-service beneficiaries discharged to skilled nursing facilities? J Am Geriatr Soc 2013; 619: 1443–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Bazzoli GJ, Shortell SM, Dubbs N, et al. A taxonomy of health networks and systems: bringing order out of chaos. Health Serv Res 1999; 336: 1683–717. [PMC free article] [PubMed] [Google Scholar]
- 47. Carlin CS, Dowd B, Feldman R. Changes in quality of health care delivery after vertical integration. Health Serv Res 2015; 504: 1043–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Scott KW, Orav EJ, Cutler DM, et al. Changes in hospital–physician affiliations in US hospitals and their effect on quality of care. Ann Intern Med 2017; 1661: 1. [DOI] [PubMed] [Google Scholar]
- 49. Correia S. REGHDFE: Stata Module to Perform Linear or Instrumental-Variable Regression Absorbing Any Number of High-Dimensional Fixed Effects. Stat Software Components Published Online First: September 17, 2018. https://ideas.repec.org/c/boc/bocode/s457874.html. Accessed October 17, 2018.
- 50. Vest J, Simon K. Hospitals’ adoption of intra-system information exchange is negatively associated with inter-system information exchange. J Am Med Inform Assoc 2018; 259: 1189–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51. Opoku-Agyeman W, Menachemi N. Are there differences in health information exchange by health system type? Health Care Manage Rev 2016; 414: 325–33. [DOI] [PubMed] [Google Scholar]
- 52. Holmgren AJ, Ford EW. Assessing the impact of health system organizational structure on hospital electronic data sharing. J Am Med Informatics Assoc. 2018;25(9):1147–1152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Adler KG, Edsall RL. EHR switch survey: responses from 305 family physicians. Fam Pract Manag 2015; 221: 13–8. [PubMed] [Google Scholar]
- 54. Javier-DesLoges JF, Ghabili Amirkhiz K, Su JJ, et al. MP80-02 leveraging structured data in an enterprise-wide electronic health record to identify risk factors for postoperative urinary retention. J Urol 2018; 199: e1089. [Google Scholar]
- 55. KLAS. Population Health 2014 Perception: Who are Providers Betting On? 2019. https://klasresearch.com/report/population-health-2014-perception/907. Accessed April 22, 2019.
- 56. Sandberg SF, Erikson C, Owen R, et al. Hennepin health: a safety-net accountable care organization for the expanded medicaid population. Health Aff 2014; 3311: 1975–84. [DOI] [PubMed] [Google Scholar]
- 57. Black Book Market Research. State of the Enterprise Health Information Exchange Industry. Tampa, FL; 2015. [Google Scholar]
- 58. Ahmad FS, Metlay JP, Barg FK, et al. Identifying hospital organizational strategies to reduce readmissions. Am J Med Qual 2013; 284: 278–85. [DOI] [PubMed] [Google Scholar]
- 59. Yanamadala S, Morrison D, Curtin C, et al. Electronic health records and quality of care: an observational study modeling impact on mortality, readmissions, and complications. Medicine (Baltimore) 2016; 95; e3332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60. DesRoches CM, Campbell EG, Vogeli C, et al. Electronic health records’ limited successes suggest more targeted uses. Health Aff 2010; 294: 639–46. [DOI] [PubMed] [Google Scholar]
- 61. Patterson ME, Marken P, Zhong Y, et al. Comprehensive electronic medical record implementation levels not associated with 30-day all-cause readmissions within Medicare beneficiaries with heart failure. Appl Clin Inform 2014; 5: 670–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62. Zlabek JA, Wickus JW, Mathiason MA. Early cost and safety benefits of an inpatient electronic health record. J Am Med Informatics Assoc 2011; 182: 169–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63. Unruh MA, Jung HY, Kaushal R, et al. Hospital participation in meaningful use and racial disparities in readmissions. Am J Manag Care 2018; 24: 38–42. [PMC free article] [PubMed] [Google Scholar]
- 64. Brice YN, Joynt KE, Tompkins CP, et al. Meaningful use and hospital performance on post-acute utilization indicators. Health Serv Res 2018; 532: 803–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65. Silow-Carroll S, Edwards JN, Rodin D. Using electronic health records to improve quality and efficiency: the experiences of leading hospitals. Issue Brief (Commonw Fund) 2012; 17: 1–40. [PubMed] [Google Scholar]
- 66. Hessels A, Flynn L, Cimiotti JP, et al. Impact of heath information technology on the quality of patient care. Online J Nurs Inform 2015; 19. [PMC free article] [PubMed] [Google Scholar]
- 67. Freedman S, Lin H, Prince J. Information technology and patient health: analyzing outcomes, populations, and mechanisms. Am J Heal Econ 2018; 41: 51–79. [Google Scholar]
- 68. Unruh MA, Jung H-YH-Y, Vest JR, et al. Meaningful use of electronic health records by outpatient physicians and readmissions of medicare fee-for-service beneficiaries. Med Care 2017; 555: 493–9. [DOI] [PubMed] [Google Scholar]
- 69. Kaushal R, Edwards A, Kern LM, et al. Association between electronic health records and health care utilization. Appl Clin Inform 2015; 6: 42–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70. Evans RS. Electronic health records: then, now, and in the future. Yearb Med Inform 2016; 25: S48–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71. Office of the National Coordinator for Health Information Technology, The Office of the National Coordinator for Health Information Technology. Connecting Health and Care for the Nation: A Shared Nationwide Interoperability Roadmap 2015. https://www.healthit.gov/sites/default/files/hie-interoperability/nationwide-interoperability-roadmap-final-version-1.0.pdf. Accessed July 2, 2019.
- 72. Kazley A, Diana M, Menachemi N. The agreement and internal consistency of national hospital EMR measures. Health Care Manag Sci 2011; 144: 307–13. [DOI] [PubMed] [Google Scholar]
- 73. Devaraj S, Kohli R. Performance impacts of information technology: is actual usage the missing link. Manage Sci 2003; 493: 273–89. [Google Scholar]
- 74. Hanauer DA, Branford GL, Greenberg G, et al. Two-year longitudinal assessment of physicians’ perceptions after replacement of a longstanding homegrown electronic health record: does a J-curve of satisfaction really exist? J Am Med Inform Assoc 2016;24(e1):e157–e165. doi:10.1093/jamia/ocw077 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 75. Gettinger A, Csatari A. Transitioning from a legacy EHR to a commercial, vendor-supplied, EHR: one academic health system’s experience. Appl Clin Inform 2012; 3: 367–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 76. Healthcare Cost and Utilization Project (HCUP) A for HR and Q. Elixhauser Comorbidity Software, Version 3.7. 2017. https://www.hcup-us.ahrq.gov/toolssoftware/comorbidity/comorbidity.jsp. Accessed May 13, 2019.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.