 This study illustrates a systematic methodology to embed medical costs into the exact flow of clinical events associated with chronic care delivery. We summarized and visualized the results using clinical and cost data, with the goal of empowering patients and care providers with actionable information as they navigate through a multitude of clinical events and medical expenses.

 We analyzed the electronic health records (EHRs) and medication cost data of 288 patients from 2009 to 2011, whose initial diagnoses included chronic kidney disease stage 3, hypertension, and diabetes.

 We developed chronological pathways of care and costs for each patient from EHR and medication cost data. Using a data-driven method called clinical pathway (CP) learning, which leverages statistical machine-learning algorithms, we categorized patients into clinically similar subgroups based on progressing clinical complexity and associated care needs. The CP-based subgroups were compared against cost-based subgroups stratified by quartiles of total medication costs, and visualized via pathways that are color-coded by costs.

 Our methods identified 3 CP-based, and 4 cost-based, patient subgroups. Two sets of subgroups from each approach indicated some clinical similarity in terms of average statistics, such as number of diagnoses and medication needs. However, the CP-based subgroups displayed significant variation in costs; conversely, large differences in clinical needs were observed among cost-based subgroups.

 This study demonstrates that CPs extracted from EHRs can be enhanced with appropriate cost information to potentially provide detailed visibility into the variability and inconsistencies in current best practices for chronic care delivery.

	This preliminary study extracts patterns and variations in medical spending associated with the flow of clinical events in the management of patients with multiple chronic conditions.

		It illustrates a systematic, generalizable methodology for embedding cost information into the clinical pathways of patient cohorts using electronic health record and cost data.

		It exposes variations in costs among clinically similar patients, and variations in clinical complexity among patients with similar costs, by comparing patient cohorts categorized by their clinical pathways and total medical spending, respectively.

		Future extensions will examine appropriate incorporation of data-driven evidence into shared decision making and innovative cost analyses.

Patient care journeys span the entire continuum of clinical and nonclinical experiences. Navigating through a multitude of these events is a mystifying and daunting experience for most patients, and particularly so for patients with multiple chronic conditions (MCCs).

 Despite being the central figure in their scenarios, research indicates that patients hold little power for steering in an appropriate direction through these passages.

 Although hospitals and individual providers recognize the need to guide patients through their treatments, often such guidance follows a top-down approach and does not actively involve patients in the decision-making process.

 However, with personalized and precision medicine gaining increasing attention, and with more patients and providers participating in the value-based payment models that emphasize patient satisfaction,

 healthcare stakeholders are seeking to facilitate shared decision making by incorporating not only treatment efficacy and clinical factors, but also—equally important&mdash;patients’ personal preferences and financial limitations. In fact, multiple research studies have found that involving patients in the decision-making process can lead to higher satisfaction, lower medical costs, and better health outcomes, among other benefits.

 Prior research also shows that patients’ knowledge of pricing information is associated with lower medical costs for common medical services.

With medical cost being such an opaque subject, providers also may not have the best guidance strategy for the treatments that they offer to their patients.

 Given these challenges, this study proposes a new approach to incorporate medical costs explicitly in the chronologically ordered, clinical pathways (CPs) of patient experiences. We especially focused on MCC patients, who are high utilizers of medical services and often participate in innovative payment programs.

 We illustrated this approach using a cost-centered perspective as well as a clinically focused perspective to show alignment in some subgroups and significant variations in others, in the categorization of pathways and patient subgroups under these 2 differing views. The long-term goal of this research is to eventually achieve accurate predictions of anticipated future events and costs following different clinical and cost pathways for improved shared decision making, and, subsequently, identify appropriate ranges of cost for targeted clinical pathways within a patient population.

CPs, commonly referred to as treatment plans that are time- and task-oriented, describe the essential steps associated with the expected clinical course in diagnosing and treating patients.

 The goals in using CPs are to optimize treatment efficiency, patient outcomes, and medical spending.

 CPs are used widely in the United States and are expected to have an even larger influence on healthcare delivery, quality, and patient outcomes, as bundled or episode-based payments gain momentum. Furthermore, using information from CPs in the operational setting has the potential to optimize not only the efficacy of treatments, but also their costs to patients, providers, and society.

Despite these anticipated benefits, to the best of our knowledge, there is a tremendous gap in the resources that will allow clinicians and patients to incorporate best practices for healthcare delivery in the form of CPs. Therefore, we aimed to begin bridging this gap by developing a data-driven approach that would take advantage of the large amount of detailed, patient-level, clinical information captured in electronic health records (EHRs), coupled with cost data that document the medical costs associated with patients’ clinical activities. This preliminary study applies an advanced, systematic, and generalizable methodology to embed medical costs, specifically, costs of medications, into the exact flow of clinical events associated with the complexities of chronic care delivery. Combined pathways of clinical and cost information are summarized and visualized for patient subgroups following common CPs, and compared against simple cost-based segmentation of the patient population.

CP learning refers to an emerging area of data-driven research to mine the common and rare patterns in the co-progression of 1 or more clinical events, depending on the availability of granular data and types of research questions, using statistical machine-learning methods.

 Our previous studies developed and applied these methods to chronic kidney disease (CKD) to identify CPs containing: a) encounters; b) diagnoses as 

International Classification of Diseases, Ninth Revision

) codes; c) procedures as Current Procedural Terminology (CPT) codes; and d) prescriptions of medications summarized by drug classes.

 This study extends these methods that combine statistical machine learning and data visualization by specifically incorporating costs associated with medications. Developing such detailed visibility into the co-progression of treatments and diseases from actual practice data, rather than abstract clinical guidelines, along with the costs associated with the events—medications, in this instance&mdash;is potentially a useful and novel approach to investigate.

We applied 2 approaches in identifying CPs with cost data. The clinically focused approach used our CP-learning algorithm to detect the differences in progression of clinical factors, but not costs, in order to divide patients into clinically similar subgroups. The costs of medications were subsequently computed for these subgroups. The cost-centered approach, which is analytically simpler, divided patients into similar subgroups based solely on the total amount of estimated co-pays on medications. Generally, the clinically focused approach categorizes patients by pathways of care, whereas the cost-centered approach does so based on the total cost of care.

We collaborated with a community nephrology practice in western Pennsylvania that specializes in the management of CKD. CKD patients—mostly an elderly and vulnerable population&mdash;commonly suffer from a number of serious comorbidities and complications as the disease progresses to end-stage renal disease (ESRD).

 Many CKD patients are Medicare beneficiaries, and the United States Renal Data System reports that CKD patients incur nearly twice the cost for their care, on average, compared with the non-CKD Medicare patient.

 In addition, characteristics of the disease have resulted in limited high-quality randomized clinical trials, further exacerbating the challenges in care delivery.

 Hence, improved management of this population using the approach illustrated in this study may generate multiple benefits, such as reduced costs and increased patient satisfaction.

We obtained detailed clinical data, over the time period from 2009 to 2011, from the EHR extracts of the nephrology practice. We identified 288 patients diagnosed with only CKD stage 3, diabetes, and hypertension, and with no other complications, at the beginning of the 2-year period. The gender ratio of female to male was 0.41 to 0.59, and the average age of patients was 73.4 years (standard deviation [SD] = 10). Race represented by the study patients included Caucasian (94.4%), African American (4.8%), and other (0.8%). Over the 2-year period, patients had, on average, 5.5 (SD = 2.1) office visits, 0.4 (SD = 1.0) hospitalizations, and 0.4 (SD = 0.2) education sessions, respectively.

CPs were created based on all records within 2 years since the initial visit, such that we could track the co-progression of clinical events and associated costs within a fixed time interval. Analysis includes the following patient care information: a) encounter types: office visit, hospitalization, education sessions; b) diagnoses as

 codes: CKD stages 1 to 5, ESRD, hypertension, diabetes, acute kidney injury, hyperparathyroidism, anemia, proteinuria, hyperkalemia, acidosis, hyperphosphatemia, glomerulonephritis, urinary obstruction, volume depletion, rhabdomyolysis; c) procedures as CPT codes: renal Doppler and ultrasound; and d) all medications taken by patients, excluding over-the-counter medications.

In the clinically focused approach, medications are summarized by drug classes for the purpose of patient subgrouping. In the cost-centered approach, total costs are calculated for each medication as is. Therefore, it is possible for 2 patients with similar clinical conditions and needs to be categorized into different levels, such as medium for complexity and high for spending, because CP-based subgroups are based on diagnoses and drug classes, whereas cost-based subgroups are based on costs from co-pays of each medication brand. For the sake of simplicity in illustrating our approach, the analysis in this paper examined only the cost of medications in the form of patient co-pays. We manually obtained the estimated co-pay of each medication from GoodRx that provides information on prescription prices under different insurance plans. To do so, we assumed that all patients were: Medicare beneficiaries, under AARP Medicare Complete Plan 1; residing in zip code 15201 of Pittsburgh, Pennsylvania; being prescribed medication dosages at the lowest appropriate level; and purchasing 30-day medication supplies.

The total estimated medication co-pays over 2 years had a wide range, from $0 to over $18,800, with an average of $1032.70 (SD = $2274.0) under the stated assumptions. In the cost-centered approach, we categorized patients into 4 quartile-based subgroups based on spending: high spenders (75% and up), medium spenders (50%-75%), low spenders (25%-50%), and zero spenders (0%-25%). The clinically focused approach detected 3 subgroups using the CP-learning algorithm based on clinical conditions and needs: high complexity, medium complexity, and low complexity. The 

 displays the descriptive statistics of each subgroup, under both clinically focused and cost-centered approaches.

The Table reveals that the 2 medium subgroups—medium spenders and medium complexity&mdash;obtained from the clinical and cost approaches are in fact clinically quite similar. For example, the subgroups had, on average, 6.2 and 5.4 visits, and 5.6 and 5.3 unique diagnoses, but differed in the number of unique drugs (9.7 vs 3.6, respectively); and as an outcome measure, 16.7% and 14.1%, respectively, of the patients progressed beyond CKD stage 3. Even the average cost, $274.6 and $272.5, respectively, differed by just over $2. However, what is strikingly different is the variation in costs between the 2 subgroups. Whereas the coefficient of variation of medium spenders was 0.73, that of the medium complexity group was 7.1&mdash;nearly a 10-fold increase.

displays the overlap in the assignment of patients to subgroups under clinically focused and cost-centered approaches. Although 97% of the high-spending patients were also part of the high-complexity group, 2 patients in the high-spending group—including the patient who spent the most, $18,801.40&mdash;were in fact assigned to the medium complexity group. One potential explanation, which needs to be verified by the clinicians, is that there might have been excessive spending of medical resources given these patients’ clinical needs, which are at the medium-complexity level.

 (parts a and b) displays the large variations in cost among clinically similar patients. Figure 2(b) displays patients’ actual CPs generated for the medium-complexity subgroup. It can be pictured as a map of visits that patients experienced during the 2-year period. Each node in Figure 2 is a unique visit characterized by encounter type, diagnoses noted, and combinations of drug classes prescribed to patients. Starting with the dense, overlapped areas at the left-hand side of Figure 2(b), with all patients diagnosed with CKD stage 3, diabetes, and hypertension, but no other complications, the fanning pathways show the divergence from the common starting point as patients’ disease progresses and complications emerge in diverse ways, for which varying treatments are provided. The size of the nodes and thickness of the edges reflect frequency of visits and transitions in the data. A larger node suggests that this is a common visit that many patients experience, and a thicker edge is an indication that the transition of visits, and accompanying change in clinical conditions or medication prescriptions, are observed among many patients. The color of each visit represents the spending category in quartiles: high (black), medium (dark blue), low (light blue), and zero (green).

There is growing recognition of the need for more precise risk-adjustment strategies, incorporation of evidence-based treatment variability, and increased use of data and information technology to facilitate patient engagement and shared decision making in promoting value-based payment systems.

 In this preliminary study, we aimed to provide a generalizable framework to estimate the costs associated with actual care delivery, and to expose variations in care and cost. In particular, the results from this study showed significant variation in costs among patients who are clinically similar. A deeper analysis of these pathways may uncover the patterns and causes for these variations within subgroups to allow appropriate incorporation of this evidence into the development of future payment models and care delivery practices.

In this study, we analyzed data on CKD patients whose complex, chronic condition is an example of the high-need, high-cost care delivery context that is a continuing challenge to the healthcare system.

 These patients require coordinated care due to their MCCs, and are a key population to be considered in policy design and implementation.

 Our proposed framework specifically targets this patient population by modeling the co-progression of multiple clinical factors, treatments, and medications. We envision that an information technology—enabled tool based on the demonstrated methodology, once developed, deployed, and rigorously evaluated, can be used at the point of care by clinicians and patients to discuss available courses of treatment options, consider their potential efficacy projected at the cohort and personal level, and, equally important, build awareness of the costs associated with the entire course of treatment. Such tools may also provide policy makers and other stakeholders at healthcare practices access to data-driven evidence for innovative cost analyses.

Although the methodology is generalizable to other health conditions and includes many clinical factors, a major limitation of our analysis is the accuracy and availability of data, particularly relevant to our cost-estimation approach. For illustrative purposes, we assigned all patients to a single Medicare plan and manually obtained the estimated co-pays from a website that provides prescription price information. Therefore, some of the drugs selected and prescribed to patients that were found to be excessively expensive may well be due to the potential discrepancy between our assumed insurance plan and patients’ actual plans.

In addition, variations in CPs observed within each subgroup may be explained by unobserved variables, such as social and behavioral factors, as well as the more extensive health information that was not included in the current analysis. For example, the nature of our EHR dataset limits our ability to infer medication adherence; thus, the generated CPs assume perfect medication adherence, which we recognize to be unrealistic. There is also the possibility that the lack of adherence may lead to divergences in the CPs, but the current data fail to capture such associations. Furthermore, patients’ conditions are gauged using 

codes recorded in the EHR; therefore, we did not distinguish patients by the severity of each condition. For instance, use of insulin, whose choice is a marker of severity in diabetes,

 is often observed among the high-spending/complexity subgroup, but our CP-learning algorithm considered all patients with diabetes to have the same severity. Availability of such detailed, relevant data in the future—such as claims data and lab results&mdash;will help to overcome these limitations.

This preliminary analysis shows that patient subgroups generated by the CP-learning methods may be able to expose variations in costs among patients who are clinically similar, and vice versa, thereby facilitating future research to develop improved treatment plans within innovative payment models.

The authors are grateful to the physicians and staff of the community nephrology practice, who generously provided data from their electronic health record for this study. They particularly thank Pradip Teredesai, MD; Nirav Patel, MD; Qizhi Xie, MD, PhD; and staff members Linda Smith and Audra Barletta, who also gave important clinical and technical information about the data and the key characteristics of chronic kidney disease and its treatments.

 Division of Health Informatics, Department of Healthcare Policy & Research, Weill Cornell Medicine of Cornell University (YZ), New York, NY; The H. John Heinz III College, Carnegie Mellon University (RP), Pittsburgh, PA.

 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.

 Concept and design (YZ, RP); acquisition of data (YZ, RP); analysis and interpretation of data (YZ, RP); drafting of the manuscript (YZ); critical revision of the manuscript for important intellectual content (YZ, RP); statistical analysis (YZ, RP).

 Yiye Zhang, PhD, Assistant Professor, Division of Health Informatics, Department of Healthcare Policy & Research, Weill Cornell Medicine of Cornell University, 425 E 61st St, Ste 301, New York, NY 10065. E-mail: yiz2014@med.cornell.edu.

1. Rein A. Navigating health care: why it’s so hard and what can be done to make it easier for the average consumer. Changes in Health Care Financing and Organization website. http://www.hcfo.org/files/hcfo/HCFONavigatingHealthCare.pdf. Published 2007. Accessed November 1, 2016.

2. Alston C, Paget L, Halvorson G, et al. Communicating with patients on health care evidence. https://nam.edu/wp-content/uploads/2015/06/VSRT-Evidence.pdf. Published September 2012. Accessed November 1, 2016.

3. Jolly SE, Navaneethan SD, Schold JD, et al. Development of a chronic kidney disease patient navigator program. 

2015;16:69. doi: 10.1186/s12882-015-0060-2.

4. Blumenthal D, Abrams MK. Tailoring complex care management for high-need, high-cost patients. 

. 2016;316(16):1657-1658. doi: 10.1001/jama.2016.12388.

Articles

METHODS

The authors illustrate a methodology for delineating variations in medical costs for patients with similar clinical conditions and needs using electronic health record data.

Data-Driven Clinical and Cost Pathways for Chronic Care Delivery

 The aim of this study was to evaluate health information technology (IT) adoption in hospitals participating in accountable care organizations (ACOs) and compare this adoption to non-ACO hospitals.

 A cross-sectional sample of US nonfederal, acute care hospitals with data from 3 matched sources:

the 2013 American Hospital Association (AHA) Annual Survey, the 2013 AHA Survey of Care Systems and Payments (CSP), and the 2014 AHA Information Technology Supplement.

 To compare health IT adoption in ACO- and non-ACO hospitals, we created measures of Meaningful Use (MU) Stage 1 and Stage 2 core and menu criteria, patient engagement—oriented health IT, and health information exchange (HIE) participation. Adoption was compared using both naïve and multivariate logit models.

 Of the 393 ACO hospitals and 810 non-ACO hospitals, a greater percentage of ACO hospitals were capable of meeting MU Stage 1 (50.9% vs 41.6%; 

<.05), having patient engagement health IT (39.8% vs 15.2%; 

<.001), and participating in HIE (49.0% vs 30.1%; 

<.001). In adjusted models, no difference was found between ACO and non-ACO hospital ability to meet MU Stage 1 or Stage 2, but ACO hospitals were more likely to have patient engagement health IT (odds ratio (OR), 2.20; 95% CI, 1.59-3.04) and be HIE participants (OR, 1.41; 95% CI, 1.03-1.92).

 ACO-participating hospitals appear to be focused more on adopting health IT that aligns with broader strategic goals rather than those that achieve MU. Aligning adoption with quality and payment reform may be a productive path forward to encourage hospital health IT adoption behavior.  

		Accountable care organization (ACO) hospitals appear to be aligning their health information technology (IT) adoption with strategic goals of quality improvement and cost reduction. 

		ACO incentives may outweigh Meaningful Use (MU) incentives, pushing hospitals to adopt different technologies, such as patient engagement- and exchange-oriented health IT. 

		Hospitals may consider adopting technologies that align with their strategic goals to engage patients, improve quality of care, and reduce costs. 

		Policy makers may consider aligning MU with quality improvement and alternative payment models for hospitals, as has been done for eligible providers.

The passage of the Affordable Care Act in 2010 authorized CMS to disburse incentive payments to accountable care organizations (ACOs) comprised of providers who voluntarily agree to care for Medicare fee-for-service beneficiaries.

 CMS developed 2 ACO program types—the Shared Savings Program and the Pioneer Program&mdash;which financially reward ACO-participating providers that surpass quality and cost benchmarks established for predefined populations based on the level of risk assumption.

 The ACO model has grown in popularity and has been extended to both private-sector payers and Medicaid.

 This expansion has resulted in over 700 ACOs covering nearly 24 million lives, demonstrating modest results in controlling costs and improving quality as shown in early analyses.

Concurrent with the growth of ACOs, the Meaningful Use (MU) policy program that provides financial incentives for eligible hospitals and providers to adopt and use health information technology (IT), and penalizes nonadoption and nonuse, has been credited with the rapid growth in the adoption and use of health IT.

 The phased integration of this program has largely dictated the specific health IT components that providers have adopted. In recognition of both success and challenges with the program, CMS recently announced that the MU program for eligible providers will transition to become part of the newly implemented Merit-based Incentive Program System (MIPS) that aims to formally align MU with quality improvement and payment reform initiatives.

 However, the MU program for hospitals will continue, as is, for the foreseeable future, despite changes being anticipated.

Although ACO incentives are not directly tied to health IT adoption, many experts believe that providers participating in ACOs must possess robust health IT to support care coordination mechanisms necessary to reach cost and quality goals.

 To better understand how ACOs are using and developing health IT, Wu et al conducted a mixed-methods study and found wide variation in ACO health IT capabilities.

 Other recent work highlights the role that health IT can play in helping to facilitate care coordination by assisting with decision support.

 Moreover, health IT can be linked across organizational boundaries through interoperable platforms, including health information exchanges (HIEs).

 This exchange of information offers providers access to comprehensive health information at the point of care.  Importantly, HIEs also can help facilitate population-level health assessments—critical for ACOs that are responsible for the care of a defined patient population.

 ACOs may also offer patient portals and other patient-facing health IT components to improve patient engagement and better involve patients in the care process, working toward the goal of reducing costs and improving quality.

In the context of these 2 significant policy initiatives—ACOs and MU&mdash;the extent to which ACO health IT strategies and MU criteria are aligned in the marketplace remains unknown. Additionally, questions remain regarding the specific health IT elements ACO provider organizations possess that address care coordination challenges relative to non-ACO providers. This study aims to: 1) characterize the health IT capabilities of ACO provider organizations, and 2) compare these organizations with non-ACO participants in an effort to understand how health IT capacity differs between the 2 in regards to MU achievement, patient engagement, and HIE. These findings can contribute to the ongoing discussions around how to align alternative payment models that support quality improvements with MU policy. 

This study used a cross-sectional design with data from 3 matched sources covering the same sampling frame: the 2013 American Hospital Association (AHA) Annual Survey (organizational and service characteristics), the 2013 AHA Survey of Care Systems and Payment (CSP) (ACO participation), and the 2014 AHA Information Technology (IT) Supplement (health IT adoption and use). The 2014 AHA IT Supplement is administered to respondents of the 2013 AHA Annual Survey, and represents hospital IT capabilities as of fall 2013. The 2013 CSP Survey, fielded in early 2013 to respondents of the 2012 AHA Annual Survey, and is used to determine hospital ACO participation status. Although the CSP Survey precedes the IT Supplement Survey, the lengthy contracts and significant amount of resources required to develop an ACO make it unlikely that these hospitals would change their ACO participation status within the timeframe between the CSP and IT surveys. Only nonfederal, acute care general hospitals within the United States were included in the analytic sample. 

ACO participation was determined from the CSP Survey question asking hospitals if their “Hospital established a separate legal entity for an accountable care organization (ACO) with the goal of being able to accept contracts to provide healthcare for a defined population?” Responses of, “Yes, my hospital has established an ACO” and “Yes, my hospital is part of an ACO” were combined into an ACO participant category (“ACO hospital”) and all other hospitals were categorized as nonparticipants (“non-ACO hospital”). The survey did not distinguish among Medicare, private sector, or Medicaid ACOs. 

Following previous work, we mapped survey items from the IT Supplement to each of the 16 MU Stage 1 and 2 core measures (eAppendix [

 allowing for the assessment of each MU item and aggregate measures of MU Stage 1 and 2. We additionally mapped Survey items to the 6 menu items in MU Stage 2 (eAppendix Table A). We also evaluated hospital health IT activity in 2 different domains particularly relevant to ACO development: patient engagement and health information exchange (HIE).

Two MU Stage 2 core criteria are relevant to patient engagement: the ability for patients to view, download, and transmit their health record; and electronic access to patient-specific education. However, the AHA IT supplement contains 5 additional items concerning health IT and patient interaction: request change to health record online, schedule an appointment online, submit patient-generated data, request a prescription refill, and secure messaging. These patient-facing components are particularly relevant for ACOs that rely on patient retention.

 The responses for the 7 components (2 MU Stage 2 criteria and 5 patient interaction questions) were summed to calculate an aggregate measure of patient engagement health IT capability. Hospitals utilizing greater than half of patient engagement components (≥4 components) were considered to possess patient engagement—oriented health IT. We conducted sensitivity analysis around this cut-off to ensure that this number of patient engagement health IT components was not too restrictive. We found a sharp drop in the number of hospitals with at least 2 or more capabilities, but found relative stability between 2 and 6 capabilities; only a small percentage of hospitals had all 7 capabilities (see eAppendix Table B). Thus, we decided that the threshold (≥4 components) was appropriate.

Given that the ACO model requires hospitals to exchange health information with providers potentially outside of their organization, it remains relevant to assess exchange capacity beyond the ability to transmit clinical care documents, required as one of the MU Stage 2 core criteria. Therefore, hospitals that noted they are “participating and actively exchanging data in at least 1 HIE/RHIO [regional health information organization]” in response to the AHA IT Survey question asking hospitals to “Please indicate your level of participation in a regional health information exchange (HIE) or regional health information organization (RHIO,)” were included.

Previous work has identified differences between respondents and nonrespondents of both the AHA IT Supplement and the CSP Survey compared with the AHA Annual Survey.

 To reduce this response bias, we applied an inverse probability weighting approach that regresses hospital characteristics (bed size, ownership type, system status, teaching status, region, urban/rural location) on whether the hospital responded to all 3 surveys.

  Predicted probabilities yielded from this model are applied as weights in all analyses. 

Using the weighting approach, we compared the aforementioned hospital characteristics of ACO hospitals and non-ACO hospitals using 

 tests to compare naïve adoption rates between ACO and non-ACO hospitals of the aggregate MU Stage 1 and Stage 2 measures, each core component, the menu items, the additional patient engagement and exchange indicators, and the aggregate patient engagement measure. We additionally evaluated separate multivariate logistic regression models to assess the effect of ACO participation on 3 indicators of health IT adoption: the ability to meet MU Stage 2 (core measures only), patient engagement health IT, and HIE participation. We chose to evaluate the stricter MU Stage 2 set of criteria rather than Stage 1 due to its focus on use of health IT rather than adoption (eAppendix Table C). These models included the set of hospital-level control variables listed above.

Inverse probability weighted hospital characteristics are shown by hospital ACO participation status in 

. Overall, the analytic sample included 1203 US nonfederal acute care hospitals that responded to all 3 surveys. Of these hospitals, weighted results revealed that 32% (393) of these hospitals were ACO participants. ACO participants were more likely to be large, nonprofit, teaching, and system members, located in urban areas and in the northeast. 

Weighted unadjusted comparisons of aggregate measures of health IT adoption and adoption of specific health IT components between ACO and non-ACO hospital participants are presented in 

. A greater proportion of ACO hospitals are capable of meeting MU Stage 1 and Stage 2, although only 7.6% of ACO hospitals and 4.8% of non-ACO hospitals could meet Stage 2. A greater proportion of ACO hospitals adopted 11 of the 16 core criteria, while no difference was found in the remaining 5 criteria. Similarly, a greater proportion of ACO hospitals had adopted 5 of the 6 menu items, and no difference was found with the sixth item. For the additional patient engagement activities, a greater proportion of ACO hospitals had adopted each of the 5 capabilities. Lastly, a greater proportion of ACO hospitals were participating in an HIE (49.0 vs 30.1; 

For the MU Stage 2 aggregate measures, the aggregate measure of patient engagement capability and HIE participation, we conducted multivariate analyses to determine the adjusted effect of ACO participation on health IT adoption (

). Adjusted analyses revealed no difference between ACO and non-ACO hospital ability to meet MU Stage 2. Alternatively, ACO hospitals were more likely to be participating in an HIE than non-ACO hospitals (odds ratio [OR], 1.41; 95% CI, 1.03-1.92). Our aggregate measure of patient engagement capability included hospitals that had at least 4 of the 7 functions that involved patient activity. ACO hospitals were significantly more likely than non-ACO hospitals to achieve this level of patient engagement health IT adoption and possess robust patient-oriented health IT (OR, 2.20; 95% CI, 1.59-3.04). 

In reporting 2014 results for the Shared Savings and Pioneer ACO programs, CMS stated that the programs had generated greater than $411 million in total savings, while ACO provider organizations outpaced quality benchmarks since program inception.

 Among the measures demonstrating significant year-to-year improvement in that report was the percentage of ACO participating providers that qualified for an electronic health record (EHR) incentive payment (77% in 2013 increased to 81% in 2014). Despite this evidence that ACO providers continue to invest in health IT, little is known about the specific health IT capabilities of ACO providers and how they differ from non-ACO providers. 

Our results suggest that, in a simple comparison, ACO hospitals are significantly more likely than non-ACO hospitals to be able to attest to MU Stage 1 and Stage 2. Interestingly, despite the percentage differences, we found no difference in ability to meet MU Stage 1 or 2 when accounting for hospital characteristics. Our findings of significant differences between ACO and non-ACO hospitals in HIE and patient engagement health IT in adjusted models suggest that ACO hospitals may be looking beyond incentives for health IT adoption to guide their investment choices, and these hospitals may be circumventing MU Stage 2 requirements to pursue a strategy of aligning health IT investments with ACO goals.

As an example of this alignment, consider the need for ACOs to conduct population health management. Health IT is a critical asset toward this strategy; it can enable tracking of patients across healthcare settings, as well as population-level analytics. The greater participation in an HIE by ACO hospitals, as seen in our study, may demonstrate ACO reliance on greater shared information for care management processes,

 including limiting duplication of services, predicting inefficient utilization, and monitoring patient cost.  However, these choices may result in intriguing dynamics in the HIE landscape. ACOs may be preferential to closed HIE networks, often called Enterprise HIEs, that limit participation and sharing of clinical data to providers within a network or health system, such as an ACO.

 The alternative HIE model, the community HIE, may offer greater potential in terms of public health benefits, yet the business case may be less clear for ACOs. Future research may focus on the competitive dynamics playing out as a result of the relationship between ACO networks and interoperability initiatives. 

Similarly, the growth of ACOs may be driving the push toward technology that empowers patients. Consumers demand the ease with which technology can facilitate access to their records and healthcare-related information. As a result, patient-facing technologies may be requisite for ACOs that need to retain patients that may not know they are part of an ACO.

 Leadership at ACO hospitals may see investment in tools that facilitate better communication with patients as a key tactic to enhance quality, increase patient retention, reduce costs, and improve outcomes.

 This indication of stronger patient orientation suggests that ACO hospitals have chosen to pursue a technology strategy that closely aligns with ACO metrics. Although investing in patient-facing health IT and population health tools may be a logical investment for ACO hospitals, non-ACO hospitals may only have the bandwidth to focus on incentives for MU adoption, as evidenced by similar results between ACO and non-ACO hospitals in our regression models.

This study faces important limitations. First, the combination of the AHA IT Supplement and the CSP Survey limits our sample, and these hospitals differ from the population of acute care hospitals in the United States. Although we do apply inverse probability weighting to address this bias, the bias cannot be eliminated. Second, our study is limited to a cross-sectional view of ACO and non-ACO hospital health IT status. The ACO model is rapidly spreading across the healthcare landscape, as is health IT adoption and use. Our use of cross-sectional data allows us to determine a baseline of health IT adoption and to better glimpse the way that ACO hospitals may be orienting their technological capability compared with non-ACO hospitals.  However, the causal direction of this relationship remains unknown, and is a ripe area for future research.

Policy makers have recently taken steps to more tightly align policy aims and health IT adoption through programs such as the Alternative Payment Models and MIPS for eligible providers. Although this alignment does not yet exist for hospitals, our findings that ACO hospitals are already adopting and exchange-oriented health IT suggests that ACO incentives may be a better driver of care change than MU incentives. As a result, alignment of MU goals with the quality and cost goals of ACOs may be a productive policy path forward pertaining to hospitals.

 Department of Family Medicine, College of Medicine, Ohio State University (DMW, ASM), Columbus, OH; Global Health Management and Policy, Tulane University School of Public Health and Tropical Medicine (AMM), New Orleans, LA.

 Concept and design (AMM, DMW); acquisition of data (DMW); analysis and interpretation of data (AMM, ASM, DMW); drafting of the manuscript (AMM, ASM, DMW); critical revision of the manuscript for important intellectual content (AMM, ASM, DMW); statistical analysis (DMW).

 Daniel M. Walker, PhD, MPH, Department of Family Medicine College of Medicine, Ohio State University, 273 Northwood-High Building, 2231 North High St, Columbus, OH 43201. E-mail: Daniel.Walker@osumc.edu.

1. CMS; HHS. Medicare program; Medicare Shared Savings Program: accountable care organizations. final rule. 

2. Casalino LP, Erb N, Joshi MS, Shortell SM. Accountable care organizations and population health organizations. 

. 2015;40(4):821-837. doi: 10.1215/03616878-3150074.

3. Lee TH, Casalino LP, Fisher ES, Wilensky GR. Creating accountable care organizations. 

. 2010;363(15):e23. doi: 10.1056/NEJMp1009040.

4. Muhlestein D. Continued growth of public and private accountable care organizations. Health Affairs website. http://healthaffairs.org/blog/2013/02/19/continued-growth-of-public-and-private-accountable-care-organizations/. Published February 19, 2013. Accessed October 2016.

5. Fulton BD, Pegany V, Keolanui B, Scheffler RM. Growth of accountable care organizations in California: number, characteristics, and state regulation. 

. 2015;40(4):669-688. doi: 10.1215/03616878-3149988.

POLICY

Hospitals participating in accountable care organizations (ACOs) have greater adoption of health information technology, particularly patient-facing technology and health information exchange, than non-ACO hospitals.

Accountable Care Organization Hospitals Differ in Health IT Capabilities

The United States’ decade-long transition from a paper- to technology-based information infrastructure has always been recognized as an initial step—a laying of the foundation&mdash;for future changes to the delivery of care. An increasingly important focal area for improvement is population health. Numerous policies and programs now require healthcare organizations to manage the risks, outcomes, utilization, and health of entire groups of individuals. Nonetheless, current health information technology (IT) systems are not ready to support population health improvements effectively and efficiently. Existing health IT systems were designed for organizations that are structurally, operationally, and culturally focused on individual care delivery, rather than improving health for a population. Opportunities exist to align health IT resources and population health management strategies to fill the gaps among technological capabilities, use and the emerging demands of population health. To realize this alignment, healthcare leaders must think differently about the types of data their organizations need, the types of partners with whom they share information, and how they can leverage new information and partnerships for evidence-based action.

Existing health information technology systems were designed for individual care delivery rather than population health.

Healthcare organizations need to expand data collection efforts beyond “sick care” information to information on actual health behaviors and social determinants.

Health information exchange can better support social determinants of health by including new partners, like social service organizations and public health agencies.

New data sources and information-sharing partners may lead to more predictive models.

Incorporating Meaningful Use (MU) into the Medicare Access & CHIP Reauthorization Act (MACRA) of 2015’s Quality Payment Program signals that healthcare organizations are expected to use health information technology (IT) to improve care. One focus of these improvements is population health, which requires managing the risks, outcomes, utilization, and health of entire groups of individuals. For example, both of the Quality Payment Program’s pathways—the Merit-Based Incentive Payment System (MIPS) and Alternative Payment Models (APM)&mdash;include population health in calculating payments. MIPS requires population health quality measures and population health—based clinical practice-improvement activities.

 APMs promote organizational accountability beyond the individual patient encounter and are expected to reduce utilization through improved health.

The changes wrought by MACRA are just one indication that population health increasingly matters. The CMS Hospital Readmission Reduction Program penalizes hospitals for excess readmission rates.

 Nonprofit hospitals and health systems now must assess and adopt strategies to address community health needs to retain their tax exempt status.

 Moreover, influential projects like the “Culture of Health” initiative of the Robert Wood Johnson Foundation and “Health in All Policies”—which is promoted by numerous public health entities&mdash; foster an environment that emphasizes population health over the provision of medical care alone.

The decades-long transition from a paper- to a technology-based information infrastructure in the United States has always been recognized as an initial step, laying a foundation for fundamental care delivery changes. Nonetheless, current health IT systems are not ready to support population health improvements effectively and efficiently. Existing health IT systems were designed for organizations that are structurally, operationally, and culturally focused on individual care delivery rather than improving health for groups of people.

 For example, electronic health records (EHRs)—primarily designed as clinical documentation tools&mdash;often lack sophisticated risk stratification and targeted case-management functionalities.

 Even when possible, healthcare organizations are not widely deploying health IT systems in ways that support population health.

Opportunities exist to align health IT resources and population health management strategies to fill the gaps among technological capabilities, use, and the emerging demands of population health. To realize this alignment, healthcare leaders must think differently about the types of data their organizations need, with whom they share data, and how they can leverage these data for evidence-based action.

Supporting decision making for population health requires expanding the fundamental types of data collected, integrated, analyzed, and used. Today, healthcare organizations collect what can be labeled “sick care” information (eg, diagnoses, encounters, symptoms, medications, procedures, tests). These data only present a “keyhole view” of a patient’s overall health; population health also requires “nonhealth” data about a population. Yet, notably absent in most current health IT systems are comprehensive data on health, behaviors, and social determinants, which are more significant drivers of health than medical care service utilization.

 For example, socioeconomic status, housing stability, nutrition habits, language, and location are all associated with poor postdischarge outcomes.

 Healthcare organizations traditionally collect and/or use these data in day-to-day operations only to a limited degree, however.

 Thus, healthcare organizations must expand data collection and usage to better understand how health within populations is created and maintained.

The good news is that many of these data exist and are available for when healthcare organizations choose to embrace them. Society is in the midst of a health data revolution. Estimates suggest that 1 in 5 adults uses a wearable fitness device,

 and similar numbers are actively measuring their own health indicators electronically.

 Beyond exercise, sleep, and physical activity monitors, retail loyalty programs generate information on purchasing behaviors, which provide objective measures of nutrition, medications, and even tobacco usage. Opportunities to leverage these information sources for educational, motivational, and communication functionality in health consumer IT abound. Beyond data about individuals, an unprecedented amount of information is available on the environment, public safety, traffic, neighborhood context, and public health.

 The opportunity to move from measuring how sick to how healthy populations are is within easy reach.

In the pre-MU era, health IT and EHRs were data silos and repositories of information that could not easily be shared between care providers. Health information exchange (HIE) was developed to share critical patient information. To date, however, HIEs have minimally supported population health initiatives. For one, social service organizations and public health agencies are rarely partners in an HIE.

 Population health requires collaboration, partnership, and cooperation with social service organizations and public health agencies because healthcare organizations lack the services, programs, or expertise to address many of the determinants of health. Likewise, many healthcare organizations’ HIE activities are narrow in scope. Factors such as limited participation in community-based HIE organizations, the growing use of enterprise HIEs, or single-vendor mediated EHR strategies limit the widespread availability of patient data in a given market. Such strategies make it difficult—if not impossible&mdash;to assemble comprehensive patient histories, aggregate data for population health, and coordinate care.

 In addition, even when available through HIEs, information from external providers is rarely integrated into clinical systems,

 resulting in limited ability to leverage exchanged information for clinical decision making.

Avenues to expand organizational participation in HIEs exist. For example, healthcare organizations can identify social service partners (eg, 2-1-1 listed programs) and assist them in connecting to an HIE network or obtaining direct secure messaging accounts. Such arrangements would facilitate patient transitions to service providers capable of addressing a broader set of health determinants. Also, these arrangements could facilitate communication from the social service organizations and supplant the need for healthcare organizations to directly capture social determinants in health IT systems at the point of care.

Similarly, simply considering who has the ability to act upon HIE information could suggest new priority partners. For example, emerging HIE event notification systems alert providers about key patient events like hospitalizations. Although small medical practices may not have the capacity to respond to these events,

 case management and home health agencies have the expertise and staff capable of coordinating care in response. Additionally, partnerships with public health agencies, which often maintain data on geographic populations, could provide access to data on social, behavioral, and environmental factors currently absent from health IT systems.

Lastly, HIE will be most effective if data-sharing partnerships accurately reflect patients’ care patterns within the community. This may require healthcare organizations to consider HIE needs beyond a single vendor and novel approaches for engaging new partners (and even competitors) to the mutual benefit of a given population.

Translating Data to Actionable Information

Today, healthcare organizations typically estimate risk using only clinical and care utilization measures, even though social, behavioral, and environmental factors are also relevant. Thus, prediction models, such as those for hospital readmission, often perform poorly.

Armed with expanded and widely shared data reflecting the contexts and behaviors that influence health, the next step is to transform these data into actionable information to achieve population health goals. A clear application is to augment current risk stratification approaches, which attempt to divide populations into groups for targeted interventions. New data sources and information-sharing partners may lead to better-performing models, and they may allow us to characterize and predict more population health—relevant outcomes. For example, many existing risk models predict outcomes like death, care costs, or care utilization. Although these are important, they are distal from the basic goals of population health. Instead, data on health behaviors and other social determinants may be a means to predict upstream factors, such as physical function and quality of life, which are more relevant to population health goals of widespread physical, mental, and social well-being.

 Finally, this new information must be put in front of the users and shared with partner organizations so they can take action.

A shift in how healthcare leaders think about data collection, data sharing, and translating data into actionable information is neither insurmountable nor technologically difficult. The capabilities exist to collect, integrate, and analyze large bodies of data relevant to human health, including social, behavioral, public health, and environmental factors. Instead, as healthcare organizations establish population health goals, leaders must ensure their organizations’ data collection and analytic capabilities align with their changing business needs. As organizations become accountable for population health, their leaders will need to initiate collaborations and agreements with nontraditional partners to obtain, share, and use social indicators and service information in order to optimally leverage health IT resources in pursuit of enhanced healthcare and population health. 

 Richard M. Fairbanks School of Public Health (JRV, CAH, BED, PH, NM), and School of Medicine (TS, SJG), and Center for Health Services and Outcomes Research (BED), Indiana University, Indianapolis, IN; Center for Biomedical Informatics, Regenstrief Institute, Inc (JRV, CAH, TS, BED, SJG, NM), Indianapolis, IN; Center for Health Information and Communication (BED), Department of Veterans Affairs, Indianapolis, IN.

 Support for this publication was provided by the Robert Wood Johnson Foundation through the Systems for Action National Coordinating Center, ID 73485.

 Dr Schleyer is an employee of the Indiana University School of Medicine, which engages in research and design of health information exchanges (HIEs) through his appointment at the Regenstrief Institute. As a part of his full-time position, he participates in a wide variety of HIE and health IT—related projects. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (BED, SJG, CAH, PKH, NM, TS, JRV); drafting of the manuscript (BED, SJG, CAH, TS, JRV); critical revision of the manuscript for important intellectual content (BED, SJG, CAH, NM, TS, JRV); administrative, technical, or logistic support (BED, CAH, PKH); and supervision (PKH, NM).

 Joshua R. Vest, PhD, MPH, Richard M. Fairbanks School of Public Health at Indiana University—Purdue University Indianapolis, 1050 Wishard Blvd, Rm 5124, Indianapolis, IN 46202-2872. E-mail: joshvest@iu.edu.

1. The Merit-Based Incentive Payment Program System (MIPS). CMS website. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/Value-Based-Programs/MACRA-MIPS-and-APMs/Quality-Payment-Program-MIPS-NPRM-Slides.pdf. Accessed July 6, 2016.

2. 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. Federal Register website. https://www.federalregister.gov/documents/2016/05/09/2016-10032/medicare-program-merit-based-incentive-payment-system-mips-and-alternative-payment-model-apm. Published May 9, 2016. Accessed November 2016. 

3. Kocher RP, Adashi EY. Hospital readmissions and the Affordable Care Act: paying for coordinated quality care. 

. 2011;306(16):1794-1795. doi: 10.1001/jama.2011.1561.

4. Additional requirements for charitable hospitals; community health needs assessments for charitable hospitals; requirement of a section 4959 excise tax return and time for filing the return. Internal Revenue Service website. https://www.irs.gov/irb/2015-5_IRB/ar08.html. Published February 2, 2015. Accessed June 16, 2016.

5. Gase LN, Pennotti R, Smith KD. “Health in all policies”: taking stock of emerging practices to incorporate health in decision making in the United States. 

. 2013;19(6):529-540. doi: 10.1097/PHH.0b013e3182980c6e.

6. Lavizzo-Mourey R. Building a culture of health. Robert Wood Johnson Foundation website. http://www.rwjf.org/content/dam/files/rwjf-web-files/Annual_Message/2014_RWJF_AnnualMessage_final.pdf. Published 2014. Accessed November 2016.

7. McDonald CJ, Overhage JM, Tierney WM, et al. The Regenstrief Medical Record System: a quarter century experience.

8. Richardson JE, Vest JR, Green CM, Kern LM, Kaushal R; HITEC Investigators. A needs assessment of health information technology for improving care coordination in three leading patient-centered medical homes. 

. 2015;22(4):815-820. doi: 10.1093/jamia/ocu039.

9. Sheikh A, Sood HS, Bates DW. Leveraging health information technology to achieve the “triple aim” of healthcare reform. 

. 2015;22(4):849-856. doi: 10.1093/jamia/ocv022.

10. Wu FM, Rundall TG, Shortell SM, Bloom JR. Using health information technology to manage a patient population in accountable care organizations. 

. 2016;30(4):581-596. doi: 10.1108/JHOM-01-2015-0003.

11. Woolf SH, Braveman P. Where health disparities begin: the role of social and economic determinants—and why current policies may make matters worse. 

. 2011;30(10):1852-1859. doi: 10.1377/hlthaff.2011.0685.

12. Calvillo-King L, Arnold D, Eubank KJ, Lo M, Yunyongying P, Stieglitz H, Halm EA. Impact of social factors on risk of readmission or mortality in pneumonia and heart failure: systematic review. 

. 2013;28(2):269-282. doi: 10.1007/s11606-012-2235-x.

13. Rosen AK, Reid R, Broemeling AM, Rakovski CC. Applying a risk-adjustment framework to primary care: can we improve on existing measures? 

14. Ask J. Where would the wearables market be without smartphones? Forrester website. http://blogs.forrester.com/julie_ask/15-09-28-where_would_the_wearables_market_be_without_smartphones. Published September 28, 2015. Accessed June 20, 2016.

15. Comstock J. Survey: 40 percent of doctors say they have recommended wearables, only 4 percent of patients say they have. Mobihealthnews website. http://www.mobihealthnews.com/content/survey-40-percent-doctors-say-they-have-recommended-wearables-only-4-percent-patients-say. Published June 20, 2016. Accessed June 20, 2016.

16. Comer KF, Grannis S, Dixon BE, Bodenhamer DJ, Wiehe SE. Incorporating geospatial capacity within clinical data systems to address social determinants of health. 

17. Nguyen OK, Chan CV, Makam A, Stieglitz H, Amarasingham R. Envisioning a social-health information exchange as a platform to support a patient-centered medical neighborhood: a feasibility study. 

. 2015;30(1):60-67. doi: 10.1007/s11606-014-2969-8.

Aligning health information technology with population health requires organizations to think differently about data needs, exchange partners, and how to leverage both for evidence-based action.

December 2016

Data-Driven Clinical and Cost Pathways for Chronic Care Delivery

Accountable Care Organization Hospitals Differ in Health IT Capabilities

Getting From Here to There: Health IT Needs for Population Health

Building Health IT Capacity to Improve HIV Infection Health Outcomes

How Health Plans Promote Health IT to Improve Behavioral Health Care

The Health Information Technology Special Issue: Current Trends and Future Directions

US Hospital Engagement in Core Domains of Interoperability

Payer—Provider Patient Registry Utilized in a Behavioral Health Home

Assessing Electronic Health Record Implementation Challenges Using Item Response Theory

Telemedicine and the Sharing Economy: The "Uber" for Healthcare