Katelyn de Lisle, BS

This paper analyzes whether hospital participation in an accountable care organization (ACO) impacts a hospital’s quality improvement and cost reduction outcomes in other value-based purchasing (VBP) programs, including the Hospital Value-Based Purchasing Program (HVBP), the Hospital Readmissions Reduction Program (HRRP), and the Hospital-Acquired Conditions (HAC) Reduction Program. 

Using VBP performance data and Leavitt Partners’ ACO data, 2 analyses were performed: 1) a descriptive comparison of VBP performance of hospital ACOs compared with non-ACO hospitals, and 2) a longitudinal analysis of hospitals that became part of an ACO during the second year of performance data. 

In the descriptive analysis, we compared VBP scores for hospital ACOs with non-ACO hospitals. To estimate the effect that becoming an ACO had on a hospital, we evaluated the performance of hospitals that became part of an ACO to all hospitals that never became part of an ACO.

For fiscal year 2016, hospital ACOs performed better than non-ACO hospitals for the HRRP, but not on the HVBP and the HAC Reduction Programs. Longitudinal analysis, however, reveals that results are varied, with evidence that hospitals joining ACOs did increasingly better than their peers for the HRRP, but had inconsistent results year-over-year with the HVBP. 

Despite similar goals, hospital participation in an ACO is not correlated with improved performance in all Medicare VBP programs. Organizations pursuing accountable care and also attempting to maximize Medicare VBP program performance must recognize the differences in program objectives and create strategies unique to each.

	Critiquing the effectiveness and overlap of value-based purchasing (VBP) programs is essential for shaping health reform. This paper analyzes whether hospital participation in an accountable care organization (ACO) impacts a hospital’s quality and cost improvement outcomes in other Medicare VBP programs. The following learnings can influence practice and policy decisions: 

		Despite similar goals, variation exists in hospital ACO participation and VBP program outcomes. Organizations must build strategies unique to the differing program objectives. 

		High ACO hospital readmissions reduction scores may signify that ACO-participating hospitals will focus priorities on measures that overlap between programs. 

		Scoring methodology changes may contribute to lack of correlation between program performance.

Since the passage of the Affordable Care Act (ACA), CMS has undertaken several initiatives to encourage providers to assume more financial and clinical responsibility for the care of the Medicare patients they serve. Although different in implementation, many of these initiatives overlap in terms of desired outcomes. Broadly speaking, accountable care organizations (ACOs) and value-based purchasing (VBP) programs seek to drive healthcare quality improvements and cost reductions. This paper analyzes whether hospital participation in an ACO program, Medicare, or otherwise correlates with a hospital’s outcomes in CMS-implemented value-based payment programs, including the Hospital Value-Based Purchasing Program (HVBP), the Hospital Readmissions Reductions Program (HRRP), and the Hospital-Acquired Conditions (HAC) Reduction Program.

ACOs are provider-led healthcare entities that bear responsibility for the financial and clinical outcomes of a patient population by coordinating care across providers.

 Although all ACOs strive to reduce costs and improve outcomes, the market is varied in terms of ACO organizational structure, ownership, and patient focus.

 ACOs are generally led by hospital providers, physician groups, or a combination of the two that seek to improve the care they provide, whether it is inpatient, outpatient, or both. Within the ACO model, provider behavior change and delivery reform is promoted through financial responsibility. Organizations and insurers enter into a contractual or risk-based agreement that is associated with the financial and clinical outcomes of a defined population; thus, accountable care entities are often classified by the payers with whom they share financial responsibility. Although patients are attributed to an ACO via their primary care physician, the risk-bearing entity is typically the hospital or provider group. As of January 2016, there were 838 ACOs, many of which have accountable care contracts with both public and private payers.

 There are 1217 total ACO-payer contracts: 32% of are commercial contracts, 51% are government, and 17% are government and commercial.

The ACO model was first legislatively encouraged within the Medicare program through the ACA, and Medicare ACOs make up a significant portion of the ACO market.

 Medicare ACOs are created by attributing patients—based on actual utilization of primary care services&mdash;to ACO organizations that will include the spectrum of healthcare providers, including physicians and hospitals. The current Medicare accountable care models are the Pioneer ACO program, which commenced in January 2012, and the Medicare Shared Savings Program (MSSP), which began in April 2012; a third program, the Next Generation ACO model, began in 2016.

Commercial payers and their provider partners have also formed commercial variations of the ACO model, including commercially developed quality measures.

 To date, the level of commercial ACO activity is greater than both of the federal programs combined, measured in either individual contracts or total covered lives.

 State-level ACO dynamics are also becoming increasingly prevalent, as Medicaid agencies struggle to find politically feasible ways to control the costs of their expanding healthcare assistance programs. Several states have adopted Medicaid versions of the ACO model, with many more either running pilot programs or actively exploring legislation that would enable accountable care initiatives.

The amount of financial risk absorbed by an ACO varies from incentive payments to upside-only bonus arrangements, and to upside- and downside contracts and full capitation. The 2 most common models are either a shared savings—including upside- and double-sided arrangements&mdash;or partial or full capitation.

 In shared savings, ACOs that spend below project costs share in a portion of earned savings; if also sharing in risk, ACOs pay a penalty for spending above projections.

 Most Medicare ACOs to date are in an upside-only shared savings model. However, some Pioneer and Next Generation ACOs have assumed more comprehensive financial responsibility in models that enable capitation, where organizations bear greater degrees of upside- and downside risk for spending compared with a negotiated rate. The private market reflects the most diversity among payment models.

Hospital Value-Based Purchasing Programs in Medicare

Similar to ACOs, providers participating in VBP programs have financial incentives that are linked to provider performance on specific care measures. Depending on a provider’s success in meeting pre-established quality and efficiency targets, they are either rewarded bonuses or penalized through reduced Medicare reimbursements. Although VBP initiatives vary across health settings, this analysis focuses on 3 hospital-based VBP strategies, which are summarized below.

In contrast to the optional ACO program, as authorized by the ACA, participation in Medicare VBP programs is mandatory for all eligible hospitals that accept Medicare payments.

 Participation is based on providing acute care and primarily includes short-term acute care hospitals and critical access hospitals, although some specialty hospitals are also included; federal Veteran’s Administration and Department of Defense hospitals are not. For each program, CMS adjusts the reimbursement rate for all services at a hospital. For example, doing poorly on readmission reductions could result in a hospital receiving only 99% of total eligible Medicare payments for all inpatient services provided to all Medicare beneficiaries. Thus, VBP program penalties represent a significant government effort to improve care quality. A summary of VBP penalties and metrics for each payment year is included in 

Under the HVBP, CMS provides incentive payments to hospitals for the quality of care they provide to Medicare patients. Quality is rewarded by how closely best clinical practices are followed and how well the hospital enhances a patient’s hospital stay experience.

 The program applies to payments on or after October 1, 2012, and affects payment for inpatient stays in 2985 hospitals across the United States.

Medicare provides quality incentive payments based on one of 2 available methods—either how well the hospital performs on each measure relative to the national average (achievement), or how much the hospital improves performance on each measure compared with their baseline (improvement).

 To assess total performance, CMS compares both a hospital’s achievement and improvement scores for each applicable HVBP measure and assigns final scores based on the higher of the two. Scores are then aggregated to create a composite score, from which incentive payments are determined.

 Hospitals are either penalized for poor performance or receive bonus payments for good performance. The size of the HVBP incentive has increased over time, from 1% of total Medicare payments in 2013 to 1.75% in 2016 (Table 1). Payment adjustments have ranged from 0.99% to 1.01% in 2013, to 0.9825% to 1.03% in 2016.

The HRRP provides a financial incentive for hospitals to reduce preventable readmissions.

 For the HRRP, specified conditions include heart attack, heart failure, pneumonia, total hip and knee repairs, and chronic obstructive pulmonary disease conditions. Rather than employing a reward system as in the HVBP, the HRRP incentive is based on a financial penalty for excess avoidable readmissions—which is a measure of the hospital’s readmission performance compared with the risk-adjusted national average.

 This excess readmissions ratio is then multiplied by the sum of base-operating DRG payments per specified condition, from which a final penalty is calculated.

Although the original fiscal year (FY) 2013 penalty could lead to a maximum of a 1% reduction on a hospital’s Medicare Diagnosis-Related Group reimbursement, the penalty increased to 2% in FY2014 and was capped at 3% in FY2015.

 Payment adjustments are based on a percentage of total payments, meaning that adjustments ranged from 0.99 to 1.0 in FY2013 to 0.97 to 1.00 in FY2015.

Hospital-Acquired Condition Reduction Program

The HAC Reduction Program is intended to improve inpatient safety by imposing financial penalties on hospitals that perform poorly with regard to hospital-acquired conditions, which are specified through CMS rule-making each year.

 Hospital HAC performance is assessed based on a hospital’s total HAC score, which ranges from 1 to 10, with 10 indicating the poorest performance. Total scores are determined by a weighted average of 2 domains: 1) patient safety, and 2) healthcare-associated infections—although other measures will likely be added in the future.

 From FY2015 onward, Medicare’s acute inpatient prospective payment system (IPPS) hospitals that rank in the highest quartile of total HAC scores (score of 7 or higher) will receive a 1% DRG payment reduction for all discharges.

Value-based purchasing data were obtained from Medicare’s acute IPPS final rule data files for FY2013 through FY2016.

 For the HVBP and the HRRP, 4 years of payment adjustment and performance data were available (FYs 2013, 2014, 2015, and 2016). For the HAC Reduction Program, 2 years of penalty data were available (FY2015 and FY2016). Because the amount of the potential penalties or bonus payments changed over time (1% of total payments in FY2013 to 3% beginning in 2015), to compare year-over-year performance, we normalized the penalties for each year to show trended data. After normalization, the mean of each year’s value is 0, with a standard deviation of 1.

Data on hospital participation in ACOs were obtained from the Leavitt Partners ACO database, which tracks organizations that are participating in accountable care arrangements and includes information on the hospitals participating in each arrangement and when the ACO went into effect. From this data, we were able to assign the quarter in which a hospital first became part of an ACO. Of the 3950 hospitals for which we were able to obtain any VBP data, 1566 are affiliated with or owned by an ACO in the Leavitt Partners database, and are thus considered to be a part of an ACO. We controlled for hospital characteristics based on number of beds, region, and ownership. Data were obtained through the end of November 2015.

We sought to understand whether ACO participation is correlated with better scores or improvements for the Medicare VBP programs. We performed 2 primary analyses: 1) a descriptive comparison of the VBP performance of hospitals that are participating in ACOs compared with those that are not, and 2) a longitudinal performance analysis of hospitals that newly become part of an ACO. In the descriptive analysis, we compared the various VBP scores for all hospitals that were part of an ACO with all hospitals that were not part of an ACO, using the most recent year (FY2016) data. We estimated raw scores and also adjusted for a variety of hospital demographic factors. The descriptive analysis was conducted for all 3 VBP programs, while the longitudinal analysis was limited to the HVBP and the HRRP due to lack of published historical data on the HAC Reduction Program. Because the HVBP scoring methodology changes each reporting year, we also repeated the analysis, retaining the second year methodology for subsequent years.

To estimate the effect that becoming an ACO had on a hospital, we evaluated the performance of hospitals that became part of an ACO during the time period for which performance data was obtained during the second year of our data against all hospitals that never became part of an ACO. Because the underlying performance data is based on data collected several years in advance, we identified hospitals that became ACOs during the performance period that led to the final performance year (PY) score for FY2014. Historical data were only available for the HVBP and the HRRP. For the HVBP, we used hospitals that became ACOs in calendar year 2012 (517 hospitals), which corresponds to performance data that affected FY2014 payments. For the HRRP, we used hospitals that became an ACO from July 2011 to June 2012 (351 hospitals), which corresponds to performance data that affected FY2014 payments. In both cases, we compared these hospitals with the 2384 hospitals that were never part of an ACO. 

The average performance of all hospitals participating in ACOs compared with those that are not participating in ACOs for FY2016 is available in 

. For the HVBP and the HRRP, a higher score is better, while for the HAC Reduction Program, a lower score is better. Hospitals participating in ACOs did worse than non-ACO hospitals for the HVBP (

 =.072), and worse for the HAC Reduction Program (

 <.001). When adjusting for hospital characteristics (beds, region of the country, ownership and teaching status), the direction of the difference remained, but the significance changed; hospitals in ACOs still did worse for the HVBP (

 = .028), and worse for the HAC Reduction Program (

 = .28). It should be noted that the differences for the VBP scores are very small because the actual scores tightly cluster around 1.00; however, even very small differences do correspond to significant differences in performance with these programs.

The results of the comparison of ACOs that newly joined ACOs during the second reporting year compared with non-ACO hospitals can be seen in 

 for the HRRP. For the HVBP, ACO hospitals outperformed non-ACO hospitals during year 1, although the difference is not significant (

= .30). During year 2, hospitals that joined an ACO in 2012 did significantly better than non-ACO hospitals (

= .005), but did not maintain that improvement into year 3 and did worse (

= .03). For the HRRP, ACO hospitals did slightly worse than non-ACO hospitals in year 1 (

= .06). For the most recent performance period, hospitals that participate in an ACO did better than their peers for the HRRP, but did worse on the VBP and HAC programs.

 compares the HVBP performance of hospitals joining ACOs during the second reporting year with non-ACOs, using only second reporting year methodology for all 4 years. If assessment methodology had remained constant from FY2014, hospitals that joined ACOs would have consistently outperformed non-ACO hospitals in PY2 (

Despite similar goals to improve healthcare quality and costs, variation exists in hospital participation in an ACO and VBP program. For FY2016, ACO-participating hospitals performed better than non-ACO hospitals for the HRRP, but not on the HVBP and the HAC Reduction Program. Longitudinal analysis, however, reveals that results are more varied, with evidence that hospitals joining ACOs did increasingly better than their peers for the HRRP, but had inconsistent results year-over-year with the HVBP.

High ACO hospital HRRP scores may signify that ACO-participating hospitals will first focus their priorities on value-based measures that overlap between programs. Reducing readmissions improves a hospital’s HRRP scores, the efficiency score in the HVBP, and clinical outcomes. Reduced readmissions also increases the possibility that an ACO will share in savings by reducing costs. MSSP ACOs are measured on their ability to reduce readmissions, and ACOs scored consistently high on this measure for the first year of the program.

 Although performance varied on other quality measures, most participants earned the maximum points possible for readmissions reductions. As ACOs have focused on readmissions to reduce costs, it is likely that ACO-participating hospitals will continue performing well on readmissions within the HRRP, as the ACO has increased incentive to reduce readmissions.

ACO-participating hospitals performed better than non-ACO hospitals from PY1 to PY2 for the HVBP, but performed worse from PY2 to PY3. The significant decrease in comparative performance during the last performance period indicates that ACO participation and VBP outcomes might not necessarily correlate. Part of this change, however, may be driven by changes in methodology between performance years of the HVBP. During each new performance period, the measured domains of the program have different weight contributions to a hospital’s overall score; new domains and additional measures are also included each year (Table 1).

The impact of the FY 2015 changes seemed quite strong as year-over-year correlation dropped from 0.61 (PY1 to PY2) to 0.42 (PY2 to PY3). Had the HVBP methodology remained consistent, hospitals that joined ACOs would have consistently outperformed non-ACO hospitals. As CMS continues to expand its focus on VBP arrangements, it must be sure that it rewards and penalizes based on factors that are truly associated with better care.

The descriptive analysis reveals that hospitals that are not a part of an ACO outperform ACO-participating hospitals in the HAC Reduction Program. Although there is a possibility that ACO hospitals see higher-acuity patients and are thus at risk for a higher frequency of hospital acquired conditions, it is difficult to assess causation on the poor performances of ACO hospitals compared with non-ACO hospitals. Various studies have suggested that the HAC Reduction Program scoring methodology is itself flawed and that the outcomes scores do not adequately reflect improvement.

 More comprehensive performance results may provide a clearer picture of the impact of ACO participation on HAC Reduction Program outcomes. 

Articles

Analyzes whether hospital participation in accountable care organizations is associated with a hospital’s quality and cost improvement outcomes in other Medicare value-based payment programs.

Katelyn de Lisle, BS

Articles

Hospital Participation in ACOs Associated With Other Value-Based Program Improvement