[Skip to Navigation]
Sign In
Figure.  Percentage Differences in Adjusted All-Cause and Avoidable Acute Care Use for Medicare Advantage Beneficiaries Cared for Under 2-Sided Risk vs Fee-for-Service (FFS) Payment Models
Percentage Differences in Adjusted All-Cause and Avoidable Acute Care Use for Medicare Advantage Beneficiaries Cared for Under 2-Sided Risk vs Fee-for-Service (FFS) Payment Models

Across all 3 categories of use, the relative differences in avoidable and all-cause events were statistically significant at the P < .05 level. Data were adjusted for age, sex, race, low-income subsidy, comorbidities, and hospital referral region. Avoidable hospitalizations and observation stays were determined based on standardized definitions available from the Agency for Healthcare Research and Quality Preventable Quality Indicators.4 Avoidable ED visits, determined using a validated claims-based algorithm,5 included those in which emergency department care was required based on the complaint or procedures performed and resources used, but the emergent nature of the condition was potentially preventable and avoidable if timely and effective ambulatory care had been received during the episode of illness.

Table.  Adjusted Rates of Acute Care Use for Medicare Advantage Beneficiaries by Payment Modela
Adjusted Rates of Acute Care Use for Medicare Advantage Beneficiaries by Payment Modela
1.
APM Measurement: progress of alternative payment models: 2019 methodology and results report. Health Care Planning and Learning Action Network. 2019. Accessed October 5, 2021. http://hcp-lan.org/workproducts/apm-methodology-2019.pdf
2.
Song  Z.  Taking account of accountable care.   Health Serv Res. 2021;56(4):573-577. doi:10.1111/1475-6773.13689PubMedGoogle ScholarCrossref
3.
Freed  M, Biniek  JF, Damico  A, Neuman  T. Medicare Advantage in 2021: enrollment update and key trends. KFF. June 21, 2021. Accessed October 5, 2021. https://www.kff.org/medicare/issue-brief/medicare-advantage-in-2021-enrollment-update-and-key-trends/
4.
Guide to prevention quality indicators: hospital admission for ambulatory care sensitive conditions. Agency for Healthcare Research and Quality. 2001. Accessed October 5, 2021. https://www.ahrq.gov/downloads/pub/ahrqqi/pqiguide.pdf
5.
Ballard  DW, Price  M, Fung  V,  et al.  Validation of an algorithm for categorizing the severity of hospital emergency department visits.   Med Care. 2010;48(1):58-63. doi:10.1097/MLR.0b013e3181bd49adPubMedGoogle ScholarCrossref
6.
Pratt  NL, Kerr  M, Barratt  JD,  et al.  The validity of the Rx-Risk Comorbidity Index using medicines mapped to the Anatomical Therapeutic Chemical (ATC) Classification System.   BMJ Open. 2018;8(4):e021122. doi:10.1136/bmjopen-2017-021122PubMedGoogle Scholar
2 Comments for this article
EXPAND ALL
Selection bias among practices accepting value-based payment
Stephen Kemble, MD | Queen's Medical Center, Honolulu
While I am sure practices taking on 2-sided risk are more motivated to keep their patients out of the ERs and hospitals, we have also seen significant churn among practices participating in ACOs. They participate while the risk is one-sided and then drop out before 2-sided risk is imposed. My suspicion is the ones who stay in for 2-sided risk have already established lower risk patient panels so that they have confidence in their ability to make it work, and those who serve higher-risk (socially disadvantaged) populations are averse to accepting 2-sided risk, creating a pre-selected favorable risk pool in the 2-sided risk group.

The authors do mention that there may be “selection bias around which primary care organizations engage in value-based payment," using the example having more tools to manage chronic disease, which may have some effect. However, they don’t mention cherry picked patient panels as a source of bias. Certainly if I were practicing in a primary care organization that was anticipating 2-sided risk, I would feel reluctant to take on or retain high-risk patients. So again, unless there is a way to accurately correct for biased patient selection and administrative costs, we really can’t say that taking on risk improves cost-effectiveness of care.
CONFLICT OF INTEREST: None Reported
READ MORE
Selection Bias?
Gordon Moore, MD, MPH | Professor of Population Medicine, Harvard Medical School, Boston, MA
This report is on an important issue. It is indeed vital to know if accepting financial risk for performance actually results in changes in provider performance that, after accounting for program administrative costs, lowers Medicare net costs. Unfortunately, this study is seriously flawed by selection bias and study design. It does not answer the question.
The results of evaluations of value based payment models thus far reported have been discouraging. Even before accounting for costs, evaluators have shown savings that have occasionally been statistically significant but not truly important in our quest to reduce the
costs of care. This sorely disappointing finding has led CMS to adopt the hypothesis that only when provider organizations are responsible for losses as well as sharing profits will risk -based payment work to enhance value .
As the authors point out, most provider organizations will only accept the downside risk formula if it is a safe bet. Voluntary participation preferentially selects for those who already believe they have the capacity to find efficiencies and reduce costs. One would thus expect their performance already to be better to start with. Moreover, with only a cross-sectional comparison rather than before-after design, the effect of a more competent and confident group would already have influenced these outcomes regardless of the association with risk-sharing.
We have no idea if sharing in risk affected the outcome measures. The selection bias and cross-sectional design in this study severely limits any causal conclusions that can be drawn about whether putting providers at risk for clinical and cost performance makes any difference in cost and quality. Their conclusion that this study "suggests that downside financial risk may play a key role in effective value-based payment arrangements" likely overstates their findings.
CONFLICT OF INTEREST: None Reported
READ MORE
Research Letter
Health Policy
March 17, 2022

Analysis of Value-Based Payment and Acute Care Use Among Medicare Advantage Beneficiaries

Author Affiliations
  • 1Harvard Medical School, Boston, Massachusetts
  • 2Harvard Business School, Boston, Massachusetts
  • 3Humana Inc, Louisville, Kentucky
  • 4Department of Medicine, Tufts University School of Medicine, Boston, Massachusetts
JAMA Netw Open. 2022;5(3):e222916. doi:10.1001/jamanetworkopen.2022.2916
Introduction

Medicare is increasingly shifting from fee-for-service reimbursement toward value-based models that reward improving quality and controlling spending.1 While researchers have evaluated these models in traditional Medicare,2 less is known about value-based models in Medicare Advantage (MA), despite greater penetration of value-based payment in MA1 and continued growth of the MA program.3 In this study, we examined the association between value-based payment and acute care use in a national population of MA beneficiaries.

Methods

This cohort study was reviewed by the Humana Healthcare Research Human Subject Protection Office, deemed not human participants research, and informed consent was waived. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

We identified beneficiaries enrolled in plans offered by a large MA organization from January 1, 2017, to December 31, 2019, and categorized them according to the payment model for their attributed primary care organization: fee-for-service (FFS); shared savings with upside-only financial risk (upside-only risk); and shared savings with upside and downside financial risk (2-sided risk). Further detail is provided in the eMethods in the Supplement.

We used claims data to identify hospitalizations, observation stays, and emergency department (ED) visits from January 1, 2019, to December 31, 2019. We segmented these outcomes into avoidable and all-cause events (eMethods in the Supplement).4,5 Next, we used quasi-Poisson regression models to estimate the association between payment model and acute care use, adjusting for age, sex, race, low-income subsidy, Rx-Risk-V score,6 and hospital referral region. The Rx-Risk-V score is a pharmacy claims–based comorbidity index and was chosen to avoid any potential bias from differences in documentation practices. To aid interpretability, we calculated adjusted rates of acute care use per 1000 patients, by payment model. Analyses were conducted between May 2021 and August 2021 using SAS Enterprise Guide v8.2 (SAS Institute). All P values were from 2-sided tests, and results were deemed significant at P < .05.

Results

In a study population of 489 796 MA beneficiaries, value-based payment was significantly associated with lower acute care use (Table). Compared with FFS, beneficiaries cared for under 2-sided risk models had lower rates of hospitalizations, observation stays, and ED visits. For example, the adjusted rate of ED visits per 1000 patients for 2-sided risk models was 375.8 (95% CI, 370.9-380.7) compared with 434.1 (95% CI, 426.5-441.9) for FFS. For all outcomes, there was no significant difference in acute care use between beneficiaries cared for under upside-only risk models and FFS.

The association between value-based payment and decreased acute care use was most pronounced for measures of avoidable acute care use. Compared with FFS, 2-sided risk models were associated with a 15.6% (95% CI, 14.2%-17.0%) relative reduction in avoidable hospitalizations, compared with 4.2% (3.4%-4.9%) for all-cause hospitalizations (Figure).

Discussion

In this study of MA beneficiaries, advanced value-based payment arrangements (ie, 2-sided risk models) were associated with lower rates of acute care use, especially those events that are potentially avoidable. These findings are consistent with evaluations of value-based payment in traditional Medicare2 and serve to expand the evidence base around value-based payment models in Medicare Advantage.1 The lack of significant differences between FFS and upside-only risk models suggests that downside financial risk may play a key role in effective value-based payment arrangements.

This study had limitations. Given the retrospective design, there is potential for residual confounding. Furthermore, it is likely there is some selection bias around which primary care organizations engage in value-based payment models. For example, groups that have invested in tools and infrastructure to manage chronic disease and population health may be more willing to bear financial risk.

Our findings suggest that organizations engaging in advanced value-based payment models in MA deliver differential outcomes to the MA beneficiaries under their care. Further research is needed to elucidate the activities of value-based primary care organizations that are associated with reductions in acute care use.

Back to top
Article Information

Accepted for Publication: January 29, 2022.

Published: March 17, 2022. doi:10.1001/jamanetworkopen.2022.2916

Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License. © 2022 Gondi S et al. JAMA Network Open.

Corresponding Author: Brian W. Powers, MD, MBA, Department of Medicine, Tufts University School of Medicine, 136 Harrison Ave, Boston, MA 02111 (BPowers5@Humana.com).

Author Contributions: Dr Powers had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Gondi, Li, Boudreau, Shrank, Powers.

Acquisition, analysis, or interpretation of data: Gondi, Li, Drzayich Antol, Shrank, Powers.

Drafting of the manuscript: Gondi, Shrank, Powers.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Gondi, Li, Shrank.

Administrative, technical, or material support: Gondi, Boudreau, Shrank, Powers.

Supervision: Drzayich Antol, Shrank, Powers.

Conflict of Interest Disclosures: Dr Gondi reported receiving personal fees from Humana, Inc, being previously employed at Commonwealth Care Alliance, and Humana, and serving as Advisor at 8VC outside the submitted work. Dr Drzayich Antol reported having equity holdings with Humana Inc. Dr Boudreau reported receiving grants from Yale University outside the submitted work. Dr Shrank reported being on the Board of Directors at GetWell Network outside the submitted work. Dr Powers reported having equity holdings with Humana and prior employment by Anthem and Fidelity Investments outside the submitted work. No other disclosures were reported.

Disclaimer: The views expressed in this article represent the authors’ views and not necessarily the views or policies of their respective affiliated institutions.

Additional Contributions: The authors thank Debbie Peikes, PhD, MPA (Humana, Inc.), for comments on an earlier draft. Dr Peikes did not receive compensation for her time.

References
1.
APM Measurement: progress of alternative payment models: 2019 methodology and results report. Health Care Planning and Learning Action Network. 2019. Accessed October 5, 2021. http://hcp-lan.org/workproducts/apm-methodology-2019.pdf
2.
Song  Z.  Taking account of accountable care.   Health Serv Res. 2021;56(4):573-577. doi:10.1111/1475-6773.13689PubMedGoogle ScholarCrossref
3.
Freed  M, Biniek  JF, Damico  A, Neuman  T. Medicare Advantage in 2021: enrollment update and key trends. KFF. June 21, 2021. Accessed October 5, 2021. https://www.kff.org/medicare/issue-brief/medicare-advantage-in-2021-enrollment-update-and-key-trends/
4.
Guide to prevention quality indicators: hospital admission for ambulatory care sensitive conditions. Agency for Healthcare Research and Quality. 2001. Accessed October 5, 2021. https://www.ahrq.gov/downloads/pub/ahrqqi/pqiguide.pdf
5.
Ballard  DW, Price  M, Fung  V,  et al.  Validation of an algorithm for categorizing the severity of hospital emergency department visits.   Med Care. 2010;48(1):58-63. doi:10.1097/MLR.0b013e3181bd49adPubMedGoogle ScholarCrossref
6.
Pratt  NL, Kerr  M, Barratt  JD,  et al.  The validity of the Rx-Risk Comorbidity Index using medicines mapped to the Anatomical Therapeutic Chemical (ATC) Classification System.   BMJ Open. 2018;8(4):e021122. doi:10.1136/bmjopen-2017-021122PubMedGoogle Scholar
×