Clinical Quality Performance of Value-Based and Fee-for-Service Models for Medicare Advantage

Abstract

This quality improvement study compares 2 reimbursement models by their mean clinical quality scores and risk-sharing levels.

Introduction

Value-based payment (VBP) and value-based care delivery models align clinician reimbursement with high-quality, cost-effective, evidence-based, patient-centered care,^1,2 while fee-for-service (FFS) models incentivize volume over health outcomes. VBP models span a continuum of financial incentive and risk-sharing arrangements, including pay-for-performance (P4P), 1-sided risk, and 2-sided risk.² Clinical quality performance was examined among Medicare Advantage (MA) members who received care from clinicians in VBP vs FFS arrangements.

Methods

This retrospective cross-sectional study analyzed unadjusted 2022 Medicare Star Ratings data for MA members of a large national payer. The Sterling Institutional Review Board deemed this study exempt from review and informed consent because it used deidentified data. The study followed the STROBE reporting guideline.

Members were designated to the VBP or FFS model based on their primary care practitioner’s contract (eMethods and eFigure in Supplement 1). Fifteen clinical quality measures representing 5 secondary and tertiary prevention categories were aggregated at the member level: 2 cancer screening, 4 diabetes, 4 heart disease, 1 osteoporosis, and 4 care coordination measures. Mean (SD) clinical quality scores were calculated according to Medicare Star Ratings technical notes.^3,4

Generalized linear model analyses with Bonferroni correction assessed differences in mean clinical quality scores between VBP and FFS and across risk-sharing levels (2-sided risk, 1-sided risk, P4P, FFS). Post hoc Tukey tests were used for pairwise comparisons.

Two-sided P < .05 indicated statistical significance. Analyses were performed from August to November 2024 using SAS Enterprise Guide 8.3 (SAS Institute).

Results

Among 3 312 536 MA members (mean [SD] age, 72.4 [9.1] years; 1 861 646 females [56.2%]), 57.1% were attributed to VBP and 42.9% to FFS model. Across all quality measures, VBP outperformed FFS (mean [SD] score difference, +6.7% [0.6%]; all P < .001) (Figure 1). Blood glucose control (mean [SD] score difference, +25.5% [0.3%]) and controlling high blood pressure (mean [SD] score difference, +23.3% [0.2%]) had markedly better performance with VBP than FFS.

Figure 1. Comparison of Clinical Quality Performance Between Value-Based Payment (VBP) and Fee-for-Service (FFS) Arrangements.

Mean and SD (denoted by error bars) scores plotted by quality measure. All comparisons are statistically significant at P < .001. BP indicates blood pressure; CVD, cardiovascular disease; and ED, emergency department.

As shown in Figure 2, when examined by risk-sharing level, 2-sided risk outperformed FFS for all measures; 1-sided risk sharing and P4P outperformed FFS for 14 and 13 measures, respectively. Two-sided risk outperformed 1-sided risk and P4P for 9 measures. Six measures related to hypertension, diabetes, and cancer demonstrated incremental improvements in performance (eg, mean [SD] score differences for controlling high blood pressure: 2-sided risk vs 1-sided risk +8.6% [0.4%]; 1-sided risk vs P4P +7.3% [0.4%]; P4P vs FFS +15.0% [0.4%]; all P < .001).

Figure 2. Comparison of Clinical Quality Performance Between Fee-for-Service (FFS) and 3 Types of Value-Based Payment (VBP) Arrangements.

Mean and SD (denoted by error bars) plotted for percent measure outcome by quality measure. BP indicates blood pressure; CVD, cardiovascular disease; ED, emergency department; P4P, pay-for-performance. Pairwise comparisons are statistically significant at P < .05.

^a2-sided risk vs 1-sided risk.

^b2-sided risk vs P4P.

^c2-sided risk vs FFS.

^d1-sided risk vs P4P.

^e1-sided risk vs FFS.

^fP4P vs FFS.

Discussion

VBP outperformed FFS for all 15 clinical quality outcomes. Across the risk-sharing continuum, clinical quality performance consistently improved as financial arrangements moved toward 2-sided risk. Incremental clinical quality improvements were observed with increased risk-sharing for 6 measures related to hypertension, diabetes, and cancer—prevalent and costly conditions. Controlling high blood pressure and blood glucose control are heavily weighted in health care quality measurement frameworks.⁵

Quality outcomes for VBP arrangements exceeded national means^3,4 for 11 of 15 measures, while FFS outperformed 4. Moreover, 2-sided risk, 1-sided risk, and P4P outperformed national means for 14, 11, and 10 measures, respectively. These results align with previous findings that commercial VBP models are mostly beneficial to care quality.⁶ Furthermore, this study addresses the need for more evidence on VBP⁶ (specifically 2-sided risk) models in the commercial insurance sector while evaluating the full risk-sharing continuum.

A study limitation is that this study supports an associational rather than causal relationship, as a randomized clinical trial was not feasible in this setting. Nevertheless, this study enhances generalizability of findings by capturing clinical outcomes using standardized, nationally recognized quality metrics. Additionally, clinicians under VBP are incentivized to report quality measures and therefore may report more completely than their FFS counterparts; however, data-reporting incentives are purpose-built into VBP because data completeness and resulting insights inform interventions to optimize quality. Lastly, while self-selection bias is also possible, inclusion of multiple risk-sharing models addresses variation in risk tolerance.

This contemporary, large-scale analysis demonstrates that VBP models, particularly with increased risk sharing, are associated with superior performance for standardized clinical quality measures. These findings support VBP’s role in advancing mutually beneficial health and health care–related outcomes for patients, clinicians, and payers.

Supplement 1.

eMethods. Additional Detail Regarding Study Sample

eFigure. Comparison of Payment Models Included in Study

Supplement 2.

Data Sharing Statement