Risk-adjusted payment and performance assessment for primary care

Abstract

Background: Many wish to change incentives for primary care practices through bundled population-based payments and substantial performance feedback and bonus payments. Recognizing patient differences in costs and outcomes is crucial, but customized risk adjustment for such purposes is underdeveloped.

Research design: Using MarketScan's claims-based data on 17.4 million commercially insured lives, we modeled bundled payment to support expected primary care activity levels (PCAL) and 9 patient outcomes for performance assessment. We evaluated models using 457,000 people assigned to 436 primary care physician panels, and among 13,000 people in a distinct multipayer medical home implementation with commercially insured, Medicare, and Medicaid patients.

Methods: Each outcome is separately predicted from age, sex, and diagnoses. We define the PCAL outcome as a subset of all costs that proxies the bundled payment needed for comprehensive primary care. Other expected outcomes are used to establish targets against which actual performance can be fairly judged. We evaluate model performance using R(2)'s at patient and practice levels, and within policy-relevant subgroups.

Results: The PCAL model explains 67% of variation in its outcome, performing well across diverse patient ages, payers, plan types, and provider specialties; it explains 72% of practice-level variation. In 9 performance measures, the outcome-specific models explain 17%-86% of variation at the practice level, often substantially outperforming a generic score like the one used for full capitation payments in Medicare: for example, with grouped R(2)'s of 47% versus 5% for predicting "prescriptions for antibiotics of concern."

Conclusions: Existing data can support the risk-adjusted bundled payment calculations and performance assessments needed to encourage desired transformations in primary care.

FIGURE 1

Distributions for 3 Models of Observed-to-Expected Ratios in the 325 Most Common HCCs for the Proxy Primary Care Spending Variable

FIGURE 2

Predicted vs. Actual PCP-level Average Per Capita PCALPractice Spending A. HCC-Predicted versus Actual PCAL B. Age-sex Predicted versus Actual PCAL Notes: PCP: primary care provider; PCAL: primary care activity level; HCC: hierarchical condition categories. For each of 436 PCPs, plotted values are person -year averages for those assigned to the practice. On the horizontal axis are 2007 PCAL predictions, converted to a relative cost by dividing by the sample mean. Vertical -axis values are defined analogously for actual PCAL outcome values (called in the text), also normalized to 1. Figure 2A uses predictions for Y from the HCC model; Figure 2B uses predictions from an age-sex model.

FIGURE 3

HCC-Predicted vs. Actual PCP-level Average Per Capita PCAL Practice Spending, by Provider Specialty Notes: Points in these scatterplots are a subset of those in Figure 2A, covering 3 subspecialties: pediatric (n=82), family practice (n=127), and internal medicine (n=63). For each PCP, values on the horizontal axis are the sum of predicted 2007 PCAL costs using the 394-category HCC model divided by the number of full-year equivalent people assigned to the practice, converted to a normalized cost by dividing by the sample mean. Vertical -axis values are defined analogously, using sums of all calculated PCAL values in the numerators. Regression lines, fit separately to pediatric (x’s), family medicine (squares) and internal medicine practices (triangles), are indicated. Variables on both axes are normalized to average 1.0.

FIGURE 4

Observed Versus Predicted Number of Prescriptions for Antibiotics of Concern (ABX) per Capita, by PCP A. Predictions Using Tailored (ABX) Model B. Predictions Using Generic (Health-Spending Calibrated) Model Note: Each dot plots average observed vs. average predicted number of prescriptions for antibiotics of concern for one practice among 436 PCPs serving 456,781 patients. Data are from the Practice-Based sample of patients assigned to mid-size practices in the Thomson Reuters 2007 MarketScan Commercial Claims. In each figure the predicted number of prescriptions for each practice is the mean of its individual level predictions. Figure 4A predictions are made using a model tailored to predict this specific outcome, while Figure 4B uses the normalized risk score from a model tailored to predict total health spending. Each model was estimated on the full sample (N~17.4 million), as a function of age, sex, 394 hierarchical condition categories (HCCs) and interaction terms.