Online Program

The Role of Mathematical Modeling in Health Care Reform
*Arlene Ash, Umass Medical School 
Randall P Ellis, Boston University 
Allan Goroll, Harvard Medical School 
Anju Joglekar, VeriskHealth 


The Patient Centered Medical Home (PCMH) is a practice reform model designed to deliver personalized, coordinated, comprehensive care that is cost-effective, efficient, and evidence-based. It uses a team approach and health information technology to provide primary care to a panel of patients. Implementing the PCMH and achieving its comprehensive care goals requires investing in practices. However, current fee-for-service payment rewards volume and procedures and underpays for evaluation and management services; it is a barrier to PCMH transformation. Payment reforms that utilize bundled payments and rewards for desired outcomes are an attractive alternative. Financing PCMHs and rewarding their performance requires 2 kinds of calculations for each practice and its panel: 1) Payments sufficient to enable the provision of comprehensive care (including support for teams and health information technology), and 2) Normative targets (expected values) for important outcomes, such as total health care costs, numbers of avoidable emergency room visits, and glucose (HbA1c) control in diabetics. Without reliable and accurate risk adjustment, PCMH practices with sicker-than-average panels will be underfinanced and inaccurately assessed. Today, most primary care payments and measures of provider performance are not risk-adjusted. For example, the HEDIS measure for diabetes control focuses on achieving a glycosylated hemoglobin (HbA1c) of 7%; it fails to recognize 1) the near-impossibility of achieving this target for, say, a homeless, morbidly obese patient whose HbA1c “starts” at 12%, and 2) the value of reducing that level to, say, 8%. This penalizes doctors who treat more difficult patients. In contrast, a risk-adjusted measure for glucose control might, for example, predict patient-specific probabilities of achieving a “success” (say, either an HbA1c < 7 or a reduction of at least 2 percentage points). A practice is then considered “good” to the extent that its panel achieves more successes than expected. Crucial to the credibility of this approach is the accuracy of the practice-level “expected” value produced by a risk-adjustment model. We discuss the PCMH as a striking example, but only one example, of the system-wide reform needed: specifically the role of 1) population-based and risk-adjusted judgments about 2) both the cost and quality of 3) integrated teams of health care providers in 4) controlling costs while improving the nation’s health.