reports

From theory to application: the implementation of risk adjustment for Minnesota Medicaid

Minneapolis, MN, USA: Park Nicollet Institute, Minnesota Department of Health, Minnesota Department of Human Services.

Park Nicollet Institute and Minnesota Department of Health, Minneapolis, MN, USA

On January 1, 2000, the state of Minnesota implemented diagnosis-based risk adjustment for its prepaid, publicly-sponsored health care programs: the Prepaid Medical Assistance  Program (PMAP) , the Medicaid program; the Prepaid General  Assistance Medical Care program (PGAMC); and MinnesotaCare.  Minnesota uses a health plan-level, concurrent (sometimes  called “retrospective”) risk assessment model for risk adjustment. The use of a concurrent risk assessment model for payment is a departure from the theoretical underpinnings of risk adjustment, which support a prospective model for payment, and from most of the risk-adjusted payment models currently in use in the US. Although the actual risk adjustment system in Minnesota is garnering increasing attention from other states, the decisions that led to its development and successful implementation are equally noteworthy. These decisions included theory-based testing of a commercially-available risk assessment grouper and a prospective payment model, evaluation of their applicability to the relevant Minnesota populations, and incentives to encourage health plan compliance with data submission requirements.
This paper describes the risk adjustment system that was ultimately implemented in Minnesota and the milestone decisions that are its foundation. Minnesota’s experience highlights four key lessons that should be useful for other states that risk adjust Medicaid payments, particularly for those populations not characterized by disability: (1) diagnosis-based risk adjustment, albeit imperfect, distributes payment more equitably among health plans than does traditional demographic payment adjustment; (2) the payment model should be based on the characteristics of the risk-adjusted population(s) and kept separate from “political” adjustments; (3) health plans, in general, will submit timely encounter data given the appropriate incentives; and (4) on-going stakeholder participation and education is critical to successful implementation. Lessons (1), (3), and (4) are well documented  and bear repeating, as this paper will show. Lesson (2) has not yet been documented and is a valuable and unique contribution to the literature.