DOCUMENTS

reports

A comparative analysis of methods of health risk assessment: final report

Published: June 6, 1995
Category: Reports
Authors: Bertko J, Braun P, Busch S, Dunn DL, Latimer E, Rosenblatt A, Stoiber T, Taira DA
Country: United States
Language: null
Type: Care Management
Settings: Academic, Health Plan

Cambridge, MA, USA: Harvard University School of Public Health.

Harvard University School of Public Health, Cambridge, MA, USA

A key component of many recent proposals for health care reform is a system of risk-adjustment payments among health plans. Such a system measures and removes the effects of plans’ relative health risks on the prices consumers face in choosing a plan. The fundamental feature that accomplishes this is the transfer of funds from one plan to another based on the underlying risk characteristics of each plan’s enrollees. In this way, consumer choice of a health plan can be based on its relative efficiency and quality rather than on the relative health status of its enrollees. Further, this system promotes the financial equity between health plans and more equitable access to them by individuals, particularly those of higher risk.
The rationale for risk adjustment is straightforward. Individuals know well their own needs and preferences for health care. Given a choice among competing health plans, they would choose a plan that is most suitable and advantageous to them. At the same time, some plans may actively seek lower risk individuals to insure and avoid those with higher risks. In both cases, health plans’ costs are affected by the particular combination of risks their enrollees represent.
As a result of selection, some health plans may face higher costs per enrollee. In the absence of accurate risk adjustment, such plans would in the longer run be forced to either limit access to needed services or increase premiums. In either case, their competitive position would be threatened, particularly if premiums were compressed through community rating. Conversely, plans with lower risk enrollees would enjoy a competitive advantage. In addition, without risk adjustment, plans may have a strong incentive to selectively enroll lower risk individuals. To both induce health plans to compete on efficiency and quality of health services and to provide equitable compensation to health plans for the risks they insure, it is necessary to have reliable and valid methods for determining risk adjustment payments.
Health risk adjustment can be thought of as a two-step process. The first step involves a risk assessment of each group of individuals to be insured. This assessment would measure the deviation of each individual’s expected cost of health care services from the average cost across all individuals.
The second step, risk adjustment refers to the methods used to compensate for the differences in risk, as measured by risk assessment. In a competitive market environment, such as that fostered by a Health Alliance purchasing cooperative, risk adjustment could involve the transfer of payments between competing carriers based on the risk assessment. Carriers with the higher risk populations would receive payments, whereas carriers with the lower risk populations would make payments. Risk adjustment also has utility as a tool for setting appropriate provider payments under capitation and, eventually, for provider profiling and outcomes measurement.
In all of these applications, the risk adjustment mechanism will only be as good as the underlying risk assessment method. A good risk assessment method should be able to predict health costs with accuracy. If the risk adjusted premium received by the plan is sufficiently close to the expected cost of a healthy or sick person’s care, it may be more costly for the plan to pursue a selection strategy than to simply accept all enrollees without regard to health risk. A good risk assessment method also cannot be so complex and costly that it cannot be applied under real life circumstances. A system should limit the ability of health plans to benefit financially by “gaming” the system. It should also be timely and allow predictability in setting premiums and determining risk transfer payments. Finally, it should provide appropriate incentives for efficient and high quality medical care.
The literature on assessing health risk assessment and risk adjustment is substantial. In particular there exist a number of competing methods for measuring the health risk of individuals and groups of individuals. Many studies have evaluated the abilities of a particular risk assessment model to accurately measure differences in risk. However, there have been few studies that have provided a comprehensive evaluation of the relative performance of different risk assessment methods.
In response to the need for a suitable risk assessment method, and in particular motivated by the heightened activity in health insurance market reform, the Society of Actuaries (SOA) funded a research study to explore the topics of health risk assessment and risk adjustment. In particular, we conducted a comparative investigation of the current methods available for health risk assessment. The study is unique in a number of ways. First, it provides a side-by-side comparison of competing models of health risk assessment using uniform methods and the same population of enrollees. Second, the enrollee data for the study come from a diverse population and describe a variety of health plans covering a range of care management approaches and all geographic areas of the country. Finally, while many studies have focused on risk assessment for elderly persons, this research examines risk assessment for a non-elderly population, including both children and adults.
We established three main objectives for the study:

  • Compare the predictive accuracy of different risk assessment methods;
  • Compare the different risk assessment methods based on other criteria, including administrative practicality, ability to resist manipulation and gaming, and incentives for efficiency; and
  • Explore the potential for risk adjustment using a list of high-cost conditions.

    Using a detailed data set developed by the SOA describing the demographic characteristics, diagnoses, medical utilization and expenditures for more than 4.5 million non-elderly individuals over a two-hear period, we tested the predictive accuracy of eight different Ambulatory Care Group (ACG) models, and five Diagnostic Cost Group (DCG) models. We also evaluated these models using other criteria including the feasibility of their implementation and the incentives they provide. In exploring the practical issues, we simulated a risk adjustment transfer process across plans using the different risk assessment methods. Finally, we developed and tested an alternative risk assessment model using a list of high-cost conditions.

Predictive Risk Modeling,Cost Burden Evaluation,Outcome Measures,Payment,United States

© The Johns Hopkins University, The Johns Hopkins Hospital, and Johns Hopkins Health System.
All rights reserved. Terms of Use Privacy Statement

Back to top