For states with limited Medicaid resources, risk adjustment helps identify best providers

Under severe budget pressures, many states are looking for ways to enable Medicaid managed care programs to profile managed care plans and better evaluate the impact of rate increases and better demonstrate the value that plans are providing.

The ways in which risk adjustment is being used have been studied under Centers for Medicare & Medicaid Services (CMS) contracts by the Center for Health Program Development and Management at the University of Maryland, Baltimore County.

John Kaelin, executive director for the Center for Health Program Development and Management, tells State Health Watch that his research first looked at nine states that have been using various risk adjustment programs and then added states that have not been using risk adjustment but would like to. He defines risk adjustment as a process by which the health status of an enrolled population is taken into consideration when determining capitation rates or other at-risk payments.

The method also can be used in managed care to evaluate patterns or outcomes of practice.

In a presentation earlier this year to the Seventh Annual Congress on Medicaid and Medicare, Mr. Kaelin said a method of risk adjustment is needed to deal with the fact that a small percentage of the population is responsible for a disproportionate share of health care expenditures.

"Let’s think about what this distribution in costs means for health care financing at the health plan or provider level," he said. "Is it likely that each plan or provider group would have a proportional distribution of each category of cost groups?"

While risk adjustment wouldn’t be necessary if populations were randomly distributed among plans and providers; in the real world, according to Mr. Kaelin, health plan membership and membership in a particular physician’s panel are not randomly distributed. As a result, he says, risk adjustment is necessary:

  • to protect plans and providers that enroll a costlier-than-average group of enrollees;
  • to minimize incentives for plans and providers from selecting or marketing to healthier enrollees;
  • to provide accurate financing for plans and providers that treat individuals with higher-than-average health care needs and costs.

In general, he says, risk adjustment affects the distribution of funds among health plans and should not increase or decrease payments in the aggregate to all plans. Risk adjustment should not determine the "size of the [payment] pie," but may affect the size of each managed care organization’s slice of the pie.

Under its first CMS contract, Mr. Kaelin’s center held two forums with representatives of nine states that are making risk-adjusted payments to collect information on their approaches and experiences. They developed a training manual on how to implement a risk-adjusted payment system and produced what he describes as the most complete documentation of state experiences in making risk-adjusted payments.

Under the second research grant, the center is responsible for preparing a checklist to be used by CMS regional offices to review risk-adjusted rate packages; developing a training manual and conducting a training session for regional staff, and holding a three-day forum for officials from states that want to learn how to use risk adjustment for payment methods, plan profiling, and program evaluation.

Survey findings

In January 2001, the center released the first comprehensive survey of states using health-based risk adjustment. Among its major findings:

  • Eight states used the Chronic Illness and Disability Payment System (CDPS/DPS), with 65 health plans affected.
  • Two states used Adjusted Clinical Groups (ACG), affecting 16 health plans.
  • Significant numbers of enrollees are covered by risk adjustment.
  • Three states risk adjust the Temporary Assistance for Needy Families population of women and children.
  • Four states risk adjust the disabled Social Security Income population.
  • Three states risk adjust both populations.

Mr. Kaelin says that the most common applications of risk adjustment are in financing (either determining capitation payments or establishing financial targets) or in profiling (comparing health plans and providers and assessing relative efficiency of plans and providers).

Implementation of a risk adjustment system involves a number of steps, starting with choosing the system that will be used.

There are two basic types of systems — categorical systems classify members into mutually exclusive groups based on their diagnostic history, while additive systems assign a risk score to each member based on his or her unique diagnostic history. Mr. Kaelin tells State Health Watch that for the states, the actual system selected is not as important as the implementation and data issues that arise following selection of a system.

Once a system has been selected, it is necessary to develop a database of diagnostic history for a plan’s members that will be used to measure their health status.

Questions to be posed and answered include the categories of service to be included, treatment of noncovered services, length of risk assignment period, risk assignment lag, and number of diagnoses.

Most risk assignment systems specify an annual risk assignment period, although some users believe that six or even three months is long enough to get an accurate measure of health status. Mr. Kaelin says there is a need to use a consistent number of diagnoses when health status is measured for payment purposes. There also is a need to define the lag between the payment period and the risk assignment period.

"With a two-year lag, the data for the risk assignment period should be complete," he says, "but many of the observed medical conditions may have been resolved. A one-year lag will be more reflective of current medical conditions, but the diagnostic data may be incomplete."

Based on choice of risk adjustment system, payments will either be based on risk groups or risk scores. With risk groups, Mr. Kaelin says, officials should consider combining the risk groups into a smaller number of actuarially stable groups.

A decision has to be made whether to make individual level payments or managed care organization level payments. Risk groups or scores can be used to determine payment for each individual. The groups and scores also can be used to determine the average health status for enrollees in each managed care organization.

While fee-for-service data can be used in new managed care programs, as programs mature they need to switch to encounter data to measure health status.

"You need to carefully evaluate your encounter data to ensure that they are complete and measure health status accurately," Mr. Kaelin says.

His survey of states using risk adjustment found that most phased in their programs using a variety of mechanisms, including implementing risk adjustment for one eligibility group and gradually expanding it to their entire population, blending risk adjustment and demographic payments, and placing corridors on changes in managed care organization revenue.

Although risk adjustment is generally budget-neutral, it can be an important tool in difficult budgetary times, Mr. Kaelin says.

Risk adjustment can help agencies understand relative differences in cost and utilization among plans. It can help evaluate the merit of individual health plan financial issues and rate adequacy concerns. And by paying plans based on health status, it can improve the equity and efficiency of the overall payment system.

The lessons learned from the research, he says, are that achieving the benefit of risk adjustment requires careful implementation, assessing the completeness and validity of the diagnostic data are essential, and involving health plans early in the process is important.

Mr. Kaelin says that those states that had the most cooperation from their plans accomplished that cooperation by talking a lot with the plans in advance of implementing risk adjustment so that the plans were able to identify potential data issues. There have been no reports of plans actually pulling out of Medicaid managed care because of problems with risk adjustment.

"Risk adjustment sits within the overall payment system and can maximize the efficiency of the payment system so that it is more fair and equitable," he tells State Health Watch. "But if there isn’t enough money in the system, risk adjustment isn’t likely to help."

[Contact Mr. Kaelin at (410) 455-6854 or at jkaelin@chpdm.umbc.edu.]