To create the right incentives, states move ahead with diagnosis-based risk adjustment in Medicaid
Risk Adjust / Medicaid
March 31, 1997
Although many kinks still need to be worked out, several states, including Maryland, Colorado, Washington, Minnesota, and Michigan, are moving ahead with diagnosis-based risk-adjusted reimbursement for Medicaid populations.
The hope is that risk adjustment will reduce incentives for risk selection or "cherry-picking" by managed care plans, and, also, that it will provide assurance of adequate compensation as disabled populations are brought into managed care.
Says Michael Rothman, director of the Office of Public and Private Initiatives for the state of Colorado: "There will not be a perfect [risk adjustment] system, so don’t wait for one. Anything is better than average fee-for-service. It sends the wrong financial message to plans."
Maryland and Colorado will soon use diagnosis-based risk adjustment for both AFDC and SSI recipients in their Medicaid programs, while other states, such as Washington and Michigan, are starting with SSI clients or other special populations first.
Medicaid is an ideal testing ground for some of the newer, diagnosis-based risk adjustment methodologies, says Tony Dreyfus of the Medicaid Working Group, a group affiliated with the Boston University School of Public Health.
"Medicaid is one of the best places to start risk adjustment. Medicaid has a lot of data consolidation in terms of expenditures and diagnoses." Risk adjustment may actually be easier to implement for the SSI disabled population because these individuals’ costs are more predictable, he says. "The high-cost people are likely to be consistently high cost," he says.
Among the many choices states must make when implementing risk adjustment is which "risk adjuster" or classification system to use. These systems assign individuals to cost categories based on diagnostic information and sometimes other factors. Ambulatory Care Groups (ACGs), Diagnostic Cost Groups (DCGs), and the Disability Payment System (DPS) are among the methodologies currently being used.
Some key developments in the states:
• In June, as part of its Medicaid 1115 waiver, Maryland will begin using
diagnosis-based risk adjustment for beneficiaries except for those who are dual eligibles and those who are in long-term care institutions.
Currently, the state uses about 18 rate cells to risk-adjust payments based on gender, age, geography, and category of eligibility. The ACG system, developed at Johns Hopkins University, is being used to develop nine rate cells that will replace those used previously. The AFDC and SSI populations will have separate rate cells, and the planned payment rates range from $40 per month to $1,100.
Barbara Shipnuck, deputy secretary for health at the Maryland Department of Health and Mental Hygiene estimates that initially, one-third of individuals will have sufficient Medicaid claims data to be classified under ACG. The state will use the old methodology for the remaining two-thirds and for new enrollees until enough encounter data is available. The plans must submit additional encounter data by October 30.
"The ACG system takes into account those individuals who are essentially healthy, says Jonathan Weiner, one of its architects. It also takes into account people who do not seek any health care.
AIDS and pregnancy/delivery will be reimbursed outside that system in Maryland. The state will pay a separate capitation rate for AIDS patients and an add-on payment for pregnancy/delivery.
Other state officials say they are planning to pay for pregnancy/delivery as it occurs, rather than attempt to include it in risk adjustment, because of the difficulty in predicting who will become pregnant, and because costs do not necessarily carry over from one year to the next.
Colorado to use DPS system
• Colorado will use the DPS system both for AFDC and SSI populations. Although DPS was developed for people with disabilities, state officials say that this system can be adapted for use with the AFDC population as well. The state is using diagnostic information reported in 1996 to develop a case-mix factor for each plan. Instead of making different payments for individuals based on individual risk assessment, states can more conveniently adjust payment to plans by determining their case mix and applying a "case-mix" factor to a base rate.
• Washington state is testing both DPS and DCGs, and plans to start risk adjustment in January 1998 for SSI recipients in managed care. Gaylan Gaither of the Professional Rates Section of the Medicaid program says that he expects 30-40% of the state’s 70,000 SSI clients to be enrolled in managed care by the end of the year, and that data collection from the plans will be the biggest challenge. So far, the program is relying on Social Security Administration data on disability determination which is cross-matched with the state’s own fee-for-service data. Mr. Gaither adds that the "plans are looking forward to risk adjustment. They have a lot of questions about the SSI population, and they’re not sure about the average capitation rate."
Medicaid is an ideal testing ground for some of the newer, diagnosis-based risk adjustment methodologies—Dreyfus, Medicaid Working Group
• Minnesota has set a target date of January 1998 to implement risk adjustment for its Medicaid managed care program. Pilot projects also will be operated for the SSI population. Virginia Weslowski, of the Minnesota Department of Health, says that the state may use ACGs, DCGs, DPS or a combination of methodologies. The state hopes to have an encounter data collection/analysis system in place within the next four months, and to gather the necessary data by October. The state also wants to analyze factors other than diagnosis that might be predictive of future expenses such as program eligibility, language skills, education level, and other socioeconomic factors. In addition, the state plans to use computer simulation to predict plan behavior under risk adjustment.
• Michigan plans to do risk adjustment for children under its Children’s Special Health Care Services program. The program serves Medicaid-eligible and non Medicaid-eligible youth under the age of 21, with some conditions covered regardless of age.
Risk adjustment will be done separately for Medicaid eligibles and non-eligibles. Medicaid-eligible children tend to be higher cost, in part because they are entitled to a broad array of services, says Steve Fitton, chief of budget and reimbursement policy for the children’s special health care division. Non-Medicaid eligible children are only covered for services related to their qualifying medical condition.
Because diseases progress differently in children and adults, the DPS system has been adapted for use with children in Michigan. The children will fall into one of four clusters based on diagnosis, cost, and other factors such as whether they have private insurance. Children under age 1 and beneficiaries over the age of 20 will have their own clusters because they tend to be higher cost. The division has issued an RFP with proposals due by the end of March. The plans will be paid DPS-based risk-adjusted rates from the beginning, with a retrospective review and adjustment six months later.
Risk-sharing and reinsurance will be part of the contract, says Mr. Fitton. "We don’t want to make or break people, or leave too much money on the table ... We’ve built in a fair amount of protection on the risk side."
Untested system
Health plans in general tend to support the concept of risk adjustment, but vary in their responses when it comes to implementation. One HMO representative in Maryland is concerned about the broad implementation of a system she believes is untested. "It is a giant experiment," says Martha Roach, executive director of the Maryland HMO Association. "None of these systems is tested. It’s important that people be put in the proper risk category, and we have absolutely no ability to know if anyone is doing it right."
"I understand [the plans] are nervous," responds Ms. Shipnuck. "Maryland is fortunate to have a high penetration rate among Medicaid recipients in managed care. The population has had experience in managed care, and we’re in a better position to move a large population forward. We’re looking at one-third of the population being subject to ACGs, and the other two-thirds being on the system that the HMOs are operating under now." Ms. Shipnuck says rates will be re-adjusted, if necessary, "when we have proof that the rate is not appropriate."
In general, says Ms. Tollen of Colorado, plans should be involved in the development of risk adjustment from the beginning. "The plans have been very involved from day one, so that they can’t come back and say we did it wrong. . . It would be
stupid of us not to have them there."
Method of implementation is key
Richard Kronick, who helped develop DPS, says that the choice of a methodology is not as important as how it is implemented. "You’re not likely to go too far wrong with any of the systems. The main determinants of success are whether the state works cooperatively with the health plans, whether the data is good, and whether a state can deal with rapid turnarounds in case loads and changes in the ways diagnoses are reported. These are problems with any risk adjustment system."
—Jana Sansbury
Contact Mr. Rothman 303-866-3327; Mr. Shipnuck at 410-767-4664; Mr. Fitton at 517-335-8950; Ms. Weslowski at 612-282-6339; Mr. Kronick at 619-534-4273; Mr. Gaither at 360-664-2653; Ms. Tollen at 303-866-3327; Mr. Dreyfus at 817-437-1550, and Ms. Roach at 301-858-5551.
To create the right incentives, states move ahead with diagnosis-based risk adjustment in Medicaid
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content
