Washington State may risk-adjust pay to plans based on health status of public employees
WA Risk-Adjust Plan
February 28, 1997
Health plans that want to provide coverage to Washington state’s 300,000 public employees may soon have to have to work with yet another factor during the process of developing their reimbursement rates —the health status of their subscribers.
Jenny Hamilton, senior health policy analyst for the state’s Health Care Authority (HCA), says it is "highly probable" that some form of health status-based risk adjustment will be implemented for the 1998 contracting cycle.
HCA officials are currently doing a simulation of how payments to the plans would change under health status-based risk-adjustment. Some of the issues that will be addressed during the simulation include budget neutrality for the HCA and the impact of health-status-based risk adjustment on enrollment and subscriber premiums. The Request For Proposals (RFP) for the 1998 contracting year is scheduled to go out in May.
The HCA has been using age and gender of enrollees to adjust payments to its participating plans for several years. But, demographic characteristics only predict about 5% of a plan’s risk, according to Ms. Hamilton, while health-status-based measures predict 20% to 30% of a plan’s risk.
"Age and sex are only threshold indicators," she says. "Health status-based measures are much better predictors." But, risk adjustment is not intended to predict 100% of a plan’s risk, Ms. Hamilton says, because doing so would remove the incentive for plans to manage care efficiently in the same way as the fee-for-service environment does.
HCA testing two systems
The HCA is testing two risk-adjustment systems: diagnostic cost groups (DCG), and ambulatory care groups (ACG). Both systems use inpatient and ambulatory diagnostic information (ICD-9-CM codes) to categorize individuals and the cost of their care. The agency may choose one of the two models or use a combination of the systems. DCGs assign individuals to a hierarchy of cost categories based on their previous year’s diagnoses. Each category includes different diagnoses with similar spending patterns. ACGs use diagnoses as well as other factors such as age and health status to group individuals into 52 mutually exclusive categories. An example would be "an individual between the ages of 2 and 5 with one or more acute minor conditions."
Whatever risk-adjustment system is used, HCA will make risk adjustments prospectively, avoiding a situation where some plans end up owing a large amount of money at the end of the contracting year and other plans build up a credit. Use of retrospective risk-adjustment payments can lead some plans to reconsider their participation in the program as recently occurred in California.
The Health Insurance Plan of California (HIPC) has had a risk-adjustment mechanism in place for a year for the 28 plans that participate in its state-run purchasing alliance. When risk-adjustment calculations were done at the end of the year, five plans were found to owe money back to the system, and one was slated to receive money. Two HMOs decided to drop out of HIPC after learning that they would owe $400,000 in "risk-transfer" payments. The other plans that owed money decided to stay in, even though one, Kaiser Permanente, owed $453,000. Sandra Shewry, executive director of California’s Major Risk Medical Insurance Board (MRMIB) says it is an encouraging sign that those plans continue to participate in the program.
She added that describing plans as "winners" and "losers" in risk adjustment is inaccurate. "It’s not really fair to characterize the paying plans as losers, because they’re actually experiencing better than average risk selection. Everyone wants to be a payer rather than a receiver. It’s always better than being the plan that experiences worse-than-average risk selection, because we’re not sure we’re able to make up all the difference with transfer payments."
"It is unlikely that plans will be underpaid, as long as they have good actuaries to help them set their rates," says James Matthisen, a consulting actuary for HCA from Foster Higgins in Seattle about HCA’s proposed system.
"The assumption is that since plans will know and understand the expected adjustment, they can get their bids in line with that. But he adds, "everyone is happy until the winners and losers are identified. We’ll see what happens when the dollars get
stated."
Originally, state officials had hoped to complete the simulation of payments by the end of 1996, but concerns about confidentiality of data prompted the state to hire a firm to "encrypt," or encode, the data at the plan level before releasing it for the simulation.
Outpatient data
Health plan executives, for the most part, praised the way HCA has moved toward health-status-based risk adjustment, acknowledging that it is probably an inevitable step in managing health care competition. But, still, they voice some concerns about data collection.
Previous risk adjustment efforts have relied heavily upon inpatient data, but experts agree that ambulatory data is critical to ensure accuracy. Sue Skillman, project director at the University of Washington, which is helping the HCA to refine a formula for risk-adjustment, says that although inpatient data is easier to collect and has greater consistency of coding, using it exclusively does not encourage plans to manage care effectively. "In effect, you promote hospitalization, because plans will be paid more if they have more inpatient hits’ on their data."
Ambulatory data historically has been more difficult to obtain. "There has been no financial incentive for plans to collect it," Ms. Skillman notes. "However, this will be an incentive. Our theory is: Give them an incentive, and they will come.’ She adds that plans have other incentives to collect this data, such as winning accreditation and measuring their performance under HEDIS. "The whole world is focused on data, and we continue to expect it to get better."
"Our number one concern is that the outcomes will be impacted not by real health status, but by the way different carriers capture their data," says Tom Beecken, director of product line services for the Group Health Cooperative of Puget Sound, which covers about 80,000 state employees. If done improperly, risk adjustment will adversely impact both plans and subscribers, but not the state, he says. Mr. Beecken also is concerned that the "timetable is aggressive and a little bit unrealistic." However, he endorses the idea of risk adjustment "one hundred percent."
Bob Power, senior director of HealthPartners, Inc., in Minnesota and a member of the advisory team for the Washington state project, says that "the typical plan manager would not want risk assessment done, because it’s a change from the status quo. When you shake up the status quo, there’s risk. It’s a business risk that they didn’t have before. But I don’t think it’s a huge risk."
Mr. Power adds that plans can become knowledgeable enough to predict where they will end up in risk adjustment. "The technology of risk adjustment is sufficiently in the public domain. It’s not a black box." He adds that anyone willing to devote the time is capable of understanding its likely effects."
The state of Maryland will soon be using ACG-based risk adjustment in its Medicaid program, and a few other states —Washington and Minnesota—are using the ACG model on a pilot basis. In addition, the Buyers Health Care Action Group, an employer coalition in Minneapolis-St. Paul is using ACG to adjust rates to provider systems. Some 90 private organizations are using ACGs either for profiling purposes or to adjust capitation rates. Finally, 10 HMOs are using ACGs to adjust rates paid to physicians.
—Jana Sansbury
Contact Ms. Hamilton at 360-923-2729; Mr. Beecken at 206-448-5205; Mr. Matthisen at 206 233-2400; Ms.Skillman at 206-543-355;, Ms. Shewry at 916-324-4695 or Mr. Power at 612-883-6534.
Washington State may risk-adjust pay to plans based on health status of public employees
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content
