Rate Setting for Managed Care

Investigators: Carole Siegel, Ph.D., Kristine Jones, Ph.D., and Judith Samuels, Ph.D.

PROJECT GOALS
The aims of this project are to design and test reimbursement methodologies for capitation or managed care scenarios emerging under state Medicaid and national health care reform. This project is composed of two studies, each using an individual approach to developing reimbursement structures for managed care. Both approaches take into account the large variation in recipient costs and aim to reduce the financial risk to payers and/or providers. The first study is a continuation of research on developing a risk adjustment payment methodology that does not depend on patient characteristics, a blended payment methodology for psychiatric care; the second is a study to develop risk groups that are intended to be used to risk adjust capitation rates based on patient characteristics. Both studies utilize data on recipient characteristics and mental health services use in Monroe and Livingston counties in NYS. The comprehensive data set includes nearly 25,000 mental health recipients and covers a three-year period from 10/1/90 to 9/30/93. It includes community mental health services licensed by NYS as well as inpatient and outpatient services at the state-operated Rochester Psychiatric Center. Private psychiatric services, which represent a small proportion of overall services utilization, are not included.

SUB-STUDY 1: Blended Payments for Psychiatric Care. Investigators: Carole Siegel, Ph.D., Kristine Jones, Ph.D. and Shang Lin, Ph.D.

PROJECT GOALS
A reimbursement methodology for managed care based on the Integrated Risk Rate (IRR), as developed by Dr. Siegel and her colleagues, is being adapted and its utility tested. It was originally designed to mitigate some of the less desirable effects of applying a diagnosis-related group method to reimbursement for psychiatric inpatient care. In a managed care framework, the structure would blend actual costs and a capitation rate and could also include average costs of a provider network. The blending formula takes into account the variance in costs of a population. The payment model is based on a hierarchical Bayes approach. The IRR will be tested in simulations of mental health managed care organizations designed to reflect current thinking and proposals in NYS for the reform of the public mental health system.

RESEARCH ACTIVITIES AND RESULTS
Managed care scenarios have been chosen to serve as prototypes of models expected to be implemented in New York State for adults receiving Medicaid and to represent scenarios under way in several states in which managed care has already been implemented. Scenarios were chosen to cover situations in which there is the possibility of competition among providers. On the benefit side, the scenarios will vary by the inclusion/ exclusion of state hospital care and by partial carve-out/full carve-out assumptions.

One scenario of managed care has been tested and favorable results were found for the IRR payment. The scenario has modeled the plans of the NYS Office of Mental Health for carve-out special needs plans for persons receiving Medicaid who are diagnosed with serious and persistent mental illness. A single payer, such as Medicaid, is assumed to contract with three managed care entities to provide services to this population in an area. Three response situations related to patient selectionrandom enrollment, a selection bias based on utilization groups, and a selection bias according to diagnosiswere considered and for each the impact of IRR was compared to that of payment by a capitation rate. In all cases, the IRR shrunk the payment error and reduced the magnitude of profits and losses. Work is ongoing to develop an implementation procedure for IRR to bring it closer in form to already accepted payment approaches. Further work will also examine the potential for gaming the payment.

In Year 4, research activities involved specifying alternative payment structures for risk
adjustment especially those currently used or being considered and determining reasonable estimates for the parameters of these structures that would act to promote client access while maintaining proper efficiencies. Thus, in Year 4, a model was developed to examine a capitation payment with a stop loss feature. A stop loss payment is one that sets an initial capitated rate that is less than expected costs but makes an additional payment if an enrollee's actual costs exceed a specified threshold. Several stop loss payments were examined based on budget neutrality constraints. Parameters that could vary included the initial capitation rate (set as some percent of average costs) and the cost sharing rate beyond the stop loss point. Budget neutrality constraints act to determine the stop loss point. Using simulated enrollment into managed care organizations (MCOs) with different selection rules, we compared the stop loss payments to payments based on capitation rates equal to average costs. In all cases, patient level error based on squared error loss was substantially reduced.

SIGNIFICANCE OF FINDINGS/POLICY IMPLICATIONS
Risk adjustment methods are clearly needed if fixed pricing methods for the reimbursement of mental health care costs are to be used to control or contain costs. The integrated risk payment has been designed to provide a fairer way of reimbursing providers or setting capitation rates for managed care entities that cover persons receiving mental health services. Within a budget-neutral framework, the payment works to ensure provider viability and to avoid the incentives for providers to "cherry pick" or to reduce the provision of appropriately required services.

Project-Related Publications and Presentations

Samuels, J., Siegel, C., and Jones K. Managed care rate setting: New methods for mental health reimbursement. Accepted for publication in Advances in Health Economics and Health Services Research, 1997.

Samuels, J., Siegel, C. and Jones, K. Managed care rate setting: New methods for mental health reimbursement. Presented at the National Institute of Mental Health Biennial Research Conference on the Economics of Mental Health in September, 1996.

SUB-STUDY 2: Grouping Recipients of Mental Health Services for Risk-Adjusted Capitation. Investigator: Judith Samuels, Ph.D.

PROJECT GOALS
The goals of this project are to (1) describe the utilization patterns of persons using psychiatric services over a three-year period, and (2) develop risk adjustment groupings useful for adjusting capitation rates of managed care scenarios emerging under state Medicaid and national health care reform.

Utilization patterns will be described for the period 10/1/90 to 9/30/93, and risk adjustment groupings useful for adjusting capitation rates of managed care scenarios will be developed. The descriptive analysis will uniquely describe a multi-year experience across all payers. It will also shed light on the relationship between service utilization and population characteristics such as demographics, type of insurance, diagnosis and chemical abuse comorbidities across a broad population of recipients.
Recursive partitioning, a nonparametric statistical methodology, is being employed to develop and test prediction rules for grouping recipients into service utilization risk categories.

RESEARCH ACTIVITIES AND RESULTS
Risk-adjustment grouping analysis was completed during the 96-97 year. The study showed that severity of mental illness and disability were the best predictors of high mental health care costs over three years. Use of the 3-year model based risk groups formed from three years of utilization data resulted in an 18% reduction in variance of costs. Single year models did not perform as well. Analysis and conclusions provide strong evidence of the potential pitfalls associated with predicting "heavy users" incorrectly and the potential fiscal dangers of inaccurate rate setting particularly for this group.

SIGNIFICANCE OF FINDINGS/POLICY IMPLICATIONS
This study provides insight into the cycles and patterns of mental health services utilization and mental illness, which will be valuable for calculating rate schedules differentiated by patient risk, whether reimbursement payments are made on fee-for-service, diagnosis-related group, capitation or any other basis.

Publications and Presentations

Papers

Samuels, J., Siegel, C., and Jones K. "Managed care rate setting: New methods for mental health reimbursement", accepted for publication in Advances in Health Economics and Health Services Research, 1997.

Samuels, J. "Risky Business: Classifying Mental Health Consumers for Capitation Reimbursement Under Managed Care," dissertation accepted at New York University. December 1996.

Presentations

Samuels, J., Siegel, C. and Jones, K. Managed Care Rate Setting: New Methods for Mental Health Reimbursement presented at:

i) National Institute of Mental Health Biennial Research Conference on the Economics of

Mental Health, September 1996

ii) 124th Annual Conference of the American Public Health Association, New York City, November, 1996.

Projects completed.

[Top]