A Statistical Method for Monitoring Social Insurance Claims

Abstract

A Statistical Method for Monitoring Social Insurance Claims Introduction One risk of primary concern to both private and social insurance administrators is that the costs of the programs will exceed the budgeted amount because of either increased costs of services and/or increased utilization. To social insurance programs the realization of such a risk results in exceeding state or federal funding levels for the program. In the private sector, the losses must be made up through adjusted premiums or from profits from other business. In either case, much of the effects of such a risk can be amerliorated by early detection of any change in costs per case or utilization rates of the beneficiaries. Indeed, one of the primary responsibilities of actuaries is to continuously monitor experience of each line of business and be prepared to react quickly to results of this monitoring [Shapiro (1982)]. The purpose of this article is to illustrate the use of statistical graphs and charts resembling control charts as a basis for monitoring cost and utilization. As such, the method represents a first step in controlling risk by providing a statistically based early warning system. Although this article uses the state administered Medicaid program as a basis for motivation and illustration, the methods presented here are applicable to monitoring other insurance programs. Medicaid, though funded in part by Federal dollars, is administered through state organizations. As a result, state governments have the responsibility for determining benefit levels, criteria for qualification, and budget. The escalation of health care costs during the past two decades has resulted in increased efforts by states to control Medicaid expenditures. To effect a cost control program some of the states have implemented a prospective payment system (PPS), similar to that of Medicare, as the basis for reimbursing providers for care. Under the PPS, the cost to be reimbursed for care is determined, prospectively, for each disease category (diagnosis related group or DRG). Thus, except for extreme cases, call outliers, the cost to Medicaid of care of any individual in a given DRG can be determined in advance. Based oin the average number of cases expected for the year for each DRG, these DRG costs can be used to establish a target budget. However, simply setting up a budget based on PPS costs is not sufficient to control costs or to meet the budget. The budget represents a set of target values for utilization rates, outlier payment amounts, and types of cases. These targets are established using the experience of previous years with adjustments made according to any anticipated changes. The actual experience of the health providers must be monitored regularly to assure compliance to these underlying budget assumptions. Regular monitoring of a process over time to attain a specified target is the basis of statistical process control methodology. Although most frequently considered a tool for monitoring quality in the manufacturing and production industries, the quantitative methods used in statistical process control can also be applied to the monitoring of Medicaid costs. Briefly, statistical process control methodology entails the following steps: 1. A set of specifications, or a goal, for the process to be monitored is est ablished. 2. Statistical monitoring tools are developed which reflect the goal. Usually these are simple charts on which data can be readily summarized. 3. Criteria are established for raising a flag when the data indicate the goal is not being obtained. 4. Communication lines are develop so that when a flag is raised, an assignable cause can be discovered. Social insurance reimbursement is not the same as the typical manufacturing process to which statistical process control methods are usually applied, however. For example, concepts of the process itself, specific goals, and control seem foreign to the monthly flow of Medicaid claims. …