Abstract
We developed a stochastic model, Proportional Distribution Method (PDM) to estimate disease-specific costs in health insurance claims with multiple diagnoses in 1996. PDM assumes a common magnitude for each diagnostic category and distribute the cost of a claim in proportion to the magnitude for each category. We demonstrated previously that, by using arithmetic means of per diem costs of claims containing a certain diagnosis with proper correction as magnitudes, PDM was able to estimate disease-specific costs in computer-generated simulation data which mimics health insurance claims with multiple diagnoses. In this article we proposed a yet another method of magnitude estimation using Excel® Solver function for optimization and refined the established method of arithmetic means with correction by introducing a correction formula for automatic correction. A Monte Carlo simulation using 100 datasets each consisting of 1000 cases with 100 diagnostic categories, which bears little resemblance to actual claims, demonstrated that PDM had achieved near-perfect accuracy by the method using Excel® Solver function and less-perfect but acceptable accuracy by the method of arithmetic means with correction. PDM is also effective in estimating disease-specific days (in and out patient) in health insurance claims but only costs are dealt with in this article.
