A Stratified Sampling Plan for Billing Accuracy in Healthcare Systems
Assistant Professor, University of Michigan-Dearborn
Professor and Chair, University of Michigan-Dearborn
Instructor, Kasetsart University, Thailand
This study presents a stratified sampling plan for estimating accuracy performance of the billing process in healthcare systems. The proposed sampling plan implements a stratification method that determines the number of strata, their boundaries, sample allocation, and accuracy estimates for the billing population with unknown strata structure.
This study presents a stratified sampling plan for estimating accuracy of billing performance for the claims submitted to third party payers in healthcare systems. The population consists of hospital claims with amounts ranging from zero, hundreds, thousands, to rare high million dollars. Accuracy of the billing process is estimated by auditing a sample of claims with two measurements: the overall percent accuracy and the total dollar accuracy. Difficulties in constructing the sampling plan arise when the number of strata and their boundaries are unknown, and when the two measurements require different sampling schemes. The proposed sampling plan is designed to perform effectively for estimating both measurements. It determines an overall sample size and tests various numbers of strata to find an appropriate stratification. The optimal stratum boundary points are found using the rectangular stratification method on the claim dollar amounts. The overall sample is then assigned to strata with a mixed strategy between the proportional and optimal allocations and finally the accuracy estimates and their precisions are obtained. The sampling plan is tested on an actual population obtained from insurance industry with simulated claim errors. The results show effectiveness of the plan for both accuracy measurements.