Understanding and Estimating Fraudulent Activities in Healthcare

Fraudulent activities are one of the pressing concerns in the healthcare system due to the financial costs of such activities. However, our understanding of these fraudulent activities is very limited; currently available estimates for healthcare fraud are within the range of 3-15%, implying a substantial amount of total economic loss. The range of available estimates also implies a high degree of uncertainty as to the extent of healthcare fraud and limited evidence for the effectiveness of available methodologies to detect healthcare fraud and estimate the rate of fraud. This concern has become more significant with the many changes occurring in the marketplace under healthcare reform given the amount of healthcare spending that is financed by public programs or government-subsidized health insurance.

In this paper, we propose a predictive modeling approach to develop methodologies for healthcare fraud management. Our methodologies are based on the following economic framework: fraud is a rational response by certain providers to misrepresent their effort on behalf of patients in order to reap monetary incentives when the probability of detection is low. Given this framework, we describe how methodologies can be used in combination with previously identified fraud predictors. We then use the Medicare 5% sample file in order to show how these methodologies can be applied with identified fraud predictors and discuss the design of empirical analysis using healthcare claims data. We also explore the implications of our study for healthcare practitioners and public policy makers in terms of reducing fraud in public programs. We conclude by commenting on the validation of fraud detection methodologies given the fact that fraud is rarely, if ever, observed and reported in a standardized manner.