Mining medical claims data with exploratory to confirmatory statistical methods

Medical billing errors constitute a major problem in the medical payment system in the USA. Little is known about how to identify the predictors that influence the error rate in hospital billings. Health services researchers have concentrated to a large extent on applying exploratory statistical methods to identify the patterns of care and analyse the variation in health services use. Despite the importance of medical claims data for detecting billing errors or fraudulent billing practises, little is known about the extent to which variations in the frequency, types, and seriousness of deficiencies reflect differences in the quality of care or broad systemic differences in the provision and use of health services. To address that gap, we propose an evidence-based approach to mining claims data to not only identify patterns of care, but also examine how individual, organisational and contextual factors may influence a particular pattern of Medicare/Medicaid fraud or abuse.