A comparison of three approaches for attributing hospitalizations to specific diseases in cost analyses

Objectives: Calculations of healthcare costs rarely disclose the specific a pproach used to allocate the cost of hospitalizations by diagnosis. However , the type of approach used can have a major impact on the findings in the case of significant comorbidities. The present analyses compared three appr oaches for attributing Medicare DRG reimbursements (which were used as surr ogates for average costs) for hospitalization by diagnosis. Methods: Medical resource utilization data from the National Hospital Disch arge Survey were analyzed using each of three allocation approaches: a) att ributing 100% of the cost of hospitalization to the disease when it was the first-listed diagnosis; b) attributing a portion of the cost of hospitaliz ation to the disease, depending on its position in the list of diagnoses an d the relevance of any comorbidities; and c) an incremental analysis of cos t based upon the hospitalization experiences of an age and gender matched c ohort. These three approaches were applied to the cost of hospitalization f or chronic obstructive pulmonary disease (COPD). Results: The first approach projected 206,098 hospitalizations at $3,449 pe r hospitalization for a projected U.S. annual total of $711 million. The se cond approach projected 681,547 hospitalizations at $3,205 per hospitalizat ion for a projected U.S, annual total of $2.2 billion. The third approach a lso projected 681,547 hospitalizations, but at $2,361 per hospitalization, for a projected U.S. annual total of $1.6 billion. Conclusions: Expanding from the example on COPD, the limitations of each ap proach are described and their applications to other conditions are present ed.