Dollars and sense: a practical guide to cost analysis for hospital epidemiology and infection control.

This paper explains practical approaches for collecting inpatient cost data for cost-of-illness and cost-effectiveness analyses. The economic definition of cost of an item is the value of the resources that are consumed in its production. Cost analysis should collect the resources hypothesized to be affected by the illness or intervention. The dollar value of these resources can also be estimated. Diagnosis-related group (DRG) reimbursements are not helpful when all study patients have the same DRG or when no DRG exists (e.g., nosocomial infection). Hospital charges are not a good surrogate for costs. Hence, data needed include resources used, charges, and cost-to-charge ratios, so that cost can be estimated. Resources used can be obtained from hospital information systems. For some resource use (e.g., physician services, pharmacy, and intravenous fluids), charges or cost-to-charge ratios may not be available, and an external standard may be needed to estimate the dollar value. For many types of resources, hospital financial systems provide both charges and cost-to-charge ratios. This yields an estimate of average cost (total cost divided by patient days) when marginal cost (change in variable cost per day of patient stay) is a better estimate of the value of the resources consumed. However, cost-to-charge ratios remain the only practical way of estimating cost in many circumstances and are commonly used in economic studies. Cost-of-illness estimates vary among the various nonrandomized study designs used. "Real-world" randomized trials are potentially useful to obtain advantages of randomization but avoid the protocol-induced biases of traditional double-blind controlled trials.