The marginal cost-effectiveness of medical technology: A panel instrumental-variables approach

We use panel instrumental variables techniques to estimate incremental mortality and cost effects of intensive procedures for treating heart attacks among the elderly. We identify incremental effects by comparing trends in procedure use, hospital costs, and mortality between hospitals that adopted intensive technologies to corresponding trends at nonadopting hospitals. We formalize the identification assumptions required for 'difference-in-differences' estimation and present empirical evidence on their val?dity. Accounting for unobserved heterogeneity substantially reduces estimated mortality effects and additional costs of intensive procedure use. Our most conservative estimate of the average cost of each additional one-year AMI survivor associated with more intensive medical treatment is at least $40,000 in 1987 dollars; more plausible estimates are $70,000 or more.