To the Editor.
—In their analysis of intensive treatment of AMI, Dr McClellan and colleagues1 warn us about the potential biases in the use of nonexperimental data for outcomes research. They have introduced an interesting methodology that is new to outcomes research; however, I believe they have made an important methodological error in their own analysis.In Table 1, the mortality rates for groups of patients with and without catheterization are given for 1-, 7-, and 30-day periods and for 1, 2, 3, and 4 years. These simple rates are misleading because the categorization factor (catheterization within 90 days) and the outcome (death) are not independent since patients will not have catheterization after they die. For example, most patients who died on day 1 are in the group without catheterization because catheterization is not often performed very early in the course of an AMI. Patients had to have catheterization
Davis K. Catheterization and Mortality in Elderly Patients With Acute Myocardial Infarction. JAMA. 1995;273(17):1332. doi:10.1001/jama.1995.03520410025014