Factors contributing to declines in cardiac mortality.

To the Editor. —Dr Hunink and colleagues 1 are to be commended for their effort in attempting to model the various contributions of preventive and secondary modalities on the observed decline in mortality associated with coronary heart disease (CHD) from 1980 through 1990. However, their application of the model extends too far into the realm of causal inference, in which mathematical models have a limited role. We believe that their inferential procedures and logical stance are in error. The authors' argument may be summarized in the following syllogism: Their major premise is that reductions in CHD mortality are caused either by changes in risk factors or by changes in treatment; their minor premise is that only 50% of the reduction in CHD mortality can be attributed directly to risk factor changes; and their conclusion is that a large portion of prevented deaths are attributable to improvements in treatment. The major