Coronary heart disease and estrogen replacement therapy. Can compliance bias explain the results of observational studies?

The overall risk/benefit of estrogen replacement therapy (ERT) is strongly dependent on assumptions about the effect of ERT on coronary heart disease (CHD). The belief that ERT causes a substantial reduction in the risk of CHD is widespread. The studies that provide support for this belief are all nonexperimental ones. Three analyses of data from two randomized clinical trials of drug treatments for CHD have examined the association of compliance with total mortality in persons who complied with the taking of placebo. In these analyses, compliance with the taking of a placebo was associated with a reduction in overall mortality of the same magnitude as the reduction in the risk of CHD in users of ERT. The benefit of compliance with placebo was not reduced by adjustment for a large number of variables, both medical and sociodemographic, that might affect mortality. Users of ERT are compliers, and the possibility that compliance bias may account for some of the apparent benefit of ERT for CHD must be taken seriously.