The US decline in stroke mortality: what does ecological analysis tell us?

We review a study in this issue that concludes, from analyses of ecological associations, that the use of medication to lower high blood pressure has caused at most a small decline in US stroke mortality rates. Our analysis suggests that other possible sources of the decline may be population-wide falls in levels of blood pressure, cigarette smoking, and coronary heart disease mortality, as well as improved treatment of cardiac and respiratory sequelae of stroke. Although the ecological method is powerful for answering questions about medical interventions' population-wide effects on disease, it must be used with care. Of particular concern are variables with meanings that differ between the ecological and the individual levels, the number of ecological units available for analysis, the sample size within the ecological units, and the range of independent variables used in ecological regression.