Decline in US stroke mortality in the era before antihypertensive therapy.

BACKGROUND AND PURPOSE: This study was done to determine if reported declines in stroke mortality in the era before antihypertensive therapy are artifactual. METHODS: This study involved analyses of national and state vital statistics data using adjusted and specific rates. RESULTS: Adoption of the third revision of the International List of Causes of Death in 1921 produced an abrupt 6.6% decrease in stroke mortality rates, but otherwise, changes in disease classification systems had little effect on stroke mortality rates. Adoption of the second revision of the joint-cause manual produced a 9.2% drop in stroke death rates, but other revisions of the joint-cause selection rules had little effect. While rates for the expanding group of states in the death registration area progressively declined, rates for fixed component areas remained constant until around 1925 and then declined. Reselection of the underlying cause from aggregate multiple cause data for 1917, 1925, and 1940 using uniform selection rules confirmed a decline after 1925. Correlation analyses of rates of change for stroke and heart disease rates did not support a shift in diagnosis to explain the divergent trends. CONCLUSIONS: The apparent decline in stroke mortality rates before 1925 is an artifact of changes in disease classification systems, joint-cause selection rules, and nonrandom incorporation of states with different mortality rates into the expanding registration area. The decline after 1925 could not be explained by changes in coding systems or joint-cause selection rules or by a shift in diagnosis from stroke to heart disease.