Background and purpose: Stroke community surveillance projects often focus on hospital data rates. We hypothesized that not including strokes which occurred in nursing homes or at home would differentially affect race/ethnic stroke rates.
Methods: Texas vital statistics data were studied to compare age-specific (45 to 59, 60 to 74, and >/=75 years) location of stroke death for African Americans (AAs), Hispanic Americans (HAs), and non-Hispanic whites (NHWs). Rate ratios are reported with 95% CIs; NHW is used as the referent group.
Results: During 1991 to 1996, there were 52 996 stroke deaths in Texas for individuals aged 45 years and older. HAs in the oldest age group (>/=75 years) were 33% more likely than NHWs to die in the hospital, and HAs aged 45 to 59 and 60 to 74 years were 4% and 10%, respectively, more likely to die in the hospital. AAs aged >/=75 years were 19% more likely to die in the hospital. HAs aged 60 to 74 years were 35% less likely to die in a nursing home, whereas HAs aged >/=75 years were 43% less likely than NHWs to die in a nursing home. AAs aged >/=75 were 33% less likely to die in a nursing home. Death at home was 19% more likely in HAs aged 60 to 74 years. Significant gender differences are also reported.
Conclusions: Using hospital data alone would overestimate stroke mortality in the HA and AA groups. Stroke community surveillance projects should account for ethnic and gender differences in location of death to avoid bias in race/ethnic and gender comparisons.