Vulnerability Among the Elderly to Extreme Heat-Associated Cardiovascular Mortality in Michigan, 2000-2009

Background: A better understanding of which populations are most vulnerable to the health effects of extreme heat will help communities adapt to climate change. Previous research has suggested that vulnerability is enhanced among individuals with certain socio-demographic characteristics, e.g., black race, advanced age, low educational attainment and low income. Additionally, vulnerability may be increased among individuals who lack adequate transportation to reach cooler locations; who reside in older homes that may have inadequate weatherproofing; and/or who live amidst areas with high levels of impervious surfaces, which increase ambient temperature. Cardiovascular mortality, particularly in cool climates such as Michigan, has been associated with extreme heat.

Objectives: Our objectives were 1) to determine if the extreme-heat-cardiovascular mortality association was present among Michigan elderly in 2000-2009 and 2) to determine which characteristics modified the association between extreme heat and cardiovascular mortality.

Methods: We used the following data: 1) geocoded death records from ten Michigan counties from the Michigan Department of Community Health (including date and causes of death, age, race and educational attainment), 2) daily mean temperature and humidity from the nearest airport weather station from the National Climatic Data Center, 3) Census block group characteristics from the 2005-2009 American Community Survey (median income among individuals aged 65 years and older and percents of residents in poverty, homes built before 1940 and residents aged 65 years and older without a vehicle), and 4) National Land Cover Data Set percent imperviousness (averaged at radii of 90m, 150m, 300m, 600m and 1.5km). For each county we calculated daily apparent temperature, which is based on temperature and dew point. We defined extreme heat exposure as four-day mean apparent temperature at or above the 97th percentile of four-day mean apparent temperature for that location from 2000-2009. We employed a time-stratified case-crossover design with conditional logistic regression to assess the association between cardiovascular death and extreme heat, and we assessed vulnerability as interactions between extreme heat and each characteristic of interest in individual models. Significant interactions were then included together in a single model. Continuous measures were mean centered and standardized for an interquartile range increase.

Results: In models with only a single interaction, the odds of cardiovascular mortality were increased during extreme heat vs. non-extreme heat among individuals of black race (19%, 95% Confidence Interval: 9%-31%), residing in areas with high (68%) percent imperviousness within 1.5km (13%, 95% CI: 5%-22%) and residing in block groups with: high (46%) proportions of homes built before 1940 (26%, 95% CI: 8%-48%) and high (43%) proportions of households with at least 0.51 occupants per room (33%, 95% CI: 5%-68%). In models with all these interaction terms included simultaneously, these effects were all attenuated and non-significant. In a model with all these terms except black race, the effects of homes built before 1940 and imperviousness within 1.5km remained significant.

Conclusions: Surrounding imperviousness and living in an older home may confer vulnerability to extreme heat in Michigan. Increased vulnerability among individuals of black race may be at least partially mediated by characteristics of the built environment. Future adaptation efforts should improve housing quality and decrease the effects of impervious surfaces on ambient temperature.