Abstracts Health Sciences

A case-crossover study of particulate matter air pollution and out-of-hospital primary cardiac arrest in King County, Washington

by Drew Griffin Levy

This dissertation evaluates the association between the incidence of out-of-hospital primary cardiac arrest (OHPCA) and daily measures of PM in a case-crossover study design. Time-series regression analyses were first conducted to determine if results seen elsewhere could be replicated locally in the same study base as the proposed case-crossover analysis. Time series analyses of the cardiac arrest case series were also conducted to corroborate the results of the case-crossover analyses. No evidence of an association between PM levels and total mortality (relative risk for a 10 mug/m 3 change in PM10 = 1.004; 95% confidence limits: 0.999, 1.008), IHD mortality (relative risk for a 10 mug/m3 change in PM10 = 1.008; 95% confidence limits: 0.999, 1.016), or out-of-hospital primary cardiac arrest (relative risk for a 10 mug/m3 change in PM10 = 1.003; 95% confidence limits: 0.909, 1.10) is found.Simulations were performed to evaluate the bias associated with various case-crossover referent selection strategies. Unbiased estimates using conditional logistic regression are obtained by defining the referent sampling window for each case to be non-overlapping time strata, brief enough in duration to be free of substantial seasonal effects, in which all subjects are separated by a period which precludes substantial autocorrelation.Case-crossover study subjects were all cases of OHPCA attended by paramedics identified in King County from October 1988 though July 1994, who were married county residents, 25 to 74 years of age, with no prior history of clinically recognized heart disease or life-threatening comorbidities (N = 362). Daily averages of regional monitoring data for nephelometry measures of PM (bsp) and PM10 from three monitoring sites were used. PM measures associated with case events were compared with PM measures from the same day of week in the same month and year using conditional logistic regression. The estimated relative risk at a lag of one day for an interquartile range (IQR) change in nephelometry measures (0.5 bsp) was 0.893 (95% C.I., 0.779 – 1.024). The estimated relative risk at a lag of one day for an IQR change in PM 10 (19.3 mug/m3) was 0.868 (95% C.I. 0.744 – 1.012). The 95% confidence intervals for all other lags also included unity.