Case-crossover analysis of short-term particulate matter exposures and stroke in the health professionals follow-up study

Abstract

Background: Stroke is a leading cause of morbidity and mortality in the United States. Associations between short-term exposures to particulate matter (PM) air pollution and stroke are inconsistent. Many prior studies have used administrative and hospitalization databases where misclassification of the type and timing of the stroke event may be problematic.

Methods: In this case-crossover study, we used a nationwide kriging model to examine short-term ambient exposure to PM₁₀ and PM_2.5 and risk of ischemic and hemorrhagic stroke among men enrolled in the Health Professionals Follow-up Study. Conditional logistic regression models were used to obtain estimates of odds ratios (OR) and 95% confidence intervals (CI) associated with an interquartile range (IQR) increase in PM_2.5 or PM₁₀. Lag periods up to 3 days prior to the stroke event were considered in addition to a 4-day average. Stratified models were used to examine effect modification by patient characteristics.

Results: Of the 727 strokes that occurred between 1999 and 2010, 539 were ischemic and 122 were hemorrhagic. We observed positive statistically significant associations between PM₁₀ and ischemic stroke (OR_lag0-3 = 1.26; 95% CI: 1.03-1.55 per IQR increase [14.46 μg/m³]), and associations were elevated for nonsmokers, aspirin nonusers, and those without a history of high cholesterol. However, we observed no evidence of a positive association between short-term exposure to PM and hemorrhagic stroke or between PM_2.5 and ischemic stroke in this cohort.

Conclusions: Our study provides evidence that ambient PM₁₀ may be associated with higher risk of ischemic stroke and highlights that ischemic and hemorrhagic strokes are heterogeneous outcomes that should be treated as such in analyses related to air pollution.

Keywords: Air pollution; Case-crossover; Particulate matter; Stroke.

Figure 1.

Odds ratios and 95% confidence intervals per change in IQR, and p-values for tests of heterogeneity for associations^a between 4-day average particulate matter exposures (Lag0–3) and ischemic stroke by potential effect modifiers for men in the Health Professionals Follow-up Study ^aORs calculated from conditional logistic regression models and presented by change in PM IQR (PM_2.5IQR =8.91 μg/m³; PM_10IQR = 14.46 μg/m³). All ORs control for mean daily temperature. p-value from chi-sq tests of model heterogeneity ^b Self-report from biennial questionnaire: For diabetes, hypertension, high cholesterol: ‘yes’ or ‘no’ is defined as ever having been diagnosed by health professional. For current aspirin use ‘yes’ is defined as 2+ tablets /week

Figure 2.

Odds ratios and 95% confidence intervals per change in IQR, and p-values for tests of heterogeneity for associations^a between 4-day average particulate matter exposures (Lag0–3) and hemorrhagic stroke by potential effect modifiers for men in the Health Professionals Follow-up Study ^aORs calculated from conditional logistic regression models and presented by change in PM IQR (PM_2.5IQR =8.91 μg/m³; PM_10IQR = 14.46 μg/m³). All ORs control for mean daily temperature. p-value from chi-sq tests of model heterogeneity ^b Self-report from biennial questionnaire: For diabetes, hypertension, high cholesterol: ‘yes’ or ‘no’ is defined as ever having been diagnosed by health professional. For current aspirin use ‘yes’ is defined as 2+ tablets /week