Short-term Effects of Ambient Ozone on Stroke Risk in South Carolina, USA

摘要

Background: Findings are inconsistent regarding the relationship between stroke and acute ozone exposures. Questions also remain regarding effect modification by race or gender. Objectives: We explored the association between acute ozone exposure and risk of stroke hospitalization among South Carolina (SC), USA residents and effect modification by race and gender. Methods: We conducted a case-crossover study with preliminary data on all in-state ischemic and hemorrhagic stroke hospitalizations for SC adult residents from 2002-2006. Maximum 8 hour daily ozone exposure estimates for 15 days prior to each hospitalization were assigned to each residential address based on daily gridded values from the US Environmental Protection Agency's (EPA's) Hierarchical Bayesian Model Derived Estimates. Selected periods across lagged days 0-7 prior to the hospitalization served as the case period of exposure, with analogous referent periods selected from lagged days 8-15. To estimate odds ratios and 95% confidence intervals, we utilized conditional logistic regression models. Results: There were 23,823 stroke hospitalizations with complete data, of which 52.6% were among females, 29.8% among African-Americans and 68.9% among Caucasians. The mean age was 68.6 (SD: 13.8) years. Median ozone for the case and control periods were 46.4 ppb (IQR: 39.5,52.8) and 45.8 ppb (IQR: 38.5,52.9) respectively. A 10 ppb increase in ozone exposure on the day prior to hospitalization (OR: 1.05; 95% CI, 1.03,1.07) as well as the cumulative average of 3 days prior to hospitalization (OR: 1.07; 95%CI: 1.05,1.10) was associated with increased risk of stroke hospitalization. There was no evidence of effect modification by race or gender. Conclusions: Preliminary results showed an increased risk of stroke hospitalization associated with acute ozone exposures among SC residents. Race nor gender modified the association. Further research will include exposure estimate validation and model adjustment for potential confounders that vary acutely.