Regina A Shih1; Beth Ann Griffin1; Adria Jewell1; Christine Eibner1; Mary Cushman2; Duanping Liao3; Charles Eaton4; Helene Margolis5; Eric Whitsel6
1 Chloe Bird， RAND， Arlington， VA
2 Univ of Vermont， Burlington， VT
3 Penn State Univ， Hershey， PA
4 Brown Univ Sch of Medicine， Pawtucket， RI
5 Univ of California， Davis， Sacramento， CA
6 Univ of North Carolina Gillings Sch of Public Health， Chapel Hill， NC
Introduction. Particulate matter air pollution (PM) may affect coagulation and risk of venous thromboembolic events (VTE) differently among women who do and do not take estrogenic hormones. A high-profile case-control study recently found a direct PM10-deep vein thrombosis association that was weaker among women taking hormone therapy. We replicated this study in the combined population of two， prospective， randomized， controlled trials of treatment with estrogen (E) 0.625 mg/d versus placebo and E plus progestin 0.25 mg/d (E+P) versus placebo.
Methods. The WHI HTs randomized 27，347 post-menopausal women aged 50–79 years at 40 clinical centers across the U.S. then followed them prospectively for an average of 5–7 years. We excluded women with a prior VTE， unavailable PM data， or current anticoagulant use， leaving 26，631 for this analysis. We used age-stratified Cox proportional hazard regression models to examine the association between time to physician-adjudicated， incident VTE and annual mean concentrations of PM10 and PM2.5 (?g/m3) spatially interpolated at geocoded participant addresses. We adjusted models for race/ethnicity， income， body mass index， history of cancer， history of lower extremity fracture， current smoking， randomized treatment arm， randomized treatment duration， and a randomized treatment arm*duration interaction. We also examined multiplicative interactions between randomized treatment arm and PM.
Results. Over the course of follow-up， we observed an overall rate of 2.8 VTE per 1，000 person-years and a hazard ratio (95% confidence limits) for VTE per 10 ?g/m3 increase in PM2.5 of 0.98 (0.63， 1.54). The hazard ratio associated with PM10 was lower and more precise than for PM2.5. Findings were suggestive of greater PM-associated hazards in the E+P than in the E trial， but randomized treatment arm*PM interactions were not statistically significant (P > 0.10).
Conclusions. Prospective findings from the Women’s Health Initiative Hormone Trials are inconsistent with prior， observational epidemiologic evidence of a strong， direct association between long-term PM exposure and VTE and clinically important modification by exposure to exogenous estrogens.