Effect modification by PM2.5/PM10 ratio on the association between PM10 and mortality

Abstract

Introduction Since PM10 is an aggregation of smaller particles with diverse toxicity, change in its composition could lead to change in its toxicity. In Seoul, Korea, there was decreasing trend in the mass of PM for recent decades, from which PM composition might have resulted. Our previous study showed decreasing trend in PM2.5/PM10 ratio (PMR).To explore variation of PM10 effect on mortality, we tested interaction effects of PMR in Seoul, Korea, 2003-2011. Methods Generalized additive models (GAM) were applied to estimate the effects of PM10 on all-cause, cardiovascular, cerebrovascular, and respiratory mortality in Seoul, Korea, 2003-2011. Long-term time trend, temperature, relative humidity, day of week, and holidays were adjusted. To consider cumulative effects, moving average of lags up to 3 days were calculated for PM10 and PMR. PMRs are then categorized into tertile. To investigate effect modification between PM10 and PMR, interaction terms were included in the GAM. Results An increase of 10ug/m3 in daily average concentration of PM10 was significantly associated with increase in all-cause, cardiovascular, cerebrovascular, and respiratory. We found positive interactions in all-cause and cause-specific mortalities, but, it was statistically significant for only respiratory mortality. In respiratory mortality, a cumulative effect of PM10 for 0-3 days was largest in the highest tertile of PMR (p-value : 0.005). Conclusions Our results could be a supporting evidence for the temporal variation of PM10 toxicity. Considering PMR seems necessary for a more precise estimate of the health effect of PM10. Furthermore, our results implies the possibility that the health effect of PM2.5 could also be changed by the proportion of more harmful particles. Further researches considering compositions are needed to estimate the health effect of PM more precisely.