Modification of the effects of air pollutants on mortality by temperature: A systematic review and meta-analysis

Temperature extremes and air pollution both pose significant threats to human health, but it remains uncertain whether pollutants' effects on mortality are modified by temperature levels. In this review, we summarized epidemiologic evidence on the modification by temperature of the acute effects of air pollutants on non-accidental and cardiovascular mortality. The EMBASE, PubMed, ProQuest Dissertations and Theses, and Elsevier Science Direct databases were used to identify papers published up to 2nd December 2014. Studies with appropriate design, exposures and outcome indicators, quantitative estimates and high/intermediate quality were included. Twenty-one studies met the inclusion criteria, of which 12 reported the effects of PM₁₀ on mortality modified by temperature, 10 studied O₃, and the rest examined NO₂, SO₂, PM_2.5, PM_10-2.5, CO and black smoke. We divided temperature into low, medium, and high categories as defined in each study. In high temperature days, a 10μg/m³ increment in PM₁₀ concentration corresponded to pooled estimates of 0.78% (95% CI: 0.44%, 1.11%) and 1.28% (0.66%, 1.91%) increase in non-accidental and cardiovascular mortality, both statistically significantly higher than the estimates in medium temperature stratum. Pooled effects of O₃ on non-accidental mortality on low and high temperature days were increases of 0.48% (0.28%, 0.69%) and 0.47% (0.32%, 0.63%) respectively, for 10μg/m³ increase in exposure, both significantly higher than the increase of 0.20% (0.07%, 0.34%) on medium temperature days. The effect of O₃ on cardiovascular mortality was strongest on high temperature days with pooled estimate of 1.63% (1.14%, 2.13%). No significant interactions between SO₂/NO₂ and temperature were detected by meta-analysis. Other pollutants were not analyzed due to the lack of suitable studies. In summary, we observed interactions between high temperature and PM₁₀ and O₃ in the effects on non-accidental and cardiovascular mortality. Low temperature modified the effects of air pollutants but not in a consistent fashion: the effect of PM₁₀ oncardiovascular mortality was diminished but the association between O₃ and non-accidental mortality was strengthened.