Inter-mortality displacement hypothesis and short-term effect of ambient air pollution on mortality in seven major cities of South Korea: a time-series analysis

Background Inter-mortality displacement (IMD) between cause-specific mortalities has not been introduced in air pollution epidemiology. Investigation into IMD would provide insights on the actual health burden of air pollution and interpretation of associations. We aimed to investigate IMD regarding short-term effect of air pollution on mortality. Methods We illustrated manifestations and interpretations of lag-mortality associations. If IMD exists, a net increase of one cause-specific death can be offset by a net decrease of other cause-specific deaths. We conducted a time-series analysis to estimate associations of ambient particulate matter smaller than 10 µm (PM10), ozone (O3), sulphur dioxide (SO2), nitrogen dioxide (NO2) and carbon monoxide (CO) with mortality, considering lags up to the previous 45 days, for seven major cities of South Korea from 2006 to 2013. Attributable mortality cases were identified. Results For O3, respiratory mortality [11 929 cases, 95% empirical confidence interval (eCI), 5358, 17 688 cases] was counterbalanced by cardiovascular mortality (-11 272 cases, 95% eCI: -22 444, -629 cases). All-cause mortality was 37 148 cases (95% eCI: 4448, 68 782 cases). For PM10, respiratory deaths were 9167 cases (95% eCI: 563, 16 521 cases), and cardiovascular deaths were 6929 cases (95% eCI: -11 793, 24 138 cases). Estimates for SO2 were comparable to those for PM10. All-cause mortality attributable to NO2 was explained by short-term mortality displacement. No associations with mortality were found for CO. Conclusions IMD may exist in the relationship between air pollution and mortality. The actual relationship between air pollution and cause-specific mortality may be masked by IMD.