Abstract
BACKGROUND: Seasonal influenza continues to pose a significant global public health challenge owing to its considerable disease burden, making it imperative to elucidate its potential driving factors. Although the roles of meteorological factors such as temperature and absolute humidity have been extensively investigated, research quantifying the burden of ambient ozone (O₃) as an environmental determinant of influenza, particularly among vulnerable populations, remains limited. The aims of this study were to assess short-term effects of photochemical oxidants (O(x), parts per billion [ppb]) in a temperate Japanese city, quantify the associated attributable burden, and evaluate variations across subgroups to enhance understanding and identify high-risk populations. METHODS: Daily time-series data on influenza cases, photochemical O(x) levels, and meteorological factors were compiled for Kawasaki City, Japan, spanning 2015 to 2019. Using time-stratified case-crossover analyses and a distributed lag non-linear model, O(x)-influenza associations were evaluated, accounting for long-term trends, seasonality, and day-of-the-week effects, and capturing both non-linear and delayed impacts. The overall attributable fractions (AFs) and those corresponding to low and high O(x) levels were subsequently calculated. Furthermore, subgroup analyses were conducted according to influenza subtype, sex, and age group. RESULTS: Data of 163,120 influenza cases were analyzed. Overall, a U-shaped exposure-response curve was observed between O(x) concentrations and influenza incidence. The overall minimum morbidity O(x) (MMO(x)) level was 48.4 ppb (94.0%), with cumulative relative risks of 2.06 (95% confidence interval [CI]: 1.50, 2.82) and 1.47 (95% CI: 0.78, 1.51) observed at low and high O(x) levels, respectively. The burden of influenza incidence attributable to non-optimal MMO(x) level was estimated at 36.6% (95% empirical CI: 17.1, 50.9), with the majority of this burden arising from low O(x) concentrations. Furthermore, patients with influenza A exhibited higher risks and AFs than did patients with influenza B, although the differences by sex and age group were not statistically significant. CONCLUSIONS: Overall, non-optimal photochemical O(x) levels are associated with a substantial burden of influenza morbidity. These findings may inform local public health initiatives by aiding the development of targeted preparedness and prevention strategies to mitigate short-term effects of air pollution.