Short-term meteorological conditions and laboratory-confirmed influenza A in children: a time-series study with adjustment for testing volume

Abstract Background Hospital testing databases are widely used to study environmental drivers of influenza, but case counts are confounded by testing volume. We aimed to describe the temporal distribution of influenza A positivity in children and to screen short-term meteorological and air pollution exposures. We also distinguished absolute positivity burden from the probability of a positive result among those tested. Methods We enrolled children aged 15 years or younger who underwent multiplex respiratory pathogen testing between 10 November 2024 and 30 November 2025, and built a daily time series anchored to illness onset. A time-stratified conditional quasi-Poisson model, stratified by year, month and day of week, used log(daily influenza A tests) as an offset. Thirty-five meteorological exposures (lag 0–7 d) and four air pollutants (lag 0–2 d) were standardised per 1 SD, with Benjamini-Hochberg correction for multiple comparisons. Results Among 1675 children, 306 were positive for influenza A (18.3%); positivity peaked in December (149/322, 46.3%) and no cases were detected between March and September. After global correction, low convective available potential energy (CAPE) showed the strongest signal (rate ratio 0.103 per 1 SD increase, 95% CI 0.049–0.218, q < 0.001); the maximum thunderstorm index and thunderstorm index also reached q < 0.05. CAPE was negatively associated with the number of positive cases, the positivity rate among those tested and in the offset model, but not with the total number of tests. Mean temperature was negatively associated with both positive cases and total tests, but not in the positivity-rate or offset models. CAPE was consistent across the full observation window, restriction to onset-to-sampling of 7 d or less, sampling-date substitution and case-crossover analysis. Conclusions In this single-centre, single-year cohort, low atmospheric instability, particularly low CAPE, was consistently associated with increased influenza A positivity in children and could not be explained by testing volume alone. The association between mean temperature and the absolute number of positive cases weakened markedly after denominator adjustment. CAPE is more plausibly a proxy for the cold-season synoptic situation than a direct cause, and these exploratory findings require multi-year, multi-centre validation.