Influenza-associated excess mortality by age, sex, and subtype/lineage, in a subtropical city in China, 2015-2018: a population-based study with a distributed lag non-linear model (Preprint)

BACKGROUND Accurate estimation of influenza death burden is of great significance for influenza prevention and control. However, few studies have considered the short-term harvesting effects of influenza on mortality when estimating influenza-associated excess deaths by cause of death, age, sex, subtype/lineage. OBJECTIVE This study aimed to estimate cause-, age- and sex-specific excess mortality associated with influenza and its subtypes/lineages in Guangzhou from 2015 to 2018. METHODS Distributed lag non-linear models were fitted to estimate the excess mortality related to influenza subtypes/lineages for different causes of death, age groups, and sex based on the daily time-series data on mortality, influenza, and meteorological factors. RESULTS A total of 199.8 thousand death certificates were included in the study. The average annual influenza-associated excess mortality rate (EMR) was 25.06 (95% empirical confidence interval [eCI], 19.85–30.16) per 100,000 persons, among which 81.2% were due to respiratory and cardiovascular (R&C) mortality (EMR: 20.36 [95% eCI:16.75–23.74]). Excess R&C deaths in people aged 60–79 years and those aged ≥80 accounted for 32.9% and 63.7%, respectively. The average annual excess R&C mortality rates attributed to influenza A(H3N2), B/Yamagata, B/Victoria, and A(H1N1) were 8.47 (95% eCI:6.60–10.30), 5.81 (95% eCI:3.35–8.25), 6.21 (95% eCI:2.31–9.97), and 0.07 (95% eCI:-5.57–5.70), respectfully. The male-to-female ratio of excess death from respiratory diseases was 1.34 (95% CI:1.17–1.54), while the ratio for cardiovascular diseases was 0.72 (95% CI:0.63–0.82). The mortality displacement of influenza A(H1N1), A(H3N2), and B/Yamagata was 2–5 days, but 5–13 days for B/Victoria. CONCLUSIONS This study suggests that the mortality burden of influenza B cannot be ignored. Including influenza A subtypes and B lineages in active surveillance and vaccination with quadrivalent vaccines would help to curb the mortality burden of influenza. The mechanisms of sex differences in influenza-associated mortality warrant further investigation. Our findings will help to better understand the magnitude and time-course of the effects of influenza on mortality.