To control COVID-19, Hong Kong, China, put in place several public health and social measures (PHSMs), including mandatory mask wearing, school closures, hand hygiene, and avoidance of gatherings. In early 2020, those measures also reduced influenza transmission (1), and according to laboratory surveillance records, influenza virus did not circulate in the community for 3 years (2). From mid-2022 through 2023, PHSMs were progressively relaxed, and on March 1, 2023, the local mask mandate was lifted. We investigated the effects of PHSMs on influenza transmission in Hong Kong.

Figure

Figure. Weekly influenza activity and several preventive measures against COVID-19 in Hong Kong during 2020–2023. The blue line represents the weekly influenza activity, measured by the product of influenza-like illness rates...

We collected weekly influenza-like illness consultation rates reported by private general practitioners and the weekly proportion of sentinel respiratory specimens that tested positive for influenza virus in Hong Kong during October 2010–May 2023. We established a proxy for influenza virus activity by multiplying rates of influenza-like illness by the proportion of influenza-positive samples following previous studies (3,4) (Appendix). We found that weekly influenza activity had decreased to almost zero since March 2020, when PHSMs against COVID-19 began (Figure). Before mandatory on-arrival quarantine of travelers started on September 26, 2022, only sporadic influenza-positive samples were detected by surveillance, all from travelers or children who had recently received live-attenuated influenza vaccine (5). After travel restrictions were removed, sporadic influenza detections increased, but overall activity remained low. After mandatory indoor and outdoor mask wearing restrictions were lifted on March 1, 2023, influenza transmission increased substantially; the first influenza season after COVID-19 in Hong Kong started and peaked on April 9, ended on May 25, and lasted for 12 weeks (6).

Because various other PHSMs were implemented concurrently with the mask mandate, resurgence of influenza activity could not be attributed to relaxation of the mask mandate alone. Therefore, we used a previous approach that estimated the time-varying effective reproductive number (R_t) (7) and a multivariable log-linear regression model on R_t that could allow for adjustment of other factors affecting influenza transmission, including depletion of susceptible persons, seasonal differences, and meteorologic predictors and preventive measures (Appendix). Because the predominating influenza strain in 2023 was influenza A(H1N1)pdm09, we identified previous influenza A(H1N1)pdm09 epidemics that had occurred during 2010–2020. To construct a preventive score, we used data from cross-sectional telephone surveys among the general adult population in Hong Kong from 2020 to 2023 as a proxy for the intensity of preventive measures, other than mask wearing, against COVID-19 (1). The preventive score included the average proportion of persons who avoided visiting crowded places, avoided going to healthcare facilities, avoided touching public objects, or used protective measures when touching public objects, and washed hands immediately after going out. Before 2020, the proportion of those preventive measures was established as baseline. When constructing a preventive score, we compared the Akaike information criterion of 4 combinations of those protective measures. Meteorologic variables provided by the Hong Kong Observatory (http://hko.gov.hk) were temperature, wind speed, and relative and absolute humidity. To quantify the effects of meteorologic variables, we fitted the models to data before the COVID-19 pandemic.

Among the 9 epidemics of 2010–2023, the estimated R_t varied from 0.62 to 1.38 (median 1.02) (Appendix Figure 1). The estimated R_t showed a decreasing pattern in each season, ranging from ≈1.2 at the beginning of an epidemic period to 0.8 at the end of an epidemic period. After model selection (Appendix), we found that a model of absolute humidity, mask wearing, and preventive score 3 (Table) explained 92% of the observed variance in estimated R_t (Appendix Table 1). Changes in absolute humidity (Appendix Figure 2, panel A), the proportion of mask wearing, and preventive score 3 (Appendix Figure 2, panel B) strongly correlated with changes in R_t. After adjusting for other factors, such as depletion of susceptible persons, between-season effects, and absolute humidity, we found that mask wearing was associated with a 25% (range 1%–43%) reduction in R_t and that other preventive measures (combined) were associated with a 77% (range 60%–88%) reduction (Table).

We found that that influenza increased after PHSMs were relaxed and influenza transmission increased shortly after the mask mandate was relaxed. Our results are consistent with those of several studies that found that PHSMs against COVID-19 may reduce influenza transmission (8) and that mask wearing may have a low to moderate protective effect against influenza virus transmission in the community (9,10).

A limitation of our analysis was that we used results of survey reports to generate a proxy of intensity of implemented PHSMs over time, which may not be accurate. Also, we used a proxy measure of influenza activity based on surveillance data, and the reliability of our analysis depended on the accuracy of this proxy. In addition, influenza vaccination coverage (Appendix Figure 5) was not included in the model because our model included the effect of vaccination via season-specific intercept. Nevertheless, our study results suggest that the resurgence of influenza after relaxation of PHSMs was most likely affected by the lifting of mask mandate and other PHSMs.

Top