Figure 1

Figure 1. Epidemiology of 5th and 6th COVID-19 outbreak waves, Hong Kong, 2022, evaluated in study of effects of school-based preventive measures on COVID-19 incidence. A) Epidemic curves of 5th (February–April 2022,...

The study was approved by the Institutional Review Board of the University of Hong Kong. We analyzed COVID-19 data reported to the Hong Kong Centre for Health Protection that included PCR-confirmed cases during January 1–November 22, 2022, and RAT-confirmed cases during February 26–November 22, 2022. Confirmative PCR was administered for RAT-confirmed cases reported during June 7–August 28, 2022. We found that age-specific incidence rate ratios for infections in children compared with adults (>18 years of age) in the 6th wave were slightly higher than in the 5th wave (Figure 1).

We divided the study period into 3 segments: school closure, summer holiday, and normal school days (days other than closures and holidays). We stratified cases into 4 age groups: 2–5 years (preschool/kindergarten students), 6–11 years (primary school students), 12–17 years (secondary school students), and >18 years (adults). We used a Poisson generalized additive regression model, adjusting for time trend of COVID-19 cases and including the age groups and study periods (Appendix), to determine the effects of school closure and summer holiday on COVID-19 transmission in school-age children.

During normal school days (Table 2), the COVID-19 incidence rate for kindergarten students was 24% (95% CI 22%–25%), for primary school students was 34% (95% CI 32%–35%), and for secondary school students was 19% (95% CI 18%–20%) higher than for adults, suggesting that school-age children had a higher infection risk than adults during normal school days. During the 5th-wave school closure, the incidence rate for kindergarten students was 31% (95% CI 29%–32%), for primary students was 42% (95% CI 41%–43%), and for secondary students was 46% (95% CI 46%–47%) lower than for adults. During the summer holiday when most schools were closed during the 6th wave, the COVID-19 incidence rate for kindergarten students was 12% (95% CI 9%–15%), for primary students was 28% (95% CI 26%–30%), and for secondary students was 32% (95% CI 30%–34%) lower than for adults. Assuming that school-based interventions had no effect on adults, effectiveness of school closure on reducing COVID-19 transmission was 31%–46% during the 5th wave and 12%–32% during the 6th wave.

Figure 2

Figure 2. School-reported data during the 6th COVID-19 wave, Hong Kong, 2022, evaluated in study of effects of school-based preventive measures on COVID-19 incidence. A) Weekly numbers of school-reported cases during the...

We collected school-related data from daily press conferences and press releases, including numbers of school-reported cases (students and staff), class suspensions, and schools reporting >1 case during periods of in-person learning (Appendix). Excluding summer holidays, weekly case numbers in the community were highly correlated with weekly numbers of school-reported student and staff cases (Pearson correlation coefficient r = 0.77, 95% CI 0.54–0.89), schools reporting >1 case (r = 0.75; 95% CI 0.51–0.88), and >2 cases of class suspension (r = 0.81, 95% CI 0.61–0.91) (Figure 2). Among 299 suspected school clusters, defined as schools that reported>2 COVID-19 cases within 7 days, a total of 66 (22%) had >5 cases and 22 (7%) had >10 cases. The largest suspected cluster recorded 53 cases in a secondary school that had ≈750 students and ≈75 staff. The second-largest suspected cluster had 35 cases in an international school that had ≈960 primary and secondary students.

We found that school-age children had a higher SARS-CoV-2 infection risk than adults in Hong Kong, consistent with another study suggesting that children were more susceptible to Omicron variants compared with adults (4). School closure and summer holiday effectively reduced incidence rates in school-age children during the 5th and 6th COVID-19 waves, aligning with modeling and simulation studies demonstrating the effectiveness of school closure in reducing COVID-19 transmission (5–7). We noted that the reduction in incidence rates for school-age children during school closure in the 5th wave was greater than that during the summer holiday in the 6th wave. Potential explanations for those results are that schools might not have been completely closed during summer holiday, possibly increasing the number of contacts between children; that Omicron BA.4/BA.5 variants were more prevalent during the 6th wave (Appendix Figure 1); or that higher ascertainment rates existed among students who had RAT used to detect COVID-19.

The strong positive correlation between school-reported data and community case numbers after school reopening indicated school reopening did not cause abnormal increases in community COVID-19 incidence. The largest suspected school cluster had 53 COVID-19 cases, comparable to other superspreading events, such as the 67-case cluster caused by Omicron BA.1 and 167-case cluster caused by Omicron BA.2 in January 2022 (8). Those results suggest that school reopening did not pose additional superspreading risks in school settings.

The first limitation of our study is that some school-reported COVID-19 infections could have originated elsewhere in the community, such as at home, instead of in schools. Although students and staff were required to conduct daily rapid tests and report positive results to schools and the government, underreporting cannot be ruled out. Second, we extracted school outbreak data from press conferences; thus, some details could have been missed. Third, we cannot exclude the possibility that some schools did not fully adhere to guidelines, particularly regarding class size and lesson duration; however, we lacked school-level data to account for that possibility. Fourth, our analysis did not account for changes in dominant virus strains (Omicron BA.2 in the 5th wave, Omicron BA.5 in the 6th wave). Finally, we considered school-based measures as a collective package and were unable to determine individual effects of specific measures on COVID-19 transmission.

In summary, we evaluated school closure and school reopening accompanied by multilayer school-based preventive measures for COVID-19 in Hong Kong, which was informative as a guide for implementing and relaxing those measures. Our results might not be directly generalizable for other respiratory pathogens because of differences in transmission and intervention effectiveness. However, our results are consistent with modeling studies suggesting that safe school reopening is possible when appropriate alternative school-based preventive measures are used (9–12). If resurgence in case numbers or emergence of variants with higher transmissibility in children occurs, school closure remains an option to reduce transmission among children.