Impact of Vaccination and Public Health Measures on the Severity of SARS-CoV-2 Omicron Infections in China: A Systematic Review and Meta-Regression Analysis

Abstract

Background: Starting in early 2022, SARS-CoV-2 Omicron has driven large outbreaks in China, a predominantly infection-naive population with high inactivated vaccine coverage. This unique context provided a substantially less-confounded opportunity to evaluate how vaccination, public health, and social measures influenced severity. Methods: We systematically reviewed 86 studies (224 severity estimates) published from 2022 to 2024, reporting symptom and clinical severity outcomes (fever, cough, and sore throat; symptomatic, severe/critical, and fatal illness) of Omicron infections in China. Using meta-regression, we evaluated the associations of study setting, age group, vaccination status, predominant subvariants, and Oxford COVID-19 Government Response Tracker (OxCGRT) indices, including the Government Response Index (GRI), Containment and Health Index (CHI), and the Stringency Index (SI), with infection outcomes, adjusting for key confounders. Results: We found the primary or booster series of inactivated vaccines conferred strong protection against severe/critical illness (pooled relative risk (RR) 0.17 [95% CI: 0.09-0.33]) but did not reduce symptom frequency (RR 0.99 [95% CI: 0.95-1.02]). Each 10-unit increase in GRI or CHI was associated with 7% (95% CI: 1-12%) and 6% (95% CI: 1-10%) lower odds of symptomatic infection and 3% (95% CI: 1-4%) lower odds of severe/critical illness. Later subvariants (BA.5, BF.7, and XBB) showed 24-38% higher odds of upper respiratory symptoms versus BA.1. Conclusions: The data collection context significantly impacted severity estimates, with higher estimates from emergency hospitals. Overall, inactivated vaccines provided strong protection against severe/critical outcomes while stringent public health measures were associated with lower severity. Our findings underscore the importance of consistent and standardized protocols to produce reliable estimates of SARS-CoV-2 severity in evolving epidemiological contexts.

Keywords: COVID-19; Omicron; SARS-CoV-2; inactivated vaccines; public health measures; severity; vaccine effectiveness.

Figure 1

PRISMA flow diagram of searching and screening approach.

Figure 2

Distributions of the computed percentage of cases with fever, cough, sore throat, symptomatic, severe/critical, and deceased cases, disaggregated by (A,F) study design, (B,G) age group, (C,H) primary and booster vaccination coverage, (D,I) predominant circulating Omicron subvariants, and (E,J) whether the study was conducted during the pre- or post-relation period. Boxes indicate interquartile ranges while bold horizontal lines in the boxes represent the median values of computed estimates. Vertical lines give the range (minimums and maximums) of computed estimates by each factor.

Figure 3

Relative decreases in the heterogeneity in symptom and clinical severity estimates when incorporating epidemiological and demographic variables including age group, antiviral treatment coverage, study period, study setting, and the predominant circulating subvariant into the meta-analysis. The asterisk symbols (*) indicate statistical significance for moderator test (Wald Z-test) with p-value < 0.05.

Figure 4

Percentage of cases with symptomatic infection, fever, cough, or sore throat, and the percentage of severe or critical cases for Omicron infections in China for vaccinated cases (at least primary series) compared with nonvaccinated cases. Squares and lines indicate the relative risk (RR) percentage of events of vaccinated cases compared with nonvaccinated cases and the corresponding 95% confidence intervals calculated based on normal approximation. The size of square is proportional to the weight of the study in relation to the pooled estimate. Pooled estimates (blue and orange colored diamonds) were estimated using random-effects meta-analysis. Arrows indicate that 95% confidence intervals are higher than the upper bound of x axis (logarithmic RR).