Swift and extensive Omicron outbreak in China after sudden exit from 'zero-COVID' policy

Abstract

In late 2022, China transitioned from a strict 'zero-COVID' policy to rapidly abandoning nearly all interventions and data reporting. This raised great concern about the presumably-rapid but unreported spread of the SARS-CoV-2 Omicron variant in a very large population of very low pre-existing immunity. By modeling a combination of case count and survey data, we show that Omicron spread extremely rapidly, at a rate of 0.42/day (95% credibility interval: [0.35, 0.51]/day), translating to an epidemic doubling time of 1.6 days ([1.6, 2.0] days) after the full exit from zero-COVID on Dec. 7, 2022. Consequently, we estimate that the vast majority of the population (97% [95%, 99%], sensitivity analysis lower limit of 90%) was infected during December, with the nation-wide epidemic peaking on Dec. 23. Overall, our results highlight the extremely high transmissibility of the variant and the importance of proper design of intervention exit strategies to avoid large infection waves.

Fig. 1. Model-predicted epidemic dynamics of the Omicron variant in China.

a The model-inferred daily cases (orange bands) during the three policy periods (zero-COVID, 20 Measures, and 10 Measures) near the end of 2022, and official case counts (points). Shades of orange show the median, 50% credibility interval (CrI), and 95% CrI. b Survey results, interpreted as population size, compared against model-predicted values for those survey categories on Dec. 26. Dashed lines are the data, and shaded curves are the posterior distribution of expected number of individuals. c Estimates of the intrinsic rate of increase, r, and reproductive number, R_exp during the exponential growth of each time period. From the model fit, points are the median, thick lines are the 50% CrI, and thin lines are the 95% CrI. d Model-inferred epidemic dynamics during 10 Measures. The median, 50% CrI, and 95% CrI are shown for people in the model’s susceptible (S), exposed (E), infected (I), and recovered (R) states (cf. Fig. 2).

Fig. 2. Schematic of the SEIR-type model.

Our model maps the four epidemiological states-susceptible (S), exposed (E), infectious (I), recovered (R)-to the three states of the online survey. Individuals who report themselves as infected and recovered (green state) are only those who previously experienced symptoms (top track of states). Those reporting as infected with symptoms (red states) could be currently infectious or recovered but still symptomatic. All others report as not yet infected (purple states), whether they are susceptible, exposed, pre-symptomatic, or asymptomatic. People in all four infectious states can infect susceptibles, with transmission rate β. A proportion f of infected people eventually develop symptoms. Other parameters are rates of becoming infectious (k_E), becoming symptomatic (k_IP), resolving symptoms (k_IS, k_S), and recovering without symptoms (k_A). Some epidemiological states are divided into substates (e.g., E₁ and E₂) to allow more realistic distributions of waiting times. The bottom track of states counts the fraction w of people who test positive with a delay governed by rate k_T; note that this does not remove them from the epidemiological or survey states (thus with a dashed arrow). The corresponding system of ordinary differential equations, along with the rate parameter values, is provided in “Methods”.

Fig. 3. Model comparison with voluntary testing survey from China CDC.

The gray band shows the incidence inferred by our main model fit, which did not include the data points plotted here. Those model predictions adjusted for testing effort and willingness to test (see “Methods”) are shown by blue and red bands for the PCR and antigen tests, respectively. Dark and light bands show 50% CrI and 95% CrI derived from the model fit.