In December 2019, the first cases of coronavirus disease (COVID-19) were reported in Wuhan, China, a megacity with a population of approximately 11 million people. To prevent the spread of this highly infectious disease, the government initiated a city-wide lockdown on January 23, 2020. However, despite these efforts, COVID-19 spread to many countries across the world, reaching pandemic levels, and continues to be a serious public health concern due to its high mortality rate. According to the large-sample analysis by Wu and McGoogan [1], China’s case fatality rate (CFR) was 2.3%—that is, 1023 deaths from 44,672 confirmed cases as of February 11, 2020, with a significant proportion of cases originating from Wuhan. The large number of infected people in Wuhan put a tight strain on essential medical resources. The city had a much higher mortality rate (according to Feb 10th statistics: CFR=4.05% [748 deaths/18,454 diagnoses]; Apr 24th statistics: CFR=7.69% [3869 deaths/50,333 diagnoses]) than the rest of China. The overall CFR of 2.3% for China was likely overestimated, due to strained medical resources and a large number of undiagnosed patients. According to a recent study, 78% of those who had been infected were asymptomatic [2]. Therefore, a large number of asymptomatic infections in Wuhan might have never been diagnosed, which contributed to the overestimated CFR. An accurate estimation of the infection rate is therefore important to assess Wuhan’s CFR precisely.

Using Markov Chain Monte Carlo methods, Wu et al [3] estimated that 75,815 individuals (95% CI 37,304-130,330) had been infected in Wuhan as of January 25, 2020. Following this, a number of foreign governments evacuated their citizens and performed thorough etiological tests on them. This group of evacuees can serve as a “random” sample to estimate the infection rate in Wuhan. With internet search as an important source of epidemiologic information on COVID-19 [4], we performed a combined analysis of the infection rates of these population samples using publicly available data (Table 1), instead of a simple pooled calculation, considering potential differences in lifestyles and pathogen exposure across different populations. The combined analysis was done using the Comprehensive Meta-Analysis Software (Biostat, Inc).

Our analysis demonstrates that there is no significant heterogeneity across different population samples (heterogeneity test P=.491). The combined infection rate is 0.013 (95% CI 0.008-0.022) (Figure 1). Based on our results, we estimate the number of infected people in Wuhan, China, to be 143,000 (range 88,000-242,000), which is significantly higher than the estimate proposed by Wu et al [3].

Our estimate indicates that a large number of infections in Wuhan were not diagnosed. The number of undiagnosed cases in late January and early February is larger than the final diagnosed count reported to date (n=50,333), which has resulted in an overestimated CFR. In addition, our study suggests that the lower CFR (0.51%) estimated by the Centre for Evidence-Based Medicine [15] does not indicate viral variants and loss of virulence. Taken together, increased awareness of the original infection rates in Wuhan, China, is critically important for appropriate public health measures at all levels, as well as to eliminate panic caused by overestimated mortality rates that may bias health policy actions by the authorities.