Transmission potential of mpox in mainland China, June-July 2023: estimating reproduction number during the initial phase of the epidemic

Introduction

On 9 August 2023, the China Centers for Disease Control and Prevention (CDC) announced 491 cases in mainland China in July, representing a marked escalation from 106 cases in June and a five-fold increase (China Centers for Disease Control, 2023b). Although concerning, this surge was largely anticipated by health experts (Murphy, 2023; Yang, 2023). Amid ambiguous prevention strategies and the stigmatization of specific patient groups—especially men who have sex with men (MSM)—the mpox epidemic intensified unnoticed.

Mpox is a viral infection mainly transmitted through direct skin-to-skin contact (The United States Centers for Disease Control and Prevention, 2023). The global mpox outbreak of 2022–2023 was primarily driven by sexual transmission within MSM networks (Thornhill et al., 2022). With limited spillover to other populations, the 2022 outbreaks predominantly affected high-risk MSM communities (Endo et al., 2022; Murayama et al., 2023), making these communities primary targets for subsequent vaccination campaigns (Chitwood et al., 2023; Vairo et al., 2023).

The global transmission of mpox was evident in 2022, but it was not until 2023 that Western Pacific nations, including Japan, South Korea, and Taiwan, saw significant case counts (Chan & Lee, 2023; Endo, Jung & Miura, 2023; Kim et al., 2023; Nishiura et al., 2023). By the end of 2022, many of these countries had minimal or no local mpox transmissions, possibly due to ongoing coronavirus disease 2019 (COVID-19)-related travel and social restrictions (Akhmetzhanov, Ponce & Thompson, 2022). Given the considerable MSM population in mainland China, the uncontrollable spread of mpox presents a significant public health concern. An effective mpox vaccine exists, however, challenges such as limited accessibility and disease-related stigma have hampered its broad dissemination (Murphy, 2023), underscoring the need to understand mpox transmission dynamics in mainland China and predict the evolution of future outbreaks.

Epidemics typically follow a consistent transmission cycle: introduction, escalation, peak, and eventual decline (Linton et al., 2022). The initial stages, often marked by exponential growth, offer important insights into understanding of disease spread dynamics. It is crucial to estimate the effective reproduction number, $R_e$, which indicates the average number of secondary transmissions per primary case, during the period in which interventions, behavioral shifts, or acquired immunity have not yet made an impact.

Prior $R_e$ evaluations for the 2022 national mpox outbreaks have revealed variability (Brand et al., 2023; Du et al., 2022; Liu et al., 2023; Yang et al., 2023). Using a sub-exponential growth model, Du et al. (2022) estimated that the $R_e$ ranged from 1.02 in Portugal to 1.95 in the United States. Conversely, other reports have noted higher $R_e$ values. For example, for the United Kingdom, $R_e$ values of 2.32 and 3.14 were reported (Liu et al., 2023; Yang et al., 2023), compared with a value of 1.18 estimated by Du et al. (2022). Whereas these studies (Liu et al., 2023; Yang et al., 2023) applied variations of a simple susceptible–exposed–infected–recovered model, they also relied on case counts that were classified by the date of reporting. Such an approach can result in an upward bias in estimates during the exponential growth phase because it may reflect not only actual transmission dynamics but also the recognition of an outbreak and the discovery of infection clusters. Administrative batching in case reporting can further bias the estimation of $R_e$. Nevertheless, a recent UK mpox study (Brand et al., 2023) employed a detailed, more complex susceptible–exposed–infected–recovered model that factored in potential biases, generating $R_e$ values that ranged from 3.02 to 5.16 during the initial stage of the outbreak. Given diverse factors, including cultural differences, differences in outbreak response, and challenges with data collection, identifying how these estimates align with $R_e$ in mainland China is crucial.

In this study, we analyzed data from the World Health Organization (WHO) and other governmental sources to enhance our understanding of mpox transmission in mainland China (Taiwan Centers for Disease Control, 2023; The Government of the Hong Kong Special Administrative Region, 2023; World Health Organization, 2023). By focusing on case counts solely attributed to mainland China and estimating the $R_e$ during the initial phase of the outbreak, we aimed to compare counts with estimates from 2022 and project potential case counts for the near future. Our ultimate goal was to inform ongoing public health interventions, and potentially optimize efficiency of these interventions, while also contributing to global epidemic preparedness. A preliminary version of this text was previously published as a preprint (Akhmetzhanov & Wu, 2023).

Methods

Results

Using the Hong Kong data, we estimated a mean delay of 3.4 days (95% credible interval (CrI) [2.7–4.4]) from symptom onset to case diagnosis and a mean delay of 6.0 days (95% CrI [5.1–7.0]) from symptom onset to case reporting. Upon integrating these delays into the case back-projection (Fig. 1), the $R_e$ was estimated at 1.57 (95% CrI [1.38–1.78]) based on estimated serial interval (SI) distribution in Miura et al. (2023), while the posterior mean ranged 1.31–1.76 using alternative SI estimates (Guo et al., 2023; Guzzetta et al., 2022; Madewell et al., 2023) (Table 1). Correspondingly, the doubling time was 15 days (95% CrI [11–21]). Projecting the mpox case counts past 5 July 2023 to cover the months of July and August, the mean count of cases that were categorized by the symptom onset date was 1,080 (95% CrI [650–1,720]) for July and 5,200 (95% CrI [1,820–12,060]) for August. Adjusting these numbers for the reporting delay, the mean numbers of reported cases were 800 (95% CrI [540–1,180]) and 3,850 (95% CrI [1,490–8,290]) for July and August, respectively.

The analysis of national outbreaks from selected countries in 2022 revealed a range in $R_e$ values. Portugal had the lowest mean $R_e$ value at 1.13 (95% CrI [1.06–1.23]), whereas Colombia had the highest value at 2.31 (95% CrI [2.04–2.60]), as illustrated in Fig. 2. The pooled mean $R_e$ was 1.53 (95% CrI [1.38–1.69]), with a between-country variance (heterogeneity) of ${\tau }^2=0.10$. However, upon evaluating the data quality provided to the WHO by individual countries, it became evident that the nations within the WHO AMRO—with the exceptions of the United States and Canada—provided only aggregated case counts without subtyping them by date of symptom onset, date of diagnosis, or date of reporting. Such reporting likely introduced higher variability and potential bias in the $R_e$ values. By excluding these countries from the analysis, the pooled mean $R_e$ dropped slightly to 1.49 (95% CrI [1.32–1.67]), and the between-country variance declined to ${\tau }^2=\mathrm{0.08.}$ Notably, the $R_e,$ value for mainland China remained similar to the pooled average regardless of whether AMRO countries were included in the analysis.

The sensitivity analysis confirmed the robustness of our findings but highlighted some variability. When considering only cases with a known symptom onset date, the $R_e$ for mainland China remained consistent at 1.56 (95% CrI [1.37–1.79]), suggesting that most cases classified by date of reporting were predominantly linked to Taiwan. Varying the time window for the identified exponential growth phase—in which the lower bound $\left(t_{lo}\right)$ varied between 16 and 26 May 2023, and the upper bound $\left(t_{up}\right)$ varied between 30 June and 10 July 2023—the median posterior $R_e$ ranged from 1.51 to 1.57 (95% CrI [1.31–1.82]; Fig. S3). The posterior predictions for reported cases ranged from 650 to 840 (95% CrI [480–1,400]) for July and 2,150 to 3,540 (95% CrI [1,110–9,690]) for August (Figs. S4, S5). For the selected 2022 outbreaks, adjusting the length of the exponential growth phase to either 30 or 60 days did not critically alter the prior estimates (Fig. S6). Importantly, when we focused solely on cases with a known date of symptom onset, the estimated $R_e$ remained unchanged, indicating the minimal impact of case records with unknown symptom onset on the $R_e$ for the selected countries.

Discussion

In our analysis, we estimated that the $R_e$ of the mpox epidemic in mainland China was within the range of 1.3–1.8, depending on the chosen serial interval distribution (Guo et al., 2023; Guzzetta et al., 2022; Madewell et al., 2023; Miura et al., 2023). This range suggests a doubling time of approximately 2 weeks. Importantly, our estimate was based on the early phase of the epidemic, in which an evident exponential growth of cases was observed and control measures had likely not yet been implemented. Given the relatively low case counts of mpox and no deaths reported during the epidemic’s early phase, significant behavioral changes in at-risk populations were likely not yet in effect, implying that mpox transmission occurred largely unhindered within affected communities during this early phase. Moreover, our projections for July-August, extrapolated into later months, might overestimate actual cases, as they do not account for the anticipated behavioral changes expected from rising public awareness and new information about the outbreak (China Centers for Disease Control, 2023a).

The $R_e$ value estimated in our study closely aligns with values derived from selected national outbreaks in 2022, suggesting the potentially universal character of the outbreaks. However, an accurate estimation of infections remains challenging given the likely underreporting of mpox cases. Several factors might contribute to this under-ascertainment. One reason may be the societal stigmatization associated with the disease that leads to individuals’ reluctance to disclose their infection for fear of facing societal prejudice or discrimination. Another possibility is that infections can result from intimate encounters outside of a committed relationship, leading individuals to withhold their health status because of the fear of repercussions in their personal lives.

Our study has several limitations. First, we did not differentiate between indigenous and imported infections in our dataset, a factor that may have influenced the estimation of $R_e$. However, by focusing exclusively on the exponential phase in which case growth is most evident, the influence of imported cases on our estimates was likely minimal. Second, the actual method of imputing cases from Taiwan in the WHO dataset may have differed from ours or could have been modified by the WHO over time. For example, the observed pattern of data imputations by date of reporting that were aggregated every 7 days was no longer feasible at the end of July and August. Third, the reporting delay distribution was estimated using Hong-Kong line list data, which may differ from the data from mainland China given the differences in healthcare systems, case ascertainment, and case investigation protocols. Fourth, the results of comparisons of $R_e$ values across different countries must be interpreted with caution. The proactive control measures adopted by some nations versus the more passive strategies employed by others can introduce variances. For instance, the outbreak in Portugal evolved at a slower pace given the government’s proactive position. The case counts were almost steady over 2–3 months before a decline was observed. Although the $R_e$ values in our study and that of Du et al. (2022) were close to 1, a granular look at the case counts combined with a time-varying $R_e$ revealed a brief $R_e$ peak of 2.7 on 10 May 2022 that then gradually declined to 1 and dropped below this threshold by the end of June 2022 (Borges et al., 2023). Our mathematical framework successfully captured this average trend across Portugal and other countries, and a pooled mean $R_e$ of 1.66 summarized our overall findings. Finally, the WHO dataset retrieved on 15 August 2023 contained incomplete case counts given that some cases were yet to be reported. At the time of writing the results of this study, a snapshot for 12 September 2023 became available. The $R_e$ estimate only slightly changed to 1.63 (95% CrI [1.43–1.86]); however, the case counts for May and June were updated and revealed possible long reporting delays (Figs. S1A and S7).

Although mpox may be associated with a lower mortality rate compared with diseases such as smallpox, its morbidity is notably high, and mpox is associated with health complications such as pneumonia, confusion, and eye infections (Patel & Patel, 2023). Mpox disproportionally affects people living with human immunodeficiency virus (HIV) (Mitjà et al., 2023). The transmission of mpox throughout mainland China in the summer of 2023 was largely unconstrained given that regulatory guidelines were first published only at the end of July. This timeline put vulnerable populations at heightened risk of infections and subsequent complications. Nevertheless, the potential impact of efficient vaccination campaigns and information programs aimed at raising awareness of the disease cannot be overlooked; such interventions could significantly reduce case numbers and prevent a larger future outbreak. In contrast to mainland China, Taiwan managed to stabilize its mpox situation in 2023, maintaining an $R_e$ value of approximately one. Drawing from its previous experience—specifically, its management of the hepatitis A virus (HAV) outbreak among MSM in 2015–2017 (Lin et al., 2019)—the Taiwan government promptly initiated a vaccination campaign, targeted at high-risk groups that was bolstered by awareness campaign. The strategy not only ensured low vaccine hesitancy but also prevented an escalating outbreak. Mainland China could adopt a similar approach with vaccination campaigns focused on high-risk social groups coupled with the dissemination of accurate health messages. Avoiding stigmatization of the disease, an oversight during the early stages of the outbreak (Murphy, 2023; Yang, 2023), could further contribute to higher vaccination rates.

Conclusions

In this study, the mean $R_e$ value estimated for the early phase of the mpox outbreak in mainland China was between 1.3 and 1.8, closely aligning with the pooled mean $R_e$ from selected outbreaks in 2022. This alignment suggests that mainland China could experience a similar trajectory of case growth as observed in other countries previously affected by mpox. These findings underscore the urgency for implementing immediate and effective control measures, including targeted vaccination coupled with information campaigns to reduce stigmatization of vulnerable groups associated with the disease.

Supplemental Information