Tracking progress towards malaria elimination in China: estimates of reproduction numbers and their spatiotemporal variation

Isobel Routledge; Shengjie Lai; Katherine E Battle; Azra C Ghani; Manuel Gomez-Rodriguez; Kyle B Gustafson; Swapnil Mishra; Joshua L Proctor; Andrew J Tatem; Zhongjie Li; Samir Bhatt

Although China is making rapid progress towards this goal, 2,675 imported cases were reported in 35 2017, highlighting the risk of re-introduction 3 . Large numbers of people move between China and 36 malaria endemic countries, both from sub-Saharan Africa and from South East Asia 8,9 , driven by 37 tourism and Chinese oversea investment 10 . Concerns remain about re-emergence of malaria, which 38 has occurred several times in the early 2000s as a result of importation and favourable climatic 39 conditions for competent vectors 11 . Therefore, in order to achieve three consecutive years of zero 40 indigenous cases (the requirement for WHO certification of elimination), a sustained and targeted 41 investment in surveillance together with efficient treatment is necessary. Figure 5) . 91 Spatial patterns of 92 As transmission declined between 2011 and 2016, we observed a reduction in the incidence of 93 locally-acquired cases which is reflected in a reduction in our estimates of the reproduction number 94 of each locally-acquired case for both species and with a more focal spatial distribution of cases 95 ( Figure 2A and 2B). We estimate a decline in the probability of a reproduction number for a P. vivax 96 case being above zero over this period ( Figure 3A and 3B), with the central parts of the province 97 being the first to reach lower risks of non-zero . The border area neighbouring Myanmar, where 98 most cases were observed, had the lowest amount of uncertainty in the estimates. P. falciparum 99 shows a decline in risk of > 0 across the province, with the more isolated areas in the north of the 100 province showing both the highest predicted risk but also the most uncertainty, due to a lack of 101 cases observed there (Supplementary Figure 6) . By 2016 all areas have reached a low risk, although 102 there is more uncertainty in these estimates compared to P. vivax, almost certainly due to the 103 smaller sample size. 104 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/628842 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/628842 doi: bioRxiv preprint Short -term predictions and temporal patterns in timeseries of Plasmodium vivax cases 114 Using a time series method to make short-term predictions, we estimate a posterior mean of 115 0.005 (95% CI = 0 -0.34) for Plasmodium vivax cases in 2019 ( Figure 4A ). We observe and overall 116 declining trend, with the fitted trend for (which estimates the general trend, separate to the 117 influence of seasonal and holiday effects) declining from 0.31 (95% CI = 0.31, 0.34) at the start of 118 2011 to 0.004 (95% CI =0.002-0.006 ) by the end of 2019 ( Figure 4B ). We estimate a small effect of 119 holiday periods to differences in observed, with Chinese New Year and National Day associated 120 with small increase risk in of 16% ( 95% CI = -112%, 152%) and 39% (95% CI = -43%, 118%) 121 ( Figure 4B ) which in this very low transmission context could increase the probability of small 122 outbreaks of local transmission in areas in which high rates of importation occur, although very wide 123 credible intervals were associated with these estimates. We did not identify a clear seaso