2023年,霍乱影响了大约100万人,并在全球造成5000多人死亡。主要在低收入和冲突环境中。近年来,新爆发的霍乱数量迅速增加。Further,冲突加剧了持续的霍乱爆发,气候变化,基础设施差,导致长期的危机。因此,对治疗和干预的需求正在迅速超过现有资源的可用性.在改善水和卫生系统之前,霍乱,一种主要通过受污染的水源传播的疾病,也经常蹂躏高收入国家。摇摇欲坠的基础设施和气候变化正在使新地点面临风险-即使在高收入国家也是如此。因此,了解霍乱的传播和预防至关重要。抗击霍乱需要多种干预措施,最常见的两种是行为教育和水处理。两剂量口服霍乱疫苗(OCV)通常被用作这些干预措施的补充。由于供应有限,各国最近已改用单剂量疫苗(OCV1)。一个挑战在于了解及时分配OCV1的位置,特别是在缺乏资源充足的公共卫生监测系统的环境中。当霍乱在这样的地方发生和传播时,及时,准确,和公开可访问的爆发数据通常无法用于疾病建模和后续决策。在这项研究中,我们证明了开放获取的数据在快速评估霍乱传播和疫苗有效性方面的价值.具体来说,我们获得了两个国家最近爆发的霍乱的非机器可读(NMR)流行曲线,海地和喀麦隆,来自发布在形势和疾病爆发新闻报道中的数字。我们使用计算数字化技术来得出霍乱病例的每周计数,当与报告的累积病例数进行比较时,导致名义上的差异(即,海地和喀麦隆的相对错误率分别为5.67%和0.54%)。鉴于这些数字化的时间序列,我们利用EpiEstim-一个开源建模平台-通过有效繁殖数(Rt)快速估计随时间变化的疾病传播。为了比较两个考虑国家的OCV1有效性,我们还使用了VaxEstim,EpiEstim的最新扩展,通过三个输入之间的关系来促进疫苗有效性的估计:基本复制数(R0),Rt,和疫苗覆盖率。这里,以海地和喀麦隆为例,我们演示了VaxEstim在低资源环境中的首次实现.重要的是,我们是第一个使用VaxEstim的数字化数据,而不是传统的流行病监测数据。在疫情初期,这两个国家的每周滚动平均Rt估计值均升高:海地为2.60[95%可信区间:2.42-2.79],喀麦隆为1.90[1.14-2.95]。这些值与海地以前对R0的估计基本一致,其中平均值从1.06到3.72,在喀麦隆,其中平均值范围为1.10至3.50。在海地和喀麦隆,这个高传输的初始周期先于一个较长的周期,在此期间,Rt在临界阈值1附近振荡。我们从VaxEstim得出的结果表明,海地的OCV1有效性高于喀麦隆(75.32%有效[54.00-86.39%]与54.88%[18.94-84.90%])。这些对OCV1有效性的估计通常与其他国家进行的实地研究得出的估计一致。因此,我们的案例研究加强了VaxEstim作为昂贵的替代品的有效性,OCV1有效性的耗时现场研究。的确,之前在南苏丹工作,孟加拉国,刚果民主共和国报告OCV1有效率约为40%至80%.这项工作强调了将爆发病例数据的NMR来源与计算技术相结合的价值,以及VaxEstim的实用性,在数据匮乏的疫情环境中,对疫苗有效性的廉价估计。
In 2023, cholera affected approximately 1 million people and caused more than 5000 deaths globally, predominantly in low-income and conflict settings. In recent years, the number of new cholera outbreaks has grown rapidly. Further, ongoing cholera outbreaks have been exacerbated by conflict, climate change, and poor infrastructure, resulting in prolonged crises. As a result, the demand for treatment and intervention is quickly outpacing existing resource availability. Prior to improved water and sanitation systems, cholera, a disease primarily transmitted via contaminated water sources, also routinely ravaged high-income countries. Crumbling infrastructure and climate change are now putting new locations at risk - even in high-income countries. Thus, understanding the transmission and prevention of cholera is critical. Combating cholera requires multiple interventions, the two most common being behavioral education and water treatment. Two-dose oral cholera vaccination (OCV) is often used as a complement to these interventions. Due to limited supply, countries have recently switched to single-dose vaccines (OCV1). One challenge lies in understanding where to allocate OCV1 in a timely manner, especially in settings lacking well-resourced public health surveillance systems. As cholera occurs and propagates in such locations, timely, accurate, and openly accessible outbreak data are typically inaccessible for disease modeling and subsequent decision-making. In this study, we demonstrated the value of open-access data to rapidly estimate cholera transmission and vaccine effectiveness. Specifically, we obtained non-machine readable (NMR) epidemic curves for recent cholera outbreaks in two countries, Haiti and Cameroon, from figures published in situation and disease outbreak news reports. We used computational digitization techniques to derive weekly counts of cholera cases, resulting in nominal differences when compared against the reported cumulative case counts (i.e., a relative error rate of 5.67% in Haiti and 0.54% in Cameroon). Given these digitized time series, we leveraged EpiEstim-an open-source modeling platform-to derive rapid estimates of time-varying disease transmission via the effective reproduction number ( R t ). To compare OCV1 effectiveness in the two considered countries, we additionally used VaxEstim, a recent extension of EpiEstim that facilitates the estimation of vaccine effectiveness via the relation among three inputs: the basic reproduction number ( R 0 ), R t , and vaccine coverage. Here, with Haiti and Cameroon as case studies, we demonstrated the first implementation of VaxEstim in low-resource settings. Importantly, we are the first to use VaxEstim with digitized data rather than traditional epidemic surveillance data. In the initial phase of the outbreak, weekly rolling average estimates of R t were elevated in both countries: 2.60 in Haiti [95% credible interval: 2.42-2.79] and 1.90 in Cameroon [1.14-2.95]. These values are largely consistent with previous estimates of R 0 in Haiti, where average values have ranged from 1.06 to 3.72, and in Cameroon, where average values have ranged from 1.10 to 3.50. In both Haiti and Cameroon, this initial period of high transmission preceded a longer period during which R t oscillated around the critical threshold of 1. Our results derived from VaxEstim suggest that Haiti had higher OCV1 effectiveness than Cameroon (75.32% effective [54.00-86.39%] vs. 54.88% [18.94-84.90%]). These estimates of OCV1 effectiveness are generally aligned with those derived from field studies conducted in other countries. Thus, our case study reinforces the validity of VaxEstim as an alternative to costly, time-consuming field studies of OCV1 effectiveness. Indeed, prior work in South Sudan, Bangladesh, and the Democratic Republic of the Congo reported OCV1 effectiveness ranging from approximately 40% to 80%. This work underscores the value of combining NMR sources of outbreak case data with computational techniques and the utility of VaxEstim for rapid, inexpensive estimation of vaccine effectiveness in data-poor outbreak settings.