METHODS: We developed and validated a robust algorithm to classify progress towards rabies elimination within sub-national administrative units, which we applied to surveillance data from Brazil and Mexico. The method combines criteria that are easy to understand, including logistic regression analysis of case detection time series, assessment of rabies virus variants, and of incursion risk. Subjecting the algorithm to robustness testing, we further employed simulated data sub-sampled at differing levels of case detection to assess the algorithm\'s performance and sensitivity to surveillance quality.
RESULTS: Our tool demonstrated clear epidemiological transitions in Mexico and Brazil: most states progressed rapidly towards elimination, but a few regressed due to incursions and control lapses. In 2015, dog-mediated rabies continued to circulate in the poorest states, with foci remaining in only 1 of 32 states in Mexico, and 2 of 27 in Brazil, posing incursion risks to the wider region. The classification tool was robust in determining epidemiological status irrespective of most levels of surveillance quality. In endemic settings, surveillance would need to detect less than 2.5% of all circulating cases to result in misclassification, whereas in settings where incursions become the main source of cases the threshold detection level for correct classification should not be less than 5%.
CONCLUSIONS: Our tool provides guidance on how to progress effectively towards elimination targets and tailor strategies to local epidemiological situations, while revealing insights into rabies dynamics. Post-campaign assessments of dog vaccination coverage in endemic states, and enhanced surveillance to verify and maintain freedom in states threatened by incursions were identified as priorities to catalyze progress towards elimination. Our finding suggests genomic surveillance should become increasingly valuable during the endgame for discriminating circulating variants and pinpointing sources of incursions.
方法:我们开发并验证了一种强大的算法,以对地方各级行政单位内的狂犬病消除进展进行分类,我们将其应用于巴西和墨西哥的监测数据。该方法结合了易于理解的标准,包括病例检测时间序列的Logistic回归分析,评估狂犬病病毒变种,以及入侵风险。对算法进行鲁棒性测试,我们进一步利用在不同病例检测水平下二次采样的模拟数据来评估算法的性能和对监测质量的敏感性.
结果:我们的工具显示了墨西哥和巴西明显的流行病学转变:大多数州在消除疾病方面进展迅速,但是由于入侵和控制失误,一些人倒退了。2015年,以狗为媒介的狂犬病继续在最贫穷的州传播,焦点仅保留在墨西哥32个州中的1个州,巴西27人中的2人,对更广泛的地区构成入侵风险。无论大多数监测质量水平如何,分类工具在确定流行病学状况方面都是强大的。在地方性环境中,监测需要检测不到所有循环病例的2.5%,以导致错误分类,而在入侵成为主要病例来源的情况下,正确分类的阈值检测水平不应低于5%。
结论:我们的工具为如何有效地实现消除目标提供指导,并根据当地流行病学情况制定策略。同时揭示狂犬病动态的见解。运动后评估流行州的狗疫苗接种覆盖率,和加强监督,以核实和维持受入侵威胁的国家的自由,被确定为优先事项,以促进消除冲突的进展。我们的发现表明,基因组监测在区分循环变异和查明侵入源的残局期间应该变得越来越有价值。