Estimating Malaria Vaccine Efficacy in the Absence of a Gold Standard Case Definition: Mendelian Factorial Design

Accurate estimates of malaria vaccine efficacy require a reliable definition of a malaria case. However, the symptoms of clinical malaria are unspecific, overlapping with other childhood illnesses. Additionally, children in endemic areas tolerate varying levels of parasitemia without symptoms. Together, this makes finding a gold-standard case definition challenging. We present a method to identify and estimate malaria vaccine efficacy that does not require an observable gold-standard case definition. Instead, we leverage genetic traits that are protective against malaria but not against other illnesses, for example, the sickle cell trait, to identify vaccine efficacy in a randomized trial. Inspired by Mendelian randomization, we introduce Mendelian factorial design, a method that augments a randomized trial with genetic variation to produce a natural factorial experiment, which identifies vaccine efficacy under realistic assumptions. A robust, covariance adjusted estimation procedure is developed for estimating vaccine efficacy on the risk ratio and incidence rate ratio scales. Simulations suggest that our estimator has good performance whereas standard methods are systematically biased. We demonstrate that a combined estimator using both our proposed estimator and the standard approach yields significant improvements when the Mendelian factor is only weakly protective. Our method can be applied in vaccine and prevention trials of other childhood diseases that have no gold-standard case definition and known genetic risk factors. Supplementary materials for this article are available online.

准确估计疟疾疫苗效力（malaria vaccine efficacy）需要对疟疾病例给出可靠的定义。然而，临床疟疾的症状缺乏特异性，可与其他儿童疾病的症状相互重叠。此外，疟疾流行区的儿童可耐受不同程度的无症状寄生虫血症（parasitemia）。综上，确立金标准病例定义（gold-standard case definition）颇具挑战。本文提出一种无需可观测的金标准病例定义即可识别并估计疟疾疫苗效力的方法。我们利用可抵御疟疾却无法防护其他疾病的遗传性状——例如镰状细胞性状（sickle cell trait）——来识别随机试验中的疫苗效力。受孟德尔随机化（Mendelian randomization）启发，我们提出孟德尔析因设计（Mendelian factorial design）：一种通过引入遗传变异对随机试验进行拓展，以构建自然析因实验的方法，可在符合实际研究假设的条件下识别疫苗效力。本文开发了一种经协变量调整的稳健估计程序，用于在风险比（risk ratio）与发病率比（incidence rate ratio）尺度上估计疫苗效力。模拟结果表明，我们的估计器性能优异，而标准方法则存在系统性偏差。我们证实，当孟德尔因子仅具备弱保护性时，结合本研究提出的估计器与标准方法的组合估计器可带来显著的性能提升。本方法可推广应用于缺乏金标准病例定义但存在已知遗传风险因素的其他儿童疾病的疫苗及预防试验。本文的补充材料可在线获取。