Replicate Testing of Clinical Endpoints Can Prevent No-Go Decisions for Beneficial Vaccines

In vaccine efficacy trials, inaccurate counting of infection cases leads to systematic under-estimation-or dilution-of vaccine efficacy. In particular, if a sufficient fraction of observed cases are false positives, apparent efficacy will be greatly reduced, leading to unwarranted no-go decisions in vaccine development. Here, we propose a range of replicate testing strategies to address this problem, considering the additional challenge of uncertainty in both infection incidence and diagnostic assay specificity/sensitivity. A strategy that counts an infection case only if a majority of replicate assays return a positive result can substantially reduce efficacy dilution for assays with non-systematic (i.e., random) errors. We also find that a cost-effective variant of this strategy, using confirmatory assays only if an initial assay is positive, yields a comparable benefit. In clinical trials, where frequent longitudinal samples are needed to detect short-lived infections, this confirmatory majority rule strategy can prevent the accumulation of false positives from magnifying efficacy dilution. When widespread public health screening is used for viruses, such as SARS-CoV-2, that have non-differentiating features or may be asymptomatic, these strategies can also serve to reduce unneeded isolations caused by false positives.

Automated summary

Inaccurate counting of infection cases in vaccine efficacy trials can lead to underestimation of vaccine efficacy. To address this issue, we propose a range of replicate testing strategies that can reduce dilution of efficacy caused by false positives. These strategies are beneficial in clinical trials and public health screening to prevent unnecessary impacts from false positives.