Likelihood-based inferences for active-arm trial with counterfactual incidence based on recency assay

Area of Science:

• Epidemiology
• Biostatistics

Background:

• HIV pre-exposure prophylaxis (PrEP) trials increasingly use recency assays to estimate incidence rates.
• Current inference methods, like the Wald test, are unreliable with small or zero HIV infections in the active arm.
• More efficacious PrEP products necessitate improved statistical methods for accurate trial analysis.

Purpose of the Study:

• To develop and present a likelihood ratio test and confidence intervals for HIV prevention trials using recency assays.
• To provide a more robust statistical inference method for scenarios with limited HIV infections.

Main Methods:

• Derivation of the likelihood ratio test for HIV incidence estimation.
• Development of likelihood-based confidence intervals for trial data.
• Comparison of the proposed method with the traditional Wald test.

Main Results:

• The likelihood-based method performs reliably when zero infections are observed, unlike the Wald test.
• P-values from the likelihood ratio test increase with the number of infections, a statistically desirable property.
• The proposed method offers improved accuracy for small sample sizes or low infection counts.

Conclusions:

• Likelihood-based inference methods are preferable to Wald-based methods in HIV PrEP trials when few infections are anticipated.
• This approach enhances the reliability of efficacy estimates for novel, highly effective HIV prevention strategies.