Estimating Antigen Test Sensitivity via Target Distribution Balancing: Development and Validation Study

Area of Science:

• Biostatistics
• Infectious Disease Diagnostics
• Medical Device Evaluation

Background:

• Lateral-flow antigen tests (ATs) are evaluated for sensitivity using percent positive agreement (PPA) against quantitative reverse transcription polymerase chain reaction (qRT-PCR).
• Observed PPA is influenced by viral load distribution (inferred from cycle threshold [Ct] values), leading to biased sensitivity estimates and complicating cross-study comparisons.
• Real-world data often lack standardized viral load sampling, deviating from regulatory guidelines and hindering accurate diagnostic test assessment.

Purpose of the Study:

• To develop and validate a statistical methodology for recalibrating AT clinical performance data.
• To enable accurate and comparable sensitivity assessments by standardizing viral load distributions.
• To correct for sampling bias in diagnostic test evaluations.

Main Methods:

• Modeled percent positive agreement (PPA) as a logistic function of qRT-PCR Ct values (PPA function).
• Applied the PPA function to a standardized reference Ct distribution to calculate adjusted sensitivity.
• Validated the methodology using clinical data from a community study, demonstrating reduced sampling bias.

Main Results:

• Analyzed paired AT and qRT-PCR data from four suppliers over two years, observing significant variations in Ct value distributions.
• Logistic regression modeling of PPA against Ct values corrected for viral load variability.
• Adjusted sensitivity estimates provided more accurate and consistent comparisons of AT performance across suppliers.

Conclusions:

• Introduced a distribution-aware framework for bias-corrected sensitivity estimation of ATs.
• The method facilitates fairer, more consistent comparisons across AT suppliers and studies, enhancing quality control and regulatory review.
• Emphasized the importance of distribution-aware evaluation for robust diagnostic test assessment.