Skewness-Corrected Confidence Intervals for Predictive Values in Enrichment Studies

Area of Science:

• Biostatistics
• Epidemiology
• Medical Diagnostics

Background:

• Positive predictive value (PPV) and negative predictive value (NPV) are crucial metrics in diagnostic test evaluation.
• Calculating confidence intervals (CIs) for PPV and NPV is straightforward in prospective studies but challenging in enrichment designs like case-control studies due to differing disease prevalence.
• Extreme conditions, such as very low or high disease prevalence, exacerbate CI estimation difficulties.

Purpose of the Study:

• To extend existing methods for calculating CIs of PPV and NPV in enrichment studies.
• To evaluate novel CI methods for PPV and NPV estimation under challenging conditions.

Main Methods:

• The study extends Li's method for deriving PPV and NPV CIs from ratio of binomial proportions CIs.
• Additional CI methods for the ratio of binomial proportions, including Gart & Nam (GN), MoverJ, and Walter, were explored and converted for PPV and NPV.
• Simulations were conducted to compare the performance of these extended methods against established methods (Fieller, Pepe, Delta) regarding skewness and coverage.

Main Results:

• No single method demonstrated universal optimality across all scenarios.
• The Gart & Nam (GN) and MoverJ methods generally exhibited favorable performance in terms of skewness and coverage compared to other methods.
• These methods offer improved CI estimation for PPV and NPV in enrichment studies, particularly under extreme conditions.

Conclusions:

• The Gart & Nam (GN) and MoverJ methods are recommended for calculating confidence intervals of PPV and NPV in enrichment studies.
• These methods provide more reliable estimates, addressing the challenges posed by differing disease prevalence and extreme conditions.
• Further research may explore additional methods or refinements for robust diagnostic accuracy assessment.