A new comparison of nested case-control and case-cohort designs and methods

Area of Science:

• Epidemiology
• Biostatistics
• Statistical Genetics

Background:

• Existing literature comparing nested case-control and case-cohort methods is insufficient for current epidemiological research.
• Conflicting conclusions in prior studies necessitate updated comparisons.
• Newly developed methods, including inverse probability weighting, require evaluation alongside traditional approaches.

Purpose of the Study:

• To compare the statistical properties and power of various analytical methods for nested case-control and case-cohort designs.
• To reconcile conflicting conclusions regarding the efficiency of standard methods.
• To evaluate the performance of inverse probability weighting methods within these designs.

Main Methods:

• Summarized two analytical methods for nested case-control studies.
• Summarized six analytical methods for case-cohort studies, all using proportional hazards regression.
• Compared statistical properties, including statistical power, under different censoring types and event proportions.

Main Results:

• Inverse probability weighting methods demonstrated greater power than standard methods for both designs, particularly when event proportions were not very low.
• The relative efficiency of nested case-control versus case-cohort designs varied with censoring type and incidence proportion.
• Nested case-control designs with inverse probability weighting were most powerful under random censoring.
• Under fixed censoring, inverse probability weighting methods showed similar efficiency between designs, while standard case-cohort methods outperformed the conditional logistic method.
• Nested case-control methods outperformed case-cohort methods when the failure event proportion was low (<10%), reducing the importance of analytic method choice.
• Standard case-cohort methods were often more powerful than the conditional logistic method for nested case-control designs when the predictor was binary.

Conclusions:

• The choice between nested case-control and case-cohort designs, and the specific analytical method, is nuanced and depends heavily on study-specific factors like censoring and event incidence.
• Inverse probability weighting methods offer a powerful alternative to standard methods, especially in scenarios with sufficient event occurrences.
• Epidemiologists should carefully consider censoring patterns and event proportions when selecting a design and analytical strategy for efficiency.