Methodological issues in case-control studies IV: Validity and efficiency of various analysis strategies for...

Area of Science:

• Epidemiology
• Biostatistics
• Statistical Modeling

Background:

• Case-control studies are crucial for investigating disease risk factors.
• Analyzing continuous data in case-control studies presents statistical challenges.
• Unconditional logistic regression is a common analytical approach.

Purpose of the Study:

• To evaluate the performance of unconditional logistic regression for continuous case-control data.
• To quantify bias in odds ratio estimates due to sample size and model choice.
• To assess the statistical power under different population conditions.

Main Methods:

• Computer simulations were employed to model various scenarios.
• The unconditional logistic regression model was applied to simulated continuous data.
• Bias and power were estimated across different exposure levels and sample sizes.

Main Results:

• The unconditional logistic regression model demonstrated robustness with continuous data.
• Small sample sizes introduced relatively small biases in odds ratio estimates.
• Using an incorrect analysis model also resulted in minimal bias.
• Statistical power remained largely unaffected by model choice.

Conclusions:

• Unconditional logistic regression is a reliable method for analyzing continuous data in case-control studies.
• The model is resilient to limitations such as small sample sizes and model misspecification.
• Researchers can confidently apply this model with minimal concern for bias or power reduction.