Diagnostic Test Accuracy Meta-analysis: A Practical Guide to Hierarchical Models

Area of Science:

• Biostatistics
• Medical Informatics
• Epidemiology

Background:

• Accurate evaluation of diagnostic tests is crucial for clinical decision-making.
• Systematic reviews and meta-analyses of diagnostic test accuracy (DTA) are vital for evidence synthesis.
• Traditional DTA meta-analysis methods can produce biased and misleading estimates.

Purpose of the Study:

• To provide a methodological guide for hierarchical random-effects models in DTA meta-analysis.
• To compare hierarchical models with traditional methods using a simulated dataset.
• To guide researchers in selecting appropriate Stata commands for DTA meta-analysis.

Main Methods:

• A methodological framework for hierarchical random-effects models in DTA meta-analysis was presented.
• A simulated dataset was used for comparative meta-analysis (traditional vs. hierarchical).
• Key concepts, interpretation of outputs, and complementary tools (outlier detection, meta-regression, bias assessment) were illustrated.

Main Results:

• Traditional meta-analysis yielded biased estimates with narrow confidence intervals, ignoring heterogeneity.
• Hierarchical models provided realistic, interpretable estimates with joint modeling of sensitivity and specificity.
• Hierarchical models revealed between-study variability, threshold effects, and allowed for covariate analysis via meta-regression.

Conclusions:

• Hierarchical random-effects models offer superior accuracy and interpretability for DTA meta-analyses.
• Correct application and interpretation of these models are essential for high-quality evidence.
• This approach enhances the reliability and robustness of synthesized diagnostic test evidence.