Pitfalls in the external validation of FRAX

Area of Science:

• Osteoporosis research
• Clinical epidemiology
• Health outcomes research

Background:

• The Fracture Risk Assessment Tool (FRAX®) is a widely used instrument for predicting fracture risk in men and women.
• Extensive validation of FRAX® has been conducted in numerous independent cohorts.
• This study critically reviews methodologies employed in post-launch evaluations of FRAX®.

Purpose of the Study:

• To critically assess the methodologies used in external validation studies of the FRAX® tool.
• To identify common methodological deficits impairing the interpretation of FRAX® performance.
• To provide guidance for future evaluations of fracture prediction models.

Main Methods:

• A critical review of published literature evaluating FRAX® calibration and performance characteristics.
• Analysis of studies comparing FRAX® with other fracture prediction models.
• Identification and critique of statistical and methodological approaches used in external cohort studies.

Main Results:

• Many studies employed inappropriate methodologies for comparing FRAX® performance.
• Common errors include discordant risk parameters, comparison with internally derived predictors, and misuse of receiver operating characteristic curves.
• These methodological deficits significantly hinder the accurate interpretation of study findings.

Conclusions:

• Existing cohort studies evaluating FRAX® performance require cautious interpretation.
• Re-evaluation of these studies using appropriate methodologies is recommended.
• Methodological rigor is crucial for the reliable assessment of fracture prediction tools.