Effects of FRAX(®) model calibration on intervention rates: a simulation study

Area of Science:

• Osteoporosis research
• Clinical epidemiology
• Health technology assessment

Background:

• The World Health Organization fracture risk assessment tool (FRAX®) predicts 10-year major osteoporotic and hip fracture probabilities.
• FRAX® models are population-specific, requiring accurate fracture epidemiology for calibration.
• Model miscalibration can lead to inappropriate clinical practice and treatment recommendations.

Purpose of the Study:

• To investigate the impact of simulated miscalibration in the FRAX® tool on intervention rates.
• To quantify the relationship between FRAX® model calibration error and changes in treatment eligibility.

Main Methods:

• Analysis of a cohort of 36,730 women and 2873 men aged 50+ with FRAX® estimates.
• Simulation of relative miscalibration errors from -50% to +50% in 10% increments.
• Assessment of changes in the number of individuals meeting intervention thresholds (≥20% major osteoporotic fracture, ≥3% hip fracture probability).

Main Results:

• Small calibration changes (e.g., 10%) substantially altered the number of individuals qualifying for treatment.
• A steep gradient was observed: a 1% calibration change resulted in a 2.5% (women) and 4.1% (men) change in intervention rates for major osteoporotic fractures.
• The relationship for hip fracture probability was closer to unity.

Conclusions:

• FRAX® model calibration is critically important for accurate clinical decision-making.
• High-quality fracture epidemiology data is essential for developing reliable FRAX® tools.
• Ensuring precise calibration minimizes misclassification and optimizes osteoporosis treatment allocation.