Related Experiment Video
Updated: May 21, 2026
04:57
Establishing a Competing Risk Regression Nomogram Model for Survival Data
Published on: October 23, 2020
Competing mortality and fracture risk assessment
1University of Manitoba, Winnipeg, MB, Canada. bleslie@sbgh.mb.ca
Summary
Accounting for competing mortality is crucial for accurate 10-year fracture probability estimates, especially in high-risk groups. Failure to do so significantly overestimates fracture risk, but a modified Kaplan-Meier method offers a simple, effective adjustment.
Area of Science:
- Osteoporosis research
- Biostatistics
- Geriatric medicine
Background:
- FRAX(®) tool estimates 10-year fracture probability, incorporating mortality as a competing risk.
- The study investigated the impact of competing mortality adjustment on fracture probability estimation methods.
Purpose of the Study:
- To compare the effect of competing mortality adjustment on nonparametric and parametric fracture probability estimation.
- To evaluate the accuracy of fracture probability estimates when competing mortality is not considered.
Main Methods:
- Utilized the Manitoba Bone Mineral Density (BMD) database (N=39,063) for individuals aged ≥50 years.
- Calculated FRAX probabilities using femoral neck BMD and assessed fractures from administrative data (N=2,543 major osteoporotic, N=549 hip fractures).
Main Results:
- Failure to account for competing mortality overestimated fracture probability by 16-56% (Kaplan-Meier) and 15-29% (Cox model) in high-mortality subgroups (men, age >80, high fracture probability, diabetes).
- Overestimation for hip fractures ranged from 18-36% (Kaplan-Meier) and 17-35% (Cox model).
- A modified Kaplan-Meier method showed close agreement (within 2%) with adjusted methods.
Conclusions:
- Inaccurate fracture probability estimates result from not accounting for competing mortality, particularly in vulnerable populations.
- A modified Kaplan-Meier approach provides a practical and accurate alternative for competing mortality risk adjustment in fracture probability calculations.
