A poisson process model for hip fracture risk

Area of Science:

• Biomechanics
• Stochastic Modeling
• Osteoporosis Research

Background:

• Current osteoporosis fracture risk assessment relies heavily on bone mass measurement, with models showing limitations in predicting individual fracture events.
• Significant uncertainty persists regarding who will experience fractures, driving the search for additional factors beyond bone mass that influence bone strength.

Purpose of the Study:

• To introduce a novel mechanistic stochastic model for characterizing an individual's hip fracture risk.
• To compute hip fracture risk as a function of age and bone mineral density (BMD).

Main Methods:

• A Poisson process models fall occurrences, with fall-induced load following a Weibull distribution, creating a compound Poisson process.
• A thinned Poisson process is defined by retaining only hip fracture events, modeling fall rate with age and hip strength with BMD.
• A Bayesian framework is employed to compute conditional densities and probabilities, comparing model outputs with clinical data.

Main Results:

• The model computes hip fracture risk based on age and BMD, demonstrating consistency with clinical observations through Bayesian analysis.
• Conditional densities of BMD and probabilities of fracture, given prior fracture history, align with clinical data.
• The analysis highlights the inherent randomness of the hip fracture process.

Conclusions:

• Improvements in hip strength estimation beyond BMD may have limited clinical impact due to the highly variable and 'noisy' nature of fracture events.
• The developed model provides a new mechanistic approach to understanding and quantifying individual hip fracture risk.