Bone densitometry categories as a salient distracting feature in the modern clinical pathways of osteoporosis care: A retrospective 20-year cohort study

Background: It is unknown as to what factors typically drive osteoporosis therapy decisions in real-world practice.

Methods: Retrospective, 20-year cohort study within the government health system of Manitoba including all women having bone densitometry (BMD) tests between 1996 and 2017. Osteoporosis prescription data was linked to registry data on fractures, clinical risk factors and BMD tests. We defined 6 possible treatment decisions by prescription data: no treatment, starting, stopping, continuing, drug hiatus and re-starting. For each decision, we tested the association between salient patient factors (age, glucocorticoid use, recent fracture, BMD hip or spine T-score ≤ -2.5, FRAX major osteoporotic fracture probability ≥20%) using multivariable logistic regression. The factors were rank-ordered by decreasing Wald χ² statistic to determine the relative importance.

Results: There were 64,181 women, 33.8% of whom started osteoporosis therapy. For patients who begin therapy after a first BMD, the rank-ordered multivariable logistic regression factor most strongly associated was the T-score ≤ -2.5 [OR of 7.59(95%CI 7.19-8.01, p < 0.001)]. This was followed by glucocorticoid use [OR 2.89(95%CI 2.59-3.22, p < 0.001)]. Increasing age and recent fracture (within 2 years) were weak predictors of therapy and high FRAX score associated with reduced odds of therapy [OR 0.80 (95%CI 0.74-0.88, p < 0.001)]. T-scores were the strongest factor predicting therapy stop/continuation/re-starting; age and prior fracture had weak or no associations.

Conclusions: Despite recommendations for fracture-risk-based approach to osteoporosis therapy, BMD T-score continues to be the dominant factor in actual practice. Age, prior fracture or global fracture risk are much less associated; it is possible that BMD T-score categories are therefore acting as a clinically salient distracting factor.