PREDITING THE RISK OF FRACTURE ASSOCIATED WITH ANTIHYPERTENSIVE TREATMENT: DEVELOPMENT OF THE STRATIFY-FRACTURE PREDICTION MODEL

Abstract

Objective: The benefits of antihypertensive treatment are well known, but less is known about the potential risk of harm from antihypertensive medication. The STRAtifying Treatments. In the multi-morbid Frail elderlY (STRATIFY) study aimed to better understand the harms of antihypertensive treatment by quantifying an individual's baseline risk of different serious adverse events. This current study aimed to develop a prediction model for an individual's risk of hospitalisation or death from fracture. Design and method: Participants aged > = 40 years, registered to a UK primary care practice within the Clinical Pr: actice Research Datalink (CPRD) GOLD database, with at least one blood pressure measurement between 130-179 mm Hg were included in this cohort study. The outcome investigated was fracture that led to hospital admission or death within 10 years after the index date. The (prespecified) predictor variables based on the literature and expert opinions were patient characteristics, history of falls and fractures (from the age of 40), diagnosis of osteoporosis and rheumatoid arthritis, other comorbidities, and prescribed medications (including antihypertensive medications). Multiple imputation was used to account for missing data. The primary analysis used a Fine-Gray competing risks approach to adjust death from other causes. Results were reported as subdistribution hazard ratios (SHR). Results: A total of 1,772,601 participants were included, with a mean age of 59 years and median follow-up of 6.1 years. The 10-year incidence of fracture was 1.4%. All antihypertensive medications were significantly associated with the risk of fracture, as were a number of patient characteristics and co-morbidities (detailed in Table 1). Potential predictors with no significant association were chronic renal disease, mobility and transfer problems, hypogonadism, vitamin D deficiency and aromatase inhibitors. SHRs and 95% confidence intervals for the predictors are shown in Table 1. Conclusions: In this study, we identified a number of predictors of fracture using UK electronic health record and quantified an individual's baseline risk of fracture. The external validation of this model will use a different database (CPRD AURUM). Our findings could inform clinicians as to when it is appropriate to prescribe and deprescribe antihypertensive treatment.