Predicting hypotension, syncope, and fracture risk in patients indicated for antihypertensive treatment: the STRATIFY models
Constantinos Koshiaris, Ariel Wang, Lucinda Archer, Richard D. Riley, Kym IE Snell, Richard J. Stevens, Amitava Banerjee, Subhashisa Swain, Andrew Clegg, Christopher E. Clark, Rupert A. Payne, FD Richard Hobbs, Margaret Ogden, Richard J. McManus & James P. Sheppard On behalf of the STRATIFY investigators
Abstract Antihypertensives are associated with increased risk of syncope, hypotension, and fractures, but the highest-risk individuals are unclear. This study aimed to develop and validate three models to predict these outcomes in patients with an indication for antihypertensive treatment. A cohort study was conducted using data from Clinical Practice Research Datalink (CPRD). Patients aged 40+ with systolic blood pressure 130-179 mmHg were included. Outcomes were first hypotension, syncope, or fracture leading to hospitalization or death within 10 years. Models were derived from CPRD GOLD data (n = 1,773,224) and validated with CPRD Aurum data (n = 3,805,366). Each model had 31-37 predictors. Validation demonstrated strong discriminative ability (10-year C-statistic: hypotension 0.824; syncope 0.819; fracture 0.790), with close agreement between predicted and observed risks for the hypotension and syncope models. Some underprediction was observed for the fracture model. These models could be used to help reassure patients about the relatively low risk of harm from antihypertensive treatment, or identify the small number of individuals with a higher risk where additional monitoring may be indicated.