Effect of antihypertensive medication reduction on short-term blood pressure control in older adults: calibration of OPTiMISE trial results to real-world primary care data

Background: While antihypertensive treatment prevents cardiovascular events, it may also increase risks such as falls in patients with frailty. The Optimising Treatment for Mild Systolic Hypertension in the Elderly (OPTiMISE) trial found that deprescribing one antihypertensive drug did not result in worse short-term blood pressure control and long-term follow-up showed no observed harm, but its generalisability to routine clinical practice remains uncertain. Objective: This study aimed to calibrate the OPTiMISE effect to a representative primary care population in England. Methods: We calibrated the OPTiMISE treatment effect using inverse probability weighting (IPW) based on trial inclusion likelihood. The trial enrolled 569 adults aged ≥80 years with controlled blood pressure on ≥2 antihypertensive drugs. A target population was reconstructed from electronic health records of 24 participating practices and extrapolated to all English adults aged ≥80 years, prescribed ≥2 antihypertensive drugs, using NHS Digital data. The primary outcome was all-cause hospitalisation or death. Weighted Cox models estimated calibrated hazard ratios (HRs). Results: The target population included 798 179 individuals (median age 84 years [81–87], 48% female). Compared with OPTiMISE participants, a higher proportion of individuals in target population were overtly frail (25% vs. 11%). After calibration using IPW, deprescribing was not associated with higher risk of hospitalisation or death [calibrated HR 0.94 (95% CI: 0.73–1.22)], similar to the long-term follow-up of OPTiMISE [HR 0.93 (95% CI: 0.76–1.12)], although with a slightly wider confidence interval. Conclusions: Calibrating OPTiMISE findings to a representative primary care population frailer than the original participants suggests that the original trial findings could be translated into a real-world population.