Better data in, better research out: the data problem in resistant hypertension research

Resistant hypertension (RH) remains one of the most heterogeneous and inconsistently defined conditions in cardiovascular medicine. Despite the increased risk of major adverse cardiovascular events and the accompanying healthcare burden, the epidemiology, underlying biology, and optimal management of RH are poorly understood [ 1 , 2 ]. One of the main reasons for this is limited data availability: inconsistent drug reporting, absent adherence data, and underpowered cohorts make robust phenotyping near impossible [ 3 , 4 ]. Without a well-defined, large-scale phenotype, genomic discovery stalls before it can start. As a result, the genetic architecture of RH remains a largely unexplored landscape. The marked difference in reported prevalence of RH across the globe underpins how poorly understood RH is. Estimates vary widely, ranging from 5% to 30% across global hypertension guidelines, and accompanying treatment strategies are often limited to the addition of further antihypertensive drugs [ 5 , 6 ]. Embedded within this variability is a more fundamental issue: a substantial proportion of RH reflects pseudo-resistance rather than true treatment resistance [ 6 ]. Where white-coat effects and measurement errors can often be rectified with ambulatory or home blood pressure monitoring, non-adherence presents a unique challenge and is one of the most important reasons for pseudo-resistant hypertension. The prevailing clinical definition of blood pressure remaining above target despite concurrent use of three or more antihypertensives assumes prescribed medications are taken as intended [ 6 , 7 ]. However, in reality, we know this is often incorrect, particularly in patients where pill burden is already high due to the polypharmacy approach to primary hypertension and the limited availability of single-pill combinations, especially in the UK [ 8 ]. In fact, adherence to antihypertensives is remarkably poor, with reported rates between 27 and 40%, an estimate limited further by the lack of a “gold-standard” adherence measure [ 9 ]. As a result, distinguishing true resistance from pseudo-resistance depends on data that is not routinely captured or that which inaccurately captures true adherence patterns.