“But why do we choose blood pressure drug classes in this order?” The question came at the end of a Grand Round presentation I had given at my hospital. I was momentarily stumped. Having worked in hypertension and nephrology for years, I knew about low-renin, salt-sensitive phenotypes. But when tenets of clinical practice are long-standing, my experience is that we absorb them completely and assume that there must be strong evidence underlying them. Had I forgotten some critical paper or important piece of this evidence that everyone else knew? My colleague Adam MacDiarmaid-Gordon stepped in to rescue me: “It goes back to this paper in The Lancet…”

That question is what led to us to undertake this work examining whether the UK algorithm for first line treatment of hypertension leads to greater reductions of blood pressure. The findings suggest that it does not. 

In the UK we are fortunate to be able to use our anonymised health care records for research. These contain vast amounts of data regarding many physiological variables, including blood pressure and kidney function. Great efforts must be taken in analyses of these data to minimise sources of bias and confounding, but they allow in depth exploration of questions related to current clinical care. By contrast, clinical trials minimise confounding due to randomisation, and are the cornerstone of defining treatment effectiveness. Modern trials are usually large, population-representative, and well conducted. The evidence they generate is robust and useful. But unfortunately many areas of clinical medicine are informed by data from an era where trials were not so well-conducted, or are based not on trials, but on clinical studies which we might not accept as an adequate base to direct medical practice if we started afresh. And importantly, that research was designed and influenced by the beliefs that clinicians and academics held at the time. In the case of hypertension, the underpinning clinical studies were informed by beliefs about what constitutes older age and the impact of racial differences in the aetiology of hypertension, ideas that are now being re-examined. 

In the face of a number of recent editorials criticising “real-world studies”, it is tempting for clinicians and policy makers to be wary of evidence from observational studies. While all agree about the importance of high quality randomised trial data, we argue that observational evidence can have a valuable role in appraising clinical practice, particularly in the light of a changing world and recognition that standards of research have progressed over time. 

We hope our results contribute to a re-evaluation of the evidence that underpins the UK guidance for initiation of blood pressure treatment. There are many additional and relevant areas of research, like an ongoing clinical trial  investigating whether using genetic markers of ancestry to guide choice of blood pressure treatment leads to better control. Indeed, there have been calls to simplify the guidance on hypertension management for many years. The potential downsides of raising questions about a long-established algorithm for managing hypertension, when a substantial number of people have untreated or sub-optimal blood pressure control need consideration. However, there are many areas of routine clinical care that can benefit from closer scrutiny and we should embrace research investigating these topics, whether the evidence is randomised or not.

Laurie A Tomlinson, associate professor, Department of Non-Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine.

Competing interests: See research paper for declaration.