Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The treatment of hypertension has been a therapeutic success. A generation or more of effective drugs deserves considerable credit for their contribution to the substantial decline in age-related incidence of stroke, ischaemic heart disease and heart failure. And because almost all the drugs are long-since off patent, the cost of success comes cheaply. Indeed, National Institute of Health and Care Excellence (NICE) has branded treatment of hypertension as not only cost effective but cost saving.1 Yet not all patients achieve their blood pressure target and are labelled as ‘resistant hypertension’.
A contentious question has been whether resistant hypertension is a pathogenetic subset of hypertension, justifying a search for ‘stratified medicines’; or is it an imaginary condition caused by doctors in white coats and patients who do not take their tablets? On the one hand, there is abundant evidence that primary aldosteronism causes 20–25% of true treatment resistance.2 But Tomaszewski et al show that the ‘untakers’ (to create a simple noun for ‘patients who are non-adherent to their prescribed therapy’) account for an almost identical, high proportion of supposed resistant hypertension. Furthermore, the authors answer not only the question ‘Is my patient taking his/her tablets’, but also ‘How can I simply and cheaply find out?’.
Non-adherence to therapy, and its recognition, is a particular problem in hypertension because of its chronicity and asymptomatic nature (at least until drug treatment is started). To an extent, it can be assumed that few patients take every tablet every day and that our evidence from outcome trials of long-term efficacy is based on average consumption around the 70% level that some trials impose (via returned tablets counts) as an eligibility criterion. Omission of occasional, even frequent, doses of single drugs may be without consequence. After all, there is no evidence of superior outcome in patients receiving longer versus short-acting members of the same class. Indeed, the most widely prescribed ACE inhibitor for hypertension, Ramipril, claimed to reduce daytime blood pressure by only 3 mm Hg while reducing complications of hypertension by 30%; it later transpired that the drug was taken at bedtime, with much larger overnight falls.3
In recent years, it has become recognised that most patients with hypertension have a complex disorder requiring treatment with multiple drugs: ‘A+C’, or ‘A+C+D’, according to the most recent NICE guidance. Doctors are encouraged to add drugs, and if a patient's BP can be controlled with 2–3 cheap generics, there is little case for checking adherence to each of these. However, once patients attract the ‘resistant’ label, the ramifications spiral with the number of drugs prescribed. Not only do the costs of drugs grow but the number of visits, referrals to specialists and consequent expensive investigations and interventions. The consequences become most serious when an intervention is ineffective and irreversible, and this unfortunately now seems to be the case for renal denervation. Despite the initial hope and promise from the Symplicity 1 and 2 studies, Symplicity 3 (which randomised 535 patients and had a sham control) reported no benefit.4
Unfortunately, the implausible message from the earlier studies was that all patients with resistant hypertension responded to renal denervation. There could still be a small subset, with renal sympathetic overactivity, who would truly benefit. But the dramatic finding by Tomaszewski et al is that the largest and most homogeneous subgroup of patients referred for renal denervation are the untakers with no detectable antihypertensive drugs in their urine. Until now there has been no simple, cheap and reliable test that clinicians can use to find out whether their patients are taking their drugs. In research studies, it is standard to issue more drugs than patients require and count tablet returns. For trials with a specific interest in adherence, a ‘Medication event monitoring system’ has been used; this device records the number of times per day that the medicine container is opened, but does not appear to improve adherence.5 We probably underuse pharmacy fill records to detect some blatant cases of non-adherence. An increasingly common clinical practice is to undertake ‘directly observed therapy’ (DOT), and most hypertension specialists have anecdotes of patients who swear to compulsive tablet taking, but collapse on the ward floor when administered a fraction of their supposed regimen. DOT is not a trivial exercise, requiring staff, time and a bed for the collapsing patient. Neither is it proof against false positives (the placebo or rest effects of a day of observation), nor against false negatives (the patient who secretes rather than swallows their tablets). I recall an ex-lawyer who was undone only when gullible enough to swallow the suggestion that his tablets were becoming stuck in his gullet and required delivery by nasogastric tube. My own ploy, for an initial outpatient assessment, has been to measure serum ACE activity in patients receiving an ACE inhibitor—and if not, to add a small dose (say, lisinopril 5 mg) to their current regimen. Since ACE inhibitors require to inhibit >99% of ACE activity to be therapeutically active, the serum activity should be close to undetectable even at trough drug levels.6
The Leicester findings, and the enabling comprehensive mass spectrometry assay for all commonly used antihypertensive drugs, solve at a stroke the problem of monitoring adherence and should rapidly transform practice. That most patients do not take all their drugs all the time was probably predictable. But that 23% of those referred for renal denervation have no detectable drug in their urine was a shock. These patients may be the extremes, both among those who are non-adherent and among those referred for renal denervation. But they account for a disproportionate and wasteful use of resource in their management. If they lie somewhere in the Munchausen spectrum, it is perhaps no surprise how many of those undergoing renal (or, we must now presume, sham renal) denervation can mount a large placebo response in clinic blood pressure. But with renal denervation appearing unlikely now to feature in clinical practice, the relevance of the Leicester findings should be generalised to any patient apparently uncontrolled despite multiple drugs.
The authors’ presentation to the British Hypertension Society in 2013 reported an inverse correlation between the number of prescribed drugs and number detected in urine, with none detectable in three patients prescribed seven drugs. Our own anecdotal experience, stimulated by these results and enabled by the Leicester assay, matches exactly. No drug was detectable in three patients referred on six or seven drugs, and blood pressures exceeding 180/120 mm Hg. One had even passed DOT! By contrast, in patients uncontrolled on five drugs, the picture is more mixed; and where true resistance is proven by detection of all drugs, a successful search for a secondary cause of hypertension often follows. The lesson may be that few patients truly require six or more drugs: what they require is simplification of the drug regimen, a search for a cause—or a psychiatrist! This fits with the experience in the British Hypertension Society's PATHWAY-2 study of resistant hypertension. Using multiple measures to exclude both white coat hypertension and non-adherence, it has taken the 10 main specialist centres in the UK almost 5 years to find 346 patients with a systolic blood pressure >140 mm Hg despite treatment with three drugs.7 Although their average entry systolic is close to 160 mm Hg, it is apparent that only a handful will finish the study above this level—the entry level to the first two studies of renal denervation. The implication is that few of the patients in these studies had true resistance to therapy. It seems reasonable now to check the urine of all patients supposedly receiving treatment with five or more drugs and a systolic blood pressure greater than 160 mm Hg.
One issue deftly side-stepped by the authors is that of consent for the test, and the clinical implementation of the result. It could be argued that the aim of checking for a drug in urine is ethically no different from measuring ACE activity in a patient prescribed an ACE inhibitor. Better may be to add consent for the test to a copy of the Morisky questionnaire (figure 1), claimed to be a simple predictive clinical tool for non-adherence.8 Neither completion nor assent to urine testing could be compulsory. But then nor is progression to expensive tests for benign, curable causes of resistant hypertension—without resistance to treatment being first demonstrated.
Philosophers define knowledge as belief with evidence. No longer need we guess whether patients have resistant hypertension. We can know.
Competing interests None.
Provenance and peer review Commissioned; internally peer reviewed.