The dilemma of placebo controlled studies: scientific evidence, guidelines, ethics and regulatory recommendations

The use of placebo in hypertension studies, although largely debated at the beginning of the antihypertensive therapy story [1,2], has come to be widely accepted [2] and has been instrumental in precisely measuring the efficacy of blood pressure lowering drugs [3] and in running trials that have shown outcome reduction by blood pressure lowering [4]. The body of evidence brought about by the investigations carried out in the last several decades has deeply changed the scenario in which placebo is used in hypertension research. The actual scenario, however, is somewhat confused because decisions about using or avoiding a placebo are often contradictory. As a result ethics committees as well as journal reviewers and editors are sometimes confronted with the dilemma whether, in a given study, placebo should be or have been avoided or conversely whether the drug under study should have been compared with a placebo rather than with another treated group. In other words the actual scenario is confronted with both underuse and overuse of placebo and thus with trials characterized by respectively undertreatment and overtreatment. Underusage of placebo results from too enthusiastic recommendations (often also issued by some guidelines) to promptly lower blood pressure in all subjects with blood pressure ≥140/90 mmHg, including those with grade 1 (or stage 1) hypertension unaccompanied by other risk factors, despite no controlled therapeutic trial has been conducted in these patients and therefore no hard evidence is available, as explicitly recognized by ESH-ESC guidelines [5]. Recommendations like these, though certainly not endangering the patient, have hindered and are hindering the possibility of running a placebo-controlled trial in subjects with just a modest BP elevation, to finally clarify the issue and provide undisputed evidence. Likewise, the statement is made in all guidelines that blood pressure should be lowered below 130/80 mmHg in diabetics, patients with previous cardiovascular events or renal insufficiency, although, e.g., the ESH-ESC guidelines honestly recognize evidence is scanty and controversial [5], and achievement of those low blood pressure values is very difficult and unfrequent [6]. Underusage of placebo in trials has also been caused by the rush to comparing different initial hypertensive regimens, in search of small and often inconsistent differences in outcome, with the frustrating results all those interested in hypertension well know [7]. In this context, placebo has too often been relegated to add on treatment (in comparison with an active agent) on top of multiple antihypertensive agents (and nowadays, multiple cardiovascular therapies, such as statins and antiplatelet agents as well) to be administered to severely complicated patients who seldom represent the relatively healthy hypertensive subjects a general practitioner takes care of. The opportunity has been lost to investigate new antihypertensive agents for their possible benefits in hypertensive patients at low or moderate added risk by comparing them to placebo on the unconfounded background, for example, of a simple monotherapy, or to use new agents in order to achieve a more ambitious blood pressure goal and prove the benefits of this achievement. However, it must be recognized that placebo is also overused. This mostly occurs, for example, in the testing of dose-response relationships and measuring the blood pressure lowering effects of new agents or fixed-combinations of two or more antihypertensive agents. This is commonly done for regulatory purpose. Regulatory authorities are commendably eager to check efficacy and safety of new drugs or drug combinations, and a comparative group under placebo is the best possible control. A placebo group is also useful in order to have a pharmacologically clean design, and measuring precisely what is the real blood pressure decrease produced by the agent being tested when the placebo effect is subtracted. If, for example, the European Medicine Agency (EMEA) Principles for Clinical Evaluation of New Antihypertensive Drugs [7] are consulted, the requirement of a placebo or no treatment period or group is explicitly mentioned. ‘The primary endpoint of studies to assess antihypertensive effect is the absolute change at the end of the dosing interval (trough) from drug-free baseline blood pressure compared to the change in the control group. ‘The effect on blood pressure…should be characterized in short-term studies (4–12 weeks), whose short duration allows for use of a placebo control’. ‘Studies conducted in a blinded fashion and with placebo controls are essential’ (and four different types of placebo-controlled studies are specifically indicated). It is certainly true that the EMEA guidelines prudently state that ‘most patients will be in the mild to moderate range’ and that ‘in general, patients with target organ damage secondary to hypertension should not be included in placebo-controlled trials of more than very brief duration’. However, it would be preferable to know what are the blood pressure values in mmHg that, in the opinion of the regulatory authorities, make a placebo period ethically acceptable: terms such as mild and moderate hypertension have long been dropped by all guidelines, and substituted by categories (grade or stage 1, grade or stage 2) based on mmHg. It would also be useful to know which hypertensive patients can receive placebo treatment for 12 weeks. It has been mentioned above that evidence for prompt drug treatment of grade 1 uncomplicated hypertensive patients is still weak, and these patients can reasonably be included in placebo-controlled trials. However, is it acceptable to allocate hypertensive patients with blood pressures ≥160/100 mmHg to placebo treatment for up to 12 weeks? ESH-ESC guidelines point out that ‘All patients in whom repeated blood pressure measurements show grade 2 or 3 hypertension are definite candidates for an antihypertensive treatment because a large number of placebo controlled trials have conclusively demonstrated that in patients with these blood pressure values blood pressure reduction lowers the incidence of cardiovascular morbid and fatal events, independently of their level of total risk (i.e. moderate, high or very high)’ [5]. Admittedly, ESH-ESC guidelines [5] suggest a treatment algorithm by which grade 2 hypertensive patients may have drug treatment deferred for several weeks (‘several’ means ‘more than two or three but not many’ according to the Oxford Dictionary), but in order to test the effectiveness of intense lifestyle changes (what would be a confounding measure in a drug testing study). The risk of including grade 2, and even grade 3 hypertensive patients in placebo-controlled studies, and of withdrawing multiple therapies (for at least eight weeks, if a real baseline is required) in order to obtain a therapy-free baseline, has considerably increased recently because of the great (and justified) interest in fixed combinations of two or three antihypertensive agents. For these studies, the EMEA guidelines [8] require a factorial study: ‘in a factorial study, placebo, and one or more doses of the test drug T and another drug D are studied alone and in the combination in a short-term randomized controlled trial… It may be necessary to study separately the low doses and low dose combinations to establish their specific usefulness’. For testing combination therapy patients with rather high baseline blood pressures (grade 2 or 3) are obvious candidates, and a fraction of these patients have the chance of being exposed to placebo or low (and uneffective) doses of the drugs being tested for up to 12 weeks. The well justified interest for fixed combinations of multiple drugs [5,9], probably the best tool for achieving a more effective control of hypertension, should stimulate the regulatory authorities to accept different experimental designs from the traditional ones. Here, scientific societies, such as the European Society of Hypertension for Europe and other national or continental societies elsewhere, may contribute ideas and designs. Probably, comparison of placebo-subtracted blood pressure differences from a drug-free baseline is an outdated and ethically doubtful approach, which will become less and less practicable in the future. Switching from pre-existing therapy to blinded therapy (without placebo) and testing not the blood pressure difference (which is not what a physician is measuring and the patient is caring about), but the different blood pressure values achieved with the various doses or drug combinations may provide regulatory authorities with the comparative information they need, and more closely match the use of antihypertensive therapy in medical practice. As additional benefit, ethics review committees, journal reviewers and editors will avoid being confronted with the unpleasant dilemma how to judge ethically controversial studies, which are however required by regulatory authorities.