Treatment of hypertension and the price to pay; adverse events and discontinuation from randomized treatment in clinical trials

Meta-analyses of randomized controlled trials of blood pressure (BP)-lowering treatment show that treatment lowers risk of cardiovascular outcomes like cerebral stroke, myocardial infarction, heart failure, renal failure, cardiovascular death and total mortality in hypertensive patients. Also novel endpoints like atrial fibrillation, incident diabetes mellitus and Alzheimer type of cognitive decline may be prevented though stronger evidence in long-lasting studies is needed for the latter. However, treatment comes with a price to pay because many patients discontinue treatment even in clinical trials, possibly because of side-effects of the treatment that is given. Side-effect may be drug specific or nondrug specific, and they are therefore termed adverse events. An adverse event is thus any untoward event that happens during the course of a clinical trial and is reported by the study investigators. Adverse events associated with BP-lowering treatment have not before been systematically meta-analyzed in large scale in the same way as the benefits of preventing the cardiovascular and renal events caused by treatment of hypertension. Thomopoulos, Parati and Zanchetti have published a series of important meta-analyses of various aspects of the treatment of hypertension in Journal of Hypertension through the past 2 years. They have created a database containing literally all important details from the randomized and controlled hypertension trials that have been published to date. In the database, they have also included adverse events reported in the trials, and that could be extracted from a majority of the publications. This issue of the Journal [1] contains a report on this aspect of the clinical trials. Among the BP-lowering trials with the design of active treatment vs. placebo, or less active treatment vs. more active treatment, Thomopoulos, Parati and Zanchetti [1] scrutinized a total of 70 trials. They have taken permanent treatment discontinuation attributed to adverse events as the index of a relevant incident. Their findings have then been standardized to SBP and DBP reductions of 10 over 5 mmHg, and also related to the fatal and nonfatal cardiovascular outcomes. In 44 randomized trials, data were published on treatment discontinuations explained by adverse events, and other six trials reported serious adverse events due to the treatment. These trials included 179 949 randomized participants who contributed with 719 796 patient-years. In these 50 randomized trials, they detected 24% reduction of major cardiovascular events vs. 89% increase in the risk of discontinuations. More precisely, 33 major cardiovascular events were prevented, and 84 excess discontinuations took place per 1000 patients for 5 years. Further meta-regression analyses indicated that both outcome reductions and excess of discontinuation caused by treatment were related to the extent of SBP and DBP reductions, but the excess discontinuation disproportionally increased with larger BP reductions than the increase in outcome risk reduction. A standard SBP reduction was found associated with a constant relative reduction, but a smaller absolute reduction of cardiovascular events, and a greater relative excess of treatment discontinuations when the achieved SBP was below 130 mmHg rather than in the higher ranges. The authors conclude [1] that the burden of adverse events should not prevent patients from the overwhelming benefits of BP treatment. However, the burden of adverse events should be considered carefully whenever the extent of the BP lowering or the BP target to be achieved are discussed. The eighth meta-analysis [1] provides both interesting and important new information that call for the clinician's awareness of patient-related problems, not only during participation in clinical trials of drugs to prevent cardiovascular disease in hypertension but also during usual care. Hypertension is reported to impose the highest global burden to disease development of any risk factor [2], and drug treatment is the only evidence-based and effective treatment. Thus, as clearly pointed out by the authors of the meta-analysis [1], these summary data on adverse events and possibly side-effects should not in any aspect be used to exclude patients with true hypertension from documented drug treatment. However, detailed and careful reporting of the adverse events may potentially disclose type of side-effects, intensity and the frequency. Details from some trials can illustrate the particularities and the impact of discontinuation in trials and complement some explanations of the adverse events. The STOP 2 study [3], a classical study with active reporting of adverse events, was not included in the meta-analyses because of a comparative design. That study [3] was done in 6614 elderly people in Sweden with quite severe hypertension randomized to comparison of treatment with calcium antagonists, angiotensin converting enzyme inhibitors and conventional treatment with diuretics and beta-blockers. BPs were lowered from 194/98 mmHg to about 159/81 mmHg and almost identically in all three groups, which was not considered as a very aggressive treatment at that time. The most common adverse event was dizziness in about 27% of participants in all three treatment arms, suggesting overtreatment and indication of more 24-h BP measurements in the elderly [4]. Despite the background of noise caused by exposing participants for a prespecified list of potential side-effects and active questioning, an excess of 17% of participants on calcium antagonist reported ankle edema, and an excess of 25% of participants on angiotensin-converting enzyme inhibitors irritating cough. These adverse events and specific drug-related side-effects have been shown to be reduced by combination or modification of the two treatment principles [5]. Discontinuations have been high in some studies, and up to half of participants may have stopped treatment for various reasons. Although in the STOP-2 study no patient was lost to follow up or refused to continue the study [3], approximately 40–50% of participants discontinued for various reasons in the Medical Research Council studies in United Kingdom [6,7]. This points to another important aspect of discontinuation in randomized trials, and that is loss of statistical power. Without the necessary number of patients finishing trials, investigators cannot expect to properly test their hypothesis and thus trials may fail. Retention of patients may be particularly difficult in double-blinded trials in which it appears easy to blame the blinded drug, the blinded comparator or matching placebo for all kinds of problems that may occur during the course of the study. About 10% discontinuation is expected in the early phase of most randomized trials because of noise related to study start and randomization, and later it may be a continuous struggle to keep patients in the trial. For example in the LIFE study [8], a campaign was launched to keep patients in the study. The attached figure (Fig. 1) shows the discontinuation rate per 6 months toward the end of the trial; during the most intense campaigning in favor of patient retention, patients even entered into the double-blinded study again (‘negative’ discontinuation rate). The experience from such situations is that discontinuations frequently are unspecific and not necessarily caused by drug-specific side-effects.FIGURE 1: Discontinuation rate in the Losartan Intervention for Endpoint reduction (LIFE) Study toward the end of the study when study leadership was concerned regarding a high discontinuation rate and launched a campaign to lower it and retain participants in the trial. Percentage is given per 6 months. In early year 2000, it was ‘negative’, indicating that participants resumed the double-blinded treatment. Previously unpublished data.However, a recent study in which the problem with treatment-induced side-effects and discontinuations were prominent was the Systolic Blood Pressure Intervention Trial (SPRINT) [9]. Many participants suffered from electrolyte disturbances, reduced renal function, dizziness and syncope, most likely caused by the extensive use of diuretics in the intensive arm. These likely drug-specific side-effects were quite serious, likely explaining the hundreds of discontinuations, and the ‘number needed to harm’ was calculated to 22 vs. 16 for the ‘number needed to treat’ (for benefits favoring the lowest treatment arm) [10]. Further, as favor in preventing heart failure may have been driving the difference in the primary endpoint in favor of the low treatment target group (BP < 120 mmHg) in SPRINT [9], benefits may have been directly exchanged with the adverse events and discontinuations [10]. The intensity of the BP-lowering treatment in SPRINT [9] may, however, not have been the problem in as much as the BP measurement technique of unattended automatic office BP [11] suggests that 10–20 mmHg needs to be added to the lower treatment arm that would then be close to 140 mmHg if compared with BP target in regular practice. Whether these data can be applied in the daily care of the hypertensive patients is uncertain, as patients who participate in randomized trials will most likely receive better care and closer follow-up than what can be achieved in daily practice. Then, it is expected that the magnitude of discontinuation is greater among patients in usual care. The authors translate the results of the meta-analysis into clinical practice, reinforcing the concept that the potential for side-effect do not want to discourage treatment but rather call on the potential consequences of treatment-associated adverse events when optimal level of care is the goal. How much and how far BP should be lowered needs to be individualized to reduce risk of producing side-effects derived from excessive BP reduction or from mounting the number of drugs given. Unnecessary increments of therapy can result in treatment discontinuation with loss of the potential benefits of the previous treatment [12]. Intensity of treatment, number and dose of drugs, should consider patient characteristic and potential cardiovascular frailty, as well as the physician–patient relationship, to avoid treatment discontinuation. In summary, the new meta-analysis points to aspect in a study that has drawn limited attention, but which can provide important information on tolerability of the given treatment, selection of drugs, intensity of treatment and investigator care in general. Patients who participate in trials are considered to have better prognosis than nonparticipants. Notwithstanding, attention should be given to the discontinuation problems, which is the focus in the meta-analysis [1]. Whether these data can be applied in the daily care of the hypertensive patients and how much it contributes to the high rates of uncontrolled hypertension worldwide is uncertain. ACKNOWLEDGEMENTS Conflicts of interest S.