To the Editor:
An excellent editorial by González-Juanatey1 on the ALLHAT study was published recently. Whereas we agree with his opinions, some of his conclusions on aspects of the design of the ALLHAT, particularly the claim that diuretics are the initial antihypertensive treatment of choice, require closer examination.
Bradford Hill laid the scientific foundations for clinical trials2 and his definition of «a carefully and ethically designed experiment to answer a precise question» remains valid. Correct interpretation of a trial should be based on the question that it was designed to answer.
The hypothesis of the ALLHAT study3 is that the incidence of coronary heart disease would be lower in patients treated with drugs other than diuretics. After 4.9 years of follow-up, the incidence of coronary events was 11.5 per 100 patients for treatment during 6 years with chlorthalidone and 11.3 and 11.4 per 100 patients for treatment with amlodipine and lisinopril. No statistically significant differences were observed. A logical interpretation of this finding is to accept the null hypothesis and claim that the new drugs are not superior to diuretics, and indeed this is stated in the commentary section of the ALLHAT study. Surprisingly, the authors then conclude that thiazide diuretics should be the initial treatment in arterial hypertension and should be used, if possible, in association with antihypertensive agents.
If the aim is to test the superiority of one treatment over another, and the results show no difference, can the authors conclude that diuretics are preferable? The claim that diuretics are the initial treatment for hypertension is also incorrect because the number of patients receiving a diuretic as the initial treatment is unknown. Patients receiving antihypertensive treatment at the start of the study (enrolled with no washout from their normal treatment) comprised 90% of the study population, but we do not know what treatment they were receiving. Around 24% of the patients were changed to a different treatment group, and at the end of the trial almost 30% were not receiving the same drug they had been randomly assigned to at the start.
The defense of the preeminence of diuretics does not sufficiently emphasize that patients treated with chlorthalidone presented a significantly higher incidence of hypokalemia, hyperglycemia, hypercholesterolemia, increased creatinine or new diagnosis of diabetes. The authors' argument that this does not influence coronary events may be fallacious because the follow-up is too short for this influence to become apparent.
The results of the ALLHAT study suggest that lower blood pressure reduces the incidence of coronary events. But such a reduction is to be expected because we can assume that the incidence of such events would be greater in a group receiving no treatment than in one receiving treatment. We should point out that the incidence of coronary events in all groups is greater than expected.3
The ALLHAT study illustrates the need to combine drugs in order to control hypertension (40% of patients in the study received a combination of drugs). But the study does not demonstrate the advantages of some combinations over others because we do not know what combination the patients received and, moreover, the combined drugs were administered in an open-label manner. There are many possible combinations, and we cannot draw conclusions on the effects of one drug or another. The interpretation of the differences for the secondary objectives does not consider the complexity of treatment, and the comparison is based on the initial randomization group, which does not ensure that the patient received the corresponding drug.
The external validity of a trial depends on the similarity between the study population and the target population in which the findings are to be applied. The study population is at high cardiovascular risk and aged over 55 years, so whether the same effects would arise in a lower risk population is mere conjecture.
The results of the ALLHAT study, as Meltzer4 noted, provide more support for the JNC IV report (1988) than for later updates, particularly, the JNC VII report, which uses the ALLHAT study to justify the choice of diuretics as the initial treatment. The lower cost of diuretics may justify their use provided there are no contraindications or express indications for another drug, but the superiority of diuretics has not been proved by the ALLHAT study.
Recent guidelines have been issued jointly by the European societies of hypertension and cardiology.5 These documents criticize the interpretation of the ALLHAT study, are less restrictive regarding pharmacological treatment, and emphasize the importance of lowering blood pressure regardless of the drug used.