Are Calcium Antagonists Harmful in Hypertensive Patients?
1995; Elsevier BV; Volume: 108; Issue: 1 Linguagem: Inglês
10.1378/chest.108.1.3
ISSN1931-3543
Autores Tópico(s)Pharmaceutical Practices and Patient Outcomes
ResumoA recent report of an increased risk of myocardial infarction in hypertensive patients treated with calcium antagonists was presented at the 35th Annual Conference on Cardiovascular Disease Epidemiology and Prevention1Psaty BM Heckhert SR Koepsell TD et al.The risk of incident myocardial infarction associated with anti-hypertensive drug therapies.Reported at the 35th Annual Conference on Cardiovascular Disease Epidemiology and Prevention. March 8-11, 1995; (Hyatt Regency Hill Country Resort, San Antonio, Texas)Google Scholar on March 10th, 1995. It received unprecedented media and public attention. In this editorial, we intend to dispel some myths, clarify some misconceptions, and offer a frame of reference to render—in due time—a fair and unbiased judgement of this hotly debated medical controversy. This was a population-based case-control study designed to examine the association between different classes of antihypertensive drugs and risk of incident myocardial infarction. The report is the result of preliminary analyses of the computerized pharmacy records of the first 291 “cases” (with a prior myocardial infarction) and 1,240 “controls” (without prior myocardial infarction) matched for age, gender, and year of diagnosis. All subjects were treated hypertensive patients followed up at the Group Health Cooperative in Seattle, Washington. The major finding of this study is that, among the group of patients who used two antihypertensive drugs, calcium antagonists were associated with a significant almost two-fold increased risk of myocardial infarction (relative risk=1.95, 95% confidence interval 1.07 to 3.54) compared with those who received a diuretic alone. At first glance, it would seem reasonable to infer that calcium antagonists are harmful and should not be used as antihypertensive drugs. Indeed, this is the message that millions of hypertensive patients received from the media coverage of this report. This inference would be appropriate if and only if the association between calcium antagonists and myocardial infarction is valid and causal in nature. We believe that there were two possible sources of bias in this study: selection and confounding bias.2Kleibaum DG Kupper LL Morgenstern H Epidemiologic research: principles and quantitative methods.New York: Van Nostrand Reinhold. 1982; : 183-193Google Scholar The first source of bias may have been selection bias because of a potential systematic imbalance between cases and controls. For example, the inclusion of postmyocardial infarction patients as the “cases” for the study may have inadvertently included “sicker” patients who required more medicines, such as calcium antagonists. Alternatively, sicker patients may have been more willing to participate in a case-control study (“self-selection”). Thus, the use of the calcium antagonists may have been a “marker” for sicker patients, rather than the cause of the increased morbidity that was observed. Another source of bias is “confounding bias” as a result of unknown confounders, unmeasured confounders, or both. For example, although the authors matched “cases” and “controls” for age, gender, and year of diagnosis of hypertension, there may have been an occult confounding factor in the patients with myocardial infarction, such as congestive heart failure, that may have contributed to the observed association of calcium antagonists with myocardial infarction. Thus, confounding bias due to unknown factors, unmeasured factors, or both could potentially invalidate the interpretation of any case-control study. A common strategy to minimize confounding bias in a case-control study is to match “cases” and “controls” for important known confounders. In the University of Washington study, “cases” and “controls” were frequency matched for age, gender, and year of diagnosis of hypertension. A major limitation of matching in a case-control study is that it may cause greater bias if matching variables are not associated or are weakly associated with the risk of myocardial infarction such that a systematic imbalance between cases and controls is artificially introduced.3Kleibaum DG Kupper LL Morgenstern H Epidemiologic research: principles and quantitative methods.New York: Van Nostrand Reinhold. 1982; : 377-402Google Scholar Another limitation of matching is that it is no longer possible to make inferences about any interactions between calcium antagonists and any of these matching variables vis-a-vis the risk of myocardial infarction, because the cases and controls were artificially matched.3Kleibaum DG Kupper LL Morgenstern H Epidemiologic research: principles and quantitative methods.New York: Van Nostrand Reinhold. 1982; : 377-402Google Scholar Thus, in summary, all case control studies are potentially vulnerable to the introduction of selection bias, confounding bias, or both, which may invalidate the conclusion(s) of the study. These limitations of study design are much less likely a problem with a randomized controlled clinical trial. The basic difference between a randomized controlled trial and a case-control study is “randomization of treatment allocation.” That is, all subjects in a randomized controlled trial are equally likely to receive active treatment or control, thereby minimizing selection bias. For example, the problem of selection bias due to greater willingness of sicker patients to participate is not a problem in a randomized controlled trial, since volunteering to participate in the trial precedes the occurrence of the disease of interest, such as a myocardial infarction. Moreover, since known as well as unknown and unmeasured potential confounders are likely to be distributed equally among treated and control subjects, confounding bias is minimized. The randomized controlled trial is the gold standard advocated by the scientific community and by regulatory agencies such as the Food and Drug Administration to judge the efficacy and safety of drug therapy and to justify public health policy recommendations regarding drug therapy. Thus, the best way to address the question of the role of antihypertensive drugs and the risk of myocardial infarction is to perform a randomized controlled clinical trial, as opposed to a case-control study. Causality implies the judgement of a cause and effect relationship. A commonly used set of criteria that can aid in the judgement concerning causality, first reported by Hill,4Hill AB The environment and disease: association or causation?.Proc R Soc Med. 1965; 58: 295-300PubMed Google Scholar includes magnitude of the association, consistency, temporality (ie, the occurrence of the disease follows the exposure to treatment), dose-response relationship, specificity, and biologic plausibility. Since a case-control study has “backward directionality” (ie, patients are only included in the study after they have developed the disease of interest), it may not, for example, clearly address the temporal association between the use of a calcium antagonist and the development of a myocardial infarction. In the University of Washington study, we may ask: When did the index myocardial infarction occur in relation to the administration of calcium antagonists? Was exposure to calcium antagonists defined as a computer record of calcium antagonists before or after the index myocardial infarction? This is particularly important because the occurrence of myocardial infarction may affect the decision to use a calcium antagonist in a hypertensive patient. Also, two other elements of causality are generally difficult to assess in a case-control study, namely dose response relationship and specificity, because exposure to treatment is not randomized in a case-control study. It is also worth noting that a single study cannot—and should not—be used to establish causality of an observed association. A judgement concerning causality will be strengthened if different studies show a similar association. To date, there are no randomized controlled clinical trials of the effects of calcium antagonists on death or myocardial infarction in hypertensive patients. Nonetheless, previous meta-analyses of calcium antagonists in patients with acute myocardial infarction suggested an excess in mortality.5Held PH Yusuf S Furberg CD Calcium channel blockers in acute myocardial infarction and unstable angina: an overview.BMJ. 1989; 299: 1187-1192Crossref PubMed Scopus (387) Google Scholar, 6Yusuf S Furberg CD Effects of calcium channel blockers on survival after myocardial infarction.Cardiovasc Drugs Ther. 1987; 1: 343-344Crossref PubMed Scopus (24) Google Scholar, 7Teo KK Yusuf S Furberg CD Effects of prophylactic antiarrhythmic drug therapy in acute myocardial infarction.JAMA. 1993; 270: 1589-1595Crossref PubMed Scopus (565) Google Scholar However, it is not appropriate to extrapolate the unfavorable results of secondary prevention clinical trials of calcium antagonists in patients with myocardial infarction5Held PH Yusuf S Furberg CD Calcium channel blockers in acute myocardial infarction and unstable angina: an overview.BMJ. 1989; 299: 1187-1192Crossref PubMed Scopus (387) Google Scholar, 6Yusuf S Furberg CD Effects of calcium channel blockers on survival after myocardial infarction.Cardiovasc Drugs Ther. 1987; 1: 343-344Crossref PubMed Scopus (24) Google Scholar, 7Teo KK Yusuf S Furberg CD Effects of prophylactic antiarrhythmic drug therapy in acute myocardial infarction.JAMA. 1993; 270: 1589-1595Crossref PubMed Scopus (565) Google Scholar, 8Habib G Roberts R Calcium channel blockers in the treatment of acute myocardial infarction.In: Eric R. Bates, ed. Thrombolysis an adjunctive therapy for acute myocardial infarction. New York: Marcel Dekker. 1993; : 167-189Google Scholar to hypertensive patients for at least two reasons. First, these trials used the three short-acting prototype calcium antagonists, nifedipine, verapamil, and diltiazem, two of which (nifedipine and diltiazem) are not even approved by the Food and Drug Administration for the treatment of hypertension. Second, the hemodynamic effects of these drugs in ambulatory patients with uncomplicated hypertension may be quite different from their effects in hospitalized patients with acute myocardial infarction. Therefore, the available evidence regarding the association of calcium antagonists with risk of myocardial infarction in hypertensive patients is limited to a single case-control study. Because of this limitation, the University of Washington study does not establish clearly causality of the association between calcium antagonists and risk of myocardial infarction in hypertensive patients. Finally, there remains a number of important unanswered questions about the University of Washington study: Was there a systematic difference in relevant baseline characteristics between cases and controls or between selected study subjects and excluded study subjects? Was the association with increased risk of myocardial infarction consistently observed with all classes and types of calcium antagonists? How successful was matching of cases and controls for age, gender, and year of diagnosis? Which potential confounders were measured and adjusted for in the analysis? In conclusion, based on the issues raised in this editorial, we should await the scrutiny of the peer review process and the publication of the full manuscript from the University of Washington before we, as an academic community, cast judgement with regard to the efficacy and safety of calcium antagonists in hypertensive patients. We believe that calcium antagonists should not be considered as harmful until the results of ongoing prospective randomized controlled trials of antihypertensive drugs in hypertensive patients, such as ALLHAT,* are completed.
Referência(s)