- Top of page
Παντɛςανθρωποιτοv ɛιδɛναιορɛγονταιφvσɛι. σημɛιονδɛητωναισθησɛωναγαπησις. Kαιγαρχωριςτηςχρɛιαςαγαπωνταιδι' αvτας, καιμαλλιστατωναλλωνηδιατωνομματων.
By nature humans thirst for knowledge. A sign of this is our love of the senses. Because irrespective of our needs we love our senses in themselves and most of all vision (eyesight).1
These first sentences of what we now call Aristotle's “Metaphysics,” emphasize the appetite for knowledge that drives us to classify and to identify patterns in the confusing multitude of facts in nature. This propensity and ability has served us well in evolutionary terms, but in summarizing large amounts of data, important details may be lost. In this issue of The Journal of Clinical Hypertension, Wang and Staessen2 provide another quantitative overview of the findings of controlled trials of antihypertensive therapy. They studied the relationship of the odds ratios of benefit with respect to clinical outcomes to the difference in systolic blood pressure (BP) change between the two interventions being compared. They observed a linear relationship between the odds ratio and the difference in systolic BP change for total mortality and a curvilinear relationship for all cardiovascular events. By calculating 95% confidence intervals around these relationships, they report that the benefit of the antihypertensive agents in the majority of studies fell within the range predicted from the BP difference.
It would be surprising not to find a relationship between BP lowering and benefit of antihypertensive therapy since we try to lower BP in order to decrease the risk imposed by the relationship between BP and risk. On the other hand, extrapolating to the implication that there are no significant differences among drugs in their effect of decreasing events is not warranted for the reasons discussed below.
Meta-analyses in general are carried out after investigators unavoidably make many subjective decisions about inclusion or exclusion of trials, evaluation of their quality, specific outcomes to consider, and analytic methods used. While these decisions greatly affect the outcome of the meta-analyses, they are subject to bias. This explains the different conclusions of the two meta-analyses published in December 2000.3,4 One reported 26% more myocardial infarctions and 25% more cases of heart failure with the use of calcium channel blockers. The other, published in the same issue of The Lancet, concluded that the evidence to evaluate differences among drug classes in their effect on decreasing risk of fatal and nonfatal events was weak. Wang and Staessen2 do not give objective reasons for excluding the Appropriate Blood Pressure Control in Diabetes (ABCD) trial, the Fosinopril versus Amlodipine Cardiovascular Events Randomized Trial (FACET), the African American Study of Kidney Disease and Hypertension (AASK), Studies on Left Ventricular Dysfunction (SOLVD) prevention trial (and other angiotensin-converting enzyme (ACE) inhibitor trials) or for lumping together small trials. Also, it is not clear whether the number of subjects or the amount of information in each study is taken into account in the analysis,5–8 thus giving equal weight to small and large trials.
As expected, the authors2 find a relationship between BP lowering and decrease in risk by assuming that all trials and drug classes are comparable. Differences among drug classes are not examined in a formal fashion and placebo-controlled trials are lumped with trials comparing active agents. The confidence intervals are wide. On the left side of Staessen and Wong's Figure 1, nearly identical BP differences could be roughly associated with a 25% increase or 25% decrease in risk. On the right side, the same BP difference can be associated with risk reductions roughly ranging from 35%–70%, a two-fold change in benefit. For these reasons, the implication that there are no significant differences among drugs is not justified. Also, the difference in the shape of the curves between mortality and total cardiovascular events is difficult to understand on pathophysiologic grounds. Does a marked change in systolic BP increase the number of events but decrease case fatality? Measuring BP on therapy was probably not possible in some patients with fatal end points. This and the use of group rather than individual data weaken the analysis. The fact that more than 50% of patients need more than one medication for BP control is not considered; also, subgroup effects (e.g., patients with diabetes) are not described.
Clinical trials and meta-analyses are only tests of hypotheses. As such, they may give true-positive, true-negative, false-positive, and false-negative results to be interpreted after consideration of their quality and their statistical significance, including statistical power, but also after evaluation of the totality of knowledge from pathophysiology, biochemistry, biology, epidemiology, and other disciplines. In this respect, all prior evidence supports the wealth of clinical trial evidence showing that lowering BP decreases events. However, there is also a significant body of basic science, pathophysiologic, clinical research, and pharmacologic evidence supporting the use of ACE inhibitors (and possibly angiotensin receptor blockers) in preference to calcium channel blockers in hypertensive patients with diabetes, heart failure, or systolic left ventricular dysfunction. This has been recognized by guidelines promulgated by the sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC VI)9 and the World Health Organization-International Society of Hypertension (WHO/ISH),10 as well as by more recent expert opinion.11,12
Considering the second sentence of “Metaphysics,” where Aristotle emphasizes vision, we could look directly at the data of several comparative trials where heart failure and myocardial infarction were decreased more by ACE inhibitors than calcium channel blockers in spite of nearly identical BP control.5,6,13 Recently released results from the Losartan Intervention for Endpoint Reduction in Hypertension (LIFE) trial14 showed superior outcomes of losartan combined with atenolol in patients with hypertension and left ventricular hypertrophy. It is clear that BP control is essential. However, to decrease the global risk imposed by hypertension and other risk factors, we need to use ACE inhibitors (and maybe angiotensin receptor blockers) in some patient subsets (e.g., diabetes, cardiovascular disease, heart failure, coronary heart disease), to use more than one agent when BP is not controlled with monotherapy, and to prescribe additional therapy with acetylsalicylic acid (aspirin), statins, or antihyperglycemic drugs as needed.