Reply

Abstract

We could not agree more with the contention of Ram 1 Ram C.V.S. Risk of new-onset diabetes mellitus in patients with hypertension treated with beta blockers. Am J Cardiol. 2008; 102: 242-244 Abstract Full Text Full Text PDF PubMed Scopus (15) Google Scholar that “meta-analysis is not an exact statistical science that can provide definitive answers to a critical clinical question.” In fact, we were the first to point to the bouillabaisse analogy of meta-analysis 2 Messerli F.H. Are calcium antagonists safe?. Lancet. 1995; 346: 767-768 Abstract PubMed Google Scholar and also to draw attention to the fact that poorly done meta-analyses can be found even in good medical journals. 3 Opie L.H. Messerli F.H. Nifedipine and mortality Grave defects in the dossier. Circulation. 1995; 92: 1068-1073 Crossref PubMed Scopus (201) Google Scholar , 4 Messerli F.H. Case-control study, meta-analysis, and bouillabaisse: putting the calcium antagonist scare into context. Ann Intern Med. 1995; 123: 888-889 Crossref PubMed Scopus (52) Google Scholar Clearly, meta-analyses are instead hypothesis generating and in many circumstances (as pointed out by Ram) may be the way to recognize important adverse outcomes that may not be pointed out by a single study. Although the limitations of meta-analyses are well known, it is important to recognize their strengths as well. For many interventions, the improvements in clinical outcomes or the adverse effects from such interventions may be clinically significant yet statistically moderate. Very large trials may be needed to “prove” the difference. Trials of this size are slow and expensive to run, and many trials may be much too small to “prove” these differences. By combining studies from different centers or different countries enrolling diverse patient subgroups, meta-analyses can make generalizations to other populations more justifiable if effectiveness or adverse outcomes are demonstrable consistently across a range of studies. Moreover, bigger and more methodologically sound studies can be weighted to have more influence than smaller, poorer studies. Furthermore, meta-analyses allow an alternative approach to traditional, narrative reviews or statistically flawed quantitative approaches such as “vote-counting” reviews. Meta-analyses offer a number of important advantages, including the ability to calculate effect-size estimates (i.e., the overall magnitude of responses) and to discriminate statistically among the effects in different subsets of studies. A goal and an inherent part of the philosophy underlying meta-analysis is that it requires the same rigor in sampling and analysis as is required in primary research. Risk of New-Onset Diabetes Mellitus in Patients With Hypertension Treated With Beta BlockersAmerican Journal of CardiologyVol. 102Issue 2PreviewIn a recent meta-analysis, Bangalore et al1 assessed the association of the use of β blockers with risk for new-onset diabetes mellitus (DM). A total of 12 randomized, controlled studies compared the use of β blockers and other agents or placebos on clinical end points.2–13 Bangalore et al concluded that β-blocker therapy is associated with an increased risk for new-onset DM compared with nondiuretic antihypertensive drugs and also compared with placebo. The antihypertensive efficacy of β blockers was inferior to that of the other agents, and β blockers provided no benefit for the end point of death or myocardial infarction and were associated with a 15% increased risk for stroke compared with other agents. Full-Text PDF