Meta-analysis clinical trials

Experimental and other study designs Clinical trials Meta-analyses... 

A recent meta-analysis of total abstinence as an outcome in clinical trials of acamprosate (Mann et al. 2004) included 17 studies and a total of more than 4,000 patients. The authors found a significant advantage for acampro-... 

Kranzler HR, Bauer LO, Hersh D, et al Carbamazepine treatment of cocaine dependence a placebo-controlled trial. Drug Alcohol Depend 38 203-211, 1995 Levin FR, Lehman AF Meta-analysis of desipramine an adjunct in the treatment of cocaine addiction. J Clin Pharmacol 11 374-378, 1991 Lima MS, Reisser AA, Soares BG, et al Antidepressants for cocaine dependence. Cochrane Database Syst Rev 4 CD002950, 2001 Ling W, Shoptaw S, Majewska D Baclofen as a cocaine anti-craving medication a preliminary clinical study 0etter). Neuropsychopharmacology 18 403 04, 1998... 

In patients with acute PE, the use of thrombolytics provides short-term benefits such as restoring pulmonary artery patency and hemodynamic stability.8,26 A recent meta-analysis of nine small randomized clinical trials showed a slightly lower risk... 

The benefit of carotid endarterectomy for prevention of recurrent stroke has been studied previously in major trials.25,26 A recent meta-analysis has been completed that has combined these clinical trials to evaluate 6,092 patients.27 Carotid endarterectomy has been shown to be beneficial for preventing ipsilateral stroke in patients with symptomatic carotid artery stenosis of 70% or greater and is recommended in these patients. In patients with symptomatic stenosis of 50% to 69%, a moderate reduction in risk is seen in clinical trials. In all patients with stenosis of 50% to 69% and a recent stroke, carotid endarterectomy is appropriate. In other patients, surgical risk factors and surgeon skill should be considered prior to surgery. The patient should have, at a minimum, a life expectancy of 5 years, and the surgical risk of stroke and/or death should be less than 6%. Carotid endarterectomy is not beneficial for symptomatic carotid stenosis less than 50% and should not be considered in these patients. 

The United States Pharmacopeial Convention, Inc. (USP) in 2000 issued the USP criteria for levels of evidence for botanical articles [117]. While issued for botanicals, the criteria have application to all therapeutic agents. The USP criteria rank evidence from I to IV, with Level I being the strongest. Within Level I, the randomized controlled clinical trial is ranked highest, followed by meta-analysis and epidemiological studies. Level II consists of the same designs, but with methodological flaws. Level III includes inconclusive studies, and Level IV is anecdotal evidence. 

P Juni, A Witschi, R Bloch, M Egger. The hazards of scoring the quality of clinical trials for meta-analysis. JAMA 282 1054-1060, 1999. 

In the decade that has passed since our article was published, the dust has settled around the issue of meta-analysis. It is no longer considered a controversial procedure. Meta-analyses of clinical trials are now routinely published in all of the top medical journals, and the National Institute for Health and Clinical Excellence (NICE), which publishes the treatment guidelines that are used by the NHS, crafts recommendations on the basis of meta-analyses that it conducts. Nevertheless, the editors were right about our article being controversial. Although some scholars in the field were persuaded by our analyses, others were sceptical, to put it mildly.2 The sceptics knew that antidepressants worked - if we had found otherwise, we must have done something wrong. Certainly there were other clinical trials of antidepressants beyond those that we had included in our analyses. Surely an analysis of those studies would point to a different conclusion. 

There were indeed clinical trials of antidepressants that we had not included in our meta-analysis, and there was also a meta-analysis of those other trials that had used some of the same methods we had used. It showed the same results that we had reported. The difference between drug and placebo in published trials of antidepressants was modest at best.3 Still, the controversy continued. 

In our meta-analysis, more than half of the clinical trials submitted to the FDA showed no difference between drug and placebo. Most reviewers of the clinical-trials literature have not had access to unpublished studies and may not even know of their existence. But the FDA and other regulatory agencies around the world knew of these data. Nevertheless, their existence is not even mentioned in the product labels, information leaflets and official Summaries of Product Characteristics (SPC) of most antidepressants. 

The most common criticism of our meta-analysis is the claim that the clinical trials we analysed were flawed, and that better results would have been found if the studies had been designed better. The trials were too short to show the real effect of antidepressants, the critics said. The people recruited to participate in them were not depressed enough, or they were too depressed. In any case, they were not representative of the patients who are generally seen in clinical practice. 

These continuation trials tell a very different story from that told by relapse-prevention trials. They show that there is little difference between antidepressant and placebo even when the clinical trial is extended over a longer period of time. Across the eight continuation trials that have been published, 79 per cent of patients on placebo and 93 per cent of patients on active medication remained well throughout the treatment period. In these long-term studies, placebo treatment was 95 per cent as effective as drug treatment. The authors of a meta-analysis of these trials concluded that the widely held - and probably erroneous - belief that the placebo response in depression is short-lived appears to be based largely on intuition and perhaps wishful thinking .17... 

Many patients are excluded from clinical trials. Critics of our meta-analysis have suggested that antidepressants might work better for these patients than they do for those who are studied in clinical trials. Let us see how plausible this concern is. 

There seems to be considerable reluctance in some parts of the medical community to acknowledge the benefits of exercise in the treatment of depression. One meta-analysis of clinical trials showed that physical exercise was as effective as psychotherapy or antidepressant medication and much better than no treatment. But the authors concluded that the effectiveness of exercise in reducing symptoms of depression cannot be determined ,45 and the editors of the journal introduced the article with an editorial comment entitled effectiveness of exercise in managing depression is not shown by meta-analysis .46 Why not Because there were flaws in the way many of the studies had been designed. To be fair, there were indeed shortcomings in the studies, but these shortcomings also characterize clinical trials of antidepressants.47 If clinical trials like these do not establish the effectiveness of physical exercise as a treatment for depression, neither do they establish the effectiveness of antidepressants. 

Editorial, A Double-Edged Sword , 2008.1 should also note that Nature, which is the primary journal of the Nature Publishing Group, responded to our meta-analysis with an excellent editorial on 6 March 2008. Citing the difficulties we had in obtaining access to complete data, they advocated a mandatory database that would provide access to the results of all trials clinical trials that are undertaken, not just those that are published. 

Currently, there are a number of systemic and intestine-selective MTP inhibitors, including lomitapide (23, BMS-201038, AEGR-733), implitapide (24), JTT-130, SLx-4090, and R-256918 (latter three structures not disclosed) believed to be in active development [60]. In a meta-analysis of three Phase II clinical trials, lomitapide as monotherapy or in combination with ezetimibe, atorvastatin, or fenofibrate significantly reduced LDL cholesterol (up to 35% as monotherapy and 66% in combination with atorvastatin) and was well tolerated with less than 2% discontinuation due to abnormal liver function [61]. Lomitapide has also been granted orphan drug status for the treatment of homozygous familial hypercholesterolemia [59]. Results of a Phase II study of JTT-130 for type 2 diabetes are expected in August 2010 [59,60]. 

Gerbarg, Z.B. and Horwitz, R.I. (1988). Resolving conflicting clinical trials guidelines for meta-analysis. J. Clin. Epidem. 41 503-509. 

Linde, K. et al.. Are the clinical effects of homeopathy placebo effects A meta-analysis of placebo-controlled trials. Lancet, 350, 834,1997. 

Kerb R, Brockmoiier J, Staffeldt B, Ploch M, Roots I. (1996). Single-dose and steady-state pharmacokinetics of hypericin and pseudohypericin. Antimicrob Agents Chemother. 40(9) 2087-93. Kim HL, Streltzer J, Goebert D. (1999). St. John s wort for depression a meta-analysis of well-defined clinical trials. J Nerv Ment Dis. 187(9) 532-38. 

Lickey ME, Gordon B. (1991). Medicine and Mental Illness. New York W. H. Freeman and Company. Linde K, Ramirez G, Mulrow CD, Pauls A, Weidenhammer W, Melchart D. (1996). St. John s wort for depression—an overview and meta-analysis of randomised clinical trials. BMJ. 313(7052) 253-58. Lingjaerde 0, Foreland AR, Magnusson A. (1999). Can winter depression be prevented by Ginkgo biloba extract A placebo-controlled trial. Acta Psychlatr Scand. 100(1) 62-66. 

Kim HL, Streltzer 3, Goebert D. (1999). St. John s wort for depression a meta-analysis of well-defined clinical trials. J Nerv Merit Dis. 187(9) 532-38. 

---