The authors of these 2 letters raise several issues regarding our meta-analysis of the pivotal trials for 3 cholinesterase inhibitors used in the treatment of Alzheimer's disease.1
Shane Kavanagh and Patricia Kabathova argue that the meta-analysis treats galantamine unfairly. In support of this claim, they point out that the majority of the galantamine trials included in the analysis did not use recommended doses. We explicitly recognized this and adjusted the meta-analysis of efficacy accordingly. They also suggest that meta-analyses focus on common treatment durations. However, there is no such thing as a “common treatment duration” in this context. In fact, the range of treatment duration is even greater than 12 weeks to 1 year, which Kavanagh and Kabathova refer to as “wide variation.” We chose to be explicit about the treatment durations that were included and handled the differences by means of a subanalysis, assessing heterogeneity and providing confidence intervals.
In addition, Kavanagh and Kabathova suggest that more weight be given to the ADAS-cog assessment than to global clinical judgement. In keeping with the approach of health regulatory agencies such as the US Food and Drug Administration,2 we provided the numbers needed to treat (NNT) and confidence intervals for both cognitive outcomes (the ADAS-cog scores) and global clinical impression. The emphasis on global outcomes was consistent with the stated goal of the analysis.
Kavanagh and Kabathova, Celio Levyman and other authors,3 all comment on the issue of comparability between the 3 drugs. However, in our meta-analysis we cautioned that individual cholinesterase inhibitors cannot be directly compared until well- designed, properly blinded, head-to-head trials are conducted. Although Levyman shares his own clinical experience of a slightly greater benefit with donepezil, Kavanagh and Kabathova cite new evidence supporting the similarity in safety profiles between galantamine and other cholinesterase inhibitors. The study they selected as an example of a head-to-head trial4 was 1 of 3 published to date and was not available when we prepared our meta-analysis. All 3 studies4,5,6 had relatively small numbers of subjects, and none was completely double-blinded. In each case, the results presented favoured the sponsor's drug. The similarities or differences among various cholinesterase inhibitors in terms of efficacy and safety remain to be established.
Levyman rightly points out that while the competing interests of the authors of the meta-analysis were fully disclosed, the primary randomized controlled trials used in the meta-analysis might themselves have been subject to bias. The bias we fear is the presentation of results unreasonably or unjustifiably favouring those of the sponsor or disfavouring those of a competitor. Although we cannot control for such bias, we did obtain and present results that were not distorted from what is found in clinical practice, that did not unduly favour the drug of any given sponsor and that did not unduly favour those drugs as a group or class. Bias also influences which trials are published, with negative trials less likely to be published.5 We addressed this potential problem by using funnel plots.
Meta-analysts and, more important, clinicians practising evidence-based medicine strongly rely on the data provided by high-quality, double-blind, randomized placebo-controlled trials. We thank Kavanagh, Kabathova and Levyman for reiterating the shortcomings in the literature to date, which led to the cautions described in our meta-analysis. Nevertheless, our conclusion, that cholinesterase inhibitors are effective as a class, and the quantification of NNT with the associated confidence intervals are supported by the data and our analysis.
Krista L. Lanctôt Departments of Psychiatry and Pharmacology Nathan Herrmann Department of Psychiatry Thomas R. Einarson Faculties of Pharmacy and Medicine University of Toronto Toronto, Ont.
Footnotes
-
Competing interests: Dr. Lanctôt has received research honoraria and/or speaker's fees and/or travel assistance from Pfizer Canada Inc., Janssen-Ortho Inc. and Neotherapeutics. Dr. Hermann has received speaker's honoraria and consulting fees from Pfizer Canada Inc., Janssen-Ortho Inc. and Novartis Pharmaceuticals, as well as research support from Janssen-Ortho Inc. None declared for Dr. Einarson.