Letters Section Editor: Robert M. Golub, MD, Senior Editor.
To the Editor: In the meta-analysis by Dr Schneider and colleagues,1 the authors describe in detail the search strategy, trials selection, and data retrieval, but there are several potential weaknesses in their methods.
Although MEDLINE and the Cochrane Controlled Trials Register databases were searched, no mention is made of EMBASE. This database contains fewer references than MEDLINE, but comprehensive searches require using both MEDLINE and EMBASE, as the overlap between the 2 databases varies between 10% to 75%, depending on the topic searched.2 Additionally, the terms selected for searching were dementia, Alzheimer disease, and clinical trial. No reference is made to truncation for possible spelling varieties. A search for Alzheimer, truncated in Ovid MEDLINE, reveals that the difference between truncation (Alzheimer$) and nontruncation (Alzheimer) amounts to almost 9000 references.
Hammerstrøm K. Atypical Antipsychotic Drugs, Dementia, and Risk of Death. JAMA. 2006;295(5):495-497. doi:10.1001/jama.295.5.496-a