A Cochrane review [Abstract] 1 included 24 studies with a total of 11 959 subjects with dementia or cognitive impairment. The relevance of the diagnostic categories used by the investigators to modern concepts of dementia and cognitive impairment is uncertain. Global Impression of Change (GIC) was the only outcome for which pooling of data was possible, involving only four studies (n=430). The OR for improvement in the piracetam group compared with placebo was 3.43 (95% CI 2.32 to 5.07). Using a fixed-effects model the OR for improvement with piracetam compared with placebo was 3.55 (95% CI 2.45 to 5.16). This estimate was derived from completers rather than from an intention-to-treat analysis as relevant data could not be extracted from the reports. In the limited data available no significant differences were found between treatment and placebo groups for cognition (immediate memory, visuospatial, MMSE, delayed memory or speech) for dependency, or for depression. The large volume of unpublished and untraceable data not available to the reviewers raises the possibility of publication bias.
Comment: The quality of evidence is downgraded by study quality (inadequate allocation concealment, short follow-up time), inconsistency (heterogeneity in patients, outcomes and results across studies), indirectness (no standardized diagnostic criteria in the studies), imprecise results (limited study size for each comparison) and publication bias (considerable amounts of unpublished and untraceable data not available to the review).
Primary/Secondary Keywords