Background:
Behavioral and psychological symptoms in dementia (BPSD) presents a clinical challenge commonly requiring drug therapy. Primary care providers will likely employ antipsychotics or anxiolytics. Yet, additional classes of agents have been studied for this problem. Because BPSD is associated with earlier nursing home referral, worse prognosis, greater costs and increased caregiver burden, the comparative efficacy of these drug classes is an important clinical question.
Objective: The purpose of this meta-analysis was to analyze data from previously published trials to compare the efficacy of several drug classes used for treating BPSD.
Methods: Using Medline, EMBASE, Psych-info, LILACS, Cochrane Controlled Trials Register, DARE-Database of Abstracts of Effectiveness, and Dissertation Abstracts, all relevant articles published between 1990-2002 were identified. Randomized controlled trials evaluating a psychotherapeutic agent for the treatment of non-cognitive dementia behavior problems were sought. After abstracts and titles were screened, two independent reviewers selected and obtained those articles appropriate for further consideration. Efficacy data, patient, and study-level characteristics were extracted from studies meeting inclusion criteria. Conventional meta-analyses were performed within each drug class providing summary estimates of effect. Cumulative meta-analysis by publication date was also performed using all drug-vs-placebo trials.
Results: 33/52 trials reported data sufficiently for analysis. 26 drug-vs-placebo trials represented 6 drug classes. 4 drug-vs-haloperidol studies included 3 drug classes. The studies were of medium methodological quality, although inter-rater agreement varied widely from 0-50%. Drug was statistically favored over placebo for 4 classes: Ê atypical antipsychotics (SMD = -0.30, 95% CI –0.46, -0.14), Ë anticonvulsants ( SMD = -0.23, 95% CI –0.44, -0.03 ) and Ì cholinesterase inhibitors ( SMD = -0.13, 95% CI –0.21, -0.04 ). Sub-analysis based upon use of the same behavior rating scale was possible for 4/5 anticonvulsant and 6/8 cholinesterase inhibitor trials. Both sub-analyses yielded results comparable to the analyses of the groups as a whole (5/5, 8/8). Haloperidol-vs-placebo analysis statistically favored Í Haloperidol ( SMD = -0.26, 95% CI –0.43, -0.10). Haloperidol-vs-anti-anxiety drugs (oxazepam, buspirone) favored haloperidol, but the results were not statistically significant. Placebo was favored over the NMDA antagonist, Memantine, the MAO inhibitor, Selegiline, and the SSRIs. Comparing the effect size estimates of the 4 drug classes favored over placebo revealed that no statistically significant difference exists between them. Cumulative meta-analysis confirmed a persistently small, but statistically significant effect size for all assembled BPSD pharmacotherapy trials since 1991 (SMD = -0.18, 95% CI –0.31, -0.04 ). Finally, funnel plot of all 26 drug-vs-placebo trials revealed no asymmetry suggesting publication bias.
Conclusions: This meta-analysis provides an updated perspective on the last decade of research into the efficacy of pharmacologic treatment of BPSD. The small effect sizes favoring atypical antipsychotics, haloperidol, anticonvulsants, and cholinesterase inhibitors over placebo are consistent with previous meta-analyses. No statistically significant difference in efficacy exists between these 4 drug classes.
Back to S045 From Pharmacotherapy to Phenomenology: New Reports and New Contributions
Back to The Eleventh International Congress