Comparative Outcomes of Commonly Used Off-Label Atypical Antipsychotics in the Treatment of Dementia-Related Psychosis: A Network Meta-analysis

Abstract

Introduction: Dementia-related psychosis (DRP) is characterized by hallucinations and delusions, which may increase the debilitating effects of underlying dementia. This network meta-analysis (NMA) evaluated the comparative efficacy, safety, and acceptability of atypical antipsychotics (AAPs) commonly used off label to treat DRP.

Methods: We included 22 eligible studies from a systematic literature review of AAPs (quetiapine, risperidone, olanzapine, aripiprazole, and brexpiprazole) used off label to treat DRP. Study outcomes were: (1) efficacy-neuropsychiatric inventory-nursing home (NPI-NH psychosis subscale), (2) safety-mortality, cerebrovascular events (CVAEs), and others (somnolence, falls, fractures, injuries, etc.), and (3) acceptability-discontinuations due to all causes, lack of efficacy, and adverse events (AEs). We used random-effects modeling to estimate pooled standardized mean differences (SMDs) for NPI-NH psychosis subscale scores and odds ratios (OR) for other dichotomous outcomes, with their respective 95% confidence intervals (CIs).

Results: Compared with placebo, aripiprazole (SMD - 0.12; 95% CI - 0.31, 0.06), and olanzapine (SMD - 0.17; 95% CI - 0.04; 0.02) demonstrated small, non-significant numerical improvements in NPI-NH psychosis scores (5 studies; n = 1891), while quetiapine (SMD 0.04; 95% CI - 0.23, 0.32) did not improve symptoms. The odds of mortality (15 studies, n = 4989) were higher for aripiprazole (OR 1.58; 95% CI 0.62, 4.04), brexpiprazole (OR 2.22; 95% CI 0.30, 16.56), olanzapine (OR 2.21; 95% CI 0.84, 5.85), quetiapine (OR 1.68; 95% CI 0.70, 4.03), and risperidone (OR 1.63; 95% CI 0.93, 2.85) than for placebo. Risperidone (OR 3.68; 95% CI 1.68, 8.95) and olanzapine (OR 4.47; 95% CI 1.36, 14.69) demonstrated significantly greater odds of CVAEs compared to placebo. Compared with placebo, odds of all-cause discontinuation were significantly lower for aripiprazole (OR 0.71; 95% CI 0.51, 0.98; 20 studies; 5744 patients) and higher for other AAPs. Aripiprazole (OR 0.5; 95% CI 0.31, 0.82) and olanzapine (OR 0.48; 95% CI 0.31, 0.74) had significantly lower odds of discontinuation due to lack of efficacy (OR 12 studies; n = 4382) compared to placebo, while results for quetiapine and risperidone were not significant. Compared with placebo, the odds of discontinuation due to AEs (19 studies, n = 5445) were higher for olanzapine (OR 2.62; 95% CI 1.75, 3.92), brexpiprazole (OR 1.80; 95% CI 0.80, 4.07), quetiapine (OR 1.25; 95% CI 0.82, 1.91), aripiprazole (OR 1.38; 95% CI 0.90, 2.13), and risperidone (OR 1.41; 95% CI 1.02, 1.94).

Conclusions: Overall results demonstrate that, compared with placebo, quetiapine is not associated with improvement in psychosis in patients with dementia, while olanzapine and aripiprazole have non-significant small numerical improvements. These off-label AAPs (quetiapine, risperidone, olanzapine, aripiprazole, and brexpiprazole) are associated with greater odds of mortality, CVAEs, and discontinuations due to AEs than placebo. These results underscore the ongoing unmet need for newer pharmacological options with a more favorable benefit-risk profile for the treatment of DRP.

Keywords: Acceptability; Atypical antipsychotics; Delusions; Dementia-related psychosis; Efficacy; Hallucination; Network meta-analysis; Safety.

Fig. 1

PRISMA study selection flowchart

Fig. 2

Forest plot for efficacy outcome of atypical antipsychotics in comparison with placebo. Forest plot for NPI-NH psychosis subscale, representing pooled standard mean differences comparing each treatment option with placebo. The diamonds denote the point estimate for each treatment compared to placebo, while the horizontal line denote the 95% CI. If the 95% CI does not cross 0 (null value), the results imply statistical significance

Fig. 3

Forest plots for primary safety outcomes of a typical antipsychotics compared with placebo. a Forest plot for mortality, representing pooled odds ratio comparing each treatment option with placebo. The diamonds denote the point estimate for each treatment compared to placebo, while the horizontal line denoted the 95% CI. If the 95% CI does not cross 1 (null value); the results imply statistical significance. b Forest plot for CVAE, representing pooled odds ratio comparing each treatment option with placebo. The diamonds denote the point estimate for each treatment compared to placebo, while the horizontal line denoted the 95% CI. If the 95% CI does not cross 1 (null value); the results imply statistical significance. c Forest plot for somnolence, representing pooled odds ratio comparing each treatment option with placebo. The diamonds denote the point estimate for each treatment compared to placebo, while the horizontal line denoted the 95% CI. If the 95% CI does not cross 1 (null value); the results imply statistical significance. d Forest plot for falls, fractures, and injuries, representing pooled odds ratio comparing each treatment option with placebo. The diamonds denote the point estimate for each treatment compared to placebo, while the horizontal line denoted the 95% CI. If the 95% CI does not cross 1 (null value); the results imply statistical significance

Fig. 4

Forest plots for acceptability outcomes of a typical antipsychotics compared with placebo. a Forest plot for all cause discontinuation, representing pooled odds ratio comparing each treatment option with placebo. The diamonds denote the point estimate for each treatment compared to placebo, while the horizontal line denoted the 95% CI. If the 95% CI does not cross 1 (null value); the results imply statistical significance. b Forest plot for discontinuation due to lack of efficacy, representing pooled odds ratio comparing each treatment option with placebo. The diamonds denote the point estimate for each treatment compared to placebo, while the horizontal line denoted the 95% CI. If the 95% CI does not cross 1 (null value); the results imply statistical significance. c Forest plot for discontinuation due to adverse events, representing pooled odds ratio comparing each treatment option with placebo. The diamonds denote the point estimate for each treatment compared to placebo, while the horizontal line denoted the 95% CI. If the 95% CI does not cross 1 (null value); the results imply statistical significance