Systematic review and meta-analysis of antipsychotic discontinuation in dementia

Antipsychotics are used to treat behavioral and psychological symptoms of dementia (BPSD). However, treatment is associated with adverse outcomes. A 2018 Cochrane review found low-quality evidence that discontinuation may have little effect on BPSD by primarily comparing the difference in number of study non-completers between treatment groups and did not report a pooled effect size for relapse risk. We performed a meta-analysis of the pooled risk ratio (RR) of relapse of symptoms following antipsychotic medication discontinuation in people with dementia. We hypothesized that trial design (enriched responder samples vs long-term user withdrawal), treatment duration, length of follow-up, withdrawal speed, and sex of participants would affect relapse risk.

We systematically searched PsycINFO, EMBASE, PubMed, and ClinicalTrials.gov databases for studies published between January 2018 and June 2024. The eight papers identified in the 2018 Cochrane review were also included in the meta-analysis. The study was registered with PROSPERO (CRD42024570329).

A total of nine studies were included, providing 682 observations with 135 events. Relapse definitions included participant removal due to worsening behavior or starting regular prescription of antipsychotic/rescue medication. A random-effects model found a pooled RR of relapse of 1.52 (95% CI: 1.18 to 1.95, p = 0.005) with stopping antipsychotics. Meta-regressions showed no effect of trial type, withdrawal speed, length of follow-up, or participant sex on RR.

While it is important to prescribe cautiously and to reduce and withdraw prescriptions in those who are considered to no longer need antipsychotics, there is a subgroup of patients for whom continuing medication is important. It was particularly surprising that the relative risk of relapse was unaffected by trial design, with similar effect sizes in both trials aiming to reduce prescription and prove efficacy. Further research is needed to identify factors associated with successful antipsychotic withdrawal.

An updated meta-analysis of trials of withdrawal of antipsychotics in dementia.Relative risk of symptom relapse increased when stopping antipsychotics.Relapse risk was not affected by trial design or prior length of treatment.Relapse risk was not affected by length of follow-up, withdrawal speed, or sex.