Heterogeneity of Treatment Response to Citalopram for Patients With Alzheimer's Disease With Aggression or Agitation: The CitAD Randomized Clinical Trial

Abstract

Objective: Pharmacological treatments for agitation and aggression in patients with Alzheimer's disease have shown limited efficacy. The authors assessed the heterogeneity of response to citalopram in the Citalopram for Agitation in Alzheimer Disease (CitAD) study to identify individuals who may be helped or harmed.

Method: In this double-blind parallel-group multicenter trial of 186 patients with Alzheimer's disease and clinically significant agitation, participants were randomly assigned to receive citalopram or placebo for 9 weeks, with the dosage titrated to 30 mg/day over the first 3 weeks. Five planned potential predictors of treatment outcome were assessed, along with six additional predictors. The authors then used a two-stage multivariate method to select the most likely predictors; grouped participants into 10 subgroups by their index scores; and estimated the citalopram treatment effect for each.

Results: Five covariates were likely predictors, and treatment effect was heterogeneous across the subgroups. Patients for whom citalopram was more effective were more likely to be outpatients, have the least cognitive impairment, have moderate agitation, and be within the middle age range (76-82 years). Patients for whom placebo was more effective were more likely to be in long-term care, have more severe cognitive impairment, have more severe agitation, and be treated with lorazepam.

Conclusions: Considering several covariates together allowed the identification of responders. Those with moderate agitation and with lower levels of cognitive impairment were more likely to benefit from citalopram, and those with more severe agitation and greater cognitive impairment were at greater risk for adverse responses. Considering the dosages used and the association of citalopram with cardiac QT prolongation, use of this agent to treat agitation may be limited to a subgroup of people with dementia.

Figure 1.

Alzheimer’s Disease Cooperative Study- Clinical Global Impression of Change (CGIC) response for pre-specified and post hoc subsets The first five were pre-specified and the next six were post hoc (see accompanying file).

Figure 2.. The average difference in CGIC response probability and its 95% confidence interval between citalopram and placebo

(95% CI after correcting for multiple comparisons “looks”) For more than 62% of the patients the 95% confidence interval was larger than the average 0.136 (seen at 100% on the x axis). This can be interpreted clinically by viewing the distribution of covariates for these patients (see Table 2) and comparing this to the distribution of all patients. For example, for a given value on the x axis, e.g., 86%ile, the corresponding value on the y axis, y=0.25, is the difference between the percent response under citalopram minus percent response under placebo (estimated nonparametrically) for the patients with the index scores below the 86%ile.

Figure 3.. Box plots of the index score deciles (CGIC).

The figure shows that 2 groups, deciles 9 and 10, show particularly strong effects favoring citalopram response compared to placebo, and one group, decile 1, shows a strong effect favoring placebo as more effective than citalopram. Most groups show essentially trivial effects on average for citalopram response.