Haloperidol for Hypoactive and Hyperactive Delirium in ICU Patients—A Post Hoc Bayesian Analysis of the AID‐ICU Trial

Background: Delirium is clinically classified into hypoactive or hyperactive motoric subtypes, which may have different responses to treatment. We assessed heterogeneity in treatment effects according to baseline motoric subtype in the AID-ICU trial, which randomised adult ICU patients with delirium to haloperidol versus placebo. Methods: In this post hoc study of the AID-ICU trial, we used adjusted Bayesian linear and logistic regression models to assess potential differential effects of haloperidol on 90-day outcomes according to motor subtypes. Effect estimates are reported as absolute differences within and between subtypes (mean differences (MD)/risk differences (RD) and differences in MDs/RDs), with probabilities of any benefit, any harm and differential effects between subtypes. Results: Of 987 participants, 540 (54.7%) had hypoactive and 447 (45.3%) had hyperactive delirium at randomisation. Haloperidol was associated with more days alive out of hospital in both subgroups, with a 76.5% probability of a larger intervention effect in those with hyperactive delirium (difference in MDs 2.9 days; 95% CrI -4.8 to 10.6). Similar results were obtained for days alive without delirium/coma and days alive without mechanical ventilation. Haloperidol was associated with lower mortality in both subgroups, with a 73% probability of larger benefit in hyperactive delirium (difference in RDs -3.7%-points; 95% CrI -15.3 to 7.8). We found no relevant differences in serious adverse reactions or use of rescue medications. Conclusion: Compared with placebo, haloperidol was associated with moderate to high probabilities of benefit in both hypoactive and hyperactive delirium, with moderate probabilities of greater effects in hyperactive delirium. Further research is warranted to confirm these results. Editorial comment: In this post hoc analysis of the AID-ICU trial, haloperidol was associated with moderate to high probabilities of benefit in both hypoactive and hyperactive delirium, with a tendency towards greater treatment effects in hyperactive delirium. Trial registration: ClinicalTrials.gov number: NCT03392376; EudraCT number, 2017-003829-15