Phase III randomised double-blind controlled trial of oral risperidone, haloperidol or placebo with rescue subcutaneous midazolam for delirium management in palliative care

Aims: Guidelines recommend targeted use of antipsychotics in delirium for specific symptoms however this approach has not been evaluated in randomised trials. To compare the efficacy of risperidone relative to placebo in the control of specific delirium symptoms in palliative care patients (communica- tion, behaviour and/or perceptual disturbances on Nursing Delirium Screening Scale) at 72 hours after study commence- ment. Secondary aims were to compare haloperidol and placebo; and risperidone and haloperidol. Methods: Dose titration occurred twice daily to effect by predefined increments to maximum 4 mg (2 mg if \u3e65). All participants had delirium precipitants managed and non- pharmacological measures. Subcutaneous midazolam rescue was available. Improvement of delirium symptoms was assessed using linear regression (average of scores on day 4), adjusted for baseline score and group. Survival between groups was assessed using the log rank test, and midazolam use by Chi squared test. Results: The trial recruited to its full sample (239 partici- pants) - 80 risperidone; 79 haloperidol; and 80 placebo. For the primary intention-to-treat analysis (with 50 resamples drawn) between risperidone and placebo (n = 160) the risperidone group had significantly greater specific delirium symptoms on average at study end, 0.57 (95% CI 0.17, 0.98, p = 0.006) than the placebo group. The haloperidol group also had significantly greater specific delirium symptoms at study end than the placebo group, 0.29 (95% CI 0.11, 0.48, p = 0.002). In a pooled analysis, those on antipsychotics had a significant reduction in survival compared to placebo (p = 0.026). Midazolam rescue use was markedly lower in placebo group vs antipsychotics on each study day, 18.2 vs. 31.3% (day 1), p = 0.035, 15.9 vs. 29.0% (day 2), p = 0.031 and 13.6 vs. 29.6% (day 3), p = 0.016. Conclusions: This adequately powered study has shown indi- vidualised management of delirium precipitants and non- pharmacological strategies results in better control of delirium symptoms without the need for midazolam rescue and better survival, than seen with the addition of risperidone or haloperidol. These results fundamentally challenge the pharmacological approach to manage delirium