Background: Studies assessing the effect of suvorexant on delirium prevention included patients treated before development of delirium, which can introduce immortal time bias. The objective of the present study was to evaluate the effect of suvorexant on delirium, comparing patients treated before the onset of delirium with patients treated within 72h of admission using the same dataset.
Methods: Data from adult patients admitted to the ICU from August 2018 to July 2021 were retrospectively analyzed. In "any time before" analysis, the incidence of delirium was compared for patients who received suvorexant at any time during their ICU stay (suvorexant) (unless delirium developed before treatment) with patients who either did not receive suvorexant or received suvorexant after development of delirium (control). This design was used in previously published studies. In "within 72h" analysis, the incidence of delirium was compared for patients who received suvorexant within 72 hours of admission (suvorexant) and patients who did not receive suvorexant or received it more than 72 hours after admission (control). Patients who developed delirium during the initial 72 hours were excluded from "within 72h" analysis (N = 799).
Results: "Within 72h" analysis included 1,255 patients, and "any time before" analysis included 2,054 patients (of 6599 admissions). The unadjusted hazard ratio of "any time before" analysis was 0.16 and the 95% confidence interval was 0.13-0.21 (p<0.01). The adjusted hazard ratio was 0.21, and the 95% confidence interval was 0.16-0.27 (p<0.01). "Within 72h" analysis had an unadjusted hazard ratio of 0.54 and the 95% confidence interval was 0.36-0.82 (p<0.01). However, this association lost statistical significance after adjustment for potential confounders (adjusted hazard ratio 1.02, 95% confidence interval 0.65-1.59, p = 0.93).
Conclusion: Reducing the effect of immortal time bias led to a significantly reduced effect of suvorexant for the prevention of delirium.
Copyright: © 2022 Shiotsuka et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.