2013 • Health Sciences
Relationship between Duration of Delirium and Short-term Mortality: A Systematic Review and Meta-Analysis of Randomized ICU Trials with Interventions Hypothesized to Reduce Delirium
Lead Presenter: Nada Al-Qadheeb
Introduction: æRegression analyses of single study cohorts suggest that increased delirium duration is associated with increased mortality. However, the relationship between delirium duration and short-term mortality in randomized controlled trials (RCTs) testing an intervention hypothesized to reduce delirium burden remains unclear. æHypothesis: æDelirium duration and hospital mortality are associated in RCTs evaluating intervention(s) suspected to reduce delirium burden. Methods: æWe searched 7 study databases to identify RCTs evaluating drug or non-drug interventions hypothesized to reduce delirium burden in critically ill adults where both delirium duration and survival status (mortality during hospitalization or æ21, 28 or 30 days after randomization) were collected. æResults: æ17 studies identified 5 different pharmacologic agents [dexmedetomidine, rivastigmine, haloperidol, quetiapine, ziprasidone], one multimodal treatment (SAT-SBT) and 1 non-pharmacologic treatment (early mobilization. The average reduction of delirium duration (vs. control) was ?3 days (n=4 studies), ?1 d (n=5), 0-1 d (n=5), -0.5 to -2 d (n=3). Across the 17 studies, the relationship between difference in delirium duration (i.e., intervention vs. control groups) and hospital mortality was not significant [OR=0.89, 95% CI (0.70,1.50)]. Excluding the 3 RCTs where the intervention increased delirium duration, the relationship between delirium duration and hospital mortality using a meta-regression technique was not significant (P=0.48). Conclusions: æAmong RCTs evaluating an intervention reducing ICU delirium burden, no significant relationship between the effect of the intervention on delirium duration and hospital mortality was found. Either no relationship exists,the relationship varies by intervention,or the effect is small and thus larger RCTs are needed to further investigate this relationship.