Randomized ICU trials do not demonstrate an association between interventions that reduce delirium duration and short-term mortality: a systematic review and meta-analysis

Abstract

Objectives: We reviewed randomized trials of adult ICU patients of interventions hypothesized to reduce delirium burden to determine whether interventions that are more effective at reducing delirium duration are associated with a reduction in short-term mortality.

Data sources: We searched CINHAHL, EMBASE, MEDLINE, and the Cochrane databases from 2001 to 2012.

Study selection: Citations were screened for randomized trials that enrolled critically ill adults, evaluated delirium at least daily, compared a drug or nondrug intervention hypothesized to reduce delirium burden with standard care (or control), and reported delirium duration and/or short-term mortality (≤ 45 d).

Data extraction: In duplicate, we abstracted trial characteristics and results and evaluated quality using the Cochrane risk of bias tool. We performed random effects model meta-analyses and meta-regressions.

Data synthesis: We included 17 trials enrolling 2,849 patients which evaluated a pharmacologic intervention (n = 13) (dexmedetomidine [n = 6], an antipsychotic [n = 4], rivastigmine [n = 2], and clonidine [n = 1]), a multimodal intervention (n = 2) (spontaneous awakening [n = 2]), or a nonpharmacologic intervention (n = 2) (early mobilization [n = 1] and increased perfusion [n = 1]). Overall, average delirium duration was lower in the intervention groups (difference = -0.64 d; 95% CI, -1.15 to -0.13; p = 0.01) being reduced by more than or equal to 3 days in three studies, 0.1 to less than 3 days in six studies, 0 day in seven studies, and less than 0 day in one study. Across interventions, for 13 studies where short-term mortality was reported, short-term mortality was not reduced (risk ratio = 0.90; 95% CI, 0.76-1.06; p = 0.19). Across 13 studies that reported mortality, meta-regression revealed that delirium duration was not associated with reduced short-term mortality (p = 0.11).

Conclusions: A review of current evidence fails to support that ICU interventions that reduce delirium duration reduce short-term mortality. Larger controlled studies are needed to establish this relationship.

Figure 1. Literature flow

Abbreviations: ICU, intensive care unit.

Figure 2. Meta-analysis of difference in duration of delirium

The data markers indicate the difference in duration of delirium in days. The size of the grey boxes indicates the weight of each study in the meta-analysis. Abbreviations: DI, daily interruption; N, number analyzed; OT, occupational therapy; PS, protocolized sedation; PT, physiotherapy; SAT; spontaneous awakening trial; SBT; spontaneous breathing trial. * These studies reported on second control groups that were not included in this meta-analysis.

Figure 3. Meta-analysis of risk ratio for short-term mortality

The data markers indicate the risk ratio for mortality. The size of the grey boxes indicates the weight of each study in the meta-analysis. Abbreviations: DI, daily interruption; n/N, number of deaths/number analyzed; OT, occupational therapy; PS, protocolized sedation; PT, physiotherapy; RR, risk ratio; SAT, spontaneous awakening trial; SBT, spontaneous breathing trial. * This study reported on a second control group that was not included in this meta-analysis.

Figure 4. Scatter plot of difference in duration of delirium versus risk ratio for death

Each circle represents a unique study that reported both the difference between treatments in duration of delirium and the risk ratio (RR) for death. The size of the circles indicates the weight of each study in the random effects model meta-regression. The solid line describes the random effects model meta-regression association across all studies of difference in duration of delirium versus RR death (P=0.11). The dashed line describes the association excluding the study where the intervention resulted in a longer duration of delirium than the control (46)) (P=0.40). The equation for each metaregression line is provided for the logarithm of RR against delirium duration.