Olsen HT et al. Nonsedation or light sedation in critically ill, mechanically ventilated patients. NEJM. DOI: 10.1056/NEJMoa1906759
Reviewed by Sanjeev Sivakumar, MD
Read the article.*
Sedating patients who require mechanical ventilation has been considered standard of care; however, clinical trials have suggested worse outcomes with use of sedatives. This was a multi-center randomized clinical trial to investigate whether no sedation, compared with light sedation with daily interruption, would result in better survival among patients requiring mechanical ventilation.
The trial was conducted at eight centers across Denmark, Sweden and Norway and included patients over 18 years of age enrolled within 24 hours of mechanical ventilation. Key exclusion criteria were patients with severe head trauma, therapeutic hypothermia, status epilepticus, coma, brain death and severe hypoxia. Stratified randomization (participating center, age, presence of shock on arrival) occurred in a 1:1 ratio to nonsedation or light sedation groups.
Trial interventions: The nonsedation group did not receive sedatives, but could receive bolus doses of morphine for analgesia and if necessary, medications similar to those used for the sedation group. Crossover between groups was not allowed. The sedation group received continuous sedative infusion with a goal of light sedation (patient remained arousable with a Richmond Agitation and Sedation Scale (RASS) score of -2 to -3). Propofol was used for the first 48 hours and replaced with midazolam thereafter. This group received sedation interruption every morning with the aim of full wakefulness, and if awake, sedation was resumed at half the dose. Sedation was not resumed among patients tolerating minimal ventilator settings (PEEP of 5 cm H2O, FIO ≤40%) unless patients were uncomfortable.
Anxiety and mild agitation with sedation withdrawal was treated with bolus doses of clonidine. Analgesic regimen of paracetamol, bolus dose opioids and epidural anesthesia (when appropriate) for refractory pain was allowed for both groups. The use of dexmedetomidine was discouraged. The Confusion Assessment Method for the ICU (CAM-ICU) was used to evaluate patients in both groups for delirium twice daily. Haloperidol or olanzapine was used to treat delirium when nonpharmacological measures failed.
Primary outcome: All-cause mortality at 90 days. The authors performed an a priori sample size estimation of 700 patients, and the study was powered to show a 25% relative risk reduction of in-hospital deaths between the two groups. A modified intention-to-treat analysis was used. Multivariable logistic-regression analysis was used to report between-group differences in primary outcome.
Secondary outcomes: Number of days until death and number of thromboembolic events up to 90 days after randomization. The following secondary outcomes were measured for up to 28 days after randomization: number of days free from coma or delirium (RASS score at least -3 and negative CAM-ICU assessment); highest score on the Risk, Injury, Failure, Loss of Kidney Function and End-stage Kidney Disease (RIFLE) assessment for acute kidney injury; length of ICU stay and number of days without mechanical ventilation. A priori statistical analysis plan to correct for multiple comparisons for secondary outcomes was not included, and secondary outcomes were reported as point estimates with unadjusted 95% confidence intervals from univariate logistic regression analysis.
Exploratory outcomes were all-cause mortality at 28 days; length of stay in ICU up to death or 90 days; number of days until patient was in no need for mechanical ventilation within 90 days; length of hospital stay; organ failure on discharge from ICU; accidental extubations requiring reintubation within 1 hour and accidental removal of central catheters requiring reinsertion within 4 hours. Multivariate logistic regression analysis with adjustment for randomization stratification factors (participant center, age >65< and SBP >70<), and for Simplified Acute Physiology (SAPS II), Sequential Organ Failure Assessment Score (SOFA), shock at admission, CKD, COPD and daily benzodiazepine use before randomization, were used for exploratory outcomes. Cox proportional hazards regression and Kaplan-Meier plot were used for survival analyses.
Results: From January 2014 to November 2017, 710 patients were randomized- 354 to nonsedation group and 356 to mild sedation group, of which 700 patients were included in the modified intention-to-treat analysis. Baseline characteristics were similar between the two groups with the exception of Acute Physiology and Chronic Health Evaluation score (APACHE II), which was 1 point higher in the nonsedation group (26 vs. 25), indicating a greater chance for in-hospital death. There was no difference in mortality at 90 days between the groups [42.7% in nonsedation group vs. 37% in sedation group; 95% CI -2.2 to 12.2; P=0.65). Secondary outcome of number of days until death up to 90 days was not significantly different (nonsedation 13 days [IQR 6-27] vs. sedation 12 days [IQR 5-28]; 95% CI -2 to 5). The number of ICU-free days and ventilator-free days did not differ significantly between trial groups. The nonsedation group had 1 more day free from delirium or coma (27 days in nonsedation group vs. 26 days in sedation group). A higher frequency of major thromboembolic events occurred in the sedation group (2.8% vs. 0.3%; 95% CI, -4.8 to -0.7). All other secondary outcomes did not differ significantly between the groups.
This study found that among mechanically ventilated ICU patients, mortality at 90 days did not differ significantly among those assigned to a plan of no sedation, compared to patients receiving light sedation with daily interruptions. The number of days free from delirium and coma was 1 day more in the nonsedation group, with fewer major thromboembolic events.
Due to the lack of a pre-specified plan for multiple comparisons, however, inferences cannot be made from observations for secondary outcomes. In contrast to several trials, this study did not find a significant difference in time on ventilator or length of ICU stay between the trial groups. A plausible explanation for this finding is that the trial groups did not differ significantly in the depth of sedation, particularly on day 1: The mean RASS score in the nonsedation group increased from -1.3 on day 1 to -0.8 on day 7 and, in sedation group, from -2.3 on day 1 to -1.8 on day 7. A higher percentage of nonsedated patients were sedated during the first week, likely accounting for lower than intended between-group difference in the degree of sedation. International guidelines categorize ‘light sedation’ as score on -2 to +1 on the RASS, which is different from this trial’s target RASS score of -2 to -3 which is actually mild to moderate sedation.
There are limitations of note. The patients were on average 70 years of age, and the ratio of PaO2 to FiO2 that mandated sedation use and therefore an exclusion criterion for study participation, was 9 or lower (≤ 68 if PaO2 measured in mmHg). This is a relatively low threshold to avoid sedation. About 15% of patients refused to participate, which is high relative to similar trials, likely because patients or family did not want to get randomized to no sedation. While crossover was “discouraged,” about 27% of patients in the nonsedation group received some sedation in the first 24 hours.
This study is important, particularly considering that optimal level of sedation for mechanically ventilated patients is under debate. While the current study showed no mortality benefit of nonsedation, individual patients may want to be awake and communicate while on mechanical ventilation instead of being sedated. The study reinforces the need to monitor sedation clinically with the aim of discontinuing it as early as safely possible. It is important to note that this was a Scandinavian study, wherein university hospitals had a nurse-to-patient ratio of 1:1 in the ICU, which is likely not the case all around the world. Awake patients could be constantly monitored in this study, which likely resulted in the relatively low number of accidental extubations and removal of central catheters. Further study into the optimal level of sedation while mechanically ventilated is warranted.