A new study in NEJM compared nonsedation with light sedation in ventilated ICU patients. This is a follow-up study, aimed at clarifying the results of a prior trial at the same center. To best understand the current study, we need to start with the first trial…
Strom et al. 2010: A protocol of no sedation for critically ill patients receiving mechanical ventilation: a randomized trial
This is a single-center RCT performed in the general ICU of Odense University Hospital in Denmark.1 Prior to the study, their standard practice was a protocol of no sedation.
This trial randomized 140 patients to receive nonsedation versus sedation, as follows:
- Nonsedation group: Boluses of morphine (2.5-5 mg) or haloperidol (1-5 mg) were allowed for management of pain or delirium. Restraints were not used. If necessary, a person was assigned to sit with the patient and provide verbal reassurance. For refractory agitation, propofol was allowed for 6 hours, with subsequent weaning. If patients had three episodes of refractory agitation, ongoing propofol infusion was allowed.
- Sedation group: Boluses of morphine (2.5-5 mg) were used for pain. Propofol infusions were used for the first 48 hours, targeting light sedation. After 48 hours, the sedative was changed to an infusion of midazolam. Daily sedation interruption was performed. When lung function improved and FiO2 decreased to 40% on 5 cm PEEP, sedation infusions were stopped and patients were transitioned to a nonsedation strategy (same as above). However, if respiratory decompensation subsequently occurred (FiO2 >50% and PEEP >8 cm), then sedation infusions were resumed.
Both groups of patients were mobilized aggressively, despite mechanical ventilation. Patients in the nonsedation group could be mobilized throughout the day, whereas patients in the sedation group could be mobilized only during their morning sedation interruption.
Patients who died or were extubated within 48 hours of randomization were excluded from the study (which may have introduced selection bias), leaving a total of only 113 patients in the study.
Subjects included intubated patients in a single medical/surgical, 18-bed ICU. There is some gender imbalance, with only 24% of the nonsedation group being women:
Patients in the nonsedation group did receive more PRN haloperidol (table below). Additionally, 18% of the patients in the nonsedation group didn’t tolerate this therapy and eventually required ongoing sedative infusions (largely for management of severe ARDS, but on one occasional due to family request).
The primary endpoint of the study was ventilator-free days. Nonsedation successfully reduced the length of ventilation, length of ICU stay, and length of hospital stay (table above). “Agitated delirium” requiring haloperidol was more common in the nonsedation group (7% vs. 20%, p=0.04). No difference was found in accidental extubation.
Mortality was increased in the sedation group (47% vs. 36%, p=0.27). What do you think about that? We’ll revisit this below.
So this was a very promising trial, but with several limitations:
- Single-center design with relatively few patients and some baseline imbalance between groups.
- Dubious generalizability (many hospitals lack the resources to maintain patients without sedation, as this requires very intensive nursing care and patient supervision). Additionally, two experienced physicians were present on the unit at all times – far from usual practice in many units.
- Use of midazolam as a sedative (given that benzodiazepines often increase delirium and duration of ventilation).
- Being underpowered to evaluate for uncommon adverse events.
NONSEDA trial 2020 (Olsen et al.): Nonsedation or light sedation in critically ill, mechanically ventilated patients.
This is a multi-center RCT involving 700 intubated patients randomized to a strategy of sedation or nonsedation.2
Exclusion criteria and enrollment occurred as shown below. Unlike Strom et al, very few patients were removed from the trial following randomization:
Baseline characteristics were well matched between groups.
Groups were managed as follows:
- Nonsedation group: Efforts were made to avoid sedation (but this was permitted if necessary).
- Sedation group: Sedative infusions were used to target a light level of sedation. Propofol was used for the first 48 hours, and subsequently this was transitioned to a midazolam infusion. However, in practice it seems like propofol was actually used for longer periods of time (table below). Sedation was interrupted every morning and then resumed at half the prior infusion rate. Once patients weaned down to a FiO2 of 40% and PEEP of 5, sedation was stopped if possible.
The primary endpoint and most of the secondary endpoints were neutral, as shown below (including mortality, ICU length of stay, ventilator-free days, and delirium/coma-free days). Fewer patients in the nonsedation group developed a major thromboembolic event (p=0.01), but this finding is of questionable significance given the multiple comparisons involved in the six secondary endpoints.
There were signals of harm with a nonsedation strategy (table below). Nonsedated patients were more likely to self-extubate and subsequently require re-intubation. Nonsedation also increased the risk of removing other lines and tubes.
deep thoughts on Strom et al. 2010 & NONSEDA 2020
#1. Why was nonsedation beneficial in Strom et al., but not beneficial NONSEDA?
There are many possible reasons that a small, single-center study may not be replicated in a multi-center RCT. For example, it’s possible that the nursing staff at Strom’s center was unusually adept at managing patients without sedation. The first study also had some baseline imbalances, with sicker patients in the sedation group. It’s impossible to know for certain.
Another clue for why NONSEDA was a neutral study may lie in the RASS scores (shown above). Patients in the sedation group truly received very light sedation (often with a RASS of -2, corresponding to brief awakening to voice). Thus, there wasn't a huge separation between groups. Achieving a target RASS of -2 reflects nearly perfect sedation, but this is a difficult target (in practice, many ICUs in the United States probably sedate patients more deeply than this). Thus, in some ways the investigators were victims of their own success – their use of sedative infusions was so precise that it wasn't possible to demonstrate differences versus nonsedation.
#2. NONSEDA validates the practice of using light sedation
Much of the literature on sedation in the ICU seems to suggest that lower levels of sedation are beneficial. And this is often true, particularly in a historical context where previously patients were often sedated to unnecessarily deep levels.
NONSEDA suggests that light sedation is safe. Compared to a nonsedation strategy, light sedation didn’t increase ventilator or ICU lengths of stay. So, we may have reached the point where further reductions in sedation intensity are no longer beneficial:
#3. NONSEDA and Strom et al. imply that we are over-utilizing opioids in the ICU (and perhaps we should be using morphine more often?).
Perhaps the most fascinating aspect of both of these trials is the minimal use of opioids. In both trials, the average use of opioids was exactly the same (~0.005 mg/kg/hour of morphine, which equates to ~8 mg of morphine daily). This is vastly less opioid than many patients receive in ICUs in the United States. In comparison, a fentanyl infusion at 50 mcg/hr provides roughly twenty times this opioid equivalent!
This suggests that we often misguidedly using opioids not as analgesics, but as weak sedatives. For example, large doses of opioid may be used to calm down patients who are not truly in pain; in this context, the opioid works due to its sedative and euphoric properties. If patients were more awake and able to communicate (as in NONSEDA), they might tell us that they're not in pain – they're just anxious.
Both Strom et al. and NONSEDA used PRN boluses of morphine. This may have allowed them to reduce opioid usage for the following reasons:
- This avoids continuous opioid infusions. Opioid infusions tend to be left running, leading to administration of large doses of opioid over time. The first step to curbing opioid use in the ICU is to eliminate opioid infusions (or, at a minimum, wean them very aggressively).
- Morphine was used. Morphine is notable among the opioids in that it has relatively less euphoric effect. Lack of a rapid euphoric/sedative effect may dissuade the inappropriate use of morphine as a sedative agent.
Traditionally, morphine has not been preferred as an analgesic in the ICU (given that it may cause histamine release). However, successful use of a PRN bolus morphine strategy in both Strom et al. and NONSEDA suggest that this is worth further consideration.
#4. NONSEDA illustrates the futility of using mortality endpoints
The futility of using mortality endpoints in critical care MC-RCTs was just explored last week on the blog. In a modern critical care context, it’s extremely unlikely that adjustments in the sedative regimen will cause robustly reproducible changes in mortality. Thus, it was highly predictable that the primary endpoint of NONSEDA would be neutral.
#5. Is nonsedation a viable clinical strategy?
Yes, nonsedation is an absolutely viable treatment option. I’ve used this approach with many patients in the past, and plan to continue this practice in the future.
The amount of analgesia and sedation which patients require is enormously variable. This may relate to variability in the gag reflex (some people completely lack a gag reflex, whereas other people have a vigorous gag reflex). For whatever reason, some patients are clearly quite comfortable being intubated with no sedation.
The NONSEDA and Strom trials remind us that nonsedation is a viable treatment option for selected patients (certainly not all). The ultimate truth of analgesia and sedation is that we must always be paying careful attention to our patients and using the minimal dose of medication required to achieve comfort. For some patients, that dose will be zero milligrams.
- NONSEDA is a multi-center RCT involving intubated patients randomized to a nonsedation strategy (with sedation only allowed for refractory agitation) versus light sedative infusion with daily interruption (largely using propofol).
- Most endpoints of the study were neutral (e.g. mortality, ventilator-free days). This may be somewhat reassuring, validating that light sedation doesn't prolong intubation.
- Nonsedation caused a small reduction in venous thromboembolism, but this came at the cost of a substantial increase in accidental extubation. Overall, nonsedation doesn’t seem like a viable strategy for most patients. However, this may work for patients who are able to tolerate the endotracheal tube surprisingly well.
- NONSEDA and Strom et al. remind us that most patients can tolerate intubation with fairly low doses of opioids. This casts some doubt over whether medical ICU patients truly need ongoing fentanyl infusions (which generally provide considerably higher amounts of opioid).
Image credit: UpSplash by Sophia Kunkel
- 1.Strøm T, Martinussen T, Toft P. A protocol of no sedation for critically ill patients receiving mechanical ventilation: a randomised trial. Lancet. 2010;375(9713):475-480. doi:10.1016/S0140-6736(09)62072-9
- 2.Olsen HT, Nedergaard HK, Strøm T, et al. Nonsedation or Light Sedation in Critically Ill, Mechanically Ventilated Patients. N Engl J Med. February 2020. doi:10.1056/nejmoa1906759
- IBCC chapter – Disseminated Intravascular Coagulation (DIC) - January 18, 2021
- PulmCrit- RCTs don't justify using convalescent plasma or antibody cocktails - January 14, 2021
- PulmCrit – Six RCTs to answer one question: what is the role of tocilizumab in COVID-19? - January 12, 2021