Staying woke II: Managing sedation in the ICU

Now that we’ve been introduced to the basic framework of ICU liberation, let’s discuss some particulars.

The topic today: sedation.

The gist: All sedation is bad. Some sedatives are extra bad.

Let’s get into the particulars.

Is deep sedation harmful?

Deeply sedating critically ill patients seems to be associated with poor outcomes. Why? Perhaps because sedated patients are difficult to wean from mechanical ventilation (having little respiratory drive), difficult to mobilize (being too sleepy), and prone to delirium (since hypnotic/amnestic agents disorder the process of consciousness and memory-formation). Regardless of mechanism, it seems clear that the ideal ICU patient would be awake, calm, with controlled pain, and non-delirious. Certainly, some patients do need sedation, but when possible, less is better than more, and none is better than some.

The evidence?

The study

Stephens 2018

  • Systematic review of “deep sedation” (defined variably, but often as a RASS [Richmond Agitation Sedation Scale] of –3 or less] occurring within the first 48 hours after intubation.
  • Early sedation increased mortality by ARR 18.4% (9.2% vs 27.6%)
  • … increased vent time by 2.1 days
  • … increased ICU length of stay by 3 days
  • Trend toward increased delirium and tracheostomy requirement

Shehabi 2012

  • Multicenter Australian prospective cohort study
  • Medical and surgical ICU patients on the vent
  • Evaluated for deep sedation (RASS -3 to -5) within the first 48 hours, with RASS evaluated every 4 hours
  • After multivariate analysis, every instance of deep sedation was associated with a 12.3 hour increase in total vent time
  • In patients with any early deep sedation, median vent time was prolonged by 5 days (7.7 vs 2.4 days)
  • In-hospital mortality was significantly higher in deeply sedated patients (HR 1.1) and 180-day mortality was higher (HR 1.08)

Tanaka 2014

  • Secondary analysis of data from a multicenter Brazilian ARDS trial
  • Medical and surgical ICU patients on the vent
  • Evaluated for deep sedation at 48 hours (using Glasgow Coma Score [GCS], which has been shown to closely correlate with RASS)

Deep sedation with GCS <9 (equivalent to a RASS <-2) was associated with…

  • Increased vent days (7 vs 5)
  • More patients needing tracheostomies (38.9 vs 22%)
  • After multivariate analysis, independently associated with mortality (OR 2.36)
  • Benzodiazepines tended to be the agent most associated with deep sedation, compared with fentanyl or Precedex.

Which agents are most harmful?

THE data

Numerous studies have specifically investigated the effects of benzodiazepines. A brief overview:

  • Carson 2006: Intermittent Ativan boluses lead to longer LOS and longer vent times than propofol infusions.
  • MENDS: Precedex drips lead to more days without delirium or coma than lorazepam infusions.
  • SEDCOM: Precedex drips lead to less delirium, shorter sedation duration, and less vent time than midazolam infusions
  • MIDEX-PRODEX: Precedex drips lead to less vent time than midazolam drips

In summary, a meta-analysis…

  • Fraser 2013 meta-analysis: Non-benzodiazepine sedation is associated with shorter LOS and less vent time than benzodiazepine-based sedation.

Is sedation needed at all?

If sedation is generally undesirable, a natural question is whether sedation is generally a necessary element of an ICU stay at all. Many providers would say “yes,” at least for intubated patients, but the data would suggest otherwise. Here are two examples of randomized trials, in both medical and surgical patients, demonstrating the feasibility and benefit of a “no sedation” approach to routine ICU care.

The study

Strom 2010

  • Single-center RCT of 140 med/surg ICU patients on the vent
  • Within 24 hours of ICU arrival, randomized to receive analgesia only (no sedation) or analgesia plus sedation with propofol (for up to 48 hours) and midazolam infusions (after 48 hours), with daily sedation interruptions.
  • Uncomfortable patients received analgesia, reorientation, sitters; no restraints were used.
  • Delirious patients received Haldol, then if needed a 6-hour course of propofol, after which it was weaned back off.
  • Up to 3 cycles of this were allowed before no-sedation was considered a “failure” and propofol was left running.
  • Both groups were mobilized daily to a chair if possible.

“No sedation” group had…

  • 4.2 fewer vent days (9.6 vs 13.8)
  • 9.7 fewer ICU days (13.1 vs 22.8) and 24 fewer hospital days (34 vs 58)
  • Non-significant trend toward reduced mortality: 16% ARR in ICU (22% vs 38%), 11% ARR in hospital (36% vs 47%)
  • No difference in accidental extubations
  • Only 18% needed crossover to the sedation group, most of them not for intractable agitation but for respiratory failure (e.g. ARDS) thought to benefit from deeper sedation.
The study

Chanques 2017

  • RCT in three French ICUs of 137 post-operative abdominal surgery patients
  • Similar protocol, with a “no sedation” group that had analgesia continued but sedation immediately halted after arriving intubated from the OR

“No sedation” group had…

  • 1.75 fewer vent days
  • 2 fewer days delirious
  • 8 fewer hospital days
  • Non-significant trend toward less mortality and fewer ICU days

Should sedation/analgesia be delivered by continuous infusion or intermittent boluses?

Even with similar sedation goals, the mode of delivery for sedatives and sedating analgesic agents can vary from continuous infusions to intermittent (usually as-needed) boluses alone. Does the method matter?

It might. Although in theory these strategies can result in similar overall drug delivery, in practice intermittent boluses often result in less sedation. It’s easy to turn up a continuous drip and leave it on for hours, days, or weeks. Giving the same amount of drug by intermittent push requires active effort, and generally doesn’t happen. As a result, simply avoiding drips for routine sedation and analgesia may be a useful practical tool for limiting drug burden.

The data?


Kollef 1998

  • Prospective cohort study of 242 mechanically-ventilated MICU patients
  • Compared patients who received continuous sedation or opioids (any type) versus intermittent bolus doses alone
  • Agents were mostly lorazepam and fentanyl

After multivariate correction, intermittent bolus group had…

  • 2.9 fewer vent days (3.3 days vs 6.1)
  • 2.4 fewer ICU days (7.2 days vs 9.6)
  • 5.6 fewer hospital days (13.8 days vs 19.4)
  • 10.4% ARR less reintubation (15.1% vs 4.7%)

Is a daily sedation interruption sufficient to limit the harm from oversedation?

The data is mixed.


Kress 2000

  • Single-center RCT of 128 vented MICU patients
  • Protocolized daily awakening trials resulted in 3.9 fewer vent days and 3.5 fewer ICU days, though questionable decrease in total drug use.
  • Also supported by Girard 2008, which combined daily sedation interruptions with spontaneous breathing trials.
But maybe not…

Mehta 2012

  • Multi-center RCT of 423 vented medical and surgical  patients using a light-sedation strategy
  • No difference in vent time, ICU/hospital stay, or delirium, although 7 fewer vent days in surgical/trauma subgroup
Cochrane weighs in…
  • Cochrane 2014: No strong evidence of benefit from protocolized daily sedation interruptions. The signal may exist but is fairly weak, especially when applied in the context of an already-sound sedation strategy.

What about protocolizing your sedation and analgesia?

Without necessarily making any specific changes in management strategy, does simply standardizing your approach to sedation using a unit-wide protocol — establishing consistent first-, second-, and third-line agents, titrating their use to pain and sedation scales, rigorously adhering to daily interruption strategies, etc — improve outcomes compared to a flexible approach determined by the whims of each provider? The answer seems to be yes.

Several studies found that…
  • Awissi (I-SAVE) 2012 and Skrobik 2010: Protocolization reduced ICU LOS 1 day, hospital LOS 28 days, and vent time 1.6 days
  • Arabi 2007: Protocolization reduced ventilator-associated pneumonia 17%, with a non-significant trend to less ICU/hospital LOS and fewer vent days
  • Mansouri 2013: Protocolization reduced ICU LOS 3 days, vent time .9 days, and mortality 11.3%


So what do we know about sedation in the ICU?

  1. Deep sedation increases ICU stay, ventilator time, and complications, even when used for short periods.
  2. Benzodiazepines are especially associated with poor outcomes, outside of special circumstances like ethanol withdrawal.
  3. Continuous infusions are worse than intermittent boluses, even with similar sedation goals.
  4. Dexmedetomidine (Precedex) may be superior to other infusions.
  5. Daily sedation interruptions may be helpful, but not as much as avoiding deep sedation altogether.

Come back next time, when we’ll talk about the potent effects of early mobilization.

4 Replies to “Staying woke II: Managing sedation in the ICU”

Leave a Reply

Your email address will not be published. Required fields are marked *