We keep changing how we sedate intubated patients. Etomidate infusions were trendy for a while, until it was noticed that they suppressed adrenal function. Then benzodiazepine infusions were popular, until research showed they increase delirium and prolong intubation. Currently fentanyl infusions are en vogue.
I’ve been using fentanyl infusions as the backbone of my sedative strategy, in keeping with the 2013 SCCM guidelines. This generally works well for patients who can be extubated quickly. However, for patients who remain on the ventilator for longer periods of time, it often leads to problems involving tolerance and withdrawal.
Hence the question: What is the evidence supporting fentanyl infusions in the ICU? Is this a safe approach to patients requiring sedation for longer periods of time (for example, over one week)?
Theoretical rationale for pain-first sedation
The 2013 SCCM guidelines advocate a pain-first approach to sedation, starting with an opioid infusion as the first-line approach to sedation. Why?
A patient who is agitated on a ventilator might be suffering from pain, delirium, or anxiety. It is often difficult to tell exactly what the problem is. Prior to these guidelines it was common for patients to be treated with a sedative (e.g. benzodiazepine infusion) without any analgesia. Unfortunately, patients with untreated pain may continue to be agitated until they are deeply sedated with benzodiazepines. Using a sedative to try to treat pain works poorly.
This is the motivation for the “pain-first” approach. Most intubated patients probably have at least some discomfort. Therefore, when approaching an agitated patient on the ventilator, it could make sense to first attempt to treat this with an analgesic. Opioids have some sedative properties as well, so if the patient has a combination of pain and anxiety then an opioid could address both issues.
Evidence basis for pain-first approach
The 2013 SCCM guidelines recommended analgesia-first sedation with a Class 2B strength (weak recommendation based on a moderate strength of evidence)(1). Four studies were cited to support this recommendation:
Breen et al. 2005
This was a multi-center RCT comparing a remifentanyl-based sedation protocol vs. midazolam-based conventional therapy among 105 intubated patients (2). Within the first ten days following intubation, patients in the remifentanyl group were extubated earlier (figure below). After ten days all study drugs were stopped and patients were transitioned to conventional sedation.
However, there was no statistically significant difference in ICU length of stay. This may reflect that patients in the remifentanyl group had a higher re-intubation rate (7/57 vs. 2/48), which was not accounted for in the primary outcome of time until initial extubation (3).
Park et al. 2007
This was a single-center study comparing outcomes among intubated patients before and after implementation of a remifentanyl-based sedative protocol. Prior to implementation, most patients were treated primarily with sedatives (either propofol or midazolam).
Overall, patients spent longer on ventilation following the introduction of remifentanyl (median time on ventilation increased from 37 to 71 hours, p=0.07). Among patients treated with either propofol or remifentanyl monotherapy, the duration on the ventilator was nearly identical:
Among numerous outcomes, one was found that favored the remifentanyl group (satisfactory sedation level while on SIMV-mode ventilation). The abstract focused on this outcome, suggesting that it was a positive study (4).
This was a multi-center cross-over study comparing a remifentanyl-based regimen with PRN propofol vs. conventional therapy among patients anticipated to be extubated within 2-3 days. Patients in the conventional therapy arm received various agents, most often infusions of midazolam and morphine (yes, morphine infusions). Patients in the remifentanyl/propofol arm were extubated more rapidly (adjacent figure).
Strom et al 2010
This was a single-center RCT which randomized 140 patients to receive no sedation vs. sedation (with propofol for 48 hours, followed by a midazolam infusion thereafter)(5). Both groups received PRN boluses of morphine (2.5-5 mg). Patients receiving no sedation spent less time on ventilation, with reduced ICU and hospital lengths of stay:
This is a landmark study showing that it is often possible to ventilate patients with no sedation and fairly low doses of opioids (the median dose used was 0.0048 mg/kg/hr, equal to ~10 mg IV morphine daily). Factors which made this possible may have included attentive 1:1 nursing care and the use of pressure-support ventilation as the default ventilator mode (6). Although it may be difficult to replicate these results under different conditions, this is a brilliant proof-of-concept study.
Why these studies cannot support fentanyl infusions
The studies above led the SCCM to recommend a pain-first approach to sedation, which in turn promoted the broad application of fentanyl infusions. For example, the guidelines state “analgesics that are short-acting and easily titratable may offer an advantage by facilitating frequent neurologic evaluations.” Although this seems like a reference to remifentanyl, it would also tend to endorse the use of fentanyl. However, these studies cannot support the use of fentanyl infusions.
Strom 2010 is a well-designed study with high internal validity. However, the way opioids were used in this study (low-dose PRN morphine boluses) is entirely different from the use of high-dose fentanyl infusions.
The remaining three studies regarding remifentanyl infusions have significant limitations:
- Bias: All were sponsored by GlaxoSmithKline, manufacturer of remifentanyl. This may have colored the way results were reported. For example, Breen 2005 found that remifentanyl use reduced the duration of ventilation, a conclusion which was proudly touted in the title of the article. Alternatively, Park 2007 didn’t find that remifentanyl use reduced the duration of ventilation, a result which was barely mentioned in the paper.
- Duration of treatment: Most patients were on opioid infusions for only a few days (with no patients treated longer than 10 days). Thus, these studies cannot support the use of longer opioid infusions.
- Straw-man comparison: Most of the patients in the “control” groups of these studies received benzodiazepine infusions, which are recognized to prolong intubation. Comparing remifentanyl vs. benzodiazepine infusion is a straw-man comparison (essentially comparing the most sophisticated opioid vs. the worst sedative). A more fair comparison would be remifentanyl vs. propofol or remifentanyl vs. dexmedetomidine. The only study that evaluated this was Park 2007, which found equivalent durations on mechanical ventilation among patients receiving propofol vs. remifentanyl.
- Protocoled vs. non-protocoled sedation: These studies compared a sedation protocol involving remifentanyl vs. “usual care” without any protocol. Having a more intensive, organized approach to sedation with a protocol could be beneficial, regardless of the agent utilized.
- Different drugs: Although the above evidence was obtained using remifentanyl, in practice this often translates into using fentanyl (which is cheaper and widely available). Unfortunately, fentanyl accumulates when infused over long time periods, making it less desirable (2).
Drawbacks of prolonged fentanyl infusions
The recommended infusion rate for fentanyl in the SCCM guidelines is 0.7-10 mcg/kg/hr. At the upper end of this range, a 70-kg patient would receive 700 mcg/hr fentanyl, equivalent to 3,360 mg of oxycodone daily.
Fentanyl is typically infused at 50-200 mcg/hr, equivalent to 240-960 mg of oxycodone daily. For perspective, Strom 2010 was able to treat the pain of intubated patients with a median opioid dose equivalent to ~20 mg oxycodone daily. Why are such high doses of fentanyl often used? One explanation may be that fentanyl is functioning as a weak sedative rather than as a strong analgesic (7).
(2) Fentanyl infusions cause drug accumulation
Fentanyl has a context-sensitive half-life, which means that:
- If a single dose of fentanyl is given, it has a short effective half-life due to redistribution of drug out of the blood and into fat tissue.
- With a fentanyl infusion, eventually fentanyl builds up in the fat tissues. Subsequently the half-life is longer, because drug doesn’t redistribute rapidly out of the blood.
If fentanyl is continued at a constant rate, serum drug levels will increase insidiously:
Opioids with higher volumes of distribution and increased lipophilicity, such as fentanyl and sufentanil, may be less appropriate for longer durations or in obese patients due to the potential for drug accumulation and prolonged duration of action – Reardon 2015
(3) High-dose fentanyl might cause hyperalgesia
It is increasingly recognized that large doses of opioids may paradoxically exacerbate pain (opioid-induced hyperalgesia). Thus, beyond a point, further increases in opioid dosing are counterproductive. Remifentanyl is known to cause hyperalgesia, and it appears that fentanyl does as well (Lyons 2015). Given the high doses described above, this could be problematic in the ICU.
(4) Tolerance, dependence and withdrawal
Exposure to continuous, high levels of opioid rapidly causes tolerance and dependence. Later on, when the patient is ready for extubation, discontinuation of fentanyl may precipitate opioid withdrawal. Some evidence suggests that transitioning from fentanyl to methadone may avoid withdrawal and thereby facilitate extubation (Wanzuita 2012).
Overall, the practice of giving patients large infusions of opioids, inducing opioid dependence, and then performing detoxification with methadone just doesn’t make sense. One wonders how much of the discomfort and insomnia experienced by such patients during their recovery could be related to lingering effects of opioid withdrawal.
(5) Gastrointestinal side-effects
Opioids commonly cause constipation, distension, and ileum. This may be particularly problematic among relatively immobile critically ill patients. Occasionally it may even cause colonic pseudo-obstruction (Ogilvie’s syndrome), with a risk of perforation.
Conclusions: Has the pendulum swung too far?
Several years ago, it was common for patients to receive benzodiazepine infusions without any analgesic. This approach (which often involved trying to treat pain with a sedative) works poorly, because it essentially amounts to sedating the patient into a coma rather than properly addressing pain.
Consequently, we have transitioned to a “pain-first” approach towards sedation. However, considering the high doses of opioid which are often required, it seems unlikely that the opioid is simply working as an analgesic. Instead, it is possible that in many cases, the opioid is primarily working via its sedative properties, for the management of untreated anxiety or delirium.
We may have come full circle. Previously, we were broadly using benzodiazepines for all patients, often mistakenly treating pain with a sedative. Now, we are broadly applying opioid infusions for all patients, sometimes mistakenly treating anxiety or delirium with an opioid.
The truth might lie in the middle: no drug is perfect. Patients might benefit most from different medications, depending on what their primary problem is (pain, anxiety, or delirium). It is possible that a cocktail of several drugs in moderation might be preferable to increasing any single agent to crushingly high levels (8).
Better approach to analgesia among intubated patients?
Adequately treating pain remains essential. However, this must be done in the safest possible manner, ideally using the lowest effective opioid dose. Potential approaches may include (9):
#1: Avoiding opioid infusions?
Patients' requirements for opioids change continuously. Utilizing PRN boluses allows for continually titrating the amount of opioid required to keep the patient comfortable. In contrast, an infusion is typically down-titrated once daily, which may cause excessive opioid administration in between:
Impact of different titration schemes on the total exposure to opioid in a patient with a steadily decreasing amount of painful stimulation (black line). For example, this could occur in a patient recovering from a painful procedure.
- Fentanyl infusion with daily titration: Fentanyl is down-titrated in rough increments, once daily in the morning. In between down-titrations, the fentanyl level will increase for a period of 24 hours due to drug accumulation.
- Fentanyl infusion without daily titration: Fentanyl will progressively accumulate over time, greatly exceeding the required dose. This demonstrates how failure to perform daily down-titration can cause a fentanyl-induced coma (Skrobik 2013).
- PRN bolus strategy: Frequent boluses based on the patient's pain level allow this to more closely approximate the patient's actual opioid requirement. In between boluses, the amount of opioid will decrease steadily.
The value of continual titration of analgesics and sedatives (rather than merely adjusting them once daily) has recently been advocated by Vincent 2016. A PRN bolus strategy may work best with an opioid that has an intermediate half-life (e.g. hydromorphone). As discussed above, Strom 2010 has already proven the efficacy and safety of utilizing PRN morphine boluses.
#2: Low-dose ketamine infusions?
Low dose ketamine infusions (e.g. 0.1-0.3 mg/kg/hr) may provide analgesia with minimal side-effects (10). Combining ketamine and PRN opioids may be a viable strategy, because ketamine could provide a baseline level of analgesia while reducing the opioid requirement. Furthermore, ketamine appears to prevent opioid tolerance and hyperalgesia, which would be especially helpful among patients ventilated for longer periods of time (Joly 2005, Visser 2006).
- There is little evidence to support the use of prolonged fentanyl infusions for sedation in the ICU (e.g. longer than 10 days).
- Over time, fentanyl infusions are often up-titrated to levels that correspond to an enormous opioid dose (e.g. 200 mcg/hour fentanyl is equivalent to 960 mg oxycodone daily). Opioid doses this high might actually exacerbate pain (opioid-induced hyperalgesia).
- When used for long periods of time, fentanyl infusions may cause drug accumulation and tolerance, with subsequent opioid withdrawal.
- It is possible that the combination of a low-dose ketamine infusion with PRN opioid boluses could achieve effective analgesia with fewer side-effects.
- Paradigm for post-intubation pain, agitation, and delirium (EMCrit)
- Post-intubation package (EMCrit)
- Intravenous olanzapine (PulmCrit)
- Dexmedetomidine: for NIV, for extubation (PulmCrit)
- Moderate evidence strength was defined in the guidelines as an RCT with significant limitations or high-quality observational study.
- Remifentanyl is an opioid given by continuous infusion with an extremely short half-life, allowing for rapid and precise titration. Unlike fentanyl, it doesn't accumulate over time and its clearance is independent of renal or hepatic function. Although remifentanyl has very desirable pharmacokinetic properties, it is more expensive than fentanyl limiting its widespread adoption.
- A more appropriate primary endpoint might have been ventilator-free time, because this would account for patients who require re-intubation.
- This paper didn’t specify a primary endpoint. It would have been very difficult to even try to claim that anyone picked this a priori as a primary endpoint.
- If absolutely needed, patients in the “no sedation” arm were allowed to be placed on propofol infusion for six hours, with subsequent discontinuation and another attempt at management without sedation. Additionally, patients could receive PRN haloperidol.
- Pressure-support ventilation is arguably the most comfortable ventilator mode, given that it allows patients to choose their own tidal volume, flow rate, and inspiratory time. This eliminates many sources of dyssynchrony with the ventilator.
- Opioids are primarily analgesics, but they do have some sedative properties. Imagine that a patient on the ventilator was anxious or delirious, but without any pain. If such a patient were treated with fentanyl, the analgesic effects of fentanyl would have no effect on them. Thus, they would require a much higher dose of fentanyl, in order to take advantage of the sedative effect of the fentanyl.
- Perhaps ketamine will be the exception to this principle, but this remains to be shown.
- Additional approaches to reducing opioid requirement include nerve blocks, epidural analgesia, and various adjunctive agents (e.g. acetaminophen). These strategies should be employed regardless of the systemic medication strategy.
- The side effect of greatest concern is delirium, which can occur at intermediate sub-dissociative doses of ketamine (see Rubin Strayer’s blog on the ketamine-brain continuum). However, at very low doses (e.g. 0.12 mg/kg/hr), the risk of psychiatric side-effects is minimal, with the ketamine level falling into the analgesic-dose range (Jouguelet-Lacoste 2015, Joly 2005, Forero 2011). Meanwhile, these doses of ketamine appear to retain a significant analgesic effect. Thus, the benefit/risk ratio of extremely low dose ketamine infusions may be substantial. Among patients with uncontrolled pain or high opioid consumption, the benefit/risk ratio may favor higher doses.