Having read it closely, I’m not so sure this paper demonstrates anything more than a fluke result. The causes of death were not predominately cardiac (which you might expect given the ‘toxicity’ of Nicotine via it’s sympathetic effects) and there was no difference in cardiovascular support – biological plausibility seems tenuous. The face validity of the paper is also pretty shaky, these are short stay ICU patients, the majority of whom were admitted for monitoring with overall median predicted mortalities of less than 1%. Such a mortality difference just secondary to a Nicotine patch seems fantastical.
However, this paper doesn’t stand alone in suggesting harm. This retrospective study (of similar methodology) suggested a statistically non-significant increase in mortality of smokers given NRT (n=67) compared with those who were not after coronary artery bypass grafting. The authors accept that the study was small and conclude it should be regarded as a pilot study with a high chance of type 2 error (false negative).
In 2012, a group from Guy’s and St Thomas’ conducted a further retrospective study looking at all smokers admitted to their mixed ICU (so should be applicable to our practice). They identified 423 smokers admitted before they stopped the prescription of NRT (in response to the Lee paper described earlier), only 73 of whom (17%) were given NRT. There was no evidence of harm.
It’s probably fair to say that the evidence of harm from NRT is not what I thought it was having looked at just one paper (and in all honesty probably just the abstract). What these papers also show is that it is the administration of NRT in critical care is interestingly not as common as you might think. Needless to say all of the above papers call for a prospective RCT in their conclusions.
So should we prescribe it more if it’s not harmful? If there is a benefit to NRT, it is likely to be in the avoidance of delirium, which is associated with an increase in mortality, increased length of stay etc.
The evidence for smoking (and therefore Nicotine withdrawal) being a risk factor for ICU delirium is surprisingly also not what you may expect from your experience.
Hsieh et al. conducted a systematic review of the link between smoking and delirium. They found the evidence was poor, with co-founders being poorly taken into account and objective measures of the degree of dependence being rarely used. They conclude that “There is currently insufficient evidence to determine if smoking is a risk factor for delirium” and call for more studies!
The only study identified by the review that was designed to directly answer the question of a link between smoking and delirium was by Lucidarme et al. This study was small, with 916 patients screened to find 44 active smokers who were expected to be ventilated for >48 hrs (suggesting a lower rate of smoking than in Sunderland!). The smokers were more likely to be agitated but not to have delirium, although in clinical terms the agitation they describe I would describe as (perhaps incorrectly) delirium given that the agitation was associated with more self-removal of tubes and catheters, and an increased rate of physical and chemical restraint.
So overall, whilst not an evidence free zone, it doesn’t look as if ‘the evidence’ is going to give the answer – there is certainly no RCT magic bullet. The most recent systematic review I could find (which also happens to be open access) sums it up when the authors conclude “… With only equivocal evidence of efficacy and signals suggesting increased toxicity, we believe that it’s (NRT) use should be limited to selected patients where the potential benefit clearly outweighs the risk”. Whilst this is true for most things in critical care, for NRT in particular this seems to be the case and it will come down to an individual judgement.
Maybe the bigger question is whether my pride will let me prescribe a patch next time she asks me….
Please leave a comment below – it might be interesting to discuss whether you think an RCT could be performed, and how such a trial may be designed?