Never let it be said we’re not ahead of the game here in Virchester. We have had a policy for the use of Patient Controlled Analgesia in the ED for longer than I have worked here. Having spent longer than average (for an EP, at least!) training in anaesthesia, I am a big fan of the pump that cares, both in terms of it providing better analgesia (common sense, that one) and reducing the time that I need to stand by the bedside of the patient squeezing in morphine. Not that I particularly mind doing this, it’s just that there are a lot of other calls on all of our time and it make sense to automate the process as much as possible. I am also a great believer in empowerment, both of the nurses and our patients, and don’t like patients being in pain unnecessarily. I share the views of Ed Gentile, who expounded his views on pain management in the continually excellent EmCrit blog and podcast.
At JC this week (it was this week, when this was originally written, but being a bit slack and allowing myself a holiday, means there’s been a bit of a delay) we searched for the evidence behind the use of patient controlled analgesia. Now, there’s lots of evidence for PCAs in the perioperative period, which we have generalised, rightly or wrongly, to acute pain in unselected ED patients. As the authors of the paper we discussed, Efficacy of Patient controlled Analgesia for Patients With Acute Abdominal Pain in the Emergency Department: A Randomized Trial, from Academic EM a couple of months ago, suggest, “there is a paucity of research on the use of PCA in the treatment of acute pain in the ED.” So have they given us anything to go on? Well, a 3 arm RCT is not to be sneezed at. The authors randomised 211 unselected patients with abdominal pain to 2 different dose schedules of morphine PCA (1mg and 1.5 mg per press) after an initial PCA delivered bolus of 0.1mg/kg morphine versus what most of us would recognise as fairly conventional care – an initial opiate bolus as above, followed by “physician-managed analgesic supplementation as needed.” Now, as much as I’d love to be there squeezing in morphine as and when you need it, we are all aware that staffing levels aren’t optimal in most EDs and we rely a lot on our excellent emergency nurses to be proactive, as well as ancillary analgesia such as IV paracetamol and NSAIDs. The randomisation was by block in a 1:1:1 ratio, into consecutively numbered opaque envelopes.
So far so good, so why not look at the abstract, and ideally read the paper as well?
Pain was assessed initially on a numerical 0-10 scale (NRS) that we are all familiar with, unless your work is exclusively with prepubescent children! Pain was reassessed every 30 minutes by the nurse looking after the patient, who also noted any adverse events (hypoxia, bradypnoea and hypotension). There was also a qualitative element to the work as patients were asked at the end whether they were satisfied with their treatment and whether they would have that method of analgesia again. Unsurprisingly, pain was better treated in the PCA groups and the vast majority of patients would use a PCA again.
As with a lot of papers, the first superficial read is encouraging. You know, the one that you do after the post has arrived, whilst having a coffee, prior to leaving for work. Abstract, CONSORT diagram, main results figures all look ok. It’s only when you sit down to appraise the paper properly that the holes start to appear. And that’s one of the reasons I love JC.
We were a bit savage today. Where was the blinding, we asked? I appreciate it would have been more work, but the study would have been improved immensely if there was even one-sided blinding (i.e. patients on the standard regime having a PCA with a syringe of saline). There could have been double blinding with little more difficulty. The authors do comment on this at the end in limitations, and I agree with them in that this is pragmatic, but if our gold-standard is PDBRCT, then that is what we should aim for, surely? The risk of bias introduced by having your patient hitched up to a magic pain pump cannot be understated; I would go so far as to suggest that it undermines the qualitative aspect of the research – i.e. the question might as well be “do you want a pain pump that you control, or do you want an overworked Emergency Nurse to look after your pain?”. Hmmm.
We were also amazed by the exclusion criteria. 78% of those approached did not take part and were excluded through various stringent criteria. I’m not convinced that in a pragmatic study, as this is, excluding ED patients who have taken opiates within the 24 hours prior to attendance is practical, as most people these days seem to have access to cocodamol. There are 250 potential participants excluded for whom no reason is given. This is greater than the sample size used.
A word on the sample size. It’s suggested in the methods that SSE have been done. I’m sure they have, but the reader will find it difficult to do likewise if there is no beta value given. We cannot tell what level of power there is. I’m also dreadfully cynical if SSE is identical to actual n, as happened here – that was lucky!
Being a rufty, tufty northern bloke, the pain scores given initially, make me think that the good folk of the Bronx are possibly a tad soft – pain score 8-10 in over 80% of patients across all the groups. Either that, or it suggests a misunderstanding of the pain score. And it certainly doesn’t go up to 11. We wondered also about the initial dosing of opiate. It was supposed to be a 0.1mg/kg bolus, and the patients weighed 80-odd kg (+/- loads!) and only got 8 mg average doses of morphine, with fairly narrow CIs. It looks like it’s not only StE’s that just gets out the 10 mg amp of morphine and gives that! There also wasn’t very much extra analgesia given to the poor patients in the control group. Extrapolating from the figures given, approximately 30% received extra morphine, raising the total dose from 7.9 to 9.6. This suggests that those who moaned the loudest got an extra 5 mg, equalling the doses given to the PCA crew.
Another weakness of the paper is the lack of an idea about diagnoses. There’s a lot of things that can go wrong intraabdominally, ranging from the quite sore that settle, eg biliary colic, to the ridiculously sore that ain’t going away like mesenteric infarction. For some conditions, opiates may not even be the best bet – I’m old enough to remember giving pethidine to pancreatitics as “morphine causes increased Sphincter of Oddi spasm!” – thankfully, by the 21st century, we have decided that giving a weak opiate with constipating and dysphoric side-effects wasn’t the greatest of ideas. Some rough idea about the final diagnoses may give a better idea to the reader.
So did it work? Well, not really. The authors acknowledge that 1.3 on the NRS is a significant change. There is exactly that between the control group and the 1.5 mg dose PCA group at 2 hours and no difference at all in the pain score at 30 minutes. The box and whisker plots reflect what I have written above – the control group got a decent slug of morphine and then a select few got more morphine – the mean NRS falls to a steady level at 30 mins and there are wide IQRs; the NRS of the PCA groups keeps on falling, because they keep getting morphine. Unsurprisingly, the PCA groups were more satisfied and would recommend PCA.
So will it change my practice. No, not really. If you’ve got pain that is severe and likely to be ongoing or recurrent then you are getting a PCA. The sicklers, the acute pancreatitics, the ones who actually look like they’ve got of pain score of greater than 7 rather than the “11s” eating a McDonalds’ with normal physiology. I think what’s more important is to have a decent departmental pain strategy, using analgesics that work on a wider range of receptors (well COX-2 and whatever we reckon paracetamol works on this week!). That’s another job to add to the list of stuff to do…
Bottom Line: PCAs provide less labour-intensive analgesia, with slightly better pain scores in the first two hours of an ED stay, at the cost of slightly more adverse events – nausea, vomiting and pruritis. They are probably worth it if the patient has a truly painful ongoing condition.
Thanks to Alison Sheehan, Registrar in Emergency Medicine for her work in leading this JC.
What advantage(s) are there to the authors using AUC (area under the curve) as an outcome measure for pain?
The authors could just use single time points for pain, but this can cause problems in situations when peak pain (or low levels of pain) might occur in a manner that does not truly reflect the burden on the patient. The idea of AUC is that it gives a measure of total pain burden, though it is dependent on how frequently the pain is measured. On balance it is a better way of assessing overall pain as it incorporates both time and magnitude of pain over a prolonged period.
Block randomisation was used in this study. Explain what is meant by block randomisation.
Firstly – I don’t care what the Americans say randomisation has an S in it! Ok…..well hopefull you understand the concept of randomisation and the reasons why we do it. Block randomisation is a particular method that can be used to avoid a potential problem if we allow randomisation to be truly random. (Eh). Hypothetically let’s randomise 100 patients to treatment A or B using a computer program. The likelihood is that it wil look like a jumbled mess BUT it’s quite possible that the first 50 would get treatment A, then 50 treatment B. Honestly, if truly random any combination is possible. So if we set off on our study and for whatever reason have to stop after 50 patients we would have no data to analyse. Block randomisation does not create a ransomisation sequence based on the individual, but rather on blocks. Example Let’s randomise in blocks of 4 (they used 12 in the study). How many ways can you create a sequence of 2 As and 2 Bs? 1. AABB 2. BBAA 3. ABAB 4. BABA 5. BAAB 6. ABBA (the dancing queen sequence) So 6 options. Now, let’s generate a random sequence of numbers between 1-6….it might come out as 1:3:6:6:5 Now, let’s convert those numbers back into blocks AABB ABAB ABBA ABBA BAAB And hey presto, we’ve got a block randomised sequence based on blocks of 4. So, the advantage here is that wherever we get to we are never more than 2 patients away from having equal numbers in each group. 🙂
Before you go don’t forget
- Subscribe to the blog (look top left for the link)
- Visit us on Facebook
- Subscribe to our PODCAST on iTunes