sinaiem.org

This journal club was a look at four papers, four ways of treating pain in the ED — from mild to severe — using four approaches a little outside our comfort zone of NSAIDS and opiates.

I began the hour with a quick look at a fifth paper (Chang and Gallagher, Annals of EM 2006, Vol. 48 No 2), previously covered in this blog, about a RCT of patient’s reduction in pain after hydromorphone (dilaudid) vs. morphine. It served as a good introduction to pain assessment tools and I liked the author’s candid writing about factors in physician ordering beyond need — such as perceived price or just the psychological barrier to ordering 10mg of morphine over 1 mg or 2 mg of dilaudid. I also had to point out how great it was to see acknowledgment in print of our practice reality — the mode for initial pain score in this study was 10/10. A good read if you have the time.

I also recommended as background reading the section on Pain Assessment Tools, page 8-11 in the July 2006 edition of EMPractice by Curtis & Morrell.  The sections on pathophysiology of pain and ED pain epidemiology was also recommended.

Our first paper, covered by the EM-1 class, was Hijazi R et al. Effect of topical alkane vapocoolant spray on pain with intravenous cannulation in patients in emergency departments: Randomised double blind placebo controlled trial. BMJ 2009 Feb 10; 338:b215. PMID: 19208703. Abstract. Paper.

This was a double-blind RCT intention-to-treat analysis in a single center, midsize Australian ED, with adults, over a six month period. Their inclusion criteria was simple — age over 18 and needing IV cannulation. Their exclusions were extensive, including anyone with moderate or severe pain, AMS, severe illness, urgent need, and more. Who does that leave? Well, maybe people reporting for transfusions. At any rate, it took them 6 months to enroll 300 patients, which I think we could do in a weekend. Then again, this was a convenience sample based on the PI’s banker-like hours.

The primary outcome measure these authors studied was pain on IV cannulation, with secondary outcomes of: discomfort with spray on administration, success rate of IV cannulation, patient willingness to try again, patient guess as to what they received, and unexpected events. They used a VA scale of 0-100 mm horizontal line, with left = no pain and R= worst pain imaginable. After cannulating, the examiners waited one minute, then mark perceived pain and initial discomfort with the spray. Then they asked about future use and guessed randomization. Demographics were recorded, and the PI attempted to reach patients 5 days out as well, three times, to assess adverse reactions.

They say the VAS beats verbal descriptors of pain. They found pain starts at 9-18 mm and and mean IV pain is 30 mm. They powered their study to detect a 10 mm difference in the groups. They assessed 304 patients and enrolled 201. No difference in demographics were seen after randomizations. There were 5 protocol violations: one in the control, and four in the intervention group. Cannulation pain median averaged 36 mm in control (19-51mm) and 12 mm in intervention group (5-40mm range) … this was significant (median difference = 24 mm, mean difference = 18mm).

Furthermore, 60% of control patients had pain above 30 mm (their threshold for ’significant’ pain), compared to 33% in intervention group… also significant. You could go on to calculate the absolute risk reduction of significant pain = 60%-32% = 28%, giving you a NNT of 3.6 patients to prevent one painful stick.

The median discomfort from sprays was 0, but 49% in intervention group reported something, as opposed to 24% in control group. Success at cannulation was unaffected by spray (no surprise). Patients planned a future choice of same spray 39% of the time in the control group, but 62% in the intervention group, which gives you an ARR of 23% for ‘not wanting same spray again’ and thus, a NNT of 4.35 patients sprayed with vapocoolant to prevent one from being dissatisfied with his or her spray.

Patients were able to guess they were in the control group 69% of the time, compared to 54% for the intervention group. The authors interpret this as meaning things were well-blinded but I say it means IV cannulation doesn’t cause much pain, so not much effect can be seen. Instead, this form of pain management has a heavy psychological component.

Two intervention patients noted some transient redness when called at 5d of followup. This is low and much better than EMLA or lidocaine or tetracaine in other trials.This is the biggest trial ever of vapocoolants, agreeing with 2 smaller ethyl chloride studies and in contrast to an ethyl chloride and fluorohydrocarbon study. The no-effect studies used larger needles, farther away, smaller.It would’ve been great if these authors had recorded patient response based on needle size, location of cannulation, or skill of operator (all probably recorded elsewhere) — it might make for a nice subgroup analysis.

• Putting this in a wider context: Vapocoolants are a pretty simple, ‘cool’ way to temporarily disable pain nerve transmission. They’re faster than EMLA and less likely to distort tissues, unlike lidocaine +/- epi.  There’s a theoretical risk of frostbite, and a risk of explosions if improperly stored. Using them costs about US$0.14 per spray, so it’s probably worth it in adults, especially jittery ones.

The next paper, for the EM-2’s, was Marvez-Valls, EG et al. The role of betamethasone in the treatment of acute exudative pharyngitis. Acad Emerg Med. 1998 Jun;5(6):567-72. PMID: 9660281. Abstract. Paper.

This was a significant paper in our field, performed in inner-city NOLA over 3 months in late 1995. Inclusion criteria was a sore throat complaint, with or without odynophagia, dysphagia, fever, cervical LAD. All you needed was pain and with large tonsils or exudate. There was no minimum pain score, but exclusion criteria included AIDS, abscess, thrush, steroids in past 3 months, diabetes, ulcerative phayngitis, not having exudates, prior allergy to steroids, inability to consent, pregnant, and not having a phone.

The authors used a 10 cm VAS with left = 0 cm = no pain. The scale hidden later during phone calls.

Patients marked their pain, then were randomized to benzathine penicillin 1.2 million IM if over 90 lbs, 0.9 million if under 90 lbs, or if allergic, polyenteric-coated erythromycin 500 mg bid x 10 days, plus 2 ml of saline or 2 mL of betamethasone for IM injection. A nurse prepared antibiotics or placebo, so the doctor and patient blinded but nurse administered. Patients were called at 24, 48 hours and more if necessary, daily for 7 days — they were asked to rate pain over the phone using the same scale, reminded of prior scores if no scale available. Callers were blinded. Side effects, missed school and work were also noted.
No pain meds were given in study, but APAP and ibuprofen were recommended at doc’s discretion. No one asked about pain med use or recorded it.

Cultures of obvious swabs were obtained and GABHS was reported. They determined a group of 50 subjects would have sufficient power to detect a 12 mm pain score difference.

They ended up with 46 patients in both groups. Initial pain in the betamethasone group was 7.5 cm, and in control, 7.9 cm. Age, gender, intervals to follow-ups, antibiotics, and allergies were all sufficiently similar. On the phone calls, 50% of patients had to be reminded of scores. Mean pain scores in betamethasone group were lower, significantly, in the first and second followup. Decrease in pain score was greater, significantly, in the first follow up in the betamethasone group. Number of hours to ‘no pain’ was significantly lower in betamethasone group. Mean decreases in pain at second followup, and number of days of school or work missed with not significantly different.

84% of patients had cultures, 49/77 were positive — 20 in the betamethasone group, 29 in placebo. Is that significant? I bet yes but it’s not reported. Of those 49 with positive cultures, 31 were GABHS, 22 of which were in placebo and 9 in betamethasone group. So 9 out of 46 patients in beta group had GABHS, wheras 22 of 46 in placebo group had GABHS.

Those with negative cultures showed no difference in pain scores with betamethasone! Culture positive patients with betamethasone had significant reduction of pain score at first and second f/u, and significant reductions in pain.

The #1 complication was pen injection site pain. One patient in betamethasone group returned after 48 hours, because of continued sore throat, got erythromycin, and had symptoms resolve after 124 hours.

To their credit, the authors note some limitations — more patients would have meant better subgroup analysis. 15 patients out of 96 had no cultures, though there was no difference in initial pain scores for those cultured vs. those not cultured. Those that ended up culture positive who got betamethasone had improved pain scores, but not for GABHS, which they believe was too small a group to detect significant change. Could steroids be more helpful with particular strains? Maybe, but this study couldn’t tell you.

It’s regrettable the injection site not assessed, and the risk of steroids maybe exacerbating pharyngitis was not covered, though we use higher doses of steroids with asthma.

Lack of quantifying analgesics was unfortunate, but they reason this should favor the null hypothesis, the hypothesis that there’s no difference – patients not getting relief would ad-lib treatment with Tylenol, and show betamethasone as less effective. So the fact that there’s a signficant difference in time to pain relief was even more remarkable.

• Putting this paper in a wider context: Studies on dexamethasone before and after this one also show a reduction of time to sore throat pain relief, compared to placebo. Dexamethasone is structurally similar to betamethasone, though their differences clinically have best been studied with regard to fetal lung maturation. There’s not enough patients to really assess safety but because of this paper and others (by Falk 1993, Wei 2002 and Tasar 2008), I ask pharyngitis patients if rapid pain relief is a priority — if it is, I give steroids.

Next up was, for the EM-3 class,Griffith et al. Metoclopramide Versus Hydromorphone for the Emergency Department Treatment of Migraine Headache. J Pain. 2008 Jan;9(1):88-94. PMID: 17981511. Abstract. Paper.

This study was motivated by the 28 million Americans — about 10% or more of the population — reporting migraines, and 15% seek ED evaluation each year. It’s a complex disease and there’s no consensus for ED therapy. A shortage of prochlorperazine (compazine) and the blackbox warning on droperidol have ED folks seeking new drugs, thus, the rise of raglan and dilaudid — metoclopramide and hydromorphone are now first-line in many places.

The authors used a retrospective cohort study to compare two drugs for initial treatment of migraine. This was performed in a single large university urban teaching hospital, exempt from patient consent. A cohort created from all consecutive ED patients over age 17 with ICD-9 dx of migraine over six months, over the winter of 2002-3. Subjects were identified if they had old criteria for migraine or if they met ICHD diagnostic criteria, first edition (second edition was not yet out). They were excluded if they had fever, nonmigraine etiology, meningeal signs, AMS, focal neuro deficit, recent trauma, pregnancy, breast-feeding.

At this hospital (Northwestern), there was no ED migraine protocol – the practice is, a doc gives a monotherapy drug, nurse assesses need for ‘rescue.’ Pain scores were self-reported, 0-10, just like we do at Sinai — with scores taken during triage, after pain med administration, and before discharge.

The primary outcome of this study was the difference of self-reported pain immediately before and after meds. A drop of 3 points was considered clinically effective and was used to dichotomize scores (this 3-point drop is 2x what they consider the minimum clinically important pain reduction) on a visual analog scale. Secondary outcomes included rescue medication needed (as judged by attending), adverse reaction, and time to discharge.
They got 200 subjects, age 18-79 (mean 36) … 86.5% female. 67% white, 25% black, 8% other. Of the 51 subjects that got hydromorphone, 94.1% got it IV and 5.9% got it IM, with most getting 1mg but a range of 0.5mg to 4 mg. All raglan patients got it IV – 37 got 10mg and 58 got 20mg (total = 95, 47.5%).
No significant age, gender differences among the three groups, but a race difference: whites were more likely to get dilaudid. Pain was slightly, nonsignificantly higher in the dilaudid group at triage (9 vs. 8.5) and time to triage was slightly better 93 minutes +/- 62.8 vs. 100.7 +/- 81.

Mean pain score after initial treatment – dilaudid went from 9 –> 6.8, reglan from 8.5 –> 4.6, others went from 7.8 –> 4.8. Pain score at discharge was 3.4 vs. 2.6 vs. 2.5, respectively, and time to discharge was 100 min vs. 57 vs. 79. There was less need for rescue in the reglan group.
The relative risk of 3+ pain reduction with reglan was 1.76 (95% CI 1.12-2.75) and after adjusting for confounders, RR = 1.6 (95%CI = 0.84-3.03, not significant)
No difference in pain reduction between 10 mg and 20mg of reglan or between 0.5mg – 4 mg doses of dilaudid, though they weren’te powered to discern a difference within groups.

The authors conclude that metoclopramide (Reglan) seems like a safe and effective and inexpensive migraine treatment, and as a bonus, it also handles nausea and vomiting. Most people use 10mg, but up to 1mg/kg is safe and larger doses should be considered. In contrast, dilauded may cause nausea and vomit, and narcotics may not help migraines and may lead to abuse. Some still do it first-line and there’s no consensus.

Limitations of this paper include study selection bias of attending diagnosis of migraine – some probably didn’t have confirmed migraines as judged by neurologists, so they applied the byzantine Headache Classification scheme. Also, because they used ED comers complaining of migraine, they got a more representative cohort than those qualifying for an RCT.

They didn’t look at who came back with pain. Also, the doc’s choice of meds and doses may have been related to severity, presence of vomiting. Other confounders could exist, though Poisson regression analysis should stop that. And indeed, because the RR is similar after confounders measured, there’s probably not much else confounding this.

• Putting this in a wider context: Like Chekhov said, when there’s a lot of therapies for something, it’s a good bet none of them work too well. We have a lot of therapies for migraine, but this and subsequent studies support the use of metoclopramide as first-line ED therapy. Other measures, like magnesium and just IV fluids, as well as opiates, seem to work to various extents, but not as quickly or as well. Fear of side effects is outweighed by efficacy, and the antiemetic potential for reglan is actually a plus.

Our final paper, for the seniors, is a review by Fitch RW, Kuhn JE. Intraarticular lidocaine versus intravenous procedural sedation with narcotics and benzodiazepines for reduction of the dislocated shoulder: a systematic review. Acad Emerg Med. 2008 Aug;15(8):757-8. PMID: 18783486. Abstract. Paper.

One of the papers reviewed was based at Elmhurst, but our practice has reverted from IAL to procedural sedation, usually opiates and benzos. This takes time to prepare and recover from, consumes resources, and risks respiratory depression. Other agents have faster onset but risk deeper sedation, myoclonus. Worst of all, none of this may not relieve pain any better than IAL.

I was curious just how this is even done — it sounds kind of easy. You direct your needle medially toward the glenoid cavity, entering laterally or posterolaterally, 2 cm below the acromion. The dislocation should cause a palpable gap in which to enter. Then you squirt 20 mL of 1% lidocaine — at least, that’s what every study here did.

IAL was first described in 1991 – Lippit showed higher success rate, lower complications. But standards haven’t changed. These authors did a lit search, and found 415 potential manuscripts, limited to humans and RCT and meta-analysis and reviews –> 60 papers. Then looking just for RCT of IAL vs. IV sedation got them to this review of six relevant papers, which they summarize, looking for success of reduction, time of procedure, and straightfoward complications.

Every paper reviewed used the same 20 mL of 1% lido dose, though different studies used different reduction techniques, populations, and ways of reporting results.  There was no significant difference in success, or pain perception, but significantly fewer complications with IAL. Time to sedation was slower with IAL but time to dispo was quicker.

The authors conclude that all the studies they looked at were small and therefore underpowered. Larger studies are therefore needed before they can make specific recommendations. But they do venture to say that where intravenous analgesia and sedation needs to be avoided, intra-articular lidocaine should be the analgesic method of choice for reducing shoulder dislocations.

While the authors address the potential for lidocaine toxicity (the toxic dose of lidocaine without epi is 7 mg / kg), they have no way of considering post-injection complications like infection. I’d also note that patient satisfaction and pain scores are problematic when dealing with amnestic agents, making comparisons tricky. Still, what we know of the windup phenomenon suggests local anesthetic is a good thing, and I look forward to the RCT between a fast-acting sedative vs. IAL.

Posted on Thursday, April 2nd, 2009 at 4:18 am by Nick. Filed under Pain Management, Headache, Journal Club, Blog.
You may post a comment.