Research Roundup – May 2026

 I think there are some very interesting papers in here this month. There are a couple big RCTs, but if you aren’t in the mood for EBM right now, I suggest at least having a look at the final paper, and imagine yourself volunteering to participate.

A whole lot of nothing

Smith JE, Cardigan R, Sanderson E, Silsby L, Rourke C, Barnard EBG, Basham P, Antonacci G, Charlewood R, Dallas N, Davies J, Goodwin E, Hawton A, Hudson C, Lucas J, Keen K, Lyon RM, Nolan B, Perkins GD, Rundell V, Smith L, Stanworth SJ, Weaver A, Woolley T, Green L; SWiFT Trial Group. Prehospital Whole Blood in Traumatic Hemorrhage – a Randomized Controlled Trial. N Engl J Med. 2026 Mar 17. doi: 10.1056/NEJMoa2516043. Epub ahead of print. PMID: 41841706

This is the first large RCT looking at the use of whole blood in trauma. My pretest probability that whole blood will help is very low. The idea that whole blood might be beneficial rests entirely on the idea that balanced transfusion is a good idea, which, at this point, is a completely unproven theory. (There are only 2 RCTs – one was negative, and mortality was 18% higher with balanced transfusion in the other.) This trial randomized 942 prehospital trauma patients who required transfusion according to the treating physician to either 2 units of whole blood or 2 units each of red blood cells and plasma. There was no difference in the composite of mortality and need for massive transfusion within 24 hours (48.7% vs 47.7%). The individual aspects of the composite were also pretty clearly unchanged. Based on my skepticism around whole blood, I tend to believe these results, but really I don’t think this trial moves the needle very much. I don’t think anyone believes that you are going to see clinical differences with just 2 units of blood. (Although, I guess someone believes it, otherwise this study wouldn’t have been run.) The 2 prehospital units were quickly swamped by identical blood products in hospital for anyone who truly needed massive transfusion, meaning that the treatment both groups received was almost identical, and so of course there were no differences between the groups. I look forward to seeing more RCTs on whole blood, but at this point it is pretty clear that this is an experimental therapy, not one that should be used routinely.

Bottom line: This first RCT of whole blood in trauma showed no benefit over component therapy. We definitely need, and will see, many more trials of whole blood in the future.

You can read more here. 

Never seen a number we didn’t want to treat

Sajdeya R, Yanez ND, Kampp M, Goodman MD, Zonies D, Togioka B, Nunn A, Winfield RD, Martin ND, Kohli A, Huynh TT, Okonkwo DO, Poblete RA, Gilmore EJ, Chesnut RM, Bunnell AE, Ohnuma T, Hashemaghaie M, Treggiari MM. Early Blood Pressure Targets in Acute Spinal Cord Injury: A Randomized Clinical Trial. JAMA Netw Open. 2025 Sep 2;8(9):e2525364. doi: 10.1001/jamanetworkopen.2025.25364. PMID: 40965887

Guidelines have always said that we should target higher blood pressures in spinal cord injuries. This has become more important over the last decade, when the concept of permissive hypotension has taken hold for most other trauma patients. However, that recommendation has always been based on flawed observational data. This is the first ever RCT, and they randomized 92 adult patients with acute spinal cord injuries to MAP targets of either 65-70 or 85-90. The headline news is that there were no statistical differences in neurologic outcomes at 6 months, but higher blood pressure targets caused more respiratory adverse events. The trial has some very significant flaws. It was stopped early. They have outcome data on less than half of their patients. It is only partially blinded. This should not be taken as the final answer. We clearly need more research, and that research could easily go in either direction. However, this is the only RCT in existence, and it shows harm with increased blood pressure targets, so for now it seems pretty clear that patients with spinal cord injuries should be treated the same as all other trauma patients. 

Bottom line: This small, flawed RCT demonstrated no benefit from higher blood pressure targets (MAP 85-90) when compared to conventional targets (MAP 65-70), but does demonstrate harm. There is tremendous uncertainty, but we should clearly favour conventional BP targets until new research is published.

Full details can be found here.

Two looks at interventional therapy for PE

Lookstein RA, Konstantinides SV, Weinberg I, Dohad SY, Rosol Z, Kopeć G, Moriarty JM, Parikh SA, Holden A, Channick RN, McDonald B, Nagarsheth KH, Yamada K, Rosovsky RP; STORM-PE Trial Investigators. Randomized Controlled Trial of Mechanical Thrombectomy With Anticoagulation Versus Anticoagulation Alone for Acute Intermediate-High Risk Pulmonary Embolism: Primary Outcomes From the STORM-PE Trial. Circulation. 2026 Jan 6;153(1):21-34. doi: 10.1161/CIRCULATIONAHA.125.077232. Epub 2025 Nov 3. PMID: 41183181

People love PE teams, and the fancy equipment they might bring to the bedside, but I have not seen any evidence that these fancy interventions actually help patients. This is a small RCT comparing a computer assisted thrombectomy device to standard care (heparin) in 100 adult patients with intermediate risk PEs. Their primary outcome was just the RV/LV ratio, and there was a statistical difference, but this is a pretty meaningless disease oriented outcome. There do not appear to be any differences in clinical outcomes. While not anywhere close to statistically significant, it is concerning to me that there were infinitely more deaths in the thrombectomy group (2 vs 0). In my mind, the biggest issue with this trial is the target population. These are not the patients I would be targeting. I don’t think that troponin or BNP, in an otherwise well patient, means anything. These patients generally have fantastic outcomes, and so subjecting them to an invasive procedure makes no sense. They excluded anyone with any hint of hemodynamic instability, which is the group of patients where these interventions actually make sense. The fact that there were 0 deaths in the control group basically proves that the intervention could not possibly help in this population. Also, this trial design does nothing to prove that this expensive computer assisted cather provides any value, given that it wasn’t compared to a standard catheter, but given that the trial was designed, run, interpreted, and written up by the company that makes the catheter, you wouldn’t really expect a fair test.

Bottom line: This is basically a pilot trial, looking at disease oriented rather than patient oriented outcomes. It is an unblinded trial, designed and run by the device manufacturer. Even with those biases, the results are completely underwhelming. No one should be using this device at this time. Overall, I don’t think there is any evidence that interventional approaches to PE actually help patients at this point.

Kjaergaard J, Bang LE, Sonne-Holm E, Wiberg S, Holmvang L, Lassen JF, Sørensen R, Høfsten DE, Ulriksen PS, Jawad S, Palm P, Søe C, Ersbøll MK, Boesgaard S, Møller JE, Thune JJ, Hassager C, Tilsted HH, Lønborg J, Egstrup M, Kristiansen OP, Seven E, Lindholm MG, Eskesen K, Fanø S, Carlsen J. Randomized trial of low-dose -, ultrasound assisted thrombolysis or heparin for pulmonary embolism. Cardiovasc Res. 2026 Jan 29:cvag038. doi: 10.1093/cvr/cvag038. PMID: 41610160

This is another RCT of interventional options for intermediate-high risk PE patients. This time they are comparing ultrasound assisted catheter directed thrombolysis to intravenous thrombolysis (both using alteplase 20mg over 6 hours) to heparin alone. Again, they use a disease oriented outcome. Both thrombolysis options reduced clot burden as compared to heparin alone, but the ultrasound assisted catheter device provided no value over simple intravenous alteplase. Clinical outcomes at 3 months were not improved, and, once again, all the deaths were in the thrombolysis groups. Like the last trial, the big problem here is that they were looking at the wrong population. Mortality was 0% with heparin alone. You can’t improve on that. Why are you adding a high risk intervention to a population with no risk of death? Despite the limitations, I do think it is important to note that, when used at the same doses, thrombolysis is equally effective when given IV and through a specialized catheter. If you are going to use thrombolytics, there is no need to use the more expensive, more invasive options. IV is good enough, and low and slow dosing makes sense. However, we clearly should not be using them in the population defined here.

Bottom line: Although thrombolysis might decrease clot burden, there is evidence of harm here, and so lytics should not be used in this population. For patients requiring thrombolysis, a peripheral IV should be as effective as the more expensive invasive approaches, and therefore should be preferred. 

More details on these two papers can be found in this post. 

Swapping meth for mirtazapine?

McKetin R, Shoptaw S, Saunders L, Nguyen L, Clare PJ, Dore GJ, Turner A, Dean OM, Kelly PJ, Arunogiri S, Koeijers J, Degan TJ, Degenhardt L, Farrell M, Goodman-Meza D, Sinclair B, Reid D, Cordaro F, Hill H, Lundin R, Hayllar J, Christmass M, Liaw W, Liu D, Woods A, Brewerton B, Holyoak E, Wu BT, Maher H, O’Dea N, Keygan J, Kontogiannis A, Palmer L, Morrison C, Wrobel A, Hyland B, Kiden G, Romeo V, Kyaw KWY, Byrne M, Colledge-Frisby S, Zahra E, Berk M. Mirtazapine for Methamphetamine Use Disorder: A Randomized Clinical Trial. JAMA Psychiatry. 2026 Apr 1:e260159. doi: 10.1001/jamapsychiatry.2026.0159. PMID: 41920558

Although it is far from perfect, the fact that we have a placebo-controlled, double-blind RCT showing a reduction in methamphetamine use in patients with methamphetamine use disorder definitely caught my eye. They randomized adult patients with methamphetamine use disorder and active drug use to either mirtazapine 30 mg daily or matching placebo, and there was a statistically significant reduction in self-reported days of methamphetamine use over the next 28 days. (There were 7 fewer days of use with mirtazapine versus 5 with placebo.) None of the secondary outcomes were positive, and there was no difference in saliva drug testing, which makes me wonder if the blinding was broken and self-report ended up biased. There were somewhat more adverse events in the mirtazapine group (mostly drowsiness and weight gain). Patients were recruited via flyers and social media, and so were a select group of patients interested in getting addictions help, and not necessarily the patients we see in the emergency department. That being said, given the lack of other options available for methamphetamine use, this is at least worth discussing with patients who are seeking help.

Bottom line: In this double-blind placebo-controlled RCT, mirtazapine resulted in less self-reported days using methamphetamine in the next 28 days. The benefit was modest, and there were adverse events, so this is not a slam dunk game changer, but it is interesting and worth discussing with patients as we await more data. 

Radiologists should not gate-keep studies for patients they have not assessed clinically

Tan H, Chandru P. Radiologist Approval for CTPA in the ED: A Single-Centre Pilot Study Did Not Demonstrate a Statistically Significant Difference in Diagnostic Yield. Emerg Med Australas. 2026 Feb;38(1):e70237. doi: 10.1111/1742-6723.70237. PMID: 41705429

Thankfully, the practice is dying, but asking radiologists for permission to order a scan has to be one of the dumbest concepts in medicine. Why are you asking someone who has not assessed the patient for permission to scan? Advice on the best protocol – of course. But permission? That is insulting. It is a waste of time. It is outright illogical. This hospital moved from a system where radiology trainees vetted all CT requests to a temporary suspension of radiologist approval for CT imaging. They retrospectively report on the yield of CT pulmonary embolism. With radiology approval, the yield was 11.4%. Without, the yield was 10.2%. This was not statistically significant, with a sample size of 953 total CTPAs (p=0.6). Really, this study doesn’t say much, given that prior to this intervention “refusals were uncommon.. Informally estimated to be near zero.” If you weren’t refusing any scans (which is the appropriate approach when you aren’t assessing the patient), you can’t possibly change anything with this ‘practice change’. However, that doesn’t make this data pointless. They don’t include any information on the time that was spent reviewing protocols in the past, but if you are never going to refuse a scan, why waste the time? We want our radiologists acting as radiologists, not gate keepers.

Bottom line: Clinicians caring for patients should be the only ones deciding whether to order tests. Radiologists should be available to provide advice about the best tests to use, but should never be making yes/no decisions about patients they have not seen. 

Non-inferiority trials are dumb

Luckey A, Balasegaram M, Barbee LA, Batteiger TA, Broadhurst H, Cohen SE, Delany-Moretlwe S, de Vries HJC, Dionne JA, Gill K, Kenyon C, Kittiyaowamarn R, Lewis D, Mueller JP, Naicker V, O’Brien S, O’Donnell JP, Phanuphak N, Spooner E, Srinivasan S, Taylor SN, Unemo M, Zwane Z, Hook EW 3rd; Zoliflodacin Phase 3 Study Group. Zoliflodacin versus ceftriaxone plus azithromycin for treatment of uncomplicated urogenital gonorrhoea: an international, randomised, controlled, open-label, phase 3, non-inferiority clinical trial. Lancet. 2026 Jan 10;407(10524):147-160. doi: 10.1016/S0140-6736(25)01953-1. Epub 2025 Dec 11. PMID: 41391465

Not a month goes by without me complaining about non-interiority trials. They have their defenders – looking at you Dr Orkin – but they are dramatically over-used, and the vast majority that I see are clearly inappropriate. My bias against non-inferiority trials is why this study had to be included despite the fact that the antibiotic of interest – zoliflodacin – is not actually available in Canada. Gonorrhea has developed resistance to almost all options used to treat it aside from ceftriaxone. Zoliflodacin is a novel oral antibiotic that is effective against gonorrhea, which obviously could be helpful. This is a large multinational open-label trial comparing zoliflodacin (3 grams PO) to ceftriaxone (500mg IM) plus azithromycin (1 gram PO) in the outpatient clinic setting. They included 930 patients and, based on their predefined non-inferiority margin, they were able to conclude that zoliflodacin is ‘non-inferior’ to the combination of ceftriaxone and azithromycin based on a test of cure at 6 days. I am not going to spend any time on the methodology, although there could be some issues, because I really just included this paper for the stats point. Despite the fact that they conclude that Zoliflodacin is ‘non-inferior’, they actually prove that it is inferior. The cure rate with ceftriaxone plus azithromycin was 96%, as compared to 91% with zoliflodacin, with an absolute risk difference of 5.3% and a 95% confidence interval of 1.4-8.6%. Because they set their non-inferiority margin at 12%, they are able to claim that zoliflodacin is “non-inferior”, but their stats actually prove that it is inferior. This is because non-inferiority trials, despite their name, do not attempt to prove non-inferiority. They just attempt to prove that the new treatment is ‘not much worse’ than the old treatment. Not much worse is not a great target in medicine, especially when the researchers just get to make up the threshold to define ‘not much worse’. ‘Not much worse’ is exactly why industry is leaning so heavily into non-inferiority designs. These trials are far easier to bias in your favour when you are trying to sell ineffective – or even inferior – products. This trial is the perfect example of why I spend so much time ranting against non-inferiority trial designs.

Bottom line: Zoliflodacin is clearly inferior to the current standard of care, despite the very misleading conclusions in this paper. That doesn’t mean it will never play a role, as antibiotic resistance gets worse, but it clearly should not be used at this time.

The No OUCH study

Ali S, Klassen TP, Candelaria P, Bhatt M, Sawyer S, Stang A, Yaskina M, Heath A, Pechlivanoglou P, Offringa M, Drendel AL, Hickes S, Poonai N; KidsCAN PERC Innovative Pediatric Clinical Trials No OUCH Study Team. Acetaminophen (Paracetamol) or Opioid Analgesia Added to Ibuprofen for Children’s Musculoskeletal Injury: Two Randomized Clinical Trials. JAMA. 2026 Mar 10;335(10):863-873. doi: 10.1001/jama.2025.25033. PMID: 41505155

I feel like we know this already: small doses of opioid are as effective as acetaminophen and ibuprofen in most conditions, with more side effects. This paper reports on two connected RCTs that enrolled 699 pediatric patients (6-17 years) with a limb injury and numerical pain score of at least 5 at one of 6 Canadian pediatric emergency departments. The reason there are 2 separate RCTs is that they allowed parents to decide if they wanted to be included in the opioid trial, and if not, they were randomised in a 2 arm trial without opioids. (You can therefore imagine some bias in the outcomes against the opioid group here, given that parents who decided not to give their kid opioids are going to be less likely to report pain and side effects that might be their fault.) There were 3 groups: ibuprofen plus hydromorphone, ibuprofen plus acetaminophen, and ibuprofen alone. Doses were relatively reasonable (ibuporfen 10 mg/kg, acetaminophen 15 mg/kg, and hydromorphone 0.05 mg/kg). The dose of hydromorphone is probably a little bit high. I personally use morphine, where the oral dose is 0.1 to 0.2 mg/kg. That translates into about 0.015 to 0.04 mg/kg of hydromorphone, which is as much as 50% less than they used here. Despite being reasonable doses, the primary issue with this study is that the analgesic effects from both ibuprofen and acetaminophen have ceilings, whereas opioids do not, and so the medications are supposed to be used very differently. This represents the maximum analgesic effect from ibuprofen or acetaminophen, but the minimum effect or starting dose for hydromorphone. Their primary outcome was a pain score at 1 hour, and the outcomes were the same in all groups (4.8 vs 4.6 vs 4.6). There were significantly more side effects with hydromorphone (28% vs 6% vs 6%), primarily dizziness, nausea, and vomiting. I tend to believe these results, because there are many many studies that say the exact same thing, but I still worry that they will be misinterpreted. Yes, the first dose of ibuprofen is just as effective whether or not you add an opioid. It should probably be the starting point. However, opioids still play an important role. You can’t add more ibuprofen, and so if the patient remains in pain, opioids are an important (and titratable) option. In addition, a lack of effect in an average patient, where only 40% of patients had fractures, and obvious deformity was an exclusion, does not eliminate an effect in patients with more extreme pain. Use your judgement. I usually start with ibuprofen, but there is definitely a subset of patients presenting in enough pain to warrant an immediate dose of morphine.

Pericardiocentesis, drainage and instilled tranexamic acid

Qandil M, Ransom P, Shammala MA, Srour A, Khafaja M, Alkhateeb N, Jafar AJN, Abughali S. Pericardiocentesis, drainage and instilled tranexamic acid: definitive management in a 25-case series of penetrating cardiac tamponade. Injury. 2026 Feb 16:113106. doi: 10.1016/j.injury.2026.113106. Epub ahead of print. PMID: 41760498

This is an interesting case series born out of horrific conditions and necessity. Although the clear gold standard therapy for patients with penetrating chest trauma and clinical signs of tamponade is an urgent thoracotomy, these authors from Gaza discuss their experience in which there was only a single cardiac surgeon for the region, who obviously could not be available 24/7 for months on end. Pericardiocentesis by emergency physicians was originally seen as a temporizing measure, but they quickly realized that many patients were doing fine without follow-up surgery. This is a case series of 25 adult patients with penetrating chest trauma, signs of obstructive shock, and ultrasound confirmed pericardial fluid. The senior emergency physician performed a pericardiocentesis, using a 16-gauage dialysis catheter, aspirating fresh blood, and then, interestingly, instilling 1 gram (10 mL) of TXA directly into the pericardial space. 24 of the 25 patients (96%) survived until hospital discharge, with the 1 death occurring in a patient with severe brain injury. They had 3 month follow-up data on 75% of the patients, and although a couple had a reaccumulation and repeat pericardiocentesis, none died after hospital discharge, and none required follow-up surgery. These are apparently consecutive patients, which is astonishing if true. This survival rate is dramatically higher than I would have expected in patients who were hypotensive from a penetrating cardiac injury on arrival to hospital in a war scenario where no surgeon was available. Obviously, a case series has many limitations, and these authors are incredibly up front about that, referring to their data as anecdotal, but it does raise some interesting questions about how invasive our trauma management truly needs to be. I have no idea whether the TXA actually added value here, but high concentration local use (similar to use in epistaxis) is more likely to be effective than intravenous use, which seems to fail time and time again. That being said, injecting a relatively large volume of fluid back into the pericardium in the setting of known tamponade concerns me, and so this unproven TXA intervention should not be taken lightly. 

They did what?

Eisendle F, Roveri G, Rauch S, Thomassen Ø, Dal Cappello T, Assmus J, Malacrida S, Kammerer T, Schweizer J, Borasio N, Dörck V, Falk M, Falla M, Fruzzetti N, Maxenti M, Mydske S, Sasso GM, Vinetti G, Wallner B, Brattebø G, Brugger H, Strapazzon G. Respiratory Gas Shifts to Delay Asphyxiation in Critical Avalanche Burial: A Randomized Clinical Trial. JAMA. 2025 Oct 8;334(19):1720–7. doi: 10.1001/jama.2025.16837. Epub ahead of print. PMID: 41060661

This trial is incredible. It is insane, but it also has great methodology. I probably should have led with it to convince people that EBM can be fun. This is a randomized, double-blind, sham controlled trial of an avalanche safety device that uses a backpack based fan to draw air directly from the snow and deliver it to the area of your face, with the intention of providing enough air for 90 minutes of ventilation while completely submerged in snow. Participants were randomized to this device or a sham that made similar noise but didn’t deliver any air, and were buried in 50 cm of snow in a prone position, and left there for up to 60 minutes! If you were lucky enough to be randomized to the safety device – they turned it off after 35 minutes anyway. 

You were left in that position until your oxygen saturation dropped below 80%, you hit the maximum time, or you (understandably) requested that this crazy procedure stop. The primary outcome was time from burial until termination, and there is no doubt that the device worked. In the control group, 11 of 12 patients terminated before they hit the max of 35 minutes. (The fact that 1 person made it to 35 minutes buried in 50 cm of snow is crazy to me, and serves as a reminder that there is air available to breath in lightly packed snow, which is why anyone is able to survive avalanche burial.) In the intervention group, 11 of the 12 made it all the way to 35 minutes, and the only failure was because of upper extremity paresthesia rather than hypoxia. After they switched the device off at 35 minutes, the results looked identical to the control group, with people developing respiratory issues and stopping within 8 minutes. (It looks to me like carbon dioxide was a bigger issue than oxygenation, but either way, I am glad I was not involved in this trial.) I was very surprised to see that this study is not run or funded by the manufacturer of the device, but just some interested volunteers at the Institute for Mountain Emergency Medicine. Honestly, this trial had better methodology than 95% of what I review here – with a sham control, a sensitivity analysis to self-matched controls, and a trial that exactly matched their clinicaltrials.gov registration – but there are some obvious limitations to a small trial like this. Looking at healthy volunteers who knew they were being monitored and knew they were going to be rescued limits generalizability. I could imagine the device being far less effective for a person hyper-ventilating and panicking in a real avalanche scenario. Also, the NNT and overall cost would be an interesting calculation. As compared to the overall number of skiers/ people engaging in mountain sports, the number trapped in avalanches has to be very small, and among those that are, there are still going to be many who can’t be saved, either because we still can’t get to them within the 90 minutes this device provides, or because they died from traumatic injuries. How many people would have to be wearing this device at all times, and at what cost, for it to end up saving one life? That being said, I was sort of surprised it worked at all, and was fascinated by this study. Kudos to the researchers for getting it done, and to the research ethics board that was brave enough to allow this study to be performed. 

Bottom line: There is enough air in snow for a basic fan to allow respiration under 50 cm of snow for up to 35 minutes. That is interesting, but perhaps not all that useful information. 

Cheesy Joke of the Month

It has been pointed out that I forgot the cheesy jokes a few month in a row. I will try to make it up by being extra cheesy:

Did you hear about the cheese factory that exploded? All that was left was de brie.