Monday, 19 November 2018

Manual vs. Mechanical CPR; which is better?

Image result for lucas device

There are theoretic advantages to a mechanical chest compression device (i.e. Lucas device). They demonstrate improved blood pressure and coronary perfusion pressure compared to regular CPR. In addition, the earlier observational studies looked promising.

However, there is a theoretical downside. Paradoxically, CPR may be interrupted more when fumbling to fit the device. Or even worse, they may increase time to defibrillation.

So, which is better?

This meta-analysis gathered the available data from out-of-hospital cardiac. Survival to hospital admission was the primary outcome of interest. They also looked at any ROSC, survival to hospital discharge and favourable neurologic outcome at discharge.

The authors found 20 studies of which 5 were RCT’s. In total, there were 21,363 patients included with just over half from the trials. The observational studies and the clinical trials were pooled and analysed separately.

For survival to admission, the RCT’s showed no advantage to the mechanical devices. Whereas the lower quality observational data demonstrated a benefit in favour of mechanical CPR.

How could this be?


In observational studies, patients are not randomised to a treatment allocation. As such, one group is usually “sicker” than the other. Perhaps the machine was utilised more when the paramedics thought the patient had a greater chance of survival (i.e. younger, witnessed arrest, bystander CPR, shockable rhythm)?

Researchers try to correct for this by doing some fancy statistics; propensity scores, stratified analysis, multivariate adjustment, logistic regression, etc. But these techniques are far from perfect. Furthermore, there are always the unknown and unmeasured confounders.

This may be irrelevant anyway.  The most important outcome, “favourable neurologic outcome at discharge” was no different in both the RCT’s and observational meta-analysis.

What’s the take home?

In all comers, it looks like there is no advantage to a mechanical compression device.

But there remains some “what if” questions.

What if we educated staff in how to fit the device very quickly while ensuring rapid defibrillation? What if it’s an overdose patient that needs prolonged good quality CPR? What if it's a bridge to the cath lab? What if the technology improves? What if we use the device as a cocktail shaker?

That’s a lot of what if’s…


Bonnes JL, Brouwer MA, Navarese EP, et al. Manual Cardiopulmonary Resuscitation Versus CPR Including a Mechanical Chest Compression Device in Out-of-Hospital Cardiac Arrest: A Comprehensive Meta-analysis From Randomized and Observational Studies. Ann Emerg Med. 2016;67:349-360. [Link]

Sunday, 18 November 2018

Battlefield acupuncture to treat low back pain in the ED; witchcraft in action

Acupuncture has been around for thousands of years… so it must be good.

We all dread treating the patient with low back pain. We don’t have a magic bullet that will provide a fix. It’s frustrating for the patient and provider.

As such, maybe we should engage in witchcraft! Perhaps something that “has been around for thousands of years.”

This was a “feasibility study” enrolling 30 patients with low back pain in New York City. They were randomised to either standard care alone vs. standard care plus “battlefield acupuncture.” Evidently, this is ear acupuncture that has been used by the US military… so, it must be good.

Primary outcome was the 10 point numerical rating scale (NRS) and the “get-up-and-go” test (GUGT). The GUGT is how much time it takes a patient to get out of a chair, walk three meters, turn around and sit back down again.


Abracadabra! The post-intervention NRS in the acupuncture group was better at 5.2 vs. 6.9 (p=0.04). The GUGT showed no difference.

But wait a minute…

This study was listed at This is where original study protocols are published. All too often researchers change the goal posts mid-study or suppress negative results. (My all-time favourite example was gabapentin for off-label use.)

A search of registration number NCT02399969 shows the GUGT as the sole primary outcome. The NRS was a secondary outcome. Research misconduct anyone? One would hope the peer review process would pick up on this… it’s not hard! At the very least, the manuscript should mention why they changed their primary outcome to give them favourable results.

But there’s even more than an allegation of research misconduct.
  • Acupuncture was not blinded. Not even a sham acupuncture arm. This introduces bias that favoured acupuncture.
  • Placebo works; in particular for pain and nausea. Acupuncture is theatrical placebo at it’s best. All the results in the NRS can easily be explained by placebo effect. The GUGT is a more objective measure and the trend favoured standard care alone.
  • A quarter of patients approached for enrolment declined to participate. Did they screen out the patients who were acupuncture non-believers?
  • There’s more… but for brevity, I’ll stop now…

What are we to conclude?

Please stop the nonsense.

Acupuncture has been around for a long time. Currently we have no consistent good quality evidence to demonstrate efficacy beyond placebo effect.

No more study is needed.  


Fox LF, Murakami M, Danesh H, et al. Battlefield acupuncture to treat low back pain in the emergency department. Am J Emerg Med. 2018;36:1045-1048. [link]
Image result for witchcraft

Wednesday, 14 November 2018

Supraglottic airway vs. endotracheal tube for out-of-hospital cardiac arrest; which is better?

This study was a large cluster-crossover clinical trial in the USA attempting to answer this research question. 27 EMS agencies were randomized in clusters to either a laryngeal tube or endotracheal tube with crossover to the alternate strategy at 3 to 5 month intervals.

Why randomize EMS agencies? Wouldn’t it be better to randomise individual patients?

Think feasibility.

Research in cardiac arrest is challenging.  The last thing paramedics want to do during a code is open a study envelope. They’ve got other things on their minds. Hence randomise EMS agencies. Of course this can introduce bias and statistical inefficiency. But such is compromise in research.

The primary outcome was 72-hour survival. They did report the much more important outcome of favourable neurologic status (mRS<3) at discharge from hospital.


3000 patients were enrolled. About half were unwitnessed and/or had no bystander CPR. 80% had non-shockable rhythms. It’s no surprise overall outcomes were terrible. It is also challenging to demonstrate a difference in treatment strategies when the vast majority were going to die regardless of intervention. But such is life… and death.

Rates of initial airway success were 90% with the laryngeal tube but a pretty dismal 50% with the endotracheal tube. Yikes!

72-hour survival was 18% with the laryngeal tube and 15% with the endotracheal tube. Rates of favourable neurologic outcome was 7% vs. 5% favouring the laryngeal tube.

What are we to think?

Unfortunately, this study did have some problems. It had poor rates of first pass success with intubation. The pragmatic study design and cluster randomisation introduced noise. They only studied one type of supraglottic airway… i.e. not the LMA.

But in the end, there probably is no compelling reason to push for endotracheal intubation in the field. This may distract from other meaningful interventions such as good quality CPR and rapid defibrillation.

It’s unclear how much of this is externally valid to a well-resourced Emergency Department. A dedicated airway doctor and lots of hands might change our outlook.  Nevertheless, this study does push our focus away from the endotracheal tube in cardiac arrest.

Image result for laryngeal tube airway 

Wang, HE, Schmicker RH, Daya MR, et al. Effect of a Strategy of Initial Laryngeal Tube Insertion vs. Endotracheal Intubation on 72-Hour Survival in Adults with Out-of-Hospital Cardiac Arrest. JAMA 2018;320(8):769-778. [link to abstract]

Saturday, 22 September 2018

Give honey for button battery ingestion? The tale of two anesthetised pigs…

Image result for american yorkshire piglets

Believe it or not, UpToDate recommends honey as first aid for button battery ingestions.

Perhaps we should look at the evidence.

We know button batteries can cause nasty and sometimes lethal caustic oesophageal injury.

This study was a cadaver and live American Yorkshire piglet model looking at potential pH neutralizing agents prior to definitive endoscopy.

In vitro, they tested the final tissue pH of cadaver oesophagus after installation of various products. They included different kinds of honey, Carafate, apple juice, orange juice, Powerade, Gatorade, maple syrup, simulated saliva and 0.9% sodium chloride control.

In the lab, the honey and Carafate seemed to work the best and underwent further study.

Bring on the live pigs!

9 anaesthetised pigs had a button battery placed in their oesophagus for an hour. (Poor creatures.) Serial irrigations of study solution occurred every ten minutes starting at the five-minute mark.

2 pigs got honey, 3 got Carafate, and 4 got saline control.

On day 7, the unfortunate piglets were euthanized and histology obtained.


Honey was the big winner! They had much less depth of injury. Half  of the saline piglets had delayed oesophageal rupture.

This little piggy went to market… to get honey! Give honey for button battery ingestion.


This recommendation is based on TWO anesthetised pigs that got honey.

This is extremely low-quality evidence.

One should only change practice based on such low-quality evidence if the treatment or therapy was considered to be extremely low risk.

Perhaps honey is extremely low risk. But are we certain?

Would care get delayed in finding and administering honey? Does honey impair endoscopy? What about aspiration or honey induced mediastinitis? Will kids get nasal-oesophageal tubes to administer honey? Would we tend to delay endoscopy because they "got the honey?” 

Perhaps risk may be minimal… But changing practice based on two honey glazed pigs??

(ok… I’d give the honey too… I can’t believe I just said that…)


Anfang RR, Jatana KR, Linn RL, et al. pH-Neutralizing Esophageal Irritations as a Novel Mitigation Strategy for Button Battery Injury. Laryngoscope 2018 Jun 11. doi: 10.1002/lary.27312. [Link to article]

Friday, 21 September 2018

Clinical Trial of Fluid Infusion Rates for Pediatric DKA

Dr Annoying Bowtie, RACP

Emergency physicians often get criticised for overhydrating children with DKA. “Too much fluids will cause their heads to explode,” as spoken from an annoying bow tie.

But there has never been good quality evidence to prove fluid administration causes cerebral edema. In fact, a study by Glaser etal. in 2001 showed no association between fluids and bad outcomes.

But as there is no definitive RCT, the debate still rages!

Hopefully this clinical trial will put matters to rest.

After 17 years, Dr Nicole Glaser is back! Now teamed up with the PECARN mega-group, they randomised 1255 children with 1389 DKA episodes to one of 4 treatment arms (please see original paper for further details)

  • Fast administration of 0.9% normal saline
  • Fast administration of 0.45% normal saline
  • Slow administration of 0.9% normal saline
  • Slow administration of 0.45% normal saline

(Just to give an idea of what is fast vs. slow, a 50 kg child would get an infusion started of 300ml/hour vs. 140ml/hour.)

The primary outcome was a decline in mental status (measured by GCS) during treatment. Secondary outcomes included clinically apparent brain injury during treatment and some measures of memory.


No difference. Neither the rate of administration nor the sodium chloride content of the IV fluids resulted in bad outcomes. 

So the debate is over?

Unfortunately, no.

98% of the children recruited in this study started out at GCS 14-15.  We know it is the sickest cohort of DKA that develop bad complications. This study was never going to show a difference. The clear majority were going to get better with any reasonable management... regardless of fluid protocols.

What about the child with a pH of 6.9 and GCS 6? Does the rate of fluid administration matter?


Unfortunately, an adequately powered RCT of severe DKA will never happen. It is simply not feasible. 

Alas, the annoying bow ties will continue...


Kupperman N, Ghetti S, Schunk J, et al. Clinical Trial of Fluid Infusion Rates for Pediatric Diabetic Ketoacidosis. N Engl J Med 2018;378:2275-87. [link to original full text article]

Thursday, 16 August 2018

Change the guidelines! An RCT of Epinephrine in Out-of-Hospital Cardiac Arrest; the PARAMEDIC2 Trial

Should we still be giving adrenalin in cardiac arrest?

Despite decades of tradition and guidelines, there has never been any good evidence showing adrenalin improves meaningful outcomes.

The PARAMEDIC2 trial was a very well conducted RCT of 8014 patients in the UK with out-of-hospital cardiac arrest. Patients were properly randomised to getting either adrenalin 1mg q3-5 minutes or matching placebo.

Primary outcome in this study was rate of survival at 30 days. But we all know the real outcome of interest is hospital discharge with a favourable neurologic outcome.


The adrenalin group had more patients with ROSC, transport to hospital and ICU resources. Rate of survival at 30 days was marginally better at 3.2% vs. 2.4%. (NNT 112)

Yes, the rates of survival were abysmal.  But remember the patients included in the study did not respond to initial resuscitative efforts (CPR and defibrillation). They were only enrolled when they got down the pathway requiring adrenalin.

But the real outcome of interest; neurologically intact survival?

No difference.

In the end, adrenalin increases resource utilization and increases survival of the neurologically devastated.

Sounds pretty bad to me. Sounds pretty bad to the consumer groups too.

Will this change the guidelines?

Old habits die hard. But there will never be a trial more definitive than this one. It would have been unethical to start this trial if the authorities (in the UK at least) were not prepared to change practice.

But adrenalin may be worse. Ask yourself; does giving a worthless treatment distract the focus from more important resuscitative efforts? This study didn't address this question but I think I know the answer.

Yes, change the guidelines!


Paper critiqued at Emergency Tasmania 2018. Special thanks to Dr Mark Reeves, FANZCA and audience for feedback.

Image result for adrenalin vial

Tuesday, 14 August 2018

Fake News: Intraosseous vascular access is associated with lower survival and neurologic recovery among patients with out-of-hospital cardiac arrest

Prior observational studies have challenged the widely held belief that IO access is just as good as IV. This paper further explored the association between IO access and poor outcomes compared to IV.

The authors performed a secondary analysis of a large trial of patients with out-of-hospital cardiac arrest. They were able to identify patients that got IV vs. IO as their means of access.

They excluded patients who had any failed attempts at either route of vascular access or who had both performed.

The primary outcome was favourable neurologic outcome on hospital discharge (mRS <3 i.e. patients could at least walk without assistance).


13,155 were included in the analysis. 5% had IO access and 95% had IV (obviously patients were not randomised to this intervention)

For all outcomes, the IV access was far superior. 7.6% of patients with IV access had favourable neurologic outcome vs. 1.5% with IO.

This is astounding! A huge difference… let’s throw away the IO’s!!!

But there is no way this is true.

No amount of fancy statistics can make up for systematic bias and residual confounding that is present in this cohort study. There is clearly a reason why paramedics went directly to IO rather than IV. The IO group was obviously a “sicker” cohort to start with and had worse outcomes.

Not all cardiac arrest is the same. We know those with witnessed arrest, bystander CPR and shockable rhythms do better. There are also many other factors that influence outcomes.

Capturing all potential confounders in a resuscitation is problematic. When under duress, data is hard to measure accurately. In addition, there are always the unknown confounders that go unmeasured and unadjusted. No statistics can truly fix bad or absent data.

Futhermore, what was the magic medicine they used to account for this five fold increase in favorable outcomes? No drugs have really been shown to work in ALS. 

In the same context of “fake news” sometimes bad evidence is worse that no evidence at all. This study has the very unfortunate possibility to mislead and cause harm. At best, it is hypothesis generating for future prospective study.

For now, reject the fake news and keep drilling your IO’s. 

 Image result for ez-io
Paper critiqued at Emergency Tasmania 2018. Special thanks to Dr Mark Reeves, FANZCA and audience for feedback.