Monday 31 July 2017

What is the purpose of log roll examination in the unconscious adult trauma patient?

I often wonder what we are trying to accomplish by performing log roll examination in patients with trauma. Will our magical fingers exclude spinal fractures?

Ok, I’ll acknowledge it clearly is useful in penetrating trauma, finding wounds, and foreign bodies. (Interestingly, the most interesting foreign body I ever found was a sawed-off shotgun. But this is another story…)

This retrospective cohort study conducted at the Alfred Hospital aimed to assess the utility of the log roll examination in unconscious (GCS <8) trauma patients for the diagnosis of soft tissue and thoracolumbar spine injury.

Examination findings as documented in the medical record were compared to the CT and/or MRI reports.

The had 402 cases that met inclusion criteria with 35% having thoracolumbar fracture. Yes, quite a banged-up cohort.

The sensitivity of log roll?


Specificity was found to be reasonable at 91%. The log-roll also found soft tissue injury, open wounds, burns and glass but no firearms.

Of course, there are several issues with this paper. The retrospective data charted in the medical record was likely incomplete. There was only a single data abstractor not blinded to the study intentions and a few other problems.

But I’m willing to believe the conclusions.

“Where CT is available, we recommend examination during log roll be limited to visual inspection only, in unconscious trauma patients.”

Some might argue… well, what’s the harm? Fair enough, but palpating is clearly a waste of time in a cohort getting CT anyway.

So, you’ll have to find another place to stick those magical fingers... Just sayin'


The fist bump: A more hygienic alternative to the handshake?

No, I don’t subscribe to the American Journal of Infection Control. This study was brought to my attention by a very recent blurb in our local newspaper the Mercury.

Should we be fist bumping rather than shaking hands? Does this reduce the spread of bacteria? For all the craziness promulgated by infection control authorities, I wonder how they have missed this breakthrough!?!

This study looked at the transfer of E. coli during a standard handshake, high five, fist bump and man hugs. (Ok… I made the last one up.)

The figure clearly shows we should be taking Barack Obama's lead and go fist bumping.

Image result for fist bump barack obama michelle

Alright… I’ll acknowledge this is a surrogate marker and there are many other methodologic issues. But I don’t care.

Peace… Doyle out!


Every minute counts: Time to delivery of initial massive transfusion cooler and its impact on mortality. A misleading observational study.

This study was a planned subanalysis of the PROPPR trial. If you can remember, this was an RCT looking at a blood products transfusion ratio of 1:1:1 vs. 1:1:2 (platelets, plasma, PRBC’s) in trauma patients with exsanguinating haemorrhage. There was no difference in the primary outcome.

But the PROPPR trial also collected observational data on time to massive transfusion (MT) activation and time to blood product cooler arrival. Could mortality and other outcomes be predicted by these times? Time for some subanalysis…

I’ll start with the conclusion of this observational study;

“…every minute from time of massive transfusion protocol activation to time of initial cooler arrival increases odds of mortality by 5%.”

You will very likely hear this quoted in the future. Unfortunately, it is not only misleading but likely wrong.

First, it is not absolute mortality but rather “odds.” This concept of “odds” is challenging to comprehend and relies on relative numbers.

In addition, there was very likely residual confounding that was responsible for the differences in mortality. The timing of MT activation and cooler arrival was not randomized.

The authors did make attempts at a few adjusted analyses for some identified confounders such as injury severity, physiology etc. and this was a bit hit and miss. There were some statistical differences and some trends. But nothing very clear. Of course, these adjustments are always difficult and are based on many assumptions.

The authors then go on to discuss how their results may be analogous to STEMI teams, door-to-balloon time, stroke teams etc. All of this is a bit silly.

I don’t have an issue with recognizing the need for massive transfusion quickly and getting blood products to the bedside as fast as possible when needed. But this is based on common sense and not this study.

So, when you hear “it is proven that there is a 5% increase in the odds of mortality for each minute delay…” you can shake your head in either a 1:1:1 or 1:1:2 fashion.


Sunday 30 July 2017

Prevalence and Clinical Import of Thoracic Injury Identified by Chest CT but not Chest X-ray in Blunt Trauma: Multicentre Descriptive Study

Uuuggggh… this paper was hard to digest. The writing and presentation of the manuscript was more complicated than playing Twister in a hot-tub. 

Case in point; the abstract alone has 629 words.  Screw the Annals of Emergency Medicine limit of 250 words in their instructions to authors.

I’ll try to distil it for you…
  • In patients with blunt trauma, how many more injuries does a chest CT find over a plain film?
  • Are these injuries clinically significant?
  • Does finding these occult injuries result in major intervention?

This was a secondary analysis of data collected for NEXUS chest and chest CT studies collected at 10 level 1 trauma centres in the USA. (So, good data in...)

Inclusion criteria were patients >14 years old with blunt trauma who had both a chest x-ray and chest CT. (98% of the time, the chest x-rays were portable supine films.)

Occult injury was defined as those seen on CT but not on plain film. These included pneumothorax, hemothorax, sternal or scapula fracture, >2 rib fractures, pulmonary contusion, T-spine fracture, diaphragm injury.


Of 14,553 patients in the NEXUS database, 5,912 met inclusion getting both plain film and chest CT. Of these, 2048 had chest injury.

What was the rate of occult injury?


What proportion of occult injuries required major intervention? (as defined by surgery, chest tube, or mechanical ventilation for pulmonary contusion).


This sounds rather impressive until you realize the great majority of these interventions were chest tubes for occult injury! 

In the end, 48 patients (08%) out of 5912 had major intervention (excluding chest tubes). A few of these were clearly nasty injuries including diaphragm rupture and major vessel injury.

Could many of these interventions been of no patient benefit and/or harmful? Could these injuries have been detected after a period of observation?

Of course.

Did over-diagnosis with CT cause harm with increased downstream intervention and resource utilization?

Quite possible.

What about the addition of point-of-care ultrasound? We know that portable supine chest x-rays are awful at finding pneumothorax and haemothorax.

This was ignored.

What are we to make of all of this?

The biggest caution is the interpretation.

You will probably hear it quoted that chest x-ray misses 70% of all injuries compared to CT without mention of the details above.


Also, remember the included patients got both plain film and CT at the discretion of the ED doctors. So, this is not all comers with blunt trauma… This was a sicker group that the doctors were worried about.

In the end, I think this paper provides more questions than answers. If you use a bigger microscope, you will find more injury. But does this help or harm patients?

The answer is still uncertain.

I think I need a dip in that hot-tub.


Mortality rates of severe TBI: impact of direct vs. non-direct transfer

There are potential benefits to transfer patients with suspected severe traumatic brain injury (TBI) directly to a trauma centre and bypass smaller local hospitals. The tertiary centres likely have more experience and expertise to manage these patients well.

But of course, this must be balanced by potential harm of prolonged transportation prior to hospital evaluation and resource implications.  

This study sought to answer which strategy is best for these patients; go to the small centres first or direct transfer to the trauma centre?

This was a registry review of patients (with no chart extraction methods reported) presenting to a level 1 trauma centre in New York. 171 patients with GCS<8 (severe TBI) were directly transferred as compared to 92 that were transferred after going to a local hospital.

The primary outcome was mortality. Can’t argue with this outcome…


There was an observed 19% mortality rate for direct transfer vs. 36% for secondary transfer.  To be clear, this was an absolute difference of 16% or a NNT of 6 for mortality.

The only thing that springs to my mind is, “holy shit!”

But wait…

This is sounds too good to be true. So, what’s wrong?

This was not a randomized trial. Therefore, this observational study will inevitably have some underlying confounders between the groups. Did they make any attempt at controlling or analysing for confounding?


To be fair, they did mention that the patients undergoing direct transfer had higher injury severity scores (ISS) but this is only a single measure and far from perfect.

I strongly suspect the cohort of patients undergoing secondary transfer were vetted at the local centres and found to be genuinely quite sick. Many likely had complete primary surveys and pan-scans. They kept the patients that were ok and only transferred the sickest ones. So of course, this group was going to have worse outcomes than those who were not screened by doctors.

But, for many reasons I could be wrong.

Unfortunately, a good RCT would be very tricky to undertake. In addition, it would arguably fail on feasibility, ethics and issues of external validity.

What should we conclude?

This study does not add much to our understanding of this complex question. It has the capacity to be grossly misinterpreted. 

Great one for journal club… but afterward you might feel like you have your own TBI.


Sunday 2 July 2017

Early Secondary Neurologic Deterioration After Blunt Spinal Trauma: Are the collars really necessary?

“Keep them still! Don’t let them move… if they turn their head an inch, they’ll be paralysed.”
-Dr Longheld Dogma

Protection of the cervical spine has been a bedrock of trauma care. This has traditionally been accomplished by strapping patients to backboards and the application of properly fitted cervical collars.

But these collars are uncomfortable, have potential harms and can make cooperative patients rather uncooperative.

Of course we can rest easy in the knowledge that  the original randomised controlled trials (RCT's) of the cervical collar took this all into account. 

What…there is no good quality evidence that the c-collars do anything?!?

Surely there must at least be a large body of observational literature to suggest they work? Heck, we’ve been using them for decades!

These authors sought to definitively answer the latter question.

They conducted a thorough literature review to find cases of neurologic deterioration in the early care of blunt spinal trauma.

In the end, they found only 41 cases from 12 papers. The quality of the evidence was “extremely low.” They conclude the following:

No reports were found of sudden, movement-provoked deterioration during the prehospital interval; this remains an unpublished phenomenon if not a medical myth. Overall, since many of these events described are consistent with the natural history of disease, it is unclear how often deterioration may be causally linked with patient movement and furthermore whether such events are preventable by motion restriction.

On the flip side, one might argue that lack of evidence does not equate to lack of efficacy. After all, these collars have not truly been studied. In addition, there may have been negative reporting bias of bad outcomes.

What should we conclude? I think you can make up your own mind… hopefully not while strapped to a collar.


Saturday 1 July 2017

Outcomes After Out-of-Hospital Cardiac Arrest Treated by BLS vs. ALS: polishing a pig

There is a developing rationale that giving ALS care during out-of-hospital cardiac arrest (OHCA) might actually be harmful. The reasons often include potential harm by trying to establish an advanced airway, drugs have never been shown to improve meaningful outcomes, and the extra time it takes. This delay may distract from other beneficial interventions.

Thus far, the best quality data comes from Captain Ian Stiell’s OPALS study out of Canada. This was a before and after design and concluded no benefit to ALS over BLS with a focus on rapid defibrillation.

Obviously a properly done RCT with allocation concealment would be ideal to answer this research question. But there are questionable ethical hurdles and genuine issues of feasibility.

These authors sought to answer the question with another hypothesis raising observational study. 

Unfortunately, it has several fatal flaws and probably should not be read beyond the abstract.

They looked at patients with OHCA who had ALS or BLS dispatched in non-rural counties in the USA. They identified the variables by Medicare billing and coding data.

This study was not randomised and there could be very good reason why one type of dispatch was chosen over another. Could the ALS cohort have been sicker?

To try to correct for this confounding by indication, the authors tried the usual propensity matching based on all kinds of variables that may or may not have been measured correctly.

In short, this study was based on a multitude of assumptions and dubious statistical analysis of poor quality billing data. There is no doubt they tried hard and put in a lot of effort. But there is only so much one can do to polish a pig. In the end, it’s still swine.


Sanghavi P, Jena A, Newhouse J, et al. Outcomes After Out-of Hospital Cardiac Arrest Treated by Basic vs. Advanced Life Support. JAMA Intern Med. doi:10.1001/jamainternmed.2014.5420