Monday 24 April 2017

Do you shenfu? Shenfu injection for septic shock.

There is probably better quality evidence for shenfu injection for sepsis than the recent craze over Vitamin C… pretty sad considering this awful paper.


It’s a “well-known traditional Chinese formulation restoring Yang from collapse, tonifying Qi for relieving desertion…” (It’s made from ginseng and aconite.)

So if you need your Yang restored and your Qi tonified then Shenfu is for you! Especially if you are feeling a bit septic?

Back to this randomized placebo controlled trial…

157 patients with sepsis were randomized to either shenfu injection vs. placebo in addition to usual care.

Inclusion criteria relied on the Third International Consensus Definition for sepsis and septic shock published in 2016… even though this study was conducted from June 2010 to November 2012. Perhaps shenfu makes you see into the future…?

The primary outcome was lots of things… something to do with cellular immunity, CD4, CD8 counts, mHLA-DR expression, ex-vivo endotoxin LPS induced TNF, IL 6 and IL 10 cytokines, length of ICU stay, vasopressor use, illness severity, etc.

Why didn’t they just measure the Yang and Qi?

Believe it or not, the authors state the study was powered to the secondary outcome of mortality at 28 days. They were looking for a “holy shit!” difference of 20% and as such only needed low numbers of patients.  (Could it be they changed their original primary outcome to a secondary one after no difference was found? No….)


Shenfu was great! (except for mortality… but who cares)

The authors conclude, “these findings suggest that shenfu injection can enhance the cellular immunity of patients with septic shock and could be a promising adjunctive treatment for patients with septic shock.”

So for all of you out there giving Vitamin C for sepsis, perhaps you should mix it with a little shenfu… shouldn’t hurt?


Geckos & Watermelons: Association Between Tracheal Intubation During Adult In-Hospital Cardiac Arrest and Survival

Generally speaking, what do you get when you perform an observational study based on registry data?

Not much.

Nevertheless, this study does pose an interesting research question. Is intubation beneficial for in-hospital cardiac arrest? Or should we just use other means of ventilation and resuscitate accordingly.

To really answer this question, one would ideally like to see a randomized trial so that confounders are equal amongst the two groups. Otherwise we might be comparing geckos to watermelons.

Back to the study...

This was an observational cohort study of patients who had in-hospital cardiac arrest. The primary outcome was survival to hospital discharge.

Of 71,615 patients who were intubated in the first 15 minutes, 60% were matched to a patient not intubated in the same minute.

Survival was lower among patients who were intubated 16.3% vs those not intubated 19.4% (RR=0.84; 95% CI 0.81-0.87; p<.001)

So this was a huge study with a pretty p value… so it must be right?!


A few things...

You can’t overpower bias with big numbers.

This study relied on registry data which is notoriously of poor quality. Garbage in, garbage out. No amount of fancy statistics can make up for this.

Once again, this was not a randomized trial so the groups were very likely different from the start. The outcomes could very well be due to one group being sicker than the other. The authors tried to correct for this confounding by some clever propensity matching. But it must be emphasized that this is far from an exact science.

So what’s the take home?

Probably not much. But this study does lay the foundations for a proper randomized trial… hopefully only including geckos.


Andersen LW, Granfeldt A, Calloway CW, et al. Association Between Tracheal Intubation During Adult In-Hospital Cardiac Arrest and Survival. JAMA 2017;317(5):494-506.

Trial of Pregabalin for Acute and Chronic Sciatica… Show me the money!

It seems half of the patients I see in the ED are taking Lyrica (pregabalin). Therefore, it was no surprise to me to see it at number 13 on the list of the biggest blockbuster drugs of 2015 making billions of dollars for Pfizer.  

But I was equally surprised to see a negative high quality RCT of pregabalin published in the highest impact journal on the planet!

Surely Pfizer was crazy when they funded and performed this study? Well… of course they weren’t… it was some other crazy authors.

This Australian NHMRC funded double-blind controlled trial randomized 209 participants to either pregabalin or placebo. The primary outcome was a leg-pain intensity score at the 8 week mark.

How much did the pregabalin help?

Zero, zip, zilch, nada…

There was no statistical improvement in the primary or any of the secondary outcomes. About the only thing we are sure of is the incidence of adverse events was significantly higher in the pregabalin group.

Although this was a small study (i.e. risk for Type II error), it probably had adequate power to detect a clinically meaningful difference in pain scores. And the results trended a favour towards placebo anyway.

Of course, no study is perfect. If I was working for Pfizer, I would point out that only a minority of the participants were likely to have neuropathic pain as measured on the PainDETECT score. So of course my drug wouldn’t work… while I sit on my horse made of gold.

As an aside…

I’ve been dubious of these neuopathic pain drugs. A disturbing study published in the NEJM demonstrated what looks like systematic research misconduct with gabapentin which is very closely related to pregabalin. This only was discovered during litigation and release of internal company documents. (This is probably my favourite journal club paper of all time…)

So what’s the take home message for the emergency docs?

Pregabalin probably does not help sciatica. But I would also be sceptical as to the benefits in a wide range of other indications… Especially with industry sponsored research.


Wednesday 19 April 2017

Risk for Clinically Relevant Adverse Cardiac Events (CRACE) in very low risk chest pain patients admitted to hospital (in the USA)

Move over MACE and say hello to CRACE!

These high profile authors from the USA sought to determine the incidence of short term life threatening events in low risk chest pain patients admitted to hospital.

Their concern was a lack of benefit from admission balanced with the potential harm (false positives, over-diagnosis, hospital acquired infections, VTE, iatrogenic etc.).

They reviewed a large database from 3 hospitals in the US Midwest from 2008-2013. They identified admitted patients that had negative serial biomarkers (old school troponins), normal vital signs and non-ischemic ECGs.

Interestingly, these authors picked a different outcome measure for this study from the traditional MACE that we have seen in many previous studies. They created “CRACE” defined as:
  • Life-threatening arrhythmia (VF, symptomatic bradycardia, or treated brady or tachy arrhythmias)
  • Inpatient STEMI
  • Cardiac or respiratory arrest
  • Death

(Noticeably absent from CRACE is non-STEMI or longer term outcomes beyond hospitalization.)

What did they find?

Of course this group was very low risk… but could you have guessed a CRACE rate of only 0.06% (4 of 7266 patients)!

As is usually the case, there were some methodological issues with this study and the reported rate is probably not a complete reflection of the truth. However this makes a compelling argument that this is an extremely low risk group that may derive more net harm from admission.

The authors are careful to mention that this study does not mean patients derive no utility from further investigation and management after the ED evaluation. But suggest this may be done as an outpatient and might provide a framework for shared decision making.

My guess is the Australians would never admit many of these low risk patients anyway. In addition, these patients would be even lower risk with the high sensitive troponin assays.

Take home? Perhaps we can sleep a little better knowing that low risk patients truly are very low risk. Hope this doesn’t seem to CRACEy…