Friday, November 21, 2014

(Don’t) Dive! Dive! For Carbon Monoxide Poisoning

It is, again, the time of year when the temperatures drop precipitously – and, so, it is again the time of year to expand the differential for atraumatic headache to include carbon monoxide poisoning.  With records already being broken in Southern states unused to such temperatures in November, this message comes earlier than usual.

And, as a reminder, there is essentially no usable evidence describing the use of hyperbaric oxygen therapy for acute CO poisoning.

The most recent Cochrane Review, from 2011, covers six studies regarding the efficacy of HBOT as compared to normobaric oxygen therapy.  And, regrettably, all have serious methodologic flaws and potential for bias.  Four of these trials are absolutely negative, with trivial differences in outcomes between the two arms.  Two trials favor HBOT – a 60-patient trial published by Thom in 1995, and the seminal 152-patient trial presented by Weaver in the NEJM in 2002.  All told, the pooled effect of HBOT on short-term neurologic sequelae provides a protective effect with an OR of 0.78 – but with a confidence interval crossing unity (0.54 to 1.12).

Thus, these data neither support nor refute the utility of HBOT for treatment acute carbon monoxide poisoning – and provide no insight into appropriate patient selection.  What is most likely, given these results, is there is a cohort of patients for whom some benefit is observed.  Probably the most reasonable patients to select for treatment include those with the most severe poisoning and who can receive treatment immediately – but, otherwise, expect extraordinarily low-yield resource utilization to attempt treatment in the remainder.

“Hyperbaric oxygen for carbon monoxide poisoning (Review)”

Wednesday, November 19, 2014

Ketamine For You! And You and You and You!

Over the last decade, ketamine has shifted from pariah status to celebrity, increasingly popular for procedural sedation, delayed-sequence intubation, treatment of intracranial hypertension, and even status epilepticus.  There is also a smattering of trials demonstrating its use for inpatient post-operative patients as an opiate-sparing mechanism – but how well does it work in the Emergency Department?

These authors from Brown evaluated use of ketamine for adjunctive pain control in a small, three-arm prospective, blinded comparison between “usual care”, and two doses of ketamine.  Patients with severe acute pain received 0.1 mg/kg of morphine up to 10mg, followed either by placebo injection, 0.15 mg/kg ketamine, or 0.3 mg/kg ketamine.  Patients were then followed for need of rescue medication.

With only 20 patients in each group, the potential for type I and type II errors are substantial.  All groups derived substantial pain relief from treatment.  Patients in the two ketamine groups, however, derived a greater degree of relief – and required a lower gross amount of rescue analgesia.  The authors eke only a few “statistically significant” p-values out of their tiny samples, but, regardless, the underlying differences likely appropriately reflect an expected acute analgesic effect from ketamine.

Unfortunately, the largest limitation of the small sample concerns the incidence of adverse events.  The largest dose of ketamine, 0.3 mg/kg, had clearly distressing effects – half the patients felt dizzy – while both ketamine groups included a handful of patients with nausea, vomiting, and confusion.  On the flip side, two patients in the standard-care group suffered possible morphine-related hypotension and respiratory depression as a result of rescue analgesia.

Ketamine may have a role in acute analgesia in the Emergency Department.  However, appropriate dosing and ideal patient selection remain unclear.

“Low-dose Ketamine Improves Pain Relief in Patients Receiving Intravenous Opioids for Acute Pain in the Emergency Department: Results of a Randomized, Double-blind, Clinical Trial”

Tuesday, November 18, 2014

Bayesian Statistics: We're Dumb as Rocks

A guest post by Justin Mazzillo, a community doc in New Hampshire.

Physicians are often required to interpret medical literature to make critical decisions on patient care. Given that it is often in a hectic and hurried environment a strong foundation of evidence-based medicine is paramount. Unfortunately, this study from JAMA showed that physicians at all levels of training have anything but that.

This group surveyed a mix of medical students, interns, residents, fellows, attending physicians and one retired physician. They were asked to answer the following question:
“If a test to detect a disease whose prevalence is 1/1000 has a false positive rate of 5%, what is the chance that a person found to have a positive result actually has the disease, assuming you know nothing about the person's symptoms or signs?”
Unfortunately three-quarters of the subjects got the answer wrong. The results were consistent across all levels of training. The most commonly given answer was almost as far from correct as possible.

I’ve withheld the answer for those who want to try out the questions themselves, and I know all the dedicated EMLoN readers will fare much better.

“Medicine’s Uncomfortable Relationship With Math: Calculating Positive Predictive Value”

[I gave this same test to our residency program – and the results were almost identical.  A few sample "answers" below. -Ryan]

Monday, November 17, 2014

Laundry Detergent is Delicious Candy

Children, rumor has it, are curious, persistent, and nefarious creatures.  Furthermore, the entire natural environment must first be tasted.  Thus, age-old cautionary guidance regarding home safety for children to prevent exposure to potentially hazardous materials.

So, naturally, laundry detergent would be kept safely out of reach of children – particularly the pre-packaged pod systems, which are candy-like and perfect for oral consumption – correct?


According to this data from the National Poison Data System, in just a single year, there were 17,230 calls for pediatric exposure to laundry detergent pods in just two years, between January 2012 and December 2013.  The majority – 79.7% of cases – were secondary to ingestion.  4.4% of patients with an ingestion were judged to necessitate hospitalization, including 102 requiring intubation.  One death was reported – but the autopsy showed intracranial bleeding of uncertain/unlikely relation to the ingestion of detergent.

The authors specifically call out Procter & Gamble, maker of Tide Pods, for failure to place adequate warning labels or safety latches on their packaging – although, frankly, the ultimate responsibility lay with parents.  Keep cleaning supplies safely out of reach!

“Pediatric Exposure to Laundry Detergent Pods”

Friday, November 14, 2014

It’s a Patient Hand-Off Miracle

Transitions of care – more frequent now in medicine than ever before – are fertile opportunities for error and miscommunication.  Most institutions have developed, at least, informal protocols to exchange patient information during hand-off.  But, certainly, everyone has some anecdotal tale of missed information leading to a near-miss or actual patient harms.

This study tells the story of I-PASS, a handoff bundle implemented and measured as an error prevention strategy by a pre- and post-intervention study design.  Across 9 pediatric residency training programs, residents were observed for six months for time spent in hand-offs, time spent in patient care, and a variety of classifications of preventable and non-preventable errors.  Then, the I-PASS bundle was introduced – a structured sign-out mnemonic, a 2-hour workshop on communication skills, a 1-hour role-playing and simulation intervention, a faculty development program, direct-observation tools, and a culture-change campaign with a logo, posters, and other promotional activities.

Following the intervention, residents were, again, observed for six-months.  And, in general, preventable medical errors decreased a small absolute amount, along with a larger absolute decrease in near misses.  2 of 9 hospitals had increases in medical errors after the interventions, and the bulk of the effect size was a result of improvements at two hospitals whose baseline error rate was double that of the other 7 facilities.

The authors, then, are very excited about their I-PASS bundle.  But, as they note at the end of their discussion: “Although bundling appears to have been effective in this instance, it prevents us from determining which elements of the intervention were most essential.”  And, on face validity, this is obvious – the structured sign-out sheet was only one of many quality improvement interventions occurring simultaneously.  A decisive change in culture will trump the minor components of implementation anytime.

The final takeaway: if your institutional audit reveals handoff-related errors are pervasive and troublesome, and if reductions in such errors are prioritized and supported with the correct resources, you will probably see a reduction.  The I-PASS tool itself is not important, but the principles demonstrated here probably are.

“Changes in Medical Errors after Implementation of a Handoff Program”

Wednesday, November 12, 2014

A Thicket of Coronary Disease Prognostications

This recent article out of JAMA garnered headlines primarily for the insight into the risk of non-obstructive coronary artery disease – headlines such as: “Risk of Heart Attack Jumps with Non-Obstructive Heart Disease” or “Increased Risk Found For People With Even ‘Minor’ Narrowing of Heart Arteries”.

Somehow, this is profound – that individuals with measurable atherosclerotic plaque are at greater danger of suffering an acute coronary syndrome than those without.  And, frankly, despite this “significantly increased risk”, the most interesting insights – from an Emergency Medicine standpoint – are tied to how low the risks of MI were, overall.

This is a Veterans Affairs database of coronary angiography findings observed on "elective" cardiac catheterization – meaning the indication for coronary angiography in all cases was not associated with an acute coronary syndrome.  Most cases were referred primarily for chest pain, with a minority for a positive functional study.  Catheterization findings were classified as non-obstructive, 20-70% stenoses, or >70%/>50% left main stenoses, subdivided into single, double, or triple vessel disease.  Center for Medicare Services data was queried to determine 1-year outcomes, specifically myocardial infarction or death from any cause.

As expected from a VA study, the cohort is mostly male and aged between 50 and 70 years.  Nearly all had a history of hypertension and hyperlipidemia, while smoking, diabetes, and obesity were well-represented.  In short, exactly the folks you’d expect to refer for catheterization in the setting of chest pain – the sort of individual every Emergency Physician would consider “high risk”.

But the catheterization only revealed obstructive coronary artery disease in about half of these patients.  And, among those, only a little more than half received an intervention associated with angiography – either PCI or CABG.  The remainder were amenable only to medical intervention.  But, even in this cohort with pervasive vascular disease, the 1-year rate of MI was only between 1.18% and 2.47%, depending on the number of vessels involved.  Then, if non-obstructive disease was found, the 1-year rate of MI falls to 0.24% to 0.59%.  And, those without CAD had a 0.11% incidence of MI within a year.

My takeaway from all this?  As a whole, even this highest-risk cohort has a combined ~1% risk of MI within a year – meaning one could theoretically discharge nearly every chest pain patient from the Emergency Department if proper short-term follow-up were in place, and the number of adverse outcomes would be a tiny fraction of a percent.  [Add:  Stephen Smith takes issue with the generalizability of this elective catheterization cohort to our Emergency Department population, and suspects we have a much higher prevalence of unstable plaques – and potential for a greater number of adverse events.]

What’s unfortunate in the data presented, however, are few obvious differences between those who had severe – even 3-vessel disease – and those who had no disease whatsoever.  In aggregate, even though the differences met statistical significance, the absolute differences were small.  Indications were similar between groups, comorbid disease was similar between groups, and, perhaps, those with more advanced disease were slightly older.  Perhaps some type of matching algorithm could be used to generate more precise, individualized estimates for individual patients, but such is just speculation.

“Nonobstructive Coronary Artery Disease and Risk of Myocardial Infarction”

Monday, November 10, 2014

The Remarkable Power of Placebo

Have you ever felt pressured to provide a patient with something at the time of discharge?  Something, anything to ease their suffering for an illness of unalterable benign progression?  Never?  You cold-hearted bastard.

This tiny trial, despite its small size, provides yet another beautiful look at the magical healing power of placebo.  Or, more accurately, rather than healing power, at least the satisfying power.  After all – most families in the ED at two in the morning are not there because their child is awake from coughing, but because the parents are.

This trial, in the same vein of several honey trials before it, compared no treatment, a placebo treatment (grape-flavored water), and agave nectar for the treatment of pediatric cough-related illness.  Agave nectar was chosen for its similarity to honey, while not carrying the hypothetical botulism risk.  With ~40 patients in each group, all patients improved during the duration of the study.  However, despite the small sample, both agave nectar and placebo provided durable advantage over no treatment across all surveyed measures of patient and parent comfort.  There was, however, no difference between placebo and agave nectar – or, if there was, it was too small to be detected in this study.

What is most remarkable about this study is the authors discussion – that placebo treatments could be considered ethical, even when no benefit to such treatment is found in controlled studies.  Providing patients – or parents – with, at least, an inexpensive, harmless treatment option takes advantage of the power of belief, to the extent that real patient/parent-oriented benefits may be observed.

Unfortunately, the lead author of this study was a paid consultant to Zarbee’s Inc (a maker of “natural” over-the-counter remedies) at the time the study was initiated, and Zarbee’s provided funding for the study.  But, thanks to their contribution to science, we now know they only work as well as you expect ….

“Placebo Effect in the Treatment of Acute Cough in Infants and Toddlers”

by Darren Cullen
Copyright Darren Cullen - Spellingmistakescostlives

Homeopathic Accident & Emergency

Friday, November 7, 2014

No Good Ever Comes of Dabigatran

Is anyone actually still using this drug?  If so, why?  There has been nothing but an endless progression of bad news associated with this medication – from Boehringer Ingelheim settling a massive lawsuit, the authors from RE-LY admitting they “missed” additional adverse events for a second time, and, now, further evidence describing flawed real-world effectiveness contrary to its supposed demonstrated efficacy.

The RE-LY trial showed non-inferiority for dabigatran at stroke prevention in non-valvular atrial fibrillation, but appeared to place patients at significantly lower risk of bleeding compared with warfarin.  One of the critiques of RE-LY, however, is the patients were not appropriately representative of the general population at-risk for atrial fibrillation.  By omitting chronic kidney disease and enrolling a generally white population in Europe, the generalizability of their findings is ultimately impaired.

And, thus, we see the fruits of such critiques.  This is a retrospective cohort of Medicare beneficiaries prescribed either dabigatran or warfarin for atrial fibrillation.  Based on propensity matched samples of 1,302 dabigatran users and 8,102 warfarin users, major bleeding of the dabigatran cohort exceeded that of the warfarin cohort – 9.0% (95% CI 7.8 - 10.2) versus 5.9% (95% CI 5.1 - 6.6).  Risks were increased in the elderly, blacks, those with chronic kidney disease, and those on concomitant anti-platelet therapy.

So, we have a lesson – one of effectiveness versus efficacy, or one that’s an indictment of the original RE-LY study protocol.  Medications should not be expected to perform the same in general use as they do in clinical trials – even those with tens of thousands of patients, such as RE-LY.  Independent, confirmatory study ought be mandatory to ensure the safety of the public.

“Risk of Bleeding With Dabigatran in Atrial Fibrillation”

Walid Gellad on Twitter points out this study in Circulation, published last week to much lesser fanfare, which uses a larger Medicare sample to come to the opposite conclusion – that dabigatran is better than, and safer than, warfarin.  Which is correct?  A subject for continued debate, to be certain.  The correct answer is probably somewhere in between – dabigatran is safer for some, but more dangerous for others.  However, given the lack of reversal – wouldn't a Factor Xa inhibitor be a better choice, regardless?

Wednesday, November 5, 2014

Scientific Writing is a Tragicomedy! Destroy!

Modern scientific writing – both in the exercises of writing and reading – is obtuse and uninviting.  Rather than clearly communicate an unbiased reflection of the conduct and findings of a particular study, the medical literature most commonly succeeds in doing the opposite.  After all, how else would I find enough to complain about on this blog?

This editorial elucidates so many joyfully preposterous notions it cannot help yet be loved.  It is best described as a no holds-barred cagematch versus all the inane pageantry of scientific writing.  Just a few of the gems, paraphrased:
  • Don’t let the authors write the abstract; they’ll just misrepresent the study!
  • Delete the introduction; uninsightful filler.
  • No one cares the brand and manufacturer of the statistical package used.
  • Unequal composite end-points and subgroup analyses should be banished.
  • The discussion section only serves authors' purposes of dubious claims through selective reporting and biased interpretation of their results.
Some elements of this brief report are, indeed, novel.  Others are simply accepted best practices long since forgotten.  Regardless, it is a refreshing reminder of how brutally poorly the current medical literature serves effective knowledge translation.

“Ill communication: What's wrong with the medical literature and how to fix it.”

Monday, November 3, 2014

What Do People Remember From Cardiac Arrest?

The anecdotal experiences and reports from survivors of cardiac arrest are diverse, yet frequently describe common themes.  Detailed memories, "near death experiences" of entering another world, and sights and sounds from the arrest context are frequently reported.  And, what better place to collect cardiac arrest events than in a hospital?

As one might imagine, the population available for such interviews is rather limited – so it requires a massive undertaking, in this case, a four-year prospective evaluation across 15 hospitals in the U.S., UK, and Austria.  Essentially, the local investigator at each institution received notification of every adult, in-hospital cardiac arrest.  Survivors were identified and interviews conducted as soon as feasible, given continued comorbid illness.

Out of 2,060 cardia arrests, only 330 were eligible for study inclusion.  Investigators conducted 140 interviews – and only 55 had any memories.  Of these, 53 had detailed memories, whether unrelated or classic “crossing-over” phenomena – but nothing relating to the circumstances of their arrest event.  Only two patients had detailed memories of the circumstances of their event – one set of memories was not able to be verified, but the authors were able to fully verify the other set of memories by interview of the resuscitation staff.

So, essentially, very rarely do patients have any recollection of their arrest event.

Interestingly, one part of this study attempted to verify the veracity of the “floating above and watching” aspect of some individual’s arrest recollections.  The authors constructed shelves in areas thought most likely to have cardiac arrest, and then placed objects on the shelf that would only be visible from a perspective near the room ceiling.  Unfortunately, 70% of the cardiac arrests in this study occurred in locations where there were no shelves, including both arrests with detailed recollection.

“AWARE—AWAreness during REsuscitation—A prospective study”