Sweet 16 (papers of the year for NYGH EMU 2017)

At this year’s North York General Emergency Medicine Update, I was asked to review the most important emergency medicine literature from the past year. These are the 10 papers I decided to talk about. If you have questions or comments, please join the discussion at the bottom of the post.

Placebo is as good as (or maybe better than) amiodarone or lidocaine for out of hospital cardiac arrest (the ALPS trial)

Kudenchuk PJ et al. Amiodarone, Lidocaine, or Placebo in Out-of-Hospital Cardiac Arrest. NEJM 2016. PMID: 27043165

Methods:

  • Large multicenter, randomized, double blind trial.
  • Included a total of 3026 patients (4653 in the intention to treat analysis) 18 years of age or older with nontraumatic out of hospital cardiac arrest and shock-refractory ventricular fibrillation or pulseless ventricular tachycardia (defined as persistent or recurrent VF or pulseless VT after one or more shocks anytime during resuscitation).
  • Were randomized to either amiodarone 300 mg IV (150 mg if < 45.4 kg), lidocaine 120mg IV (60 mg if < 45.5 kg), or placebo (0.9% Normal Saline IV).

Results:

  • The primary outcome was survival to hospital discharge, and there was no statistically significant difference: Amiodarone 24.4%; Lidocaine 23.7%; Placebo 21%
  • There were statistically significant differences favouring treatment in the subgroups that had witnessed arrests.
  • There was a statistically significant difference in survival to hospital admission: Amiodarone 45.7%; Lidocaine 47.0%; Placebo 39.7%

Caveats:

  • The trial was powered to find a 6.3% difference between the groups. A smaller, but real difference between the groups is possible, and wouldn’t have been identified here.
  • Time to treatment was 19 minutes on average, which could limit the effectiveness of the treatments.
  • Including patients with unwitnessed cardiac arrests (and therefore poor outcomes) could dilute out a true treatment effect.
  • There were a very large number of protocol violations. This means that they excluded 1627 patients from the analysis who were initially randomized because they were later discovered not to have VF or VT. This is the argument for focusing on a per-protocol analysis rather than an intention to treat analysis. However, this very much limits the external validity of this paper, because those patients would almost certainly be included in real world practice. (There was no difference at all between the groups in the intention to treat analysis.)

Why I chose this paper:

  • This is a large, well done trial that confirms a recurrent theme in resuscitation: giving drugs is unlikely to help dead patients. A lot of the commentary about this paper has focused on the possibility that it was underpowered or has focused on the subgroups. Through that lens, it might appear that is reasonable to continue using antiarrhythmics in ACLS despite this negative trial. However, my concern (aside from the fact that the results from subgroups are almost never verified in future research) is that we are overlooking very significant harms. In this trial, there was a significant increase in survival to hospital admission, but no change in survival to discharge. That means we are putting a lot of people into the ICU with no hope that they are going to leave. Aside from the obvious harms to the medical system, I think this is a massive harm to the individual patients. I know of nobody who wants to spent the last week of their life in the ICU with no hope of recovery. This is one of the worst things we can do to people with modern medicine, and we need to be more cognizant of it as a harm.

My bottom line:

  • I will not use amiodarone or lidocaine routinely as part of ACLS

Other FOAMed commentaries:


Procainamide is better than amiodarone for stable ventricular tachycardia

Ortiz M, Martín A, Arribas F. Randomized comparison of intravenous procainamide vs. intravenous amiodarone for the acute treatment of tolerated wide QRS tachycardia: the PROCAMIO study. European heart journal. 2016. PMID: 27354046

Methods:

  • A multicenter, randomized, open-label trial.
  • They included 74 patients with stable ventricular tachycardia (SBP >/= 90 mm Hg, absence of dyspnea at rest, absence of peripheral hypoperfusion and no severe anginal symptoms).
  • Were randomized to either procainamide 10 mg/kg IV over 20 minutes or amiodarone 5 mg/kg over 20 minutes.

Results:

  • The primary outcome (major cardiac events within 40 minutes) favoured procainamide: procainamide = 9% vs. amiodarone = 41% (absolute difference = 32% NNH = 3).
  • Procainamide was also better in terms of terminating the tachycardia: procainamide = 67% vs. amiodarone = 38% (absolute difference 29% NNT = 3.3).

Caveats:

  • The major limitation here is that the trial was not blinded. How that affects the results would depend on the views of the treating physicians, but amiodarone is generally the preferred drug, so the lack of blinding probably favours amiodarone.
  • The study was stopped early due to very slow recruitment (they had planned on including 302 patients). Stopping the trial early might result in an overestimation of the benefit of procainamide.
  • Although the absolute differences are large, the trial is small, resulting in large confidence intervals. The fragility index for the primary outcome is only 3.
  • The method of randomization (closed envelopes) is not ideal.

Why I chose this paper:

  • In my experience, amiodarone is the heavily favoured medication in this scenario, but this study indicates a significant advantage (NNT 3) to using procainamide.

My bottom line:

  • My preferred medication in ventricular tachycardia is still propofol, followed by a good dose of electricity, but in cases I decide to manage medically, this one was a game changer for me: I will use procainamide first line.

Other FOAMed commentaries:


Does size matter? (Ketorolac has a ceiling dose)

Motov S, Yasavolian M, Likourezos A. Comparison of Intravenous Ketorolac at Three Single-Dose Regimens for Treating Acute Pain in the Emergency Department: A Randomized Controlled Trial. Annals of emergency medicine. 2016. PMID: 27993418

Methods:

  • A single-center, randomized, double-blind trial
  • Included 240 patients 18 – 65 years of age presenting to the ED with pain that was ≥ 5 on a standard 10 point pain scale in whom the attending physician though ketorolac would be an appropriate medication choice.
  • Patients were randomized to 10 mg, 15 mg, or 30 mg of ketorolac given intravenously.

Results:

  • The primary outcome of pain reduction at 30 minutes was identical for all three groups. On a 10 point scale, pain was improved by 2.5 with 10 mg, 2.4 with 15 mg, and 3.0 with 30 mg.
  • As a secondary outcome, there were no differences in adverse events between the groups.

Caveats:

  • The study was not powered to identify adverse events.
  • This is a single center study, which could affect external validity.
  • They used a convenience sample, and it is possible that patients who present to the emergency department at night are different in some way.
  • Although there was no difference at 30 minutes, it’s possible that a high dose would have a longer duration of action.
  • Is 10 mg a low enough dose? Would 5 mg have resulted in the same result? Are these changes in pain any better than we would have seen with placebo?

Why I chose this paper:

  • Sometimes it is difficult to change practice in response to new evidence. When we are presented with easy, low hanging fruit like this, we should jump at the opportunity. I changed to 10 mg the day I saw this paper and have not looked back. Although this paper doesn’t demonstrate any harm, it was underpowered, and the rule of toxicology is that the dose makes the poison.
  • If you were wondering if the same 10 mg ceiling dose applies to intramuscular injections, it does. (See Staquet 1989 or Minotti 1998 for example.)

My bottom line:

  • There is no reason to be using any dose of ketorolac higher than 10 mg.

Other FOAMed commentaries:


Alternative facts: PE and syncope

Prandoni P, Lensing AW, Prins MH et al. Prevalence of Pulmonary Embolism among Patients Hospitalized for Syncope. N Engl J Med. 375(16):1524-1531. 2016. PMID: 27797317 [free full text]

Methods:

  • Multicenter cross-sectional study in Italian hospitals
  • Among 560 adult patients with first time syncope who were being admitted to hospital, they performed a workup for PE for all patients, which could include risk stratification and D-dimer, CTPA, or VQ scan.

Results:

  • The primary outcome was the prevalence of PE in this group of patients, and is reported as 17.3%.

Caveats:

  • Although they report on the 560 patients who were admitted, there were actually 2584 patients who presented with first time syncope over the study period. The vast majority were sent home. Selection bias (only sicker patients were admitted to the hospital) is one of the critical shortcomings of this paper.
  • The population might differ significantly from what we are used to in North America, as European PE studies always have much higher rule in rates.
  • They don’t give us patient level data on how many individuals diagnosed with PE had specific signs, symptoms, or risk factors for VTE, but it is clearly a lot. In the group ultimately diagnosed with PE, 40% had signs of DVT on exam, 45% were tachypneic, 33% were tachycardic, 36% were hypotensive, 11% had previous VTE, and 20% had active cancer. These are not surprise PEs in otherwise healthy looking patients. These are sick patients, requiring admission, who fainted, but also have abnormal vital signs and signs, symptoms, and risk factors for PE. I don’t think we would have missed any of these PEs (they actually didn’t do the workup in the emergency department, so we can’t know.)
  • The CT or VQ scan could happen up to 48 hours after admission. In elderly patients with VTE risk factors, might some of these PEs have developed during the 2 days of admission, rather than being the cause of admission?
  • It is not clear how many of these PEs were clinically relevant, as compared to just being examples of overdiagnosis. (Although, as most of them sound like they were clinically obvious, with vital sign abnormalities, this might not be a huge problem.)
  • Using CTPA or VQ as a gold standard is a problem, as there is a high rate of false positives, or at least disagreement among radiologists reading the same studies. (See Hutchinson 2015)
  • These results simply don’t pass the “sniff test”. We all see a huge number of syncope patients, and there simply isn’t a large cohort of patients coming back with missed PEs.

Why I chose this paper:

  • This paper got a lot of press when it came out, and it would be tempting to alter your practice based on the headlines screaming that 1 in 6 syncope patients have a PE. Unfortunately, any increase in testing in this population is likely to do far more harm than good. According to the authors, only 24 patients (out of 2584 presenting with syncope) had no clear symptoms of PE. This translates to 0.9%. Aside from the fact that a radiologic finding in the absence of signs or symptoms is probably a false positive or example of overdiagnosis, 0.9% is below the test threshold for PE, and so you will be harming more patients than you help if you start working up syncope patients who don’t otherwise have signs or symptoms of PE. If we consider this study in the context of our pre-existing literature, it becomes even more clear that you should not change your practice. We have very large syncope cohorts, and in none of those cohorts were there large numbers of subsequent PEs, bad outcomes, or unexplained deaths.

My bottom line:

  • This paper should not change your practice in any way. If a syncope patient has signs, symptoms, and risk factors for PE, work them up. If they don’t, don’t.

Other FOAMed commentaries:


We will keep trying, but treating blood pressure doesn’t seem to help in intracerebral hemorrhage.

ATACH-2 trial: Qureshi AI, Palesch YY, Barsan WG. Intensive Blood-Pressure Lowering in Patients with Acute Cerebral Hemorrhage. The New England journal of medicine. 2016. PMID: 27276234 [free full text]

Methods:

  • A randomized, multicenter, open label trial.
  • They included 1000 adult patients with intracerebral hemorrhage (<60cm^3) and a GCS greater than 4 who presented with 4.5 hours of symptom onset and had at least 1 systolic blood pressure reading over 180.
  • These patients were randomized to a target systolic blood pressure of either 110-139 or 140-179, using nicardipine as the first line agent.

Results:

  • For the primary outcome of death or disability (a modified Rankin score 4-6), there was no difference between the groups. (37.7% and 38.7%)
  • The study was stopped early due to futility (original target was 1280 patients).
  • There were no really important differences among the secondary outcomes, but in an adjusted analysis, there were more adverse events in the intensive treatment arm (26% vs 20%, p=0.05).

Caveats:

  • The one big limitation in this trial was that it was open label, but generally that would introduce bias favouring the treatment arm.
  • The trial could be under-powered, as they based their power calculation on an assumed event rate of 60%, and it was only 38%.
  • There were relatively extensive exclusion criteria, which could limit the generalizability.
  • 12.5% of the intensive treatment group failed to reach the blood pressure target at 2 hours, as compared to only 0.8% of the standard treatment group.

Why I chose this paper:

  • Despite all previous research, including the large RCT INTERACT 2, showing no benefit with intensive blood pressure management, we still seem to have a fascination with the blood pressure in intracerebral hemorrhage. Although the results here aren’t surprising, they reinforce that we have much more important things to focus on than the blood pressure.

My bottom line:

  • There is no need to aggressively manage blood pressure in patients with intracerebral hemorrhage.

Other FOAMed commentaries:


Bonus paper: Platelets also don’t help intracranial hemorrhage

Baharoglu MI, Cordonnier C, Al-Shahi Salman R. Platelet transfusion versus standard care after acute stroke due to spontaneous cerebral haemorrhage associated with antiplatelet therapy (PATCH): a randomised, open-label, phase 3 trial. Lancet (London, England). 387(10038):2605-13. 2016. PMID 27178479

Methods:

  • A randomized, multicenter, open label trial. (The outcome assessor was blinded.)
  • They included 190 adult patients with non-traumatic intracerebral hemorrhage and a GCS of at least 8 who had been on antiplatelet therapy.
  • These patients were randomized to platelet transfusion or standard care.

Results:

  • The platelet transfusion group did worse. The primary outcome was a shift towards death and disability on the modified Rankin scale, and it was worse in the treatment group (OR 1.84, 95% C.I. 1.1-3.08, p=0.02).

Caveats:

  • This is an open label trial.
  • The primary outcome was modified after data collection, but before the results were unmasked.
  • This is a relatively small trial (although the best evidence we have on the topic).

Why I chose this paper:

  • Although platelet transfusion has never been standard practice where I work, I understand that it has been routinely taught elsewhere. This trial, representing the best evidence we have seen on the topic, indicates that we could be harming our patients.

My bottom line:

  • Patients taking antiplatelet agents should not receive platelet transfusions to manage their intracerebral hemorrhage.

Other FOAMed commentaries:


Antibiotics for abscesses… maybe

Talan DA, Mower WR, Krishnadasan A. Trimethoprim-Sulfamethoxazole versus Placebo for Uncomplicated Skin Abscess. The New England journal of medicine. 374(9):823-32. 2016. PMID: 26962903

Methods:

  • A multicenter, randomized, double-blind trial.
  • Included 1247 emergency department patients older than 18 years of age with a cutaneous abscess less than 1 week old and greater than 2 cm.
  • After incision and drainange, they compared trimethoprim–sulfamethoxazole (320mg/1600mg – or 2 double strength tabs) twice a day for 7 days to placebo.
  • Clinical cure was defined as a lack of any of the following: fever, an increase in erythema by >25% from baseline, worsening of wound swelling or tenderness by day 3 or 4; fever, no improvement at the end of the treatment period (day 8–10); or more than minimal erythema, swelling, or tenderness by the test-of-cure visit (day 14–21).

Results:

  • The primary outcome of clinical cure at 14-21 days was 80.5% with TMP/SMX and 73.6% with placebo. This equates to a 6.9% (95%CI 2.1 – 11.7%) absolute difference, or a number needed to treat of 14.
  • The TMP/SMX group also had fewer new abscesses (3.4 vs 8.6%) and fewer infected family members (1.7 vs 4.1%).
  • There were more GI side effects in the antibiotics group (42.7 vs 36.1%).

Caveats:

  • In absolute numbers, the improvement in the primary outcome is similar to the adverse events (6.9% improvement vs 6.6% GI side effects).
  • The trial is underpowered to detect more serious side effects, like Steven Johnson’s Syndrome.
  • The dose of TMP/SMX used is twice the usual dose (expect a lot of phone calls from pharmacists if you start prescribing this.)
  • By the descriptions in the paper, the enrolled patients do not seem to have just simple abscesses. The median surrounding cellulitis was 6.5×5.0cm, 20% of the cohort with a cellulitis area greater than 75cm2, 18% were febrile, and many patients “met other guideline criteria for antibiotics treatment”.
  • 45% of participants had wound cultures positive for MRSA.
  • There is a long list of exclusions that needs to be considered.
  • Despite being a modern trial, it wasn’t listed on clinicaltrials.gov.

Why I chose this paper:

  • The use of antibiotics for abscesses would certainly be a game changer. This paper has changed my practice, but not universally. I think these patients were sicker than the average patient that I see, as illustrated by the low cure rates with placebo, greater extent of cellulitis, and high rate of MRSA. I still treat healthy patients with minimal cellulitis without antibiotics, but discuss antibiotics with patients who might be more like those in this study.

My bottom line:

  • Antibiotics may provide a small number of patients with benefit, but that needs to be balanced against the expected increase in side effects. This is a good issue for clinical judgement and shared decision making.

Other FOAMed commentaries:


NPO status for sedation

Beach ML et al. Major Adverse Events and Relationship to Nil per Os Status in Pediatric Sedation/Anesthesia Outside the Operating Room: A Report of the Pediatric Sedation Research Consortium. Anesthesiology. 124(1):80-8. 2016. PMID: 26551974

Methods:

  • This is an observational study. It is a retrospective look at a prospectively gathered database that included all pediatric procedural sedations that occurred at one of the 42 sites that are part of the Pediatric Sedation Research Consortium.
  • There were a total of 139,142 encounters identified.
  • They compared those where the patient was known to be NPO (solids for at least eight hours, non clear fluids for at least six hours, and clear fluids for at least two hours) to patients who were known to fail those NPO criteria.

Results:

  • The was no statistical association between NPO status and adverse events (either aspiration, or major adverse events, which were listed as co-primary outcomes.)
  • Aspiration occurred in 8 patients who were NPO and 2 who weren’t, event rates of 0.8/10,000 and 1/10,000 respectively (odds ratio, 0.81; 95% CI, 0.08 to 4.08; P = 0.79).
  • There were only 75 major complications (a tiny fraction of a percent): 62 unplanned admissions, 10 aspirations, 3 cardiac arrests, and no deaths.

Caveats:

  • This is an observational study, so can only demonstrate associations, not causation. However, a lack of association is more believable than the demonstration of an association.
  • There are a number of possible confounders, such as ASA status, type of procedure, comorbidities, and urgency of the procedure.
  • There are so few events that the 95% confidence intervals are all large (although the simple fact that there are so few events is probably the main take away for me.)
  • NPO status was not available for 31,195 patients.

Why I chose this paper:

  • This paper probably shouldn’t be considered practice changing. We knew this already. The ACEP clinical policy states, “do not delay procedural sedation in adults or pediatrics in the ED based on fasting time. Preprocedural fasting for any duration has not demonstrated a reduction in the risk of emesis or aspiration when administering procedural sedation and analgesia.” However, I know many departments are still handcuffed with rules about NPO times. This just hurts our patients. This data, although not perfect, illustrates that aspiration, and adverse events in general, are incredibly rare, and we should not be prolonging patient stays based on their NPO status.

My bottom line:

  • Within reason, I will perform a sedation whatever the patient’s NPO status. (I do require them to spit the food out of their mouth before I start.)

Other FOAMed commentaries:


The 5 second rule debunked

Miranda RC, Schaffner DW. Longer Contact Times Increase Cross-Contamination of Enterobacter aerogenes from Surfaces to Food. Applied and environmental microbiology. 2016; 82(21):6490-6496. PMID: 27590818 [free full text]

Methods:

  • If you really want to know the details, read the paper. It’s free. Essentially they threw watermelon, dry bread, bread with butter, and gummy candies on the ground for various amounts of time. They used different types of ground (carpet, tile, wood). There was a prespecified quantity of bacteria on the ground.

Results:

  • The amount of bacteria transferred varied by the type of food, material of the floor, and time of contact, but there was still a significant amount of bacteria on the food by 1 second.

Caveats:

  • This is a surrogate outcome. It doesn’t tell you if this amount of bacteria would actually cause illness.

Why I chose this paper:

  • For fun.

My bottom line:

  • Please don’t eat food off the ground.

Apple juice and preferred fluids for gastro?

Freedman SB, Willan AR, Boutis K, Schuh S. Effect of Dilute Apple Juice and Preferred Fluids vs Electrolyte Maintenance Solution on Treatment Failure Among Children With Mild Gastroenteritis: A Randomized Clinical Trial. JAMA. 315(18):1966-74. 2016. PMID: 27131100

Methods:

  • A randomized, single-blind, non-inferiority trial.
  • They enrolled 647 pediatric patients (aged 6-60 months) presenting to a single tertiary pediatric emergency department with 3 or more episodes of vomiting or diarrhea in the preceding 24 hours, less than 96 hours of symptoms, and minimal dehydration.
  • They were randomized to either half-strength apple juice while in the emergency department followed by the child’s preferred fluid at home or an electrolyte rehydration solution.
  • Children received 5-mL aliquots of the assigned fluid every 2 to 5 minutes. Those who vomited received oral ondansetron.

Results:

  • The primary outcome was a composite of hospitalization, need for IV rehydration, subsequent unscheduled health care visit, protracted symptoms, crossover, or dehydration.
  • The apple juice strategy was superior: 16.7% (95%CI 12.8%-21.2%) treatment failure with half-strength apple juice and preferred fluid vs. 25.0% (95%CI 20.4%-30.1%) with electrolyte solution.
  • As a secondary outcome, there was less intravenous hydration in the half-strength apple juice and preferred fluids group (0.9% vs. 6.8%) at index ED visit (p<0.001).

Caveats:

  • These children were only minimally dehydrated. (This was by design. They were testing this fluid strategy for maintenance not for rehydration.)
  • This was a convenience sample, although it isn’t clear how night time patients would differ.
  • When using a composite outcome, you have to be careful, because all components of the composite might not be equally important.
  • This is a single center study, and not just any single center, but The Hospital for Sick Children in Toronto. The external validity to other settings, especially other countries with different resources, might be questioned.

Why I chose this paper:

  • Have you ever tasted the electrolyte solutions? If so, it’s no mystery to you why the children never seem too happy.

My bottom line:

  • I have changed my practice. I suggest preferred fluids for all my patients with gastro.

Other FOAMed commentaries:

Author: Justin Morgenstern

Community emerg doc, FOAM enthusiast, evidence junkie “One special advantage of the skeptical attitude of mind is that a man is never vexed to find that after all he has been in the wrong.” - William Osler

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s