Can we electrically cardiovert stable AF in the ED?

True Story:

A 56 year old anaesthetist presents to your department, he states he started having palpitations while cycling up a large hill.  He took his pulse which felt irregular and so he cycled to the ED.  His HR 107, BP 130/75 RR 18 Sats 98% ORA Apyrexial.  His ECG shows AF, with no ST changes.  He ‘demands’ electrical cardioversion.

For the basics the brilliant RCEM learning website, has a great article on ED management of Atrial Fibrillation.   It doesn’t go into much detail about the nuts and bolts of electrical cardioversion in the ED.

Lets try and go into a bit more detail.

Now what does NICE say?  The relevant verse is here:

1.7.2 In people with atrial fibrillation presenting acutely without life-threatening haemodynamic instability, offer rate or rhythm control if the onset of the arrhythmia is less than 48 hours, and start rate control if it is more than 48 hours or is uncertain. [new 2014]  

1.7.3 Consider either pharmacological or electrical cardioversion depending on clinical circumstances and resources in people with new-onset atrial fibrillation who will be treated with a rhythm control strategy. [new 2014]  [page 23]


We can consider EC or PC depending on the clinical circumstances and resources.  The implication here is that EC is more resource heavy.  I would argue that it might be in the ED, but that across the organisation as a whole it is less resource intensive to sort in our department than to clog up an acute ward. This is backed up by studies in Ottawa8, Rochester6 and Pennsylvania which show that length of stay (the biggest determinate of health care cost) is less if we just get on and do it9.

So NICE asks us to consider the clinical circumstances of people with new-onset AF.  So what clinical circumstances do we need to consider?

Cause of AF

AF caused by electrolyte abnormalities, ACS, Sepsis, Structural heart disease, or Thyrotoxicosis is unlikely to resolve with a single shock.  After all you will need to treat the underlying cause first.  It is only really appropriate to offer ED EC to patients with idiopathic AF.  It is also important to note that up to 60% of patients with idiopathic AF revert spontaneously with no treatment at all.

Suitability for Procedural sedationbeaker

This depends on the skills present in your department, as well as the patient’s history.  It is unlikely that it is going to be appropriate to perform EC on a patient who has an ASA grade 3 or more.  Similarly you aren’t going to want to sedate someone whose airway looks like Beaker from the muppets.

Choice of sedation technique and agents depends on your skills and preferences.

Risk of thromboembolic complications (before and after EC)

Most guidance, and most studies seem to have a 48 hr cut off mark.  If the AF has been going for less than that time the received wisdom is that you aren’t going to shoot off a thrombus.  There is some opinion out there that EC might stun the right atrial appendage and a thrombus forms AFTER cardioversion.  This remains to be proven.

If the AF has no definitive time of onset, then the safest thing to do is anticoagulation first (depending on centres this can be clexane + warfarin and a clinic, or an admission and IV heparin), and an elective cardioversion later (usually 4 weeks).

What is the incidence of Thromboembolic events (TE) after EC?  I did find a meta-analysis3 from 1998 (Berger et al) which looked at 32 studies, giving 4621 participants.  92 patients had a thromboembolic episode (2%).  The duration of the AF was less than 2 days in the TE group in only 3 patients (0.06%).

Duration of AF Number of patients with TE episode/4621
<2 days 3 (0.06%)
3 days – 1 month 21 (0.45%)
>1 month 42 (0.9%)
Unknown 26 (0.5%)


The paper doesn’t tell us how many patients were in each group, so the data has little meaning when broken down as above, so we can only really take from this the overall percentage of 2% for thromboembolic events.

Berger et al was more interested in the timing of thromboembolic event after the cardioversion, this peaks in the first few days following the cardioversion no matter the duration of the AF.  I also found a study looking at NOACs (novel oral anticoagulants eg dabigatran) vs warfarin for anticoagulating patients prior to cardioversion but they got a 30 day thromboembolic rate of 0.8% with warfarin10.

A retrospective study from Denmark of 16 274 DC cardioversions7 without and with anticoagulation was published this year.  Unfortunately they didn’t look at the timing of the AF, but conclude quite reasonably that the risk of TE’s is higher in non-anticoagulated people with AF than people who are.

360/16274 Anticoagulation        n=11 190 No Anticoagulation n=5084 NNT
0-30 days0-360 days 32 [0.28%]144 [1.28%] 54  [1%]216 [4.24%] 12934

Weigner et al looked specifically at incidence of thromboembolic events (TE’s) after cardioverting recent onset AF (<48 hours). They conducted a prospective observational study which was published in 19972.  They identified 375 adults with recent onset AF, 66.7% reverted spontaneously, the rest had some form of active management (electrical or pharmacological or both).  Three patients out of the 375 had a TE event, none of them received electrical cardioversion, all were managed with anticoagulation, and various drug combinations.  All 3 of these patients had essentially benign echoes during that admission, suggesting that the ‘safe’ (drugs and anticoagulation) approach isn’t actually that safe.  All 3 patients were elderly people that had multiple co-morbidities that even the most EC fanatical ED doctor would baulk at doing a cardioversion on.

A 2013 Finnish observational study of acute cardioversion5 (EC within 48 hours of AF onset) pegs the rate of thromboembolic events at less than 1%.  There were 38 thromboembolic events within 30 days in a total of 5116 cardioversions, in 2497 patients, this gives us an incidence rate of 1.5% per patient (or 0.7% per cardioversion).  The study goes on to suggest that using either the CHADS2 score or CHA2DS2-VASC score to risk stratify allows us to reduce that risk further if we only attempt to cardiovert low scoring individuals without anticoagulating them first (it remains to be shown if anticoagulating the high risk group actually reduces thromboembolic events).

afdecThey produce this brilliant decision tree, which I think I might use when discussing the options with patients.  The risk of thromboembolism with delayed elective cardioversion and anticoagulation is also about 1% (confirmed by our study into NOACs10 and the big Danish study7).

I can see myself talking patients through this decision tree to help them make an informed choice about their options.

The question about whether we should be offering this cardioverted group anticoagulation is difficult.  On the one hand the Finnish research suggests that CHADS2VASC scores or similar can help us estimate the risk, but the large study from Denmark seems to suggest blanket prescription (but neglected to sub categorise the time of onset of AF).

This may again come down to giving your patient the choice between nothing, something for 4 weeks and something for longer, depending on their CHADS2VASC score.  So the risk of TE’s following ED EC seems to be about 1%, which is similar to the risk of doing an in-hospital EC with anticoagulation.  So if the risk is the same, is one approach better than another?

Is electrical cardioversion more effective than chemical cardioversion for acute onset AF?

Strangely there is not much to go on here.  Most trials use combination therapy if they study EC at all.  I could only find 3 trials that were in any way useful, and only one compared EC to PC directly.

The first trial I found was a single centre randomised controlled trial that compared electricity alone to pharmacological cardioversion from Italy in 20111.  Patients were excluded if their CHADS2 score was >2, if they were unstable, or if it was felt that the AF was secondary to another process (eg Hyperthyroidism, Sepsis, ACS etc).  All patients had an ED transthoracic ECHO, looking for significant valvular disease.  However the authors neglect to say how many patients they excluded because of their ECHO findings.

N= 247 EC 121 PC 126
Success 108 [89.3%] 93 [73.8%] HR 0.34(CI 0.17-0.68 p= 0.02
Adverse Events 1 6 HR 2(0.22-18.29) p=0.29
Recurrence of AF @2/12 26.3% 

31 lost to follow up


52 lost to follow up

HR 0.9 CI 0.45-1.8 p = 0.86]


Patients in the EC arm were more likely to be successfully cardioverted than in the PC arm.  There were more adverse events (flutter, bradycardia and ischeamia) in the PC arm (but this was not significant).  The recurrence rates were broadly similar, but there were a lot more PC patients lost to follow up than EC, so this may have skewed the results in the favour of EC.

The other trial evidence is not as impressive.  Jacoby’s chart review9 compared 30 ED electrical cardioversions with ‘eligible’ controls.  This study also included people with known paroxysmal AF that were managed with a rhythm control strategy (so a few of them got shocked on more than one occasion).  The ED cardioversions were successful 97% of the time.  Follow up was only to 3 weeks, but found no thromboembolic events, however one patient did end up having a mitral valve replacement.  This study was primarily looking at length of stay, and cost per admission.  EC was found to have shorter lengths of stay (23 h vs 56 h), the median cost of admission was also significantly less in the EC group [$1,598 vs $4,274].  This was effectively a small observational study of relatively poor quality, but the conclusions agree with the Italian study.

stiellThis brings me on to my third study – Stiells Ottawa Aggressive protocol for AF6,4.  This was a consecutive cohort study looking at the safety and efficacy of using procainamide and then using electrical cardioversion if cardioversion had not been achieved.

This study included 628 patients with AF. Procainamide successfully cardioverted in 376 (60%) of patients. In the 252 patients where procainamide had failed, EC was attempted in 223 patients of which there was a 92% success rate (203 success, 20 fail). Cardioversion was not attempted in the remainder of the patients (29 patients).

No one was anticoagulated in this study if the onset time was less than 48 hours.  No one in the study had a stroke, no one died, and only 8.6% of patients relapsed, but the follow up was only 7 days.


So in an ED with the right skill set it’s perfectly possible to electrically cardiovert patients with acute onset AF with or without pharmacological cotherapy, in fact electricity might have a better success rate.  The risk of thromboembolic events is also similar to doing a delayed cardioversion with anticoagulation, but recent data suggests it may be worth considering prescribing anticoagulation for a short time afterwards and risk stratifying our patients based on CHADS2VASC score.  If their score is <2 I would offer them EC. If there score is >2 I think I’d try to limit their risk by admitting them for anticoagulation.

Pros and Cons

Pros to ED EC Cons to ED EC
As successful as in-hospital careSame relapse rate

Possibly slightly higher success rate

No side effects from rate control drugs

Risk of thromboembolism 1% same as in hospital care

Can go home faster

Hospital care still an option if ED EC fails

Cheaper for the hospital

Risks of procedural sedationMight not need it (up to 60% revert spontaneously)

Possibly higher risk of thromboembolism after EC if sent home with no anticoagulation

More resource intensive in the ED



  1. Bellone, Andrea, et al. “Cardioversion of acute atrial fibrillation in the emergency department: a prospective randomised trial.”Emergency Medicine Journal (2011): emj-2010.  []
  2. Weigner MJ, Caulfield TA, Danias PG, Silverman DI, Manning WJ. Risk for Clinical Thromboembolism Associated with Conversion to Sinus Rhythm in Patients with Atrial Fibrillation Lasting Less Than 48 Hours. Ann Intern Med. 1997;126:615-620. doi:10.7326/0003-4819-126-8-199704150-00005 []
  3. Berger, Marvin, and Paul Schweitzer. “Timing of thromboembolic events after electical cardioversion of atrial fibrillation or flutter: a retrospective analysis.”The American journal of cardiology12 (1998): 1545-1547.
  4. Safety of Urgent Cardioversion for Patients With Recent-Onset Atrial Fibrillation and Flutter Stiell, Ian G. et al. Canadian Journal of Cardiology , Volume 31 , Issue 3 , 239 – 241 []
  5. Airaksinen, KE Juhani, et al. “Thromboembolic complications after cardioversion of acute atrial fibrillation: the FinCV (Finnish CardioVersion) study.”Journal of the American College of Cardiology 13 (2013): 1187-1192. []
  6. Stiell, Ian G., et al. “Association of the Ottawa Aggressive Protocol with rapid discharge of emergency department patients with recent-onset atrial fibrillation or flutter.”Cjem 3 (2010): 181-91.  []
  7. Hansen, Morten Lock, et al. “Thromboembolic risk in 16 274 atrial fibrillation patients undergoing direct current cardioversion with and without oral anticoagulant therapy.”Europace 1 (2015): 18-23.
  8. Decker, Wyatt W., et al. “A prospective, randomized trial of an emergency department observation unit for acute onset atrial fibrillation.”Annals of emergency medicine 4 (2008): 322-328.
  9. Jacoby, Jeanne L., et al. “Synchronized emergency department cardioversion of atrial dysrhythmias saves time, money and resources.”The Journal of emergency medicine 1 (2005): 27-30.
  10. Nagarakanti, Rangadham, et al. “Dabigatran Versus Warfarin in Patients With Atrial Fibrillation An Analysis of Patients Undergoing Cardioversion.”Circulation2 (2011): 131-136.


Squeeze your face in, like you’ve sucked a lemon, then say the following ‘Early Goal Directed Therapy doesn’t work’  We’ve had three massive, well run RCTs published in the last 6 months which compared “usual care” to “EGDT” for severe sepsis.  They all show no difference between EGDT and “Usual Care”.

Headline figures

n 1600

796 EGDT

804 “Usual”


439 EGDT

446 Protocol Standard

456 Usual


630 EGDT

630 Usual


130 EGDT

133 Usual



147/792 18.6% EGDT

150/796 18.8% Usual


[90 day mortality]

129/405  31.9% EGDT

128/415 30.8% Protocol

139/412 33.7 % Usual

[90 day mortality]

184/623 29.5% EGDT

181/620 29.2% Usual


[90 day mortality]

50/130 44.3%EGDT

70/133 56.9% Usual


[60 day mortality]


All of the recent MCT RCT’s showed no significant difference between EGDT (the River’s protocol) and “Usual” care.

Pooling Results EGDT Vs “Usual”

n 4201

1865 EGDT

1890 Usual

Outcome Mortality 90 days 460/1865 24.66% EGDT

470/1890 24.86% Usual


So if there is no difference between EGDT and ‘normal’ care, can we just ditch the sepsis six, buffalo, or whatever acronym reminder is used in your department to manage sepsis early?  Some trusts are creating sepsis nurses, or sepsis champions who go around helping fight sepsis on the wards.  Is all this expenditure, time, and effort (for god sakes think of the sheer number of blood cultures alone).  Not worth it?

Well you could collapse down into a nihilistic funk if you want, but if you take a slightly longer view things are looking up for sepsis mortality, and this is probably because “Usual Care” now in 2015 is different to “Usual Care” in 2001.

I found a prospective meta-analysis looking at sepsis mortality between 1958 and 1997 (n= 10 694).  This gives a headline overall mortality of 49.7% but there was a trend to improving mortality.  So if we take 49% as a historical mortality, something has happened to drop mortality between 1997 and 2015 by almost half.

Mortality from Sepsis pre 1997 49.7%, mortality from Sepsis in 2015 25%.

In fact the news has got even better (at least in Australasia), a retrospective observational study from 2000 to 2012 in 171 ICUs in Australasia showed an absolute mortality decrease from 35% to 18.4%.  This at least agrees with ARISE’s slightly better numbers than PROCESS or PROMISE.  It might be interesting to find out why outcomes in Australasia are better than in the US and UK.

There are massive caveats to this, the 3 big RCTs excluded lots of patients for logistical and clinical issues.  Do we DNAR more people now than we used to?  A lot patients got excluded from all 3 trials because the treating physician felt ‘aggressive care was unsuitable’.  That’s going to skew the results slightly.

What do we do differently now?  What’s making the difference?  Antibiotics haven’t changed, if anything our ability to treat systemic bacterial infections has got a little less impressive given the rise of resistant organisms.  The widespread implementation of early warning scores, awareness raising, and early fluid resuscitation and early antibiotics are probably the biggest difference. Certainly the main component of the EGDT that’s been adopted (at least where I have worked) is the ‘E’.  In terms of critical care the same tools are used, just as they always were, the actual targets or ‘goals’ might be different, but intensivitists are still aiming to perfuse tissues and organs, and it’s entirely probable that they are keeping similar numbers in their heads as Emanuel Rivers did 15 years ago.  What we may have here is the biggest contamination confounding variable ever.

Maybe we can tentatively conclude that sepsis treatment has improved in the US, UK and Australasia.  The surviving sepsis campaign must surely take the lion’s share of the credit here, at least for giving hospitals implementation and treatment strategies, as well as raising the awareness of systemic infection.

There is much we still don’t understand, exactly how much fluid to give remains elusive, and the pathogenesis of septic shock on biochemical level is still poorly understood.  The FEAST trial suggested that we still don’t understand fully the effects of giving simple intravenous fluid boluses to people in shock.

In conclusion ARISE, PROMISE and PROCESS probably don’t show a benefit because EGDT is now very similar to “usual care”.


  • Rivers, Emanuel, et al. “Early goal-directed therapy in the treatment of severe sepsis and septic shock.”New England Journal of Medicine 19 (2001): 1368-1377.
  • Peake, Sandra L., et al. “Goal-directed resuscitation for patients with early septic shock.”The New England journal of medicine 16 (2014): 1496.
  •   Trial of Early, Goal-Directed Resuscitation for Septic Shock. (ProMISe Trial). N Engl J Med 2015;epublished March 17th
  • Friedman G, Silva E, Vincent JL (1998) Has the mortality of septic shock changed with time? Crit Care Med 26:2078–2086
  • Systemic Inflammatory Response Syndrome Criteria in Defining Severe Sepsis. N Engl J Med 2015;epublished March 17th
  • Yealy, Donald M., et al. “A randomized trial of protocol-based care for early septic shock.”The New England journal of medicine 18 (2014): 1683-1693.

The aggressive agitated elderly patient

The alarm has gone off in the CDU toilet.  Inside the toilet was an 87 year old man in only his hospital issue paper y-fronts demanding someone call the police because he is being “held against his will”.  He is refusing to leave the toilet, refusing to let you switch off the godawful crash alarm.  One nurse is hanging for dear life on the other side of the door to keep it open.  2 other patients with dementia in the same bay keep on wandering past, wanting to “help”.

What is your approach to the agitated patient?

There are 2 arms to any case such as this, first off we have to ensure the safety of the patient, other patients in the vicinity, other staff, and ourselves.  The next thing we need to establish is why has this person become agitated?

Medical causes of potential agitation are wide and varied.  Delirium in itself should be thought of as a medical emergency, as it carries a potentially high mortality (though estimates vary hugely 20%-70%).

This mnemonic may help – “SHED WIMP HELPS”

Substances, Hypoxia, Events (vascular), Dehydration, Withdrawl, Infection, Myocardial Infarction, Psychiatric, Head Injury , Endocrine, Low BM,  Pain, Seizures.

The key information here is going to be what you glean from the ambulance record, and whoever normally looks after your patient.  Is their behaviour normal for them?  Is it out of character?  If you are seeing that patient for the first time in the ED you are in the best position to assess them, as very often family members, carers, and other information will be lost prior to the RMO’s clerking them on the medical ward 6 or 7 hour later (on a good day).  If they are more confused than normal this does need investigating.

So say we’ve established that the patient has dementia, that they are in CDU for some form of social care assessment, and that to all intense and purposes there is no medical problem.

The best way to nurse someone with delirium or dementia is to keep them in a brightly lit room, with visual cues as to where they are, and what time it is.  It’s important to have consistent contact with the same individuals, such as carers from a nursing home, or the same nurse.  It’s important that the environment is calm, quiet, and free of interruptions.  Try to make sure that the patient has access to their hearing aids, and glasses, to avoid sensory disturbances.  There should be easy access to food and drink.  If you were going to design an environment to provoke a patient with dementia into confusion and agitation you’d design a place pretty much like an emergency department.

It’s noisy, there is no consistency of staff, patients are moved often, there are often absolutely no visual cues as to what time of day or night it is, if there are clocks they are always wrong.  The trolleys or wheelchairs are not comfortable, there is often no access to food or drink.

If they’ve trapped themselves in the toilet, or a running around the ward trying to escape you might feel under pressure to try some medication.  For elderly patients with dementia most of the literature tends to recommend either haloperidol or midazolam, but there is very little good quality data for elderly patients.  There is a Cochrane review of both antipsychotics and benzodiazepines in depression, they conclude that there is no difference in efficacy between atypical and typical antipsychotics, but no evidence that benzos help with non-alcohol related delirium.  So be cautious with both, haloperidol is thought to be safer because it doesn’t cause as much respiratory depression. IV doses are safer and more predictable if you can get a line in.

IF you can’t get a line in, this is what I’d put in my dart gun, either of…

Haloperidol 1-5mg IM Avoid if QTc is prolonged (risk of Torsade)Risk of dystonic reaction
Midazolam 2.5mg IM Risk of respiratory depression


After some time using de-escalation techniques, offers of food, drink, escape, to downright begging, we tried some IM medication.  I got punched in the face giving our gentleman 2.5mg of IM haloperidol.  About 5 minutes later his daughter arrived.  She took one look at her dad held out her hand, sighed, said “come over here Dad”, her father immediately let himself out of the toilet, and came to his daughter’s outstretched hand.

As she walked him back to his bed I heard him say “Thank god you came, I don’t know why they were keeping my trapped in that loo”

I love my job.


Paediatric Gastroenteritis

vomitWarning.  This is occasionally devolves into a bit of a rant, however it’s a rant with a sound evidence base.

Can we please give kids with gastroenteritis some anti-emetic?  If you happen to be reading this in the US, Canada, New Zealand or Austrailia where standard practice is a little different, I apologise.  Let me give you some background.

In the UK children who get gastroenteritis and come to ED get assessed, usually because parents have been trying and failing to hydrate them.  We tend to give them an oral fluid challenge.  Usually ORS at 5 or 10mls ever 5 or 10 minutes.  They get a full history, examination, their hydration status is documented and usually we wait for them to urinate.  If they don’t vomit, urinate (and the urine dip is okay), we shout hurrah, fist bump (or would but we are British) and send the parent on their way with a prescription or advice to get or make ORS, and to continue with the little and often amount of fluid required.

If they ‘fail’ this challenge by vomiting we are often forced to admit them, and continue with the cycle until they wee.  Paediatric admission units have one or two of these children on the go at any one time, and we just keep ploughing on until the child stops vomiting, or they get dehydrated enough to require NG or IV fluid.

For some strange reasons in this group of patients there is a lot of resistance to trying an oral fluid challenge with an anti-emetic.

I have yet to hear a coherent reason as to why this but the commonest one is that it ‘might mask symptoms’.  I can’t understand why this could be the case, as anti-emetics work by blocking receptors in the CTZ.  I’m not sure how this would stop vomiting secondary to some other serious disease process.  If someone has a closed head injury like a subdural they are going to continue to vomit no matter what you do, and will have other signs.  Similarly if a child has a metabolic disorder, their BM is going to be low (or really ‘freakin’ high), or there are going to be other clues in the history.  Also I’m suggesting giving a dose of anti-emetic to kids that we have diagnosed with acute gastroenteritis, which implies that you have assessed the patient, taken a history and examined them.  The anti-emetic the literature seems to favour is ondansetron.

So will we miss something? 

Looking for evidence for this is tricky, as it’s hard to prove a negative, especially when the ‘things we might miss’ are very rare metabolic disorders.  There is some research that backs me up.

Sturm, Jesse J., et al. “Ondansetron use in the pediatric emergency department and effects on hospitalization and return rates: are we masking alternative diagnoses?.” Annals of emergency medicine 55.5 (2010): 415-422.

Sturm conducted a retrospective review of visits to paediatric EDs in Atlanta, USA, between 2005 and 2007.  34 117 charts were reviewed, and ondansetron was used for 19857 patients.  They found that there was no significant change in the diagnosis at discharge between children given ondansetron and those who weren’t, they were also less likely to be admitted.  Children who were given ondansetron were more likely to return, and then be readmitted, but the admission rate globally was less in the ondansetron group than the nothing group.

Okay so it means we probably won’t miss anything but does it actually work?

YES. – NNT is about 5. That’s better than steroids in COPD (NNT 10) and Aspirin in STEMI (NNT 42).

Well the key single RCT was published in the NEJM in 2006, this was a prospective, double blind randomized controlled trial.

P 215 children 6 month – 10 years in the Paeds ED with gastroenteritis AND mild dehydration.
I 1 single dose of orally disintegrating ondansetron
O Primary:Proportion who vomiting while having rehydration


  1. Number of vomits,
  2. Incidence of IV rehydration
  3. Admission rates.
Primary:14% vs 35% RR 0.4 95% CI 0.26-0.61


  1.  0.18 mean vomits Vs 0.65 p<0.0001
  2. 14 % Vs 31% RR 0.46 CI 0.26-0.79 p=0.003
  3. 4% Vs 5% not significant


You can get the study here.

What do these results mean?  Well it looks like the group that were given a single dose of ondansetron we more likely to pass their fluid challenge, less likely to need IV therapy, but were not necessarily more likely to go home.  I like this last result.  I think it means that if a child was still dehydrated, and needed further observation that this was what was happening rather than taking false reassurance from being given a medication.

Children were given 2mg PO ondansetron 8-15kg, 4mg 15-30kg, 8mg if >30kg.

Caveats?  The children participating in the study were assigned a dehydration score by a single rater, which was based on largely clinical, and subjective measures such as skin turgor.  The dehydration score they used is pretty much the same table as exists in paeds textbooks and APLS manuals so seems a reasonable method to use however, it does introduce the potential for bias.  The other concern for me about this study was the number of patients that were excluded prior to randomization, 3067 children were considered but only 243 were asked to enrol, another potential source for bias.

This study on it’s own yields a NNT of 5.  We need to treat 5 children with acute gastroenteritis with ondansetron to stop 1 kid vomiting.

This is all very promising but is there any other data to support it’s use?

I’m glad you asked….

Then a Cochrane review was published in the BMJ in 2012 which looked at the literature from 1980 to 2012.  Ondansetron They found 10 studies and compared ondansetron (oral, and IV) to granestron, dexamethasone, and other antiemetics.  They looked at ondansetron vs placebo for cessation of vomiting, initiation of IV rehydration and hospital admission rates.  Now the review could only find 4 studies looking at the effectiveness of PO Ondansetron vs placebo, but their headline result for cessation of vomiting was RR 1.44 95% CI 1.29-1.61 NNT = 4.  One study was reliable but VERY pro ondansetron and threw the results out a bit, but with that excluded you still got an impress result RR 1.33 9%% CI 1.19- 1.49 NNT =5.  There was no statistically significant difference in hospitalization rates within 72 hours, suggesting that children’s admission might be delayed rather than avoided if you trial them ondansetron.  They did manage to find a reduction in resorting to IV rehydration [RR 0.57 NNT 6].

SO there is good evidence to suggest that it is safe and effective to use oral ondansetron in a vomiting child with gastroenteritis, we will probably decrease length of stay, increase success of oral rehydration, and maybe save some money for the trust.  No one will thank you for it though…

…apart from the kid’s parent, oh and the kid.

Rant ends.



  • Carter, Ben, and Zbys Fedorowicz. “Antiemetic treatment for acute gastroenteritis in children: an updated Cochrane systematic review with meta-analysis and mixed treatment comparison in a Bayesian framework.” BMJ open2.4 (2012).
  • Sturm, Jesse J., et al. “Ondansetron use in the pediatric emergency department and effects on hospitalization and return rates: are we masking alternative diagnoses?.” Annals of emergency medicine 55.5 (2010): 415-422.
  • Freedman, Stephen B., et al. “Oral ondansetron for gastroenteritis in a pediatric emergency department

At the end of an algorithm with no where to go..

Pre-alert at 4am: “Status 1 23 yo male, seizures, now GCS 3, BM 6.4, HR 130 BP 90/60 Sats 95% on 15L, NPA insitu.  Given 15mg IM midazolam.  Seizures have stopped ETA 10 minutes”

On arrival, slightly sheepish paramedic, also says ‘errr he’s in VT but we only found out as we were driving in’

On arrival I do a quick assessment, while the boss takes over the airway.

A, patent, NPA insitu, B equal AE, sats 95% on 15LNRB, good respiratory effort, C BP 80/50 HR 180, monitor looks like broad complex tachy.

1 x 200 J synched shock – Still VT.

I look to the boss  “s***” we say together.

Here is the ECG:

click to enlarge

What are you going to do?




We got an anaesthetist to manage the airway, and we gave a series of further shocks, and changed the positions of the pads, and we loaded him with Amiodarone.

I pulled up his past medical history, he was involved in a RTC when he was a teenager, had had ORIFs of his right tib and fib, and right femur.  No heart conditions.

His girlfriend, who was (understandably) distraught said he was not on any regular medications, and though he had been depressed in the past, wasn’t depressed now.  I then asked her if she was on any medication he might have taken.  She was taking dothiepin (a TCA).

He’d been in the department now for 30 minutes, still in VT, still not responding to the medications we had.  In the end we decided that he was behaving like a TCA overdose, and we had enough evidence to try treating him as such.

The treatment for TCA OD is sodium bicarbonate, as it provides a sodium load for the Na+ channel blocking affect of the TCA (which is why TCAs affect cardiac conduction).

Start with 50ml of 8.4% NaHCO3- over 15 minutes.  We used about 6 vials of this stuff, and his ECG turned into this:

click to enlarge
click to enlarge

After an infusion of HCO3- 4 hours later his ECG looked like this:


About TCA OD:

– TCAs – [STEM]”line” or [STEM]”mine”.  Commonest UK ones –  Amitriptyline, Nortriptyline.

TCAs [tricyclics – 3 actions] block the reuptake of noradrenaline centrally and peripherally, they have an anticholinergic effect, and also block Na channels in brain and myocardium.

Patients who have taken TCAs get anticholinergic effects: warm, dry, hyperthermia, tachycardic, agitation, blurred vision, hallucinations, urinary retention, ileus.  At higher concentrations the symptoms get worse, then affect their consciousness level, then their cardiac conduction.

All ODs need an ECG, look for a long QRS; >100ms increases the risk of fits and coma.  Greater than 160ms is associated with ventricular arrhythmias.

Treatment (ABCs as given)

50mmol of 8.4% over 15 minutes, repeated as required, infusions of bicarbonate can be given.  If you’ve ventilated the patient you can also hypoventilate them to a pH of 7.45-7.5, which seems to decrease the affinity of the TCA to myocardial Na+ channels.


What happens if the bicarb isn’t working?  [thanks to @bedrocteam] for pointing this out:

– Mg2+ is recommended if the bicarbonate isn’t working, though the evidence is getting quite thin at this point[1].  There is also case reports of intralipid being used, when all else has failed [2].

  1. Emamhadi, Mohammadali, Babak Mostafazadeh, and Marzieh Hassanijirdehi. “Tricyclic antidepressant poisoning treated by magnesium sulfate: a randomized, clinical trial.” Drug and chemical toxicology 35.3 (2012): 300-303.
  2. Engels, Paul T., and Jonathan S. Davidow. “Intravenous fat emulsion to reverse haemodynamic instability from intentional amitriptyline overdose.” Resuscitation81.8 (2010): 1037-1039.

A tale of 2 dissections…

Thoracic dissection is a diagnostic challenge to say the least.  Coming from a relatively CT-frugal ED background, it’s not normally in many nature to order CTA’s in people who have pain ‘going through to their back’.  Mostly because I’d order about 9 million a day (that may be a slight overestimate).  However I had a ‘remember that lady’ moment a few months ago from a very good friend on the MAU team.  This good friend is a very able colleague (certainly smarter than me), told me about a case he missed, a case I sent him.  His choice of words was interesting because he didn’t imply I had missed it too (though I feel I had).

It was busy, I was the boss on overnight.  I was trying to make some room in my department, so I was trying to ‘improve flow’ now I don’t care how perfect a doctor you are; we all do this.  In about 30 minutes I’d sent a non-toxic OD’s to the mental health team, and someone with flu home.  Then I went to see this delightful, charming 89 year old lady, who had some chest pain, unfortunately about 3 minutes into the consultation I was grabbed to go and calm down a drunk fat girl with a personality disorder and a cut foot, who had been ‘disrespected’ by the security staff.  I sewed up her ankle while she simulated oral sex on the handle of bay operating light.

After finishing that I went back to my 89 year old.  She had developed chest pain whilst walking up some stairs, the pain had lasted maybe a minute, she told me she nearly blacked out.  The family called her an ambulance.  I noticed from ambulance sheet, her initial BP was 60/40, but it had rapidly improved to a ‘normal figure’ 145/80 while she was with us.  She’d had vasovagals before she told me, and apart from treatment for hypertension, and IHD.  ECG –  flattened lateral T’s, nothing exciting.  The CXR, showed a widened mediastinum, which I noticed, I then looked back at her old CXRs, and I saw one from a month ago that ‘looked the same’ (she had a widened ectatic, unfolded aorta).  So off she went to medicine for a 12 hour troponin, and someone to stop her bendroflumethiazide.  My friend clerked her in, and did the same.  All the way through this she was pain free, and her obs were totally unremarkable.


A CXR with a widened ectatic thoracic aorta.
A CXR with a widened ectatic thoracic aorta. (this is not her CXR, but is close enough for the purposes of this case)

The CXR was reported that morning as showing a widened mediastinum, the medical team arranged urgent CTA and a large type B dissection was noted on the scan.  She went for interventional radiology EVAR, her aorta was perforated, and she died in ICU later that day.

Now I find out about this through my friend, on the grapevine.  The patient was the subject of an M+M meeting which I didn’t attend (as in my current institution we keep the ED M+M and Medicine M+M separate).  I felt awful about this, I’d missed the diagnosis, my colleague had missed the diagnosis, and had we managed to pick it up would she have survived?  It’s difficult to say.  There is evidence that suggests for every hour a diagnosis of AAD is missed the mortality goes up by 1%, however the mainstay of treatment in this group is aggressive BP control, and her BP was normal to low.  Her history was not pathognomonic for AAD, straining at stool (like King George II), chest pain radiating through to back, but I had the CXR which was also abnormal, but not normally abnormal.

On reflection, I made at least 2 cognitive errors; Attribution bias (I saw her old CXR, which showed an ectatic aorta, and I felt her current CXR showed a continuation of that process); and confirmation bias – I felt that the St John low BP finding confirmed my thought that she had had a vasovagal or angina from exertion, neglecting to fit this in with the pain that she very well described to me.

The second case occurred on a busy Saturday evening a few weeks later.  I had picked up a card for a 29 year old gentleman in our monitored bay that said “Leg pain”.  I started looking through his old clinic letters, and scans (a wonderful byproduct of where I work’s beautifully integrated IT system), and I found that he was being treated for idiopathic hypertension.  This letter from a renal physician commented on a normal MRI of renal arteries, drilling down to that report referenced an MRI of his heart which showed a dilated aortic root.  I clicked through a few DNA’s on the computer system, and as I was pondering if his hypertension could be related to his leg pain.  I got grabbed by one of my colleagues who had noticed this guy was being a bit odd.

The guy worked as a woodsman, and had been brought in by his friend.  His friend said he’d come home, complained of pain in his leg, got sweaty and confused, so he’d brought him here.  Now we were all thinking ‘toxins’ at this point but he pointed to his chest, and his belly and said ‘hurts’ then kept trying to sit up, move around and generally make himself more comfortable.  He looked awful.  That was the most history I got from him, after that he had a profound expressive and receptive dysphasia.  The patient’s flatmate, in one of those wonderful moments of honesty, told me he was the only one in the flat that didn’t take synthetic cannabinoids.

This man’s BP was 220/160 he had all of his pulses, no delays, a normal ECG, he got a portable CXR which showed a normal mediastinum, to my eye (and the radiologist the next morning).  His left leg was a little cooler perhaps than his right.


An example of a normal CXR
An example of a normal CXR, (this is not his CXR, but it is close enough for the purposes of this case)

I was pretty worried about this gentleman, and asked for my first ever CTA, then I grabbed one of the ED consultants to have a look at him with me.   She also suggested we start banging in labetalol at this point so we did (it didn’t really work).  I took him up to scan, and after a few attempts to get him to lie flat and stay still we got our diagnostic images (with very very small doses of midazolam).  They showed a type A dissection.

He came back down to ED, and went to theatre about 25 minutes later.  He survived his repair without having to have a valve replacement, but did have some embolic events post op (and probably pre-op), but he’s alive, and independent.

Now I had these cases within about 3 weeks of each other, I’m not sure, but I wonder if my failures with the first, primed me to pick up the second.  I certainly did a fair amount of reading on dissection after I found out about my error.  What has struck me about that is how difficult and nebulous the symptoms can be, but it seems any combination of acute onset pain, and neurology could be a dissection’s only symptoms (and you may not even get that).


About Acute Aortic Dissection

Clinical Features % of cases
Severe/worst ever 90%
‘abrupt’ 84-90%
Sharp 64%
Tearing 50%
Migrating 16%
Down the back 46%
Differential BP >20 mmH in arms or missing pulse 15-30%
Altered or syncope 13%
Hemiplegia 5%
Focal neurology 17%
Abdo pain 43% (descending0 22% (ascending)



CXR: Widened mediastinum in 56-63% of patients.  Abnormal aortic contour 48-71%.

Transthoracic echo 75% diagnostic type A, 40% type B

CT sensitivity is 83-98% (but probably better than this now due to high res scanners).

I think the clinical features make more sense if you imagine the process going on in your patient.  The pain is from the initial tear, the pulse differential, or BP differential is due to the involvement of the arch vessels, and depends entirely on the physical shape of the tear.  It’s the same with neurological symptoms, you might get hemispheric signs if a major vessel is blocked, or you might get random neurology that doesn’t make sense because of embolic phenomena (which is by it’s nature random).  As the tear elongates it will affect lower branches, giving appropriate syndromes (renal failure/infarction, mesenteric ischaemia).


There is some discussion in the literature of using d-dimers has a ‘rule-out’ test for dissection.  As you are forming clot inside the false lumen, in theory the d-dimer should be pretty high.  Certainly a negative d-dimer might rule out a dissection, but I don’t think there are any prospective trials out there to say for definite that this is a safe strategy.   I did find a recent meta-analysis (of observational studies) with approximately 500 participants which suggests it might be sensitive enough to use as a rule out test, (this is quite an interesting topic, and I think I’ll revisit it later).


Type A Vs Type B



Type A dissections need go to theatre, patients need to have either an AVR, and graft, or just a graft to pin the false lumen back.  Other options the cardiothoracic surgeon has is to fenestrate the false lumen (basically cut through it) which allows blood to flow through both lumens.  Type B dissections (which start after the arch) used to be controlled with BP control only, sometimes some centres are using large stents, to push the false lumen closed.  The theory behind controlling the BP is to decrease the pressure flowing through the false lumen and stop the dissection tearing any further.

Management in the ED is mostly around identifying the problem and making the diagnosis, it’s an often missed or delayed diagnosis because the features are so nebulous and changeable, for us you should ask yourself “Could this be a dissection?” for every one of your chest pains.  If you do suspect it, talk to a boss, and consider more investigation.

Once you’ve found one, start lowing the BP, labetalol as a bolus and infusion is recommended in most centres, GTN infusions can also be used (or used together).  Nitroprusside is also still in the textbooks, I’ve never seen this used, but I’m sure other people have.  Aim for a target of <140/90.  Slipping in an arterial line would also be a good idea while the cardiothoracic people are getting ready for theatre.  It will aid your BP management, and make the anaesthetist moderately grateful (as it will make their RSI safer).

Further information

–          Excellent SMACC podcast on dissection from Rob Rogers

–          LIFTL review – here



Useful literature and references

Shimony, Avi, et al. “Meta-analysis of usefulness of d-dimer to diagnose acute aortic dissection.” The American journal of cardiology 107.8 (2011): 1227-1234.

Harris, Kevin M., et al. “Correlates of Delayed Recognition and Treatment of Acute Type A Aortic Dissection The International Registry of Acute Aortic Dissection (IRAD).” Circulation 124.18 (2011): 1911-1918.

Howard, Dominic PJ, et al. “Population-based study of incidence and outcome of acute aortic dissection and pre-morbid risk-factor control: 10-year results from the Oxford vascular study.” Circulation (2013): CIRCULATIONAHA-112.

Howard, Dominic PJ, et al. “Incidence, risk factors, outcome and projected future burden of acute aortic dissection.” Annals of Cardiothoracic Surgery 3.3 (2014): 278-284.

Coyle, Siobhan, et al. “Diagnostic Testing in Acute Aortic Dissection.” Current Emergency and Hospital Medicine Reports 2.2 (2014): 97-103.

Things I’m going to miss…

As I am sat in Christchurch airports departure lounge, I’m reflecting on my time here.  I’ve loved my time away, and there are some things they do here in Christchurch that if I could slip into my carry on and bring to the UK I would.

1.  Departmental radiologists

no radiologists were harmed in the taking of this photograph

This is our handover room, and in it sits one of 3 full time radiology consultants who do real time reporting and scan protocolling for the ED.  I have seen them leave this room, come out to talk to us about a finding, and EVEN examine patients.  They are truly awesome, and every Friday they lead us through a greatest hits of our radiology cases from the week. 


2.  Full time departmental Social Workers, alongside physio and OT

These guys activate emergency care packages, inform people of voluntary services, help with child protection, vulnerable adult protection, they are present during major trauma  Every morning and afternoon, if someone needs help getting back onto their feet, I can call “the MDT” and get my patients MOVING.  If they can’t get people home, they get them into respite care or a residential home (with the social workers help).


Friends of the Emergency Department

3.  Volunteers

“Friends of the Emergency Department” or FEDS, talk to and listen to patients, their families, their children.  Grab cups of tea, snack boxes, they make the place so much more civilised, and make the ED more pleasant (and less frightening for the patients).


4.  Consistent teams!!

We are rota’d together in teams of 7 doctors.  I know each individuals strengths, and interests so if someone has a thing for paediatric emergency medicine, or if someone is a proto-plastic surgeon and likes suturing, when it’s busy we can divvy up the tasks, swap patients and keep the department flowing.  As I’m working with these people day in day out, I’ve got the incentive to train them well, and upskill them as much as possible, and they can teach me what they’ve learnt from their other placements. THIS SHOULD BE STANDARD IN EVERY DEPARTMENT.


5.  An integrated IT system

concertoThe CDHB uses something called concerto which means I can look through old discharge summaries, clinic notes, ECGs, bloods, Xray reports, outpatient pharmacy prescribing (both what’s dispensed and what’s prescribed), and GP diagnoses.  I can have a patients old ECG, and cardiology clinic letter printed out and by their bedside to compare to their new ECG before they arrive.



6.  This is called a nitrous mixer.

It is great for procedural sedation and analgesia, as people don’t need to coordinate the ‘sucking-in’ from an on-demand valve during the painful bit.  It’s useful for kids too.



7.  This is called Topicaine Gel.  It is awesome for wound analgesia (and haemostasis) for kids and adults.


8.  Acute Demand (intergration with outpatient diagnostics and primary care)

Outpatient IVs?  Outpatient Biliary USS? Home visits?  Home GP Visit?  If they can keep you out of an acute inpatient bed.  They will move heaven and earth to do so. 24/7, just a phone call away (and present at morning and afternoon handover).  This is all funded by the local health authority and saves them millions (patients love it too).  Who follows these up?  The patients own GP, if they don’t have a GP, Acute Demand’s GPs follow them up.


%d bloggers like this: