If you build it, will they come (back)?

This is a review of 2.5 years worth of lower back pain.  Basically, every single patient with an ICD-9 code had their chart reviewed to ask the question, “If you got an MRI in the ED, are you less likely to return?”

They specifically did not look at yield, rather, just whether or not patients initially thought to have plain, old back pain came back.
The answer, contrary to those of us who think more is more, is no.   Repeat visits for those that had an MRI was 4.3%, repeat visits for those without an MRI at initial visit- 4.6%
However, it does give you some insight into who returns – they were more likely male (5.9% vs 2.8%) – further demonstrating that males are in fact the weaker sex.  They were more likely English speaking (4.9% vs 2.1%) and, speaking of which, non-English speakers were less likely to have an MRI (12.2% vs 7.4%), and self-pay patients were also less likely to have an MRI (9.8% vs 5.1%).   Its unclear if this further demonstrates a touch of bias on the provider side, if English-speaking patients speak up more about wanting diagnostics, or if non-English speakers just go to another hospital since this was a single center study.
As a secondary measure, they also looked at disposition for these patients.  Only 16% were discharged after MRI (vs 82% of those without MRI), 74% of those MRI’ed were placed in observation status (vs 11% of those not), 9% of those MRI’ed were admitted vs 3%, and 4% of those not MRI’ed left prior to completing treatment. Now, for some fancy math.
of the 797 patients who received an MRI over this time period, if only 11% were placed in obs status (rather than 74%), that is only 88 patients rather than 591 patients placed in obs status. that’s 503 patients with an unnecessarily padded bill. At $35.8013 per RVU, were talking tens of thousands of dollars billed that’s wasted.
Or, to put it another way, if you are in an area with a good payor mix, why not just obs & MRI them all?
Mythbusting, Improving Outcomes, Radiology, Radiology

Spinal Abscess: The Baystate Review

This is a review of all spinal abscesses at Baystate (total 162), from 2005 – 2015.  They compare 88 randomly selected controls whom had similar ICD-codes less the spinal abscess plus an MRI that was negative for acute infectious process. 

Interesting take home points, much of which is consistent with prior (albeit scant) literature:

-73% of patients are over age 50.

-more likely to have their second visit (50.6% vs 29.6% of controls) – though this 50.6% of patients with a second visit is surprisingly low for me – no word on how many were sent home from the ED, and had an MRI as an outpatient that were not included in this calculation.; or maybe we’re getting better at finding the needle in the haystack?  Or maybe we’re MRI’ing everyone?

-Many received antibiotics within the month: (35.2% vs 6.8% of controls) – this signifies a huge red flag for me.  If a patient revisits the ED and recently had pyelo (or anything infectious really), and now presents with back pain, probe a bit more for the possibility of vertebral osteo or discitis. 

-percentage of patients with history of IVDA: 20.4% vs 4.6% … this number seems low, but also is somewhat in line with prior studies – thus making me wonder how many I’ve missed…

– percentage of patients with alcoholism with a spinal abscess: 19% vs 8% – the more I get interested in ID, the more I realize that alcoholism is basically a form of immunosuppression.

-percentage of spinal abscess patients with obesity 21.6% vs 2.3%; I’m surprised only 2.3% of controls were obese.  Not sure what role this plays as being a diabetic in and of itself was not associated with a higher increased risk in this study.

-fever was present 62.4% in those with a spinal abscess vs 13.6% of those without; this includes self reported fever, which I have to wonder how often we sweep this aside when the patient is afebrile in the ED.

-16% had no identifiable risk factors; a third of the patients  presented with back pain, fever, neurologic deficits vs 6%

-Other symptoms and signs related to potential spinal cord impingement were seen with similar frequencies and of similar durations among cases and controls- meaning, focal deficits seen in both groups.

-noncontiguous co-infection: 53.7% of time (pneumonia, distant osteo, endocarditis… of those with a co-infection, 20% had more than one).

-blood cultures were positive 63.4% of the time, and >75% of the time it was staph Aureus. 

-Majority of lesions were found in the L-spine at 56.2%  – which means almost half are elsewhere!

-while “admits” for spinal abscess were up from 2.5 to 8 in 10,000 admissions from 2005 to 2015, I have to believe that number is somewhat inflated as admits like chest pain, pneumonia and renal colic probably decreased, while MRI became more readily available. 

All in all, this paper is pretty much in line with others on this topic, and strengthens the signal a bit for certain key points: a good number of spinal abscesses are not in the L-spine; many patients are older than you think, and, among other things: its more than just IVDA. 

Critical Care, Improving Outcomes, Mythbusting

Procalcitonin: Holy Grail, or Holy Sh*t ?

Procalcitonin is marketed as, “a marker of broad routine use, both for differential diagnosis of bacterial infection as well as for antibiotic stewardship.

But is it?  This study looks at 107 ICUs that had >25 sepsis cases in 2012, and had an ability to perform procalcitonin (PCT) levels on their septic patients, and essentially looked to compare the outcomes of those that had PCT ordered and those that did not.  All in all, there were about 17,000 septic patients without a PCT ordered, and about 3800 patients with a slightly lighter wallet and slightly more anemic after their admission than their comparators.

There was little difference in baseline characteristics – save for those having PCT ordered more likely hailing from the West (27.9% of PCT orders vs 12.7% of those not getting PCT ordered) and the opposite holding true for the South (55.3% without vs 49% with PCT).  PCT was slightly less ordered at teaching facilities (37.8% of septic patients without PCT orders vs 31.9% of those with a PCT ordered).  All other OR were <1.25.

There was no difference in length of stay and no differences in mortality.

There was an increase in days of antibiotic treatment for those in whom a PCT was ordered (relative risk increase 1.17), and with that an accompanying increase in Cdiff (OR 1.42) .  Of course, 1 PCT begets another (33% of the time, and about 3 days later).  Patients with serial PCT orders had higher rates of antibiotic use, higher Cdiff, and again, no mortality benefit.

Stop the madness.  Indiscriminately ordering tests that will not change management should not be done.  And they certainly should not be repeated.


Ditch the steroids for urticaria

Summer is here. People are going out and eating strange food, getting stung by strange things, and in general, coming into the ED for interesting rashes.

Contrary to a previous post suggesting a longer taper for contact dermatitis, this paper suggests that steroids are well, quite bunk for urticarial rashes.

This French, two-center, randomized, double-blinded study look at 5mg of levocetirizine vs 5mg of levocetirizine plus 4 days of 40mg prednisone for acute urticaria of under 24 hours duration. Patients were excluded if they had angioedema or anaphylaxis – among a multitude of other things, such that only 100 of 710 potentially eligible patients were enrolled (exclusion criteria comprised of: a history of diabetes, peptic ulcer disease, pregnancy, and chronic kidney disease, among many others).

The authors note that in Italy (strangely, why not in France?) , 93% of patients presenting to the ED for urticaria received steroids in 2011 – which I imagine is probably slightly higher than what I would expect for US EDs. So they reasonably suggest that, hey, maybe we can lower steroid-related ADR’s if we find a way to decrease the number of times we prescribe them.

Basically, after 1 hour, 2 days, and 5 days, there was no difference in itch control or rash resolution between the two groups, and if anything, a trend towards better outcomes with levocitirizine alone (For instance, 62% with an itch score of zero for steroids + levocitirizine vs 76% in levocitirizine only at 2 days). Relapse rates were 30% vs 24% (steroids+levo vs levo only) during the first 5 days, which was not statistically significant.

Sadly, and oddly, they report that “one patient in the placebo group had vomiting and abdominal pain in the ED 1 hour after initiation of treatment. These symptoms were related to anaphylaxis, and the patient rapidly improved after a steroid injection.” … which as we know, probably was not anaphylaxis if it improved “rapidly” after steroids (and without adrenaline). But I digress…

Previously, there have only been two studies looking at steroid use for urticaria; This one showing earlier resolution with 3 days of prednisolene (unsure if this is prednisolone, prednisone, or something else) against 10mg of loratadine (94% vs 66% resolution at 3 days), and this one showing lower itch scores with 4 days of 40mg prednisone plus 25mg of hydroxyzine every 4-8 hours than hydroxyzine treatment alone.

I think this particular study lends some credibility to those of us that are steroid-averse, and probably lends to a reasonable discussion with patients that are concerned about getting steroids. Likewise, it is probably reasonable to do a single dose dexamethasone treatement and chase it with a short course of levocitirizine.


TXA’s post-partum fragility

Warning: An American on TXA.

The WOMAN trial was an international randomized double-blind placebo controlled trial across 193 hospitals in 21 countries looking at TXA use for morbidity and mortality for post-partum hemorrhage.

First off, this was a massive undertaking as this was an enormous trial – about 10,000 patients in each treatment arm, with fantastic follow up- only 31 of about 20,060 total patients were lost to follow up. If the treating provider was unclear as to the utility of TXA for post-partum hemorrhage control, the patient was randomized 1:1 to placebo or 1g of TXA, and if the bleeding continued after 30 min or stopped and restarted within 24 hours of the first dose, a second dose (1g TXA) or placebo was given. Baseline characteristics were quite similar between the placebo and TXA arms.

So since social media is clamoring, and since this is published in the Lancet, there must be a mortality benefit, right?

This is where I point out three things:

1) there was no difference in mortality (2.3% vs 2.6% – favoring placebo)

2) TXA had 0.4% fewer patients who experienced death due to post-partum hemorrhage

3) this was accompanied by a p value of 0.045, and a fragility index of…..

Screen Shot 2017-04-27 at 10.58.18 PM


A study of 20,000+ patients, and a p-value of 0.045, and a fragility index of zero.  And again, the most patient-centered outcome possible- mortality – favored placebo.

TXA is not the magic bullet in this instance.  There is a weak signal of benefit if you are proceeding to laparotomy for bleeding – particularly for caesaean delivery (1.5% vs 2.4% mortality benefit, fragility index of 4), but that is an exploratory analysis needing further review, otherwise, this is a flimsy trial.  While TXA remains inexpensive, it is worth a go- particularly in low-resource areas after laparotomy (provided TXA is still inexpensive there), but by no means does a clinician not giving it deserve to be chided – the evidence is incredibly fragile and not worthy of social media’s  “life-saving” claims – at least for post-partum hemorrhage.

Cardiology, Improving Outcomes, Mythbusting

SVT: treat, wait, re-evaluate

What do you *really* need to do with your SVT patients? Well, this is a retrospective observational study of 633 consecutive SVT patients over 10 years seen in a single ED. This was more hypothesis generating than anything – they basically provide patient characteristics and try to tease out if labs / imaging were necessary.

Their mean age was 55, 62% of patients were female, 55% had prior SVT history, 31% had at least one cardiovascular risk factors (dyslipidemia, hypertension, diabetes, CHF, or vascular disease), and 9% had ischemic heart disease.

Some interesting lab nuggets:

-0.4% had a hemoglobin < 8g/L

-1.5% had a sodium >150 mmol/L, none <126

-no patient with severe hyperthyroidism

Chest Xray was obtained 30% of the time, and while it was abnormal 21.6% of the time (41 of 190), none of the time did it alter ED treatment – despite showing 14 cases of pulmonary edema, 4 cases of pneumonia, and 3 pleural effusions.

The authors conclude that patients with uncomplicated SVT are over-investigated, and that most have normal or near-normal results. While I tend to agree – for the 25 year old in SVT without a concerning story – the 55 year old diaphoretic (14% were diaphoretic) female with ischemic heart disease I’m going to work up. Chest films were only ordered on 30% of these patients – frankly in a US hospital, I’m thankful its not higher.

I know Billy Mallon loves his TSH, but why not get a better history to see if there are other concerning symptoms before sending off TSH… Speaking of which, maybe we could decrease those Chest films if we fixed the patient a bit, then reassessed to see if imaging is wanted. (ie, are you still short of breath?).

Finally, I think this study is plagued by premature closure, as they only searched for cases with a discharge diagnosis of paroxysmal supraventricular tachycardia. They’re likely missing at least a few patients who came in with SVT and were found to have actually have another diagnosis.

Ultimately, while this study should not change practice by any means, it should give us pause before shotgunning labs & chest films until after we treat the patient, re-evaluate, and get a better history. This could probably be said for many other diagnoses besides SVT.


Let’s stop the sepsis high-five.

82 y/o F from SNF, AMS, “foul smelling urine.”

80/50, 103.1 PR, 120HR 98%.

An initial POCUS showed a collapsing IVC, you give 30 cc/kg LR, 650 Tylenol PR, vancomycin loading dose, and 3.375g Zosyn. BP is now 110/74, HR is 80 bpm, labs show an obvious UTI, and you call to admit the patient.

Or as I like to call this, the “sepsis-high five.”


This is a single center study of 828 patients looking at time delay between first and second antibiotic. They broke it down to 6, 8, 12, and 24 hour drugs, and considered a delay of 25% significant (90 minutes for a q6 hour drug, 180 minutes for a BID drug, etc). The primary outcome being in-hospital mortality.

So what’d they find? Well, unsurprisingly, they found that 72% of patients had a delay with 6hr dosing between their first dose and second dose compared to 47% for Q8hr antibiotics, and 25% for Q12hr antibiotics. (<5% for Q24hr drugs- but they had 6 hours to hang the dang drug!)… They acknowledge several issues: we often order a one time dose in the ED, but the upstairs care only knows Q6hr drugs at 12-6-12-6 dosing. I think this is most likely the case in this study, since the next dose for a 6, 8, and 12 hour drug were 9.55, 9.6, and 10.6hr respectively. They also acknowledge that it is possible that delayed second antibiotics are not inherently contributory to adverse patient outcomes, but simply a surrogate for patients who generally received less attention and care overall, particularly given 43% of delays occurred in ED boarding in their institute.

Paradoxically, those that received the q6hour antibiotic (cough, ZoSyn, cough), had high rates of adherence to the sepsis bundle (fluid loading, early pressors, early abx, etc) – but also high rates of >25% delays to second doses. True, it only took an hour and a half to have a delay for zosyn, but that is the point here: while yes, you provided solid care up front, are your system failures (preset administration times from your inpatient order sets) hurting your patients? Or is it a sign that those getting tighter adherence to everything (Q6hr drugs, early pressors, etc), and that those patients are getting better care with better adherence to second dose antibiotics as merely a proxy?

Regardless, there was a 1.6x increased odds of mortality for those with >25% delay, and a 2.4x high rate of mechanical ventilation, with an OR of 72 for a missed second dose of a q 6hr drug, 24 for a q 8hr drug, and 7 for a q 12hr drug.

Call it better care. Call it a reason to do a ZoSyn continuous infusion. Call it an inpatient problem. Regardless, it should not be ignored.

Me? I maintain that if the patient is still in my emergency department, its still my patient no matter how long they have been there. Hence, I’ve been known to periodically order a second dose of ZoSyn – especially for the critically ill and those being transferred. So, much like after a successful intubation post-arrest there is much work to do, there is much work to do post resuscitation for a septic patient.