Critical Care

Who ya gonna call? #VancZosyn!

If there’s some strange cough in your resus room,

Who you gonna call? Vanc-ZoSyn!
If something’s fevered… and it don’t look good,
Who you gonna call? Vanc-ZoSyn!

I ain’t afraid of no Staph.
I ain’t afraid of no Strep.

If high lactates are running through your EMR,
Who you gonna call? Vanc-ZoSyn!


There’s been some FOAM rumblings about Vanc/ZoSyn causing AKI, but this was the first time it has been compared directly head to head with Vancomycin-Cefepime. This was a retrospective matched cohort study with 279 patients in each arm – one received combination therapy with vancomycin-cefepime (VC), the other received vancomycin-piptazobactam (VPT) for > 48 hours. Patients were excluded if their baseline serum creatinine was >1.2mg/dl or they were receiving RRT. Patients receiving VC were matched to patients receiving VPT based on severity of illness, ICU status, duration of combination therapy, vancomycin dose and number of concomitant nephrotoxins. The primary outcome was the incidence of RIFLE criteria-defined AKI, with a slew of secondary outcomes performed as well.

So, wait, what’s so special about RIFLE anyway? Glad you asked: In general, the worse the acute kidney injury, the higher the mortality.

Since this study shows an 11% AKI rate with VC and 29% AKI rate with VPT, maybe we can improve our mortality if we simply switch from zosyn to cefepime?

Except that this group reports mortality was actually worse in the VC group (though not statistically significant – 8.6% vs 5.7%). That’s right – the group with more AKI had less mortality. In other news, ICU stay was decreased (6 vs 8 days), which was statistically significant., and only ~1% of patients in both arms required long term hemodialysis.

While I was getting ready to click submit on this blog post, I found a second paper (published Nov 28, 2016) that looked at a matched cohort of 1633 VPT vs 578 VC patients, with essentially similar results – 21.4% AKI in VPT vs 12.5% VC.  This second paper found similar LoS, but also a similar trend in mortality-  6.9% for the VPT arm and 9.2 for VC.

So… I’m not certain what to make of this – but it seems more than fair to question whether drug induced AKI is a meaningful surrogate marker for sepsis mortality.  We need a long term look at mortality between VC vs VPT to see if VPT induced AKI follows the same trends. Maybe we’re trading a slight bump short term mortality for improved long term mortality with VC (or maybe not).  In the meantime, I think we need to pump the brakes on shouting about Vanc/Zosyn AKI until we sort this out a bit more.

Improving Outcomes, Pediatrics

Baby LPs, ultrasounds, and fragility



How fitting that the SMACCdub talk, What Scares You, has recently been released, and, to some extent, discusses pediatric bleeding. Well, this paper discusses high risk peds (febrile infant <60 days) and (post LP) bleeding, and whether or not ultrasound assisted guidance helps.

SPOILER ALERT: (it probably does).

From February 2007-December 2007 (wow, talk about a knowledge translation delay), the authors attempted to enroll 46 total patients to either standard LP without ultrasound vs ultrasound assisted LP. Here’s one key point – while ultrasound guidance means direct visualization of the needle into the desired space (like for central lines or paracentesis), ultrasound assisted means that landmarks were sonographically visualized, and then they marked the skin and estimated how deep was too deep for the needle, then performed the LP (without direct visualization.-Basically they performed an ultrasound to determine a “maximum safe depth” to limit needle advancement to avoid traumatic taps, since this is a common element of LP failure in this age group.

Patients with known spinal abnormality or VP shunt were excluded, and the procedures were done by either a house officer or pediatric NP with MD oversight (so, I’m not certain how applicable this is to those of us with significant experience in this age group). Unfortunately, the study was terminated prior to reaching their goal of enrolling 23 patients into each group due to academic calendar demands of the lead author (21 vs 22 patients in either arm – meh.) Success was defined as <10k RBC and whether or not CSF was obtained. Their 5 month historical failure rate was 44%.

The groups did not differ in terms of prematurity, patient weight or length, there was a lower median age in ultrasound assisted group (38 days vs 45 days p=0.02), which may give them a bit more of an uphill battle. The results are seen below:



On first glance, these look good – less frequent traumatic taps, more frequently obtaining CSF with NNTs of 3.7 and 5.6 respectively. However, with such a small sample size, a Fragility index of 1, and having house officers and NP’s do the tap (with an unclear level of experience), I’m not certain this is broadly applicable to all providers, particularly when you add that 19 sono-assisted attempts are not enough to reach 80%  success in this study.  With that said, we commonly perform interventions with much lower NNTs with higher risks to the patient than a few ultrasonic waves. This is a cant hurt, will probably help intervention that we should probably be utilizing more frequently for all of our patients, not just our pediatric population.

For a great review on this topic check out sonomojo for more on ultrasound use for LPs.

Improving Outcomes, Improving Throughput

How soon to safely discharge the opiate OD?

For those of us wondering how long we need to keep the thrashing, agitated, cursing “narcan’ed” patient in our ED, look no further. This is essentially a review of the literature on whether or not adopting a treat & release policy for opiate overdose in the prehospital realm is safe & feasible.

They review 4 papers, 3 of which contain data points from 1994-2003, of which two were non-US studies. In general, they look at some short time period for bouncebacks, (6-12 hours), and if the patient does not come back to receive chest compressions or repeat narcan dosing, they considered it a win. Ultimately, out of 3875 patients that received narcan in the field and were able to AMA after a 15-20 minute observation period, only 3 had a recurrence that resulted in death.

Digging a bit deeper, part of the problem here is that two of the papers had exclusion criteria that does not necessarily fit what happens in practice. One paper excluded patients brought to the hospital, while another excluded those with polysubstance abuse. I’m not sure about your patient population, but the heroin abusers I’ve like to chase with China White with a stick of xanax. Fortunately, the two US studies were more likely applicable, with almost no exclusion criteria – and of which zero patients out of 1550 prehospital treat & release patients died within 12 hours.

So how does this apply to the ED? It is important to note that there are clinical decision rules to help guide who can go home relatively quickly.  If patients can ambulate, has normal vitals and a GCS of 15, then your miss rate is likely well below 1% for them to return in the next few hours from this particular overdose. So, by the time a patient is reversed with narcan, you write the chart and get discharge papers ready, if they remain alert, oriented, competent and reasonable, they can likely go. However, it should be noted that there were a small number of patients who returned within a few weeks with various other issues – one patient hung themselves within 48 hours. Another overdosed 4 days later. All in all, still <1% dying within 30 days, but this is potentially a teachable moment. Patients do have the right to make bad decisions, but that shouldn’t necessarily allow us to stigmatize them and not at least offer them the help they likely need.

Cardiology, Improving Outcomes, Improving Throughput

A chicken in every pot, and a cardiologist in every box.

I think the HEART score is useful, and an incredible start to getting everyone on the same page. Getting both an ED and consultant group to universally agree on a protocol, and implement observation / admission protocols off of it are probably a bit more difficult.

And this is only with a few “soft” variables– what exactly differentiates slightly from moderately suspicious anyway? As we all know, stories change (and not just from patients!).

This paper retrospectively looked at 6 months worth of ED chest pain charts which had a cardiology consult and tries to extract a HEART score based off the ED documentation as well as the cardiology consultation.

Unfortunately the retrospective nature and lack of a standardized “flow sheet” for history probably greatly contributes to cardiology/EP disagreement in the HEART score (like, say, documenting tobacco usage in the chart). History between EP and cardiology was in agreement 47% of the time, EKG interpretation agreement at 76%, and risk factor agreement at 85%. Overall HEART score agreement between EP’s and cardiology occurred 70% of the time, primarily with some mixture of cardiology consistently downplaying elements and/or EP’s upselling some.

Of those who had a phone consultation with cardiology, only 5.4% were discharged, vs 45% discharged when physically seen by cardiology. Only 9% were admitted after in-person cardiology evaluation vs 77% for those with phone consultation. Of those who received further testing, 45% of the cardiology phone consultations were discharged, vs 87% discharge rate for those who received additional testing after an in-person cardiology consultation…. Seems like cardiology is scared to discharge without seeing the patient, and that we are probably upselling the patient a bit.

Regardless, this is hypothesis generating at best, particularly with such low numbers to evaluate (33 patients evaluated by cardiology and EP’s over this 3 month period!), and frankly, the retrospective data extraction without a clear checklist for HEART scores makes me question the validity of their conclusions. Nonetheless, I hold hope that cardiology and EM can live in harmony at some point in the future.


PA’s- making Ada Plumer Proud.

Many years ago, nursing was not allowed to place IVs. Now, in some places, they place ultrasound guided PICCs, and in a handful of places, ultrasound guided central lines.   Nursing can titrate vasopressors, and in some facilities, they run ACLS during codes.


So…. Have you ever experienced or asked “for the doctor (or consultant)?”

Have you ever been concerned and wondered do PA’s provide appropriate care in the ED?

This study is done at the world famous Our Lady of Lourdes in Camden,NJ, with none other than the EM famous Al Sacchetti- and should aim to answer some of these concerns, at least in the pediatric population

Over a 24 month period, over 10,000 patients age 6 or younger were restrospectively evaluated for bounce back rates and broken down into 3 groups based on their provider: attending only care, PA only care, or co-evaluation by both EP and PA.  Here’s the twist- in this department, policy permits PA’s to evaluate treat and discharge patients of any age independent of the attending physician.  There are no specific protocols for assigning specific patients to specific providers, though they do state that PA’s do not perform LP’s in the study department, and that febrile infants <8 weeks were brought to the immediate attention of the attending physician by the triage nurse.  Essentially, the PA seemingly functions at a high level and fairly autonomous.

So what’d they find?  Are you more likely to “bounceback” based on who you’re seen by?

Well, as one would likely expect, a higher percentage of higher acuity patients were seen by the attending physician (85% ESI-2, 70% ESI-3, 60% ESI 4/5.), and the younger the patient, the more likely they were an “attending only” case.

Bounce backs?  Only 0.4% of PA only cases vs 0.6% of attending only cases bounced back and were admitted – not statistically significant – and below the national average of 0.83%.  Only 0.9% of PA only cases were admitted vs 4.1% of attending only cases (and 3.4% of PA/MD cases.  This was statistically significant, and likely reflects higher acuity of the cases the MD is involved in.  There was a higher rate of return visits in MD only cases – 8% vs 6.8% (statistically significant) – but the rate of return for combined MD/PA cases was highest – 9.3%.

Amongst ESI 2,3, & 4s, bounceback rates for EP only eval was consistently higher than PA only eval, and bounceback rates for MD/PA co-evaluation was consistently higher than both PA only and MD only eval.

So, what’s the take home?  For one, its that PA’s can provide high level care without a significant drop off in care… and that in order for medicine to progress, we have to cognitively offload to expand our boundaries.  This may include expanding services to nursing or PAs (as discussed above)…

And here is the most important message:

Dont be that person – if someone (nursing, PCA’s, PA/NP’s, etc) comes to you with a concern – go see a patient.  We’re all on the same team.  Take it as a compliment – as evidenced by this paper it’s likely a complex or clinically ambiguous case with a higher bounceback rate, and who doesnt like a challenge!



The Ketamine vs All battle continues

ketamine. Ketamine. Ketamine! KETAMINE!!!!

Is there anything the SoMe wonderdrug can not do? I mean, isn’t the answer *always* ketamine?

This is a single center, randomized, prospective parallel study for patients aged 18-70 with moderate to severe acute, traumatic, orthopaedic pain with pain scale >80 on VAS. With about 25 patients per treatment arm, they compared intranasal Ketamine (1 mg/kg), IV morphine (0.1mg/kg), and IM morphine (0.15 mg/kg).


The details:

Rule ins: Must have GCS 15, weight 50-110kg, systolic bp 90-160 mmHg, HR <100 bpm.

Rule outs: head trauma, history of regular opiates or psychiatric disorder, analgesia within 3 hours, “a large meal ingested within the previous hour,” any LOC, dizziness, vomiting, or nausea were excluded as well.


The group measured pain scales every 5 minutes for 1 hour, as well as ADRs at end of 1 hour. They deemed a 15 mm score reduction in pain deemed significant.


Time to onset was fastest in the IV morphine route at 8.9 minutes, Ketamine came in second at 14.3 minutes, and IM morphine, unsurprisingly, brought up the rear with 26 minutes. The time of onset between Ketamine and the two morphine routes was insignificant, though the time to onset between IM and IV morphine was statistically significant.  Also not surprisingly, Ketamine had significantly higher ADRs (difficulty concentrating 58% vs ~21%, confusion 50% vs ~15%).  While pain reduction at one hour was similar across all treatments, there was a trend for decreased patient satisfaction with Ketamine (58% satisfaction vs ~70% with morphine – this was not statistically significant).


While I would love to say this trial adds to the data for usage of Ketamine… not quite. It really does not look at the patients for which we would **want** to use ketamine – namely, say those with poor access that an IN analgesic may work wonders; those with an opiate habit; and seriously, what trauma patient doesn’t come in a bit tachycardic?  And while yes, the results are about in line with what we’ve seen in the past and sort of come to expect (reasonable analgesia, somewhat decreased patient satisfaction, higher ADRs) this just is not a real world study that we can point to and say, “this is why we need IN Ketamine in our protocols.”

Improving Throughput, Mythbusting, Radiology, Radiology

Whaddaya Mean You Can’t Learn POCUS?

After taking a few one day ultrasound courses, a common theme amongst classmates is something to the effect of, “well, I get it now, but what about next week when the instructor isn’t there to guide me?”

Admittedly, I have wondered about the same thing; and wondered about the retention of knowledge and ability to apply what you have learned at these 1-2 day crash courses.

So you think you cant learn ultrasound? Or that you can not retain it after a weekend course? Well, those damn whippersnappers from Oregon are putting the non-believers to shame.

Medicine interns at Oregon Health & Science University were taught point of care ultrasound 3 months into their first year, as one day of a 5 day medical “boot-camp.” The day-long program consisted of 15-20 minutes of didactic training, and was followed by a 40 minute hands on session. Learners were placed in groups of 2-3 individuals and taught one-hour modules consisting of: the basics (knobology, physics, etc), CLUE protocol, FAST exam, hydronephrosis eval, and aorta & neck anatomy.  The 40 minute hands on portion was divided into 20 minutes for completing modules demonstrating pathology on SonoSim machines and 20 min for facilitator-led hands-on practice with volunteer models. (example- 20 didactic minutes learning FAST, 20 minutes on simulation, then 20 minutes on a real-live person!).  This was followed by two optional 1 hour courses done within 6 months.

A 30 question multiple choice test was administered prior to the course to all 33 interns, testing image interpretation, image acquisition/optimization, and clinical applications of ultrasound. The test was re-administered 6 months later; there was a significant drop out rate (27%), and it was untracked as to whom took the optional one hour courses.

Survey says?

Mean pretest scores – 61%

Mean post-test scores- 85%

Mean 6 month post-test scores – 79%

Great news – We probably intuitively know & retain much more than we think, but just have to continue to pick up the probe to hone our craft.  Bad news, I’m not certain that an ability to retain enough knowledge to improve a multiple choice test score is the same as making a correct clinical decision off of limited ultrasound skills.  Admittedly, POCUS in the wrong hands can be a problem, and making clinical decisions based off limited ultrasound skills and knowledge is a difficult leap to take, but its one we invariably have to make in order to grow as a clinician.

So, yeah, don’t tell me you can’t learn ultrasound.