Critical Care, GI, Improving Outcomes, Mythbusting

Less transfusions, the better, platelet style.

Over 4 years, the Mayo clinic reviewed over 40,000 ICU patients, and sought to determine if prophylactic platelets for critically ill thrombocytopenics matter.

Of these 40,000 ICU patients, they excluded anyone who received RBCs within 24 hours prior to the PLT count that triggered transfusion “to minimize the risk of including those with active bleeding” … The primary outcome was essentially to determine if transfusing platelets led to more RBC transfusions, and, secondarily, all-cause mortality, ICU and hospital free days (among others). Seems backwards, but whatever.

So what falls out of this data?

For all comers, in a propensity matched cohort, one transfusion begets another – (46.3 % of those with platelet transfusions also had an RBC transfusion vs 10.4% of those without platelet transfusions who had an RBC transfusion). Interestingly, those transfused platelets had less ICU free days (22.7 vs 20.8), more hospital free days (15.8 vs 13). When you look specifically at the ~5000 patients with <50k platelet counts and compare those who were/were not transfused platelets, there was no change in ICU mortality, 30 day mortality, ICU-free days, or hospital free days. While this was underpowered to determine statistical significance for those with platelet counts <50k, it is not hard to imagine a larger study to suggest similar benefits of not transfusing these patients- particularly since this study saw fewer hospital free days and fewer ICU free stays (10.2 vs 7.8 days and 19.9 vs 18.3 days respectively) – favoring a more restrictive transfusion strategy (but again, not meeting statistical significance, perhaps due to so few patients with <50k PLT).

This is not the first study I have seen suggesting empiric transfusion or outright canceling of procedures based on platelet counts between 50k and 150k is essentially bunk, and that prophylactic platelet therapy is of little benefit, if not outright harmful. There is even a flicker of a signal that prophylactic platelet transfusions >20k may not be beneficial – but this has yet to be definitively shown true -yet.

I can not agree more with the last words to the authors, “Finally, it must be acknowledged that while clinical trajectories did not improve for the cohort as a whole after platelet transfusion, it is possible that certain subpopulations may indeed benefit from the intervention, though these subgroups have yet to be identified.”


Delayed endocarditis diagnosis? The patient can have as many diseases as they please.

This is exploratory look at patients diagnosed with endocarditis at admission versus those with a delayed diagnosis. Granted, this is not a US study, and over a 9 year period at a single center, but does provide an interesting look at how we manage these patients….

They looked at those with an admitting diagnosis of endocarditis that eventually went on to have this as a final diagnosis as well (54 patients), and compared them to patients with a non-endocarditis initial diagnosis to those who eventually had a final diagnosis of endocarditis (64 patients).

Even in the two slam-dunk groups- the IVDA & those with valve replacements, the diagnosis was delayed in the 38% of the time for those with a history of IVDA. For those with a valve replacement there were also significant delays with native valves delayed 63% of time, vs prosthetics delayed 24% of time…. Are we really bad at diagnosing this? Let’s peel back this onion.

Cases were placed into 3 categories: (1) complications of endocarditis, but not immediately recognized as endocarditis – 70% of cases (2) infectious disease unrelated to endocarditis (14% of cases) – ie, hepatitis (3) inconsistent non-infectious disease (16% of cases).

Of those in the “complications” category, only 10% were unlikely to be dosed with antibiotics – they were admitted for “stroke” or CHF/ pulmonary edema. This is clearly understandable. Do we need STAT echos for the pulmonary edema patient? Or for the stroke? STAT echo’s right away for all of these patients – or perhaps routinely ordering them on all CHF / stroke patients may prove more costly and harmful than its worth.

The author’s make the argument that there is significant mortality involved with an initial missed diagnosis in their cohort- 75% vs 25% (!!!). I’m not certain these represent a complete whiff on the part of the treating clinicians. Rather, I would argue that these patients had their complications recognized and treated appropriately (ie, the pneumonia and UTI’s got antibiotics initially), and that these patients were likely sicker to begin with and that is why they had all the additional complications and higher mortality.

While perhaps a heightened awareness of valve replacement patients, and/or awareness of the disease process may help, but sadly, when you are looking for a needle in the haystack, having a 100% sensitive and specific diagnostic algorithm is unreasonable. When can certainly do better, but how much better without causing harms to the rest of the department remains debatable.

Mythbusting, Radiology

POCUS gel – hot or not?

“Wow thats cold!”

It’s a common statement after applying the probe to a patient, and wouldn’t you love to know the answer to the age old question, “Does gel temperature matter for patient satisfaction?”… While we have some data suggesting improved patient satisfaction scores with POCUS, but could those scores be better if the gel was warmed? Or perhaps would all the improvement in satisfaction scores be lost with cool gel?

This group performed bedside ultrasound using heated gel (102F) or room-temperature gel (82.3F – quite the warm department!).  The investigators even went so far as to trying to blind those performing the study with a heat-resistant glove (!) – and even validated gel temperature through weekly quality assurance measurements throughout the study period.  The investigators informed all patients that the study entailed investigation into various measures to improve patient satisfaction with POCUS, but did not inform them of their specific focus on gel temperature.  After POCUS was performed, patients completed the following survey: “How satisfied are you with the experience of having a bedside ultrasound today?” (on a 100-mm visual analogue scale (VAS) for satisfaction), as well as, “Are you satisfied with the care you received today in the emergency department (yes or no), as well as “How professional was the provider who performed your bedside ultrasound?” (A Likert scale spanning 1-5)

All in all, 59 patients underwent randomization to POCUS with room-temperature gel and 61 underwent randomization to heated gel.

In the end, heated gel made no difference for any of the three questions posed to patients. While I think those of us utilizing warm gel were probably making space in the blanket warmer, it’s nice to know that this intervention, while nice, does not make much of a difference, and a “special” warmer specific for gel in your department is probably unnecessary.


Do Prehospital Antibiotics Matter?

In short, probably not, but still not completely disproven.

This randomised controlled open-label trial looked at giving 2 grams of IV ceftriaxone to patients that met SIRS criteria (save for WBC- testing unavailable to EMS) with suspected infectious illness. Patients were randomly assigned (1:1) to the intervention group or usual care group using block-randomisation with blocks of 4. This study took place across ten large regional ambulance services serving 34 secondary and tertiary care hospitals in the Netherlands over a 2 year period. They screened 3228 patients of which 2698 were eligible (pregnancy, beta-lactam or ceftriaxone allergy, suspected prosthetic joint infection, among others); 1150 in the usual care arm (IV fluids, supplemental oxygen prn), and 1548 in the intervention group (2g ceftriaxone plus usual care). 13 patients in each arm were excluded from final analysis or excluded due to withdrawn consent or being lost to follow up. The primary outcome was all-cause mortality at 28 days.

So, while they screened over 3,000 patients over 2 years (a massive undertaking!), unfortunately, only 37 (3.3%) patients in usual care and 66 (4.3%) patients in the early antibiotics group had septic shock. Perhaps you could make an argument that the intervention group was slightly sicker with 22% vs 17% having 2 or more qSOFA criteria. Despite a median time to antibiotics of 70 minutes in the ED (thus, probably close to 90+minutes faster in the intervention cohort), and with 14% having antibiotics >3hrs from presentation and 14% having none at all (suspected viral syndrome) – there was 8% mortality in both arms at 28 days and 12% at 90 days in both arms. No difference.

When you look at mortality for septic shock it was 27% (10/37) in the prehospital antibiotic cohort vs 28.8% (19/66). Again, not statistically significant. While prehospital antibiotics might make a difference in a larger cohort, its probably going to be very hard to ever do that study – this was a 2 year study looking at over 3,000 patients and they were barely able to accumulate over 100 septic shock patients.

While an American might argue “they only gave ceftriaxone, you need a real drug like Pip-tazo and vancomycin!” – slow down. The authors acknowledge that ceftriaxone may not have been appropriate because it was “a big gun” that they could all agree on and most patients were rapidly narrowed to receive, most commonly, amoxicillin–clavulanic acid with ciprofloxacin and ceftriaxone the second and third most common antibiotics given. They did not have culture reports back at time of publication, but having low mortality, and 9% of each cohort were not given antibiotics from the ED due to suspected viral illness makes me suspect that they do not have nearly the resistance problem (or concerns) that the Americans do, likely do to appropriate stewardship. Likewise, while one may be concerned about missed diagnosis due to premature closure, there was a miss rate of 1.4% in the intervention group vs 1.7% in the usual care group, also not statistically significant.

In the end, the authors provide a sensical view of the current state of prehospital antibiotics, “Studies showing that early antibiotic treatment is beneficial for reducing mortality found this positive association mainly in patients with more severe illness and a (time to antibiotic) of more than 5–6 hours… However, we currently do not advise antibiotic administration in the ambulance to patients with suspected sepsis.“

While it is certainly plausible that prehospital antibiotics may be beneficial for those with septic shock, it is a near certainty that, at least in the USA, sepsis hysteria would further ensue and the inertia of giving everyone a dose of broad spectrum antibiotics will likely occur – not to mention our continued fixation with iatrogenic salt-water drowning. The cost to the system – including other patients in the department – of responding to these prehospital alerts for those not in shock will likely be the hidden cost infrequently published or discussed by administrations.

Improving Outcomes, Improving Throughput, Radiology

Parting the SEA with the almighty H&P (& rapid MRI).

Necessity is the mother of invention, and sometimes, necessity comes in the form of hospital administration after a bad outcome. The authors of this paper, essentially developed a rapid MRI protocol for suspected spinal epidural abscess after “several cases of SEA associated with delayed diagnoses and poor outcomes prompted the chairs of the departments of emergency medicine, neurosciences, medicine, and radiology, and members of the Division of Healthcare Quality, to develop a multidisciplinary, clinical decision support tool and imaging protocol with the goal of facilitating early recognition of SEA.”

Wow. Talk about moving mountains. If you’re department is anything like mine, it takes hours to agree on where we’re getting take out from; I cant imagine adding in 4 entire departments into the lunch-ordering mix, let alone all agree on a protocol.

They took a relatively simple approach – if you have new or worsening back/neck pain AND a history of spinal abscess or current/recent (6 months) bacteremia, straight to MRI. I think the recent bacteremia often gets lost in the work up, so I appreciate that they put this front and center. If there is no recent spinal infection or recent/current bacteremia, They looked at risk factors- and I’ll make this simple and break it into 2 categories: people putting things where they dont belong (IVDA, vascular catheters, spinal procedures/injections) and the recurrently ill: ED visit or antimicrobial treatment within 30 days or an infectious process elsewhere. If yes, head to MRI.

I’m torn a bit on this- while I want to applaud the authors for not dwelling on a variety of risk factors that only a small portion of the population has – alcoholism, HIV, severe COPD, the undomiciled, HepC, oncology patients, transplant patients, etc; to say that this group is pretty much captured in the recent ED visit category probably misses a fair amount of patients on the first go-round. And here is the problem of trying to find a needle in the haystack – its hard to increase sensitivity and specificity without causing a delay at some other portion of the food chain – every stat MRI for so many additional back pain patients pushes out another patient and potentially extends at least 1 other patients length of stay.


Despite an increase from 56 MRI’s in the 7 months pre-intervention to 147 in the 7 months post-intervention, yield for a positive MRI (defined solely as SEA and not vertebral osteomyelitis or infectious discitis), went from 16.1% to 17.7%.

On first glance, that’s not a lot of improvement in yield, but they screened 3 times as many patients without losing yield! This is rather impressive. However, they tripled their ED MRI rate, and, even though they drastically cut turn around times from 8.6 hours to 4.4 hours from time of MRI order to radiology report, thats still well over 4 hours for patients with back pain in a highly optimized system. And while yes, they missed fewer SEAs, they probably still have a good percentage that they missed on first visit – the various forms of immunocompromised – the severe COPDer on repeated steroid prescriptions, the HepC patient, the elderly – these are likely missed on the first go round.

I think this is a great step towards creating a policy towards SEA workup. It needs some refinement, but is the best I’ve seen yet. It poses some issues for smaller facilities that do not have 24/7 MRI capabilities, as well as for consultants (neurology essentially becoming a house officer for ID and neurosurgery), and poses a big time crunch for the ED (again, neurology took control of these cases once the decision to MRI was performed, which the hospitalists must be thankful for!). In the end, there is no such thing as zero miss, but Baystate, with this study, demonstrates that, at least for one day, the H&P is not dead.

Critical Care, Mythbusting

Even Pharma is getting in on Vanco-PipTazo AKI

This was an entertaining 9 page meta-analysis espousing the therapeutic harm of vancomycin and pip-tazo in the form of acute kidney injury.  With a conflict of interest page that reads like a pharmaceutical mutual fund (The Medicines Company, Cubist, Pfizer, Merck, Forest/Allergan, Melinta), it’s no wonder that they infer increased mortality due to AKI, yet conveniently COMPLETELY ignore that the same papers they reference show no mortality difference – and if anything a trend towards mortality benefit for vanco-PipTazo.  Likewise, with dialysis rates <2%, the induced kidney injury is less likely to cause harm than a suboptimal drug that wont kill your bug.

They also fail to mention cefepime neurotoxicity.

There are other ways to go about this. Like, say, reviewing the damn cultures.

But in the end, since The Medicines Company and Melinta have new broad spectrum antibiotics on the market or on the way, it probably behooves them to run a slight smear campaign on current treatment regimens. Therefore, forgive me for considering the possibility that the authors intentions may not be pure.

Improving Outcomes

Review the damn cultures.

This is a multi-center observational cohort study performed over 5 years at two hospital systems. They reviewed over 1800 cases of gram negative bacteremia. About 20% of patients with a prior gram negative bacteremia (within the year) received antibiotics to which their prior cultures were resistant.

This is embarrassing. Just review the prior damn cultures. The answer isnt VancoPime reflexively for everyone. Hell, add on a dose of gentamicin, or whatever the prior cultures are sensitive to. Just write a note in the chart and explain it to the oncoming team.

Side note: bout 25% of admits within the last 90 days were resistant to ceftriaxone and cipro, with an 80% or better percentage for resistance (for ceftriaxone, ceftazidime, meropenem, cipro, or gentamicin – 61% for pip-tazo) to the same antibiotic if the same organism was isolated.

Regardless, blaming the surviving sepsis guidelines or the federal government, or whoever is simply trying to pin your own laziness on someone else. It takes no fewer than 5 minutes – and probably closer to 30 seconds – to review prior cultures. In the critically ill, this is utterly and completely unacceptable.