Choosing Wisely Hepatology, Eh?

The Choosing Wisely campaign is quite popular in theory, if not in practice – ranging widely across the specialties from Pediatric Hospital Medicine to our own, beloved, Emergency Medicine.

This list is from the Canadian Association for the Study of the Liver, and two of their five recommendations are somewhat relevant to EM. Without further ado:

Statement 1: Don’t order serum ammonia to diagnose or manage hepatic encephalopathy

This was their most highly ranked recommendation when members were surveyed at their annual meeting. They cite multiple confounders regarding ammonia levels, factors affecting accuracy of the measurement, and state “elevated ammonia levels do not add any diagnostic, staging, or prognostic value.” The diagnosis, they feel, ought to be made based on clinical history and response to therapy alone.

Statement 2: Don’t routinely transfuse fresh frozen plasma, vitamin K, or platelets to reverse abnormal tests of coagulation in patients with cirrhosis prior to abdominal paracentesis, endoscopic variceal band ligation, or any other minor invasive procedures

This is another one of my favorite pet topics – transfusion intended to “restore normal hemostasis” in a dysfunctional, but somewhat already rebalanced coagulation system. As they say, “Routine tests of coagulation do not reflect bleeding risk in patients with cirrhosis and bleeding complications of these procedures are rare.” In fact, I’ve seen several articles approaching even liver resection in the context of elevated coagulation parameters absent any major bleeding complications – so this ought certainly apply to minor procedures, including those in the Emergency Department.

No doubt the uptake of these recommendations will be highly variable among hospitals and specialty groups, but lists like these are great tools with which to start the conversation.

“Choosing Wisely Canada-Top Five List in Hepatology”
https://www.ncbi.nlm.nih.gov/pubmed/30596626

Vital Signs = Vital

That is how the authors frame it, after all: “‘Vital signs are vital’ is a common refrain in emergency medicine.”

And, these authors add to the body of work further exploring this axiom. In this simple, retrospective data analysis, they evaluate all adult visits to their Emergency Department to determine the effect of abnormal vital signs at disposition on short-term outcomes.

For discharges, about 3% of their cohort returned to the same ED within 72 hours. Only a handful – a little less than 15% – had any vital sign abnormalities at discharge. And, yes, those with vital sign abnormalities were slightly more likely to return than those who did not, with relative risk ratios centered generally around 1.2. Then, a little more than a quarter of patients were admitted on their return visit – and, again, vital sign abnormalities increased the likelihood of subsequent admission by a small amount. In this case, fever was more likely than the other abnormal vital signs to tip the scales towards admission.

Similarly, an analysis of inpatient visits and subsequent escalations in care noted vital sign abnormalities exhibited a greater risk of upgrade, with RRs centered around 2.

Overall, however, the vast majority of patients who were either admitted or discharged with abnormal vital signs did well. Abnormal vital signs are always worth recognizing and dedicating a bit of cognitive effort, but the aren’t strong enough predictors of subsequent outcomes to drive changes in management.

“Association of Vital Signs and Process Outcomes in Emergency Department Patients”
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6526877/

The EXTEND Alteplase Meta-Analysis

Did you miss the publication of EXTEND a couple weeks ago – a publication I helpfully labeled as “shenanigans“? Well, these same authors have wasted little time performing a systematic review and meta-analysis of individual patient data in the 4.5-9 hour timeframe. Their search, specifically limited to hemispheric stroke and pretreatment perfusion/diffusion evaluation, identifies: EXTEND, ECASS4-EXTEND, and EPITHET.

EXTEND we’ve already heard from – and, since most of the patients for this IPD meta-analysis come from EXTEND, it should be no surprise the overall results effectively mirror EXTEND. EPITHET, of which you may have some faint familiarity, has been pulled from the dusty archives of 2008. Then, there’s ECASS4-EXTEND, of which you probably hadn’t heard, since it was published with zero fanfare about a month ago.

So, what is ECASS4-EXTEND? These were again 4.5-9h patients screened with MRI and enrolled between 2014 and 2017, with early termination recommended by the Data Safety Monitoring Board when enrollment slowed to a trickle following publication of the endovascular trials. Before discontinuation, these authors enrolled 120 and analyzed 116, 60 receiving tPA and 56 placebo. Most of them were “wake up” strokes, and the “time-to-treatment” variable is again facetiously estimated by taking the midpoint between sleep onset and time of waking. There are small increases in patients with reduced disability in the tPA arm, but these unsurprisingly do not reach statistical significance. Likewise, deaths within 90 days are double – 11.5% versus 6.8% – another technically non-significant result. The authors, naturally, focus on the promise of the treatment if a sufficient sample were recruited, rather than the potential threat to patient safety.

And then there’s this all-too-familiar editorial failure:

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.

…in direct contradiction to the third author having this affiliation:

Medical Affairs, Boehringer Ingelheim Pharma GmbH & Co KG, Ingelheim, Germany

And this little snippet in the body of the article:

Role of the funding source
… The trial was supported with a restricted grant from Boehringer Ingelheim (Germany), the funder. The funder approved the study design…. Two employees of the funder were members of the steering committee and thus involved in data interpretation and preparation of the publication.

Finally, amusingly enough, ECASS4-EXTEND doesn’t technically meet criteria for their inclusion in the systematic review and IPD meta-analysis – they report they searched for trials “published in English between Jan 1, 2006, and March 1, 2019”, while ECASS4-EXTEND was published on April 4th.

Nitpicking aside, despite the relative frequency and prominence of these publications, this is mostly much ado about nothing – it should be obvious from the early termination of ECASS4-EXTEND these data primarily reflect a cohort we’re sending to endovascular therapy. Therefore, what we really need for these data to be relevant is a confirmatory trial performed specifically in the resource-austere settings thrombectomy might not be available.

“Extending thrombolysis to 4·5–9 h and wake-up stroke using perfusion imaging: a systematic review and meta-analysis of individual patient data”
https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(19)31053-0/fulltext

“Extending the time window for intravenous thrombolysis in acute ischemic stroke using magnetic resonance imaging-based patient selection”
https://www.ncbi.nlm.nih.gov/pubmed/30947642

The United Colors of Sepsis

Here it is: sepsis writ Big Data.

And, considering it’s Big Data, it’s also a big publication: a 15 page primary publication, plus 90+ pages of online supplement – dense with figures, raw data, and methods both routine and novel for the evaluation of large data sets.

At the minimum, to put a general handle on it, this work primarily demonstrates the heterogeneity of sepsis. As any clinician knows, “sepsis” – with its ever-morphing definition – ranges widely from those generally well in the Emergency Department to those critically ill in the Intensive Care Unit. In an academic sense, this means the patients enrolled and evaluated in various trials for the treatment of sepsis may be quite different from one another, and results seen in one trial or setting may generalize poorly to another. This has obvious implications when trying to determine a general set of care guidelines from these disparate bits of data, and resulting in further issues down the road when said guidelines become enshrined in quality measures.

Overall, these authors ultimately define four phenotypes of sepsis, helpfully assigned descriptive labels using the letters of the greek alphabet. These four phenotypes of sepsis are derived from retrospective administrative data, then validated on additional retrospective administrative data, and finally the raw data from several prominent clinical trials in sepsis, including ACCESS, PROWESS, and ProCESS. The four phenotypes were derived by clustering and refinement, and are described by the authors as effectively: a mild type with low mortality; a cohort of those with chronic illness; a cohort with systemic inflammation and pulmonary disease; and a final cohort with liver dysfunction, shock, and high mortality.

We are quite far, however, from needing to apply these phenotypes in a clinical fashion. Any classification model is highly dependent upon the inputs, and in this study the inputs are the sorts of routine clinical data available from the electronic health record: vital signs, demographics, and basic labs. Missing data was common, including, for example, lactate levels, which was not obtained on 80% of patients in their model. These inputs then dictate how many different clusters you obtain, how the relative accuracy of classification diminishes with greater numbers of clusters, as well whether the model begins to overfit the derivation data set.

Then, this is a little bit of a fuzzy application in the sense these data represent as much different types of patients with sepsis, as it represents different types of sepsis. Consider the varying etiologies of sepsis, including influenza pneumonia, streptococcal toxic shock, or gram-negative bacteremia. These different etiologies would obviously result in different host responses depending on individual patient features. These phenotypes derived here effectively mash up causative agent with the underlying host, muddying clinical application.

If clinical utility is limited, then what might the best utility for this work? Well, this goes back to the idea above regarding translating work from clinical trials to different settings. A community Emergency Department might primarily see alpha-sepsis, a community ICU might see a lot of beta-sepsis, while an academic ICU might see predominantly delta-sepsis. These are important concepts to consider – and potentially subgroup-analyses to perform – when evaluating the outcomes of clinical trials. These authors do several simulations of clinical trials while varying the composition of phenotypes of sepsis, and note potentially important effects on primary outcomes. Pathways of care or resuscitation protocols could potentially be more readily compared between trial populations if these phenotypes were calculated.

This is a challenging work to process – but an important first step in better recognizing the heterogeneity in potential benefits and harms resulting from various interventions. The accompanying editorial does really a very excellent job of describing their methods, outcomes, and utility, as well.

“Derivation, Validation, and Potential Treatment Implications of Novel Clinical Phenotypes for Sepsis”
https://jamanetwork.com/journals/jama/fullarticle/2733996

“New Phenotypes for Sepsis”
https://jamanetwork.com/journals/jama/fullarticle/2733994

ECLS in AMI

Acute myocardial infarction complicated by cardiogenic shock has dismal outcomes. Pharmacologic therapy, intra-aortic balloon pumps, ventricular assist devices, et cetera, all have limitations. So, why not extracorporeal life support (ECLS)? Well, for one, it’s resource-intensive, expensive, and little high-quality evidence supports its use. But, on the other hand, it’s fancy and magical and reputed to beget miracles.

Published as a brief research report, this small article describes an open-label pilot of 42 patients suffering cardiogenic shock, randomized to ECLS or no mechanical circulatory support. The primary outcome was left-ventricular ejection fraction at 30 days, with multiple secondary outcomes crammed into their short bit.

The short answer: zero difference – median 50% vs. 50.8%, and no reason to tell you which is which because they’re the same. Mortality favored ECLS at 19% vs. 33%, but these outcomes and their survival curves are so entwined there is no reliable difference to be made from this small sample – and moreso even because the control group had greater illness burden at baseline. Process outcomes, such as intensive care unit length-of-stay and duration of mechanical ventilation favored the control cohort, as expected, given the relative resource intensity of ECLS.

The new sexy thing is always alluring and presumed to be better, but as these authors conclude: “This raises an urgent call for randomized controlled trials assessing survival as primary endpoint.”

“Extracorporeal Life Support in Cardiogenic Shock Complicating
Acute Myocardial Infarction”
https://www.ncbi.nlm.nih.gov/pubmed/31072581

Progesterone for First-Trimester Bleeding

Emergency Department evaluation for patients with first-trimester bleeding is fairly straightforward. Most of the time, an ultrasound identifies an unremarkable intrauterine pregnancy and patients are provided with expectant management and best wishes. However, there is some evidence progesterone supplementation – in this trial, an intravaginal progesterone supplement – may help implantation and prevent pregnancy loss. This, the Progesterone in Spontaneous Miscarriage (PRISM) trial, is the first, large, high-quality investigation of this intervention.

Over the course of two years, 12,862 women with bleeding before 12 weeks of pregnancy were screened, and 4,153 enrolled across 48 hospitals in the United Kingdom. Enrolled patients were randomized to either 400mg intravaginal micronized progesterone twice daily through 16 weeks of pregnancy, or identical placebo. The primary outcome was live birth after at least 34 weeks, with secondary outcomes being other early pregnancy milestones, as well.

Per the authors, and hewing fast to their frequentist analysis, this is a negative trial. The primary outcome occurred in 75% of the progesterone cohort and 72% of placebo, a relative rate of 1.03 (1.00 to 1.07) and a p-value of 0.08. The authors conclusion: “treatment with progesterone did not result in significant improvement in the incidence of live births among women with vaginal bleeding during the first 12 weeks of pregnancy.”

Maybe?

It is always curious to look at the statistical analysis portion of these articles and consider the decisions leading to the inability to detect a difference. In this trial they state their choice of sample size was driven by the “minimally important absolute difference of 5 percentage points between the progesterone group and the placebo group in the incidence of live births after at least 34 weeks of gestation (65% vs. 60%)”. If we had a medication for use in sepsis that was inexpensive and readily available, would we require a 5% difference in mortality for its use? Anti-hypertensives, taken for five years, prevent heart attack, stroke, and all-cause mortality with numbers-needed-to-treat swimming right around a 1% difference – and these are taken in massive numbers at the population level. Without delving into any sort of personhood argument, a lost pregnancy is effectively a mortality benefit – and, despite the massive scale required, it might have rather been more appropriate to choose a smaller “minimally important absolute difference.”

If the observed difference of 2 to 3% were to be confirmed, these are NNTs in the 33-50 range to prevent pregnancy loss. While this would not be rank as a profoundly effective intervention, we are rather talking about producing an actual new human being. No harms were detected, although a trial such as this would be expected to be even further from powered to detect very rare events such as congenital deformities. I would expect the debate regarding this to continue, and I wouldn’t be surprised if you found some groups encouraging its use or initiation in the ED.

“A Randomized Trial of Progesterone in Women with Bleeding in Early Pregnancy”
https://www.nejm.org/doi/full/10.1056/NEJMoa1813730

EXTEND Alteplase Shenanigans!

Do you remember EXTEND-IA? Or EXTEND-IA TNK? This is, well, their neglected little brother, regular old EXTEND, stumbling along to “completion” and publication in the New England Journal of Medicine, as is apparently their birthright.

EXTEND-IA was part of the enormously important series of trials launching the endovascular revolution for acute ischemic stroke. EXTEND-IA TNK is another piece of evidence probably pushing us slowly, inexorably, towards tenecteplase rather than alteplase. This, despite its provocatively titled editorial, is not a grand event.

This trial, which started enrolling way back in 2010, essentially mirrors EXTEND-IA, but gives alteplase to patients with a mismatch on perfusion imaging, rather than referring them to thrombectomy. Over 8 years at 16 centers, mostly in Asia-Pacific, the authors were able to randomize a mere 113 patients to alteplase and 112 to placebo. The primary outcome, a modified Rankin Scale of 0 or 1 at 90 days, favored the alteplase cohort, 35.4% to 29.5%. Deaths, partly related to a 6% absolute excess of intracranial hemorrhage, were higher in those treated with alteplase, 11.5% vs 8.9%. The efficacy results do not meet statistical significance prior to adjustment, but the median NIHSS was 12 for alteplase and 10 for placebo. So, you can probably guess the bulk of their discussion focuses on their adjusted effect size, which does reach statistical significance at 1.44 (1.01 – 2.06). Interestingly enough, this wasn’t their original planned adjusted analysis – the 95% CI for that traditional logistic regression crosses unity at 0.99 – leading to questions whether this fortuitous p-value is innocent serendipity, or found because it was findable.

Regardless, this trial – stopped early, per the authors, because of the publication of WAKE-UP – is already mostly obsolete. Systems of stroke care have changed immensely since this trial was planned. About 80% of patients in this trial had large vessel occlusions on imaging – patients who in this modern era would simply go straight to thrombectomy. These results do not support the use of alteplase as an alternative to thrombectomy, as recanalization rates – as we’ve known forever – are simply not good enough with medical therapy. Therefore, in modern systems of stroke care, this trial probably has zero effect on care. The better approach to tailoring treatment to individual patient heterogeneity in our modern systems is to find new ways of integrating MRI into the rapid assessment of stroke.

However, much of the world does not have access to timely thrombectomy for stroke, for a variety of reasons. In rest of the world, in that narrow slice with a modern system for acute evaluation with perfusion imaging and alteplase administration, but not timely thrombectomy, then you could consider changing protocols to include alteplase administration like here in EXTEND. It is not clear from these data whether generalization of these data to such lower-resource settings would accurately reflect effectiveness and safety, but that is the conceivable application of these results. Then, you have to consider the typical disclaimers affecting the reliability of their presented findings:

Dr. Parsons reports receiving consulting fees from Apollo Medical Imaging Technology, Boehringer Ingelheim, Canon Medical Systems, and Siemens; Dr. Wong, receiving grant support, paid to Royal Brisbane and Women’s Hospital, from Boehringer Ingelheim; Dr. Sabet, receiving travel support from Boehringer Ingelheim; Dr. Christensen, holding stock in Ischema- view; Dr. Mitchell, receiving lecture fees from Medtronic USA and Stryker; Dr. Thijs, receiving advisory board fees from Amgen and Bristol-Myers Squibb, advisory board fees and lecture fees from Bayer and Pfizer, advisory board fees, lecture fees, and travel support from Boehringer Ingelheim, and advisory board fees and travel support from Medtronic; Dr. Meretoja, receiving advisory board fees, lecture fees, and travel support from Boehringer Ingelheim and Stryker; Dr. Davis, receiving advisory board fees from AstraZeneca and Boehringer Ingelheim; and Dr. Donnan, receiving advisory board fees from AstraZeneca Australia, Bayer, Boehringer Ingelheim, Merck, Pfizer, and Servier.

At the minimum, at least, it is another bit of evidence regarding the importance of salvageable brain for the utility of any intervention for stroke – a principle that probably ought be applied for those treated within 4.5 hours of stroke, as well.

“Thrombolysis Guided by Perfusion Imaging up to 9 Hours
after Onset of Stroke”
https://www.nejm.org/doi/full/10.1056/NEJMoa1813046

“Image-Guided Intravenous Alteplase for Stroke — Shattering a Time Window”
https://www.nejm.org/doi/full/10.1056/NEJMe1904791

Addendum 5/15/09: Minor updates in response to Twitter discussion and comments below.

Oops, Missed the Culprit

This is a rather curious study I’d never specifically see coming, which makes it potentially brilliant. We in Emergency Medicine identify our non-STEMI patients, tidy them up for admission, and bid them bon voyage. We assume their post-Emergency Department care will include non-urgent coronary angiography, and this will properly address their underlying anatomic cause.

It turns out, this may not be the case more frequently than I would have thought. This is a very interesting study in which a convenience sample of patients with non-STEMI were recruited to undergo a cardiac MRI prior to coronary angiography. Cardiologists performing the angiography were blinded to the cardiac MRI results. Following angiography, the cardiologists would identify the culprit artery, if there were one, and its associated infarct territory on an anatomic model. These results were then subsequently compared to the results of the cardiac MRI.

There were 114 patients who underwent MRI and coronary angiography. Of these, angiography identified a culprit lesion in 72 (63%). Of these, 47 were consistent with the ischemic region identified on MRI. The remainder found either a different infarct artery on MRI (10), a non-CAD diagnosis such as myocarditis (9), or no apparent ischemic hyperenhancement (6). In the remaining 42 (37%) who did not have a culprit identified on angiography, 25 of these had an ischemic territory identified on MRI, while the others had a non-CAD related diagnosis or did not find hyperenhancement. Effectively, as compared with the MRI results, the coronary angiography accurately identified a culprit or lack of culprit in 62 of 114 (54%) of cases.

There are many limitations to this article – the least of which being an author is an inventor on the patent of delayed-enhancement MRI, and holds at the minimum a professional conflict of interest in showing this technology adds value. It is non-obvious this cardiac MRI technology should be considered the “gold standard” for identification of the culprit lesion territory, or that their methods of collecting culprit lesion infarct territory is reliable. Finally, this is a small sample from just three institutions and may not be generalizable.

If these data are valid, however, it certainly raises significant issues for many studies featuring revascularization as part of a composite “major adverse cardiac events” endpoint. Frequently, much of downstream MACE consists of coronary revascularization – for good or ill – and these data might further cast doubt upon the accuracy and effectiveness of revascularization. One could imagine an MRI might even be a necessary step in evaluation of non-STEMI to avoid unnecessary or erroneous coronary intervention.

Thought-provoking, but certainly far too early to suggest change in practice.

“Identifying the Infarct-Related Artery in Patients With Non–ST-Segment–Elevation Myocardial Infarction”

https://www.ahajournals.org/doi/full/10.1161/CIRCINTERVENTIONS.118.007305

The CT and Syncope

The Choosing Wisely Campaign lists non-contrast CT of the head as one of their low-value procedures for low-risk patients presenting with syncope. However, despite these recommendations, these authors voice concerns up to two-thirds of patients still undergo advanced imaging.

In this systematic review looking at both practice patterns and yield, the authors identify 17 studies of both hospitalized and Emergency Department patients addressing this topic. Pooling together 1,669 ED patients, 55% underwent CT with a yield of 3.8%. Pooling 1,289 hospitalized patients, CTs were performed 45% of the time with a yield of 1.2%. Considering the general morbidity and mortality associated with intracranial conditions, these are not fantastic yield numbers, but are not entirely unreasonable.

There is a bit of trouble with these numbers, however: even though their systematic review went up to 2017, most of the included studies were published before 2011. Even their citation of “two-thirds receiving head CTs” was published in 2009, well before the 2014 Choosing Wisely statement. Then, the bulk of the included studies were retrospective, blurring the reliability of their inclusion and outcomes measurement.

I think these data probably effectively illustrate the rarity of serious outcomes in syncope, but provide little insight into the current scope of the problem with respect to overuse. An intracranial process or intracranial injury associated with syncope ought to be considered in each case, but better data describing features predictive of underlying intracranial injury is needed to better separate high-risk from low-risk.

“The Yield of Computed Tomography of the Head Among Patients Presenting With Syncope: A Systematic Review”
https://www.ncbi.nlm.nih.gov/pubmed/31006937

Add Clopidogrel to Aspirin, Temporarily

In the Emergency Department, we’re not necessarily spending a lot of our time sending home patients with minor stroke or high-risk transient ischemic attack – or, even if we are, we’re usually not doing it independently. That said, our scope of practice always seems to be expanding, observation units are frequently run by emergency physicians, and hospitals are looking to take advantage of opportunities to discharge patients rather than admit.

Regardless, this is a meta-analysis supporting the findings of CHANCE, looking specifically at the short-term use of aspirin and clopidogrel after a minor stroke or high-risk TIA. CHANCE enrolled 5,170 patients, while this meta-analysis effectively combines these with POINT, while also tossing in the patients from FASTER.

The main takeaway here is the combination of the evidence from the long-term treatment observing it was effectively a wash between stroke prevention and bleeding complications, and the observation that stroke risk was highest immediately following the index event. A reasonable interpretation, as highlighted by the guidelines, is to use dual-antiplatelet therapy short-term after the index event. The evidence does not specifically describe the optimal duration, but anywhere between 10 and 21 days seems reasonable.

Just wanted to toss this one out there with a little more prominence, as this isn’t particularly new, but I also wouldn’t want it to be overlooked.

“Clopidogrel plus aspirin versus aspirin alone for acute minor ischaemic stroke or high risk transient ischaemic attack: systematic review and meta-analysis”
https://www.bmj.com/content/363/bmj.k5108

“Guideline: Starting dual antiplatelet therapy ≤ 24 h after high-risk TIA or minor ischemic stroke is recommended.”
https://www.ncbi.nlm.nih.gov/pubmed/30986828