Month: February 2019

The “Fragility Index” in Emergency Medicine

This little paper is an interesting look at trials in Emergency Medicine and their robustness – or lack thereof.

The “fragility index“, as described by its proprietors, is effectively a “number needed to treat” for confidence intervals and p-values. In simplest form, the fragility index is the number of patients needed to change their primary outcome in order to nullify an otherwise statistically significant result. Effectively, if a trial result would change if only two patients had different outcomes, the fragility would be 2 – and this would reflect a trial whose results are not terribly robust.

The authors then go on to use their tool to evaluate 180 trials across emergency medicine and emergency medicine journals. The medial fragility index across all trials: 4. Of course, this is not terribly unexpected, as the median sample size across all trials was only 140 – not usually enough to illuminate a reliable result except in only the most impressive of effect sizes.

The tool itself is not a novel reinvention of statistical significance or analysis, but just another mechanism for knowledge translation to reflect the relative stability or robustness of a result from a trial. Interpretation of frequentist statistics, p-values, and confidence intervals can frequently be rather opaque or misleading, and reporting a fragility index can be considered one approach to conceptualizing the strength of a result. The real solution, also noted by these authors, is simply to interpret individual trial results in a Bayesian context – adding one trial to a Bayesian prior, when available, to see how a trial shifts the current evidence.

“The Results of Randomized Controlled Trials in Emergency Medicine Are Frequently Fragile”

https://www.ncbi.nlm.nih.gov/pubmed/30551894

More Bad News for Influenza Antivirals

Deep in the throes of influenza season, I’m sure the oseltamivir is flying off the shelves around the country. In Japan, however, it’s baloxavir that’s flying off the shelves. Unfortunately, as was presaged by the data from their definitive clinical trial, resistance to baloxavir is rapidly increasing.

And, now, tucked into this retrospective look at “early” versus “late” oseltamivir treatment in the critically ill – additional data regarding its general futility. In this 1,330 patient ICU cohort of patients who received osteltamivir within 48 hours of symptom onset (“early”) or later (“late”), overall mortality was 46.8% – and no different between the two groups. There are obvious issues here with regards to confounding and baseline differences, but it should be apparent a beneficial treatment … provides some benefit.

The authors did observe an absolute 10% survival advantage associated with “early” treatment in those infected with A/H3N2 – but as this accounted for a minority of their cases, overall, the entire cohort was a wash. This is consistent with another review specific to data from the 2009A/H1N1 pandemic. Mortality in included studies was only 8%, but no survival advantage was seen in those treated with oseltamivir. While universal and indiscriminate treatment with neuraminidase inhibitors is engrained in the conflict-of-interest-infested IDSA guidelines, one can only hope these data points encourage additional prospective evaluation into the true narrow value of our tools for the treatment of influenza.

“Effect of early oseltamivir treatment on mortality in critically ill patients with different types of influenza: a multi-season cohort study”

https://www.ncbi.nlm.nih.gov/pubmed/30753349

Lactate is Dead! Long Live Lactate?

Our use of serum lactates as targets for resuscitation in sepsis is more than a little flawed. Once upon a time, we resuscitated using central venous oxygenation as part of the Rivers’ trial. Whether those targets were actually a valid part of the multi-pronged bundle remains an excellent and open question. Of course, CVO2 requires invasive monitoring – and serum lactate became our less-invasive surrogate. And, yes, patients with high lactates do poorly – but that doesn’t specifically address the assumption lactate-guided resuscitation is tied to outcomes, or the optimal resuscitation strategy.

This multi-center trial out of Latin America looks at another marker of perfusion status, capillary refill time, that is likewise observationally associated with mortality in sepsis. In a randomized, open-label trial, the first eight hours of resuscitation was guided either by lactate levels or capillary refill time. Resuscitation in both arms used a specific protocol of fluids, fluid-responsiveness assessments, vasopressors, and inodilators.

Without unpacking these specifics in too great of detail, as will be done by many other critical care physicians, the results are quite interesting: of 424 patients randomized, they observed 34.9% mortality in the CRT-guided cohort compared with 43.4% in the lactate-guided cohort. Other secondary outcomes, including lactates at 48 and 72 hours, SOFA scores at 72 hours, generally favored the CRT cohort.

Is this the end of lacate? Certainly, in a resource austere setting, it would generally indicate there’s no rush to adopt lactate use in the context of a just-as-good, zero-cost means of assessment. The accompanying editorial wonders aloud: why not use both? While this seems like a reasonable idea, it probably doesn’t go far enough – why not use all the data for an individual patient to determine their optimal treatment, rather than our current one-size-fits-all nuclear option? Reliance on any single approach to resuscitation – perhaps mandated by “quality” measures – is almost certain to be short-sighted. While I do not advocate a return to the wild west of late recognition and neglect, these data should add further fuel to a reassessment of our golden idols and targets in the treatment of sepsis.

“Perfusion Status vs Serum Lactate Levels on 28-Day Mortality Among Patients With Septic Shock: The ANDROMEDA-SHOCK Randomized Clinical Trial”

https://jamanetwork.com/journals/jama/fullarticle/2724361

Who Recanalizes with Just tPA?

The original argument: tPA helps all strokes, we must give it to everyone as quickly as possible!
The updated argument: tPA doesn’t not help all strokes, so it should still be given!

Specifically, as applies to the cohort of patients with large vessel occlusions being considered for mechanical thrombectomy. This small, pooled registry sample looked at cases from four centers, evaluating the rate and predictive characteristics for recanalization prior to cerebral angiography. The stated purpose of their study was to develop a predictive score, with the reasonable goal of reducing unnecessary tPA exposures prior to thrombectomy.

The numbers, in their score derivation and validation cohorts:

  • ICA: 6.4%/1.0%
  • M1 proximal: 16.1%/13.7%
  • M1 distal: 30.3%/30.7%
  • M2: 33.7%/34.0%

But, an even more powerful a predictor was thrombus length, as measured by T2 MRI susceptibility vessel sign. Recanalization was seen at over 80% for clots <5mm, 30% for 6-10mm, and below 10% for clots longer than 10mm, with particular futility for >20mm.

Interesting data – and a nice look at how not all sites of occlusion and clots are created equal. Whether, and how, we ought to treat them differently remains uncertain until the results of a prospective trial.

“Post-Thrombolysis Recanalization in Stroke Referrals for Thrombectomy”
https://www.ahajournals.org/doi/pdf/10.1161/STROKEAHA.118.022335

Disutility, Thy Name is ANEXXA-4

About two and a half years ago, we were introduced to andexanet alfa (Andexxa), a modified recombinant form of factor Xa designed as a reversal option for factor Xa inhibitors. The mechanism of action is simple: andexanet mimics native factor Xa, providing the various Xa inhibitors (rivaroxaban, apixaban, edoxiban, betrixiban, and enoxaparin) an alternative target. When sufficient Xa inhibitor is bound to andexanet, the native version is freed to return to its normal work in hemostasis.

The problem, however, is the reversal is not durable. The half-life of andexanet is approximately 1 hour, after which point the factor Xa inhibitor levels rise and hemostasis is again impaired. This necessitates a bolus and an infusion. A bolus and infusion that costs ~$3,000 per 100mg vial – and requires 900mg for a “low dose” protocol or 1,800mg for a “high dose” protocol. And, it all vanishes to dust when the infusion completes.

The first NEJM publication regarding andexanet featured just the first 67 patients from ANNEXA-4, an open-label, single-arm descriptive study of patients given andexanet for major bleeding. Now, presented at the International Stroke Conference 2019 in Honolulu, we have the results from the full 352 patient cohort – and they’re every bit as uninspiring as the preview.

Nearly all patients were receiving rivaroxaban (half-life 7 to 13 hours) or apixaban (half-life 12 hours), with a handful on enoxaparin. As in the preview, the bolus of andexanet effectively dropped circulating levels of the factor Xa inhibitor down to negligible amounts. Again, as seen before, after cessation of the bolus, factor Xa levels returned to a level consistent with their terminal half-lives. Case in point: “At 4, 8, and 12 hours after andexanet infusion, the median value for anti–factor Xa activity was reduced from baseline by 32%, 34%, and 38%, respectively, for apixaban and by 42%, 48%, and 62%, respectively, for rivaroxaban.” This is just normal metabolism, not andexanet magic.

Death occurred in 14% of patients within 30 days, and there were thrombotic events in 10%. Hemostasis at 12 hours, their primary outcome, was 82% “in 204 of 249 patients who could be evaluated”. Specifically, they excluded a third of their population because they were effectively enrolled in error, as they met bleeding criteria but not Factor Xa inhibitor level criteria. This is similar to the idarucizumab open-label demonstration, in which many patients who were not actually coagulopathic were treated with anticoagulation reversal. This represents tremendous waste and cost.

Finally, the nail in the coffin, this admission in the abstract and the text: “Overall, there was no significant relationship between hemostatic efficacy and a reduction in anti–factor Xa activity during andexanet treatment.” The primary outcome wasn’t even correlated with their intervention!

As you can tell from the tone of this post, I am profoundly unimpressed with the value demonstrated here. There’s no evidence this is clinically useful, nor a potentially preferred strategy to use of prothrombin concentrate complexes to replete missing factor. The company and the FDA effectively admit the same:

The most important limitation of this trial is that it did not include a randomized comparison with a control group. At the time of study initiation, it was determined that a randomized, controlled trial would have logistic and ethical challenges, given the perceived risks of placebo assignment in this highly vulnerable population. However, continued use of unapproved agents, despite a lack of rigorous clinical data, has changed the equipoise for a trial. Thus, under the guidance of the FDA and as a condition of accelerated approval in the United States, the sponsor is conducting a randomized trial (ClinicalTrials.gov number, NCT03661528) that is expected to begin later this year.

I’m sure more will be made to unpack even further unfavorable details as time passes – and, until further reliable evidence can be presented, I’d pass on andexanet, as well.

“Full Study Report of Andexanet Alfa for Bleeding Associated with Factor Xa Inhibitors”

https://www.nejm.org/doi/full/10.1056/NEJMoa1814051


Telehealth Triage for EMS

We’ve all seen non-emergency patients in our Emergency Departments. Further still, we’ve seen those same non-emergency patients arrive via emergency medical transport. Per these authors, the estimated burden of non-emergency or medically unnecessary ambulance transport ranges from 33-50%.

Houston, Texas, has a program of prehospital telehealth support provided by online board-certified Emergency Physicians. This article describes their retrospective cohort from 2015 through 2017, in which 15,067 patient encounters occurred from a total of 865,000 EMS incidents. Patients were eligible for telehealth if they met certain vital sign, chief complaint, and age criteria, and could be transported via non-ambulance alternative.

The good news: nearly everyone the EPs consulted with over telehealth was diverted from ambulance utilization. Only 11.2% of patients were ultimately transported by EMS –while basically the entire remainder utilized a taxi service. The bad news: nearly everyone was still transported to an Emergency Department. Only 5.0% of patients accepted same-day or next-day referral for follow-up at an affiliated outpatient health center. The primary advantage of this service, then, is increasing the availability of the limited ambulance services resource to respond to higher-acuity patients.

There are more than a few issues at play here in our current system with regard to providing, effectively, an unpaid, low-fidelity evaluation and assuming the liability risk. However, in systems with different structures and payment models where the overall costs to the system from ambulance utilization outweigh the other costs, this has a great deal of potential. Whether these results are generalizable is a reasonable concern, although the proportion of patients being referred to non-ambulance transport is not surprising. The entry criteria for telehealth consultation substantially narrowed the eligible population specifically to those whose complaints likely did not merit emergency transport. Finally, whether the EPs needed to be involved is another matter that could potentially be solved by more robust protocols in place to defer transport.

“Telehealth Impact on Primary Care Related Ambulance Transports”

https://www.ncbi.nlm.nih.gov/pubmed/30626250