Month: May 2018

Dialysis Purgatory

This little article is not terribly generalizable and the results are eminently predictable – but this barbaric practice was my daily routine a couple years ago while working in Houston.

To received routinely scheduled dialysis, someone must, of course, pay for it. For those U.S. citizens without the means to pay, various federal mechanisms provide coverage. For non-U.S. citizens in this country without the means to pay, there is no external payor source – unless “necessary for the treatment of an emergency medical condition.” Therefore, many facilities simply restrict dialysis to non-citizen patients in extremis.

As you might expect, this is bad and bad for you. This retrospective cohort study follows 5-year survival and resource utilization, comparing those at hospitals where non-U.S. citizens could be eligible for scheduled dialysis with those at facilities where only emergency dialysis was available. The sample sizes are small – 169 in the “emergency-only” cohort and 42 in “standard” – but at 5 year follow-up, over half the “emergency-only” cohort had died, as compared with about 10% of the standard hemodialysis. Various statistical analyses and propensity matching further quantify the exact excess hazard of emergency dialysis.

Patients receiving emergency dialysis received, obviously, fewer sessions per month – 6.2 instead of 10.3. However, these savings are potentially offset by a ten-fold increase in acute care days, probably associated with those episodes of emergency dialysis. In the most morbid sense, unfortunately, total cost-savings probably favor the emergency dialysis cohort owing to the greatly increased mortality.

There are uncertainties and holes in these sorts of retrospective studies, particularly in a cohort whose deaths are potentially not documented in our national registries. The face validity for the overall findings is strong, however, even if the specific numbers are not reliable.

Regardless, just as a manner of respecting basic human decency, it is simple cruelty to layer this additional suffering onto the already miserable state of being dialysis-dependent. All feasible efforts should be made to provide access to regular dialysis, considering the burden on the health system infrastructure is likely similar.

“Association of Emergency-Only vs Standard Hemodialysis With Mortality and Health Care Use Among Undocumented Immigrants With End-stage Renal Disease”
https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2665387

If It Bleeds, It Can Get TXA?

When trauma bleeds: TXA! When women bleed: TXA! When the nose bleeds: TXA! When your freckles need lightening: TXA!

But, what about inside the brain?

This is TICH-2, an international, randomized, placebo-controlled trial testing tranexamic acid versus placebo for patients with primary intracerebral hemorrhage. The intervention arm received 1g of IV TXA as a bolus, followed by 1g over the subsequent 8 hours. The primary outcome was functional status at day 90 measured – inappropriately so, of course – as shift on the modified Rankin Scale. We’ve critiqued the ordinal shift several times as, effectively, statistically magnifying unimportant differences as a crutch for trials struggling to find a difference using a traditional, dichotomous endpoint.

However, regardless, their efforts are for naught: their primary endpoint still failed to reach statistical significance. Across five years and 2,325 randomized participants, nearly all patient-oriented outcomes showed no difference: 29% of TXA patients were mRS ≥2 at 90 days, compared with 29% of placebo. The numbers of deaths by day 90 were virtually identical, as were measures of quality of life, functional status, and days at home. Adjusted analyses and various subgroups generated odds ratios whose confidence intervals almost broke free of unity, but not quite.

The major quirk – over two-thirds of the trial was randomized greater than 3 hours from onset. The trauma literature focuses on early anti-fibrinolytic treatment, and it is reasonable to suggest the delay in treatment was too great to demonstrate a benefit. Then, even though no patient-oriented benefit was observed, hematoma expansion was attenuated in the TXA cohort. This is not the first time an ICH trial has seen benefits with regard to hematoma expansion absent patient-oriented outcome improvements, but it still seems a valid surrogate for, at least, a small effect size for which this trial may (or not) be underpowered to detect.

My takeaway is this trial hasn’t done much to move the needle with regard to evaluating TXA in ICH. It does show, at least, as administered in this trial, it is unlikely to have substantial benefit. However, TXA is inexpensive and seems to demonstrate a reasonable margin of safety. It is still reasonable to consider its use in as timely a fashion as possible, with the expectation the true NNT may be ~50 to 200, while awaiting further data from other trials currently underway.

“Tranexamic acid for hyperacute primary IntraCerebral Haemorrhage (TICH-2): an international randomised, placebo-controlled, phase 3 superiority trial”
https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(18)31033-X/fulltext

Anti-Calcitonin

The use of procalcitonin to guide antibiotic therapy has been gradually increasing over the past several years – driven, in no small part, by increased recognition of the harms of antibiotic overuse. However, what evidence we have regarding its utility is primarily derived from manufacturer-sponsored trials – including virtual carpet-bombing of the literature by their sponsored representatives.

So, what happens when the manufacturer isn’t part of the trial?

No benefit.

This is the ProACT trial, an individual-randomized comparison between a procalcitonin-guided arm and “usual care” in patients with suspected lower respiratory tract infection for whom the indication for antibiotics is unclear. Physicians caring for patients randomized to the procalcitonin arm were provided results tied to antibiotic use recommendations – “strongly discouraged”, “discouraged”, “encouraged”, “strongly encouraged” – on initial presentation in the Emergency Department, and then in serial fashion for those admitted to the hospital. In those in the “usual care” arm, procalcitonin results were obtained, but not provided to the treating clinicians.

Then: Across 14 hospitals and 1,656 patients, there were no statistically significant differences between antibiotic-free days or adverse outcomes between the two arms. Done? Done.

Except, as skeptical as I might be regarding procalcitonin-guided therapy, there are big holes in these data as the definitive word on its disutility. Unlike other trials, these centers provided only passive guidance to clinicians regarding the procalcitonin algorithm. This resulted in only 72.9% of physicians adhering to protocol, with the greatest numbers of violations being antibiotic use in patients for whom it were discouraged, including 30% of those for whom antibiotic use was “strongly discouraged”:

Even though the “per-guideline” analysis also shows no difference, this is mostly because the bulk of the procalcitonin “per-guideline” population were those who appropriately received antibiotics – effectively eliminating the possibility of showing a difference in antibiotic use.

There are a few signals within these data reflecting the potential advantages of a procalcitonin-guided algorithm, should the protocol actually be followed. There were small differences in prescribing favoring the procalcitonin arm for almost every final clinical diagnosis – excepting about 15% absolute advantages for “acute bronchitis” and for those with non-specific diagnoses. It is likely these represented the cases for which the appropriateness of antibiotics was lowest, and probably also represent the majority of protocol violations. That said, one could easily make the argument this advantage only exists as a result of culturally-ingrained poor antibiotic prescribing habits for these sorts of borderline cases.

In short, these data clearly show there is no advantage to introducing procalcitonin into practice specifically in the fashion demonstrated here – but these cannot be generalized to say a different implementation or application of procalcitonin has no value.  On the flip side, however, places that have implemented procalcitonin-driven stewardship programs also struggle with inappropriate and high-volume test ordering.  There is work yet to be done for both proponents and skeptics of its value.

“Procalcitonin-Guided Use of Antibiotics for Lower Respiratory Tract Infection”
https://www.nejm.org/doi/full/10.1056/NEJMoa1802670

Wake Up and Smell the tPA

What happens when you wake up and you’re paralyzed from a stroke? Well, usually nothing. “Unknown time of onset” takes you – for better or worse – out of the game for alteplase, but not necessarily for endovascular therapy should a large-vessel occlusion be identified. Those large vessel occlusions, in the setting of a favorable CT perfusion profile, seem to benefit from endovascular therapy.

But, getting back to the “wake up stroke” – these have had our neurologists gnashing their teeth for some time. They have hypothesized many of these strokes have occurred just before waking and might otherwise be eligible for treatment. Absent reliable presenting information regarding the time of onset, these authors look to MRI – using presence of DWI lesion without corresponding FLAIR signal as a surrogate for tissue viability/stroke recency. Exclusion criteria in addition to the usual alteplase culprits were extremes of age, premorbid functional disability, NIHSS >25, thrombectomy candidates, and those with infarct volumes greater than 1/3rd the MCA territory. The primary outcome was 0 or 1 on the mRS at 90 days, like most trials.

These authors in this multicenter, placebo-controlled planned to enroll 800 patients, but ran out of money after five years and 503 patients. To get to these 503, the authors needed to screen 1362 potential strokes. These 859 exclusions were for various reasons, but over half were because the FLAIR matched the DWI lesion – indicated a completed infarct. Another 137 had negative DWI – i.e., not stroke – and various others had hemorrhage, failed to meet criteria for infarct size, or a scattering of other exclusions. Even despite these exclusions, another 79 snuck through as protocol violations, including 48 who should have been excluded based on imaging criteria.

Now, the meat: About 95% of those included were of the “wake up” variety, nearly all from overnight sleep. Baseline clinical features were generally well-matched. Median NIHSS was 6 in each group, although median lesion volume on DWI was 2.0 mL in the alteplase cohort as compared to 2.5 with placebo. At 90 days, 53.3% of the alteplase cohort achieved an mRS of 0-1 as compared with 41.8% with placebo. Bleeding complications, as typical, favored the placebo cohort – with absolute advantages ranging from 1.6% to 3.6%, depending on the definition of hemorrhage used. Death at 90 days also favored placebo at 1.2% versus 4.1%.

It is difficult to know what to do with these data unless your system is specifically equipped to replicate the conditions of this trial with rapid MRI. Even then, there are some oddities and specific warnings to unpack. If adhering to this protocol, the majority of patients screened will not be eligible for treatment. The number of patients who had completed their infarction was similar to those who had DWI/FLAIR mismatch, and another third had other imaging or clinical findings excluding them from treatment. Incorporating MRI into workflow may not yet represent a high-value approach.

Then, the authors performed a pre-specified subgroup analysis stratifying based on NIHSS – and the 109 analyzed patients with a NIHSS >10 did terrible. Only 13.5% of those in the tPA cohort and 12.3% of those receiving placebo achieved mRS 0 or 1. Unpacking these stratifications further, the authors provide us a whole host of breakdowns:

Generally, not too much should be read into these secondary outcomes, but they are useful for generating equipoise for other investigations.  That said, these data should be at least a useful cautionary tale regarding the value of tPA in the setting of mild, but disabling stroke – as these 175 patients represent at least six times more patients than from NINDS, and are of higher quality evidence than any of the Get With the Guidelines publications trying to build the case for tPA in mild stroke.

One takeaway that should definitely not be generated from this is: “well, if there’s an absolute increase in good outcome of 12% on those screened with MRI, then treating all ‘wake up’ patients after screening with just CT could generate about a 6% absolute benefit, and that should be offered to patients.”  Unfortunately, I suspect we will hear such calls – probably based on parsing out the low NIHSS patients in the subgroups above, and trying to toss out the ~30% with a large-vessel occlusion identified on MRA as patients who should be triaged to endovascular.  Again, trying to pick and choose the secondary outcomes that suit your narrative is fraught with peril, and the fact remains such a treatment strategy is also likely to generate harms greater than those seen in this trial.  These data ought to have a very narrow application – but shareholders and executives don’t realized dividends when alteplase isn’t flying off the shelves for expanded indications.

“MRI-Guided Thrombolysis for Stroke with Unknown Time of Onset”
https://www.nejm.org/doi/full/10.1056/NEJMoa1804355

Snacking Before Bedtime

How long before a procedural sedation is fasting required? You know, of course, the American Society of Anesthesiologists guidelines specify: a mini- mum fasting period of 2 hours for clear liquids, 4 hours for breast milk, 6 hours for infant formula and light meals, and 8 hours for solids containing meat or fatty foods.

Of course, anecdotally – if anecdotally means hundreds of thousands of safe sedations – Emergency Physicians have known these restrictions are nonsense.

But, guidelines are best written off published evidence – so, we have a pre-planned analysis of the relationship between fasting time and vomiting from a Canadian cohort study of pediatric sedation. With 6,295 sedations included in their analysis, almost half of whom did not meet solids fasting guidelines, these authors found no relationship between fasting time and vomiting. There were, even, only six instances of intra-procedure vomiting, and fasting duration ranged from 1.7 hours to 17.5 hours – but they all received ketamine. None of the intra-procedure, or ~300 peri-procedural episodes of vomiting, resulted in pulmonary aspiration. No relationship was found between fasting time and any other type of adverse event, either.

So, another useful piece of literature to wave around in committee meetings – both to eliminate any fasting restrictions, and, again, to help demonstrate the safety of EP-performed procedural sedation.

“Association of Preprocedural Fasting With Outcomes of Emergency Department Sedation in Children”
https://jamanetwork.com/journals/jamapediatrics/article-abstract/2680050

Whoa! Fosfomycin in Prime Time!

For many years, I’ve tossed out the idea of using fosfomycin for uncomplicated urinary tract infections to various trainees – the vast majority of whom looked at me as though I had three heads. Even now, it’s easy to find folks who’ve never heard of fosfomycin, despite its mention in the most recent guidelines for UTIs. In the United States, the land of low-value health care, fosfomycin is preposterously expensive for a single dose – and rarely used.

The story, however, is a little different outside the U.S. Thus, the question – which is a better options, fosfomycin or nitrofurantoin? The answer: in Israel, Switzerland, and Poland, nitrofurantoin, probably.

This is an open-label trial with 513 patients randomized either to five days of nitrofurantoin or a single 3g dose of fosfomycin. Outcomes included clinical and microbiologic cure, and both favored nitrofurantoin by an absolute margin of ~10%. Oddly enough, their primary outcome was a 28-day clinical cure – which starts to stretch the measurement window into the range of subsequent, unrelated infection, rather than response to the initial therapy. This is apparent when looking at the 14-day and 28-day microbiologic response, in which bacterial counts were clearly creeping back up after an initial nadir.

Regardless, both agents are options – and fine options, depending on local resistance patterns, suspected pathogens, and other contextual clinical features. That said, in the U.S., most of the current appropriate prescribing is for trimethoprim-sulfamethoxazole – so, a similar trial comparing this with these alternative agents would need to be performed to better inform practice here.

“Effect of 5-Day Nitrofurantoin vs Single-Dose Fosfomycin on Clinical Resolution of Uncomplicated Lower Urinary Tract Infection in Women”
https://jamanetwork.com/journals/jama/article-abstract/2679131

Don’t Rely on the EHR to Think For You

“The Wells and revised Geneva scores can be approximated with high accuracy through the automated extraction of structured EHR data elements in patients who underwent CTPA in the emergency department.”

Can it be done? Can the computer automatically discern your intent and extract pulmonary embolism risk-stratification from the structured data? And, with “high accuracy” as these authors tout in their conclusion?

IFF:  “High accuracy” means ~90%. That means one out of every ten in their sample was misclassified as low- or high-risk for PE. This is clinically useless.

The Wells classification, of course, depends highly upon the 3 points assigned for “PE is most likely diagnosis.” So, these authors assigned 3 points positive for every case.  This sort of probably works in a population that was selected explicitly because they underwent CTPA in the ED, but is obviously a foundationally broken kludge.  Revised Geneva does not have a “gestalt” element, but there are still subjective examination features that may not make it into structured data – and, obviously, it performed just as well (poorly) as the Wells tool.

To put it mildly, these authors are overselling their work a little bit. The electronic health record will always depend on the data entered – and it’s setting itself up for failure if it depends on specific elements entered by the clinician contemporaneously during the evaluation. Tools such as these have promise – but perhaps not this specific application.

“Automated Pulmonary Embolism Risk Classification and Guideline Adherence for Computed Tomography Pulmonary Angiography Ordering”
https://onlinelibrary.wiley.com/doi/abs/10.1111/acem.13442

The Futility of Alteplase

This article is mostly a story about tenecteplase, but, effectively, it’s also a scathing indictment of alteplase – you know, the miraculous “clot-buster” we’ve been using for the past 23 years.

Because, despite rumors to the contrary, it doesn’t actually bust clots.

This isn’t news to anyone who actually follows the stroke literature closely. Indeed, the entire endovascular/thrombectomy industry is constructed upon this edifice of failure. And, as we see in this 202-patient study comparing recanalization rates after large-vessel occlusion, tenecteplase appears to be more efficacious than alteplase – 22% vs. 10%.

That is to say, in a population of 24 ICA occlusions, 3 basilar occlusions, 60 M1 occlusions, and 14 M2 occlusions, alteplase successfully “busted the clot” in 10. Most of the difference in recanalization was driven by the M1 and M2 patients, where tenecteplase had a 26% success rate and alteplase languished at 8%. These rates for alteplase are a little lower than most prior literature, so it is reasonable to be suspicious of the superiority margin associated with tenecteplase.

But, regardless, as you can see, both are dismal – and, for the past 20+ years, prior to the advent of even limited endovascular availability, we’ve just been pushing alteplase on these large vessels to no beneficial effect – except to Genentech and their shareholders.

“Tenecteplase versus Alteplase before Thrombectomy
for Ischemic Stroke”
https://www.nejm.org/doi/full/10.1056/NEJMoa1716405