Monday, January 23, 2012

Progress In Combating Publication Bias

...if from abysmally terrible to embarrassingly bad represents progress.

A certain subgroup of trials registered with are required to report their results within one year of conclusion of study.  These mandatory-reporting requirements include clinical trials of FDA-approved drugs, devices, or biological agents that have at least one study site in the U.S.  In the future, this will expand to include unapproved drugs.  These requirements, ideally, should help reduce publication and sponsorship bias by ensuring result availability regardless of ability to obtain publication or the desire of a pharmaceutical corporation to publish negative results.

And, so far, these authors discover that it is a tremendous success - trials subject to the mandatory reporting complied with the requirement in twice as many of identified trials as compared with registered trials that were not required to report results.

Unfortunately, twice as many was only 22% compared with 10%.  So, there's still quite a ways to go before we have full transparency in clinical trial reporting - but it's "progress."

"Compliance with mandatory reporting of clinical trial results on cross sectional study"

Saturday, January 21, 2012

Observation For Anticoagulated Head Trauma

Coming in a future issue of Annals, the Editor's capsule summary: "Delayed intracranial hemorrhage is common after minor head injury when patients are receiving warfarin. A minimum protocol of 24-hour observation followed by repeated scanning is necessary to detect most such occurrences."

Now, this isn't a terribly management agnostic statement.  It does not specifically state this is something we need to start doing - but it rather implies that, if you don't, you'll be missing this "common" phenomenon.  It isn't an alien concept - since 2002, the European Federation of Neurological Societies has recommended admission for observation after minor head trauma - but it's certainly not the standard of care here.  So, for the Annals editors to state that observation and repeat scanning is "necessary", they must obviously have excellent evidence.

Or they have an observational case series consisting of 87 patients from Italy.

These authors present a prospective case series of all patients at their institution who were admitted for observation specifically for minor head trauma while on oral anticoagulation.  At the time of repeat CT scanning 24 hours later, the authors report five of them had new bleeding detected.  In addition, two patients who were discharged after two negative CT scans returned with symptomatic bleeding, one at two days, and one at eight days.

So, should we be observing and rescanning every anticoagulated minor head trauma patient as these authors suggest (and as they do in Europe)?  If you practice in a zero-miss litigation environment, this article and ACEP's apparent embrace of the results will hamstring your decision-making.  This data is completely inadequate to change clinical practice, and inconsistent with prior literature documenting delayed hemorrhage in only 2 of 137 patients.

Clearly, some patients will have delayed bleeding - a subset of which will be clinically significant.  However, we simply cannot expose all anticoagulated patients with minor head trauma to the harms and costs of hospitalization.  Better studies are required to prospectively determine the risk profile of patients who require further observation in a hospital setting, rather than a watchful discharge home.

"Management of Minor Head Injury in Patients Receiving Oral Anticoagulant Therapy: A Prospective Study of a 24-Hour Observation Protocol"

Friday, January 20, 2012

Correspondence in Nature Reviews Neurology

After publishing an article that alluded to the "antipathy" of emergency physicians towards TPA in acute ischemic stroke, the editors were nice enough to publish my correspondence defending the reasonableness of a cautious attitude.

"Skepticism about thrombolytics in stroke is not unreasonable."

Thursday, January 19, 2012

EMLit on ERCast

It's like podcasting with Jack Buck - always a pleasure.  Rob Orman does great work, and he's got a huge archive of invaluable wisdom accumulated through his podcasts.  I hope you enjoy our most recent episode together.

"Decision Tools: PERC, NEXUS and CURB-65"

Lies, Damned Lies, and Tamiflu (oseltamivir)

Receiving quite a bit of press yesterday, and rightfully so, the Cochrane Collaboration published their analysis of oseltamivir - the miracle influenza antiviral that (at great cost) is part of our nation's strategic stockpile for an influenza pandemic.  The story is interesting both regarding what they found, and what they didn't find.

As for the data from the review, the numbers are similar to what we've been basing our practice upon - oseltamivir significantly shortens the length of time until symptom improvement from 160 hours to 139 hours.  However, it did not demonstrate any difference in hospitalization rates.  Additionally - whether through study bias or by direct medication effects - the oseltamivir groups were significantly less likely to have a confirmed diagnosis of influenza.

So, this suggests that it's a little troubling that we've gone to all the expense to stockpile this expensive medication that does not appear to reduce hospitalizations from influenza - and it remains an individual decision whether that extra day of symptom improvement is worth exposure to the side effects of the medication.  But the reason this is national news is that Roche pharmaceuticals refused to supply the promised clinical data requested by the Cochrane Collaboration; the published analysis is based on 15 oseltamivir studies with complete information, and excludes 42 other studies with discrepancies in the data.  This sort of behavior is just another representative sample of the unethical, but completely understandable, profit-motivation of pharmaceutical corporations protecting their financial interests.

I would be greatly surprised if the clinical data Roche is holding onto supports oseltamivir efficacy.

"Neuraminidase inhibitors for preventing and treating influenza in healthy adults and children - a review of clinical study reports"

"Flu Drugs: Search for evidence goes on"

Tuesday, January 17, 2012

100,000 Incorrect TIA Diagnoses Every Year

...if we extrapolate the results from this single-center study to the entire United States.

In Annals this month, a chart review of the "discordance" between the final neurologist "gold standard" diagnosis and the provisional Emergency Department diagnosis.  Apparently, in Cleveland, 36% of the patients receiving an initial diagnosis of transient ischemic attack in the Emergency Department are subsequently evaluated by a neurologist and given an alternative diagnosis.  As the authors note in their introduction, the diagnosis of TIA is made 300,000 yearly - and if 36% of those cases are made incorrectly, then we're theoretically admitting 100,000 patients for extensive and expensive evaluation.

So, are they right?  Well, if the three neurologists responsible for 93% of their 427 evaluations are representative of the entire country, perhaps.  Or, if their chart review methods are adequate - as the authors note, one of their chart audits changed an abstracted diagnosis of TIA to "right hip pain" - then, perhaps.  If you ignore that ED physicians have a few minutes of history, examination, and limited imaging available at their disposal - compared with the neurologists that can subsequently perform any manner of inpatient studies that might uncover an alternative diagnosis mimicking a TIA - then, perhaps.

If neurologists are walking into the ED and evaluating patients under the same constraints as we are and producing this level of discordance, then we have a problem.  But, I don't think this study tells us anything we can use to evaluate ED physicians' ability to appropriately include or exclude TIA in the differential for neurologic complaints of a transient nature.

Somewhat disappointing that a small, retrospective chart review with results that might not be internally or externally valid are in the premier, #1 Impact Factor journal of our profession.

"Variables Associated With Discordance Between Emergency Physician and Neurologist Diagnoses of Transient Ischemic Attacks in the Emergency Department"

Sunday, January 15, 2012

PERC - Still (Mostly) Useless

...except, perhaps, in a risk-management sense - but, only if we keep beating it down into its narrowest application due to its terrible specificity.

This most recent Annals publishes a systematic review of the Pulmonary Embolism Rule-Out Criteria, a decision instrument recommended in ACEP's pulmonary embolism clinical guidelines as a reasonable tool to risk-stratify a patient into a so-called "zero-risk" population that does not require any testing - not even a D-dimer.  And, I think they do a reasonable job including studies and summarizing the data, especially considering the width of the error bars on a lot of these studies.

The key points - pooled sensitivity is 97% when applied to a low-risk (Wells, Geneva, gestalt, whichever) population with a negative LR of 0.18.  This means, if you had someone who you already didn't think had a PE and they meet PERC criteria, it helps you with your medicolegal documentation, since it's in ACEP's guidelines.  The negative LR is strong enough to be helpful - but when you're already looking at single-digit percentage risk for PE, the absolute reduction in risk is quite small.

The important point to hammer home is the positive LR is only 1.23, which makes it the D-dimer of decision instruments.  Please don't justify further work-up just because they fail PERC - it barely moves the needle with its terrible specificity.  You need to have another clinical justification for further work-up in pulmonary embolism.

As an aside, in this era of over-testing and over-diagnosis of PE, the diagnosis of PE isn't necessarily the ideal endpoint - what we should be following are patient-oriented outcomes such as death/heart failure in untreated PE in PERC-negative patients to truly make it a valid tool.

"Diagnostic Accuracy of Pulmonary Embolism Rule-Out Criteria: A Systematic Review and Meta-analysis"

Friday, January 13, 2012

Heart Failure, Informatics, and The Future

Studies like these are a window into the future of medicine - electronic health records beget clinician decision-support tools that allow highly complex risk-stratification tools to guide clinical practice.  Tools like NEXUS will wither on the vine as oversimplifications of complex clinical decisions - oversimplifications that were needed in a pre-EHR era where decision instruments needed to be memorized.

This study is a prospective observational validation of the "Acute Heart Failure Index" rule - derived in Pittsburgh, applied at Columbia.  The AHFI branch points for risk stratification described below, in this extraordinarily complex flow diagram:

Essentially, the research assistants in the ED applied an electronic version of this tool to all patients given by the Emergency Physician a diagnosis of decompensated heart failure - and then followed them for the primary outcome(s) of death or readmission within 30 days.  In the end, in their small sample size, they find 10% of their low-risk population meets the combined endpoint, while 30.2% of their high-risk population meets their combined endpoint.  Neither group had a very high mortality - most of the difference between groups comes from re-admissions within 30 days.

So, what makes this study important isn't the AHFI, or that it is reasonable to suggest further research might validate this rule as an aid to clinical decision-making - it's the progression forwards of using CDS in EHR to synthesize complex medical data into potentially meaningful clinical guidance.

"Validating the acute heart failure index for patients presenting to the emergency department with decompensated heart failure"

Wednesday, January 11, 2012

Can We Stop Placing NG Tubes?

One of the worst-tolerated procedures in Emergency Medicine - placement of the NG tube.  Unfortunately, when I call my GI fellow on-call for any upper GI bleeding, the first question is invariably - what did the NG lavage show?

There is good evidence demonstrating that positive NG lavage tends to identify the presence of high-risk lesions found on subsequent endoscopy.  There is also evidence that endoscopic treatment of high-risk lesions decreases rebleeding and mortality.  So, if NG lavage identifies high-risk lesions, and endoscopic treatment of high-risk lesions decreases mortality, then patients who undergo NG lavage for their upper GI bleeds should have lower mortality, right?

This is a retrospective review of all the patients admitted to the West LA VA with a diagnosis of upper GI bleeding - a sample of 632 meeting inclusion criteria.  Of these, 255 did not undergo NGL and 378 did.  What's interesting in this article is that the authors attempted to statistically create two identical cohorts using propensity scoring.  They ended up with two nearly identically matched cohorts of 193 patients from the original 632 based on demographics, triage, lab values, physiologic characteristics, and medical interventions.

Between these two groups, they found no significant difference between mortality, hospital stay, emergency surgery, and blood transfusion requirements.  There was a statistically significant difference in the number of patients who underwent endoscopy - patients who didn't receive NGL had 60% endoscopy vs. 72.3% in the NGL group.  This is mildly interesting - considering that, in theory, the identification and endoscopic treatment of high-risk lesions is associated with increased survival - and if you're doing less endoscopy on an identical patient cohort, you should be missing the opportunity to treat those lesions.  Yet, there was no significant difference outcomes between cohorts.

So, yes, if you wanted to stop placing NG tubes because they're uncomfortable for patients and apparently don't change ultimate outcomes - certainly, that may be reasonable.  Some gastroenterology literature suggests patient-specific risk factors are more important in predicting the impact of endoscopic intervention on outcomes, rather than the limited information derived from the NG lavage.

However, this is just statistical calisthenics in an attempt to replicated a randomized-controlled trial and doesn't give us the prospective evidence needed to change practice.  Or argue over the phone with the GI fellow.

"Impact of nasogastric lavage on outcomes in acute GI bleeding"

Monday, January 9, 2012

So, NEXUS Is Invalid?

Another doom and gloom trauma article that wants to take one of our most cherished tools away from us regarding the evaluation of the blunt trauma patients.  Certainly, nothing is sacred, but these authors want to take NEXUS out to the woodshed and make sure every trauma patient gets a CT of the c-spine.

The premise of their argument is reasonable - NEXUS was derived in an era of plain films for radiographic clearance of the cervical spine, and now, many studies have observed that CT with 3D reconstruction picks up potentially significant injuries that could be missed by plain x-rays.  Therefore, the gold standard incorporating plain radiography for NEXUS renders it invalid due to missed injuries.

These authors performed a prospective evaluation of the NEXUS rules by applying them to 2,606 adult trauma patients, all of whom underwent 16 multidetector CT scanning with 2mm thick axial cuts.  They found 157 patients with a total of 258 fractures - and note that 26 patients had fractures identified despite meeting NEXUS criteria.  Of these 26, 16 were managed in a c-collar, 2 underwent operative stabilization, and 1 had a halo placed.  Therefore, they simply conclude that NEXUS is not externally valid to their trauma population and everyone should receive a CT of the c-spine based on mechanism.

Finding flaws with NEXUS - excellent, let's identify the subset at higher-risk so we can prevent missed injuries.  However, this article doesn't help us at all.  They don't do any sort of descriptive analysis of the NEXUS-negative patients who end up with significant injuries with which to educate our practice.  They simply conclude with the blanket statement that the dollar cost of performing all the CTs is less than the dollar cost of potential malpractice payouts.

In an era where we're trying to cut healthcare costs and reduce the practices of defensive medicine, this is precisely the sort of article that we don't need.  This is fantastic data presented in a non-constructive fashion that will likely, as the authors seem to intend, ensure the 97% of NEXUS-negative patients who had no injuries get their CT of the c-spine.

"National Emergency X-Radiography Utilization Study Criteria Is Inadequate to Rule Out Fracture After Significant Blunt Trauma Compared With Computed Tomography"

Saturday, January 7, 2012

Cardiology Corner - More Brugada Tidbits

Most physicians are aware of the Brugada Syndrome cardiac repolarization phenotype - the most recognizable being Type 1, or "coved" type.

Type 2 and Type 3, however, are essentially indistinguishable from an incomplete right bundle branch block with ST-segment elevation and a positive T-wave.  These authors, based on a small case series, took 38 patients referred for ajmaline provocation testing and compared their baseline ECGs.  Of the 14 patients who converted to Type 1 following ajmaline infusion, they found the baseline angle of the R' wave differed significantly - with an alpha angle cut-off of 50 degrees and a beta angle cut-off of 58 degrees.

A little esoteric, but fascinating.

"New Electrocardiographic Criteria for Discriminating Between Brugada Types 2 and 3 Patterns and Incomplete Right Bundle Branch Block"

Thursday, January 5, 2012

Who Knows If Older Platelets Are More Harmful

It might be true, but there's no way to know from this study - another illuminating example of just how difficult it is to perform trauma research.

Given that increased platelet transfusion in trauma has been linked to sepsis, ARDS, and other untoward outcomes, these authors decided to retrospectively evaluate whether the age of the platelet had any effect on sepsis, ARDS, ARF, liver failure, and mortality.  And, the answer - like I said, who knows?  The group that received four-day old platelets had the highest ISS - mostly attribute to head AIS >3 - in addition to an unlimited number of accounted for and unaccounted for confounding variables.

If you believe their adjustments, their proportional hazard regression model shakes out platelet and blood product age-related variables as significant associations with complications - most of which is sepsis.  So, while the authors are probably right, there are limitations.

"Impact of the Duration of Platelet Storage in Critically Ill Trauma Patients"

TYRAPS69S6HZ claim code (don't ask).

Tuesday, January 3, 2012

Too Many Traumatic Arrests Are Transported

Traumatic arrest in the field - except in the narrowest of circumstances - has universally dismal outcomes.  Yet, As the authors of this study observe, a great number of these patients continue to be transported to hospitals.

This is a retrospective review of a prospective trauma registry at Sinai in Chicago in which all traumatic patients with pre-hospital arrest were considered.  Patients were excluded for pediatrics, medical causes, drowning/electrocution injuries, and if the prehospital time was less than 15 minutes.  Essentially, they were looking at guidelines from the ACS Committee on Trauma for termination of resuscitation in the out of hospital setting - pulseless, apneic, no organized ECG activity, or unresponsive to 15 minutes of resuscitation.

They identified 428 patients in their cohort - and found that 294 of them were transported in violation of guidelines.  Of the inappropriately transported patients, 93% were declared dead in the ED and the remaining 6.8% (20 patients) survived the ED.  Of those 20, 12 died in surgery, 8 made it to the ICU, and 7 died.  A single, neurologically devastated, patient survived to discharge to a long-term care facility with a GCS of 6.

The total hospital charges incurred for the futile resuscitation of these patients totaled $3.8 million - a figure that excludes the EMS charges as well as the long-term care facility charges for the patient with GCS 6.

And this is just a single hospital.

"The Consequences of Noncompliance With Guidelines for Withholding or Terminating Resuscitation in Traumatic Cardiac Arrest Patients"

Monday, January 2, 2012

Must We Use IV Paracetamol/Acetaminophen?

I've yet to be terribly impressed with the "new" pain control options available to clinicians these days.  We've got tapentadol (Nucynta), which works just about as well as ibuprofen.  We've got companies working on a purified hydrocodone derivative that's 10 times stronger and equally more dependence forming.  And then we have intravenous paracetamol/acetaminophen.

So, it works.  Studies, like this one, show it's reasonably effective and has a minimal side effect profile - at least compared to the mild incidence of nausea seen with IV morphine.  It's slightly faster acting, achieves more reliable plasma levels than oral paracetamol/acetaminophen, and it's presumably as safe - although the safety of any intravenous drug is compromised due to extravasation risks and potential administration errors.  Oral paracetamol/acetaminophen bills a patient a few dollars while IV administration bills around a hundred, and I continue to wonder whether these sorts of "innovations" are worthwhile advances in pain control outside of extremely narrow indications.  I believe we now stock this and intravenous ibuprofen at our hospital - and goodness knows I've never seen anyone use them.  While relief in suffering is undoubtedly one of our most important roles in healthcare, we have to weigh the few moments of physical suffering against the long-term consequences/suffering of the hospital bills that may be passed along to our patients.

Anyone have a favorable experience with these new non-narcotic medications?

"Intravenous paracetamol versus morphine for renal colic in the emergency department: a randomised double-blind controlled trial"

Saturday, December 31, 2011

The Risks of Missing Dialysis

Hemodialysis patients have an elevated risk of death - and it's even higher for patients on scheduled dialysis during their "weekend."

Most scheduled plans are every other day Mon-Wed-Fri or Tue-Thu-Sat, which leads to a two-day interval between dialysis - resulting in an extra day of fluid retention and electrolyte abnormalities.  Bad hearts + extra fluid results in a much higher incidence of essentially any kind of mortality or morbidity associated with cardiovascular causes - significantly more myocardial infarction, congestive heart failure, stroke, and dysrhythmia.  Overall, there were 22.1 vs. 18.0 deaths per 100 person-years on the long-interval days than the others.  Retrospective registry data-mining, but it probably illuminates a logical truth.

This particular article caught my eye because we have a significant population at our county facility that comes for "compassionate dialysis".  Non-U.S. citizens that do not qualify for scheduled dialysis, they "live" a tortured existence in which they can only receive "emergency dialysis", as in, we routinely wait until they're at the precipice of death - with strict criteria of pulmonary edema, K+ > 6.0, bicarbonate less than 10, etc. - before pulling them back a small increment and sending them home to repeat the cycle in another week.  Barbaric.  I can't even imagine what their outcomes are like....

"Long Interdialytic Interval and Mortality among Patients Receiving Hemodialysis"

Thursday, December 29, 2011

Yet Another Highly Sensitive Troponin - In JAMA

...peddling the same tired phenomenon of magical thinking regarding the diagnostic miracle of highly sensitive troponins.  However, this one is different because it's been picked up by the AP, CBS News, Forbes, etc. saying: "Doctors are buzzing over a new blood test that might rule out a heart attack earlier than ever before" and other such insanity.  Yes, our hearts are in atrial flutter around the water cooler about a new assay that changes sensitivity from 79.4% to 82.3% at hour 0 and 94.0% to 98.2% at hour 3.

Unless you actually read the article.

Somehow, contrary to every other high-sensitivity troponin study, this particular highly-sensitive troponin had increased specificity as well - which simply doesn't make sense.  If you're testing for the presence of the exact same myocardial strain/necrosis byproduct as a conventional assay, it is absolutely inevitable that you will detect a greater number of >99th percentile values in situations not reflective of acute coronary syndrome.  The only way to increase both sensitivity and specificity is to measure something entirely different.

Or, if it suits your study aims, you can manipulate the outcomes on the back end.  In this study, the final diagnosis of ACS "was adjudicated by 2 independent cardiologists" whose diagnostic acumen is enhanced by financial support including Brahms AG, Abbott Diagnostics, St Jude Medical, Actavis, Terumo, AstraZeneca, Novartis, Sanofi-Aventis, Roche Diagnostics, and Siemens.

I am additionally not impressed by their results reporting - sensitivity and specificity, followed by the irrelevant positive predictive and negative predictive values.  Since the PPV and NPV are determined by the incidence of disease in their cohort, they're giving us numbers that are potentially not externally valid.  Rather, they should be reporting positive and negative likelihood or odds ratios - which are relatively cognitively unwieldy, but at least not misleading, but conceptually facile, like PPV and NPV.

And this is from JAMA.  Oi.

"Serial Changes in Highly Sensitive Troponin I Assay and Early Diagnosis of Myocardial Infarction"

Tuesday, December 27, 2011

How Frequently Is The Cath Lab Cancelled?

In North Carolina - a fair bit, actually.

This is a 14-hospital registry of cardiac catheterization activations for which the authors retrospectively evaluated how many were subsequently cancelled after activation.  They don't delve into a great deal of detail regarding specific findings that accounted for the cancellation - they simply observe the broad categories of cancellation.

Of all cath lab activations, it was judged that 15% were "inappropriate", with the gold standard being the consulting cardiologist opinion.  Of the cancellations, 40% were based on the EMS ECG, 31% were ED ECG, and the remainder were "not cath lab candidates".  The author's main focus in their conclusion is on the difference between EMS ECG cancellation and ED ECG cancellation due to ECG reinterpretation following activation.

What's more interesting from the paper, however, is when they break it down to the precise cohorts of activation and arrival - and note that 24.7% of EMS activations were subsequently judged inappropriate.  It is also interesting that 13% of non-PCI center activations were inappropriate vs 8% of PCI center activations.  Reading between the lines, there's probably some experiential component to the differences in activation rates, but this study doesn't specifically look at volume and training.

"Rates of Cardiac Catheterization Cancelation for ST Elevation Myocardial Infarction after Activation by Emergency Medical Services or Emergency Physicians: Results from the North Carolina Catheterization Laboratory Activation Registry (CLAR)"

Sunday, December 25, 2011

Happy Holidays!

Holiday break - intermittent and ineloquent blogging will be the norm.  I count 209 blog posts for the year - more than enough to keep anyone busy reading the archives.

But, if you're done with those, Life In The Fast Lane has a lovely Christmas-themed blog post with great articles including:

"What was wrong with Tiny Tim?"

"Children’s Nomenclatural Adventurism and Medical Evaluation study"

"No poinsettia this Christmas"