U.S. Physicians are Awful at Prescribing Antibiotics

…and the Emergency Department is one of the worst offenders.

This is an analysis of the National Hospital Ambulatory Medical Care Survey, a representative sampling of ambulatory settings across the United States.  These authors simply reviewed all the antibiotic prescriptions and diagnosis codes for adult visits to offices, outpatient departments, and Emergency Departments.  10% of visits result in antibiotic prescriptions – and 61% of these prescriptions were broad-spectrum agents (amoxicillin/clavulanate, quinolones, etc.).  The largest category of antibiotic prescribing was for acute respiratory infections – and only 32% of those prescriptions were for diagnosis codes where antibiotics were typically indicated.  88% of respiratory diagnoses for which antibiotics were rarely indicated (e.g., bronchitis) received a broad-spectrum antibiotic.

This is retrospective, and the NHAMCS database has limitations – but this is farcical.  We’re passing out antibiotics without regard to the consequences – and we’re overusing broad-spectrum agents when narrow-spectrum agents are likely appropriate.  We’re far behind Europe in antibiotic stewardship, and the end result is certainly net population harm from over-treatment and induction of microbial resistance.

And, this doesn’t even account for pediatric visits – which are probably even worse.

Tragically, physician reimbursement is tied to patient satisfaction – or is an emphasized part of a healthcare business model in for-profit settings – and the evidence clearly indicates patients are more satisfied when they receive antibiotics.(pubmed, pubmed, archives of pediatrics)

Yet another example of perverted incentives degrading medical practice.

“Antibiotic prescribing for adults in ambulatory care in the USA, 2007 – 09”
www.ncbi.nlm.nih.gov/pubmed/23887867‎

CPR, Epinephrine … Vasopressin and Steroids?

Considering we’re still mighty skeptical regarding the ill effects of epinephrine on coronary and cerebral blood flow during resuscitation, I have to say I’m a little doubtful regarding the addition of a second vasopressor, along with steroids.

But, these authors, building on their prior work, attempt a randomized, placebo-controlled evaluation of epinephrine versus a combination of epinephrine, vasopressin, and methylprednisolone – along with a 7-day course of additional stress-dose steroids vs. placebo if post-ROSC hypotension was observed.  At hospital discharge, there were over twice as many neurologically intact survivors in the combination group as the epinephrine group – 18/130 vs. 7/138 – and thusly the authors conclude:

“Among patients with cardiac arrest requiring vasopressors, combined vasopressin-epinephrine and methylprednisolone during CPR and stress-dose hydrocortisone in postresuscitation shock, compared with epinephrine/saline placebo, resulted in improved survival to hospital discharge with favorable neurological status.”

Regrettably, with such a concisely worded conclusion, the authors devote barely two sentences to their limitations.  Indeed, for a study with so much to discuss, the authors compose a discussion section that occupies far less than even a full page.

There are a couple glaring problems with this study – not the least of which are the baseline differences between groups.  Despite randomization, the epinephrine group was saddled with quite different causes of cardiac arrest, almost certainly favoring the intervention group.  A randomization of additional patients with hypotension as their primary cause of arrest to the steroid group is almost certainly an allocation of a more favorable cohort, whereas “metabolic” causes of arrest are probably not corticosteroid deficient.  Similarly, the epinephrine group had far more asystole than the combination group – another poor prognostic feature.  Indeed, in their multivariate logistric regression (supplemental appendix), the cause of arrest and initial rhythm had statistically similar association with good outcome as intervention group membership.

The second issue is the problem of multiple interventions.  It is not clear whether the observed effect, if present, is secondary to the vasopressin-epinephrine-methylprednisolone cocktail during resuscitation or the stress-dose hydrocortisone given to nearly all survivors of the intervention group.  55% of the epinephrine group is alive 4 hours after ROSC vs. 66% of the intervention group, which, along with their physiologic data, implies the resuscitation intervention has some treatment effect.  Then, it’s unclear what favorable effect the stress-dose steroids has – particularly considering some of the epinephrine-only group then received open-label stress-dose hydrocortisone.  After resuscitation, different numbers of each group underwent PCI and similar numbers in each group received therapeutic hypothermia – but not all, leading to potential further confounding through selection bias.

Ultimately, it’s a mess – and it’s difficult to generalize these findings from a heterogenous and unbalanced cohort to routine practice.  The authors should be applauded for their ambitious goals, but a larger study, with a more effective randomization protocol, is yet needed.

“Vasopressin, Steroids, and Epinephrine and Neurologically Favorable Survival After In-Hospital Cardiac Arrest”
www.ncbi.nlm.nih.gov/pubmed/23860985‎

Letter in Stroke

My correspondence regarding this article was published in Stroke a couple days ago.  I’d like to say the author response is earth-shattering and insightful, but while it is not, they were kind enough to respectfully reply.

Sadly, the correspondence and response is available only to subscribers – but I have PDFs available for educational purposes….

“Letter by Radecki Regarding Article, ‘Safety of Thrombolysis in Stroke Mimics: Results From a Multicenter Cohort Study'”http://stroke.ahajournals.org/content/early/2013/08/08/STROKEAHA.113.002040.full.pdf+html

Observation of Minor TBI Prevents Harms

This study regarding the observation of children following minor traumatic brain injury is a little bit oddly spun by its authors and the medical news.

As we all know, most children presenting to the Emergency Department for minor head trauma do not have a clinically significant injury.  Regardless, a significant portion of these children receive non-therapeutic cranial radiation to further assure parents and clinicians alike.  The PECARN group, a few years back, published a rough decision instrument to help classify ~50% of these patients as “very low risk” (<0.05% risk of TBI) to give clinicians a tool to obviate CT scanning.

This group at Boston Children’s prospectively evaluated clinicians’ use of immediate CT scanning versus delayed CT scanning (observation).  They find, of course, that observing children in the ED for a short period, rather than making an immediate decision regarding CT use, resulted in decreased use of CT.  Thusly, the press releases state “Waiting and Watching Can Reduce Use of Brain Scans for Kids in the Emergency Department“.

But, watching and waiting doesn’t benefit the children in this cohort – other than preventing avoidable harms.  The eight children who had CT scans showing clinically important injuries were easily identified by clinicians as requiring immediate CT.  The period of observation doesn’t change the short-term clinical outcome of any of the patients – it only “treats” the risk-aversion of clinicians and parents.  “Watching and waiting” may reduce scans – but discharging the entire observation cohort immediately would have reduced scans even further, without missed cTBI (although the study is underpowered to truly detect all events down to an appropriate “zero-miss” threshold).

While I agree this is an important clinical problem to address, I simply find an odd discordance between the patient-oriented features and the resource utilization-oriented outcome measured.

“Effect of the Duration of Emergency Department Observation on Computed Tomography Use in Children With Minor Blunt Head Trauma”
www.ncbi.nlm.nih.gov/pubmed/23910481

More Sales Representatives, More Stents

In this breaking news update: sales representatives sell things!  Thusly, their company stays in business, and the employment of the sales representative continues.

This is a retrospective review of a Canadian hospital’s cardiac catheterization practices, evaluating the association between presence of sales representatives for stent manufacturers and use of each company’s stents during PCI.  Each day, during normal business hours, potentially a single sales representative from one of five stent manufacturers could be present in the lounge or in one of three cardiac catheterization laboratories.  Certain manufacturers specialized in bare metal stents, drug-eluting stents, or antibody-coated stents.

Unsurprisingly enough, cases performed in the presence of a sales representative resulted in increased use of that particular representative’s stents.  Additionally, for cases where DES were deployed, on average, more stents were placed during PCI when a drug representative was present.  Increased stenting, increased per-patient average cost.

It is a retrospective review, and there are baseline differences between the indications for catheterization – but, I think the observed association is probably real.  The authors also note, after these promotional visits were discontinued, all variation in stent use disappeared.

Further evidence of the suggestibility of physicians to marketing influences – supporting efforts to expunge them from our practice settings.

“The impact of industry representative’s visits on utilization of coronary stents”
www.ncbi.nlm.nih.gov/pubmed/23895808‎

Red Pill, Blue Pill, Video Pill

As these authors note, upper GI hemorrhage is responsible for almost 600,000 Emergency Department visits yearly – and there is some value and interest in risk-stratifying the suspect lesion with direct visualization.  Enter the gastroenterologist.

But, wait!  What if you could replace the on-call gastroenterologist and his endoscope with – a pill?  That was the question these researchers, funded by an unrestricted grant from the capsule endoscopy manufacturers, tried to address.

Sadly, their study design is woefully inadequate – except for producing positive findings to return the favor to their funding source.  A convenience sampling of 126 Emergency Physicians attending a conference watched four videos clipped only to footage of the stomach, three of which had blood present, and one of which did not.  These physicians missed a few (94% sensitivity) and overcalled a few more (87% specific) from these handpicked test videos.

So, we have a surrogate endpoint for patient-oriented outcomes, an idealized simulated setting that is non-equivalent to clinical practice, and conflicts of interest with the manufacturer.  The authors mention high “cost of capsule endoscopy” – and, at this point, I cannot see how this study does anything other than mislead readers this might be appropriate for an Emergency Department setting.

“Emergency Physicians Accurately Interpret Video Capsule Endoscopy Findings in Suspected Upper Gastrointestinal Hemorrhage: A Video Survey”
www.ncbi.nlm.nih.gov/pubmed/23859585

Where Is My: Coffee. Where is it.

Most modern vices seem to be, at the minimum, associated with some substantial harms.  Excessive sun exposure, rich western diets, scotch, sloth, etc.  And, then there’s coffee.

This is a review article covering the evidence behind various cardiovascular associations uncovered regarding the consumption of coffee.  After noting coffee contains thousands of compounds, the most prominent of which are caffeine, alcohols, antioxidants, and anti-inflammatories, the authors review the effects on various cardio-metabolic risk factors.

In brief, coffee consumption conferred:

  • No observed effect on blood pressure.
  • Decreased association with Type II diabetes.
  • Uncertain relationship with serum lipds.
  • A U-shaped relationship with congestive heart failure.
  • Decreased incidence of coronary heart disease.
  • Fewer cardiac arrhythmias.
  • Reduced risk of stroke.
  • Decreased risk of death.

Obviously, many of these findings are observational and potentially confounded by many other factors.  But, at the least – despair not of your coffee addiction.

“Effects of Habitual Coffee Consumption on Cardiometabolic Disease, Cardiovascular Health, and All-cause Mortality”
www.ncbi.nlm.nih.gov/pubmed/23871889

tPA Proponents Want to Have It Both Ways

I’m sure I sound like a bit of a broken record – yet again covering further attempts by the stroke neurology literature to continue expanding use of tPA for acute stroke beyond its initial, narrowly selected treatment population.  This observational, retrospective review from SITS-ISTR would like you to know:

“…this is the first observational study demonstrating that intravenous alteplase therapy within 4.5 to 6 hours of stroke onset for patients compliant with other EU approval criteria resulted in comparable rates of SICH, mortality, and functional independence to treatment initiated within 3 hours. This observation persisted in the multivariate analysis after adjustment for baseline imbalances.”

No difference in outcomes between 0-3h, 3-4.5h, and 4.5-6h – in both guideline-compliant and protocol-violation cohorts, and both adjusted and unadjusted results.  The authors, all receiving honoraria or grant support from Boehringer Ingelheim would like you to believe this analysis, also funded by Boehringer Ingelheim, supports a benefit for tPA up to six hours.

…but didn’t we just cover Jeff Saver’s JAMA article that “proved” an obvious time-to-treatment effect, favoring faster treatment?

Which of these observational, retrospective, registry reviews provides us the truth?  Is there a time-to-treatment effect that decays benefit with time, or, as this registry suggests, no difference between any time frames?

I suppose it depends on your specific professional conflicts-of-interest.

“Results of Intravenous Thrombolysis Within 4.5 to 6 Hours and Updated Results Within 3 to 4.5 Hours of Onset of Acute Ischemic Stroke Recorded in the Safe Implementation of Treatment in Stroke International Stroke Thrombolysis Register (SITS-ISTR)”
www.ncbi.nlm.nih.gov/pubmed/23689267‎

Half of What You Know is Wrong

…but we don’t know which half.  This highly entertaining study dredged the New England Journal of Medicine for the last decade, asking a simple question:  does new literature confirm or refute current practice?

They identified 1,344 articles concerning a specific medical practice.  Of these, 363 tested an established medical practice.  38% confirmed current practice, 40% rejected current practice, and 22% were inconclusive.  Examples of rejected medical practice included:

  • Primary rhythm control strategy in patients with atrial fibrillation.
  • The use of aprotinin during cardiac surgery.
  • Cyclooxygenase 2 inhibitors due to cardiovascular events.
  • Tight glycemic control vs. more permissive standards.
  • Benefit of stenting for patients with stable coronary disease.

…and so on.  It’s a fascinating list – and this is just one journal.

Of further interest, the majority of articles concerning specific medical practice concerned the development of a new medical drug or intervention, and most of these were positive.  I expect we shall see half of those similarly rejected by follow-up investigation in the next decade….

“A Decade of Reversal: An Analysis of 146 Contradicted Medical Practices”

Most Positive LPs Aren’t SAH

There’s still active debate regarding whether CTs are now sensitive enough to pick up all subarachnoid hemorrhage, and whether lumbar puncture is about to go the way of the dodo.  Their argument is based on the premise that, despite the imperfect gold standard of this practice changing study, very few negative CTs result in positive lumbar punctures.

However, as this chart review study shows, in real-world practice, most positive LPs may in fact be false positives.  This is a 10 year review at Barnes Jewish Hospital, where they were able to identify 57 patients with negative CT and positive LP, who subsequently underwent angiography for diagnostic confirmation.  Of these, three patients had positive findings – only two of which were ultimately determined to be true positives.

This is a little different than previous reviews, in which 53% of negative CT/positive LP represented true positives.  These authors suggest the previous higher-yield results are likely the result of dependence on xanthrochromia as part of the diagnostic evaluation, and a cohort with a longer duration of symptoms prior to LP.

The truth is likely somewhere in the middle, splitting the difference between useless and half-wrong.  This probably aids the “CT, no LP” camp – of which I tend towards – by demonstrating the preponderance of false positives inherent to LP for SAH.

“Yield of Cather Angiography After Computed Tomography Negative, Lumbar Puncture Positive Subarachnoid Hemorrhage”
www.ncbi.nlm.nih.gov/pubmed/23619131‎