Simple SBI Prediction – Hopeless

It remains a noble endeavour to attempt to identify the risk of serious bacteria infections in children.  That said, many have tried, and many have failed.

These authors from the Netherlands and the United Kingdom try, yet again.  They note the best performing decision instrument incorporates 26 variables – which they feel is unworkably unwieldy in a clinical setting – and attempt to derive their own, tighter instrument.  Unfortunately, the clinical variables that shake out of their prediction methodology all have odds ratios less than 6 – leading to a prediction model that can be calibrated only either for horrible sensitivity or horrible specificity.  The sensitive model will lead to over-testing of an otherwise well population, and the specific model will essentially pick up only the cases that were clinically obvious.

It's becoming pretty clear over the years that attempting to reduce the number of discrete clinical variables in the febrile SBI decision-instrument is a dead-end strategy.  Complex clinical problems simply defy dimension reduction.  Furthermore, the true test of a decision instrument also ought not just be statistical evaluation in a vacuum, but comparison with clinical judgement.


"Clinical prediction model to aid emergency doctors managing febrile children at risk of serious bacterial infections: diagnostic study"
www.bmj.com/content/346/bmj.f1706

Fewer CTs, More Ultrasounds in Children

It's pretty clear that children shouldn't be receiving CT scans, whenever possible.  Despite this, the rate of CT for the diagnosis of pediatric appendicitis continues to rise.

This is a retrospective review from the Medical University of South Carolina that describes their implementation of an imaging protocol designed to encourage ultrasound use.  They report before-and-after statistics for their protocol – and, unsurprisingly, they're pro-ultrasound.

Their protocol is generally simple – if it's clinical appendicitis, consult surgery.  If it's uncertain, do ultrasound first – if ultrasound equivocal, do CT.  If the patient appears unwell, skip ultrasound and do CT to evaluate for perforation.  Their institution started out with 82% of patients undergoing appendectomy having received CT, with this percentage dropping to 20% following implementation of the protocol.  Their negative appendectomy rate was stable at 5% after implementation, as well.  They also note the cost of a pediatric CT is $6500 compared with ultrasound at $1100.

The main disadvantage of their protocol was the low sensitivity exhibited by ultrasound – 61% – and the length of stay resulting from patients who required both ultrasound and CT – nearly 8 hours.  Considering ultrasound sensitivity depends on the experience of the operator, efforts to implement this strategy might benefit from upgrading local ultrasound capabilities.

"Clinical Practice Guidelines for Pediatric Appendicitis Evaluation Can Decrease Computed Tomography Utilization While Maintaining Diagnostic Accuracy"
www.ncbi.nlm.nih.gov/pubmed/23611916‎

Prophylactic Platelet Transfusions Needed?

Every so often, we're referred a patient from the hematology clinic for prophylactic platelet transfusion when their infusion center is full.  As these authors state, it is common practice to routinely prophylactically transfuse platelets in the presence of severe thrombocytopenia.

And, the abstract conclusion here is:  "The results of our study support the need for the continued use of prophylaxis with platelet transfusion and show the benefit of such prophylaxis for reducing bleeding, as compared with no prophylaxis."

This conclusion is the dubious interpretation of a combined endpoint of uncertain clinical significance.  Their primary endpoint was the summation of WHO 2, 3 and 4 bleeding.  Nearly all their bleeding was WHO 2 – essentially, troublesome, but self-limited bleeding not requiring transfusion.  With regards to major bleeding, the per-protocol rates were 5/258 in the no-prophylaxis group and 1/230 in the prophylaxis group – falling short of statistical significance, but reasonably clinically significant.

There was a 0.2 unit PRBCs per patient difference in transfusion between the prophylaxis group and the no prophylaxis group.  There was, obviously, a 1.3 unit platelets per patient difference between groups.  Hospitalization days were similar.  No patient died from bleeding.  One patient in the prophylaxis group suffered a transfusion-related anaphylactic event.

So, yes – platelet transfusions reduce bleeding.  I'm just not certain these authors have demonstrated the net benefit – given 66 patients receiving prophylactic transfusions, 1 fewer non-fatal important bleeding event.  Clinically insignificant net PRBC difference, significantly higher usage of platelet transfusions.  I think there is plenty of room to debate cost and benefit – particularly considering interesting hypothesis-generating subgroup variation.

The authors declared multiple conflicts-of-interest, although it's not clear to me if any of them would be relevant to the current study.

"A No-Prophylaxis Platelet-Transfusion Strategy for Hematologic Cancers"
http://www.nejm.org/doi/full/10.1056/NEJMoa1212772

Here Comes Copeptin

Are you interested in making your AMI diagnostic evaluation even less specific?  Good!  Because Brahms Thermo Fisher et al want to sell you a rapid copeptin assay to help with that.

Copeptin is a stable, terminal portion of the arginine vasopressin peptide.  This peptide is released from the pituitary in response to cardiovascular hemodynamic stress and has a theoretical role in the diagnosis of acute myocardial infarction.  The advantage copeptin may have over conventional troponin assays is detectable release in circulation preceding troponin.

In 1971 patients collected through their multi-center trial, 156 were diagnosed with AMI (7.9%).  Upon presentation to the Emergency Department, 40 were STEMI.  281 patients had cTnI greater than 40ng/L, 97 of whom were subsequently diagnosed with nSTEMI.  1646 had cTnI less than 40ng/L, 19 of whom eventually were diagnosed with nSTEMI – a miss rate of 1.1%.  Pretty good – but, obviously, we practice in a zero-miss world.  Adding copeptin to this troponin-negative population with a cut-off of 14pmol/L decreased the miss rate to 0.5%.  The specificity, of course, was useless – only 10 of 493 patients with positive initial copeptin and negative inital troponin went on to receive a diagnosis of nSTEMI.

So, the question is – would a negative copeptin change your practice?  Is there a clinically important difference between a 1 in 100 miss rate vs. a 1 in 200 miss rate?  These authors think adding copeptin to troponin will allow you to discharge a patient after the inital biomarker result – but I think this minimal incremental improvement in diagnostic performance doesn't change whatever pathway the patient was already on, nor add much to a discussion of shared decision-making with the patient.  They also don't address a performance advantage compared to high-sensitivity troponin assays (which have, of course, their own issues).

These authors are pretty high on copeptin – but, then again, many of them are employed by or sponsored by the manufacturers of the copeptin assay.

"Copeptin Helps in the Early Detection Of Patients with Acute Myocardial Infarction: the primary results of the CHOPIN Trial"
http://www.ncbi.nlm.nih.gov/pubmed/23643595

Proto Magazine Letter

My recently-published short invited response in Proto Magazine, a Massachusetts General Hospital publication, to an article on the state of current medical journals:  "Probing Deeper"

They did, however, unexpectedly edit out a portion of my response – an entire paragraph originally between the current 2nd and 3rd paragraphs:

In 2005-2006, The Lancet derived 41% of its revenue through sales of over 11 million reprints.[1]  The NEJM, which published more industry-funded studies thanThe Lancet – 78% vs. 58% – undoubtedly derives even more.[2]  Ironically, Jeffery Drazen, editor-in-chief of NEJM, is quoted as saying "Our most important job is vetting information.”  Dr. Drazen infamously failed to do so when privy to information regarding increased mortality in rofecoxib's (Vioxx) VIGOR trial – a publication for which NEJM sold Merck 900,000 reprints.[3][4]

And, here are my references:

1.  Dorsey ER, George BP, Dayoub, EJ, Ravina BM.  Finances of the publishers of the most highly cited US medical journals.  J Med Libr Assoc. 2011 Jul;99(3):255-8.

2. Lundh A, Barbateskovic M, Hróbjartsson A, Gøtzsche PC.  Conflicts of interest at medical journals: the influence of industry-supported randomised trials on journal impact factors and revenue – cohort study.  PLoS Med. 2010 Oct;7(10):e1000354.

3.  Armstrong D.   Bitter pill: how the New England Journal missed warning signs on Vioxx.  Wall Street Journal 2006 May 15:A1.

4.  Smith R.  Lapses at the New England Journal of Medicine.  J R Soc Med. 2006 Aug;99(8):380-2. 

Questioning "Atypical Angina"

The prevailing notion has been that women present with symptoms of angina that are "atypical" from men – headaches, jaw pain, generalized malaise – rather than definitive anginal-type chest pain or pressure.  These authors would like to suggest this global characterization is incorrect.

These authors enrolled a sample of 128 men and 109 women who underwent coronary angiogram following an abnormal stress test.  Patients with obstructive coronary artery disease on their angiogram were surveyed regarding the symptoms that prompted them to seek care.  Of this cohort, 89 men had obstructive disease compared with 50 women.

Overall, there was no significant statistical difference in the rate of most descriptors used by men or women.  Surprisingly, women were statistically more likely to use "typical" terms such as "discomfort", "crushing", "pressing" and "aching" to describe their chest pain.  Therefore, these authors conclude the clinical construct of "atypical angina" in women is incorrect.

I would tend to agree – excepting their study suffers from selection bias.  If patients are only referred for testing due to suspected coronary artery disease, then the population with "atypical" symptoms might not be fully captured.  That being said, it does look as though the female population in their study encompassed a number of patients who potentially were referred for atypical symptoms, considering the yield of their coronary angiography was much lower in women.  It would have been interesting to compare the referral symptoms to the subset with demonstrated obstructive CAD.

"Reconstructing Angina: Cardiac Symptoms Are the Same in Women and Men"
www.ncbi.nlm.nih.gov/pubmed/23567974‎

Falling Short on Pneumonia Prediction

These authors address a real problem: which coughing adults have pneumonia?  Unfortunately, after evaluating 2,820 of them – they still don't really know.

This is an interesting article because it pulls together a symptom profile along with two of the other non-specific inflammatory markers being touted as important diagnostic tools: CRP and procalcitonin.  Primary care physicians enrolled adults presenting with acute cough, and used plain radiography as their gold standard for diagnosis of pneumonia.

In short:

• "Symptoms and signs" suggestive of pneumonia (fever, tachycardia, abnormal lung exam) all had positive OR between 2.0 and 5.3, and combined offered an AUC of 0.70.
• Adding CRP as a continuous variable to symptoms and signs gave an OR of 1.2 and increased the AUC to 0.78.
• Adding procalcitonin as a continuous variable to symptoms and signs gave an OR of 1.1 and increased the AUC to 0.72.

Using CRP as a dichotomous cut-off at 30 mg/L, in addition to the independent symptom predictors, gave them the discriminating ability to produce a low, intermediate and high risk group: 0.7%, 3.8%, and 18.2% chance of pneumonia.  A high-risk group where fewer than one in five have the disease?  The authors recommend consideration of empiric antibiotic therapy in this group, but I prefer their other recommendation to consider radiography as confirmation in this subset.  The remainder ought to be candidates for observation, as false positives and harms from additional testing are likely to outweigh true positives.

Again, refuting the terrible JAMA distortion, procalcitonin had no useful discriminatory diagnostic value.

"Use of serum C reactive protein and procalcitonin concentrations in addition to symptoms and signs to predict pneumonia in patients presenting to primary care with acute cough: diagnostic study"

http://www.bmj.com/content/346/bmj.f2450

Azithromycin & Cardiovascular Risk, Belabored

Last year, I noted a study concerning a report of excess deaths associated with azithromycin use.  This study, a retrospective, observational cohort from Tennessee Medicaid data suggested a death rate double that of other antibiotics.  This led to the FDA issuing a warning regarding azithromycin use.

I thought all this fuss was absurd – the data quality was one step above junk and the absolute magnitude of the proposed harms was trivial.

Now, we have the counterpoint – a retrospective, observational cohort from Denmark, using their national health system database to compare prescriptions for azithromycin to penicillin V over the last 13 years.  In their cohort, there's an obvious increase in risk of death from cardiovascular causes simply from being prescribed any antibiotics – but no difference between azithromycin and penicillin V.  This seems to indicate either the systemic infectious process contributes to excess cardiovascular risk, or that respiratory symptoms are being misdiagnosed as infectious rather than cardiovascular.  The absolute effect in their propensity matched cohorts is also tiny – a handful of patients or fewer spread across a million prescription events.

The accompanying opinion seems to attempt to justify the FDA review based on the wide confidence intervals in the Danish study – the OR for death from cardiovascular causes vs. penicillin V is 1.06 (0.54 – 2.10) and doesn't statistically contradict the Tennessee study.  However, yet again, I would point to the reason behind the wide confidence intervals – the nearly trivial absolute magnitude of the harms, which amount to fractions of a patient per 1000 patient-years.

Again, plenty of reasons to responsibly reduce azithromycin prescriptions – but this cardiovascular hullabaloo probably isn't one of them.

"Use of Azithromycin and Death from Cardiovascular Causes"
http://www.nejm.org/doi/full/10.1056/NEJMoa1300799

Levamisole-Induced Vasculitis

Let the good times roll – unless, of course, those good times are cut with levamisole.

This short case report from my good friends across the street at Baylor showcases a couple lovely pictures of the purpuric and necrotizing skin lesions associated with the anti-helminth levamisole – which, for some reason, is an increasingly popular additive to cocaine.  They are quite distressing and usually present following several courses of failed outpatient antibiotics.

I don't think I specifically ever saw the exact patient these authors report upon, but it's not a unique presentation in our county healthcare system.  This article is open-access for all to view without an institutional subscription.

"Levamisole-adulterated Cocaine Induced Vasculitis with Skin Ulcerations"
http://www.escholarship.org/uc/item/4rd630zt

Bonus link nomination for best article title ever:

"Riding the Escalator: How Dangerous is it Really?"

Mixed "Cost-Conscious" Ordering Results

It's a little bit of a messy study, sadly, because it's probably a lovely idea.

These authors performed a before-and-after interventional trial in which they measured laboratory test ordering rates.  After a six-month baseline phase, the intervention phase consisted of displaying the 2008 Medicare allowable charge for a subset of frequent lab tests.  The theory, of course, is that displaying price information in the context of test ordering will alter physician behavior.

Most of the orders were placed on internal medicine services – and yes, there was a decrease in the number of orders with cost information displayed.  At the same time, however, the tests without cost information increased.  The net result, overall, was a decrease in total testing.  Interestingly, the impact seemed to mostly include a reduction by replacing CMP orders by BMPs.  $3.79 per patient-day costs were reduced during the intervention period.

So, the impact was mixed – slightly expensive tests were replaced by slightly less expensive tests.  More evaluation is necessary to determine whether these reductions have unanticipated impact on patient outcomes.

"Impact of Providing Fee Data on Laboratory Test Ordering"
www.ncbi.nlm.nih.gov/pubmed/23588900