Category: Screening

Procalcitonin, Still Auditioning For a Role in Neonatal Sepsis

A single test to rule out bacteria infection would be a lovely invention.  But, in the absence of such, we’ll settle for a test to rule out serious bacterial infection.  But, alas, procalcitonin – despite its sponsored proponents – is not that test.

This is a systematic review and meta-analysis pooling 2,317 patients from seven studies evaluating its use in detection of SBI in a neonatal population of less than 91 days of life.  Most commonly, the discriminatory value reported used was 0.3 ng/mL, with five studies reporting this data and the other two providing this data upon request.  All told, infants with PCT >0.3 ng/mL had a 42.7% prevalence of SBI, while those with PCT below the cut-off had a 12.5% prevalence.

So, this works out to a relative risk of 3.97 (95% CI 3.41 to 4.62) given a PCT greater than the cut-off.  Unfortunately, a prior review of the “Rochester criteria” for infants aged 29 to 90 days noted this cohort of low-risk patients had a prevalence of SBI of only 2.7%, with a RR for SBI of 30.6 (95% CI 7.0-68.13).  Noting 2.7% to be superior to 12.5% as a rule-out mechanism, it would seem prudent to retain the Rochester criteria rather than rely on PCT.

It may be reasonable to incorporate PCT into future decision instruments for risk-stratification, but such validated rules are not yet available.

“Use of Serum Procalcitonin in Evaluation of Febrile Infants: A Meta-Analysis of 2,317 Patients”
http://www.ncbi.nlm.nih.gov/pubmed/25281186

Still Adrift in Ignorance Over Blood Cultures

While supervising residents, one of the frequent diagnostic suggestions in undifferentiated febrile patients is: blood cultures.  As an Emergency Physician, the utility of blood cultures – short of diagnosing endocarditis or another primary hematogenous source – is vanishingly small.  After all, the source of infection is nearly universally somewhere else – lung, urine, CSF, skin & soft tissue – and relying on the blood to give you the answer two days later is an unreliable and impractical proposition.

This study is yet another attempt at identifying patients with high likelihood of bacteremia, retrospectively analyzing 5,499 patients at Odense University Hospital for whom blood cultures were drawn.  This cohort, representing roughly half of all patients presenting to the Emergency Department, had positive blood culture results 7.6% of the time.  CRP, temperature, and SIRS criteria were evaluated as potential predictive variables – and, unfortunately, the positive likelihood ratios of each were only between 2 and 3, and the negative likelihood ratios associated with each were all 0.4.  The authors combine these criteria and promote their absence as a rule-out, with a negative predictive value of 99.5% – but, common sense ought obviate trying to diagnose bacteremia in an afebrile patient with no SIRS criteria, and the NPV performance is more related to the low prevalence of disease than the utility of their criteria.

Really, the most interesting element of this study: the massive volume of blood cultures performed, with 92% of them true negative or false positive.  Costs for blood cultures vary by facility, and range from $15-$50, with patient charges typically a significant multiplier beyond.  A low yield might be important if the diagnoses were changing management and improving outcomes, but the vast majority of culture results are clinically unimportant.  These authors have not described particularly strong positive predictors – but they’ve illustrated the massive scope of the problem.

“How do bacteraemic patients present to the emergency department and what is the diagnostic validity of the clinical parameters; temperature, C-reactive protein and systemic inflammatory response syndrome?”

Highly Sensitive Troponins – False Positive Bonanza

The “highly sensitive” troponin has received a great deal of publicity, hyped ad nauseum, see: “Simple test could help rule out heart attacks in the ER.”

But, as sensitivity increases – invariably, specificity decreases.  However, that is not the fault of the test – it is a failure of clinicians to ask the correct question of the test.  When asking “does this patient have an acute myocardial infarction?”(most commonly Type 1 MI in the ED), our training and education has been outpaced by assay technology – the test no longer provides a dichotomous “yes” or “no”.

This publication provides a lovely window into precisely the added value of the hsTnI compared with conventional TnI, both assays by Abbott Laboratories.  In this study, the authors simultaneously drew research samples of blood any time a cTnI was ordered.  The sample was frozen, and then analyzed at least 1 month following presentation.  Authors performed hospital records review, telephone follow-up, and vital records search to evaluate adverse events in patients with hsTnI or cTnI elevation.

Overall, they enrolled 808 patients, 40 of which received an adjucated diagnosis of “acute coronary syndrome” – 26 with AMI and 14 with unstable angina.  61 patients had acute heart failure, 7 had volume overload, 7 had pulmonary emboli, and 41 had other non-ACS cardiac diagnoses.

All told, there were 105 elevated cTnI samples – and 164 elevated hsTnI samples.  This means, essentially – in the acute setting, asking our question of interest – there were 50% greater false positives associated with hsTnI.  No patients would have been reclassified as nSTEMI based on the hsTnI result.  The authors sum this up nicely in their discussion:

“The preponderance of novel elevations (roughly 10% in this study) will be observed mainly in subjects with non-ACS conditions.”

The authors go on to note the value in detecting these novel or detectable troponin levels – essentially, non-ACS, subclinical disease – with a much poorer long-term prognosis.  This is almost certainly the case, although it will require further investigation to reliably demonstrate cost-effective management strategies based on these results.

“Troponin Elevations Only Detected With a High-sensitivity Assay: Clinical Correlations and Prognostic Significance”
http://www.ncbi.nlm.nih.gov/pubmed/25112512

Addendum:  As Stephen Smith points out, it may be possible to use the greater precision of hsTnI at the low end of the assay to more accurately adjudicate some MI.  Great insight!

Lives Saved … or Profiteering by Overdiagnosis?

Following an initial acute ischemic stroke, a search for the cause must be undertaken – for small vessel vasculitis, atherosclerotic emboli, thrombi from the systemic circulation, and so forth, beyond the domain of the Emergency Physician.  However, what the Emergency Physician does encounter is the sequelae of this search, in the form of oral anticoagulants.

These two articles from the New England Journal of Medicine, on their own, seem to reflect advances in diagnostic yield following acute ischemic stroke or transient ischemic attack.  The authors point out approximately 25% of patients suffering AIS and half of those suffering TIA never receive an ultimate identified etiology for stroke – and are classified as “cryptogenic”.  The authors in each of these studies suppose this may be due to the paroxysmal nature of atrial fibrillation, and that short-term electrocardiographic monitoring is missing this diagnosis.  In each study, some type of long-term monitoring technology is utilized, and, ultimately, the rate of diagnosis for paroxysmal atrial fibrillation jumps from 1-3% in each cohort to 8-12% in each cohort.

The catch – scads of authors for each report conflict-of-interest with both manufacturers of novel oral anticoagulants, or device manufacturers likely related to continuous ambulatory monitoring.  There is clear benefit to each of these parties, considering potential expanded indication for both monitoring and for anticoagulation.  These articles will likely be used to support both activities, despite not measuring any patient-oriented benefit.  How much of a primary or recurrent stroke risk is attributable to these very-infrequent paroxysms of atrial fibrillation?  Do they benefit equally from anticoagulation?

Given the conflict-of-interest enshrined in these articles, I am certain the advertised presumption will be they do.  They may, of course, be right – or, this may turn into yet another example of overdiagnosis and high-cost, low-yield medicine.

“Atrial Fibrillation in Patients with Cryptogenic Stroke”
http://www.nejm.org/doi/full/10.1056/NEJMoa1311376

“Cryptogenic Stroke and Underlying Atrial Fibrillation”
http://www.nejm.org/doi/full/10.1056/NEJMoa1313600

Intermediate Lactate Values, Lowering the Bar for Cryptic Shock

A guest post by Rory Spiegel (@CaptainBasilEM) who blogs on nihilism and the art of doing nothing at emnerd.com.

Serum lactate has been the darling of Emergency Medicine/Critical Care since Manny Rivers first introduced EGDT to the Emergency Department. Since then we have used it as a screening tool, a means to guide therapy and even to prognosticate outcomes. Despite our universal acceptance of its utility, very little high quality data has been published on its diagnostic properties. I reviewed this evidence in more depth in a past post and will limit this to the question, “Can serum lactate identify a group of patients who are in cryptic shock, despite clinically appearing well?” The Surviving Sepsis Campaign recommends using a lactate level of 4 mmol/L as the threshold for identifying cryptic shock, but lactate has a continuous curvilinear association with mortality and a 4 mmol/L threshold seems like an arbitrary cutoff.

In an attempt to answer this question Puskarich et al conducted a systematic review, published in the Journal of Critical Care, examining the ability of intermediate lactate values (2.0-3.9 mmol/L) to predict cryptic shock and death. Eight studies were included in this review. A total of 11,062 patients with intermediate lactate levels were examined. The authors appropriately decided that given the heterogeneity of these datasets, a formal meta-analysis was not appropriate. Instead they settled for descriptive statistics of each individual trial. In summary they found patients with intermediate lactate values who were normotensive had a 30 day mortality rate of 14.9% (mortality in individual trials ranged from 3.2-16.4%). Obviously the patients with intermediate lactate levels that were concurrently hypotensive fared far worse (30 day mortalities of 35-37%).

This review fails to define the clinical utility of the association between elevated lactate levels and risk of death.  In the few studies included in this review which published diagnostic test characteristics, lactate performed surprisingly poorly. Howell et al found lactate had an AOC of 0.71 for predicting 30 day mortality. Shapiro et al reported a similar AOC of 0.67. In fact in the Shapiro study when a cutoff of 2.5 mmol/L was used as screening tool for cryptic shock, it had a sensitivity of 59% and a specificity of 71%. Even a threshold of 4 mmol/L though very specific (92%) had a sensitivity of 36%, a far lower sensitivity than one would be traditionally accepted for a screening test.

More importantly, this data does not allow us to determine how a lactate threshold of 2.5 mmol/L  performs in the true cryptic shock patient. This is the patient who has end organ hypoperfusion without any clinically obvious signs. In most of the patients with elevated lactates, they appear clinically ill and thus the lactate is only confirming what we already know, that this patient needs aggressive intervention. If lactate is to prove useful as a true screening tool (at whatever threshold), it should be able to identify the patient clandestinely experiencing septic shock before any obvious signs of of end-organ damage (AMS, hypotension, AKI) become apparent. Unfortunately we have little data supporting its use in this manner. Even the secondary analysis of the Jones trial, finding similar mortalities between hypotensive patients and normotensive patients with elevated lactate (above 4mmol/L), fails to impress. Although the cryptic shock patients were not hypotensive in the strictest sense, they were by no means physiologically normal. On the contrary they were older, more tachycardic, with faster respiratory rates, and experienced significantly more intra-abdominal infections (30% vs 16%) than their hypotensive counterparts. And though they were not hypotensive (<90 mmHg), their blood pressures were not necessarily normal. The median blood pressure in the cryptic shock group was 108mmHG with an IQR of 92-126. To put it simply, these patients were sick. They did not require a lactate level to identify them as in need of aggressive therapy. There was nothing cryptic about them….

“Prognosis of Emergency Department Patients with Suspected Infection and Intermediate Lactate Levels: A Systematic Review”
http://www.jccjournal.org/article/S0883-9441(14)00002-1/abstract

How I (Hardly Ever) Scan For Pulmonary Embolism

There’s probably no diagnosis in the Emergency Department that confounds residents more than the practice variation between attendings regarding the evaluation for pulmonary embolism.  Some folks send d-Dimers with reckless abandon on patients with near-zero pretest probability, others make emotional decisions to “take PE off the table” when faced with no other explanation, and then there’s a group that only very rarely pursues the diagnosis.

I rarely pursue the diagnosis – mostly because the epidemiological evidence suggests we’re only harming folks by making additional diagnoses of pulmonary embolism.  Therefore, in a patient who is otherwise physiologically intact, a diagnosis of pulmonary embolism is more likely to result in iatrogenic bleeding risk rather than treatment benefit.  And, then, there’s the backwards fashion in which I use d-Dimer: I order it at the same time as the CTA in an otherwise intermediate- or high-risk patient, and then cancel the CTA if the d-Dimer is normal.

I use this strategy based on this prospectively collected data from the Kaiser system, published obscurely in The Permanente Journal several years back.  These authors evaluated 744 patients over 16 months who underwent CTA for rule-out PE, 347 of which had latex agglutination d-Dimer levels less than 1.0 µg/mL.  In this cohort of 347, there were seven positive scans – six of which were ultimately found to be false positives.  A handful of patients were lost, but the remainder had zero events in the three-month follow-up period.

So – d-Dimer negative, cancel the CTA, regardless of the pretest probability.  So far, so good!

“Computed Tomography Angiography in Patients Evaluated for Acute Pulmonary Embolism with Low Serum D-dimer Levels: A Prospective Study”
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2911823/

Negative Tests Fail to Reassure Patients

This article touches in a topic that we encounter all the time in Emergency Medicine – testing with the intent of “reassurance”.  The assumption is, wouldn’t a patient with symptom concerns be less anxious regarding their illness if they received a favorable negative test result?

That assumption, according to this meta-analysis and systematic review, is wrong.  These authors gathered together 14 trials evaluating the effect of non-diagnostic testing on downstream patient outcomes.  These tests included endoscopy for mild dyspepsia, radiography for low back pain, and cardiac event recording for palpitations.  This is a difficult article to interpret, particularly because there’s so much heterogeneity between the included studies, but the general conclusion is that tests performed in the setting of low pretest probability do not decrease subsequent primary care utilization, symptom recurrence, or anxiety regarding illness.

It’s rarely easy to tell a patient no testing is indicated – but this is yet another example illustrating the minimal benefits to over-testing.

“Reassurance After Diagnostic Testing With a Low Pretest Probability of Serious Disease”
http://www.ncbi.nlm.nih.gov/pubmed/23440131

EM Lit of Note on KevinMD.com

Featured today as a guest blog, revisiting the JAMA Clinical Evidence synopsis critiqued last month on this blog, here and here.

It’s rather an experiment in discovering just how influential social media has become – open access, crowdsourced “peer review” – and whether this mechanism for addressing conflict-of-interest in the prominent medical journals is more effective than simply attempting a letter to the editor.

KevinMD.com – “The filtering of medical evidence has clearly failed“

The Sad Reality of Chest Pain Observations

Chest pain observation units run by the Emergency Department are fairly popular – and it’s easy to see why.  It eliminates the need to fight a hospitalist for admission, allows for complete coverage of medicolegal liability, captures another set of billing codes for ED revenue, and keeps the cardiologists happy with a steady stream of interpretation and consultation revenue.

Duke University has one of these such chest pain observation units, and this study is a retrospective evaluation of the subgroup of patients aged less than 40 years.  Of the 2,231 patients observed for suspected acute coronary syndrome, 362 met eligibility based on age.  Of these 362 patients, median age 36, 238 underwent stress testing and the remainder underwent serial enzymes.

From this cohort, there was a single true positive – defined as a patient who underwent a coronary angiogram with an intervention performed.

There were, however, 14 false positives – indeterminate or positive stress tests and one set of positive biomarkers, leading to five negative invasive coronary angiograms.

The authors sum it up quite nicely:  “The extremely risk- adverse physician cannot totally exclude the possibility of ACS based on age, but it seems that routine observation for such patients may cause the potential for as much harm as good.”

“Utility of Observation Units For Young Emergency Department Chest Pain Patients”
www.ncbi.nlm.nih.gov/pubmed/22975283

The NICE Traffic Light Fails

Teasing out serious infection in children – while minimizing testing and unnecessary interventions – remains a challenge.  To this end, the National Institute for Health and Clinical Excellence in the United Kingdom created a “Traffic Light” clinical assessment tool.  This tool, which uses colour, activity, respiratory, hydration, and other features to give a low-, intermediate-, or high-risk assessment.

These authors attempted to validate the tool by retrospectively applying it to a prospective registry of over 15,000 febrile children aged less than 5 years.  The primary outcome was correctly classifying a serious bacterial infection as intermediate- or high-risk.  And the answer: 85.8% sensitivity and 28.5% specificity.  Meh.

108 of the 157 missed cases of SBI were urinary tract infections – for which the authors suggest perhaps urinalysis could be added to the NICE traffic light.  This would increase sensitivity to 92.1%, but drop specificity to 22.3% – if you agree with the blanket categorization of UTI as SBI.

Regardless, the AUC for SBI was 0.64 without the UA and 0.61 with the UA – not good at all.

“Accuracy of the “traffic light” clinical decision rule for serious bacterial infections in young children with fever: a retrospective cohort study”
www.ncbi.nlm.nih.gov/pubmed/23407730