Is the 6-Hour CT for SAH Debate Over?

There has been a fair bit of back and forth about the validity of early CT in the setting of “thunderclap headache” to obviate a lumbar puncture in the search for aneurysmal subarachnoid hemorrhage.  David Newman and Kevin Klauer debated this subject a few years ago – and that was in all-comers, not simply those in the few hours following onset.

This most recent meta-analysis and systematic review gathers together all the published literature regarding early CT and the incidence of SAH on follow-up.  Including 8,907 patients from five publications, based on a few assumptions from retrospective studies, there were up to 13 missed cases of aneurysmal SAH occurring despite a negative CT within 6 hours of onset.  Worst-case sensitivity based on these data, then, was 0.987 (95% CI 0.971-0.994).

The prevalence of SAH in patients presenting with true thunderclap headache is estimated at ~10%.  The post-test odds, then, after a negative CT, are on the order of 0.1% – in line with David Newman’s posit of requiring 1000 LPs to catch one missed SAH.  The problem, then, lies in taking the next step in Bayesian reasoning – how likely is the positive LP to be true SAH?  If prevalence has dropped to 1 in 1000 after a negative CT, and the specificity of LP for SAH is only 65%, even a positive result barely budges the likelihood of disease.

How do you consent a patient for an invasive procedure in a setting in which a positive result has only the tiniest fraction of a chance of being real – and the treatments based on findings of follow-up examinations may be more likely to harm the patient than the magnitude of benefit associated with detection of a true positive?

“Sensitivity of Early Brain Computed Tomography to Exclude Aneurysmal Subarachnoid Hemorrhage: A Systematic Review and Meta-Analysis”

CTCA in the ED, Getting Less Sexy By The Day

As I’ve written before, the CT coronary angiograms is a funny test.  The idea of having a non-invasive method of detecting previously unknown coronary artery disease is compelling.  The practical application, however, has been limited by a low specificity – further exacerbated by those encouraging its use in a population with low pretest probability.

However, the few major studies regarding it tend to view CTCA in a favorable light – the result of comparing CTCA-based strategies to modern over-triage and over-testing of potential acute coronary syndrome.  These studies, ACRIN-PA and ROMICAT, showed significant improvements in direct discharge from the ED and in length-of-stay, not so much due to being a superior strategy of benefit to patients, but by obviating unnecessary care inflicted upon them.

The general gist of this trial is framed in the “era of high-sensitivity troponins” – referring to new developments in assays allowing a safer rapid rule-out in the Emergency Department.  This trial, as opposed to the others, also occurs in the Netherlands, a setting in which direct discharge from the ED is no anathema.  The “standard care” arms of ACRIN-PA and ROMICAT-2 had discharge rates from the ED of ~20% or less, while this trial discharges nearly 60%.  Yet, despite such recklessness displayed in this trial, these patients are ultimately just as safe.  And, when such an insanity-reduction initiative is undertaken, the advantages of CTCA diminish.

And, frankly, nearly all low-risk patients can be discharged safely from the Emergency Department.  The appropriate urgent follow-up test, if any, is a trickier proposition – and CCTA may yet be appropriate for some.  However, as a routine, ED-based strategy, it should probably be considered low-value care.

“Coronary CT Angiography for Suspected ACS in the Era of High-Sensitivity Troponins”

Give More tPA, Pretty Please?

There’s been another scientific update from Genentech-by-proxy, this time lamenting the low utilization for tPA in ischemic stroke patients.  This guideline panel from the AHA notes administering tPA in a safe and timely fashion to stroke patients is a non-trivial organizational exercise, but, what really gets their goat are the various exclusion criteria.  These criteria – minor symptoms, elevated blood pressure, acute trauma, etc. – are aimed at reducing the overall harms of tPA use, but to “help” as many patients as possible, the stroke neurologists believe it is necessary to break all the eggs.

This document, most prominently, is an entertaining exercise in linguistic calisthenics.  Clearly, these authors would like to treat as many patients as possible with tPA.  In their pursuit of these justifications, absent evidence, they torture the English language into providing he most diverse possible assortment of non-committal positivity.  To wit:

  • “There should be no exclusion …”
  • “Intravenous alteplase treatment is reasonable …”
  • “Intravenous alteplase administration for ischemic stroke may be considered …”
  • “…  it is reasonable that urgent intravenous alteplase treatment not be delayed …”
  • “Intravenous alteplase may be considered on a case-by-case basis.”
  • “Intravenous alteplase may be reasonable …”
  • “Use of intravenous alteplase in carefully selected patients … may be considered …”
  • “In acute ischemic stroke patients … intravenous alteplase may be carefully considered …”
  • “… administration of intravenous alteplase is reasonable and probably recommended …
  • “Intravenous alteplase treatment is probably recommended …”
  • “Patients … may benefit from intravenous alteplase …”
  • “… intravenous alteplase treatment appears safe and may be beneficial …”
  • “… intravenous alteplase may be as effective … and may be a reasonable option …”
  • “Intravenous alteplase is probably indicated …”

I.e., this scientific update may be a reasonably probably carefully considered option.  Or, as I’ve heard Jerry Hoffman say, each instance of the word “may” should be paired with its logical partner – “may not” – for the appropriately even-handed reading.

Other gems:

The risk of symptomatic intracranial hemorrhage in the SM population is quite low; thus, starting intravenous alteplase is probably recommended in pref- erence over delaying treatment to pursue additional diagnostic studies (Class IIa; Level of Evidence B).

The authors devote half a page to overstating the safety margin of tPA in stroke mimics by focusing on a single 100-patient cohort.  The clinical anecdotes of the two patient suffering sICH, and being lucky enough to survive, in this cohort are provided as apparently definitive reassurance.


For patients with mild but disabling stroke symptoms, intravenous alteplase is indicated within 3 hours from symptom onset of ischemic stroke. There should be no exclusion for patients with mild but nonetheless disabling stroke symptoms in the opinion of the treating physician from treatment with intravenous alteplase because there is proven clinical benefit for those patients (Class I; Level of Evidence A).

The pursuit of treating mild, but nonetheless disability symptoms is not new – and not specifically offensive.  However, they give this recommendation “Class I, Level of Evidence A”, which is the strongest level of support, based on, apparently, multiple randomized clinical trials or meta-analyses.  Except, however, their justification in the text for this recommendation is merely:

Alteplase may be beneficial for milder stroke cases judged as potentially disabling despite low NIHSS scores. The NINDS trialists explored 5 different definitions of minor stroke in a post hoc analysis and found benefit for alteplase across all definitions. However, data are not available on the effect of alteplase for milder stroke cases judged as not potentially disabling at presentation. Because nearly 3000 such cases of ischemic stroke were excluded from the 2 NINDS trials for mild symptoms, any analysis of mild symptoms within the 2 NINDS trials is difficult to interpret.

Why let the true level of evidence affect the final recommendation categorization?

If left to my own devices, this post could meander onward for an eternity.  I will, then, now step aside to allow the motivated reader to move along to source document itself.

“Scientific Rationale for the Inclusion and Exclusion Criteria for Intravenous Alteplase in Acute Ischemic Stroke”

Still Making Sense of High-Sensitivity Troponin

Figuring out the value of a troponin measurement is both simple, and complicated.  Big numbers are still straightforward and trouble.  Small numbers – and even smaller still – are increasingly difficult to parse.

What does seem to be true, at least: the closer to zero, the better.

This is a multi-center evaluation of a 0- and -1hour troponin delta strategy, based on a hs-cTnT from Roche.  These authors prospectively enrolled 1,458 patients with chest pain of peak intensity within 6 hours.  Samples were then taken at 0, 1, 2, and then at least 4 hours after presentation.  The outcome of interest was an independently adjudicated final diagnosis of acute myocardial infarction.  And, this study probably shows just about what every similar study has shown: initial sensitivity is ~90%, with marginal increases as additional time points are added, while specificity suffers predictably due to the nature of the assay.

However, there are a couple concerning potential sources of bias.  The authors enrolled 1,458 patients – but exclude 176 of them from analysis for a variety of reasons.  In a study trying to catch rare events to demonstrate near-100% sensitivity, over 10% of patients dropping out is an important consideration.  There were also issues with slow enrollment, compared to previous studies, and the patient flow diagram is extremely sparse.  Over two years, the centers involved likely had many thousands of chest pain presentations.  No information regarding the missed enrollments is presented.

There are also issues with the adjudication downstream, which was based on the results of various follow-up examinations as well as, oddly enough, a different troponin assay: s-cTnI-ultra.  213 (17%) patients received a final diagnosis of AMI, while 167 (13%) received a diagnosis of unstable angina.  The clinical significance of their definition of unstable angina remains unclear to me – myocardial ischemia without cellular injury associated with chest pain at rest.  The authors reference these UA patients as being at low risk for poor long term outcomes, which seems clinically discontinuous with the sort of “critical near-occlusion” working definition I’m familiar with for true UA.  Regardless, the safety of their strategy is only reasonable if UA is a relatively benign catch-all diagnosis for troponin-negative chest pain, so I will accept their categorization.

There were also diverse and perverse conflicts-of-interest described with the manufacturer of the assay involved.

Regardless, as previously stated, these data are consistent with prior demonstrations – so, yes, using these assays at presentation, or as 1- or 2-hour deltas in the Emergency Department, will result in a very low miss rate when paired with low pretest likelihoods of disease.  Furthermore, anything missed by these assays will be such a minute injury pattern as to be extremely low-risk for short term cardiac mortality.

Yes, Virginia, you can discharge chest pain.

“Multicenter Evaluation of a 0-Hour/1-Hour Algorithm in the Diagnosis of Myocardial Infarction With High-Sensitivity Cardiac Troponin T”

Again With No Antibiotics, This Time for UTI

Frequent readers of this blog may have noticed a bit of an anti-antibiotic tendency.  Diverticulitis!  Strep throat!  All manner of upper respiratory symptoms!

How about urinary tract infections?

This German study randomized ambulatory women with urinary tract infection symptoms and positive findings on urine dipstick to either fosfomycin plus placebo tablets for three days, or simply ibuprofen for three days.  Patients were then reassessed after three days, and those with treatment failure were provided an additional course of antibiotics.  “Co-primary” endpoints were antibiotics utilization and the AUC of sums of daily symptom scores.

The results are, like last week’s URI trial, a little mixed.  The authors included 484 patients in their intention-to-treat analysis, and 77% of them ultimately had culture-positive UTIs.  A lot – 69% – of patients randomized to ibuprofen had spontaneous resolution of their symptoms and avoided antibiotic use for their UTI.  However, obviously, those who did not improve spontaneously, and ultimately were given antibiotics, did worse than their fosfomycin counterparts – and the symptom scores clearly favored the antibiotic cohort.  Furthermore, 5 of 241 of patients randomized to ibuprofen advanced to pyelonephritis, and one more patient suffered ulcer-related bleeding due to ibuprofen.

I’m not sure how many women would opt for the trial of ibuprofen as part of a shared decision-making conversation, were practice to be based on this specific trial.  That said, it does raise a bit of an interesting question regarding potential strategies to reduce antibiotic use.  Would a 24- or 48-hour “waiting period” help?  If routine urine cultures weren’t already grossly low-value care, could waiting for those results help triage appropriate use of antibiotics?  Could a different symptom adjunct, such as pyridium, help reduce the difference in symptom scores while awaiting spontaneous resolution?

Regardless, it is is yet again an insight into the general effectiveness of the human body’s natural antibacterial defense mechanisms.  How much of modern medicine is critically important – and how much is simply are mildly harmful minor ameliorations of mostly self-limited disease processes?

“Ibuprofen versus fosfomycin for uncomplicated urinary tract infection in women: randomised controlled trial”

The “Don’t Anticoagulate Pulmonary Embolism” Guideline

We’ve expressed a lot of angst around these parts regarding the homogenous treatment of acute pulmonary embolism.  With the advent of angiography, and then computed tomography with ever-increasing levels of sensitivity, the physiologic spectrum of disease for pulmonary embolism has massively increased.

In olden times, pulmonary embolism actually presented as pulmonary infarction with the “classic” triad of pleuritic pain, hemoptysis, and signs of deep venous thrombosis.  Now, nearly 10% are virtually incidental subsegmental PE of uncertain clinical significance – yet, recommendations for treatment remained systemic anticoagulation.

Until now.

This new guideline states patients with subsegmental PE, without another identifiable VTE source, and at low risk for recurrent VTE, have the option of watchful waiting.  They cite no new groundbreaking evidence, but generally recognize the low rates of recurrent VTE in retrospective and observational studies.  They also recognize a diagnosis of subsegmental PE is quite likely to be a false-positive, as covered in my last ACEPNow column, unless the following conditions are met:

We suggest that a diagnosis of subsegmental PE is more likely to be correct (i.e. a true-positive) if: (1) the CT pulmonary angiogram (CTPA) is of high quality with good opacification of the distal pulmonary arteries; (2) there are multiple intraluminal defects; (3) defects involve more proximal sub-segmental arteries (i.e. are larger); (4) defects are seen on more than one image; (5) defects are surrounded by contrast rather than appearing to be adherent to the pulmonary artery; (6) defects are seen on more than one projection; (7) patients are symptomatic, as opposed to PE being an incidental finding; (8) there is a high clinical pre-test probability for PE; and D-Dimer level is elevated, particularly if the increase is marked and otherwise unexplained.

Patients discharged without anticoagulation should be provided prospective guidance on seeking care for new or progressive symptoms.  These recommendations are appropriately GRADE category 2C, reflecting moderate/weak certainty and a low level of evidence – but, it at least provides a framework to have a reasonable conversation and shared decision-making with a patient.

It also appropriate raises a question about testing for PE:  if a patient has a PE meeting criteria for non-treatment, does it need to be found in the first place?  Should the acceptable miss rate for PE – assuming the quoted prevalence of subsegmental disease – be ~10%?

“Antithrombotic Therapy for VTE Disease: CHEST Guideline”

Radiation Therapy for the Common Cold

We overuse computed tomography for many things – popular topics in the literature (and on this blog) are mostly minor head injury, renal colic, and CT pulmonary angiograms.  But, it is not simply these modalities that have increased in the preceding decades, as this research letter shows.  CT use has also increased for such apparently benign conditions as non-acute upper respiratory symptoms.

This is a National Hospital Ambulatory Medical Care Survey analysis, with all the sampling limitations inherent to such a data source, evaluating CT usage for upper and lower respiratory complaints between 2001 and 2010.  CT usage in 2001 for such complaints ranged from a 0.5% of visits for non-acute URI, to 3.1% of visits for acute LRTI symptoms.  In 2010, such usage ranged from 3.6% to 12.1%.  Despite all this added cost and extensive evaluation, management of these patients remained unchanged: both antibiotic use and admission rates were steady.

The NHAMCS is an imprecise tool to full discern the reasons for visit on a granular level, but the relative increase in advanced imaging is consistent with increases in CT usage for other indications.  Obviously, the radiation itself has no known therapeutic potential – so, therefore, the clear conclusion is simply the unfortunate presence of additional low-value care.

“Use of Computed Tomography in Emergency Departments in the United States: A Decade of Coughs and Colds”

The Antibiotic Mandatory Waiting Period?

It has become generally accepted within the medical community the vast majority of cases of pediatric otitis media will resolve without antibiotics.  However, the tradition of treating OM with antibiotics is slow to wane, fueled by momentum and parental expectations.  Some success has been achieved with delayed-prescription strategies, where parents are provided with antibiotics, but encouraged to wait a few days and observe for spontaneous improvement.

These authors applied the same school of thought to benign upper respiratory tract infections – sinusitis, pharyngitis, and bronchitis.  They randomized 398 patients with common URTI-spectrum symptoms to one of four treatment strategies:  immediate antibiotic initiation; antibiotics provided immediately, but patient encouraged to wait-and-see a few days for spontaneous improvement; antibiotics available for pick-up three days later, if desired; and no antibiotics.  The primary outcome was duration and severity of symptoms, with various secondary outcomes of absenteeism, satisfaction, and antibiotic utilization.

The results of this study are a little bit mixed.  The patients initiating antibiotics immediately had shorter symptoms duration than any other strategy.  The sample sizes are small, and the standard deviation of symptoms in each cohort is huge, but it’s probably reasonable to estimate antibiotic use truncated moderate or severe symptoms by about a day or a day and a half from a 5-6 day illness duration.  Twelve of 98 randomized to no prescription ultimately crossed over to antibiotics.

But the remainder, despite their randomization to benign neglect, improved regardless, without any detectable difference in safety outcomes.  After all, the majority of these infections are either viral, or self-limited bacterial infections handled by the body’s natural immune system without complications.

The interesting outcomes, however, were the two delayed-antibiotic strategies.  Compared to the 91% antibiotic usage rate of the immediate antibiotic group, the patient-initiated and delayed-collection strategies resulted in 33% and 23% antibiotic usage rates, respectively.  Symptom duration, as to be expected, was mildly attenuated, falling between immediate antibiotic use and no antibiotics.

Is this the happy medium strategy needed to finally divorce ourselves from our addiction to unnecessary care for URTI disease?  “Choosing Wisely” as a general philosophy doesn’t seem to have had the desired effect – how about an executive action to mandate the same waiting period for antibiotics as we have for guns?

“Prescription Strategies in Acute Uncomplicated Respiratory Infections: A Randomized Clinical Trial”

Survival of the Intoxicated

We are mostly all familiar with the beneficial social outcomes associated with occasional alcohol use.  However, we also see the detrimental effects of chronic use in the context of comorbid disease states.

But, it’s apparently not all bad news in a medical sense.

This retrospective evaluation of patients admitted to medical and surgical critical care units in teaching hospitals in Boston simply tracked outcomes segregated by admission blood alcohol level.  Evaluating 11,850 admissions, most admitted patients had undetectable BAC, while 638, 703, and 1,226 patients fell into their three strata of increasing BAC.

Unsurprisingly, as most patients with the highest BAC had the fewest comorbid factors and were overwhelmingly admitted due to trauma, they had the lowest mortality in the unadjusted analysis.  However, across several different types of multivariate adjustment and propensity scoring, the same survival advantage held, although substantially weakened.

There is some school of thought alcohol confers a neuroprotective and anti-inflammatory effect in the setting of acute illness.  These retrospective data, fraught with potential confounders, are inadequate to conclusively confirm or refute such a hypothesis – but it is reasonable to suggest further study of underlying mechanisms may be warranted.

“Association between blood alcohol concentration and mortality in critical illness”

Informatics Trek III: The Search For Sepsis

Big data!  It’s all the rage with tweens these days.  Hoverboards, Yik Yak, and predictive analytics are all kids talk about now.

This “big data” application, more specifically, involves the use of an institutional database to derive predictors for mortality in sepsis.  Many decision instruments for various sepsis syndromes already exist – CART, MEDS, mREMS, CURB-65, to name a few – but all suffer from the same flaw: how reliable can a rule with just a handful of predictors be when applied to the complex heterogeneity of humanity?

Machine-learning applications of predictive analytics attempt to create, essentially, Decision Instruments 2.0.  Rather than using linear statistical methods to simply weight a small handful of different predictors, most of these applications utilize the entire data set and some form of clustering.  Most generally, these models replace typical variable weighted scoring with, essentially, a weighted neighborhood scheme, in which similarity to other points helps predict outcomes.

Long story short, this study out of Yale utilized 5,278 visits for acute sepsis and a random forest model to create a training set and a validation set.  The random forest model included all available data points from the electronic health record, while other models used up to 20 predictors based on expert input and prior literature.  For their primary outcome of predicting in-hospital death, the AUC for the random forest model was 0.86 (CI 0.82-0.90), while none of the rest of the models exceeded an AUC of 0.76.

This still simply at the technology demonstration phase, and requires further development to become actionable clinical information.  However, I believe models and techniques like this are our next best paradigm in guiding diagnostic and treatment decisions for our heterogenous patient population.  Many challenges yet remain, particularly in the realm of data quality, but I am excited to see more teams engaged in development of similar tools.

“Prediction of In-hospital Mortality in Emergency Department Patients with Sepsis: A Local Big Data Driven, Machine Learning Approach”