Thursday, July 9, 2015

The stethoscope – “it’s not a lie if you believe it.”

The skillful use of the stethoscope is a key element of the physical exam. Introduced in 1816, it afforded an unprecedented “view” inside the body. Since then it has evolved to become synonymous with being a physician (as well as nurse, paramedic,  tech, etc. 
But technology has progressed since Laënnec pressed his crude prototype against the chest of his first patient. We have discovered what electromagnetic radiation can reveal in an x-ray; we discovered that subatomic particles can be excited in a magnetic field, producing images of exquisite detail. And we discovered, uh, sound.  {Citation needed}

But regardless of all these new technologies, we all keep carting around our space-age versions of Laënnec’s hollow wood stick, draped around our necks, ready to bonk us in the eye if we make any sudden moves.   Why?

We all occasionally joke that these diagnostic devices serve mostly as “doctor jewelry,” and that we will always use a CXT, CT, MR, or whatever, to confirm our auscultations. However, if the conversation turns serious, we will solemnly avow the sacrosanct role of the hollow stick. “Nothing can replace a thorough Hx and P,” we intone, “nothing.”

We are, in that moment, the most harmful species of liar – the liar that believes the lie.

"Jerry, just remember. It's not a lie... if you believe it."
We need to stop, take a deep breath, look ourselves in the eye, and admit that we have not been truthful with thine own self. Dare to say the following; first while alone in your locked bathroom, then in a safe 12-step environment, and then eventually in public:

The stethoscope is a vestigial element of the physical exam, and ought to be retired.

Sort of the tone I'm shooting for.

First off, the physical exam in general is not that helpful.

Accordingly, the stethoscope is a vestigial element of a whole ritual of limited utility. In one study, the history was 20 times more useful than the physical exam. Heck, just using the ECG, CXR, and basic labs was more useful than the physical exam! In another study, looking at the ability to diagnose nerve root compression in back pain, the exam only improved the accuracy marginally, budging the AUC from 0.8 to 0.83. (OTOH, in appendicitis, tenderness in the RLQ kind of blows history out of the water...)

Second, the stethoscope sucks (or blows?)

While revolutionary in its day, the 'scope has seen its diagnostic performance plummet as medicine progressed. Pneumonia provides the best example of the limited contemporary accuracy of the stethoscope.

A systematic review in the Emergency Medicine Journal concluded that "pneumonia cannot reliably be confirmed or excluded by auscultation, or indeed physical examination, alone.” The most recent study in that review looked at patients coming into the ED with “chest symptoms,” and looked at how the diagnostic impression changed from pre- to post-auscultation, and then post-discharge. The vast majority of the time, about 95%, the clinical impression did not change after using the sacred hollow stick
Might have better luck with a Harry Potter replica wand.
Pneumonia, in particular, was diagnosed correctly in 45% of patients with history alone, and only improved to 49% after the exam.

An earlier study looked only at CXR-verified pneumonia, and blinded the auscultators to the clinical history. It wasn’t pretty – the sensitivity of the 3 examiners ranged from 47% to 69%. Specificity wasn’t much better.

This all comports with a JAMA metaanalysis from 1997 that concluded that “no individual clinical findings, or combinations of findings, that can rule in the diagnosis of pneumonia.” Specifically, while one study was wildly bullish on egophony, all the auscultation signs were of variable sensitivity, and generally low yield.

(BTW, CHF is even worse for the proponents of ‘scopes. While crackles modestly raise the likelihood of CHF, and wheezes modestly decrease such likelihood, there is some serious overlap! Between 1/5 to 1/3 of patients manifest wheezing with acute CHF. By the same token, wheezing is surprisingly unhelpful in diagnosing or excluding obstructive lung disease.)

Compare this to using ultrasound

While some authors have stated that “highly-skilled sonographers“ are required to diagnose pneumonia with ultrasound, a recent study required pediatric EM doctors to undergo only 1 hour of training. Despite this limited education, the sensitivity and specificity of US for CXR-proven consolidation was 86% and 89%. However, another study, using CT scan as the gold standard, found US to be far more sensitive than CXR. A raft of other studies have come out in the last few years (do they ultrasound anything but lungs in Italy?), backing up these results.

So, why are you carrying that MRSA biofilm-vehicle around the same neck your kids touch?

If you really need some jewelry to identify you as a doc, you can get a stethoscope earring. Cute as heck, and unlikely to get covered in MRSA, or give you a black eye.

Thursday, June 18, 2015

Oral contrast for abdominal CT?

PO, IV contrast for abdominal CT
This material is educational, and is not intended to serve as hospital protocol.
Mostly, it's here to serve a quick reference for myself, and for impromptu discussions with PAs and residents.


Suspected SBO
Oral contrast is contraindicated in suspected SBO.
IV contrast is preferred, but not required.

Suspected appendicitis
Oral contrast is not required, but may be beneficial in extremely thin individuals (e.g. BMI < 18), or non-obese children.

Suspected diverticulitis
Oral contrast is not required

Unclear etiology
PO contrast is suggested by some experts in undifferentiated abd pain if the patient
1)   Is very thin (e.g. < 120 pounds, BMI < 18);
2)   Has had a Roux-en-Y gastric bypass; or
3)   Has and inflammatory bowel disease that could produce a fistula

Discussion and References

Per the American College of Radiology guidelines, in a patient with suspected SBO, PO contrast is contraindicated. They explain that: “Oral contrast will not reach the site of obstruction, wastes time, adds expense, can induce further patient discomfort, will not add to diagnostic accuracy, and can lead to complications, particularly vomiting and aspiration.”    

Multiple studies have shown that the use of PO contrast does not increase the accuracy of the diagnosis of appendicitis (references below). An editorial in a recent radiology journal concluded that “routine administration of oral contrast medium may not be necessary in the setting of suspected acute appendicitis.” American College of Radiology guidelines state that PO contrast “may not be needed,” leave the decision to use PO contrast to “institutional preference.”
Some have suggested that PO contrast might be needed in extremely thin individuals (e.g. BMI < 18). See the discussion below in "Unclear etiology."

A 2005 systematic review from the American Journal of Surgery concluded that: “Noncontrast CT techniques to diagnose appendicitis showed equivalent or better diagnostic performance compared with CT scanning with oral contrast.”

A 2009 randomized study in the American Journal of Radiology found that “nontraumatic abdominal pain imaged using 64-MDCT with isotropic reformations had similar characteristics for the diagnosis of appendicitis when IV contrast material alone was used and when oral and IV contrast media were used.”

A 2011 retrospective study in the Journal of Surgical Research found that oral contrast did not reach the cecum in 1/3 of patients. Furthermore, “there appears to be no diagnostic compromise in those without contrast in the terminal ileum.
A 2014 prospective study in the Annals of Surgery concluded that: “Enteral contrast should be eliminated in IV-enhanced CT scans performed for suspected appendicitis.”

The American College of Radiology guidelines state that PO contrast “may be helpful for bowel luminal visualization.” Despite this, they conclude that “regardless of the [the use or lack of IV or PO contrast], the accuracy is high for depicting findings of acute diverticulitis.”

A 2006 prospective study in Emergency Radiology looked at patients with undifferentiated abdominal pain in the ED; 1/5 of them had LLQ tenderness, and diverticulitis was the second-most common suspected diagnosis. All patients had CT scans with and without PO contrast; no IV contrast was used. Although there were discordant interpretations between the (+) PO and (-) PO CT scans, the authors found that a “significant portion of the discordance was attributable to interobserver variability.” Thus, it would appear that a CT scan without either PO or IV contrast could be accurate in the diagnosis of diverticulitis.

Unclear etiology
Oral contrast may not be needed for undifferentiated abdominal pain, as per the discussion of the study in Emergency Radiology noted above.

However, in a 2012 study in Emergency Radiology that examined the use of PO contrast in undifferentiated non-traumatic abd pain, the authors excluded 2 groups of patients.  They excluded all “subjectively thin” patients, but did not provide cut-offs. They noted the research that suggests this exclusion is unnecessary, and may be dropped in the future.  They also excluded patients at high risk for having an intra-abdominal fistula. Primarily, they targeted patients who had had a Roux-en-Y gastric bypass, or who had IBD, disposing them to fistula formation.

Thursday, June 4, 2015

Is the shoulder in? Use ultrasound!

It was a classic “good news/bad news” sign out.

The  attending who was signing out to me admitted that this wasn't the cleanest sign-out, but they had a plan.

The bad news: the patient likely had a shoulder dislocation, but the read on the X-ray was equivocal.
"Can I order a clinical correlation?"
Given this unclear picture, the plan was to try sedation, and pop it back in place. 

The good news: the attending had already tasked a senior resident and the orthopedics PA-C to perform the sedation and (attempted) reduction, and they were champing at the bit to do the procedure. 

My evaluation

I looked at the X-ray, and re-examined the patient. As he was young chubby guy, it was difficult to be sure of the exam. Since both my exam and the X-ray were so unhelpful, I stalled the resident and PA-C while I grabbed the ultrasound.

Now, ultrasound is not usually thought of as a great test for shoulder dislocation. The usual approach to definitively excluding a glenphumeral dislocation (especially posterior dislocation!) is to obtain an axillary view, shooting up through the armpit. And of course we can always get a CT of the joint.

But you can get much the same image with the ultrasound. So, placing the probe on the posterior shoulder, I aimed anteriorly, with the left side of the screen oriented to the lateral aspect, like this:

This produced a clear image of the glenohumeral joint:

As usual, US is best appreciated dynamically, with this clip showing the patient rotating his arm:


I hope you can see how the head of the humerus is well-seated in the glenoid. But despite this dazzling proof, x-rays and a CT were needed to convince ortho that the shoulder was not dislocated, and that no sedation and tugging were needed.

The (ionizing radiation) proof

The axillary view of the shoulder X-ray (needs to be specifically ordered), oriented similarly to the US:

 The CT of the shoulder, with the slices and orientation similar to the US:

If you want to check out some other US examples of dislocated shoulders, you can also check out this great video from ALiEM:

Ultrasound for the ... Wait, what?!

I would have liked to say this was an ultrasound win, but I'm not so sure. The downsides of this approach, in retrospect, were:
  • The patient didn't get to enjoy some of our high-grade ketamine;
  • We didn't get to high-five each other after our "subtle reduction;" and
  • The resident was cheated out of a procedure!
So, know the downsides ahead of time.

Sunday, March 8, 2015

High-flow nasal cannulas: Not yet ready for ED use.

What is high-flow nasal cannula (HFNC) therapy, and, more importantly, does it work? A recent segment on EM:RAP went into a fair amount of detail about the putative mechanism, so I’ll leave that alone.

What that segment left out, though, was any discussion of the published evidence pointing to the benefit of HFNC. And indeed, while there are a lot of anecdotal reports and personal testimonials, the actual data hasn’t been clearly discussed. Here’s a quick review of what we currently know.

1. Preterm infants

Although the “diet-PEEP” argument would suggest that HFNC might have a large role in tiny patients, it isn’t clear that HFNC helps prevent intubation in preemies. A Cochrane review suggested that using HFNC post-extubation, instead of CPAP, might lead to higher re-intubation rates. An RCT that was conducted after that review, however, found that HFNC was roughly equivalent to nasal CPAP for respiratory support in this age group. Data from other recent trials haven’t been as convincing, though, and many neonatologists still find the evidence “insufficient” to use HFNC in place of more established therapies

2. Infants

It may help prevent intubation of little kids with bronchiolitis, although the data is weak. The two most relevant studies were retrospective chart reviews, using a before-and-after design, looking at overall rates for intubation in the time period after HFNC was introduced to the pediatric service. Nonetheless, McKiernan found that intubation rates for bronchiolitis dropped from 23 % to 9%, and Schibler found the rate plummeted for 37% to 7%!
McKiernan 2010
This may end up being one of the best-supported roles for HFNC, and high-quality studies are in progress that could help clarify the issue.

3. Adults

The initial trials in adults have demonstrated modest improvements in oxygenation, but haven't studied patient oriented-outcomes. For example, one study found that oxygenation mildly increased after HFNC initiation, but no control group was used.

Sztrymf 2012

In one controlled trial, versus face mask, fewer desaturations were seen with HFNC. Despite this, there was no statistical difference in the use of rescue CPAP/BiPAP. A few other studies have further noted that HFNC seems to be more comfortable than CPAP.

A single-author review, otherwise very bullish on HFNC therapy, conceded: “although some clinicians may have the impression that in some instances, use of HFNC has avoided intubation, this has not been shown in a controlled trial.”

The largest current review on the use of HFNC concludes that (my emphasis):
“While theoretical advantages exist over standard nasal cannula and face mask oxygen, current evidence does not definitively demonstrate superiority to other methods of respiratory support. Few studies have focused on clinical outcomes beyond common respiratory parameters. Given the potential lack of consistency of positive pressure generated with current HFNC systems, NIV such as CPAP and bilevel positive airway pressure should still be considered first line therapy in moderately distressed patients in whom supplementation oxygen is insufficient and when a consistent positive pressure is indicated.”

Bringing it home!

Being an “early-adopter” is cool – if you’re lining up to get the new iPhone or Zune! In medicine, however, it doesn’t often pay to jump on a bandwagon before the data is in. (Want to buy some Xigris cheap?) We are being encouraged to try a new therapy that uses proprietary (proprietary = $) devices, with soft indications, scant evidence, but with touted outcomes such as “improved comfort,” instead of mortality or rates of intubation. We should be cautious.

Particularly concerning is the uncritical enthusiasm for the use of this device in situations that either clearly call for other therapies, or for no therapy. For example, some describe the utility of HFNC in patients who are “extremely hypoxic,” but there is little evidence that HFNC improves outcomes in this population.

HFNC is probably more useful for precisely titrating FIO2 in the (mythical?) CO2 retainer. But if there is a concern about the PaCO2, why not use a proven therapy like NPPV that we know saves lives?

Lastly, some clinicians promote the use of HFNC for CHF, since there is (wink, wink) a “PEEP component,” but that’s a patient who needs CPAP or BiPAP as well, since we already have proven a mortality benefit in that population as well.

Sure, you can relax, talk with family, and eat while wearing HFNC, but if you are so dead set on wolfing down a sandwich, you probably don’t need an expensive, unproven therapy. You need 2 liters per minute, and a floor bed!

Monday, December 29, 2014

Femoral pulse during CPR - Arterial or Venous?

There are many uncertainties about how each of us will die.

Based on my experience in the ED, however, many of us will spend our last minutes on Earth with a stranger's hand shoved into our groin.

Your own hand? Fine.
I'm talking about the checking the femoral pulse during CPR, of course. It's a well-enshrined part of our resuscitation practices, but what it is it actually telling us?

The Theory
The idea, of course, is that effective CPR will produce an arterial pulse, albeit weak, that will generate a cardiac output about 20% of normal. Palpating a femoral pulse during compressions supposedly verifies that the CPR is being effectively delivered.

Evidence that the femoral pulse is venousHowever, there is some uncertainty about what a palpable femoral pulse actually represents. Hilty used ultrasound in a study of central line placement during cardiac arrest, and noted that 9/20 patients appeared to have femoral venous pulsations, rather than arterial.
Note that this was in the dark ages of ED ultrasound (1997), back before Christian Doppler was born. 

Based on this finding, as well as on a case report of two kids getting open-chest cardiac compressions, many people now believe that the femoral pulse during CPR is just the venous back-flow. A video posted by an emergency physician in Qatar supports this view, where he demonstrates interrogation of the femoral vessels with power doppler during CPR:

The femoral vein shows much brighter signal than the artery, suggesting that the venous flow far exceed the arterial. Of course, since this is power doppler, we don't know the direction of the flow.

Evidence that the femoral pulse is arterial?
Cardiac arrest, asystole when EMS arrived in the ED, the LUCAS dutifully chugging along.  

Probe placed in the right groin, angled cephalad slightly. First, with color doppler:


The femoral artery seems to show a fairly well-defined arterial pulse, while the femoral vein has a turbulent, almost "yin-yang-like" character, that does not suggest effective flow. So based on this clip, it seems like the femoral pulse indeed reflects arterial impulses, not venous.

(BTW, here's a clip of the same view, while the LUCAS was taking a break. Just so you know that there was no spontaneous cardiac activity mucking up the doppler.)


So, what does the femoral pulse tell us about CPR quality?
Not much, probably.
The scanty and conflicting "evidence" (i.e. collection of anecdotes) reviewed here doesn't make it clear if pulsations in the groin are coming from the artery, the vein, or perhaps even both. This looks like a promising avenue for an emergency ultrasound study!

In the meantime, assessing the quality of CPR is likely best done with end-tidal CO2, although a recent Ultrasound Podcast episode suggested using focused echo  to optimize compressions. 

Monday, December 15, 2014

Septic hip in Kids, part 2: This time it's personal...

I wrote about the "textbook" approach to differentiating the septic hip from plain ol' synovitis in a previous post Septic hip in kids: 5 myths of ED evaluation. Today, I want to review a real case of mine, and highlight both the weakness of the older methods, and the advantages of the newer perspective.

The Case

A 4 year-old male complained of pain in his hip, bad enough that he couldn't put his weight on it. It had started 1 day ago, but had worsened despite acetaminophen. There was no significant recent trauma. He has seemed ill yesterday, though, and while the parents had not taken a temperature, he had "felt hot," and had some mild chills. His temperature in the ED was 100.2.

His right hip was held in slight flexion and external rotation, and he could not bear weight. Labs were obtained, and showed a WBC of 12.2, and (after hours of waiting) and ESR of 28. The CRP doesn't get run at night.

Were the Kocher criteria helpful? (Spoiler: No!)

Kocher-type criteria are applied to help decide on the need to perform an ultrasound, and then decide on the need for aspirating a hip effusion if it is found on US. How did our patient do?
  • Non-weight-bearing - 1 point
  • Temp < 38.5 - 0 points
  • WBC > 12 - 1 point
  • ESR < 40 - 0 point
  • CRP - not available
So, 2 points, which means that he has a probability of having septic arthritis of...
Sultan 2010
... somewhere around 11 - 60%. But notice that if the WBC had been just trivially lower, say 11.8 instead of 12.2, we would have only had 1 point, and the risk would have dropped to 3- 36%.

Heck, say we also had given him a strong analgesic, and he was then able to walk a little. He would have then had 0 predictors, and so only had a risk of somewhere between < 0.2% and ... 17%???

Clearly, these criteria are not helpful at "ruling-out" the possibility of septic arthritis, so I decided to decide by looking at the hip.

Was Ultrasound helpful? (Spoiler: Yes!)

Quick guide to placing the probe :
From Tsung and Blaivas

First I checked the "good" hip, the left side:


I then checked the "bad" hip, the right:


Looked like there was a difference!

Generally, any effusion greater than 5 mm, or 2 mm greater than the contralateral side, is considered positive. This looks pretty unambiguous, and the radiologogist agreed, and told me that a formal ultrasound was not needed in this case.

Clinical course:

The hip was aspirated by IR in the ED (Those guys are total champs at this, but I'm doing this next time! Very straight-forward.), and the patient admitted. The cell count cam back quite high, around 100,000 white cells/ml, which would have placed this patient as a "true positive" case of septic arthritis in almost any study.

However, by the time this result came back, the patient was tearing around the pediatric floor, completely asymptomatic. The culture never grew out an organism, and he was discharged as a transient synovitis.

Bottom Line:

This was the first time I had a patient where I was truly concerned about septic arthritis of a pediatric hip. The use of US clarified the diagnosis early in the visit, and pointed to a greater role in future case. As I highlighted in my prior post, use of point-of-care US in the ED has the potential to "flip" the older diagnostic and therapeutic pathway - use it!

Saturday, December 13, 2014

Pulmonary Embolus: See the echo, and believe the ECG!

Too many people are nihilistic about studying the ECG for signs of PE, and believe that the ECG is too non-specific to play a role. In particular, two beliefs stick in my craw:

“The most common sign of PE on the ECG is tachycardia”
This is not necessarily true. For example, Ferrari found that TWI in V1-V4 was far more common in PE (68% of patients) than was sinus tachycardia (only 36%). Likewise, the average heart rate in Kosuge’s 2007 study was only 94!

“PE = S1Q3T3
S1Q3T3 can be very specific for PE, and is helpful to note. But it isn’t the only thing to look for on the ECG, and the poor sensitivity of this sign could mean missing a PE. Kosuge has found that only 20-22% of PE patients had an S1Q3T3, while the TWI in leads III and V1 was far more sensitive.

The Case
An 80-something year old male was brought to the ED by EMS, feeling weak. 

He said that this had been going on for about a week, and was getting worse. Only when directly queried did he admit to orthostasis, and in fact had syncoped while shopping the day prior. He denied any chest pain, but endorsed some mild dyspnea.

He had a history of CAD, PCI, and a remote history of a PE. His memory and the records were vague on this last point, and he was not on anticoagulation.

His vitals and exam were unremarkable. Since this could have been ACS, or even mild CHF, a troponin and BNP were ordered.

The ECGs
An ECG was immediately obtained:
The baseline wanders, but there is a clear S1Q3T3. Furthermore, there is T wave
inversion (TWI) in the anterior leads, from V1 to V5,. These findings suggest RV strain

Kosuge showed in 2007 that, in patients with anterior TWI and symptoms suggestive of either ACS or PE, TWI in both leads III and V1 strongly favored PE over ACS.1 

A recent update from Kosuge confirms and extends those results, demonstrating that TWI in both leads III and V1 and/or peak TWI in leads V1 or V2 was extremely sensitive and specific for PE (versus ACS due to LAD occlusion).2

So, the ECG proves it, right? Unfortunately, when the emergencu physician reviewed the ECG from 5 months prior...

An arguable S1Q3T3, and TWI in III and V1-V4.
… and from 7 years prior...

'Bout the same.
... it seemed like the ECG findings were, perhaps, simply chronic. Well, when in doubt, break out the ultrasound!

The Echos
Parasternal long-axis

There is a dilated and hypokinetic RV, while the anterior wall contracts nicely.

Parasternal short axis

Septal bowing, or D-shaped septum (“Movahed’s sign?”)

Apical 4-chamber

Markedly dilated, hypokinetic RV free wall, with preserved apical contractility
(a.k.a. McConnell’s sign)

Triscuspid valve – color Doppler

Moderate regurgitation,  max velocity 5 m/s by CW Doppler, indicating very high pulmonary artery pressure

Clinical course
Given the patient’s CKD, and the stable hemodynamic status, unfractionated heparin was started, and the patient was admitted. A V/Q scan the following day confirmed an acute PE, apparently with significant chronic emboli as well! Anticoagulation was bridged to oral therapy, and the patient was discharged back home.

Bottom line:
The ECG can be very helpful is suggesting PE. Many physicians are nihilistic about studying the ECG for signs of PE, and believe that the ECG is too non-specific to play a role. 

In particular, two elements of the conventional wisdom are often uttered without considering the evidence:

“The most common sign of PE on the ECG is tachycardia”
This is not necessarily true. For example, Ferrari found that TWI in V1-V4 was far more common in PE (68% of patients) than was sinus tachycardia (only 36%).3 Likewise, the average heart rate in Kosuge’s study was only 94!

“No S1Q3T3 = no signs of PE”
S1Q3T3 can be very specific for PE, and is helpful to note. But it isn’t the only thing to look for on the ECG, and the poor sensitivity of this sign could mean missing a PE. Kosuge found that only 20-22% of PE patients had a PE,1,2 while the TWI in leads III and V1 was far more sensitive.

Perhaps the ECG isn’t completely sensitive for picking up every tiny sub-segmental embol-ette.  Remember, though, that the ECG isn’t that sensitive for picking up every little troponin leak either! But the ECG is darn good for picking up the STEMI patient that needs emergent therapy, and the analogy with PE is likely true as well – the ECG will suggest the diagnosis in patients who are at high risk from large PEs.

1.         Kosuge M, Kimura K, Ishikawa T, et al. Electrocardiographic Differentiation Between Acute Pulmonary Embolism and Acute Coronary Syndromes on the Basis of Negative T Waves. Am J Cardiol. 2007;99(6):817-821. doi:10.1016/j.amjcard.2006.10.043.
2.         Kosuge M, Ebina T, Hibi K, et al. Differences in negative T waves between acute pulmonary embolism and acute coronary syndrome. Circ J Off J Jpn Circ Soc. 2014;78(2):483-489.
3.         Ferrari E, Imbert A, Chevalier T, Mihoubi A, Morand P, Baudouy M. The ecg in pulmonary embolism : Predictive value of negative t waves in precordial leads—80 case reports. Chest. 1997;111(3):537-543. doi:10.1378/chest.111.3.537.