Conservative vs interventional Treatment for Spontaneous Pneumothorax: Don’t just do something… stand there!

This study in the New England Journal of Medicine is going to change how we treat spontaneous pneumothorax. If you haven’t read it, you should.

(Disclaimer: the lead author, Dr Simon Brown is a colleague of mine at the Royal Hobart Hospital. He is bigger than me and knows where I live. Hmmmm….)

By way of background, there are conflicting guidelines regarding the treatment of primary spontaneous pneumothorax. The Americans are quite aggressive with intervention, the British less so. Any time you see such variation, it usually means there is no high-quality evidence to inform practice.

Fortunately, we now have that evidence.

The Primary Spontaneous Pneumothorax (PSP) trial was a prospective randomised open-label non-inferiority study that enrolled 316 patients in 39 hospitals in Australia & New Zulund over 6 years.

They included stable patients between the ages of 14-50 with at least a 32% (sum of interpleural distance greater than 6cm) primary spontaneous pneumothorax.

Those randomised to “intervention” had a <12 French Seldinger style chest tube placed. They got a repeat chest x-ray in one hour. If the lung was re-expanded and there was no air leak, they clamped the tube. Another x-ray was taken 4 hours later. If all was fine, the tube was pulled and the patient discharged. If not, they got admitted.

The “conservative” arm was observed for 4 hours and a repeat chest x-ray was taken. If they did not require oxygen and they were walking around comfortably, they could be discharged. However, intervention could be allowed for if the patient had clinically significant symptoms despite adequate analgesia, couldn’t walk, unwilling to continue, became unstable or if the chest x-ray was getting worse.

The primary outcome was full lung expansion by 8 weeks. There were lots of arguably more important secondary outcomes.

Results?

As expected, patients were generally young, male and thin. Half were smokers.

Mean pneumothorax size was 65%! Holy crap… that’s big.

For the primary outcome, 98% of the intervention arm had full lung re-expansion at 8 weeks vs. 94% in the conservative arm. This was within their prespecified non-inferiority margin of -9% as such, this is a positive study.

However, even more impressive are the secondary outcomes in favour of a conservative approach.

85% of patients randomised to conservative management got no intervention. The intervention arm had many more adverse events (26% vs. 8%) mostly related to the chest tube. Mean time to resolution of symptoms was 15 days in both groups. Recurrent pneumothorax was double in the intervention arm; 16% vs. 8%.

Median hospital length of stay was 3.8 days in the intervention group compared to 0.2 in the conservative arm!

All of this is quite a good argument for us to change our practice.

For what it is worth, the manuscript goes into length regarding the limitations of loss to follow-up. Even going so far as to perform an analysis assuming all of them were treatment failures making this study “fragile.” While perhaps technically correct, I think this was overzealous peer review that was ultimately biased.

A 26-year-old male, who got no treatment and feels fine is probably less likely to follow up as directed. However, patients in the intervention arm (who got painful tubes and spent 4 days in hospital) were more likely to return for the 8-week visit.

What’s the take home point?

Looks like we’ll be doing less chest tubes in stable patients with primary spontaneous pneumothorax.

 

Bummer… I like putting in chest tubes.

Covering:

Brown S.G.A, Ball E.L, Perrin K, Conservative versus Interventional Treatment for Spontaneous Pneumothorax. N Eng J Med. 2020;382:405-15. [link to article]

Diagnosis of PE with D-Dimer Adjusted to Clinical Probability- A game changer!

Intern, “the D-dimer just came back at 0.6… what should I do?”

Me, “… faaaaaaaaaaaaaaaaaaaaaaaaarrrrrrrk!”

Fortunately, this study is probably a game changer.

These Canadian’s conducted a prospective study in which PE was considered to be ruled out if the pretest clinical suspicion was low (Well’s score0-4) and a D-dimer of less than 1.0mg/L or if the clinical suspicion was moderate (Well’s score 4.5-6) and D-dimer less than 0.5mg/L.

Just over two thousand patients were enrolled. Their overall rate of PE was 7.4%.

Of the 1325 patients with low or moderate clinical suspicion for PE and a negative D-dimer (as above), NONE (95%CI 0.0 to 0.29%) of them had venous thromboembolism during the 3-month follow up.   

This is amazing.

Once again, two-thirds of all patients were excluded from imaging and the diagnostic performance of this strategy was perfect.

Furthermore, this publication from the NEJM is one in line of papers supporting the notion of adjusting D-dimer thresholds. Consistency in the literature is reassuring.

Slam dunk?

Ok… no studies are perfect. What were the two major limitations?

The quality of a study looking at diagnostic performance of a test is only as good its gold standard. Unfortunately, patients that were ruled out by this strategy did not have definitive imaging. This study relied on 3-month follow up (sometimes by telephone) and may have led to the underdiagnosis of PE.

Secondly, there were 34 deaths during follow-up.

Yikes!

But they claim, “no deaths were attributed by the central adjudication committee to pulmonary embolism.” Table S5 in the supplementary appendix shows that most died of cancer or cardiovascular disease. I very much doubt many (if any) of the deaths had autopsy performed. We all know PE can be quite tricky to diagnose. I still wonder if some of these deaths were due to missed PE.

Despite these limitations, I still think adjusting D-dimer thresholds makes sense. And now it is supported by numerous studies published in high impact journals.

Intern, “the D-dimer just came back at 0.6… what should I do?”

Me, “Never order a D-dimer without asking me first!”

Oh… I meant, “send them home…” And hopefully they don’t die of “cancer.”

Covering:

Kearon C, de Wit, Parpia S, et al. Diagnosis of Pulmonary Embolism with D-dimer Adjusted to Clinical Probability. N Engl J Med 2019;381:2125-34. [link to article]

Roc vs. Sux for RSI… a storm in a teacup

Far too much time and effort has been spent debating the winner of the rocuronium vs. succinylcholine contest. But it is a bit too hard to ignore this paper claiming to be the first RCT of these drugs in an emergency setting.

This was a multicenter, non-inferiority RCT of 1248 patients with out-of-hospital cardiac arrest in France. Tout bon!

Rocuronium dosed at 1.2m/kg was compared to succinylcholine at 1mg/kg. The authors note the patients were blinded to the assigned drugs. Yes… really… Patients with no health insurance were excluded. Yes… really…

About 90% of the intubations were conducted by a specialist emergency doctor. They only used direct laryngoscopy (no video) and half the time intubated in some pretty creative places… i.e. on the ground.

The primary outcome was first pass intubation success rate. Hardly a patient oriented outcome, but at least it is reasonably objective and measurable. They also looked at some secondary outcomes.

A non-inferiority margin of 7% was chosen.

Results?

It turns out that rocuronium was not non-inferior to succinylcholine!

Howzat…?

First pass success rate was 79% in the sux group vs. 74% in the roc group. The difference in the 95% confidence interval was up to 9%. This is greater than their non-inferiority margin.

Sux is the winner!

Not really.

There were more adverse events in the succinylcholine group.

Unfortunately, this study will not end the Sux vs. Roc debate. C'est la vie... But either way it doesn’t really matter.

Both drugs are fine. And in your practice, you can always use both.

Don’t forget some contraindications to giving sux (mostly hyperkalemia) and don’t forget to provide post intubation sedation after giving a proper dose of roc.

Covering:

Buihard B, Chollet-Xemard C, Lakhnati P, et al. Effect of Rocuronium vs Succinylcholine on Endotracheal Intubation Success Rate Among Patients Undergoing Out-of Hospital Rapid Sequence Intubation. JAMA 2019;322(23):2303-2312 [link to article]

Comparison of manual pulse palpation, cardiac ultrasonography and Doppler ultrasonography to check the pulse in cardiopulmonary arrest patients

This study aimed to determine which method of pulse check is the best during CPR. Should we do the traditional manual pulse check, use our ultrasound to look at the heart (POCUS) or use a Doppler on the femoral artery.

Of course, to do such a study would require comparison against a gold standard and calculation of test performance by reporting sensitivity and specificity.

This study did none of this.

In addition, the manuscript is incredibly poorly presented and should not have been published in its current form (example below). Given the sloppiness of the paper, one should really question the veracity of the research conducted.

A more detailed USG was performed when the rhythm returned to normal or when the patient accepted exitus.

As this study looks to be fatally flawed, I should probably stop here… but I will mention a few things.

This study allegedly enrolled 137 patients in a single very busy Turkish ED. Two ED residents with a year of POCUS experience did all of the echo’s. All three methods to check the pulse were performed at the same time; at the start of resuscitation, at 15 minutes and at the end of efforts or ROSC. They tried to maintain blinding by curtain set up as seen in the picture.

They claim the echo’s only took 4 seconds to perform. Contrasted with doppler taking 8 seconds and manual palpation 10 seconds. This sounds impressive (and contrary to other published reports) until you consider the two ED residents were not blinded to the study aims. Nor did they mention how the times were measured.

The study claims the echo was the best (as it was always right by design) and there were a bunch of false positive and negatives with the Doppler and manual palpation.

I’m a big proponent of POCUS and have already incorporated echo as my default pulse check. I can get the images quickly and I feel a I get a better sense of cardiac contractility than with other methods. I occasionally find important pathology such as tamponade, RV strain, dissection, catastrophic valvular regurgitation or cardiac standstill.

AHA/ACC, ILCOR and ERC guidelines support POCUS in arrest, but caution it should not interfere with high quality CPR. To be fair, there has been little high-quality research. There is no doubt we should encourage further formal study with robust methods and design. 

We don’t want our patients to accept exitus.

Covering:

Zengin S, Gumusboga H, Sabak M, et al. Comparison of manual pulse palpation, cardiac ultrasonography, and Doppler ultrasonography to check the pulse in cardiopulmonary arrest patients. Resusitation 2018;133:59-64. [link toarticle]

Pregnancy-Adapted YEARS Algorithm for Diagnosis of Suspected Pulmonary Embolism

There must be something in the coffee houses that make the Dutch so sensible.

Just like the HEART score, they have created another simple clinical decision instrument that has face validity (regardless if it is actually proven). This time, it’s for investigating potential PE in pregnancy.

This multicentre, international trial was conducted at 18 hospitals over 5 years. They included pregnant adults who had been referred to the ED or obstetrical ward with suspected PE.

They aimed to determine whether their pregnancy-adapted YEARS algorithm could be used to avoid diagnostic imaging in this at-risk population.

Simply put, the algorithm is as follows:

• Three criteria were assessed in all patients; clinical signs of DVT, haemoptysis, whether PE was thought to be the most likely diagnosis. D-dimer was measured.
• Patients with clinical signs of DVT underwent ultrasound.
• If a patient did not meet any of the three YEARS criteria and the D-dimer was less than 1000ng/ml a diagnosis of PE was considered to be ruled out.
• If a patient had any of the YEARS criteria and the D-Dimer was less than 500ng/ml then PE was ruled out.
• All the remaining patients were referred for CTPA.

The primary outcome was the occurrence of symptomatic DVT or PE within 3 months. (Patients did not undergo routine screening.) Secondary outcome was the proportion of patients where CTPA was not indicated.

Results?

About 500 women were included. 20 (4%) of patients had a PE diagnosed at baseline. Only one patient (0.2%) was diagnosed with a DVT in the follow up period. There were no PE’s and no deaths. CTPA was avoided (not indicated) in about 40% of patients. The efficiency of the algorithm was better during the first trimester (65% avoided CTPA).

Major limitations?

Gold standard screening tests were not performed to look for subsequent DVT or PE. Much like it is hard to find a fever if you don’t take a temperature, it is hard to find VTE if you are not testing for it. But they reportedly did look if patients had symptoms… which is the practical and ethical approach in this RCT.

It is hard for this study to claim safety regarding mortality as death from PE in pregnancy is reasonably low. It will never be feasible to enrol enough patients in an RCT to asses mortality benefit.

What’s the take home?

Using clinical gestalt (i.e. the 3 YEARS criteria) and a pregnancy adjusted D-dimer is a sensible approach. 

Covering.

Van der Pol LM, Tromeur C, Bistervels IM, et al. Pregnancy-Adapted YEARS Algorithm for Diagnosis of Suspected Pulmonary Embolism. N Engl J Med 2019;380:1139-49. [link to article]

IV cefazolin plus probenecid vs oral cephalexin for the treatment of skin and soft tissue infections

It is traditionally thought that IV antibiotics are superior to those given orally. But IV administration comes with extra time, cost, resources, and possibly not necessary.

To answer the question, these authors designed an elegant bias-busting study. It was a prospective, multicentre, double-dummy blind, randomised non-inferiority trial conducted in two Canadian ED’s.

However, the authors soon learned valuable lessons in performing an RCT; feasibility and generalisability.

They had quite strict inclusion and exclusion criteria that hindered patient recruitment. In addition, this also made any conclusions rather difficult to generalise to a broader patient population.

They included adult patients who had what they thought was mild to moderate cellulitis (a bit subjective). They excluded very mild patients, those with renal disease, prior MRSA, prior antibiotic use in the preceding 7 days, infections requiring I&D or debridement, two or more signs of SIRS, bites, PVD, perioperative wounds, etc, etc.

They even excluded patients with BMI greater than 30. That’s most of Canada!

Arguably defeating the whole purpose of the study, participants were brought back to the ED every day for 7 days of treatment. This is hardly decreasing resources and simulating real conditions.

The primary outcome was failure rate at 72 hours. About 300 patients were thought needed to power the study to a 10% non-inferior margin.

Why 72 hours? A great pearl… 

“This end point is beyond the clinical extension of redness that may be seen in the first 24-48 hours of therapy which would not typically represent treatment failure.”

Results?

Over 4 years, they screened 2855 patients for eligibility but finally gave up after randomising 206 patients.

The proportion of patients failing therapy at 72 hours was not statistically different and about 5% in both groups. The authors conclude cephalexin is just as good as IV cefazolin for mild to moderate cellulitis.

It is probably true that we overdo IV antibiotics for many infectious disease processes. But it is disappointing to see such a well-designed study design ultimately unable to provide us with much useful information.

Covering:

Dalen D, Fry A, Campbell, S, et al. IV cefazolin plus probenecid vs oral cephalexin for the treatment of skin and soft tissue infections. Emerg Med J 2018;35:492-429 [link to article]

Duration of ECG monitoring of ED patients with Syncope

How long do we need to monitor patients after an episode of syncope? What a great question…

These authors used data that came from a large multicentre study that developed the Canadian Syncope Risk Score (CSRS). They enrolled consecutive adult patients who presented within 24 hours of a simple syncopal episode.

They collected data on baseline characteristics and were eventually able to risk stratify patients by using their newly derived syncope score into low, intermediate and high-risk groups.

The primary outcome was serious arrhythmic conditions (arrhythmias, interventions for arrhythmias, and unexplained death) over 30 days.

The 30 day follow up was a tremendous undertaking. They had structured review of all available medical records, subsequent visits, hospitalisations, and deaths. They performed a scripted telephone follow up at 30 days and they looked at administrative records for subsequent visits.

A whopping 5581 patients were analysed. As expected, only a small number 7.5% had serious outcomes. The syncope score seemed to work pretty well. The low risk group had 0.4% 30-day arrhythmic outcomes. The intermediate and high risk were 8% and 25% respectively.

The authors state that one-half of arrhythmic events were identified within 2 hours of ED arrival in low risk patients and within 6 hours in the intermediate and high-risk patients.

Based on this, the authors suggest either 2 or 6 hours of cardiac monitoring is enough after syncope but outpatient cardiac monitoring for 15 days in higher risk patients.

There were a few problems…

Only 11% of the patients were hospitalised and the median ED length of stay was only 4 hours. It is no surprise that they found half of the events early on when they were actually looking. It is quite possible that many events were missed when the patients were not being monitored at home. However, they should have at least they should have picked up on mortality.

Remember the extraordinary work they did in assessing the 30-day outcomes? Unfortunately, quality of data is often inversely proportional to the effort and complexity in obtaining it. In other words, they may have been making conclusions based on bad data.  It’s hard to know.

There have been several attempts at creating a clinical decision instrument for syncope. Unfortunately, these instruments don’t tend to work in complex disease processes. The CSRS has not  been externally validated nor compared to clinician gestalt.

What’s the take home point?

From an evidence-based medicine standpoint, I don’t believe this single study is enough to justifiably change practice. But in our hearts, we all know that low risk patients probably don’t need to be watched very long… and high-risk patients need to be monitored longer. 2 or 6 hours? Who knows…

Covering

Thiruganasambandamooooorhylinglangadong, V, Rowe B, Sivilotti M, et al. Duration of Electrocardiographic Monitoring in Emergency Department Patients with Syncope. Circulation 2019;139:1396-1406. [link to article]