Journal Club Summary, May 2016: Revisiting antibiotics for abscess; Reconsidering the diagnostic evaluation of Acute Heart Failure; the Power of the probe in diagnosing undifferentiated hypotension.

Many thanks to Mike and Karen Lambert for hosting on a beautiful evening, and to Matt DeStefani, Frank Lee, Mike Stanek, Mike Kennedy, Liz Regan and Bristol Schmitz for their cogent, concise article analyses.

Article 1: Talan DA, et al: Trimethoprim-Sulfamethoxazole versus Placebo for Uncomplicated Skin Abscess. N Engl J Med 2016;374:823–832.

Brief background; CA-MRSA abscess is the culprit behind the huge spike in SSTIs seen in our emergency departments over the past 20 years. While historically abscesses have been successfully treated with I/D alone, CA-MRSA is virulent and pesky, and small studies have suggested a role for antibiotics in addition to I/D, both for improved resolution of infection and to help prevent abscess recurrence. These studies have been underpowered, as the I/D success rate even in the age of CA-MRSA is around 80%, making it difficult to demonstrate small differences in cure rates in small studies. The most recent 2015 IDSA guidelines on SSTI recommend I/D and no antibiotics for simple, uncomplicated abscess.

So, a shout out to Dave Talan and Greg Moran, UCLA EM, for publishing this double blind RCT of 1265 patients >12 yo who presented with uncomplicated abscess to 5 US emergency departments and who were treated as outpatients. While immunosuppression was an exclusion criteria, 11% of included patients did have diabetes and 8% had prior MRSA. About half of wound cultures grew MRSA. Average abscess size was 2.5 x 2 cm, with an average area of erythema of 7 x 5 cm.

The primary endpoint was clinical cure of abscess 7-14 days after end of treatment with either 7 days of 2 DS Bactrim bid or placebo. Clinically important secondary outcomes included the need for surgical drainage, hospitalization, development of new infection in patient or household contacts, and invasive infections.

Results: Clinical cure was 80.5% in the Bactrim group, 73.6% in the placebo group; an ARR 7% for cure with a NNT of 14. There was still a high cure rate even without antibiotics...likely even higher assuming those lost to follow-up were clinically cured. Bactrim was superior to placebo for most secondary outcomes, including a 7% ARR in skin infections at new sites (10.3% vs. 3.1%). Subsequent invasive infections were rare in both groups. Adverse events were mild and most commonly GI (43% in Bactrim group vs 36% in placebo group). No C. difficile was reported.

So why not just give everyone with an abscess Bactrim after the I/D? Well, for starters, the majority of abscesses still resolved without antibiotics. Somehow, they didn’t see any, but you have to consider C. difficile. Also cost, the nuisance of non-specific GI side effects, and the potentiation of other medications, especially drug/drug interactions for Bactrim with Coumadin and oral hypoglycemics. They also didn’t see Stevens Johnson Syndrome, which is a rare but potentially life threatening complication of Bactrim. Antibiotic resistance is always a concern.

Maybe this is a time for shared decision making? Pretty simple talking points; 7% ARR for cure or recurrence, but baseline at least 74% cure. They didn’t look at subgroups...maybe push more for antibiotics in a patient with DM or with a large area of cellulitis? Could consider a “wait and see” prescription, although treatment failures may need additional I/D and should probably be re-seen.

NB, if prescribing antibiotics, remember they used 2 Bactrim DS bid for 7 days. Current IDSA guidelines say either 1 or 2 DS tabs bid is acceptable.

Show of hands? A minority of people in the room would take the Bactrim if they had an uncomplicated abscess. Interestingly, nobody in the room admitted to ever having a MRSA abscess....hmmmm.

Article 2: Martindale JL, et al: Diagnosing Acute Heart Failure in the Emergency Department: A Systematic Review and Meta-analysis. Acad Emerg Med 2016;23(3):223-42

It’s an age old question...how to accurately diagnose acute heart failure?

The ED diagnosis of AHF based on history, exam, CXR and ECG is discordant with the final discharge diagnosis in nearly one out of four cases.

In this review of 57 studies with 18,000 patients, 46% of whom were diagnosed with AHF, the authors examined the test characteristics of history and physical exam findings, ECG, CXR, BNP, Ultrasound, Echo, and Bioimpedence devices in patients presenting to the ED with dyspnea.

Not surprisingly, history and exam don’t do so well. There are many overlapping historical and exam features in our patients who often have multiple co-morbidities causing dyspnea. Unfortunately, authors also looked at each feature independently, instead of considering the power of gestalt of multiple history/exam variable. In any case, S3 has the highest LR (+) of 4, but this is found in only 13% of patients with AHF.

ECG: not helpful.

CXR: pulmonary edema on CXR has an LR (+) of 4.8, but CXR has a known poor sensitivity for AHF and a normal CXR does not help rule out CHF.

Echo: if reduced EF, echo can be helpful, except it doesn’t take into account 50% of CHF patients with preserved EF...also may be able to get the EF from recent echo/EMR, so may not add new information. Reduced EF as determined by visual estimation had a LR (+) of 4.1.

BNP was useful as a rule out, with a LR (-) of 0.11 when BNP was < 100 pg/ml. BNP had a LR (+) of 7 when BNP was > 1000 pg/ml, but values in between had unreliable LRs. Point made in the article that dichotomizing a continuous variable such as BNP is not the optimal use of the test. Also, most of the studies evaluating BNP excluded patients with renal failure, leading to fewer false-positives and inflating the test’s specificity.

There have been limited studies of Bioimpedence devices, with heterogeneity in the data analysis, but some initial promising results with segmental bioelectrical impedence analysis.

The overall winner....Lung Ultrasound with a LR (+) of 7.4 when B lines were visualized in multiple lung zones (hint, Mike Lambert likes using the windows at the medial costosternal articulations) and LR (-) of 0.16 when B-line pattern was absent.

Biggest limitation of this paper (beyond its mind numbing length)? Gold/criterion standard diagnosis of AHF is still clinical and consensus based, incorporating the clinical data and objective test results...leading to potential incorporation bias (study evaluates features that are also used to define the final diagnosis).

Bottom line: Use Ultrasound, to both rule in and rule out AHF. Gestalt history and physical may be helpful, but no individual H/P finding except for the elusive S3 is reliable. Very low or very high BNPs can help. Low EF on echo is useful, but misses 50% of patients with EF preserved AHF.

Article 3: Shokoohi H, et al: Bedside Ultrasound Reduces Diagnostic Uncertainty and Guides Resuscitation in Patients With Undifferentiated Hypotension. Crit Care Med 2015;43:2562-9.

In an ideal world, as the ultrasound probe becomes your sixth digit, and a rapid US protocol for the patient with undifferentiated shock will improve diagnostic accuracy and certainty.

In this single center study, attendings trained in critical care US or US fellows performed ultrasounds (4 view cardiac, 8 view lungs, IVC @ hepatic vein, & FAST/Aorta scan) on 118 adult ED patients with undifferentiated shock defined as SBP < 90 mm Hg after at least 1 liter NS IV. Patients with obvious sources of hypotension, or trauma-related hypotension were excluded. Clinicians caring for patients completed a pre/post US form indicating how the results of the protocolized US performed by an independent physician influenced their diagnostic certainty, diagnostic ability, and treatment/resource utilization. Primary outcome  was the change in the treating physician’s diagnostic certainty pre/post US and the concordance of post-US ED diagnosis with final chart review diagnosis.

There was a 28% decrease in diagnostic uncertainty after US, and an increase in the proportion of patients with a definitive diagnosis from 0.8% to 12.7%. Overall, the leading diagnosis after US demonstrated excellent concordance with the final diagnosis. A quarter of patients had significant changes in management after US, and significant changes in diagnostic imaging, consultation, and ED disposition. A subgroup of cases was described where ED US identified serious and time sensitive diagnoses (eg PE, ruptured splenic artery aneurysm, AAA).

Limitations: uncertain how much clinical information was available to physician prior to performing the scans, and this would potentially impact the post-test certainty of diagnosis. Likewise, initial resuscitation measures, including average amount of initial fluids given, was not included. How long do these scans take?? Lambert thinks maybe 10 minutes if you do it a bunch. At the same time, for many patients, after a targeted history, the complete protocol is probably not necessary, and scan time therefore decreases. This study didn’t evaluate the impact on clinical, patient oriented outcomes, including the impact of US on morbidity, mortality, ICU stay, etc. Finally, pre-US protocol diagnosis accuracy and post-US diagnosis accuracy were not directly compared, again likely secondary to lack of standardization of initial clinical information and initial resuscitation measures.

Bottom line: Kerwin-make US a habit. Barounis-it will make you a faster clinician. Team US in general: this is an important subgroup of patients, with potentially time sensitive diagnoses. US: do more, be better, save lives.