Why you should consider antibiotics for abscess, even without cellulitis.

Back on my general surgery clerkship, my very first time on the wards, my resident taught me about abscesses. He had a… unique style. “Ok, when I say abscess, you say ‘Gotta get the pus out.'” I laughed. “No seriously. Every time. If you hear the word abscess, you say it.” I laughed again. He was not joking.

Three years later, my attending asked me a question: “Do you think he could have an absc-”
“I uhhh… nothing. So call surgery?”

Point is, I and everyone else under the sun was taught that the management for abscess is drainage, and it’s never been clear for skin abscesses whether, in the absence of cellulitis, antibiotics are useful as adjunctive therapy. So today we’re going to review an RCT by Talan and colleagues of trimethoprim-sulfamethoxazole (TMP-SMX) vs placebo following incision and drainage of skin abscess.

Patients, Intervention, Comparator, and Outcomes

The trial enrolled patients 12 years or older with cutaneous abscess (diagnosed clinically) present for at least one week and at least 2cm in diameter whose clinicians intended to treat them as outpatients. Investeigators excluded patients with indwelling hardware, with suspected bone or joint involvement, with another indication for systemic antibiotics (diabetic foot, bite wound, etc.) or with immunosuppression including AIDS or neutropenia (HIV alone was not an exclusion criteria). Interestingly, IV drug use in the past month was a criterion for exclusion – we are left to guess at the rationale (likelihood of having taken antibiotics and recently and resultant resistance?), but it’s a shame to see a group at high risk for SSTI excluded from the trial. Patients with contraindications to TMP-SMX (G6PD deficiency, allergy, or GFR < 50) were obviously ineligible.

Patients in the intervention arm got four TMP-SMX single-strength pills twice daily, while the control arm got an equivalent number of placebos. All patients received incision and drainage (and all participating clinicians got trial-specific training on I&D in an attempt to ensure uniformity). Primary outcome was clinical cure at 14-21 days. Buried in a table is the definition of cure as the absence of failure – it’s a long list of possible failures, including increase in erythema at day 3-4, failure to get at least somewhat less tender and swollen by day 8-10, and failure to essentially resolve at day 14-21, all of which seem reasonable endpoints. Patients lost to follow up were considered to have treatment failure. Secondary outcomes include repeat surgical procedures, systemic infection (sepsis/bacteremia), and hospitalization.

A Bit of Context

Abscesses are incredibly common, and have become exponentially more so over recent decades. This is because of one bug, and the good news is that it’s one you already love to hate: MRSA.


Pallin et al., Ann Emerg Med. 2008;51:291-298.

Over 12 years, we have gone from 1.1 to 3.5 million ED presentations for skin and soft tissue infection, which has tracked directly with the rise of community-acquired MRSA (CA-MRSA). The majority of the increase has been in abscess, while cellulitis has remained largely consistent. The above graph uses data about antibiotic prescription to track MRSA incidence, mostly because that’s much easier to get than actual microbiologic data. Still, the microbiology has also been done for those of a mind to read a study with phrases like “Pulsed-field type USA300 isolates accounted for 97 percent of MRSA isolates.” For the rest of you, 11 university hospital-associated EDs had an overall MRSA prevalence of 59% in their incident skin and soft tissue infections, which I would describe as “moderately horrifying.”

There are a few prior studies exploring the question of antibiotics status post incision and drainage of abscess. Most of these are pretty limited, with a few retrospective trials and a slew of single-center, small RCTs. Some used beta lactam therapy, which we just established will fail to cover our most important pathogen. While we’re on the subject though, it’s important to note that TMP-SMX has excellent CA-MRSA coverage (not so hot for the hospital-acquired stuff), but actually has relatively poor strep coverage, so keep that in mind if your patient really does have a cellulitis that could be strep. Consider clindamycin to split the difference.

The most sophisticated prior study is a multi-center RCT by Schmitz and colleagues looking at the exact same intervention, TMP-SMX versus placebo status post incision and drainage. As you might imagine from the fact that we’re still asking this question, it’s a negative trial. We’ll talk more about why that is once we finish with the study currently in question.


1265 patients were enrolled, 98.6% of whom were randomized. 64.7% of those patients were 100% adherent to therapy, similar between the two groups. Median age was 35. 11% of patients were diabetic. Race was not reported. All characteristics were similar in both groups. Abscesses were on the whole small, with greatest dimension measuring just 2.5cm on average on ultrasound, with 6.5cm average greatest dimension of erythema.

It’s a positive trial – abscess cure rate at the 14-21 day visits was 80.5% for TMP-SMX vs 73.6% for placebo in modified intention-to-treat analysis (6.9% difference; 95% confidence interval [CI], 2.1% to 11.7%; P=0.005). This bumps to 9.8% difference if you assume that those who did not follow up had cure rather than treatment failure (not an unreasonable assumption if they did not present for follow up visits, but good to see the study holds up with the assumption that lost equates failure). Only four patients, two in each group, developed locally invasive infection. The treatment group also had a 5.2% decrease in subsequent surgical drainage and 7.2% reduction in infections at a new site, both significant.

Importantly, despite a slightly higher risk of GI side effects in the treatment group compared to the placebo group (42.7% vs 36.1%), rates of discontinuation due to adverse effects were similar (1.9% vs 0.6%), and AKI and hyperkalemia were minimal in both groups. While 2.5% of patients on TMP-SMX got a rash (a non-significant increase from placebo), no cases of Stevens-Johnson were reported. No C difficile colitis was reported.

So on the whole, a modest but present increased rate of cure, number needed to treat of 14, with no clear trade off in terms of adverse events. Plus, TMP-SMX is dirt cheap. Granted cure rates are > 80% from I&D alone, but when you have a cheap and low-risk intervention that can avoid the need for a repeat procedure, it’s a pretty easy sell. The trial itself is pretty unimpeachable, though the authors note it’s not unreasonable to wonder if sicker patients that treating physicians may have felt to be too high-risk to enroll are not represented here.

One burning question remains for me even after all of the above is ironed out. Why do we have a significant result in this trial when the Schmitz trial failed to find a significant difference in treatment failure? Well, first we should note that there was a non-significant trend towards effectiveness of TMP-SMX in that trial, P = 0.12. The rest of the picture gets much clearer when we look at the power calculations of the two trials (stay with me – I promise I’m not getting TOO far into the weeds with stats here). Schmitz et al. powered their study to detect a 15% difference between the groups with 80% power – in short, they were completely underpowered to pick up the small 6.9% change in clinical cure rate (9.6% relative risk reduction) that the Talan paper reports. So these results are not clearly at odds with prior negative trials, but are rather a reflection of the fact that there are nearly six times as many patients in this trial.

Bottom Line

TMP-SMX is superior to placebo for skin abscess >2cm, with increased rates of clinical cure at day 14-21 in treated patients, NNT ~14. It does not appear to cause more adverse effects compared with placebo.

One thought on “Why you should consider antibiotics for abscess, even without cellulitis.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s