Spoiler alert: The study was negative. 13.2% of patients in the cefazolin arm and 14.9% in the placebo arm developed SSI, (absolute risk difference, −1.7 [95% CI, −8.0 to 4.6], P=0.60). Thus, with a negative study, I headed to the sample size calculations section where the authors stated that they powered the study with an estimated SSI rate of 3.3% in the cefazolin arm and 10% in the placebo arm based on the SSI rates in clean-contamined procedures and recent Dutch retrospective studies, respectively.
First, this is an underpowered study. It's true that they hit their target sample size, but their estimates were completed using rosy expectations for the benefits of cefazolin and were not selected based on clinically meaningful reductions in SSI. Many might think a 2% absolute reduction in SSI is clinically meaningful. Second, the authors suggest that the high rates of SSI might have resulted from very low thresholds for starting antibiotics if there was “the slightest suspicion of a SSI.” Since CDC definitions define SSI based on receipt of antibiotic treatment, this behavior could have biased the rates. Finally, deep SSI rates were 0.4% (1 patient) in the cefazolin group and a far higher 2.9% (7 patients) in the placebo group, (absolute risk difference, −2.5 [95% CI, −5.7 to 0.4]). Since the study was powered for the primary outcome, all SSI, not much was made of the big difference in deep SSI.
The end result might be disappointing but this is no discredit to the authors and clinicians behind the study. Randomized, controlled trials are very hard to design and implement, so thumbs up for all their efforts. With that said, I would prescribe cefazolin as a peri-operative antibiotic during implant removal below the knee since a 2% absolute reduction in all SSI and a 2.5% reduction in deep SSI are both clinically meaningful benefits. I will then wait for another larger study powered using clinically meaningful SSI targets, including deep SSI.
