Skin Antisepsis: When New Evidence Emerges, Reevaluate Your Practice

Commentary We thank Dr. Datti for drawing attention to the PREPARE (A Pragmatic Randomized Trial Evaluating Preoperative Alcohol Skin Solutions in Fractured Extremities) trial results published in The New England Journal of Medicine (NEJM). We categorically disagree with his assertions: The current 90-day surveillance period for the primary outcome was appropriate because the rationale for skin antisepsis preventing infections that present months later is mechanistically questionable, and prolonged surveillance would have been susceptible to the competing risk of other exogenous causes. Sample-size estimates are designed to balance the risk of type-I and II errors. Although we observed a significant effect sizesmaller than hypothesized (odds ratio [OR], 0.76 instead of 0.64), our estimate was based on the best available evidence2,3. Similarly, the primary statistical approach used null-hypothesis testing with significance set at p < 0.05. The p value for the primary comparison was below this threshold, and the null hypothesis was rejected. NEJM copy editing rounded the confidence interval (CI) to 1.00. As indicated by Dr. Datti, the Appendix of the published study1 reports the CI with greater precision and confirms it was <1.000. The interpretation of the fragility index is incorrect and dangerous. In this cohort of 6,785 patients with closed fractures, 185 patients experienced a surgical site infection (SSI). It is true that a few more infections in the iodine group would alter the statistical conclusion; however, it must be noted that an appropriately designed trial intentionally recruits just enough patients to observe an effect that can reject the null hypothesis (according to the sample-size estimate). It should also be noted that the fragility index is designed for use with a Fisher exact test. It cannot be applied to the estimates obtained from the multilevel statistical model used to account for the correlation between the recruitment clusters and the alternating treatment periods. In addition, a properly conducted trial should never have a large fragility index because it would be unethical to keep randomizing patients to receive an inferior treatment beyond the necessary confidence to reject the null hypothesis. We have referenced a few elegant editorials on this topic4,5. A subgroup effect assesses whether a treatment works better or worse in certain subpopulations. We used the ICEMAN (Instrument for assessing the Credibility of Effect Modification Analyses) criteria for considering potential subgroup effects6. Bacterial pathogens (an outcome) and surgical trauma (an intervention) occur downstream from the exposure (antisepsis); these are not baseline characteristics that meet credible subgroup criteria. The assertion that the magnitude of benefit should increase with the severity of the clinical condition is flawed. It is unclear why one would assume that skin antisepsis is expected to be more effective in a highly contaminated open fracture with a diverse and high concentration of bacteria compared with a closed-fracture surgical site with only skin flora. Yet, the estimated absolute SSI reductions in the closed and fracture cohorts were similar (0.8% and 0.9%, respectively); this similar effect did not reach significance in the open fracture group because the baseline risk was higher—presumably from the environmental contamination that allowed bacteria to infiltrate the wound hours prior to the skin antisepsis. Finally, we agree that clinical medicine is a probabilistic science. We also performed a Bayesian analysis, which was reported in the Appendix of the published study1. Using a neutral moderate prior that assumes no treatment benefit from iodine povacrylex, our trial data suggest a 97% probability of any treatment benefit (OR, <1.0) for patients with closed fracture and a 74% probability for patients with open fracture. These probabilities of benefit far exceed the theoretical concern for iodine-related thyroid toxicity in adult patients or the case reports of life-threatening anaphylaxis from chlorhexidine antisepsis7. We conclude by inviting readers to reflect on the clinical decision. Skin antisepsis is a mandatory step prior to surgical incision. The surgeon must use a solution, and in most hospitals in North America, alcohol-based solutions of chlorhexidine gluconate and iodine povacrylex are shelved next to each other. The availability, ease of use, and cost are essentially identical. Yet, the surgeon must still choose one. The PREPARE trial provides the best available evidence to guide this choice for orthopaedic surgeons. The results suggest a clinically important benefit to skin antisepsis with iodine povacrylex in alcohol over chlorhexidine gluconate in alcohol for closed fractures, and no harm, with potential benefit, for open fractures. When new data emerge, we reevaluate our practice—and we hope that you do, too.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.