Methodological considerations in interpreting postoperative antibiotic prophylaxis after clean thyroid and breast surgery

Sir,

The article by Imran et al. ¹ addresses an important antimicrobial stewardship question in a resource-constrained surgical setting: whether postoperative antibiotics reduce surgical site infections after clean thyroid and breast operations. The study is timely and clinically relevant, particularly because unnecessary antibiotic use remains common despite global concerns regarding antimicrobial resistance. Its randomised design, complete 30-day follow-up, and use of standard surgical site infection definitions are notable strengths.

However, several methodological issues merit consideration when interpreting the findings. The trial is described as a non-inferiority study, but the primary result is reported as a relative risk with a two-sided p-value. For a non-inferiority design, interpretation should centre on the absolute risk difference and its confidence interval in relation to the pre-specified non-inferiority margin. ² In the trial mentioned, surgical site infection occurred in 10 of 109 patients in the postoperative antibiotic group and 9 of 109 in the no postoperative antibiotic group. Although the point estimate suggests little difference, the manuscript does not report whether the upper confidence limit for the absolute excess risk remained within the stated 2% margin. Therefore, the conclusion that omission of postoperative antibiotics does not increase infection risk may require more cautious phrasing.

The sample-size assumptions also appear difficult to reconcile with the stated objective. The authors used expected infection rates of 1% with antibiotics and 9% without antibiotics, while adopting a 2% non-inferiority margin. Such assumptions imply a larger expected excess risk than the margin considered clinically acceptable. Clear justification of the chosen margin, based on clinical acceptability and prior evidence, is central to non-inferiority trial validity.^2,3

Additional reporting details would strengthen interpretability. The pre-incision antibiotic regimen included either ceftriaxone or cefazoline, but allocation, balance between groups, and redosing criteria were not described. The postoperative oral antibiotic regimen was also insufficiently specified. Because thyroid and breast procedures have different baseline infection risks, pooled analysis without reported procedure-specific outcomes may obscure heterogeneity. Similarly, drain use was mentioned but not analysed, despite being a common reason for prolonged prophylaxis. Finally, absence of blinding, particularly for outcome assessment, could influence diagnosis of superficial infection, where clinical judgement is often required. ⁴

These concerns do not negate the clinical importance of the study. Rather, they suggest that its findings should be interpreted as supportive evidence against routine postoperative antibiotics, while recognising uncertainty around formal non-inferiority, intervention standardisation, and subgroup applicability. Transparent reporting of absolute effect estimates, per-protocol sensitivity analysis, antibiotic exposure details, drain-stratified outcomes, and procedure-specific results would further strengthen the manuscript's contribution to surgical antimicrobial stewardship.