Meta-analysis of oral antibiotics, in combination with preoperative intravenous antibiotics and mechanical bowel preparation the day before surgery, compared with intravenous antibiotics and mechanical bowel preparation alone to reduce surgical-site infections in elective colorectal surgery (BJS Open 2018; 2: 185-194).

We read with great interest the systematic review and meta‐analysis by McSorley et al.1 regarding the current controversy regarding the usefulness of mechanical bowel preparation before elective colorectal surgery. By pooling 22 studies totalling 57 207 patients, the authors demonstrated that preoperative oral antibiotics, in combination with intravenous antibiotics and mechanical bowel preparation, were associated with signicantly lower rates of surgical‐site infection than intravenous antibiotics plus mechanical bowel preparation.

In their quantitative analysis, the authors pooled 14 RCTs involving 3014 patients, but also eight cohort studies with a total of 54 193 patients. Among the latter, the authors included the following studies: the study of Morris and colleagues2, comprising 8415 patients from the National Surgical Quality Improvement Program (NSQIP) colectomy cohort from 2011 to 2012; the study of Moghadamyeghaneh and co‐workers3, comprising 5021 patients from NSQIP from 2012 to 2013; the study of Kiran et al.4 , comprising 8442 patients from NSQIP from 2012; the study of Koller and colleagues5, comprising 32 359 patients from NSQIP from 2012 to 2014; and the study of Scarborough and co‐workers6, comprising 4999 patients from NSQIP from 2012.

In addition, McSorley et al. included the study of Hendren and colleagues7, which included 4331 patients from the Michigan Surgical Quality Collaborative (MSQC). However, we note that the list of participating hospitals in NSQIP in Michigan (https://www.facs.org/search/nsqip-participants?state=MI) includes hospitals listed by the MSQC (https://msqc.org/about/member-hospitals/). Therefore, it is likely that patients reported by the MSQC are the same as those reported in the NSQIP. Only the studies of Konishi and co‐workers8 (556 patients) and Cannon et al.9 (9940 patients from the Veterans Affairs Surgical Quality Improvement Program) might not have been duplicates.

Therefore, we raise the concern that McSorley and colleagues pooled duplicate patients in their quantitative analysis, at least for the period after 2012, resulting in an increase in the weight of the NSQIP cohort for the investigated outcomes. Duplicate data should be avoided in meta‐analyses10, 11. However, it appears that the overall conclusions drawn by McSorley et al. were not affected by this methodological pitfall, as the findings of the subgroup analyses including only RCTs showed similar trends.

In conclusion, the effect of antibiotics and mechanical bowel preparation on surgical‐site infection during elective colorectal surgery cannot be assessed properly by analysis of the pooled data originating from the included cohort studies.