Discussion

In our evaluation of observational studies, OAP reduces the risk of SSI after colorectal surgery by 50%. Combining OAP with MBP had a comparable effect on SSI risk. Although these findings seem conclusive, we must address several limitations. We included only 1 of the studies performed on the ACS-NSQIP database because we were unable to extract the proportion of unique participants across all publications, which inevitably reduced precision. Nevertheless, all studies reported a protective effect of OAP; therefore, we considered the direction of the effect reliable. The magnitude of the effect, however, could not be directly determined because of the limitations that apply to the ACS-NSQIP database, which we believe affected all the studies performed on these data. The database contains a limited number of variables, which likely hampered adequate adjustment for confounders. Thus, residual confounding cannot be excluded. Secondly, the grouping of participants may be unreliable because only MBP administered in the hospital was properly documented. This could imply the presence of misclassification bias when a part of the OAP only group did receive MBP at home. In addition, all studies excluded patients with data missing for the determinant, which may have introduced selection bias. Another issue is that the percentage of patients in the OAP only group was very low compared to the other preparation strategies. Albeit the aORs all demonstrate a protective effect of OAP, several studies clearly lacked power to determine the effectiveness of OAP without MBP. More importantly, these low numbers may also reflect the presence of confounding by indication. The choice of bowel preparation generally depends on surgeon’s preference and on patient’s prognosis. In most studies, a preference for combining OAP with MBP is seen. Not adding MBP to OAP could be because patients are unable to tolerate MBP because they are less fit, or that surgery was performed subacutely. In both cases, SSI risk was higher. This could have led to an underestimation of the effectiveness of OAP only, and it is also impossible to disentangle the impact of MBP when comparing OAP only with OAP and MBP combined because of unknown differences in patient characteristics that influence SSI risk.

That OAP is also effective without MBP was confirmed by a study that investigated OAP in a setting where MBP was not used. In this study, the risk of confounding by indication was present but negligible; OAP was implemented as standard of care.^{Reference Mulder, Crolla and Kluytmans-van den Bergh16} Although confounding due to residual time variation could not be completely excluded, a reduction in deep SSI and mortality of 42% was reported. Findings from the network meta-analysis also demonstrate that OAP can be administered without MBP. Although this conclusion was based on indirect associations, OAP alone appeared to be a better strategy than OAP with MBP in reducing organ-space infections. In contrast, a single-center RCT from Israel found no difference between OAP and OAP combined with MBP regarding SSI risk, suggesting that the MBP component can be safely omitted.^{Reference Zmora, Mahajna and Bar-Zakai19}

We also demonstrated that the impact of MBP with OAP is similar to that of OAP alone. Considering the absence of a beneficial effect of MBP alone, the only rationale for continuation of MBP in combination with OAP is because it was hypothesized that the antibiotics were not effective in an uncleansed colon. Based on our findings, we consider the added value of MBP to be questionable at best. This is relevant because, in addition to the higher costs, MBP not only poses a risk of electrolyte disturbance, its administration is also a significant burden to the patient. High-quality evidence is needed to confirm the efficacy of OAP without MBP. An RCT that includes an OAP-only arm and is powered to detect a 40% reduction in SSI risk may bring us closer to closing the research gap on the use of OAP and the necessity of MBP.