The review included 7 RCTs involving 1,454 participants in total.
Two of the 7 studies were judged to be of a poor methodological quality (each scoring 2 points).
Anastomotic leakage (7 RCTs).
There was a statistically significant higher rate of leakage in the bowel preparation group than in the no-preparation group (OR 1.74, 95% CI: 1.05, 2.90). There was no evidence of heterogeneity (Q=4.41). When the poor-quality RCTs were removed from the analysis, the results still favoured no preparation but were no longer statistically significant (OR 1.55, 95% CI: 0.82, 2.92). A subgroup analysis of the 4 RCTs using PEG for bowel preparation found a significantly higher rate of leakage in comparison with no preparation (OR 1.92, 95% CI: 1.03, 3.60), whereas the difference between preparation and no preparation in studies using other methods was not statistically significant (OR 1.43, 95% CI: 0.59, 3.45). The funnel plot for anastomotic leakage was symmetrical, which suggests no evidence of publication bias.
Other outcomes.
The bowel preparation group had higher rates of wound infection (OR 1.33, 95% CI: 0.88, 2.30; 7 RCTs), other extra-abdominal septic complications (OR 1.12, 95% CI: 0.69, 1.83; 3 RCTs), post-operative mortality, reoperation and non-septic complications, but none of these differences were statistically significant. These results were not substantially altered by the removal of the poor-quality RCTs.