Predicting conversion in laparoscopic colorectal surgery

Publisher Website
Google Scholar
Abstract
Background: Laparoscopic colorectal surgery has clear advantages over open surgery; however, the effectiveness of the approach depends on the conversion rate. The objective of this work was to prospectively validate a model that would predict conversion in laparoscopic colorectal surgery. Methods: A simple clinical model for predicting conversion in laparoscopic colorectal surgery was previously developed based on a multivariable logistic regression analysis of 367 procedures. This model was applied prospectively to a follow-up group of 248 procedures by the same team, including 54 procedures performed by one new fellowship-trained surgeon. Results: Patients in the follow-up group were more likely to have cancer (56% vs 44%, p = 0.007) and were more obese (median, 71.0 vs 66.0 kg; p < 0.001). The rate of conversion in the follow-up group was unchanged (8.9% vs 9.0%, p > 0.05). Despite expected trends toward increasing risk of conversion with weight level (p > 0.05) and malignancy (10.1% vs 7.3%, p > 0.05), the model did not distinguish well between groups at risk for conversion. Contrary to the model, however, the fellowship-trained surgeon had a conversion rate that was not higher than that of the other, more experienced surgeons (7.3% vs 9.3%, p > 0.05) even though he was less experienced, and operating on patients who were more obese (median, 75.0 vs 70 kg; p = 0.02) and more likely to have cancer (59% vs 55%, p > 0.05). Recalculated conversion scores that excluded the inexperience point for the fellowship-trained surgeon showed a good fit for the model. Considering the original and follow-up experience together (615 cases), the model clearly stratifies patients into low (0 points), medium (1–2 points), and high risk (3–4 points) for conversion, with respective rates of 2.9%, 8.1%, and 20% (p = 0.001). Conclusion: This model appears to be a valid predictor of conversion to open surgery. Fellowship training may provide sufficient experience so that learning curve issues are redundant in early practice. This model now requires validation by other centers.