Background: Surgical outcomes are determined by complex interactions among a variety of factors including patient characteristics, diagnosis, and type of procedure.
Objective: The aim of this study was to prioritize the effect and relative importance of the surgeon (in terms of identity of a surgeon and surgeon volume), patient characteristics, and the intraoperative details on complications of colorectal surgery including readmission, reoperation, sepsis, anastomotic leak, small-bowel obstruction, surgical site infection, abscess, need for transfusion, and portal and deep vein thrombosis.
Design: This study uses a novel classification methodology to measure the influence of various risk factors on postoperative complications in a large outcomes database.
Methods: Using prospectively collected information from the departmental outcomes database from 2010 to 2011, we examined the records of 3552 patients who underwent colorectal surgery. Instead of traditional statistical methods, we used a family of 7000 bootstrap classification models to examine and quantify the impact of various factors on the most common serious surgical complications. For each complication, an ensemble of multivariate classification models was designed to determine the relative importance of potential factors that may influence outcomes of surgery. This is a new technique for analyzing outcomes data that produces more accurate results and a more reliable ranking of study variables in order of their importance in producing complications.
Patients: Patients who underwent colorectal surgery in 2010 and 2011 were included.
Settings: This study was conducted at a tertiary referral department at a major medical center.
Main outcome: Postoperative complications were the primary outcomes measured.
Results: Factors sorted themselves into 2 groups: a highly important group (operative time, BMI, age, identity of the surgeon, type of surgery) and a group of low importance (sex, comorbidity, laparoscopy, and emergency). ASA score and diagnosis were of intermediate importance. The outcomes most influenced by variations in the highly important factors included readmission, transfusion, surgical site infection, and abscesses.
Limitations: This study was limited by the use of data from a single tertiary referral department at a major medical center.
Conclusions: Body mass index, operative time, and the surgeon who performed the operation are the 3 most important factors influencing readmission rates, rates of transfusions, and surgical site infection. Identification of these contributing factors can help minimize complications.