1. Patients and ethical approval
Patients received CRC resection surgery at our institute form June 2015 -May 2019 were enrolled in this study. Data were retrospectively collected from a maintained, institutional review board-approved CRC database. The study was conducted ethically according to the Helsinki Declaration and the Ethics Committee of Xinhua Hospital approved this study (Approval No.XHEC-D-2020-0128).
2. Inclusion and exclusion criteria
The inclusion criteria were as follows: (1) patients aged more than 75 years old; (2) received selective CRC resection surgery. The exclusion criteria were the patients who underwent palliative surgery (colonic bypass or stoma only), multiple primary tumor, emergency surgery, aged less than 75 years old
3. Clinical evaluation and definition
Patient characteristics include age, gender, primary tumor site, surgical procedures such as laparoscopy, anastomosis and stoma, postoperative complications and postoperative hospital stay. The definition postoperative outcome was indicated that postoperative complications, postoperative hospital stay and postoperative mortality.
Postoperative complications were reviewed for type and scored using the Clavien-Dindo classification system(17): Grade 1, deviation from normal postoperative course without antibiotics or invasive intervention; Grade 2, pharmacological treatment; Grade 3, requiring surgical, endoscopic, or radiological intervention; Grade 4, life-threatening complications requiring ICU management; and Grade 5, death of a patient. In our study, surgery-related complications included postoperative ileus, SSI, anastomotic leakage, bleeding. The primary outcome was the incidence of postoperative complications.
4. Statistical analysis
Categorical and continuous variables were presented as frequencies and percentages, and median and standard deviations (SD), respectively. Unpaired Student’s t-test and Chi-squared test were used for the comparison of continuous and categorical variables.
Factors associated with postoperative complications were assessed by both univariate and multivariate logistic regression model. Results were expressed as a odds ratio (OR) with a 95% confidence interval. To assess factors associated with postoperative hospital stay, we used negative binomial regression model. As the variance of postoperative hospital stay (104.0464) was greater than the mean(14.8009). This meant over-dispersion existed. Therefore, a negative binomial regression was conducted. Results were expressed as an incidence-rate ratios (IRR) with a 95% confidence interval. IRR means incidence-rate ratios. In our negative binomial regression model of postoperative hospital stay, IRR revealed the altered proportion of postoperative hospital stay compared to the reference.
Akaike’s Information Criterion (AIC) and Bayesian Information Criterion (BIC) were calculated to validate multivariate logistic regression model and negative binomial regression model. The Hosmer-Lemeshow goodness-of-fit statistic was estimated after multivariate logistic regression models. To evaluate the goodness of fit of multivariate logistic regression models, calibration belt was plotted with its associated statistical test. To evaluate the goodness of fit of negative binomial regression models, chi-squared goodness-of-fit test was performed.
All statistical analyses and plotting were performed using STATA 14 software. A P-value < 0.05 was considered statistically significant in this study.