Statistical software release 15

All analyses were conducted using Stata Statistical software release 15^{(44 )}. Bivariate associations between knowledge or beliefs and demographics were tested using chi-square, and associations between knowledge or beliefs and consumption variables (regular soda and diet drink consumption; high v. low consumers for each beverage) were tested via a series of multilevel logistic regression models with random effect for school to adjust standard errors for correlations within and between schools. Models also adjusted for school type, state/territory, sex, age and level of disadvantage. OR were reported to describe the odds of being a high soda or diet drink consumer according to knowledge and beliefs regarding these beverages. The primary interest was the associations between knowledge (soda) or beliefs (diet soda) regarding, the health effects of a beverage and consumption of the same type of beverage. However, for completeness we also looked at associations between knowledge of the health effects of consuming soda with diet drink consumption, and vice-a-versa. The number of analyses undertaken and the large data set inflated the risk of Type I error, as such, we employed a conservative probability level (P < 0·001)^{(45 ,46 )}.

The study collected data from the patient information system and utilized the “encode” command in Stata Statistical Software Release 15 to pseudonymize the data. Based on previous studies and a literature review, clinical variables were identified as potentially significant determinants of clinical outcomes (target variables considered: hospital length of stay [LOS], operation time, and complications). These clinical feature variables included surgery technique, ASA physical status classification, demographic information, and preoperative C-reactive protein levels. The study also compiled the names of surgeons who performed the surgeries during the study period, with a focus on evaluating their years of experience and the number of surgeries they had performed with the respective surgical technique at the time of surgery. Surgeons were grouped based on the number of cases they had performed, with five surgeons performing the majority of the surgeries and others grouped together if they had performed less than ten cases. Patient outcomes were tracked for durations ranging up to one year post-surgery. For our cohort, the average follow-up time was recorded at 7 ± 4 months.