4. Statistical Methods

All categorical variables are expressed as numbers (percentages), and all continuous variables are expressed as means ± standard deviations. All data analyses were conducted using NCSS 2021 software (NCSS, East Kaysville, UT, USA) with 2-sided tests and a type 1 error rate of 0.05. The learning curve was plotted according to TT and the number of procedures. The Curve Fitting-General package (which is a part of NCSS 2021) was used to fit the following mathematical model to the learning curve: Y= A× X^B

where Y = TT of each procedure, A = the TT taken for the first procedure (i.e., the first case), X= the procedure index, and B= the index of learning. The estimated B value (and its 95% confidence interval) was then used to calculate the estimated learning rate (LR) as per the following mathematical formula:

A general linear mixed-effect model was used to determine the effect of the fixed factors on each ROSA time interval. The fixed factors included in the model were stratified categories of clinical experience, sex, number of screws (4, 6, or 8), presence of scoliosis (yes or no), and T spine involvement (yes or no). Two interaction terms (number of screws× presence of scoliosis, T spine involvement × presence of scoliosis) were also included in the model. The age, body mass index (BMI), and BMI of each patient were measured as covariates, and the procedure index was measured as the random factor.