Stata software version 14

Data from the trial will be analyzed using STATA software version 14.0 (Stata Corp, College Station, Texas, USA). All data will be checked for accuracy and a missing data analysis will be undertaken. The trial will adhere to the ‘CONSORT statement: extension to cluster randomized trials’ [34 (link)]. The data will be analyzed on an intention-to-treat basis, based on patient assignment.
A mixed linear regression model will be adopted to compare the significant differences between groups over a 12-month period. The mixed linear regression model will be adjusted by baseline value and potential confounding variables, for example, the sites and staff education levels [35 (link)]. Two-sided tests will be performed for all analyses and the level of significance will be set at P < 0.05. Where appropriate, 95% confidence interval will also be reported along with the P values.

All analyses were performed using Stata software version 14.0 (StataCorp LP). Two-tailed test with P < 0.05 was considered statistically significant. Predictors with more than 20% of the values missing were excluded from the primary analysis. To address the missing data, however, all other missing values were imputed with a multiple imputation technique^{33 (link),34 (link)}. For the predictors with less than 10% of missing values, the complete case was accepted and considered in the primary analysis^{35 (link)}. Nevertheless, the multiple imputation analysis was examined as a sensitivity analysis.

The inter-observer reproducibility of PD-L1 scoring was evaluated by kappa statistics. Simple and weighted kappa (κ) values were calculated, and agreement was described according to Landis et al [18 (link)] as moderate, substantial, and almost perfect for κ values of 0.41–0.60, 0.61–0.80, and 0.81–1, respectively.
The endpoint OS was defined as time from operation to death of any cause or last follow-up. RFS was defined as time from operation to death of any cause or recurrence of CC. Patients later diagnosed with another cancer were censored at the date of diagnosis (N = 102). The median age was used as cut-off to dichotomize the parameter age. Survival curves were generated according to the Kaplan-Meier method and the log-rank test was used to test for differences between groups. The multivariable Cox-regression model was used to test for independent prognostic value with hazard ratio (HR) of 1.0 as reference and a 95% confidence interval (CI). A cut-off significance level of 0.10 was pre-specified for a variable to be included in the multivariable Cox regression model.
Chi²-statistics were used to test associations between clinicopathological variables. A p-value < 0.05 was considered significant. The statistical analyses were performed using STATA software version 14.0 (StataCorp, Texas, USA), and all statistical tests were two-sided.

The 1957 referred patients from 11 primary care clinics were categorized as enrolled or not enrolled in the TX CORD secondary prevention study. The implementation index scores for the primary care clinics were converted into categorical variables with quartile splits collapsed into three categories: (a) relatively high level of implementation (upper quartile of the index), (b) medium level of implementation (middle two quartiles), and (c) relatively low level of implementation (lower quartile of the index). We then conducted a mixed effects logistic regression test with a random intercept term for primary care clinics to examine the association between implementation index levels and enrollment of children with overweight and obesity into the TX CORD secondary prevention study, separately for each year. Similar regression models were also used to examine the associations of underlying constructs, including acceptability, adoption, appropriateness, and feasibility with the enrollment of children into the TX CORD secondary prevention study. Significance was set at P value< 0.05. STATA software version 14.0 (College Station, TX) was used for data analysis.