Stata software 14

The combined odds ratios (ORs) and 95% confidence interval (CI) were used to evaluate the strength of the association with the risk of leukemia. Pooled ORs were performed for allelic comparison (a vs. A), dominant (aa + Aa vs. AA), recessive (aa vs. Aa + AA) and codominant (aa vs. AA and Aa vs. AA) models (“a” and “A” represents the mutant allele and the wild-type allele, respectively). Heterogeneity among the included studies was assessed using the I² statistic. A random-effects model or fixed-effects model was used to calculate the pooled OR in the presence or absence of heterogeneity, respectively. To detect possible sources of heterogeneity and potential differences among subgroups, meta-regression and subgroup analyses were carried out with the stratification of different ethnicities, age groups and subtypes of leukemia. The significance of the pooled OR was determined through a Z-test, and p < 0.05 was considered to be statistically significant. Publication bias was investigated using funnel plots and Egger’s regression test. We also conducted sensitivity analysis to test the robustness of associations by sequentially omitting each of the included studies one at a time. All the data analysis was performed using STATA software 14.0 (StataCorp, College Station, TX, United States) and Review Manager 5.3 (Cochrane Collaboration, Oxford, United Kingdom).

Statistical analysis was performed using Stata Software 14.0 (Stata, College Station, TX). Pooled HRs (high/low) and their associated 95% CIs were used to analyze the prognostic role of CXCL5 expression in various cancers. The heterogeneity among studies was evaluated using Cochran’s Q and I² statistics. A p value less than 0.10 or an I² value larger than 50% were considered statistically significant. The fixed-effect model was used for analysis without significant heterogeneity between studies (p > 0.10, I² < 50%). Otherwise, the random-effect model was chosen. To explore the source of heterogeneity, subgroup analysis and meta-regression were preformed through classifying the included studies into subgroups according to similar features. We also conducted sensitivity analysis to test the effect of each study on the overall pooled results. The publication bias was evaluated by using both Begg’s test and Egger’s test. A p value less than 0.05 was considered statistically significant.

We expressed continuous and categorical variables as mean (SD) and percentages (%). Comparisons were made between frequencies and means using t-tests and Chi-square tests. Multiple mixed-effect linear regression was used to examine the association between blood glucose levels and lipids and lipid ratio parameters. In the multiple mixed-effect linear regression model, the independents and dependent variables have a linear relationship, and the dependent variable must be continuous and at least interval-scale. Besides, the residuals follow the normal probability distribution and are independent. In this study, three separate models were used to control for each covariate: 1) adjusted for the demographic variables, 2) adjusted for demographics and health-related behaviors, and 3) adjusted for demographics, health-related behaviors, and physical health. Gender was fitted as a random intercept model. The regression models were shown with beta coefficients and 95% confidence intervals. All the analyses were performed using STATA software 14.0.

We used the Stata software 14.0 for data analysis. Descriptive analyses were conducted to examine the differences in social network characteristics between patients with controlled/uncontrolled hypertension and between patients with behavioral adherence/nonadherence. Chi-squared tests and Mann–Whitney tests were performed as appropriate. Multivariate logistic and tobit regression models were used to explore the association of network variables with uncontrolled hypertension (yes/no) and behavioral adherence score. In this study, we conducted four models for each dependent variable. Model 1 was adjusted for age and sex, while model 2 was adjusted for age, sex, and quality of life (PCS-12 and MCS-12). Model 3 was additionally adjusted for body mass index and number of comorbidities, while model 4 was additionally adjusted for smoking and alcohol use. Multicollinearity was checked by using variance inflation factor (VIF). VIF results showed no multicollinearity among variables. Akaike information criterion (AIC) and Bayesian information criterion (BIC) were then calculated to examine the sensitivity of the models. The level of statistical significance was set at the 5% level.

We analyzed the data using Stata software 14.0 (StataCorp, TX, USA, 2015). We first present the sample distribution of adolescent boys’ sociodemographic characteristics and bullying scores, gender role attitudes, and perceived friend support by gender groups (see Table 1 and Table 2). Second, three regression models were used to examine the relationship between gender attitudes and bullying and whether perceived friend support mediated the relationship (Table 3). Since bullying is a dichotomous variable and friend support is an ordinal variable, this study used binary logistic regression models and an ordinal logistic regression model to predict bullying and friend support, respectively. Third, the Sobel–Goodman test was used to examine the mediating effect of perceived friend support.