Statistics 25

Descriptive data are presented as mean ± standard deviations (SD) and analyzed by SPSS Statistics 25.0. Normality of all variables was tested by the Kolmogorov–Smirnov test. Independent t tests and chi-square analyses were conducted where applicable to compare the continuous variables and categorical variables at baseline, respectively. Differences of anthropometric and metabolic indicators before and after intervention in each group were analyzed via Paired t test. For comparison of percent changes in variables between MUO and MHO groups, analysis of covariance (ANCOVA) with metabolically healthy status as the between-subjects factors with inclusion of baseline level of dependent variable and other confounding variables as covariates. Interactions of sex and age with groups (MUO and MHO) were also considered. As no evidence of interactions were observed, the analysis was conducted using the whole sample. Significance was set at a 2-tailed p value < 0.05.

All strains with three biological replicates were used in this study. Statistical analysis was performed by one-way analysis of variance (ANOVA) using SPSS Statistics 25.0, and the data were displayed as mean ± SD (standard deviation) (n = 3). The difference was considered statistically significant when P < 0.05 and extremely significant when p < 0.01.

The fracture load values were imported into a statistics program (Statistics 25.0; SPSS, Stanford, CA). The Kolmogorov–Smirnov test was applied to test for a normal distribution. The fracture loads of individual materials and the standard deviations were calculated by descriptive statistics. One-way ANOVA followed by Scheffé’s post hoc test determined the relationship between the thickness and the fracture load. Regression analysis was used to determine the fit (coefficients of determination (R²)) for the linear, quadratic, and cubic curve shapes versus the material thickness for each material group. To calculate the regression curves of the three different materials, the fracture load at a material thickness of 0.0 mm was defined as an additional supporting point. This fracture load value was 0 N for all materials by definition.

SPSS Statistics 25.0 software was used to analyze all data. For data not normally distributed, the differences between two groups was analyzed using the Mann–Whitney U test, while multi-group analysis was analyzed using the Kruskal–Wallis H test. To evaluate the diagnostic value of CTCs, a receiver operating characteristic (ROC) curve was established. To assess the predictive power, the area under the curve (AUC) was analyzed. A two-tailed P value of < 0.05 was considered statistically significant. Graphs were plotted using GraphPad Prism 5.

All statistical analysis was carried out with the aid of SPSS Statistics 25.0. Continuous variables were expressed as mean ± standard deviation (SD) if data is normally distributed or as median (Quartile Deviation) if information is not consistent with the normal distribution. Categorical variables were present as frequencies (n) with percentages (%). For analyzing continuous variables, the Student' t-test was used when the normal distribution was conformed. Otherwise, the non-parametric Mann-Whitney U-test was applied if a skewed distribution was met. For categorical variables, chi-square or Fisher's exact-test was selected. The binary logistic regression analysis was employed for the univariable and multivariable analyses. When the efforts were made to construct the multivariable predictive model, all candidate variables derived from the univariable analysis (with a p-valve < 0.1) and those possible predictive variables were selected. Besides, the receiver operating curve (ROC) and the Hosmer-Lemeshow test were used to further evaluate the predictive model. A two-sided p-valve < 0.05 was considered statistically significant.

Data are given as boxplots. Statistical differences were analyzed by two-tailed Mann–Whitney U Test or paired t-test (SPSS Statistics 25.0 program). A value of p < 0.05 was regarded as significant.