Statistical packages for social sciences version 22

The Statistical Packages for Social Sciences version 22.0 (IBM, Armonk, NY) were used in data analysis. In current study, receiver operating characteristics (ROC) curve was performed to calculate the most appropriate cutoff point for SLDH level. χ² test (or Fischer's exact test, if indicated) was used to test the baseline balance between low SLDH and high SLDH groups. Actuarial survival rates for OS, PFS and DMFS were obtained using Kaplan-Meier method. In addition, differences in survival between groups were calculated using log-rank test. Univariate and multivariate analysis were performed using Cox proportional hazards model (multivariate analysis consisted of variables with P value <0.05 in univariate analysis). Proportional hazard assumption was assessed using log-minus-log plots. Any result with two-sided P value <0.05 was considered statistically significant.
In addition, propensity score matching was performed using R version 3.4.0 (The R Foundation of Statistical Computing, Vienna, Austria) to adjust for bias and confounding. Variables used to calculate the propensity score index were: age, sex, tumor classification, nodal classification, TNM staging and radiation dose. This was performed using nearest neighbor 1:1 matching in MatchIt package.

The Statistical Packages for Social Sciences version 22.0 (IBM, Armonk, NY) was used for all statistical analysis. Receiver operating characteristics (ROC) curve was performed to determine the most appropriate threshold value for PNI and SII. χ² test (or Fisher’s exact test, if indicated) was used to explore the baseline balance between low PNI/SII and high PNI/SII groups. Correlation between variables was assessed using Pearson correlation coefficient. Actuarial rates for OS, PFS, and DMFS were generated using Kaplan–Meier method. Furthermore, differences between curves were analyzed with log-rank test. Univariate and multivariate analysis were calculated using Cox proportional hazards model (multivariate analysis consisted of variables with P value < 0.05 in univariate analysis). Proportional hazard assumption was explored using log-minus-log plots. A two-tailed P value < 0.05 was considered statistically significant.
In addition, PSM and correlogram was done using R version 3.4.0 (The R Foundation of Statistical Computing, Vienna, Austria). PSM was carried out using nearest neighbor 1-to-1 matching in MatchIt package and correlogram was created using corrplot package.