3.1. OS, DFS, and DSS
We found no significant differences in OS (HR = 1.1226; 95% CI 0.9263; 1.3604), DFS (HR = 1.0797; 95% CI 0.8765; 1.3300), or DSS (HR = 0.8652; 95% CI 0.6531; 1.1462) between SNB and END. Heterogeneity was not detected in analyses of pooled OS, DFS, and DSS data (I2 < 50) (Figure 2). Neither the Egger nor Begg test revealed any publication bias in terms of OS (p = 0.8415) or DSS (p = 0.7608) (Figure 3A, 3B). Although mild bias was apparently present in the DFS for SNB (p = 0.004459; Figure 3C), the Duval and Tweedie trim and fill method revealed no significant difference between the observed and adjusted values (1.0797, p=0.4711 vs. 1.0390; p = 0.7086). Thus, we concluded that the DFS data were not biased. In the sensitivity analyses, the OS, DFS, and DSS data did not change on omission of any individual study (Figure 4).