Table 2. The results of meta-analyses and sensitivity analyses.
| Analysis | No. ofstudies | Pooled hazardratio (95% CI) | I 2 statistic(%) | x 2 p-value forheterogeneity | p-value foroverall effect | Analyticalmodel |
| Primary analyses | ||||||
| OS | ||||||
| AFP-L3% elevation6–9,17,19–23,25,26 | 12 | 1.65 (1.45–1.89) | 18 | 0.26 | p<0.00001 | REM |
| DFS | ||||||
| AFP-L3% elevation7,8,10,18–20,24 | 7 | 1.80 (1.49–2.17) | 0 | 0.57 | p<0.00001 | REM |
| Sensitivity analyses | ||||||
| AFP-L3% elevation and OS | ||||||
| Exclusion of study with the largest effect size21 | 11 | 1.70 (1.50–1.94) | 0 | 0.45 | p<0.00001 | REM |
| Sample size ≥1007–9,17,19–21,23,25,26 | 10 | 1.57 (1.40–1.77) | 0 | 0.44 | p<0.00001 | REM |
| NOS scoring ≥66–9,19–23,25,26 | 11 | 1.62 (1.42–1.85) | 14 | 0.30 | p<0.00001 | REM |
| AFP-L3% elevation and DFS | Not applicable |
CI, confidence interval; REM, random-effect model; DFS, disease-free survival; OS, overall survival; AFP-L3, lens culinaris agglutinin-reactive fraction of α-fetoprotein; NOS, The Newcastle-Ottawa Scale Score.