Risk of bias
Risk of bias within studies
The risk of bias within the five randomized controlled trials was evaluated (Supplementary File 7 ) by the bias risk assessment tool provided in the Cochrane Evaluation Manual, and 30 items were defined as low risk, 11 items were unclear, and only eight items were defined as high risk, while the risk of bias within the nine single-arm studies could not be evaluated. All the studies were open-label trials because of the inevitable occurrences of treatment-related adverse events.
Publication bias
Publication bias of OS, ORR, DCR, clinical characteristics subgroups analyses of OS with HPV positive or HPV negative, any grade and grade ≥3 AE calculated by Egger’s test in Stata 16 software were 0.384, 0.679, 0.248, 0.849, 0.965, 0.704, 0.667, respectively. Publication bias of any grade AEs of PD-1/PD-L1 inhibitors, any grade AEs of PD-1 inhibitors, grade ≥3 AEs of PD-1/PD-L1 inhibitors, grade ≥3 AEs of PD-1 inhibitors were calculated by Egger’s Test, as shown in Supplementary File 8 .
The quality of evidence
The quality of evidence from 27 outcomes across studies was evaluated as high, 21outcomes’ were moderate by GRADE system (20), which were shown in Supplementary File 9 .