We generated network diagrams for different outcomes by the Stata software (version 15.1, Stata, Corp, College Station, TX), to elucidate the direct and indirect comparisons among different treatments in the included studies 9. Network meta-analyses of PFS and OS were conducted within a bayesian framework, which is more accurate than frequentist approaches 10, using Markov Chain Monte Carlo methods with the help of “gemtc” (version 0.8.4) and “rjags” (version 4.1.0) package of R-4.0.0 software. Hazard ratio (HR) and corresponding 95% credible interval (CrI) were used to assess the comparative efficacy between two treatments. The I2 statistic was used to demonstrate the heterogeneity of included studies, with I2 ≤50% denotes no or low heterogeneity and fixed effects model was applied, while I2 >50% indicates obvious heterogeneity and the random effects model was used. With three Markov chains, 250000 sample iterations were generated with 50000 burn-ins and a thinning interval of 1 in both PFS and OS analyses. We visually inspected the trace plot and density plot that showed the fit of the three chains to evaluate the convergence of iterations, and conformed to the Brooks-Gelman-Rubin diagnosis 11. The posterior ranking probability of each treatment was established by calculating the surface under the cumulative ranking (SUCRA) value, which equals 0 when an intervention is definite to be the worst, and larger value indicates higher likelihood of a given treatment being better 10. We assessed global inconsistency by comparing the fit of consistency and inconsistency models 12, and also applied the node-splitting method to detect the local inconsistency in any closed loops, with P < 0.05 denotes the existence of inconsistency between direct and indirect evidence 13, 14. We also performed sensitivity analyses by changing the effects model. Additionally, for studies that were not eligible for network meta-analysis, their data were summarized narratively using a qualitative data synthesis approach.

Note: The content above has been extracted from a research article, so it may not display correctly.

Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.

We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.