e23012 Background: A meta-analysis is a formal, quantitative study design in epidemiology and clinical medicine that systematically assesses and consolidates previous research findings to derive comprehensive conclusions on a specific topic. Here, we aimed to establish a new tool for the visualization options useful for conducting a meta-analysis specifically designed for oncology trials. Methods: A meta-analysis can be based on binary data, continuous data, or time-to-event data. The most crucial statistical models, including the random effect model and the fixed effect model, are outlined alongside their respective statistical methods. Furthermore, we implemented graphical representations such as the forest plot, funnel plots, and Z-scope plots. While the forest plot effectively illustrates heterogeneity and pooled results, a funnel plot can reveal potential publication bias, and a Z-score plot shows the robustness of the used sample number. Results: The models and visualization options outlined have been integrated into a new, online web portal which can be accessed without the need for registration. The web application operates on an Ubuntu server running Apache and enables users to conduct meta-analyses using results typically used in oncology trials including binary (total and event numbers), continuous (mean and standard deviation data), and time-to-event data (hazard rate and CI data). Study results from common spreadsheet applications like Excel can be directly input into the system. The online platform utilizes the meta and metafor packages in the R programming environment (R version 4.2.2) for calculations and plots. The Shiny user interface is powered by shinyjs, shinydashboard, and rhandsontable R packages. The portal generates a forest plot to summarize meta-analysis results, a funnel plot for visual detection of potential bias, and a Z-score plot showing the sufficiency of the cumulative sample number. The analysis results of the tool were validated using IBM SPSS v28. The new bioinformatic tool is available at www.metaanalysisonline.com and requires no programming knowledge or command line use. Conclusions: Our platform aims to provide a user-friendly meta-analysis tool for epidemiological studies and oncology clinical trials, offering a swift and reproducible way to integrate the results from multiple studies.
Read full abstract