PERSystMA – Instructions

Welcome to the Meta-Analysis of Proportions App (PERSyst MA version 1.0). The app is based on the R package meta, version 7.0-0. This application is designed to facilitate meta-analysis of proportions through a user-friendly interface, making it accessible regardless of your statistical software proficiency. Here’s a step-by-step guide to help you get started.

 

1) Data Entry:

  • Upon launching the app, you will see two options for data entry: Manual and Upload CSV. Choose the most convenient option for your analysis.
    • Manual: Select this option to enter your study data manually. You will be able to input study names, sample sizes, event counts, and subgroup information (if applicable) directly into the interface.
    • Upload CSV: Select this option if you have a CSV file containing your study data. The CSV file should have the following columns: study name, sample size, event count, and subgroup (if any). Ensure there is no header in the file, and the columns follow this exact order, with values separated by commas.
  • Use the Add Row or Remove Row buttons to adjust the number of studies in your manual entry form if you’re entering data manually.
  • Be sure no cell for sample size or events is left blank, and the number of events is not higher than the sample size.

 

2) Settings:

  • Subgroup Analysis: Decide whether you want to perform a subgroup analysis. If “Yes” is selected, ensure your data (either entered manually or uploaded via CSV) includes subgroup information.
  • Synthesis Model: Select the appropriate synthesis model for your analysis. You can choose between common (fixed), random, or both (suggested only for educational purposes).
  • Method for the Meta-analysis: Select among Inverse of Variance (standard method using the inverse of variance for weighting studies) or Generalized Linear Mixed Model – GLMM (advanced method that uses a random intercept logistic regression model). If GLMM is used, heterogeneity will be quantified using maximum-likelihood as the variance estimator, and the meta-analysis will apply the logit transformation.
  • Transformation of Proportions: Select the appropriate transformation.
  • Heterogeneity Estimator: Choose the estimator for assessing heterogeneity among studies.
  • Prediction Interval: If using a random effects model, decide whether to include a prediction interval in your analysis. A prediction interval is estimated only if you have three or more studies.

 

3) Running the Analysis and Viewing the Results:

  • Click on Run to perform the meta-analysis based on your inputs and settings. If you update any parameter, make sure to click on Run again to update the meta-analysis results.
  • To adjust the forest plot, set the axis interval for your forest plot between 0 and 100%.
  • The results, including a forest plot and a summary of the meta-analysis findings, will be displayed in the main panel. To enhance reporting on the meta-analysis of proportions, below the forest plot, you will see a suggested text for the methods section.
  • Use the Download Forest Plot button to save the forest plot as a PDF vector file.