Skip to main content

Table 1 An example of the steps in a dynamic meta-analysis

From: Dynamic meta-analysis: a method of using global evidence for local decision making

Step Action Result
1. Meta-analysis of all studies Browse Metadataset by intervention and/or outcome, make selections, and click “Start your analysis” in the Shiny app Crop yield: 0% different after cover crops (non-significant)
2. Subgroup analysis of selected studies Select filters and then click “Update your analysis” Crop yield: 13% lower after cover crops (non-significant) (brassicas in the USA, with irrigated cash crops)
3. Meta-regression of all studies, with results for selected studies With the same selections, click “Meta-regression” Crop yield: 9% lower after cover crops (significant) (brassicas in the USA, but irrigation was not included in the best model)
4. Recalibration of selected studies Move the sliders on the tab for “Study summaries and weights” and then click “Update your analysis” Crop yield: 17% lower (significant) (brassicas in the USA, with irrigated cash crops, and with a relevance weight of 0.5 assigned to one study)
5. Sensitivity analysis Permute the settings (e.g. methods for handling missing data) and then compare the results Crop yield: significantly lower than 0%