Package results of a method on the OHDSI Methods Benchmark

Stores the results of a method on the OHDSI Methods Benchmark in a standardized format, for example for use in the Method Evaluation Shiny app.

packageCustomBenchmarkResults(
  estimates,
  negativeControls,
  synthesisSummary,
  mdrr = NULL,
  analysisRef,
  databaseName,
  exportFolder
)

Arguments

estimates

A data frame containing the estimates. See details for required columns.

negativeControls

A data frame containing the negative controls. See details for required columns.

synthesisSummary

A data frame with the summary of the positive control synthesis as generated by the synthesizePositiveControls function.

mdrr

The MDRR as computed using the computeMdrr function. Should contain the following columns: "exposureId", "outcomeId", "mdrr". For comparative analyses, the "mdrr" column can be replaced by a "mdrrTarget" and "mdrrComparator" column.

analysisRef

A file describing the various analyses that were performed. See details for required columns.

databaseName

A character string to identify the database the method was executed on.

exportFolder

The folder where the output CSV files will written.

Details

The estimates argument should have the following columns: "targetId", "outcomeId", "analysisId", "logRr", "seLogRr", "ci95Lb", "ci95Ub". The negativeControls argument should have the following columns: "targetId", "outcomeId", "type". The analysisRef argument should have the following columns: "analysisId", "method", "comparative", "nesting", "firstExposureOnly" The targetId and outcomeId fields identify the specific control, and should correspond to those in the negativeControls and synthesisSummary objects. The type fields should be either "Outcome control" or "Exposure control". The analysisId field is an integer that identifies a specific variant of the method. For example, if the method is 'CohortMethod', analysisId = 1 could identify a set of settings using propensity score matching, and analysisId = 2 could identify a set of settings using stratification. logRr, seLogRr, ci95Lb, and ci95Ub correspond to the log of the effect estimate (e.g. the hazard ratio), the standard error, and the upper and lower bound of the effect size estimate, as produced by the method. method is a character string identifyign the method (e.g. "CohortMethod"). comparative is a boolean indicating whether the analysis can also be considerd to perform comparative effect estimation (comparing the target to the comparator). nesting is a boolean indicating whether the analysis is nested in the nesting cohorts identified in the gold standard. firstExposureOnly is a boolean indicating whether only the first exposure was used in the analysis.