Skip to contents

Perform Disproportionality Analysis

Source: R/disproportionality_analysis.R
disproportionality_analysis.Rd

performs disproportionality analysis and returns the results, which can be plotted using render_forest(). Look at the DiAna website fore extensive tutorials to practically use the function (vignette("Disproportionality Analysis", package = "DiAna")) and to design and run disproportionality analysis based on expected biases (vignette("Causal Inference", package = "DiAna")) Contrary to disproportionality_comparison(), it calculates the association based on the Drug and Reac database, and not based on number of reports. See disproportionality_trend() for investigating how disproportionality changed through time.

Usage

disproportionality_analysis(
  drug_selected,
  reac_selected,
  temp_drug = Drug,
  temp_reac = Reac,
  meddra_level = "pt",
  drug_level = "substance",
  restriction = "none",
  minimum_cases = 3,
  frequentist_threshold = 1,
  log2_threshold = 0,
  multiple_comparison = TRUE,
  store_pids = FALSE,
  save_in_excel = FALSE,
  file_name = "disproportionality_results"
)

Arguments

drug_selected

A list of drugs for analysis. Can be a list of lists (to collapse terms together).

reac_selected

A list of adverse events for analysis. Can be a list of lists (to collapse terms together).

temp_drug

Drug dataset. Can be set to sample_Drug for testing

temp_reac

Reac dataset. Can be set to sample_Reac for testing

meddra_level

The desired MedDRA level for analysis (default is "pt").

drug_level

The desired drug level for analysis (default is "substance"). If set to "custom" allows a list of lists for reac_selected (collapsing multiple terms).

restriction

Primary IDs to consider for analysis (default is "none", which includes the entire population). If set to Demo[!RB_duplicates_only_susp]$primaryid, for example, allows to exclude duplicates according to one of the deduplication algorithms.

minimum_cases

Threshold of minimum cases for calculating identifyin a signal (default is 3).

frequentist_threshold

Threshold for defining the significance of the lower limit of the Reporting Odds Ratio (default is 1).

log2_threshold

Threshold for defining the significance of the lower limit of the Information Component (default is 0).

multiple_comparison

Logical specifying whether to perform Bonferroni correction for multiple testing on the ROR. Default to TRUE. Particularly important when running the disproportionality on many combinations.

store_pids

Logical specifying whether to store primaryids recording the drug and primaryids recording the event as lists. Default to FALSE.

save_in_excel

Whether to save the outcome in an excel. Defaults to TRUE

file_name

The name of the Excel file to save the results. Default is "Descriptives.xlsx". It only works if save_in_excel is TRUE.

Value

A data.table containing disproportionality analysis results.

See also

Other disproportionality functions: disproportionality_comparison(), disproportionality_trend()

Examples

disproportionality_analysis(
  drug_selected = "paracetamol",
  reac_selected = "overdose",
  temp_drug = sample_Drug,
  temp_reac = sample_Reac
)
#>      substance    event   D_E  D_nE     D  nD_E     E nD_nE ROR_median
#>         <fctr>    <ord> <num> <num> <num> <num> <num> <num>      <num>
#> 1: paracetamol overdose     5    46    51    10    15   939      10.14
#>    ROR_lower ROR_upper IC_median IC_lower IC_upper              label_ROR
#>        <num>     <num>     <num>    <num>    <num>                 <char>
#> 1:      2.61     34.19      2.12     0.55      3.1 10.14 (2.61-34.19) [5]
#>               label_IC Bonferroni ROR_signal IC_signal
#>                 <char>     <lgcl>      <ord>     <ord>
#> 1: 2.12 (0.55-3.1) [5]       TRUE        SDR       SDR