Skip to main content

Table 5 Sensitivity and false discovery rate (FDR) for the likelihood ratio test (LRT), normal approximation, proportional reporting ratio (PRR), and reporting odds ratio methods while using the Benjamini-Hochberg (BH) method for adjustment of multiple testing under different parameter settings. The simulation takes counts from 250 random AEs from each \(n_i.\) size category and randomly assigns \(20\%\) of drugs from each AE to follow a proportion that differs from the null hypothesis proportion by a value of \(\Delta\)

From: Statistical methods for exploring spontaneous adverse event reporting databases for drug-host factor interactions

 

\(\boldsymbol{n}_{\boldsymbol{i}\cdot}\) size

\(\Delta\)

LRT

Normal approx.

PRR

ROR

Sensitivity

small

0.025

0.00821

0.0106

0.0297

0.0185

small

0.05

0.0146

0.0192

0.0449

0.0260

small

0.1

0.0501

0.0625

0.108

0.0700

small

0.2

0.244

0.278

0.351

0.305

medium

0.025

0.00598

0.00831

0.0238

0.0125

medium

0.05

0.0139

0.0194

0.0436

0.0199

medium

0.1

0.0621

0.0784

0.130

0.0718

medium

0.2

0.298

0.340

0.435

0.336

large

0.025

0.0031

0.00417

0.0156

0.00645

large

0.05

0.0143

0.0180

0.0401

0.0169

large

0.1

0.0759

0.0899

0.150

0.0828

large

0.2

0.315

0.356

0.465

0.358

FDR

small

0.025

0.0299

0.0345

0.110

0.0877

small

0.05

0.0299

0.0345

0.110

0.0882

small

0.1

0.0298

0.0348

0.106

0.0866

small

0.2

0.0303

0.0356

0.0978

0.0822

medium

0.025

0.0303

0.0338

0.129

0.103

medium

0.05

0.0302

0.0338

0.127

0.102

medium

0.1

0.0302

0.0342

0.121

0.0999

medium

0.2

0.0299

0.0348

0.103

0.0875

large

0.025

0.0383

0.0308

0.263

0.215

large

0.05

0.0357

0.0308

0.241

0.203

large

0.1

0.0317

0.0316

0.187

0.160

large

0.2

0.0304

0.0338

0.125

0.100