Skip to main content

Table 4 Empirical family-wise error rate (FWER) for the likelihood ratio test (LRT), normal approximation, proportional reporting ratio (PRR), and reporting odds ratio while using either Max-Stat or Benjamini-Hochberg (BH) method for adjustment of multiple testing when all null hypotheses are true. The inference methods were applied at \(\alpha =0.05\). The result is based on 500,000 simulation runs based on the FAERS dataset regarding sex for different sample sizes when there is no difference between host factor groups (\(\Delta =0\))

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

Method

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

LRT

Normal approx.

PRR

ROR

BH

small

0.0363

0.0421

0.136

0.107

medium

0.0376

0.0416

0.160

0.127

large

0.0491

0.0388

0.341

0.271

Max-Stat

small

0.0465

0.0416

0.136

0.106

medium

0.0479

0.0412

0.159

0.126

large

0.0492

0.0385

0.335

0.266