Measures of frequency
The same data give different numbers depending on whether you count patients or events and divide by people or by time. Enter counts from one dataset to see the incidence proportion, the exposure-adjusted incidence rate (EAIR), the exposure-adjusted event rate (EAER) and the one-year risk they imply.
Inputs
Results
Enter N, patients with an event and person-time.
Absolute and relative, always both
A relative effect without an absolute one invites alarm; an absolute effect without a comparator invites complacency. Compare two groups by person-time or by patients, or convert a published relative risk into excess cases.
Inputs
Results
Enter both groups.
Zero events: what the data still allow
"No cases were observed" is true and incomplete. The informative statement is the largest frequency that zero events in this denominator cannot exclude.
Inputs
Results
Enter the denominator.
Observed versus expected
Expected = background rate × person-time in the risk window, summed over strata. Use age- (and sex-) specific rates that describe the exposed population; a flat all-ages rate is the commonest error.
Exposure and background
| Stratum | Persons / doses | Background rate |
|---|
Observed and sensitivity
Results
Add strata with exposure and background rates, then the observed count.
SmPC section 4.8 frequency categories
Very common ≥1/10 · common ≥1/100 to <1/10 · uncommon ≥1/1,000 to <1/100 · rare ≥1/10,000 to <1/1,000 · very rare <1/10,000 · not known (cannot be estimated from the available data). The workbench assigns the crude category and flags the rows an assessor is likely to test.
Adverse reactions
| Term | Source | Drug n | Drug N | Placebo n | Placebo N | Long-term rate /100 PY (optional) |
|---|
Results
Add rows.
Contextualized risk statement
One sentence per context question. The builder assembles the paragraph, checks it against the ICH E2C(R2) §16.4 elements and flags words that overstate or understate a risk.
Six questions
Draft paragraph
§16.4 element coverage
Wording check
Highlights words from the book's over/understatement list (Table 24.2) in the draft above, or paste any text below.
How to use the workbench
The workbench follows the book's six context questions: How often? Compared with what? How bad? For whom and when? Can it be prevented or managed? How sure are we? Modules 1–5 produce the numbers; module 6 turns them into a paragraph. Each module drafts wording you can copy, and Print report collects everything into one record for your file.
Load Kestrelinib demo fills every module with the book's worked examples (serious infections, herpes zoster, GI perforation, ARV-60 Guillain–Barré syndrome, the Set A adverse-reaction table and the VTE statement) so you can compare the output with the chapters.
Methods
- Rates: exact (Garwood) Poisson confidence intervals; proportions: exact Clopper–Pearson intervals.
- Rate ratio: exact conditional interval (binomial split of events given person-time). Rate and risk differences: Wald intervals. Risk ratio: log-scale interval. Fisher's exact test (two-sided) for proportions.
- Number needed to harm: 1 / difference, with the interval inverted from the difference interval; when that interval spans zero it is shown in the Altman form (NNH … to ∞ to NNT(benefit) …).
- Observed-versus-expected: expected = Σ exposure × window × stratum rate; O/E interval from the exact Poisson interval of the observed count divided by the expected count (expected treated as fixed).
- Zero events: exact upper bound 1 − (α/2)1/n, one-sided bound 1 − α1/n, and the rule of three (3/n).
- One-year risk from a constant rate: 1 − e−rate. This assumes a constant hazard and no competing events (Chapter 5).
Privacy
Everything runs in your browser. Nothing is uploaded. Inputs are kept only in this browser's local storage so they survive a page reload; Clear all removes them.
The book
Risk in Context: Practical Risk Contextualization in Pharmacovigilance — From Safety Findings to Regulatory Conclusions — Dr. Hafez Selim, MD, PhD. Selim Medical Press, 2026. Companion titles: Practical Benefit–Risk Assessment in Pharmacovigilance and Time to Onset in Drug Safety.
Version 1.0 · © 2026 Hafez Selim, Selim Medical Press. Educational use only; see the terms shown at start-up.