> ## Content Index
> Fetch the complete content index at: https://www.elenapintea.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Interactions as safety signals: how DDI case reports feed pharmacovigilance
- URL: https://www.elenapintea.com/interactions-as-safety-signals-how-ddi-case-reports-feed-pharmacovigilance/
- Published: 2026-10-14T16:40:45.000Z
- Updated: 2026-11-19T16:24:24.000Z
- Description: Some of the best-known drug withdrawals were not about a single drug. They were about combinations.
- Author: Elena Pintea
- Tags: Drug Interactions

Some of the best-known drug withdrawals were not about a single drug. They were about combinations. Terfenadine, an antihistamine, caused fatal arrhythmias mainly when combined with CYP3A4 inhibitors such as ketoconazole (Honig et al., 1993). Cerivastatin was withdrawn in 2001 after cases of fatal rhabdomyolysis, many involving co-prescription with gemfibrozil (Furberg & Pitt, 2001).

In both cases, drug–drug interactions (DDIs) were detected largely through post-marketing case reports. This article explains how interaction cases become safety signals and why the quality of each report matters.

### What is a safety signal?

A signal is information, from one or more sources, suggesting a new potentially causal association between a medicine and an event, or a new aspect of a known association (Council for International Organizations of Medical Sciences \[CIOMS\], 2010). It is a hypothesis that needs investigation, not a conclusion.

For interactions, the "new aspect" is often the combination itself. Each drug may be well known and safe on its own, while together they produce an effect that neither label describes clearly enough.

### Why interactions are hard to detect in clinical trials

Pre-approval trials usually exclude many patients who take several medicines, especially older patients with multiple conditions. Formal DDI studies test the most predictable combinations, based on known metabolic pathways.

In real life, patients combine drugs in ways no trial anticipated. That is why spontaneous reporting remains essential for interactions.

### The signal management process

Signal management follows a structured sequence, described in European good pharmacovigilance practice (GVP Module IX) and applied in similar form by Swissmedic and companies worldwide (European Medicines Agency \[EMA\], 2017):

1. **Detection:** finding a possible new association in the data.
2. **Validation:** checking whether there is enough evidence to justify further analysis.
3. **Prioritisation:** deciding how urgently the signal needs attention, based on seriousness and public health impact.
4. **Assessment:** reviewing all available evidence, including cases, literature, mechanism and clinical studies.
5. **Recommendation for action:** for example updating the product information, adding a contraindication or issuing a safety communication.

### How interaction signals are found

Safety databases use statistical methods to find drug–event pairs reported more often than expected. Common measures include the proportional reporting ratio (PRR) and the reporting odds ratio (ROR; Evans et al., 2001).

Interactions need an extra step. Analysts compare how often an event is reported with *both* drugs against how often it appears with each drug alone. Specialised methods, such as the shrinkage-based measures developed at the Uppsala Monitoring Centre for the WHO global database, are designed for this purpose (Norén et al., 2008).

Statistics only point to possible signals. Each one still needs clinical review of the individual cases.

### What makes an interaction case useful

An interaction can only be detected if the second drug appears in the report. This is where the reporter's work matters most.

A useful interaction case includes:

- **all concomitant medicines**, including self-medication and herbal products;
- **start and stop dates** for each drug, so overlap can be confirmed;
- **doses**, since interactions are often dose-dependent;
- **relevant patient factors**, such as kidney function, age or known genotype;
- **the outcome after a change**, for example improvement when one of the two drugs was stopped.

In my pharmacy work, I record the full medication list when documenting a suspected adverse reaction. The pharmacy file often shows products from several prescribers and over-the-counter purchases that no single doctor sees.

### Pitfalls in interpreting interaction signals

Spontaneous data have well-known limitations, and interaction analysis adds more:

- **Co-prescription bias:** some drugs are frequently prescribed together, so they appear together in reports regardless of any interaction.
- **Confounding by indication:** the underlying disease, not the combination, may explain the event.
- **Masking:** a drug with very many reports can hide signals for other drugs in the same database.
- **Notoriety bias:** publicity about a risk can trigger a wave of reports (Pariente et al., 2007).

A good assessment weighs these factors and uses pharmacological plausibility. My training in mechanisms of drug action and toxicity during my MSc helps in this step: a credible mechanism, such as enzyme inhibition, makes a signal more convincing.

### From signal to action

When an interaction signal is confirmed, the outcome is often an update to the interactions and warnings sections of the product information. For serious risks, it can mean a contraindication, a direct communication to healthcare professionals or, rarely, withdrawal.

Pharmacists then put these changes into practice. That closes the loop between the patient at the counter and the safety team that assessed the signal.

### Why this matters to employers

Safety scientists working on signal detection need to understand both the statistics and the clinical reality behind each report. My background lets me read a case narrative and judge whether an interaction is plausible, what information is missing, and what follow-up would make the case assessable.

### Key takeaways

- Many important interactions are found after approval through spontaneous reports.
- Interaction signals depend on complete concomitant medication data with dates.
- Statistical findings need clinical and mechanistic assessment before any action.

### References

Council for International Organizations of Medical Sciences. (2010). *Practical aspects of signal detection in pharmacovigilance: Report of CIOMS Working Group VIII*. CIOMS.

European Medicines Agency. (2017). *Guideline on good pharmacovigilance practices (GVP): Module IX – Signal management (Rev 1)*. https://www.ema.europa.eu

Evans, S. J. W., Waller, P. C., & Davis, S. (2001). Use of proportional reporting ratios (PRRs) for signal generation from spontaneous adverse drug reaction reports. *Pharmacoepidemiology and Drug Safety, 10*(6), 483–486\. https://doi.org/10.1002/pds.677

Furberg, C. D., & Pitt, B. (2001). Withdrawal of cerivastatin from the world market. *Current Controlled Trials in Cardiovascular Medicine, 2*(5), 205–207.

Honig, P. K., Wortham, D. C., Zamani, K., Conner, D. P., Mullin, J. C., & Cantilena, L. R. (1993). Terfenadine-ketoconazole interaction: Pharmacokinetic and electrocardiographic consequences. *JAMA, 269*(12), 1513–1518.

Norén, G. N., Sundberg, R., Bate, A., & Edwards, I. R. (2008). A statistical methodology for drug–drug interaction surveillance. *Statistics in Medicine, 27*(16), 3057–3070\. https://doi.org/10.1002/sim.3247

Pariente, A., Gregoire, F., Fourrier-Reglat, A., Haramburu, F., & Moore, N. (2007). Impact of safety alerts on measures of disproportionality in spontaneous reporting databases: The notoriety bias. *Drug Safety, 30*(10), 891–898.