How to code adverse events and medications (MedDRA, WHODrug)
Turn the free text sites type into MedDRA and WHODrug codes you can analyse. Dictionary versions, coding rules and review in six steps.
Interactive tour on demo data. Click through at your own pace, no email asked.
All guides and toursInside the demo
A 30-minute call on your protocol, your sites and the modules you need.
- Fix the dictionary versions per study and write them in the data management plan.
- Code adverse events at the lowest level term and keep the full MedDRA path.
- Query the site when a verbatim holds two events or no medical meaning.
- Have a second person review the codes before database lock.
You code by turning the free text a site enters, such as "bad headache since yesterday", into a standard term from a medical dictionary: MedDRA for adverse events and medical history, WHODrug for concomitant medications. Your statistician can then count events across sites and languages, and every safety table holds up.
What a coder picks
Coders pick a lowest level term (LLT) in MedDRA and the dictionary supplies the rest of the path up to the system organ class. WHODrug links a trade or generic name to its active ingredient and ATC class. Both publish a new version twice a year, so your study fixes the version it uses and records it in the data management plan. For the levels of each dictionary and how they nest, read medical coding in clinical trials.
How to code in 6 steps
- Fix the versions. Record the MedDRA and WHODrug versions in the data management plan, and decide when you upgrade.
- Choose the items to code. Adverse events, medical history and concomitant medications are the usual three. Add supporting fields, such as the indication or the route, to give the coder context.
- Write coding conventions. Agree how to handle combined terms, abbreviations and signs that belong to a diagnosis already coded.
- Code at the lowest level term. Pick the LLT that matches the verbatim, check the full path to the SOC, and store the dictionary version with the code.
- Query unclear verbatims. "Nausea and vomiting" holds two events. "Pain" needs a location. Ask the site to split or clarify.
- Review before lock. A second coder or a medical reviewer checks consistency: the same verbatim should get the same code across the study.
Mistakes to avoid
- Changing dictionary versions mid-study without a plan. Codes can move between terms. Recode and document the change.
- Coding a combined verbatim to one term. You lose an event from the safety table.
- Coding outside the EDC. Exporting verbatims to a spreadsheet and importing codes back breaks the audit trail.
Medical coding in Datacapt
In Datacapt you upload the MedDRA package once in the global settings, and Datacapt reads its version and language from the file. You map a free-text item to a dictionary with up to 10 supporting items, and answers already collected join the coder's queue. Coders search the dictionary and see the full path from LLT to SOC before they apply a code. Each code stores the path, the version, the coder and the time in the audit trail, and coders work from the verbatims without opening the eCRF.
Keep reading
Who should code: the site or the sponsor?
The sponsor or its CRO. Sites enter the verbatim as the patient or the record describes it, and trained coders apply the dictionary.
What is autocoding?
An autocoding tool matches a verbatim to a dictionary term, or to a synonym your team approved, and a coder reviews each match. In Datacapt your coders apply each code, and they search the dictionary with partial words.
Do I need a MedDRA licence?
Yes. MedDRA requires a subscription from the MSSO, and WHODrug a licence from the Uppsala Monitoring Centre. Academic and non-profit users can get reduced or free access.