Medical coding turns the free text your investigators type into the eCRF into the standardized terms of a medical dictionary. The same adverse event reported in three different wordings then arrives at analysis as one value. Until version 4.24 you had to export the data, code it in an external tool and re-import it. The whole workflow now lives in Datacapt. This article covers the vocabulary, who does what, and the questions that come up most often.

⚠️ Warning: MedDRA is the only dictionary type available in this first version, and a MedDRA license is mandatory. Datacapt ships no dictionary content. You download the packages from the MedDRA website with your own subscription.

Definitions

Verbatim: one free-text answer typed by an investigator into the eCRF, exactly as written. "Severe headache since yesterday" is a verbatim. Every answer collected on a coded item becomes its own verbatim, so three subjects reporting the same text give three verbatims.

Dictionary: the standardized vocabulary your coders pick terms from. You upload it once, at platform level, and every study of your tenant can use it. One dictionary is one language and one version, for example MedDRA 29.0 in English. Several versions can be active at the same time.

MedDRA (Medical Dictionary for Regulatory Activities): the international reference dictionary for adverse events and medical conditions. Its terms sit in a hierarchy of five levels, from the Lowest Level Term (LLT) up to the System Organ Class (SOC). A coder always picks an LLT, and Datacapt stores its full path: LLT, PT, HLT, HLGT, SOC.

Mapping: the study-level declaration that a given eCRF question must be coded, and against which dictionary. One mapping is one item to code, up to 10 supporting items, and one dictionary. Until an item is mapped, nothing is collected and nothing can be coded.

Supporting item: another question from the same subsection, attached to a mapping to give the coder context: the start date, the severity, the outcome. Supporting items can be of any type. The coder sees them next to the verbatim, and they are never coded themselves.

Coder workspace: the Medical coding section of a study, where every verbatim collected on a mapped item is listed with its context and a dictionary lookup sits alongside. The coder never needs access to the eCRF.

Code: the dictionary term attached to a verbatim, stored with its full path, the dictionary version, the mapping name, who coded it and when. A verbatim can carry several codes, one per medical concept, and an optional interpretation note on each.

💡 Tip: a verbatim has two statuses only. Uncoded means no code is attached yet. Coded means at least one code is. Removing the last code sends the verbatim back to Uncoded.

Who does what

Medical coding involves three steps, done by three people with three different permissions. Each has its own article.

Step	Where	Permission	Who, in most teams
1. Upload the dictionary, once for the whole platform	Global Settings › Dictionaries	Medical coding dictionaries (Access to view, Edit to upload, deprecate and delete)	Platform administrator
2. Create the mappings, per study	Study › Medical coding › Mappings	Medical Coding Mappings (Access to view, Edit to create and delete)	Data manager
3. Code the verbatims	Study › Medical coding	Medical Coding Coder (Access to view the verbatims, Edit to search a dictionary and assign a code)	Medical coder

💡 Tip: give your coder the Coder permission only. They see every verbatim with its supporting items in the coder workspace, and nothing else in the study. See Roles and Permissions Management to add the permissions to a role.

Frequently asked questions

Do I need my own MedDRA license?

Yes. Datacapt ships no dictionary content. You download the MedDRA package from the MedDRA website with your own subscription, then upload it to Datacapt exactly as delivered, with the password MedDRA gave you. Unzipped, renamed or repackaged files fail processing.

Which dictionaries can I use?

MedDRA only, in this first version. One upload is one language and one version, so MedDRA 29.0 in English and in French are two dictionaries. Several versions can be active at once, which long studies need since MedDRA publishes about two releases a year. Version and language are read from the package itself, so they can never be mistyped.

Which eCRF questions can be coded?

Free-text questions only. A free-text item inside a repeated measure can be coded, and each occurrence produces its own verbatim. A free-text item inside a matrix cannot be coded in this version. Any other question type can still join a mapping as a supporting item.

Does the coder need access to the eCRF?

No. The coder workspace shows each verbatim with its supporting items, subject, visit and form. Everything needed to pick a term sits on that one screen.

My study has been running for two years. Are the answers already collected coded too?

Yes. Creating a mapping is retroactive: every answer already filled on that item is pulled into the coder's queue, along with the ones that arrive afterwards. On a large study they take a moment to appear.

What happens to my mapping if the eCRF changes?

It depends on the change. If you delete the item, or change its type so it is no longer free text, the mapping is flagged Broken in red: collection stops, and the verbatims collected so far and their codes are deleted. If you change the title, the variable or a supporting item, the mapping is flagged Modified in yellow: collection continues and the codes are kept. Moving the item elsewhere in the eCRF has no impact.

Hide a coded question rather than delete it. Hiding keeps the coding work intact.

Can I edit a mapping?

No. You create it or you delete it, and deleting a mapping deletes every code assigned through it, with no way back. A warning modal tells you so before you confirm. The same item can be mapped a second time only against a different dictionary.

Can one verbatim carry several codes?

Yes, one per medical concept. "Headache and vomiting" gets two codes. Each code can carry its own optional interpretation note. The status covers the whole set, and the verbatim stays one row in the list.

Can I change or remove a code?

Yes. The coder removes a code from the verbatim and applies another. Removing the last code sends the verbatim back to Uncoded. Both actions are traced in the audit trail with the previous and the new value.

Is there a review or approval step, and can I export the codes?

Not yet. In this first version coding stops at Coded. Review, approval and export come in a next iteration.

Does medical coding work on a blinded study?

Blinded studies are out of scope for now.

Is medical coding audited?

Yes. Dictionary uploads, deprecations and deletions, mapping creations and deletions, and every code assigned or removed are recorded in the audit trail. See Read the platform audit trail.

👉🏻 Next step

Upload and manage medical dictionaries
Set up medical coding for your study
Code verbatims against a dictionary

What medical coding is, and how it works in Datacapt