Medical coding is the clinical data management activity that converts the free-text terms recorded during a trial — the adverse events a participant reports, the medications they take, their medical history — into standardised terms drawn from controlled dictionaries. A site might record “pounding headache,” “bad head pain,” and “cephalalgia” for what is clinically the same event. Coding maps all three to a single standard term so the data can be counted, grouped, and analysed consistently.
Without coding, free text cannot be aggregated reliably: you cannot count how many participants experienced a given event if the same event is written five different ways. This guide explains how medical coding works in clinical data management, the two dictionaries that dominate the field, the coding workflow, and how quality is controlled.
What is medical coding in Clinical Data Management?
Medical coding in clinical data management is the assignment of verbatim terms reported in a trial, such as adverse events, medical history and concomitant medications, to standard terms in a controlled dictionary: mainly MedDRA for events and conditions, and WHODrug for medicines. The coded terms let the same event or drug be grouped, counted, analysed and reported consistently across sites and studies.
Medical coding is the process of assigning a standardised dictionary term to a verbatim term — the exact words entered by the site. The verbatim term is preserved as the original record, and the assigned dictionary term sits alongside it, enabling consistent grouping and analysis. Coding takes place during the conduct stage, as part of standardising and consolidating data into an analysis-ready database.
It is worth clearing up a common confusion early: medical coding in clinical research is not the same as the medical billing coding used in healthcare administration (ICD and CPT codes for reimbursement). In clinical trials, coding uses different dictionaries built specifically for regulatory safety analysis, and its purpose is data standardisation, not billing.
Why medical coding matters
Coding is what makes trial data analysable and comparable. Standardised terms allow a sponsor to count and group events accurately, detect safety signals across a study or programme, and present data to regulators in a consistent, expected form. Because the same dictionaries are used across the industry, coded data can also be compared across trials and pooled for analysis.
The safety dimension is the most important. Adverse events must be grouped reliably for a signal to be visible — if related events are scattered under inconsistent terms, a genuine safety pattern can be missed. Accurate, consistent coding is therefore a patient-safety activity as much as a data-quality one.
The two main dictionaries: MedDRA and WHODrug
Two controlled dictionaries handle the large majority of coding in clinical trials.
MedDRA — adverse events and medical conditions
MedDRA (the Medical Dictionary for Regulatory Activities) is the standard for coding adverse events, medical history, and indications. It is maintained under ICH and is organised as a five-level hierarchy: from the broad System Organ Class (SOC) at the top, down through High Level Group Terms (HLGT), High Level Terms (HLT), and Preferred Terms (PT), to the most specific Lowest Level Terms (LLT) at the bottom. A verbatim term is coded to an LLT, which rolls up to a single Preferred Term — the level at which events are usually counted and reported. MedDRA is updated twice a year, which makes version control an active part of coding.
In practice, coding is done at LLT level and analysis happens higher up. Safety tables usually count events by Preferred Term and group them by System Organ Class. A Preferred Term can sit under more than one SOC, but it always has one primary SOC, which is the one normally used in summary tables. For example, a verbatim term such as “bad head pain” is coded to an LLT that rolls up to the Preferred Term Headache, whose primary SOC is Nervous system disorders.
| MedDRA level | Abbreviation | Role in coding and analysis |
|---|---|---|
| System Organ Class | SOC | Broadest grouping, by body system, cause or purpose. Used to organise safety tables. |
| High Level Group Term | HLGT | Groups related High Level Terms. |
| High Level Term | HLT | Groups related Preferred Terms. |
| Preferred Term | PT | One distinct medical concept. The usual level for counting and reporting events. |
| Lowest Level Term | LLT | Most specific level, including synonyms and lexical variants. The level a verbatim term is coded to. |
WHODrug — medications
WHODrug is the global standard for coding medications — concomitant medications, prior therapies, and trial treatments. Maintained by the Uppsala Monitoring Centre, it links each drug to standardised information including its active ingredients and its Anatomical Therapeutic Chemical (ATC) classification, which allows medications to be grouped by therapeutic class. Like MedDRA, WHODrug is released in regular versions, and trials must manage which version applies.
WHODrug coding: how medications are coded
WHODrug coding matches each medication the site records, whether a trade name or a generic name, to a drug name in WHODrug Global. The coded entry identifies the active ingredients, so the same medicine reported under different trade names can be grouped together.
The coded drug is also assigned an ATC class. Many active ingredients have more than one ATC code, so the choice depends on why the participant takes the drug. Acetylsalicylic acid (aspirin), for example, can be classed as an analgesic or as an antithrombotic agent, and the indication recorded by the site decides which applies. If the indication is missing or unclear, the coder raises a query rather than guessing. WHODrug Global, like MedDRA, is updated twice a year.
The medical coding workflow in clinical trials
Coding combines automation with human judgement, and follows a consistent sequence:
- Auto-coding — the system attempts to match each verbatim term automatically against the dictionary, often using a synonym list of previously approved matches. Clean, common terms code automatically.
- Manual coding — terms the system cannot match are coded by a trained medical coder, who selects the correct dictionary term based on the clinical meaning.
- Coding queries — when a verbatim term is ambiguous, misspelled, or describes more than one event, the coder raises a query to the site to clarify what was meant, rather than guessing.
- Medical review — coded terms are reviewed for clinical appropriateness, often by a medical monitor, particularly for events that influence the safety analysis.
- Approval and synonym update — approved matches are added to the synonym list so the same verbatim term codes automatically in future, steadily improving the auto-coding rate.
Underpinning the workflow are coding conventions — documented rules that keep decisions consistent between coders and over time, so the same verbatim term is always coded the same way.
Medical coding is usually carried out by specialist medical coders, either in the sponsor's data management team or at a CRO, with a physician or medical monitor reviewing terms that affect the safety analysis. The data manager tracks coding status and open coding queries so that coding is complete before database lock. See our guide to clinical data management roles and responsibilities for how the medical coder fits into the wider team.
Dictionary versioning and re-coding
Because MedDRA and WHODrug are updated on a regular schedule, a trial must decide which version it uses and when, if ever, to move to a newer one. Moving to a newer version — sometimes called up-versioning — can change how certain terms are classified, so it is a controlled activity with its own impact assessment. Long-running studies in particular need a clear strategy for whether to re-code data to a later dictionary version before database lock.
What is upversioning?
Upversioning (also written up-versioning) is moving a study's coded data from the dictionary version it was coded with to a newer MedDRA or WHODrug release. Terms that are already coded are checked against the new version and recoded where a term has been changed, moved in the hierarchy or made non-current. The resulting changes to groupings are reviewed and documented.
Because MedDRA and WHODrug Global are each updated twice a year, the data management plan or coding conventions usually state which versions a study uses and when upversioning happens, for example before database lock. Pooled analyses across several studies also need all data coded to a common version.
Quality control of coding
Coding quality is checked rather than assumed. Quality control looks for consistency — the same verbatim term coded the same way throughout the study — and for clinical accuracy, confirming that the assigned term genuinely reflects what the site reported. Coded data is typically reviewed in batches, with particular attention to events that carry safety weight. This QC is closely linked to the wider data-management quality and audit-trail framework, and to the medical and safety reviews that depend on correctly coded data.
Where coding fits in the lifecycle
Coding runs through the conduct stage alongside data cleaning, and feeds directly into two downstream activities: the consolidation of data into a standardised, analysis-ready database, and safety reconciliation, where coded adverse-event terms in the clinical database are compared against the safety database. Coding must be complete and reviewed before a clean file can be declared and the database locked. For data managers, coding is also one of the recognised specialist skills within the discipline — see our guide to clinical data management skills for how it fits alongside the other competencies in the field.
Frequently asked questions
What is medical coding in clinical data management?
Medical coding assigns a standardised dictionary term to the free-text (verbatim) terms recorded during a trial, such as adverse events and medications. The verbatim term is kept as the original record, while the coded term enables consistent grouping, counting, and analysis. It is performed during the conduct stage and must be completed before database lock.
What dictionaries are used for medical coding in clinical trials?
Two dominate. MedDRA, maintained under ICH, codes adverse events, medical history, and indications through a five-level hierarchy from System Organ Class down to Lowest Level Term. WHODrug, maintained by the Uppsala Monitoring Centre, codes medications and links them to active ingredients and ATC therapeutic classes. Both are released in regular versions that trials must manage.
What is the difference between MedDRA and WHODrug?
MedDRA codes medical concepts: adverse events, medical history and indications, organised in a five-level hierarchy from System Organ Class down to Lowest Level Term. WHODrug codes medicines: concomitant medications, prior therapies and trial treatments, linked to active ingredients and ATC therapeutic classes. A typical trial uses both, MedDRA for what happened to the participant and WHODrug for what they took.
What is the difference between auto-coding and manual coding?
Auto-coding is the system's automatic matching of verbatim terms against the dictionary, usually via an approved synonym list, and handles clean, common terms. Manual coding is performed by a trained coder for terms the system cannot match, selecting the correct dictionary term based on clinical meaning. Approved manual matches are added to the synonym list, gradually increasing the auto-coding rate.
Who performs medical coding in clinical trials?
Medical coding is usually done by trained medical coders, working in a sponsor's data management team or at a CRO. Autocoding handles common terms, coders code the rest manually and raise coding queries to sites, and a physician or medical monitor reviews terms that matter for safety. The data manager oversees coding status and makes sure coding is complete before database lock.
Is clinical data management coding the same as medical billing coding?
No. Medical coding in clinical research uses dictionaries such as MedDRA and WHODrug for the purpose of standardising trial data for safety analysis and regulatory reporting. Medical billing coding uses systems such as ICD and CPT for healthcare reimbursement. They share the word “coding” but serve entirely different purposes with different dictionaries.
Why does dictionary versioning matter?
MedDRA and WHODrug are updated on a regular schedule, and newer versions can change how some terms are classified. A trial must decide which version it uses and whether to re-code to a newer one before lock. Because up-versioning can alter groupings, it is a controlled activity with its own impact assessment, especially important for long-running studies.
What is upversioning in medical coding?
Upversioning is moving coded trial data to a newer version of MedDRA or WHODrug. Both dictionaries are updated twice a year, and a new version can change or move terms, or make them non-current. Coded terms are checked against the new version and recoded where needed, and the impact on groupings is reviewed and documented, often before database lock or before studies are pooled.
Learn coding in context with TriTiCon
Medical coding is one part of turning raw collected data into a standardised, analysis-ready database. TriTiCon's Introduction to Clinical Data Management course introduces data consolidation, standardisation and medical coding concepts within the core collect, clean and consolidate process. The Clinical Data Management Conduct Process then follows how data is cleaned, queried and consolidated while a trial is live. You can explore the full range of clinical data management courses to see how the modules build on each other.
