The clinical data management process is the structured lifecycle that turns raw observations from a clinical trial into a clean, reliable, analysis-ready dataset. Done well, it protects data integrity from the first data point captured to the final locked database. Done poorly, it becomes one of the most common and preventable source of delays and data quality problems in a study.
This guide is a map of that lifecycle. It follows the three stages used in practice — set-up, conduct, and close-out — and shows how they connect, pointing you to deeper guides for the technology and the individual activities along the way. The framework is universal: the same principles apply whether you work on a small Phase I study or a large global trial, and wherever in the world you sit.
What is the clinical data management process?
Clinical data management is the discipline of collecting, cleaning, and managing trial data so that it meets the quality standards required for regulatory decision-making. The process is governed by frameworks such as ICH E6 Good Clinical Practice and, for electronic records, regulations like 21 CFR Part 11 in the US and EudraLex Annex 11 in the EU. Underpinning all of it are the ALCOA+ principles: data should be attributable, legible, contemporaneous, original, accurate, complete, consistent, enduring, and available.
In practice, the work falls into three sequential stages, each with its own deliverables and quality gates:
- Set-up — everything that must be in place before a single data point is collected.
- Conduct — collecting, cleaning, and consolidating data while the trial runs.
- Close-out — confirming the data is clean, locking the database, and delivering it for analysis and archiving.
Each stage builds on the previous one. Shortcuts taken during set-up almost always resurface as expensive problems during conduct and close-out, which is why experienced teams invest heavily up front.
Stage 1: Set-up — preparing to collect data
The set-up stage defines how data will be captured, validated, and governed before any data exists. In practice it delivers a short checklist: the data collection system is built and tested, the governing documents are written, and the people using the system are trained. Only when all three are in place can data collection safely begin.
The data capture technology sits at the centre of set-up. Most trials run on an electronic data capture (EDC) system, where the electronic Case Report Form (eCRF), edit checks, and audit trail are configured to match the protocol, and integrations with laboratories and vendors are established. Because that technology layer is a substantial subject in its own right, this guide does not repeat it here — see our complete guide to EDC systems in clinical trials for how these platforms work, and our guide to EDC validation requirements for what validating and releasing one involves.
The governing document is the Data Management Plan, which sets out how data will be handled, cleaned, coded, and reviewed across the whole trial — covered in full in our dedicated guide to the clinical data management plan. Set-up closes once the system is validated and released and users are trained. That training step is a genuine gate: sites cannot begin entering data until designated users are trained and granted access.
Stage 2: Conduct — collecting, cleaning, and consolidating
The conduct stage runs for as long as the trial collects data. Its purpose is to ensure that the data flowing in reaches the quality required for analysis, and that data from many sources is consolidated into one coherent, standardised database. This is where the lifecycle does most of its work.
Data collection and the source data concept
Data arrives through several channels: site entry into the eCRF, electronic transfers from laboratories and vendors, and patient-reported outcomes from ePRO and eCOA tools. A foundational concept here is source data — the original record of an observation. Regulatory expectations require that data be traceable back to its source, which is why source data verification and review remain central activities throughout the stage.
Data cleaning and data quality
Data cleaning is the heart of the conduct stage. It combines the automated edit checks built during set-up — which flag values that are missing, out of range, or inconsistent and raise queries to the site — with manual review by data managers and medical reviewers that catches what automation cannot. The detailed mechanics of designing and running those checks belong to data validation, which we cover in its own guide. At the process level, what matters is the rhythm: cleaning is continuous, not a phase at the end, and the goal is a clearly defined level of data quality — data fit for the decisions that depend on it, not perfection at any cost.
Medical coding, reconciliation, and consolidation
As data accumulates, several specialised activities run in parallel. Medical coding standardises free-text terms for adverse events and medications against dictionaries such as MedDRA and WHODrug. Reconciliation compares data held in different systems — for example serious adverse event data in the safety database against the clinical database, plus laboratory and external vendor data — to confirm they agree. Finally, data consolidation merges these sources and applies standards such as CDISC (notably the SDTM model) so that the resulting database is structured and ready for analysis. Each of these has its own dedicated guide; here they mark the transition from raw collection to an analysis-ready dataset.
Stage 3: Close-out — locking and delivering the data
The close-out stage confirms that data processing is complete and that the database can be trusted as final. This is where the process delivers its ultimate product: a clean, locked dataset. It is also the stage least covered elsewhere, so it is worth understanding in detail.
Declaring the clean file
Before locking anything, the team confirms that data collection, cleaning, and consolidation are complete. Any residual issues are documented and accounted for, protocol deviations are assessed, and a clean file meeting brings stakeholders together to formally agree that the data meets the required standard. Declaring the clean file is the gate that authorises everything that follows.
Database lock and trial unblinding
Database lock freezes the data so that no further changes can be made without a controlled, documented process. For blinded trials, unblinding follows lock so that treatment assignments can be revealed for analysis. In longer or adaptive studies, teams may use interim or partial locks to release a defined subset of data for an analysis while the rest of the trial continues.
Archiving, decommissioning, and timelines
After delivery, data and documentation are archived to meet regulatory retention requirements, user access is revoked, and the data collection and transfer systems are decommissioned. Close-out timelines are typically measured from last-patient-last-visit to database lock, and reducing that interval — without cutting corners on quality — is one of the clearest ways an experienced data management team adds value.
Roles and timelines across the process
The process is run by a team rather than an individual: data managers own the plan and the cleaning workflow, programmers build edit checks and integrations, medical coders and reviewers handle clinical judgment, and quality functions provide oversight. One principle sits above that division of labour, though: the investigator owns the data. Only the site can change site-entered data, so when cleaning surfaces a problem the correction is never made by the sponsor directly — it is raised as a query for the investigator to resolve. We cover who does what in our guide to clinical data management roles and responsibilities. Across all three stages, timelines are driven by trial complexity, the number of data sources, and how much was prepared during set-up — which is why the process rewards front-loaded effort more than any other discipline in clinical development.
Frequently asked questions
What are the main stages of the clinical data management process?
The process has three sequential stages. Set-up prepares everything needed before data collection, including the eCRF, the Data Management Plan, system validation, and user training. Conduct covers data collection, cleaning, coding, reconciliation, and consolidation while the trial runs. Close-out confirms the data is clean, locks the database, and delivers it for analysis and archiving.
What is the difference between data cleaning and database lock?
Data cleaning is the ongoing activity of identifying and resolving data issues during the conduct stage, using automated edit checks and manual review. Database lock happens once, at close-out, after cleaning is complete. It freezes the dataset so no further changes can be made without a controlled process, marking the point at which the data is considered final and ready for analysis.
What is a clean file in clinical data management?
A clean file is a dataset that has been fully collected, cleaned, coded, and reconciled, with all outstanding queries resolved and any residual issues documented and accounted for. Declaring the clean file is a formal step, usually confirmed at a clean file meeting, and it is the prerequisite for locking the database.
Why is the set-up stage so important?
Set-up defines how data will be captured, validated, and governed before any data exists. Decisions made here — eCRF design, edit checks, the Data Management Plan, and system validation — shape the entire trial. Problems left unresolved during set-up almost always resurface as more expensive and time-consuming issues during conduct and close-out, so thorough preparation directly reduces downstream risk.
Which regulations and standards govern the process?
The process is built on ICH E6 Good Clinical Practice, with electronic records governed by regulations such as 21 CFR Part 11. The ALCOA principles define the qualities data must have, and standards from CDISC structure the data for analysis. Dictionaries such as MedDRA and WHODrug standardise medical coding, supporting consistency and regulatory acceptability across the trial.
Learn the process in depth with TriTiCon
This guide is a map of the full process. If you want to build practical, applicable expertise stage by stage, TriTiCon's training modules follow exactly this structure: The Clinical Data Management Set-up Process, The Clinical Data Management Conduct Process, and The Clinical Data Management Close-out Process. Each module defines the key concepts, highlights regulatory requirements, and shows you how to apply the fundamentals across different trial and company settings.
