# Data Management Plan Template for Clinical Trials By the end of this article you will be able to build a complete [data management](/blog/guides/data-management-basics-principles-processes-and-best-practices) plan (DMP) for a clinical trial using a reusable template, fill in every section with trial-specific content, and run a quality check before the plan goes to your sponsor, funder, or ethics committee. You do not need special software. You need a plain-text or word processor, a copy of your trial protocol, and access to your funder's data policy if one applies. If you plan to use an electronic data capture (EDC) system, have its name and version on hand. ## What a Clinical Trial Data Management Plan Actually Is A data management plan is a written document that explains how a study's data will be collected, organized, used, stored, protected, preserved, and shared, and it names the people responsible for each step [1]. For a clinical trial, the plan covers everything from the case report forms (CRFs, the structured forms used to record each participant's study data) to the rules for de-identifying records and the repository where de-identified data will land after the trial ends. A good plan does three jobs. First, it demonstrates governance, meaning it shows how clinical, laboratory, and molecular data will be handled under the regulations and policies of every stakeholder involved, including funders, institutions, and ethics bodies [1]. Second, it satisfies funder requirements. The National Institutes of Health (NIH) requires a Data Management and Sharing Plan (DMSP) with funding applications, and the policy has been in effect since January 2023 [2]. Third, it surfaces problems early. Writing down how data will be collected and shared tends to expose conflicts, for example between a sponsor's sharing timeline and a consent form that promised narrower use, while they are still cheap to fix [1]. The habit of writing these plans is not universal. A survey of the 20 most prolific public and 20 most prolific private clinical trial sponsors operating in Europe found wide variation in whether organizations publish data sharing policies, data management and sharing plans, and data use agreements at all, with substantial heterogeneity in what those documents contain [3]. In other words, you cannot count on copying a sponsor's house style. You need a template of your own that you adapt per trial. ## The Template Format: Nine Sections That Cover Every Reviewer's Question Most funder guidance, including the NIH's supplemental guidance on DMSP elements, lists categories rather than a strict form, so the structure below is a practical superset [2]. A clinical trial [data management plan template](/blog/guides/data-management-plan-template-a-practical-guide-for-researchers) in this format has survived review at academic medical centers and maps cleanly onto DMPTool, the free online planning platform that generates funder-specific templates [4][5]. | Section | What it answers | Common pitfall | |--|--|--| | 1. Administrative overview | Trial title, registration number, principal investigator, funder, DMP version and date | Missing version number, so reviewers cannot tell which draft they approved | | 2. Data types and sources | Every data stream: eCRF entries, lab results, imaging, patient-reported outcomes | Forgetting ancillary data such as adapters or wearables | | 3. Standards and metadata | Data dictionaries, common data elements, controlled terminology, units | Free-text fields where coded values belong | | 4. Collection tools and workflow | EDC system, paper source documents, [lab notebook](/knowledge/molecular-biology/lab-notebook) templates, visit schedules | No fallback described when the EDC is offline | | 5. Quality control | Validation rules, query management, monitoring, database lock criteria | Query resolution deadline left undefined | | 6. Storage, security, backup | Where files live, access control, backup schedule, retention period | Backup cadence not stated in numbers | | 7. Participant privacy | De-identification, coding of participant IDs, handling of vulnerable populations | Re-identification risk from dates or rare conditions | | 8. Sharing and preservation | Repository choice, what gets shared, when, under what access terms | Sharing commitment made without checking consent language | | 9. Roles, resources, budget | Named data manager, training plan, costed line items | "The study team" listed instead of named individuals | Keep the plan short enough that people actually read it. Five to ten pages is a realistic target for a single-center trial, with longer versions justified for multicenter studies. ## Step-by-Step Walkthrough: Filling In the Template ![Five-step flow of filling the DMP template: create the file, list data types, name standards, describe tools, and write quality control rules.](/article-images/illustrations/clinical-trial-data-management-plan-template-dmp-walkthrough.webp) Work through these eight steps in order. Each step produces text you paste into the template file. The worked example follows a fictional single-center trial called TRIAL-A1, a randomized comparison of two postoperative pain management regimens in 120 adults. ### Step 1. Create the template file Start a plain-text file called `DMP_TRIAL-A1_v0.1.md`. Markdown keeps the template portable, and you can paste the same content into DMPTool or a sponsor form later. Here is the skeleton, which doubles as a free clinical trial data management plan template you can copy: ```markdown # Data Management Plan: [TRIAL NAME] **Protocol number:** [XXX] **ClinicalTrials.gov ID:** [NCTXXXXXXX] **Principal Investigator:** [NAME] **Funder and award number:** [NAME, NUMBER] **DMP version / date:** [v0.1 / YYYY-MM-DD] ## 1. Administrative Overview [One paragraph describing the trial design in two sentences.] ## 2. Data Types and Sources | Data type | Source tool | Format | Volume estimate | |---|---|---|---| ## 3. Standards and Metadata [Data dictionary reference, controlled terminology, units.] ## 4. Data Collection Tools and Workflow [EDC name and version, paper fallback, training.] ## 5. Quality Control [Edit checks, query rules, monitoring plan, lock criteria.] ## 6. Storage, Security, and Backup [Locations, access control, backup cadence, retention.] ## 7. Participant Privacy and Confidentiality [De-identification method, coding scheme, special protections.] ## 8. Data Sharing and Preservation [Repository, timing, access terms, embargo.] ## 9. Roles, Resources, and Budget [Named roles, training, costed items.] ``` ### Step 2. List every data type and source Pull this from the protocol's schedule of assessments. For TRIAL-A1 that means eCRF entries (demographics, pain scores on a 0 to 100 visual analog scale, opioid use), serum laboratory results from the central lab, adverse event records, and participant diaries. ```markdown ## 2. Data Types and Sources | Data type | Source tool | Format | Volume estimate | |---|---|---|---| | Baseline and follow-up eCRF | Redcap 14 (study-hosted) | Export CSV | ~120 subjects x 5 visits | | Pain VAS scores | eCRF, entered at visit | Numeric 0-100 | 600 values | | Central lab chemistry | Vendor portal | CSV, HL7 batch | ~4 analytes x 600 draws | | Adverse events | eCRF AE form | Coded (MedDRA terms) | ~300 records | | Paper diary (rescue med use) | Participant diary, transcribed | Scanned PDF + eCRF | ~120 scans | ``` Expected output is a table with no empty rows and no data stream the protocol mentions that the table omits. The check: read the protocol's schedule of assessments line by line and tick each one off against your table. ### Step 3. Name your standards This is where most drafts go vague. State the exact dictionary, terminology, and units. The NIH HEAL Initiative's specialized DMPTool template emphasizes implementing common data elements, standardizing metadata, and protecting sensitive information as core planning activities, and you can borrow that structure even outside the HEAL program [4]. ```markdown ## 3. Standards and Metadata - Data dictionary: DMP Appendix A, one row per eCRF field (variable name, label, type, units, permissible values). - Adverse event coding: MedDRA version 27.0, locked at first patient enrolled, upgrades require a documented impact review. - Lab units: SI units as reported by the central lab, conversion factors stored in the data dictionary. - Participant identifiers: single sequential code (PT-001 through PT-120), assigned at randomization, stored apart from the contact log. ``` ### Step 4. Describe collection tools and the paper fallback Name the EDC and version. Then describe what happens at sites that still generate paper, because many do. If your protocol has any hands-on measurement or dispensing steps, those records often start as paper source documents or bound notebooks. A structured lab notebook template or a downloadable PDF source sheet keeps those records consistent, and the plan should state that scanned PDFs of paper source are filed in the trial master file within a stated number of business days of each visit. ```markdown ## 4. Data Collection Tools and Workflow - Primary EDC: Redcap 14, hosted on [institution] servers. - Entry window: within 3 business days of each visit. - Paper fallback: if EDC downtime exceeds 24 hours, sites record data on the approved paper CRF (template PDF v2.0), then transcribe within 2 business days of restoration. Original paper is filed in the investigator site file. - Source notebooks: pharmacy accountability logged in a bound notebook with preprinted page numbers, archived on study closeout. - Training: site staff complete EDC training and sign the delegation log before their first data entry. ``` ### Step 5. Write the quality control rules Quality control is where a data management plan earns its keep. Define range and logic checks, query turnaround, monitoring, and the criteria for locking the database. An illustrative edit check looks like this in the EDC: ``` Rule AE-04: If AE_START_DATE < RANDOMIZATION_DATE then fire query: "Adverse event start date precedes randomization. Please verify dates." ``` And a range check on a lab value: ``` Rule LB-12: If HEMOGLOBIN not between 3.0 and 25.0 g/dL then fire query: "Value outside plausible range. Please review source." ``` Set explicit service levels, for example that the data manager issues queries within two business days of detection and sites respond within five. Define database lock conditions in a numbered list, such as all queries closed, all monitoring visits complete, and all protocol deviations coded. ### Step 6. Specify storage, security, and backup Give concrete numbers. A vague sentence like "data are backed up regularly" will draw a query from any experienced reviewer. Replace it with locations, cadence, encryption status, and the retention period your funder and institution require. Note that retention rules vary by funder and by country, so the exact figure should come from your own award terms rather than from this article. ### Step 7. Address participant privacy For a clinical trial this section does real work. Describe how personal information is protected, which matters most for participants from vulnerable populations, and describe the governance framework covering use and transfer of data and samples nationally and internationally [1]. Practical content includes the participant coding scheme, who holds the link between codes and identities, rules against entering direct identifiers into the EDC, and special handling of the recruitment log. ### Step 8. Write the sharing and preservation section State what will be shared, in what form, where, and under what access conditions. The NIH DMSP guidance expects you to name a repository and describe timing and access terms, and published recommendations for maximizing compliance stress that naming the repository and addressing special data types explicitly are among the highest-value moves you can make [2]. For sensitive projects, the HEAL template shows how to match data types with sharing restrictions to HEAL-compliant repositories, an approach you can mirror with whatever repository your funder designates [4]. The access terms usually take the form of a data use agreement (DUA), a contract that governs what recipients may do with the data. Sponsors vary enormously here. The European sponsor survey found data use agreements present in only a subset of organizations and described recurring heterogeneity in their content, so never assume a sponsor's DUA template will match your consent language [3]. Read them together. ## Worked Example: Sections 7 and 8 Filled In ```markdown ## 7. Participant Privacy and Confidentiality - Participants are coded PT-001 to PT-120 at randomization. - The code-to-identity link is held by the coordinating center registrar in an access-controlled file, separate from all research data systems. - No direct identifiers (name, address, exact dates of birth, medical record number) are entered into the EDC. Dates are recorded as day/month/year. Ages in years. - Incarcerated participants and participants who cannot consent independently are not enrolled in this trial. - All study staff sign a confidentiality agreement before system access is granted. ## 8. Data Sharing and Preservation - Shared dataset: de-identified eCRF export plus the data dictionary, lab results keyed to participant code, and the annotated CRF. - Timing: deposited within 12 months of primary publication or end of funding, whichever comes first. - Repository: [funder-designated repository], with a persistent identifier requested at deposit. - Access: controlled-access. Requesters sign a data use agreement limiting use to non-commercial research and prohibiting attempts at re-identification. - Retention: study records retained per the award terms and institutional policy after database lock. ``` ## Where the DMP Ends and the Statistical Analysis Plan Begins A frequent point of confusion is the boundary between the data management plan and the statistical analysis plan (SAP). The SAP is a separate, protocol-level document that defines estimands, analysis populations, and statistical methods, written before data are examined. The TELESCOPE 2 trial, a stepped-wedge cluster randomized trial in Brazilian intensive care units, published its SAP as a standalone document covering its 25 intensive care units and staggered randomization sequences, which illustrates the level of statistical detail that belongs in the SAP rather than the DMP [6]. Keep design and analysis detail in the SAP and keep data flow, quality, and stewardship in the DMP, and cross-reference between them. The same separation shows up in methodological guidance for therapy research, which recommends a pre-specified statistical analysis plan alongside mechanisms for data registration and sharing as distinct deliverables [7]. ## Safety Data Deserve Their Own Paragraph If your trial collects adverse events, add a short subsection under quality control describing how safety data will be reviewed in aggregate, not just case by case. Industry practice is moving this way. A survey of clinical safety professionals and safety data scientists at 21 companies described the Aggregate Safety Assessment Planning (ASAP) process, in which a multidisciplinary Safety Management Team plans coordinated assessments of accumulating safety data across ongoing and completed trials, reflecting a shift in emphasis from individual expedited reports toward aggregate causality assessment [8]. Your DMP does not need to implement a full ASAP process, but it should say who reviews cumulative safety data, how often, and where the accumulating dataset lives. ## Timing and Team: When the Plan Gets Written Write the DMP during the planning phase, before first patient enrolled, not during setup crunch. Operational research on investigator-initiated multicenter trials mapped the entire planning phase into a time-bound blueprint of roughly 15 months from pre-award through first enrollment, and it treats Data Management as one of several parallel workstreams, alongside Data Systems Integration, Operations and Clinical Affairs, and Regulatory work, each with quarter-by-quarter deliverables and a dedicated project manager tracking predecessor-successor dependencies [9]. The practical translation for a smaller trial is simple. Give the DMP an owner by name, give it a deadline on the study timeline, and treat it as a gating item for EDC build and site training rather than paperwork to finish later. For multicenter trials that will share participant-level data with other studies, add the coordination machinery early. A toolkit developed across more than a decade of multinational individual participant data meta-analysis projects emphasizes clearly documented data sharing and authorship agreements, a dedicated management team, and standard operating procedures, all negotiated before data start moving [10]. ## If Your Funder Uses a Template, Start From It Three practical options, in order of preference: 1. **Funder-specific platform template.** DMPTool provides funder-templated forms, including specialized versions such as the NIH HEAL template for complex multisource studies [4]. Surveys at an academic medical center found that only a minority of researchers had used DMPTool while a large majority wanted tailored templates, which tells you the tailored template exists precisely because it saves time [5]. 2. **Sponsor template.** Some private and public sponsors publish their own data management and sharing plan templates, although the European survey showed these are not consistently available or uniform across organizations, so verify what your specific sponsor provides rather than assuming [3]. 3. **Generic template.** The nine-section skeleton above works when neither of the first two applies. The same template-first logic applies beyond the DMP itself. Community-engaged teams have shown that a structured recruitment plan template, refined through iterative review with community advisory boards, produces better grant language for recruitment and retention sections, and an 8-section version is publicly described with sections covering recruitment strategy and stakeholder communication [11]. If your DMP touches recruitment data, referencing a community-informed recruitment plan template keeps the two documents consistent. ## Checking the Result: A Ten-Point Review Before You Submit Run this checklist against the finished draft: 1. Every data type in the protocol's schedule of assessments appears in Section 2. 2. Every named tool has a version number. 3. Every quality control rule has a numeric threshold and an action. 4. Every query turnaround time is stated in business days. 5. The participant coding scheme is described, and the location of the re-identification key is stated. 6. A named repository appears in Section 8, not "an appropriate repository." 7. Sharing timing is stated as an event plus a duration, for example "within 12 months of end of funding." 8. The consent form's data sharing language matches the sharing section, word for word on scope. 9. Every role in Section 9 has a named individual, not a job category alone. 10. The document carries a version number and date on page one. Two people should run the list independently, then reconcile. That mirrors the double-extraction habit used in structured document reviews of sponsor data policies, where two reviewers extract from the same documents independently before reconciling [3]. ## Common Mistakes and Limitations - **Writing the DMP after the database is built.** Late plans force retrofitted standards, and retrospective fixes are exactly what the NIH DMSP policy is designed to prevent [2]. - **Confusing the DMP with the SAP.** Reviewers notice when analysis methods show up in a data plan. Put them in the SAP. - **Promising open sharing without checking the consent form.** If participants consented to defined uses only, the sharing section must match, and the mismatch is a real ethical and regulatory problem, not a formatting one [1]. - **Naming no repository.** Institutional surveys found low confidence in identifying appropriate repositories even among researchers who understood the policy, so this is the single most common weak point in real plans [5]. - **Skipping the budget.** Data management costs money, and low confidence in budgeting for data management and sharing activities showed up in the same institutional survey [5]. List concrete items: EDC hosting, repository deposit fees where applicable, and data manager effort. - **Leaving "the study team" as the responsible party.** Plans that survive review name the data manager, the registrar holding the re-identification key, and the person authorized to approve data requests [1]. - **Ignoring heterogeneity across sites and centers.** Multicenter trials need one data dictionary for all sites, and published methodological guidance recommends standardized templates and consistent criteria across centers precisely because protocol heterogeneity undermines comparability [7]. - **Treating safety data like any other data.** Aggregate safety review has its own cadence and its own multidisciplinary team, and the plan should reflect that [8]. - **Overlooking the human side.** Recruitment and retention failures often trace back to design decisions made before any data were collected, and community input on study design helps teams avoid pitfalls and build trust, especially with underrepresented groups [12]. Individual trials always have quirks that no template anticipates, and your institutional data office or sponsored programs office should review the final document before submission. ## Frequently Asked Questions ### What is a clinical trial data management plan template? It is a pre-structured document with labeled sections for describing how a trial's data will be collected, quality-checked, stored, protected, and shared. You fill in trial-specific content rather than writing from a blank page. ### Is there a free clinical trial data management plan template I can use? Yes. DMPTool.org provides free funder-templated forms, and the generic nine-section markdown skeleton in this article can be copied at no cost. Funder-specific templates are the best starting point when one exists. ### How does a data management plan differ from a statistical analysis plan? The DMP covers data flow, quality, security, and sharing. The SAP covers analysis populations, estimands, and statistical methods. They are separate documents that cross-reference each other. ### When should the DMP be written? During the planning phase, before the database is built and before first patient enrolled. Planning-phase workstreams for multicenter trials are typically mapped month by month, and the DMP is one of the deliverables in that window. ### Who reviews or approves the DMP? Typically the sponsor or funder, and for funder-required plans the document is submitted with the application. Institutional data offices and, where relevant, ethics committees may also review it. ### What belongs in the data sharing section? What will be shared, in what de-identified form, in which named repository, on what timeline, and under what access terms such as a data use agreement. The consent form's sharing language must match. ### How do I handle sensitive participant information in the plan? Describe the participant coding scheme, state who holds the re-identification key and where, forbid direct identifiers in research systems, and document special protections for vulnerable populations. ### How long should a clinical trial DMP be? Five to ten pages is realistic for a single-center trial. Longer is justified for multicenter studies, but every section should still answer a specific question rather than pad the document. ## Sources 1. [A data management plan for the NESHIE observational study.](https://pubmed.ncbi.nlm.nih.gov/38130874/) 2. [Ten simple rules for maximizing the recommendations of the NIH data management and sharing plan.](https://pubmed.ncbi.nlm.nih.gov/35921268/) 3. [Data sharing policies, requirements, and support from public and private clinical trial sponsors: a survey on top sponsors of clinical trials in Europe.](https://pubmed.ncbi.nlm.nih.gov/42575483/) 4. [Navigating the new NIH Data Management Sharing plan: the NIH HEAL Initiative Data2Action DMPTool template.](https://pubmed.ncbi.nlm.nih.gov/40401004/) 5. [Supporting researcher compliance: assessing institutional readiness for the 2023 NIH Data Management and Sharing Policy.](https://pubmed.ncbi.nlm.nih.gov/42889533/) 6. [Statistical analysis plan for a stepped-wedge cluster randomized clinical trial for the evaluation of the clinical impact of different telemedicine practices in intensive care units - TELESCOPE 2 trial.](https://pubmed.ncbi.nlm.nih.gov/42912019/) 7. [Designing clinical research on hyperbaric oxygen therapy: a narrative review and practical methodological guide.](https://pubmed.ncbi.nlm.nih.gov/42443991/) 8. [Aggregate Safety Assessment Planning in Clinical Trial Drug Development: Survey of Clinical Safety Professionals and Safety Data Scientists Advancing Premarket Safety Analytics at 21 Companies.](https://pubmed.ncbi.nlm.nih.gov/42879004/) 9. [Optimizing the study planning phase in an investigator-initiated multicenter clinical trial: Mapping the process into a blueprint.](https://pubmed.ncbi.nlm.nih.gov/43011273/) 10. [How to plan and manage an individual participant data meta-analysis. An illustrative toolkit.](https://pubmed.ncbi.nlm.nih.gov/37700398/) 11. [A community-informed recruitment plan template to increase recruitment of racial and ethnic groups historically excluded and underrepresented in clinical research.](https://pubmed.ncbi.nlm.nih.gov/36572240/) 12. [Recruitment and retention for chronic pain clinical trials: a narrative review.](https://pubmed.ncbi.nlm.nih.gov/38304397/)