Completion guidelines tell site staff exactly how to fill in each form so two coordinators enter the same facts the same way. This guide gives you a structure to copy, field-level wording for a demo study, and a rollout plan that gets the document read.
Free sandbox · No credit card · 21 CFR Part 11 aligned
Adverse event
Adverse event term
AETERMDiagnosis if known, else the symptom. One event per row.
HeadacheStart date
AESTDTCPartial dates allowed: unknown day as UNK.
12 Mar 2026Severity
AESEVIntensity, not seriousness.
Serious
AESERWhat to take from this guide
Purpose
Every free-text field and every judgement call invites inconsistency. Is a "headache" that lasts all week one adverse event or seven? Do you enter a drug by brand or generic name? What does a coordinator record when a participant refuses a blood draw: blank, zero or "not done"? Without a shared convention, each site answers differently, and the data manager discovers it as a flood of queries or, worse, as unusable data at analysis. A completion guideline pays for itself the first time it prevents a query that would have taken three emails to close.
It also helps people who are new. Site staff change, sub-investigators join and coordinators cover each other's leave. A concise guide with examples is faster to learn from than the protocol and the data management plan together. And it supports inspection: it shows that you told sites how data should be recorded and when you changed your mind. The related data management plan template normally references the completion guidelines as one of its controlled documents.
This page is different from guidance on how to design the form itself (how to design an eCRF) and from paper CRF to eCRF conversion. Those are about building the form. The CCG is about using it correctly once it exists.
Template
| Section | What goes in it | Length |
|---|---|---|
| 1. Document control | Study ID, protocol version, CCG version and date, author, approver, change history | Half a page |
| 2. Purpose and how to use | What the document is for, how it relates to the protocol and the data management plan, who to ask | A few lines |
| 3. General conventions | Dates (format and partial dates), times, units, decimals, "not done" versus "not applicable" versus "unknown", rounding, corrections and reasons for change | One page |
| 4. Visit and form map | Table of visits and forms, which are repeating, and which are completed by whom | One page |
| 5. Form-by-form instructions | For each form: purpose, when to complete, field-level notes for fields that need them | Bulk of document |
| 6. Adverse events and medications | Severity versus seriousness, relationship, onset and end conventions, naming drugs, recording changes | One to two pages |
| 7. Queries and corrections | How queries are raised, how to answer, what a reason for change should say, response times | Half a page |
| 8. Contacts and appendices | Who to contact, abbreviations, example completed forms | As needed |
Keep section 5 to fields that need a rule. Self-evident fields (height, weight) need no entry.
Worked example
These examples are for an invented study, DEMO-WALK-01, a walking programme trial with two sites. They show the tone: direct, one rule per field, an example where it helps. Adapt them to your own protocol and conventions.
Dates: enter as day, month, year. If the day is unknown enter UNK, for example UNK Mar 2026; never guess. Not done vs not applicable: if an assessment was planned but not performed, mark "not done" and give the reason; if it did not apply to the participant, mark "not applicable". Never leave a required field blank without one of the two. Corrections: change the value and give a reason for change (for example "transcription error"). Do not overwrite without a reason.
Blood pressure: enter the second of two seated readings taken after five minutes of rest, in mmHg, to the nearest whole number. If only one reading was taken, enter it and note in the comment. Weight: kilograms, one decimal, without shoes and with light clothing. BMI is calculated automatically; do not enter it.
Event term: record a diagnosis if known; otherwise the main symptom. One event per row. Severity: intensity (mild, moderate, severe); this is not the same as seriousness. Serious: select Yes only if the event meets a seriousness criterion in the protocol; then notify the sponsor within the protocol timeline and see the adverse event reporting page for how that connects. Changes in severity: follow the protocol convention. Here, record a new row when severity increases.
Name: generic name where known; if only a brand name is known record it as written and the data manager will code it. Dose and unit: as prescribed, not as taken. Start and stop: use UNK for unknown parts of the date; use "ongoing" only if still taken at the visit.
If a visit falls outside the window, still enter the data on the visit form and record a protocol deviation; do not alter the date to fit.
Create your forms from templates in the free sandbox, add edit checks for the rules you wrote, and test them before sites do. Free, every feature, no credit card.
Writing rules
A completion note is a tiny piece of user guidance. Apply the same discipline you would to a label on a medicine pack.
Start with the action: "Enter", "Select", "Leave blank", "Do not".
Show a correct entry and, for tricky fields, a wrong one.
Every field that may be unknown needs a convention (UNK, not done, not applicable).
Match terms in the protocol and the form so site staff are not translating.
Point to the section rather than copying. Copies drift out of date.
If a rule can be tested (a range, a date order), add an edit check. See edit checks.
Someone who will use it should read it before it is released.
Rollout
Version 1.0 approved by the data manager and the study lead, linked to a protocol version.
Send the CCG with the blank eCRF PDF export, which prints every approved form empty with a form index and visit-by-form matrix, so sites can read the instruction beside the form.
Walk site staff through the key conventions at the initiation visit or a short call. Keep the attendance and the version they received. Add staff to the delegation log.
Capture is the data system; keep the signed CCG in your investigator site file or study document system, not in the EDC.
After the first ten participants, review the most frequent queries and add or sharpen instructions. Issue a new version and ask sites to confirm receipt.
When a protocol amendment changes a form, update the CCG in the same change. See eCRF versioning mid-study.
Where Capture helps
Documents get filed; fields get filled. The most reliable completion guideline is the one built into the form: clear question wording, units, required flags and rules that raise a query the moment a value is wrong.
Blank eCRF · no subject data
Visit by form matrix
| Form | SCR | D1 | W2 | W4 | W8 | EOS |
|---|---|---|---|---|---|---|
| Demographics | X | |||||
| Vital signs | X | X | X | X | X | X |
| Laboratory | X | X | X | |||
| ECG | X | X |
3. Was the ECG performed?
☐ Yes☐ No
If No, complete question 4.
A study-specific document that tells site staff how to fill in each form, including conventions for dates, unknown or missing data, units, adverse events and corrections.
GCP expects data to be recorded consistently and accurately; a completion guideline is a common way to achieve that, though the format is yours. Your sponsor or CRO may have an SOP that requires one.
As short as possible. Cover general conventions and only the fields that cause confusion or queries, with examples.
Capture is the data system, not a document library. Distribute the CCG with the blank eCRF PDF export and file it in your investigator site file.
Usually the data manager, with input from the study lead and a site coordinator, approved by the data manager and the study lead. Your SOPs define roles.
The data management plan describes the data process for the study team. The completion guidelines are site-facing instructions for entering data.
Keep exploring
Free sandbox with every feature. No credit card, and you pay only when you go live with real participants.