EDC for diabetic retinopathy clinical trials
Retinal trials are run eye by eye: one study eye, a fellow eye, graded photographs, letter scores and injection procedures. Capture keeps each as its own dated, audit-trailed record. Build and test free, no credit card.
- Study eye and fellow eye
- DRSS, BCVA and OCT fields
- Masked roles
Free sandbox · No credit card · 21 CFR Part 11 compliant
Visual function
Study eye
EYEBCVA, study eye
BCVACentral subfield thickness (OCT)
CSTDRSS level from reading centre report
DRSSEntered from the report, not graded here
Post-injection IOP
IOPPIAbove the range you set: confirm and record action
What a diabetic retinopathy trial needs from its EDC
- Eye-level data: laterality is part of every ophthalmic record. One participant can contribute a study eye and a fellow eye, and a wrong-eye entry is a classic, avoidable error.
- A severity scale graded from photographs: registered trials in non-proliferative disease commonly use a proportion of eyes with a two-step or greater improvement on the Diabetic Retinopathy Severity Scale (DRSS) at about one year. Check current regulator guidance for your design.
- Function and structure side by side: best-corrected visual acuity (BCVA) in ETDRS letters and central subfield thickness on OCT are usually collected at every visit, whether as secondary or as exploratory measures.
- Procedure records: intravitreal injection and sham visits need a lot, time, post-injection pressure and complication record, kept in a form masked staff do not see.
- Systemic diabetes context: HbA1c, blood pressure, renal function and diabetes duration shape eligibility and the interpretation of results.
Endpoints and data
Where the primary endpoint actually comes from
In diabetic retinopathy the key outcome is not a number a nurse reads off a device. Colour fundus photographs, and often widefield imaging or fluorescein angiography, are sent to a central reading centre, where trained graders assign a DRSS level to each eye. Published phase 2 and 3 programmes in non-proliferative disease have used the share of study eyes improving by at least two steps at around week 52, with worsening events such as progression to proliferative disease or centre-involving macular oedema tracked as key secondaries. A sponsor slide from one programme says the FDA accepted a two-step DRSS change as a pivotal endpoint, but we could not retrieve the guidance itself, so treat that as a lead to verify rather than a rule.
What this means for the EDC: the graded level arrives as a report, not a measurement. Capture does not read images, and no reading-centre integration is claimed. Your options are for staff to enter the level, date and image-quality flag from the report into a dedicated form, or to load the reading-centre file through your own data transfer process. Either way, keep the image date, the grading date and the grader batch as separate fields so a late or ungradable image shows up as a gap you can query, not a silent blank. In Capture, forms move from draft to approved and are then locked for live use, and every change keeps an audit record with the old value, new value and reason. See source data verification for deciding which fields get checked against source.
Function is captured the traditional way. BCVA is a letter count from a standardised chart at a fixed distance after refraction, and many protocols require certified examiners. Record the examiner role, the refraction date and the test distance so changes in method are visible in the export. Central subfield thickness from OCT belongs on the same form, with a field for the device model if your protocol allows more than one. A numeric range check can flag a letter score above the chart maximum or a thickness that is implausible, which saves a query round later. For the general idea of endpoint hierarchies, see clinical trial endpoints.
Study eye, fellow eye and the unit of analysis
Most retinal protocols name one study eye per participant and treat the fellow eye as a safety and context observation. If both eyes are eligible, the protocol must say how the study eye is chosen (worse eye, investigator choice, randomised), and that decision belongs on the screening form. Enter the study eye once at screening, and use skip logic so every later ophthalmic form asks about it in the same way. If the analysis is by participant rather than by eye, make that explicit in the data dictionary the programmer receives.
| Assessment | Screen | Day 1 | Wk 4 | Wk 8 | Wk 16 | Wk 28 | Wk 52 |
|---|---|---|---|---|---|---|---|
| Informed consent | |||||||
| BCVA and refraction | |||||||
| OCT thickness | |||||||
| Fundus photographs for DRSS | |||||||
| Study treatment or sham | |||||||
| HbA1c, BP and labs |
x = done at this visit. Illustrative only: visit spacing and imaging frequency come from the protocol.
Protocol to build
Retinopathy protocol elements and how they map to Capture
| Protocol element | What the data looks like | Where it lives in Capture |
|---|---|---|
| Study eye selection | Right or left, rule used to choose | Screening form, then skip logic on every ophthalmic form |
| DRSS grade | Level, image date, grading date, ungradable flag | Reading-centre form entered from the report; edit check on allowed levels |
| BCVA | ETDRS letters, distance, examiner, refraction date | Visit form with numeric range check |
| Anatomy | OCT central subfield thickness, device | Same visit form; range check |
| Injection or sham procedure | Lot, time, eye, IOP after, complications | Repeating row per procedure, in a form masked roles cannot open |
| Rescue or laser | Date, type, reason | Concomitant medication style log, plus a procedure form |
| Systemic status | HbA1c, BP, renal function | HbA1c log, vital signs, lab results |
| Ocular adverse events | Eye, onset, severity, relationship | Adverse event form with an eye field |
Capture does not grade images or calculate DRSS. It holds the dated record that the reading centre report produces.
Build the study eye form and test it with mock participants
Create the ophthalmic visit form, enter fake right and left eyes, and check the masked view. Free sandbox, no credit card.
Masking and safety
Keeping injection and sham records away from the masked team
Many retinopathy trials compare an intravitreal agent with sham injection, so the examiner measuring BCVA and the person reading OCT must not learn which arm a participant is in. The injecting clinician is unmasked and works from a separate record. Capture supports blinded roles that never receive treatment-arm values, enforced in the database through masked views and row-level security, not just hidden in the interface. Use one form for the procedure and give access only to the unmasked role. The masked team still sees that a visit happened and its outcome measures. See blinded and unblinded role management.
Post-injection safety is time-bound: intraocular pressure shortly after the injection, intraocular inflammation, endophthalmitis signs, retinal detachment, and vitreous haemorrhage are the events investigators watch for. A range edit check on post-injection pressure raises an automatic query when a value crosses the threshold you set, so a high reading is not left without an action recorded. It is a data prompt, not a clinical alert; clinical response follows the protocol. Serious ocular events flow into the serious adverse event workflow, with the SAE report form template as a starting point.
Participants with diabetes are often older and have comorbidity, so consent and visit burden deserve thought. eConsent with on-screen signature and email confirmation can be completed at home before the screening visit, and long-term follow-up studies can use the long-term follow-up consent template. A patient-reported vision questionnaire, if your protocol includes one, is a licensed-instrument question: the licence stays with your study documents, and the sandbox is free to test the layout.
Build sequence
From retinal protocol to a running study
- 1
List the ocular measures
For each visit, write which eye, which measure, which device and which role collects it. Mark every item that comes from a reading centre report.
- 2
Draft the forms
Upload the protocol (PDF, DOCX or DOC) and let the AI study builder propose visits and forms. Nothing is saved until a person reviews it.
- 3
Fix the eye logic
Capture study eye once at screening and reuse it with skip logic so later forms cannot silently switch eyes.
- 4
Add edit checks
Range checks on letters, thickness, post-injection pressure and HbA1c, and a date-order rule so a grading date cannot precede the image date.
- 5
Separate masked and unmasked access
Put procedures in a form only the unmasked role opens, then test the masked view with a dummy participant.
- 6
Test, approve and go live
Enter completers, an ungradable image, a rescue laser and an early withdrawal, then export. Approved forms lock for live use.
Before first participant
Diabetic retinopathy readiness checklist
Study eye rule written down
Screening form records the eye and how it was chosen; later forms reuse it.
Reading-centre data path agreed
Who enters or loads DRSS results, with image date, grading date and ungradable flag.
Examiner qualification fields
Role and refraction date captured for BCVA so method changes can be seen.
Masking tested
A masked-role login checked against the unmasked procedure form.
Rescue therapy defined
Laser, injection and surgery recorded with date and reason for the analysis rules.
Export reviewed
CSV with data dictionary opened by the statistician, or SDTM with Define-XML if needed.
Can Capture grade fundus photographs or calculate DRSS?
No. Capture does not read images and no reading-centre integration is claimed. Your team enters or loads the graded level and dates from the reading centre report, and the audit trail records every change.
How do I handle two eyes per participant?
Record the study eye once at screening and use skip logic so every later form refers to it. If fellow-eye safety data are collected, give them their own repeating rows with an eye field.
Can examiners stay masked to treatment?
Yes. Blinded roles never receive treatment-arm values, enforced by masked database views. Put the injection or sham record in a form only the unmasked role can open.
Can it flag a high post-injection eye pressure?
An edit check with a range rule you define raises an automatic query when a value crosses it. It is a data prompt, not a clinical alert.
Is a diabetic macular oedema trial different?
The tools are the same, but the primary endpoint is more often BCVA change, so letter-score quality and examiner method matter more than the severity scale.
Can I get SDTM datasets?
Yes. Capture exports CDISC SDTM as SAS XPT files with Define-XML, alongside CSV and Excel with an automatic data dictionary.
How much does it cost?
The sandbox is free with nearly every feature, no credit card. Plans and prices are on the pricing page.
Keep exploring
Related pages
EDC for diabetes trials
Glycaemic endpoints and systemic data.
EDC for myopia control trials
Another eye-level study design.
EDC for endocrinology trials
Metabolic umbrella page.
IOP and visual field template
Eye-level pressure and field fields.
Blinded and unblinded roles
Keep the examiner masked.
Adverse event template
Severity, relationship and sign-off.
Pricing
Free sandbox.
Start your retinopathy study free
Free sandbox with nearly every feature. No credit card.