The Alzheimer's Disease Assessment Scale, cognitive subscale, has been a primary cognitive endpoint in Alzheimer's trials for decades. Capture each task score, the word list used and the rater, so change from baseline holds up.
Free sandbox · No credit card · 21 CFR Part 11 aligned
Word recall (0 to 10)
Mean errors over three trials
6.3Orientation (0 to 8)
Word recognition (0 to 12)
Delayed word recall (0 to 10)
Number cancellation (0 to 5)
ADAS-Cog at a glance
The instrument
The ADAS-Cog was designed to measure the cognitive symptoms that change as Alzheimer's disease progresses: memory, language, praxis and orientation. Unlike the 30-point screens it has enough range to register change across a year or more of treatment, which is why it became the standard cognitive efficacy measure in Alzheimer's drug development.
Its weaknesses in early disease (floor effects on some tasks in people with only mild impairment) led to extended versions. ADAS-Cog 13 adds delayed recall and number cancellation; other variants add further executive tasks. Sponsors choose the version to match the disease stage of the population, and that choice has to be fixed in the protocol and statistical analysis plan.
Record every task score, not only the total. Task-level data lets data managers check each score against its allowed range at entry, supports sensitivity analyses on sub-domains, and makes source data verification against the worksheet straightforward. The total is then derived in analysis from the exported task scores.
Much of the variance in ADAS-Cog change scores comes from differences between raters and administration drift over time. Keep the same rater for a participant wherever possible, retrain periodically, and review rater attribution before interim and final analyses. Capture attributes every value to the person who entered it in the field-level audit trail.
Baseline
31.3
Week 24
33.1
Change
+1.8
8 participants with a rater change
3 participants with a missing task
Structure
All tasks are scored so that higher means more impairment. The first eleven make up ADAS-Cog 11; the last two are the ADAS-Cog 13 additions.
| Task | Range | Version |
|---|---|---|
| Word recall | 0 to 10 | 11 and 13 |
| Commands | 0 to 5 | 11 and 13 |
| Constructional praxis | 0 to 5 | 11 and 13 |
| Naming objects and fingers | 0 to 5 | 11 and 13 |
| Ideational praxis | 0 to 5 | 11 and 13 |
| Orientation | 0 to 8 | 11 and 13 |
| Word recognition | 0 to 12 | 11 and 13 |
| Remembering test instructions | 0 to 5 | 11 and 13 |
| Spoken language ability | 0 to 5 | 11 and 13 |
| Word-finding difficulty | 0 to 5 | 11 and 13 |
| Comprehension | 0 to 5 | 11 and 13 |
| Delayed word recall | 0 to 10 | 13 only |
| Number cancellation | 0 to 5 | 13 only |
Structure only. Obtain the administration manual, word lists and any required permissions for the version you use.
Build it in Capture
Each task is a numeric field with its allowed range as an edit check, so an impossible value raises an auto-query at entry. A dropdown records the word list and a date field records when the assessment was done; the person who entered each value is in the audit trail automatically.
Changed by Data manager · reason: critical safety value added
Schedule
Longer instruments run less often. A common pattern is the ADAS-Cog at baseline and every three to six months, with shorter screens at screening and functional and caregiver measures alongside.
| Assessment | Scr | BL | M3 | M6 | M9 | M12 | M18 |
|---|---|---|---|---|---|---|---|
| MMSE | |||||||
| ADAS-Cog 13 | |||||||
| Word list | A | B | C | A | B | C | |
| Functional scale | |||||||
| Caregiver report |
Letters show the word list used at each visit.
Set up the tasks, word list tracking and SDV in a free sandbox.
Data quality
Treatment effects on the ADAS-Cog are usually a few points against a background of steady decline, so small amounts of measurement noise can hide them. The main sources are well known and mostly controllable through study design and data review.
A participant assessed by different raters at baseline and follow-up picks up the difference between the raters as well as any change in cognition. Plan rater continuity, record who administered each assessment, and retrain raters periodically across long studies.
The word recall task is averaged across trials and several tasks are scored as errors rather than correct answers. Arithmetic and direction-of-scoring errors are common on paper worksheets. Range checks catch impossible values; source data verification of the tasks that drive the endpoint catches plausible but wrong ones.
A participant who is too fatigued to finish, or a task skipped by mistake, leaves a gap in the total. Record a reason when a task is not done (Capture exports distinguish not done from missing) and pre-specify whether totals can be computed with a missing task.
Before go-live
ADAS-Cog 11, 13 or another variant, stated in protocol and analysis plan.
Administration manual and word lists obtained; any licence recorded on the form.
Rotation schedule written down and recorded per visit.
Certification before first participant, with a retraining plan.
Impossible values queried at entry.
Verify the tasks that carry the endpoint against the worksheet.
Each task has its own range and task scores are summed. ADAS-Cog 11 totals 0 to 70 and ADAS-Cog 13 totals 0 to 85. Higher scores mean worse cognition.
Delayed word recall (0 to 10) and number cancellation (0 to 5), which improve sensitivity in milder disease.
Task scores. They can be range-checked at entry, verified against the worksheet and used in sensitivity analyses. The total is derived in analysis.
Different lists at successive visits reduce learning effects. Recording the list confirms the protocol rotation was followed.
Trained raters. Keeping the same rater per participant and retraining periodically reduces variability.
Building the form in the free sandbox is free. Obtain the instrument materials and any permissions separately.
Keep exploring
MMSE eCRF template
Staging and eligibility.
MoCA eCRF template
Screening mild cognitive impairment.
EDC for Alzheimer's trials
The full dementia trial build.
Audit trail software
Rater attribution on every value.
Sample size calculator
Plan the endpoint.
What is an endpoint?
Primary, secondary and exploratory.
Task-level capture, range checks and rater attribution. Free sandbox, no credit card.