The dog is imprinted on the scent signature of the handler's out-of-range blood glucose — primarily hypoglycemia — and alerts the handler (or a third party) so they can test and treat. Diabetic alert is certifiable detection work: the odour is collectable, imprintable, and blind-testable. A DAD complements — never replaces — glucose monitoring (CGM/meter).
6.1 Certifiable tasks
| Task | Trigger | Standard |
|---|---|---|
| Low-glucose alert ⚠ | Target odour (live or sample) | Persistent passive alert (e.g., nose-bump sequence, bringsel, paw touch) delivered to the handler until acknowledged |
| Night alert ⚠ | Target odour while handler asleep | Wakes the handler (licking, pawing, insistent nudging); escalates to a household member if trained and the handler doesn't respond |
| Third-party alert (recommended) | Handler non-responsive or cued | Finds and alerts the designated person |
| Glucose-kit retrieval (recommended) | Cue or chained to alert | Retrieves labeled kit/juice pouch to the handler's hand |
| High-glucose alert (optional) | Hyperglycemia odour | Distinct alert behaviour from the low alert |
6.2 Scent sample protocol
Sample quality is the foundation of the entire discipline; sloppy samples produce unreliable dogs.
- Collection: during a meter-confirmed hypoglycemic episode (typically < 4.0 mmol/L, per the handler's care plan), the handler swabs saliva or breath onto a cotton pad, or wears gauze against the skin; the sample is sealed immediately.
- Labeling: date, time, meter reading. Unlabeled samples are discarded.
- Storage: frozen in individually sealed containers; used within 6 months of collection.
- Controls: in-range samples from the same handler, collected and stored identically, serve as blanks/distractors — so the dog discriminates state, not person.
- Handling: tongs/gloves; hot and control samples never stored or transported together.
6.3 Training methodology requirements
- Imprint on hot samples → discrimination against in-range samples and distractors → alert-chain shaping (find → alert → persist until acknowledged → reward from handler).
- Persistence is trained explicitly: a single ignored nudge is not an alert; the dog must escalate.
- Handler-blind sessions before formal testing (third party stages sample lineups).
- Live-alert reinforcement protocol: every real alert is meter-verified and rewarded appropriately, building the live-alert log (§6.5).
6.4 Task test — discrimination protocol
Mirrors the Service Paws 15-presentation structure (Chapter 5) for consistency across detection disciplines.
Test 1 — Sample discrimination (blind) ⚠
- 15 presentations at 5-minute intervals: 7 hot samples, 8 blanks/controls (in-range samples and distractors), randomized, placed by a third party; handler and trainer blind to the order.
- Alert within 60 seconds of presentation; alert behaviour must be the trained chain, delivered to the handler.
- Pass: all 7 hots alerted; at most 1 false alert on blanks/controls; a re-presented item follows the ≤ 3 total attempts rule.
Test 2 — Alert chain and retrieval
- Simulated alert scenario ×3: dog performs the complete chain (alert → persist → acknowledged) each time.
- If kit retrieval is certified: 5 retrieval trials, ≥ 4/5, delivered to hand, packaging undamaged.
Test 3 — Night alert (if certified) ⚠
- Documented via the live-alert log (§6.5) with at least 2 meter-corroborated nighttime alerts, or a staged sleep-trial with a hot sample placed at the sleeping handler's bedside: dog wakes the handler within 5 minutes; 2/2 staged trials.
6.5 Live-alert log ⚠
Required for certification and renewals:
- Minimum 4 consecutive weeks of logging before the task test, with ≥ 10 documented alerts corroborated by a meter/CGM reading taken at alert time.
- Each entry: date/time, dog's behaviour, glucose reading, action taken.
- CGM users: export or screenshots referenced against log entries where practical.
- The log demonstrates real-world function; the blind test demonstrates the dog is alerting to odour. Both are required.
6.6 Honest-claims note
Real-world detection windows and sensitivity vary between dogs and days. Certification attests that this dog passed blind discrimination testing and has a corroborated live-alert record — not that it will detect every event. Attestation letters must use this framing, and handlers are informed in writing at issuance.
6.7 Maintenance
- Weekly sample refreshers (rotating fresh samples); monthly blind mini-lineups (logged).
- Ongoing live-alert log strongly recommended; required again for any re-assessment.
6.8 Records for this chapter
Discrimination scoresheet with placement key (Ch. 16); live-alert log; sample collection/storage notes; video when overseen remotely. On pass, Service Paws records the task testing date.