Instrument · rev 7.63

CGM + glucometer

Both instruments are compromised. Fix the reference before you trust the sensor.

Do this today — depends on nothing

Alert	Now	Should be
Urgent Low Soon · fires ~20 min before a projected <55	OFF	ON
Falling Fast	OFF	ON
Urgent Low	55	55 ✓
High	250	200 — inert at 250; make it a canary

Zero cost. Zero risk. The low alerts stay on as cheap insurance — asymmetry favors keeping them.

The rationale — and its limit

C-peptide 1.4 (fasting, June 11 2026 — pre-Frey) with GMI 5.7%. Retained endogenous insulin secretion plus ⟨assumed⟩ impaired glucagon counterregulation would be the configuration that produces a low you cannot sense or self-correct. (Reconciled from the stale 2.1, which was the June 2025 value; 1.4 is still in range, so "retained secretion" holds, but nearer the floor.)

⚠ But glucagon-impairment is UNMEASURABLE — Garbacz 7/15: never tested, no routine clinical test exists. The low alert stays on regardless; do not describe this as established physiology.

Anatomy note — corrected, then demoted

The procedure was an open Frey (head cored out), not a Puestow. The Frey resects the alpha-poor head/uncinate and spares the alpha-rich body and tail — so alpha mass is largely preserved.

But spared mass ≠ preserved function. Kumar 2018 (PMID 30029953, n=30): beta-cell function unchanged at 3 months, alpha-cell function deteriorated post-Frey, independent of mass. So the anatomy line is directionally reassuring, not dispositive. What actually falsifies the hypoglycemia-caution premise is the 180-day behavioral record — zero confirmed real hypos on four glucose-lowering botanicals — not the anatomy.

The pairing protocol is contaminated

The reference standard, not the sensor

You are pairing the G7 against a home glucometer at Hct 25.2. At low hematocrit these meters over-read — fewer red cells impede the reaction, so glucose reads falsely high.

The chain: your stick reads high → the sensor looks low → you conclude a negative sensor bias → you gate four reintroductions on an error that lives in the reference standard.

✅ RESOLVED 8/3 — METER VALIDATED. Now on the Accu-Chek Guide, labeled hematocrit range 10–65% (Roche spec) — wider than the retired 20–60% meter and clearing Hct 25.2 with real margin. Accuracy standard is Accu-Chek's 10/10 rule. The G7 is now the instrument being checked and the meter is the reference — the correct direction. Every gate that named "a validated CGM" is released.

Five-day protocol

Fresh sensor. Discard day 1 and day 10 permanently. Warm-up and end-of-wear are known low-fidelity periods.
Stop calibrating. The G7 requires zero fingersticks. Manual calibrations launder fingerstick error into the sensor trace.
Pair, don't correct. Sensor first, then stick. Enter nothing into the app.
Pair at ~4:30 AM and ~5:45 PM — and whenever the sensor reads below 70. Bias is not constant across the range, and the low range is the one that matters.
One meter, one strip lot, record Hct with every pair.

Interim glycemic instrument

HbA1c	Do not use	Falsely low post-transfusion. Wait ~10–12 weeks.
Fructosamine	Use	2–3 week mean, RBC-independent. Invalid below albumin 3.0; yours is 4.0, so it holds. Draw simultaneous albumin.

The 180-day aggregate survives: GMI 5.7% and 5.8% across two windows, zero readings above 250.

CGM + glucometer · rev 7.63 · 2026-09-04 · content unchanged since rev 7.35 · METER VALIDATED — Accu-Chek Guide, Hct 10–65%