The app said 6.4%, the lab 7.1% — which one is wrong?

The screen said 6.4%; the paper in my hand, 7.1%. Same week, same person. I looked for a bug, then saw two kinds of number.

A lab HbA1c measures how much of the hemoglobin carries attached glucose. It builds across the lifespan of red blood cells, measured or not.

The estimate is calculated, not measured. ForMyGluco runs the ADAG equation in reverse, solving for an A1C from the 90-day window average, only with at least 10 readings on 7 different days. That regression was derived to run the other way: A1C to average glucose (Nathan et al., 2008).

GMI, a formula the app does not use, is built for continuous glucose monitoring. Bergenstal and colleagues compared it with laboratory A1C in 528 people from four randomized trials on one manufacturer’s sensors: they differed by 0.3 A1C points or more in 51% of these people, by 0.5 or more in 28% (Bergenstal et al., 2018).

Both averages were measured densely: in ADAG, seven-point fingerstick profiles at least three days a week; in Bergenstal, around the clock. The readings I enter are sparser and not random: for months, mostly on waking; most of the day never entered my average.

The lab has limits too: ADA Standards of Care 2026, Section 6, says conditions affecting red blood cell turnover can interfere with A1C accuracy.

On the HbA1c screen the label reads ESTIMATE · 90 DAYS, beside the difference from a lab result under 120 days old; the estimate does not replace the lab result.

Questions for an appointment: does this gap mean anything to you? At which hours would my readings tell you more? Could anything in my blood work affect A1C? Interpretation and targets are the doctor’s.