Normal HbA1c but rising? Why the trend beats the threshold

Two people walk in with the same HbA1c: 5.6%. On paper they’re identical, and a standard panel waves them both through with the same word. Normal.

One of them has sat at 5.6 for a decade. The other was 5.0 three years ago, 5.3 last year, 5.6 now. That second person is on a slope, and the slope is pointed somewhere they don’t want to go — across the 5.7 prediabetes line within the year. Same number, different story. The threshold can’t tell them apart. The trend tells them apart in a half-second.

The line is the last thing to move

Most of medicine runs on thresholds. Fasting glucose over 126 mg/dL: diabetes. Blood pressure over 130/80: hypertension. HbA1c over 6.5%: a label switches on, a treatment switches on, and below the line you’re “fine.” These cutoffs are useful. They’re also the last thing to move.

By the time you trip a threshold, the drift underneath it has usually been running for years. A cutoff is a verdict on a process that’s mostly already happened, and the years of slow creep that got you there are exactly where the leverage was. A system that only asks whether you’re over or under the line is blind to all of it.

A single result tells you whether the building is on fire right now. It can’t tell you that someone has been walking toward the door for the last twenty minutes. Check once a year and that’s the trade: you’ll catch the fire, you’ll miss the walk to the door. Watching the slope is how you read the walk.

My own number talked me out of caring

I’ll be the cautionary example. My fasting insulin came back at the low end of normal a few years ago, I glanced at it, and I filed myself under “fine.” It was fine, as a single reading. What I didn’t do was look back. Two earlier draws would have shown the same marker drifting up draw over draw, inside the reference band the whole way, normal at every check. The band was never going to flag it. Only the line connecting the three points would have, and I wasn’t looking at the line.

That’s the trap with population reference ranges. They describe a crowd, and you’re an N of one. Your own previous draws are a better reference range than the band printed next to the result, because they’re measured on you. A reading inside normal that’s marching steadily toward the edge is a finding. The band will keep calling it nothing right up until the day it calls it a disease.

Two signals beat one

A trend on a single marker is good. A trend that lines up with a second signal is better, because that’s where you get the why, not just the what.

Here’s a real shape of it. Your fasting insulin has crept up across the last three draws, 6 to 9 to 12 µIU/mL. On its own, that’s a worrying line and not much else. Put it next to your fasting glucose, still flat and normal but holding its ground only because the insulin is working harder to keep it there. Put both next to your training log: the climb starts when you stopped lifting for a three-month stretch and your steps fell off. Now the line has an author: lost muscle and movement eroding insulin sensitivity, with glucose still papering over it. None of those three sources could have told you that alone. The insulin trend says something’s wrong; the three read together say what, and what reverses it: get back under a bar and the numbers start settling by your next draw.

Your watch sees one part of you, your ring another, your blood a third, and each, taken once and alone, is the weakest version of itself. The strongest version is all of them on one timeline, where a flat-looking number on one device gets explained by a moving one on another. A threshold can never make that move, because a threshold knows one marker and only knows today.

The fair objection

The honest counterargument: a trend is just noise that happened to line up. Three points that climb can be lab variation, a bad night’s sleep before the draw, a different time of day, the assay drifting. Read too much into a wiggle and you’ll chase ghosts and order panels you don’t need.

That’s a real risk, and it’s the reason to be disciplined, not the reason to give up. The fix is to confirm motion before you act on it: re-measure, hold the conditions steady, and ask whether a second, independent signal is moving the same way. A glucose trend that also shows up as a rising HbA1c and a rising fasting insulin isn’t three coincidences in a row. It’s one thing, seen three ways. Cross-source agreement is what separates a real slope from a wobble.

What to actually do with this

Act on the slope, not the line. A marker moving the wrong way fast deserves your attention before it’s ever called “abnormal,” because abnormal is the expensive end of a trip that was cheap to interrupt at the start.

Give direction its credit. A still-high number that’s falling is good news, and a threshold will never hand you that win, because all it sees is that you’re still over the line. The line is going the right way. That’s worth knowing, so you keep doing whatever’s bending it.

And confirm before you panic or celebrate. One point is a position. The slope is the direction, and the direction is what you act on.

So go back and connect the dots you already have. Pull your last three draws of one marker you care about and look at the line, not the latest value. If it’s drifting and you’re still inside the band, that’s not nothing. That’s the second person at 5.6%, three years out from 5.0%, with plenty of room left to turn. The point of catching the slope early is that the correction is still small.