The Glucose-Mortality Curve
The scariest finding in glucose epidemiology is also the most useful: mortality risk doesn't wait for the diabetes diagnosis. It begins climbing while your numbers are still "normal" — somewhere north of 100 mg/dL fasting, or 5.7% HbA1c, the curve bends upward. This page walks the actual pooled-cohort data, explains what each number catches, and then spends equal effort on the part most headlines skip: how to read those curves without turning one lab value into a personal forecast.
What the evidence supports
- Mortality risk begins to rise once fasting glucose climbs above roughly 100 mg/dL (5.6 mmol/L) — well below the diabetes threshold.
- HbA1c tracks cardiovascular risk in a graded way across the entire "normal" range, not just above it.
- Two-hour post-load glucose catches people at risk whom fasting glucose misses entirely.
What remains uncertain
- These are observational curves — people with higher glucose differ in many other ways, and adjustment can never fully isolate glucose's own effect.
- Whether treating mildly elevated glucose in the absence of diabetes changes mortality is a live question; the risk curve is not the same thing as a treatment target.
- Very low fasting glucose also associates with higher mortality in cohorts — a marker of illness and frailty, not evidence that lower is always better.
Evidence last reviewed: August 15, 2026. Conclusions may change as new research is published.
where the risk line bends
The Shape of the Curve
Plot mortality against fasting glucose across a population and you get a J: lowest risk in the normal range, a gentle climb that starts inside "normal," and a steep rise through prediabetes into diabetes. The shape matters more than any single point on it, because it tells you there is no cliff — no threshold where health suddenly becomes disease. There is a slope, and you want to be on the flat part of it. The two surprises in the data are where the slope starts (earlier than most people assume) and what happens at the very bottom of the curve (very low glucose marks illness rather than causing it — the J's other arm).
Fasting Glucose: The Pooled Numbers
The reference dataset for this question pooled 820,900 adults from 97 cohorts without baseline diabetes and followed them for over twelve million person-years (Emerging Risk Factors Collaboration, NEJM, 2011). The headline finding, stated carefully: mortality — especially vascular death — began rising once fasting glucose climbed above about 100 mg/dL (5.6 mmol/L), long before any diabetes diagnosis. The increases were modest but real in the 100–125 band and grew steeper with each step. Because the analysis adjusted for age, sex, smoking, blood pressure, and cholesterol, the slope is not simply "people with high glucose also have high blood pressure" — though no adjustment is perfect, and the residual-confounding caveat still applies.
| Fasting glucose band | Diagnostic meaning | Vascular mortality vs 70–99 mg/dL | Read |
|---|---|---|---|
| < 70 mg/dL | Below reference range | Some cohorts show higher risk — mostly marking illness and frailty | Mixed |
| 70–99 mg/dL | Normal | Reference (hazard ratio 1.0) | Reference |
| 100–125 mg/dL | Impaired fasting glucose | ~11% higher in the pooled analysis (HR 1.11) | Modest rise |
| 125–140 mg/dL | Upper impaired range | ~38% higher (HR 1.38) | Clear rise |
| ≥ 140 mg/dL | Diabetic-range on repeat testing | ~86% higher (HR 1.86) | Steep rise |
Two honest glosses on the table. First, these are hazard ratios for vascular mortality — the strongest and most consistent signal in the data; non-vascular causes showed weaker, noisier associations. Second, the numbers describe groups, not individuals: an 11% higher group average is a modest effect size, and it says nothing about any single person's fate. The curve is a reason to prefer the flat part — not a forecast.
HbA1c: The Slow-Moving Signal
Fasting glucose is a snapshot; HbA1c is a three-month average, and its relationship with risk is if anything smoother. In the ARIC cohort of over 11,000 adults without diabetes, cardiovascular risk climbed in step with HbA1c across the entire sub-threshold range — each band above 5.0% carried more risk than the last, and the association survived adjustment for fasting glucose and standard risk factors (Selvin et al., NEJM, 2010). By the time HbA1c reaches the 5.7% prediabetes line, the slope is already visible behind it. The practical consequence: a "normal" HbA1c of 5.6% is not a clean bill of health in the same way 5.1% is — the continuum extends all the way down, which is exactly why the quarterly audit tracks the trend across years rather than checking a pass/fail box once.
The Two-Hour Number They Usually Skip
The least-used measurement in this story is arguably the most informative. The DECODE collaboration pooled 22 European cohorts and found that two-hour post-load glucose — the value after a standardized 75-gram drink — predicted mortality better than fasting glucose did, and it identified people at elevated risk whose fasting numbers were entirely normal (DECODE Study Group, Lancet, 1999). The mechanism is intuitive: fasting glucose reflects the overnight liver trickle, while the two-hour number reflects the whole system — absorption, insulin response, muscle uptake — under real load. It is the canary for the process the insulin-resistance topic tracks in detail, and it is the reason that topic treats "normal fasting glucose" as reassuring but incomplete. You can ask for an oral glucose tolerance test; many clinicians reserve it for specific situations, which is worth a conversation rather than a demand.
⚠️ These are population curves, not personal forecasts
A hazard ratio describes groups. It cannot tell you what your 104 mg/dL means for your future — most people with mildly elevated glucose never develop diabetes, and risk is modifiable: the strongest evidence for bending an individual's curve comes from the lifestyle trials the insulin-resistance topic documents. The sane reading order is: know your numbers, note the trend, act on the levers — and keep the epidemiology in its place as motivation, not as prophecy. Single lab values also bounce: judge trends across measurements, not any one draw.
What Moves the Curve
- 🛏️ Sleep holds the baseline down. One short night measurably impairs next-day glucose handling — the Sleep pillar's repair science shows the overnight shift of this very curve.
- 🏋️ Muscle is the drain. Every unit of contractile tissue is glucose-disposal capacity; training keeps the post-meal wave small and brief. The walking topic and the strength topics own the dose-response.
- 🍽️ Meal architecture shapes the wave. Protein-first sequencing and post-meal movement are the two cheapest curve-flatteners — the next two pages in this series.
- 🔑 Early action beats late. The intervention trials that changed outcomes worked in prediabetes, not after the diagnosis — the decade-before-diabetes logic of the insulin-resistance topic.
- 📉 And some things are not levers. Chasing a flat sensor curve, or pushing fasting glucose from 90 to 80 with obsessive means, moves nothing on the population curve — the J's left arm is not an invitation.
Questions, Answered Briefly
- 😟 My fasting glucose is 98 — should I worry? No. It sits on the flat part of the curve. What deserves attention is the trend: 98 this year, 96 last year, 102 next — that's when the audit conversation starts.
- 🧪 My HbA1c is 5.6 — is that normal? Diagnostically yes — below the 5.7% prediabetes line. Epidemiologically it sits where the risk slope is already gently positive, which is why the trend across years matters more than the single value.
- 📐 Which number matters most? They catch different things: the two-hour value catches the earliest trouble, HbA1c averages it, fasting glucose is the cheapest screen. If they disagree, the higher-risk reading is the honest tiebreaker — and the OGTT settles it.
- 📉 Is lower always better? No — very low fasting glucose in cohorts associates with higher mortality, largely because serious illness and frailty drive glucose down. Low is not a target; the flat part is.
- 🏃 Can I move my numbers without medication? For the prediabetic range, the lifestyle-trial evidence is among the strongest in metabolic research — the insulin-resistance topic documents it, and the levers are the same ones on this page.
The Bottom Line
- The curve is a slope, not a cliff — vascular risk starts climbing near 100 mg/dL fasting, with roughly 11% higher risk in the 100–125 band and steeper steps after.
- HbA1c risks rise across the whole "normal" range — a graded continuum, not a threshold, which is why trends beat single values.
- The two-hour number catches what the others miss — post-load glucose predicted mortality better than fasting glucose in pooled cohorts.
- Read the curve as motivation, not prophecy — risk is modifiable, group statistics are not personal forecasts, and most people with mild elevations never develop diabetes.
Related Topics
- Emerging Risk Factors Collaboration (Seshasai et al.), "Diabetes mellitus, fasting glucose, and risk of cause-specific death," New England Journal of Medicine (2011)
- Selvin et al., "Glycated hemoglobin, diabetes, and cardiovascular risk in nondiabetic adults," New England Journal of Medicine (2010)
- DECODE Study Group, "Glucose tolerance and mortality: comparison of WHO and ADA diagnostic criteria," The Lancet (1999)
- American Diabetes Association, "Standards of care in diabetes — classification and diagnosis" (2024)
- Centers for Disease Control and Prevention, "National Diabetes Statistics Report" (2023)