Biomarkers · Glycemic & Metabolic
Glycated Hemoglobin (HbA1c)
Both low and high values go with higher risk of death. In NHANES 1999–2018 with linked mortality to 2019, each standard deviation above average goes with 18% higher risk of death. Risk rises at both ends: not comparable
HbA1c and risk of death
NHANES, our fit with 95% CIPublished curve
Published curve higher than the NHANES fitNHANES fit higher
Risk of death relative to each curve's own reference value, so the two curves need not cross 1 at the same point. The dotted vertical line marks the NHANES median.
Risk of death per SD (NHANES)1.18×95% CI 1.16–1.21 · 49,067 adults, 7,174 deaths
Published risk of death1.52×lowest HbA1c decile (median 6.4%) vs decile with lowest risk of death (median 7.5%) (95% CI 1.32–1.76) · UK GPRD, two cohorts of patients aged >=50 with type 2 diabetes intensified to combination oral therapy (27,965) or insulin regimens (20,005), 1986-2008
Across studies, risk per SD1.18×NHANES 1.18 [1.16–1.21] · CRELES 1.20 [1.10–1.31]
HbA1c across cohorts
11.11.25HR per SD (95% CI)
NHANES · United States, 20–8449,067 people · 7,174 deaths1.18 1.16–1.21
CRELES · Costa Rica, 60+2,460 people · 470 deaths1.20 1.10–1.31
CHARLS · China, 50+7,811 people · 1,219 deaths1.16 1.09–1.24
HbA1c by age and sex
Each bar runs from the 5th to the 95th percentile, the darker part from the 25th to the 75th, and the tick is the median. Not a reference range or a target.
The percentiles as a table
| Sex | Age | 5th | 25th | Median | 75th | 95th | n | Cycle / source |
|---|---|---|---|---|---|---|---|---|
| All | 20-39 | 4.70 | 5.00 | 5.30 | 5.50 | 6.00 | 1,452 | NHANES 2017-2018 |
| All | 40-59 | 5.00 | 5.30 | 5.50 | 5.80 | 7.60 | 1,583 | NHANES 2017-2018 |
| All | 60+ | 5.10 | 5.50 | 5.74 | 6.20 | 7.80 | 1,936 | NHANES 2017-2018 |
| All | all | 4.90 | 5.20 | 5.50 | 5.80 | 7.40 | 4,971 | NHANES 2017-2018 |
| Men | 20-39 | 4.70 | 5.10 | 5.30 | 5.50 | 5.90 | 697 | NHANES 2017-2018 |
| Men | 40-59 | 4.90 | 5.30 | 5.60 | 5.90 | 7.97 | 739 | NHANES 2017-2018 |
| Men | 60+ | 5.10 | 5.50 | 5.80 | 6.40 | 7.80 | 979 | NHANES 2017-2018 |
| Men | all | 4.90 | 5.20 | 5.50 | 5.80 | 7.40 | 2,415 | NHANES 2017-2018 |
| Women | 20-39 | 4.74 | 5.00 | 5.20 | 5.50 | 6.10 | 755 | NHANES 2017-2018 |
| Women | 40-59 | 5.00 | 5.30 | 5.50 | 5.80 | 7.40 | 844 | NHANES 2017-2018 |
| Women | 60+ | 5.10 | 5.50 | 5.70 | 6.10 | 7.70 | 957 | NHANES 2017-2018 |
| Women | all | 4.90 | 5.20 | 5.50 | 5.80 | 7.20 | 2,556 | NHANES 2017-2018 |
What studies found when this marker was changed
They describe what happened in those study groups, not a recommendation. See all catalogued interventions.
Measured changes in HbA1c, study by study
8 effects with a numeric size, each from one study arm: the dot is the reported change, the line its 95% interval where one was given. Study results in the groups studied, not predictions for a person.
| Type | Intervention | Reported effect | Evidence | Source |
|---|---|---|---|---|
| behavioral | Digital health technology (apps/connected meters) combined with human coaching for diabetes self-management | HbA1c -0.31% (95% CI -0.45 to -0.16) vs usual care; high-intensity coaching -0.45%, low-intensity -0.28% (n.s.) | meta-analysis | Kerr D, et al. J Med Internet Res. 2024. Digital Interventions for… |
| clinical | Real-time or flash continuous glucose monitoring instead of fingerstick monitoring in adults with type 2 diabetes | HbA1c reduced by about 0.32% (rt-CGM 0.34%, flash CGM 0.33%); moderate certainty | meta-analysis | Uhl S, et al. J Clin Endocrinol Metab. 2024. Effectiveness of Continuous… |
| diet | Low-carbohydrate / Ketogenic dietary intervention | Reduced HbA1c by 0.29 % vs control diets in T2D (meta-analysis of RCTs), largest at 3 months and attenuating thereafter | meta-analysis | Mongkolsucharitkul P, et al. Effectiveness of low-carbohydrate diets on type 2… |
| diet | Eating only between 12 pm and 8 pm daily without calorie counting for 6 months in adults with obesity and type 2 diabetes | HbA1c -0.91% (95% CI -1.61 to -0.20) relative to controls at 6 months (daily calorie restriction -0.94%) | RCT | Pavlou V, et al. JAMA Netw Open. 2023. Effect of Time-Restricted Eating on… |
| diet | Two non-consecutive meal-replacement fasting days per week plus 5 days habitual intake for 16 weeks in newly diagnosed overweight/obese type 2 diabetes | HbA1c LSM change -1.9% at 16 weeks, vs -1.6% with metformin and -1.5% with empagliflozin (adjusted differences -0.3% and -0.4%) | RCT | Guo L, et al. JAMA Netw Open. 2024. A 5:2 Intermittent Fasting Meal Replacement… |
| exercise | Combined aerobic and resistance training (3x/week) | Combined training lowered HbA1c by 0.34% (95% CI -0.64% to -0.03%) vs control over 9 months | RCT | Church TS, et al. JAMA. 2010;304:2253-2262. |
| exercise | Structured aerobic exercise training programmes of at least 12 weeks in adults with type 2 diabetes | HbA1c -0.73% (95% CI -1.06 to -0.40) with structured aerobic exercise; any structured training -0.67%; >150 min/week -0.89% | meta-analysis | Umpierre D, et al. JAMA. 2011. Physical activity advice only or structured… |
| lifestyle | Intensive lifestyle intervention (Look AHEAD) in type 2 diabetes | Mean HbA1c fell from 7.3 to 6.6% at 1 year vs 7.3 to 7.2% with diabetes support and education | RCT | Look AHEAD Research Group, et al. Reduction in weight and cardiovascular… |
| pharmacologic | GLP-1 Receptor Agonists (Semaglutide) or dual GIP/GLP-1 agonist Tirzepatide | Vs placebo, tirzepatide lowered HbA1c by 2.3% and semaglutide by 1.5%; body weight -9.1 kg and -2.8 kg (network meta-analysis, 64 RCTs) | meta-analysis | Ren X, et al. Efficacy and safety of GLP-1 agonists in the treatment of T2DM: A… |
| pharmacologic | Dual GIP/GLP-1 receptor agonist (Tirzepatide 5-15mg) | HbA1c -2.01, -2.24 and -2.30 percentage points with 5, 10 and 15 mg at 40 weeks (vs -1.86 with semaglutide 1 mg) | RCT | Frias JP, et al. N Engl J Med. 2021;385:503-515. |
Things that worsen it
| Type | Condition or exposure | Reported effect | Evidence | Source |
|---|---|---|---|---|
| clinical | Progressive beta-cell failure and unmanaged insulin resistance | HbA1c began to rise after 1.5 years with rising insulin needs and declining beta-cell reserve (effect size not reported) | prospective cohort | Levitt Katz LE, et al. Glycemic control in youth with type 2 diabetes declines… |
| diet | Chronic high-glycemic ultra-processed dietary pattern | Each one-point higher UPF score associated with 0.418% higher HbA1c (95% CI 0.333-0.503) in diabetic adults | cross-sectional | Otay Lule N, et al. Ultra-Processed Food Consumption Is Independently… |
Sources
- Risk curve: Currie CJ 2010 (PMID 20110121): all-cause HR 1.52 (1.32-1.76) lowest HbA1c decile (median 6.4%) vs decile with lowest risk of death (median 7.5%); highest decile (median 10.5%) HR 1.79 (1.56-2.06). Scaled to US values (NHANES).
- Currie CJ, et al. Survival as a function of HbA1c in people with type 2 diabetes: a retrospective cohort study. Lancet. 2010;375(9713):481-489.
- NHANES curve: our Cox model adjusted for age and sex, NHANES 1999–2018 with linked mortality to 2019.
Map Glycated Hemoglobin (HbA1c) against another biomarker · Featured: HbA1c × body mass index.
Related markers: Fasting Plasma Glucose · Fasting Serum Insulin · Homeostatic Model Assessment of Insulin Resistance (HOMA-IR) · Proinsulin