Glucose, HbA1c and tests for diabetes

A glucose result is only a number until you know what it was taken to show. This page explains the common tests used to diagnose and monitor diabetes: fasting and random plasma glucose, the oral glucose tolerance test (OGTT), haemoglobin A1c (HbA1c), capillary ketones, C-peptide, urine ketones, fructosamine and time in range. Each one measures a different aspect of glucose control, and each has its own strengths and limitations. The reference ranges, thresholds and first-line statements quoted here come from current WHO, ADA and NICE publications; where guidelines differ, both values are stated.

1. State the WHO and ADA thresholds that define normal glucose tolerance, prediabetes and diabetes, on fasting glucose, the OGTT and HbA1c.
2. Explain what HbA1c reflects, how its two reporting units relate, and name the conditions that make it unreliable.
3. Interpret a result sheet: which test to trust when results disagree, and which confirmatory test is needed.
4. Recognise patterns such as isolated impaired fasting glucose, insulin deficiency and ketosis, and say what each suggests.
5. State when capillary ketones, C-peptide, urine ketones, fructosamine and time in range are the right test to use.

A diagnosis of diabetes should rest on a test of plasma glucose or HbA1c, interpreted against a laboratory method that has been validated for that purpose. One abnormal result is enough in a symptomatic person; in an asymptomatic person the result must be confirmed on a separate day. ^[1]

Fasting plasma glucose (FPG) is measured after at least 8 hours without food or drink other than water. It reflects the liver's overnight glucose production and the basal insulin level. ^[2] Random (casual) plasma glucose is any sample taken without regard to the last meal; it is diagnostic only when symptoms of diabetes are present. ^[1] Whole-blood glucose (for example from a meter) runs lower than plasma glucose, so the whole-blood diagnostic threshold is lower. ^[1]

The OGTT measures the plasma glucose response to a standard glucose load. The person fasts overnight, then drinks 75 g of anhydrous glucose dissolved in water; venous plasma glucose is measured at baseline and at 2 hours. It is the most sensitive of the glucose tests for diabetes but is the most cumbersome and the least reproducible, so it is reserved for cases where fasting glucose or HbA1c is inconclusive (for example in pregnancy or suspected gestational diabetes). ^[1]

HbA1c is the percentage of haemoglobin that has glucose attached. Because erythrocytes live about 120 days, the value reflects average glucose over the preceding 2 to 3 months. ^[5] It has three advantages: the person does not need to fast, there is no glucose load, and the sample can be taken at any time. ^[2] Its disadvantages are that it is affected by anything that changes red-cell lifespan or haemoglobin type, and that it changes slowly, so it cannot show sudden improvement or deterioration in control. ^[3][5]

HbA1c is reported in two units. Percent (%) is the older NGSP/DCCT unit still used in the United States. Millimoles per mole (mmol/mol) is the standardised IFCC unit used in the United Kingdom and most of Europe. The two scales are related by a fixed equation; clinically, 5.7% equals 39 mmol/mol, 6.5% equals 48 mmol/mol and 7.0% equals 53 mmol/mol. ^{[2][1] [3] [4]}

Capillary (fingertip) ketone meters measure beta-hydroxybutyrate, the ketone body that dominates in diabetic ketoacidosis. A venous sample in the central laboratory or a capillary point-of-care sample can be used, and point-of-care testing is faster. ^[4] Urine ketone strips measure acetoacetate, which is present in smaller amounts in DKA and is overestimated during resolution, so a blood result is preferred when available. ^[4]

C-peptide is cleaved from proinsulin in equimolar amounts with insulin, so its concentration reflects endogenous insulin secretion. ^[6] A random or fasting sample can usually be used in clinical practice if the timing relative to food and the concurrent glucose are recorded; formal stimulation tests (mixed-meal tolerance or glucagon stimulation) are more accurate but slower. ^[6] C-peptide is low or absent in absolute insulin deficiency and raised in insulin resistance. ^[6] It cannot be interpreted without the concurrent glucose, and it is affected by renal clearance. ^[7]

Urine ketone dipsticks detect acetoacetate semiquantitatively. They are cheap, widely available and remain useful where blood ketone testing is not, but they lag behind blood results and can be affected by drugs such as captopril and valproate. ^[4] A result of 2+ or more on a standard urine stick counts as significant ketonuria. ^[4]

Fructosamine is glycated serum protein, mainly albumin. Because albumin turns over faster than erythrocytes, it reflects average glucose over the preceding 2 to 3 weeks. ^[5] It is therefore useful for detecting recent change in control and for patients in whom HbA1c is invalid. ^[3]

Time in range is the percentage of CGM (continuous glucose monitor) readings that fall between 70 and 180 mg/dL (3.9 to 10.0 mmol/L). It gives a day-to-day picture that HbA1c cannot, because it also reports time below and above range and glucose variability. ^{[8] [4]}

The ranges below are the adult cut-offs quoted in the sources used for this page. They are not universal: different laboratories use different assays, and age, sex, pregnancy, altitude, fasting, posture and time of day all move values. The interval printed on the laboratory report is the one to use. ^[2][5]

Values outside the reference interval can be explained by the conditions below. A high or low result is a reason to look again, not a diagnosis on its own. ^[2]

Named patterns make it easier to reason from results rather than memorise cut-offs.

• Isolated impaired fasting glucose: fasting glucose 5.6-6.9 mmol/L with a normal 2-hour value and HbA1c below 48 mmol/mol. It predicts progression to diabetes; lifestyle intervention and follow-up are advised. ^[1][2]
• Isolated impaired glucose tolerance: normal fasting glucose with a 2-hour value of 7.8-11.0 mmol/L. It is the glucose abnormality most reliably picked up by the OGTT. ^[1][2]
• Combined prediabetes: IFG plus IGT; the risk of progression and of cardiovascular disease is higher than for either alone. ^[1]
• Intermediate hyperglycaemia (prediabetes) is the WHO term for impaired fasting glucose and impaired glucose tolerance. ^[9]- Diabetic-range glucose with non-diabetic HbA1c: consider a haemoglobin variant, haemolysis, recent bleeding or transfusion, pregnancy, or a short duration of hyperglycaemia (fulminant diabetes). Repeat glucose and use fructosamine or quality-controlled glucose profiles. ^[3][5]
• Diabetic-range HbA1c with normal glucose: consider iron, B12 or folate deficiency with decreased red-cell turnover, or chronic kidney disease. Recheck the glucose basis of the result and seek specialist advice. ^[5]
• Ketosis without acidosis: capillary ketones 3.0 mmol/L or more without a low pH or low bicarbonate. This is not DKA; common causes are starvation, a ketogenic diet or a missed insulin dose in an otherwise well person. ^[4]
• Diabetic ketoacidosis: ketones 3.0 mmol/L or more with pH below 7.3, bicarbonate below 18 mmol/L and usually glucose 11.1 mmol/L or more (or a prior history of diabetes). ^[4]
• Hyperosmolar hyperglycaemic state: glucose 600 mg/dL (33.3 mmol/L) or more, effective osmolality above 300 mOsm/kg, total osmolality above 320 mOsm/kg, ketones below 3.0 mmol/L, pH 7.3 or higher, bicarbonate 15 mmol/L or higher. ^[4]- Absolute insulin deficiency: fasting C-peptide below 0.08 nmol/L, especially with ketosis. This points to type 1 diabetes or another insulin-deficient state and means insulin is required. ^[6]
• Time in range below 50% in older or high-risk adults still meet the relaxed target; in most adults the aim is above 70%. ^[8]

The values below are invented for teaching and labelled as illustrative. They are not real patient data and must not be treated as clinical results.

1. The fasting glucose is below 5.6 mmol/L, the 2-hour value is below 7.8 mmol/L and the HbA1c is below 39 mmol/mol, so no test crosses its diabetes or prediabetes threshold. ^{[1][2] [3]}
2. Capillary ketones are not elevated, so there is no ketosis. ^[4]
3. No further glucose testing is needed; repeat only if risk factors change or symptoms appear. ^[2]

Conclusion: This adult does not have diabetes or prediabetes, and the next step is routine risk reduction rather than further glucose testing.

1. Fasting glucose, the 2-hour value and HbA1c all meet the ADA/WHO diagnostic thresholds for diabetes, with concordant results. ^[1][2]
2. OGTT is not required to make the diagnosis when fasting glucose and HbA1c both agree in a symptomatic or high-risk adult. ^[1]
3. Ketones are not elevated, so there is no ketoacidosis; the next step is lifestyle, cardiovascular risk and medication planning, with referral as appropriate. ^[4]

Conclusion: This adult has diabetes on three consistent tests and no ketosis, so the next step is comprehensive diabetes assessment and treatment initiation rather than urgent ketone management.

1. The person has impaired fasting glucose, impaired glucose tolerance and an HbA1c in the prediabetes range, all in agreement. ^[1][2]
2. No value reaches a diabetic threshold, so this is non-diabetic hyperglycaemia rather than diabetes. ^[1][2]
3. Ketones are absent, so there is no ketosis; the aim is to delay or prevent progression with lifestyle intervention and annual reassessment. ^[1][2][4]

Conclusion: This adult has prediabetes with consistent results across three tests, so the next step is intensive lifestyle intervention and annual glucose review.

• Trusting a single glucose result in an asymptomatic person. Confirmation on a separate day is required. ^[1]
• Using HbA1c where it is not valid: in children and young people, suspected type 1 diabetes, acute illness, pregnancy, or any condition that affects red-cell lifespan or haemoglobin type. Use glucose-based criteria instead. ^[1]
• Comparing HbA1c from different methods without checking IFCC standardisation; always use a method calibrated to the IFCC standard. ^[3]
• Ignoring a mismatch between HbA1c and glucose. Investigate discrepancies and seek specialist advice. ^[3]
• Treating red-cell or albumin disorders as if they did not exist. Iron deficiency raises HbA1c; haemolysis and pregnancy lower it; nephrotic syndrome lowers fructosamine. ^[5]
• Forgetting sample type. Whole-blood (meter) glucose runs lower than plasma glucose, and whole-blood thresholds are lower. ^[1]
• Using urine ketones as the only ketone test when a blood meter is available; urine lags and can read high during resolution. ^[4]
• Interpreting C-peptide without the concurrent glucose, or without considering renal clearance. ^[6][7]
• Quoting population thresholds as if they were an individual's target. HbA1c targets of 48 or 53 mmol/mol are management goals from NICE, not diagnostic thresholds, and they are individualised. ^[3]
• Applying adult cut-offs to children or pregnancy. Gestational diabetes uses lower glucose thresholds, and children should be diagnosed on glucose, not HbA1c. ^[1]

• Diabetes can be diagnosed on fasting plasma glucose (7.0 mmol/L or more), 2-hour OGTT glucose (11.1 mmol/L or more), random glucose (11.1 mmol/L or more with symptoms) or HbA1c (6.5% / 48 mmol/mol or more), from WHO and ADA criteria. ^[1][2]
• One abnormal result is diagnostic in a symptomatic person; in an asymptomatic person confirm on a separate day. ^[1]
• HbA1c reflects average glucose over 2 to 3 months, does not require fasting, and is reported as percent (NGSP/DCCT) or mmol/mol (IFCC). ^[5][2]
• HbA1c is unreliable when red-cell lifespan or haemoglobin type is altered, such as in iron, B12 or folate deficiency, haemolysis, pregnancy and renal disease. ^[5]
• NICE advises using quality-controlled glucose profiles, total glycated haemoglobin or fructosamine when HbA1c is invalid. ^[3]
• Fructosamine reflects glucose control over the previous 2 to 3 weeks and is not affected by red-cell disorders. ^[5]
• Capillary beta-hydroxybutyrate (3.0 mmol/L or more) defines the ketosis of DKA, with pH below 7.3 and bicarbonate below 18 mmol/L. ^[4]
• Urine ketones of 2+ or more are a surrogate for significant ketonuria where blood ketones are unavailable. ^[4]
• C-peptide reflects endogenous insulin; fasting below 0.08 nmol/L suggests absolute insulin deficiency. ^[6]
• Time in range should be above 70% (70 to 180 mg/dL) for most adults, with time below range under 4% and below 54 mg/dL under 1%. ^[8]
• Every laboratory method and every individual patient may need its own interval; the printed local range always wins. ^[2][5]