Diagnostic criteria and tests for diabetes mellitus
blood glucose, oral glucose tolerance test, and glycated hemoglobin

The diagnosis of diabetes mellitus is not based on an abstract concept of hyperglycemia, but on biochemical thresholds defined according to the relationship between increasing blood glucose and the risk of complications, especially microvascular complications, with particular reference to retinopathy. Over time, the diagnostic process has become progressively standardized and now recognizes four main diagnostic pathways in nonpregnant individuals: fasting plasma glucose, 2-hour plasma glucose after an oral glucose load, glycated hemoglobin, and, in the presence of a compatible clinical picture, random plasma glucose. These tools are not perfectly interchangeable because they assess different pathophysiological aspects of glucose dysregulation: hepatic glucose production and basal insulin secretion, the ability to manage a dynamic glucose load, average glycemic exposure over the preceding months, and the magnitude of acute hyperglycemia in a symptomatic setting.

From a practical standpoint, the clinical value of diabetes diagnostics lies not merely in assigning a label, but in placing the patient along a continuum from normoglycemia through intermediate abnormalities to overt diabetes, while also distinguishing situations in which one test is more reliable than another. Correct interpretation therefore requires attention to the method, sampling site, patient conditions, possible need for confirmation, and contexts in which certain tests may be misleading, such as hemoglobinopathies, anemia, pregnancy, advanced kidney failure, cystic fibrosis, or the post-transplant setting. This page focuses on the official diagnostic criteria for diabetes in nonpregnant individuals and the reasoned interpretation of the main tests used in clinical practice.

The diagnosis of diabetes arose from the need to identify a threshold beyond which hyperglycemia no longer represents merely a metabolic variation, but becomes a determinant of target-organ damage. This principle explains why modern diagnostic criteria are not based solely on statistical conventions, but on studies correlating glycemic levels with the development of diabetic retinopathy, neuropathy, nephropathy, and other clinically relevant outcomes. Fasting plasma glucose primarily captures the balance between basal insulin secretion and hepatic glucose production during the overnight fast; 2-hour glucose after a load instead assesses the body's ability to handle a standardized glucose challenge, revealing early defects in first-phase insulin secretion, insulin resistance, and inadequate suppression of hepatic glucose production. Glycated hemoglobin, finally, does not measure blood glucose at a specific moment, but the average exposure of red blood cells to glucose over the preceding 2–3 months.

This physiological diversity explains why the same individual may have a normal result on one test and an abnormal result on another. This is not necessarily an error, but reflects the fact that dysglycemia does not progress uniformly across all metabolic compartments. In some individuals, the first sign is an increase in post-load glucose; in others, fasting hyperglycemia predominates; and in still others, HbA1c rises before basal glucose exceeds the diabetic threshold. International guidelines therefore recognize several different tests as diagnostic, provided they are performed correctly and interpreted in the appropriate clinical context.

An essential principle is that a diagnosis of diabetes, except in the presence of unequivocal hyperglycemia with classic symptoms or a metabolic crisis, requires confirmation. This prevents inappropriate classification caused by biological variability, preanalytical error, intercurrent illness, or analytical interference. It is equally important to remember that standard criteria apply to nonpregnant individuals. Pregnancy is a specific context in which metabolism changes physiologically and gestational diabetes is diagnosed using dedicated criteria. The clinician must therefore ask not only “how high is the blood glucose?” but also “in what biological context and with which test was it measured?”

In nonpregnant individuals, the major international scientific societies recognize diabetes when one of the defined thresholds is exceeded for fasting plasma glucose, 2-hour plasma glucose after an oral glucose tolerance test (OGTT), glycated hemoglobin (HbA1c), or random plasma glucose in the presence of classic symptoms of hyperglycemia. The rationale is both pragmatic and pathophysiological: to permit diagnosis through different but standardized tools capable of detecting different disease phenotypes. Glycemic measurement must, however, be performed on venous plasma, because capillary values or readings obtained with a home glucose meter are not valid for establishing the formal diagnosis.

The spectrum of intermediate abnormalities, often termed prediabetes, also has major clinical significance. It is not a “minor” or harmless form of dysglycemia, but a state associated with increased risk of progression to overt diabetes and cardiometabolic comorbidities. Here too, however, the different tests do not overlap perfectly. An individual may have impaired fasting glucose, impaired glucose tolerance, or HbA1c in the intermediate range without all three conditions coexisting. This has practical implications for screening, counseling, and follow-up frequency.

Operationally, the diabetic threshold alone is not sufficient to define the type of diabetes. Biochemical criteria answer the question “is diabetes present or not?” but do not automatically establish whether it is type 1 diabetes mellitus, type 2 diabetes mellitus, monogenic diabetes, pancreatogenic diabetes, diabetes due to an endocrinopathy, or drug-induced diabetes. After biochemical diagnosis, the medical history, age at onset, body weight, clinical phenotype, autoantibodies, C-peptide, family pattern, and pathological context must therefore be integrated.

Official diagnostic criteria for diabetes in nonpregnant individuals

• HbA1c of 6.5% or higher, or 48 mmol/mol, measured using a standardized and certified method
• Fasting plasma glucose of 126 mg/dL or higher, equivalent to 7.0 mmol/L, after at least 8 hours of fasting
• 2-hour plasma glucose of 200 mg/dL or higher, equivalent to 11.1 mmol/L, during an OGTT with 75 g of anhydrous glucose
• Random plasma glucose of 200 mg/dL or higher, equivalent to 11.1 mmol/L, in the presence of classic symptoms of hyperglycemia or a hyperglycemic crisis

Intermediate abnormalities of glucose regulation

• HbA1c between 5.7% and 6.4%, equivalent to 39–47 mmol/mol
• Fasting plasma glucose between 100 and 125 mg/dL, equivalent to 5.6–6.9 mmol/L
• 2-hour plasma glucose during an OGTT between 140 and 199 mg/dL, equivalent to 7.8–11.0 mmol/L

These are the criteria that should be reported when referring to diagnostic criteria in the strict sense. Everything that follows, such as autoantibodies, C-peptide, lipid profile, kidney function, or continuous glucose monitoring, may be extremely useful for defining subtype, pathophysiology, and management, but does not replace or modify the official diagnostic threshold for diabetes in nonpregnant individuals.

Fasting plasma glucose is the simplest, least expensive, and most readily available test for diagnosing diabetes. It measures the glucose concentration in venous plasma after at least 8 hours of fasting and largely reflects the balance among hepatic glucose production, basal insulin secretion, and hepatic insulin sensitivity. When the liver continues to produce excessive glucose during fasting and the beta cell can no longer compensate adequately, morning blood glucose tends to rise until it exceeds the pathological threshold.

Its clinical usefulness is substantial because it allows convenient large-scale screening and is less demanding than the OGTT. However, it is not the most sensitive test for detecting all early forms of dysglycemia. Some individuals, especially in the early stages of type 2 diabetes, may maintain fasting glucose within the normal range or only mildly elevated despite a marked abnormality in the postprandial response and a clearly abnormal OGTT. In these situations, fasting glucose alone may underestimate the extent of the metabolic disorder.

From a technical standpoint, correct specimen processing is essential. Glucose continues to be consumed by blood cells after collection; if the specimen remains at room temperature for too long without appropriate preanalytical handling, the measured value may be falsely low. It is also important to distinguish laboratory plasma glucose from capillary glucose measured with a meter, which is useful for self-monitoring but not for establishing a diagnosis. Clinical interpretation of fasting plasma glucose must also consider context: acute illness, corticosteroid therapy, infectious stress, or hospitalization may cause transient hyperglycemia and require caution before a definitive diagnosis is made.

The oral glucose tolerance test (OGTT) is the most comprehensive dynamic test for assessing the body's ability to dispose of a standardized glucose load. After an overnight fast, baseline blood glucose is measured and the patient drinks 75 g of anhydrous glucose dissolved in water; in nonpregnant individuals, the main diagnostic value is the 2-hour plasma glucose. This test explores a different pathophysiological dimension from fasting glucose because it challenges the entire glucoregulatory system: intestinal absorption, incretins, early and late insulin secretion, peripheral insulin sensitivity, suppression of glucagon, and hepatic glucose production.

The OGTT is the most sensitive test for identifying impaired glucose tolerance and many early forms of diabetes that are missed by fasting glucose alone. It therefore retains a central role in several specialist settings. It is particularly useful when clinical suspicion and simpler tests are discordant, when HbA1c is uncertain or potentially unreliable, and in specific scenarios such as cystic fibrosis and post-transplant diabetes, in which guidelines recognize the superior sensitivity of the OGTT compared with HbA1c and isolated glucose measurements.

Correct preparation is required for an interpretable result. During the preceding days, the patient should consume a diet with adequate carbohydrate content and should not deliberately restrict carbohydrates, because a marked reduction in intake may alter the test response. Smoking, eating, and strenuous physical activity are not permitted during the test because they may affect blood glucose. The OGTT also has practical limitations: it is longer, less acceptable, less reproducible than a simple fasting sample, and more vulnerable to contextual factors such as immobility, acute illness, medications, and poor adherence to preparation. It is therefore used selectively, not as the sole universal test, but as a high-value tool when a more refined assessment of glucose tolerance is needed.

Glycated hemoglobin (HbA1c) expresses the proportion of hemoglobin that has undergone nonenzymatic glycation in relation to the average exposure of red blood cells to circulating glucose. It is therefore an integrated indicator of glycemic trends over the preceding weeks, with greater relative weight for the more recent period. Its major advantage is practical and clinical: it does not require fasting, has less within-person variability than blood glucose, is convenient for screening and follow-up, and correlates robustly with the risk of microvascular complications.

For diagnostic use, HbA1c must be measured using methods standardized and traceable to international reference programs. This is crucial because the strength of HbA1c as a diagnostic test depends on analytical quality. When the laboratory uses certified methods, the 6.5% threshold has solid clinical significance; conversely, if the method is unreliable or the patient's biological context alters red blood cell turnover, the result may be misleading.

The main limitation of HbA1c is biological even before it is technical. All conditions that alter red blood cell survival or interfere with the assay can distort the result. Hemolytic anemia, recent hemorrhage, transfusion, erythropoietin treatment, certain hemoglobinopathies, advanced kidney failure, pregnancy, and other conditions may produce artificially low or high values, dissociating HbA1c from actual glycemic exposure. In such cases, diagnosis should not automatically rely on HbA1c, but should return to direct measurements of plasma glucose. HbA1c is therefore extremely useful when the context is appropriate, but must never be interpreted as a universally infallible test.

Random plasma glucose has a specific diagnostic role when the patient has classic symptoms of hyperglycemia, such as polyuria, polydipsia, unexplained weight loss, blurred vision, marked fatigue, or when a hyperglycemic crisis occurs, such as diabetic ketoacidosis or hyperosmolar hyperglycemic state. In this setting, a plasma glucose level of 200 mg/dL or higher permits diagnosis without waiting for a second confirmatory test. The reason is straightforward: the clinical picture makes hyperglycemia unequivocal and clinically relevant.

This criterion is particularly important in rapidly evolving forms, such as classic type 1 diabetes, rapidly progressive adult autoimmune diabetes, or diabetes induced by immune checkpoint inhibitors. In these situations, waiting for delayed confirmation could expose the patient to metabolic deterioration or a dangerous treatment delay. Random plasma glucose is not, however, a shortcut for every scenario: in the absence of symptoms or a metabolic crisis, an isolated elevated value is not sufficient for a definitive diagnosis and requires confirmation by repeating the same test or obtaining another concordant diagnostic criterion.

It is also important to avoid confusing stress hyperglycemia with diabetes. During severe infections, myocardial infarction, stroke, trauma, major surgery, or treatment with catecholamines and glucocorticoids, blood glucose may rise substantially because of the counterregulatory response without necessarily indicating pre-existing chronic diabetes. Subsequent assessment after clinical stabilization with fasting plasma glucose, HbA1c, and sometimes OGTT can distinguish transient dysglycemia from true diabetes.

In an asymptomatic individual, a diagnosis of diabetes requires confirmation. In general, it is preferable to repeat the same test that was abnormal because this strategy reduces interpretive ambiguity and verifies reproducibility. If, for example, one HbA1c value is in the diabetic range and a second HbA1c confirms that the threshold has been exceeded, the diagnosis is established even if other tests remain below their respective thresholds. Similarly, a diabetic fasting plasma glucose confirmed on a second sample establishes the diagnosis even if HbA1c is not yet diagnostic.

Two different tests performed in parallel may, however, be discordant, for example HbA1c in the diabetic range with nondiagnostic fasting glucose, or an abnormal OGTT with HbA1c in the intermediate range. In such cases, the results should not be mechanically combined; rather, it should be recognized that each test captures a different expression of disease. Guidelines recommend giving weight to the test that exceeds the diagnostic threshold, especially if confirmed on repeat testing. Clinically, discordance should also prompt consideration of a biological or analytical issue that makes one test less reliable.

Confirmation is not merely a laboratory formality. It is an epistemological step that separates fluctuation from persistent disease. A single abnormal value may be influenced by improper fasting, intercurrent infection, unusual physical activity, acute stress, preanalytical variability, or collection errors. Once the diagnosis is confirmed, the clinician can proceed more confidently to the next phase: etiological classification of diabetes and assessment of complications or comorbidities already present at onset.

All diagnostic tests for diabetes can be distorted by interference, but HbA1c is especially vulnerable to interpretive errors when used outside the proper context. Its reliability declines whenever red blood cell survival is substantially shortened or prolonged, or when a hemoglobin variant interferes with the assay. Thus, an individual with chronic hemolysis, recent bleeding, transfusion, hemoglobinopathy, advanced kidney failure, or erythropoiesis-stimulating therapy may have an HbA1c value that does not faithfully represent true hyperglycemia. In these settings, plasma glucose and, if necessary, the OGTT provide more robust reference measures.

Blood glucose is not immune to problems either. The preanalytical phase is critical: delays in plasma separation, incorrect tube use, or improper specimen storage may artificially lower the value because of ongoing cellular glycolysis. Glycemia may also be transiently influenced by acute stress, infection, myocardial infarction, stroke, parenteral nutrition, hyperglycemia-inducing drugs, and concomitant endocrine disorders. Diagnosis must therefore always be contextualized rather than based on an isolated number.

In routine clinical language, HbA1c is often considered more “stable” and therefore absolutely superior, but this is true only when the patient's biology makes it interpretable. Selection of the best diagnostic test is therefore a reasoned decision: in a person with strong clinical suspicion and possible erythrocyte-related interference, direct glucose measurement takes priority; in a stable outpatient setting without confounders, HbA1c offers enormous practicality; and when postprandial dysglycemia is suspected or in certain special conditions, the OGTT retains a decisive advantage.

The standard criteria for diabetes in nonpregnant individuals should not be applied automatically in every context. Pregnancy is the most important example. At the initial prenatal assessment, particularly in high-risk women, standard tests may identify previously undiagnosed overt diabetes; however, gestational diabetes is defined through dedicated pathways and thresholds, generally based on an obstetric OGTT rather than HbA1c alone. This is because pregnancy physiology profoundly alters glucose homeostasis and the diagnostic objective also includes prevention of adverse maternal and fetal outcomes.

In cystic fibrosis, the OGTT is the preferred test for screening and diagnosis of cystic fibrosis-related diabetes because fasting glucose and HbA1c may remain deceptively little altered despite substantial postprandial glycemic instability. Similarly, in post-transplant diabetes, the OGTT has a particularly important role because immunosuppressive agents, perioperative stress, and metabolic variability may make static tests less sensitive. In pediatric patients as well, test results must be interpreted within the clinical phenotype, considering the importance of symptomatic presentation, autoimmunity, and monogenic forms.

The hospital setting requires additional caution. During hospitalization, particularly in critical care or after surgery, hyperglycemia may reflect the stress response rather than chronic diabetes. Admission HbA1c may help determine whether dysglycemia predated hospitalization, but even this interpretation must be cautious when anemia, transfusions, or kidney failure coexist. When it is not already evident, definitive diagnosis often requires reassessment after the acute phase. In other words, the numerical criteria remain the same, but their reliability depends on the clinical context in which they are applied.

In real-world practice, diagnosis of diabetes almost never results from a single action, but from an orderly diagnostic pathway. The starting point may be opportunistic screening, the presence of risk factors, an incidental finding of hyperglycemia, compatible symptoms, or suspicion generated by other clinical conditions. The first level usually consists of fasting plasma glucose and/or HbA1c because these are simple, reproducible, and convenient. If either exceeds the diabetic threshold in an asymptomatic patient, confirmation follows; if the results are borderline or discordant but suspicion remains high, the OGTT becomes the most informative clarifying test.

Once diabetes has been documented, the diagnostic process does not end but changes objective. The next steps are to distinguish the type of diabetes, assess severity at onset, and identify complications or target organs already involved. If the patient is lean, young, has weight loss, ketosis, or rapid progression to insulin requirement, autoantibodies and C-peptide become central. If there is a strong vertical family history, young age at onset, and absence of autoimmunity, monogenic diabetes should be considered. If the condition emerges after pancreatitis, pancreatic resection, transplantation, steroid therapy, or an endocrinopathy, the etiological assessment follows a different direction.

The core of the diagnostic process is therefore a logical sequence: clinical suspicion, biochemical documentation of hyperglycemia according to official criteria, confirmation when necessary, etiological attribution, and initial clinical-metabolic staging. This approach avoids both excessive simplification, which reduces everything to a single number, and the opposite excess of unnecessary premature investigations before the biochemical diagnosis has even been confirmed. In diabetology, as throughout internal medicine, a good diagnosis is one that organizes reasoning and makes the next step coherent.