ADA Diabetes Diagnosis Criteria and Testing Methods

The American Diabetes Association uses four laboratory tests to diagnose diabetes, each with a specific numeric threshold: a fasting plasma glucose of 126 mg/dL or higher, a two-hour oral glucose tolerance test (OGTT) result of 200 mg/dL or higher, a hemoglobin A1c (HbA1c) of 6.5% or higher, or a random plasma glucose of 200 mg/dL or higher in someone with classic symptoms like excessive thirst and frequent urination. These criteria are published in the ADA’s annually updated Standards of Care in Diabetes, which serves as the main clinical reference for how diabetes is identified and classified in the United States. But the numbers, straightforward as they look, come with important caveats about who they work best for, when they can mislead, and how different forms of diabetes require different diagnostic thinking.

The Four Diagnostic Tests

Any one of the four ADA-recognized tests can confirm a diabetes diagnosis, but they do not always agree with each other, and each has practical tradeoffs. The fasting plasma glucose (FPG) test requires you to skip food for at least eight hours, and a result of 126 mg/dL or above on two separate occasions meets the diagnostic bar. The OGTT involves drinking a 75-gram glucose solution and having blood drawn two hours later; a value of 200 mg/dL or above is diagnostic. The HbA1c test, which reflects average blood sugar over roughly two to three months, requires no fasting and can be drawn at any time of day, making it the most convenient option for many clinics. A random plasma glucose above 200 mg/dL paired with symptoms like unexplained weight loss or blurry vision rounds out the list.

When results from different tests conflict, the ADA recommends repeating the abnormal test or running a second type of test. In practice, many clinicians rely heavily on HbA1c because it does not require fasting and is less sensitive to day-to-day fluctuations. The ADA and the American Association for Clinical Chemistry recommend that HbA1c used for diagnosis be performed in an accredited laboratory rather than on a point-of-care device, since handheld analyzers can introduce enough variability to push a borderline result across the diagnostic line in either direction.¹

Prediabetes and the Gray Zone

Below the diabetes thresholds but above normal, the ADA defines a prediabetes range: fasting glucose of 100 to 125 mg/dL, a two-hour OGTT of 140 to 199 mg/dL, or an HbA1c of 5.7 to 6.4%. These ranges signal that blood sugar regulation is already impaired and the risk of progressing to full diabetes is elevated. A meta-analysis pooling data from multiple studies found that people with an HbA1c in the 6.0 to 6.4% range developed diabetes at a rate of about 36 per 1,000 person-years, a pace roughly comparable to people with impaired fasting glucose or impaired glucose tolerance defined by older glucose-based criteria.²

What many people do not realize is that HbA1c-based prediabetes and glucose-based prediabetes do not always identify the same individuals. You can have a normal fasting glucose but an HbA1c in the prediabetes range, or vice versa. The overlap is imperfect because HbA1c captures average glucose over months while fasting glucose is a single-moment snapshot. If your clinician flags prediabetes on one test and you feel skeptical because another test looks fine, that discordance is common and does not mean either result is wrong.

Who Should Be Screened

The ADA’s screening recommendations have broadened over time. The current guidance says testing should begin at age 35 for all adults, or earlier if you have a body mass index of 25 or greater along with at least one additional risk factor such as a family history of diabetes, a sedentary lifestyle, a history of gestational diabetes, or membership in a high-risk racial or ethnic group. That 2022 change, which lowered the age trigger from 45 to 35, was designed to catch more cases earlier.³

The ADA criteria cast a wider net than the US Preventive Services Task Force guidelines, which historically focused on overweight and obese adults in a narrower age range. Modeling studies have estimated that the ADA approach detects roughly a quarter more undiagnosed diabetes cases and about 40% more prediabetes cases than the USPSTF approach, with especially large gains among younger adults, Black Americans, and Hispanic Americans.⁴ If you are under 45 and your doctor has never mentioned diabetes screening, it is worth asking whether you meet the ADA’s expanded criteria.

When HbA1c Gets It Wrong

HbA1c measures how much glucose has attached to hemoglobin inside red blood cells. That measurement depends on two things: your actual blood sugar levels and how long your red blood cells live. Normal red blood cells circulate for roughly 120 days. If yours live shorter or longer than that, HbA1c can give a misleading number even if your glucose levels are identical to someone else’s.

Hemoglobin Variants and Blood Disorders

Sickle cell disease, sickle cell trait, and alpha thalassemia all alter red blood cell turnover and can produce falsely low HbA1c readings, potentially masking real diabetes.⁵ A study among Tanzanian adults found that people with sickle cell trait had dramatically reduced odds of being diagnosed with prediabetes or diabetes by HbA1c compared to those without the trait, yet their odds of diagnosis by oral glucose tolerance testing were no different.⁶ G6PD deficiency showed a similar trend, though the statistical evidence was weaker. Because these conditions are far more common in people of African, Mediterranean, and Southeast Asian descent, relying exclusively on HbA1c in these populations risks systematically missing diabetes cases.

A shorter red blood cell lifespan means hemoglobin spends less time being exposed to glucose, which drives HbA1c down regardless of what blood sugar is actually doing.⁷ Modeling work has confirmed that the relationship between HbA1c and average glucose shifts predictably with red blood cell lifespan: at a given percentage of HbA1c, the actual average glucose needed to produce that number changes in proportion to how long cells survive.⁸

Kidney Disease

Chronic kidney disease (CKD) introduces its own distortions. Anemia from reduced erythropoietin production, the use of erythropoiesis-stimulating agents, and altered red blood cell survival all conspire to make HbA1c unreliable in people with moderate-to-advanced CKD.⁹ Studies have found that HbA1c significantly underestimates blood sugar control in diabetic patients with pre-dialysis CKD, and that glycated albumin provides a more accurate picture in this group.¹⁰

Racial and Ethnic Differences

Even apart from hemoglobinopathies, researchers have documented that HbA1c can read differently across racial and ethnic groups at the same average glucose level. The reasons remain incompletely understood, but likely involve a combination of differences in red blood cell survival, how glucose moves between blood plasma and the interior of red cells, and genetic variation in how readily hemoglobin gets glycated.¹¹ This does not mean HbA1c is useless in any population, but it does mean that a borderline result in a Black or Hispanic patient deserves a second look, ideally with a glucose-based test to confirm or rule out the diagnosis.

Gestational Diabetes Screening

Diagnosing diabetes during pregnancy follows a different set of rules, and there is an ongoing debate in obstetric medicine about the right approach. The ADA endorses two strategies: a one-step method and a two-step method. In the one-step approach, a fasting 75-gram OGTT is performed at 24 to 28 weeks of pregnancy, and only one abnormal value out of three time points (fasting at or above 92 mg/dL, one hour at or above 180 mg/dL, two hours at or above 153 mg/dL) is enough for a diagnosis.¹²

The two-step method, still preferred by the American College of Obstetricians and Gynecologists, starts with a non-fasting 50-gram glucose challenge. If the one-hour result is 135 mg/dL or above, a full fasting 100-gram OGTT follows, and two or more abnormal values out of four time points are needed for the diagnosis.¹³ The one-step method catches more cases because its bar is lower, which means more women receive a gestational diabetes diagnosis, are monitored more closely, and undergo dietary or pharmacologic intervention. A meta-analysis found that both approaches identified gestational diabetes associated with higher risks of pre-eclampsia, cesarean delivery, and large-for-gestational-age babies, but the one-step diagnosis also flagged additional risks like neonatal intensive care admission and gestational hypertension.¹⁴ The trade-off is straightforward: the one-step approach labels more women as having gestational diabetes, leading to more treatment, which may improve some outcomes but also increases the burden of medical intervention on pregnancies that the two-step approach would have classified as normal.

Telling Diabetes Types Apart

The ADA’s diagnostic thresholds confirm that someone has diabetes, but they do not tell you which type. This matters because type 1, type 2, latent autoimmune diabetes in adults (LADA), and monogenic forms like MODY all require different treatment strategies. Getting the type wrong can mean prescribing the wrong medication for years.

The classic picture is familiar: type 1 diabetes usually appears in childhood or adolescence with rapid onset, while type 2 appears in overweight adults with gradual onset. But the real world is messier. A substantial fraction of people with adult-onset diabetes have features of both: they look like type 2 patients at diagnosis because they do not initially need insulin, yet they carry autoimmune markers associated with type 1. This slowly progressing autoimmune form, LADA, accounts for roughly 2 to 12% of all adult-onset diabetes depending on the population studied.¹⁵

Two laboratory tools help sort things out. Autoantibody testing (most commonly GAD65 antibodies) identifies the immune attack characteristic of type 1 and LADA. C-peptide, a protein released in equal amounts alongside insulin, gauges how much insulin the pancreas is still making. Research has identified C-peptide cut-off values that help distinguish diabetes types: a level below about 0.95 ng/mL suggests type 1 over type 2, while a level above about 1.65 ng/mL points toward type 2 over monogenic forms.¹⁶ For LADA specifically, an international expert panel recommended using C-peptide levels to guide treatment: very low values call for a full insulin regimen as in type 1, intermediate values represent a gray area requiring individualized decisions, and higher values allow an approach more similar to type 2 management with ongoing monitoring.¹⁷

Monogenic diabetes, including the various MODY subtypes, is another category frequently misclassified. The hallmarks include early-onset diabetes, a strong family history following an autosomal dominant pattern, absence of insulin resistance features, and no islet cell autoimmunity.¹⁸ One Asian registry set its genetic testing criteria at onset before age 35, a BMI below 32.5, and negative GAD antibodies.¹⁹ Because MODY can sometimes be managed with sulfonylureas rather than insulin, an accurate diagnosis changes the entire treatment plan. If you were diagnosed with type 2 diabetes at a young age, are lean, and have a parent and grandparent with diabetes, it is reasonable to ask about genetic testing.

High Blood Sugar in the Hospital

A diagnosis of diabetes is supposed to reflect a chronic metabolic condition, but blood sugar often spikes during acute illness. The body’s stress response floods the bloodstream with glucose as part of the fight-or-flight reaction, and this so-called stress hyperglycemia is extremely common during hospitalizations for infections, heart attacks, and surgeries. The diagnostic challenge is figuring out whether an elevated glucose reading in the hospital means underlying diabetes that just has not been caught yet, or a temporary stress response that will resolve when the illness does.

A large Scottish retrospective study found that while high blood sugar during hospitalization was associated with a higher risk of later diabetes diagnosis, most cases turned out to be stress hyperglycemia rather than true type 2 diabetes. Still, the researchers recommended follow-up testing for anyone admitted with a glucose above 200 mg/dL (11.1 mmol/L).²⁰ A separate study of over 2,000 patients hospitalized with pneumonia found that moderate-to-severe hyperglycemia on admission was strongly associated with a new diabetes diagnosis over the following five years, suggesting that opportunistic evaluation for diabetes during hospital stays could catch people who would otherwise slip through the cracks.²¹ If you are discharged from a hospital and nobody mentioned following up on an abnormal glucose reading, an HbA1c test a few weeks later can settle the question.

Why Fasting Glucose Climbs in Type 2 Diabetes

Understanding why fasting glucose rises helps make sense of the diagnostic thresholds. In a healthy person, the liver constantly produces glucose between meals to keep the brain and other organs fed, and insulin holds that production in check. In type 2 diabetes, the liver becomes resistant to insulin’s braking signal, and glucose production runs higher than it should, especially overnight. This excessive hepatic glucose production is a major contributor to the elevated fasting glucose readings that clinicians use for diagnosis.²² The liver’s glucose output in people with type 2 diabetes tends to increase in direct proportion to the degree of fasting hyperglycemia, driven particularly by increased gluconeogenesis, the process of building new glucose molecules from scratch.²³ This is part of the reason the fasting glucose test is effective at catching type 2 diabetes: it measures the metabolic process that breaks down earliest and most consistently in the disease.

Continuous Glucose Monitors as a Diagnostic Tool

Continuous glucose monitors, the wearable sensors that track blood sugar every few minutes, are well established for managing diabetes but are not yet part of the standard ADA diagnostic criteria. That may be changing. A consensus report has evaluated the potential role of CGM in screening for early-stage type 1 diabetes in people already known to carry autoantibodies. Metrics like the amount of time spent above 140 mg/dL could flag individuals progressing toward clinical diabetes and trigger confirmatory blood testing earlier than periodic lab visits would.²⁴

For prediabetes, CGM is a tool with obvious appeal. It captures glucose spikes after meals that a single fasting blood draw would miss entirely, and it provides a richer picture of daily glucose patterns. Researchers have begun examining CGM-derived markers to distinguish people with normal metabolism from those with prediabetes or early diabetes, with some novel metrics showing strong ability to separate the two groups in initial studies.²⁵ The practical barriers are cost, insurance coverage, and the lack of validated diagnostic thresholds. A review of published evidence noted that future studies are still needed to validate CGM metrics specifically for prediabetes before the technology can move into a formal diagnostic role.²⁶

The Cost Question

Screening large numbers of people for diabetes is not free, and the economics shape how aggressively health systems pursue it. A systematic review of cost-effectiveness studies found that the cost per new diabetes case detected through screening was roughly $100 in some models, but the downstream lifetime costs associated with screening-detected diabetes (including earlier treatment and monitoring) could exceed the cost of no screening by several thousand dollars per person identified. The math shifts favorably, though, when early detection leads to reductions in cardiovascular risk and routine medical costs, and some scenarios showed net savings per detected case when those benefits were factored in.²⁷ The broader point is that catching diabetes early through ADA criteria only generates value if it leads to actual intervention. A diagnosis without follow-up treatment is all cost and no benefit, which is why the ADA’s screening recommendations are paired with guidelines on immediate lifestyle counseling and, when warranted, pharmacologic treatment for prediabetes and newly diagnosed diabetes alike.