Diagnosis and Classification of DiabetesMellitusAMERICAN DIABETES ASSOCIATION

DEFINITION ANDDESCRIPTION OF DIABETESMELLITUS — Diabetes is a group ofmetabolic diseases characterized by hy-perglycemia resulting from defects in in-sulin secretion, insulin action, or both.The chronic hyperglycemia of diabetes isassociated with long-term damage, dys-function, and failure of differentorgans,especially the eyes, kidneys, nerves, heart,and blood vessels.

Several pathogenic processes are in-volved in the development of diabetes.These range from autoimmune destruc-tion of the �-cells of the pancreas withconsequent insulin deficiency to abnor-malities that result in resistance to insulinaction. The basis of the abnormalities incarbohydrate, fat, and protein metabo-lism in diabetes is deficient action of in-sulin on target tissues. Deficient insulinaction results from inadequate insulin se-cretion and/or diminished tissue re-sponses to insulin at one or more points inthe complex pathways of hormone action.Impairment of insulin secretion and de-fects in insulin action frequently coexistin the same patient, and it is often unclearwhich abnormality, if either alone, is theprimary cause of the hyperglycemia.

Symptoms of marked hyperglycemiainclude polyuria, polydipsia, weight loss,sometimes with polyphagia, and blurredvision. Impairment of growth and suscep-tibility to certain infections may also ac-company chronic hyperglycemia. Acute,life-threatening consequences of uncon-trolled diabetes are hyperglycemia withketoacidosis or the nonketotic hyperos-molar syndrome.

Long-term complications of diabetesinclude retinopathy with potential loss ofvision; nephropathy leading to renal fail-ure; peripheral neuropathy with risk offoot ulcers, amputations, and Charcotjoints; and autonomic neuropathy caus-ing gastrointestinal, genitourinary, and

cardiovascular symptoms and sexual dys-function. Patients with diabetes have an in-creased incidence of atheroscleroticcardiovascular, peripheral arterial, and ce-rebrovascular disease. Hypertension andabnormalities of lipoprotein metabolism areoften found in people with diabetes.

The vast majority of cases of diabetesfall into two broad etiopathogenetic cate-gories (discussed in greater detail below).In one category, type 1 diabetes, the causeis an absolute deficiency of insulin secre-tion. Individuals at increased risk of de-veloping this type of diabetes can often beidentified by serological evidence of anautoimmune pathologic process occur-ring in the pancreatic islets and by geneticmarkers. In the other, much more preva-lent category, type 2 diabetes, the cause isa combination of resistance to insulin ac-tion and an inadequate compensatory in-sulin secretory response. In the lattercategory, a degree of hyperglycemia suffi-cient to cause pathologic and functionalchanges in various target tissues, butwithout clinical symptoms, may bepresent for a long period of time beforediabetes is detected. During this asymp-tomatic period, it is possible to demon-strate an abnormality in carbohydratemetabolism by measurement of plasmaglucose in the fasting state or after a chal-lenge with an oral glucose load.

The degree of hyperglycemia (if any)may change over time, depending on theextent of the underlying disease process(Fig. 1). A disease process may be presentbut may not have progressed far enoughto cause hyperglycemia. The same diseaseprocess can cause impaired fasting glu-cose (IFG) and/or impaired glucose toler-ance (IGT) without fulfilling the criteriafor the diagnosis of diabetes. In some in-dividuals with diabetes, adequate glyce-mic control can be achieved with weightreduction, exercise, and/or oral glucose-lowering agents. These individuals there-

fore do not require insulin. Otherindividuals who have some residual insu-lin secretion but require exogenous insu-lin for adequate glycemic control cansurvive without it. Individuals with ex-tensive �-cell destruction and thereforeno residual insulin secretion require insu-lin for survival. The severity of the meta-bolic abnormality can progress, regress,or stay the same. Thus, the degree of hy-perglycemia reflects the severity of the un-derlying metabolic process and itstreatment more than the nature of theprocess itself.

CLASSIFICATION OFDIABETES MELLITUS ANDOTHER CATEGORIESOF GLUCOSEREGULATION — Assigning a type ofdiabetes to an individual often dependson the circumstances present at the timeof diagnosis, and many diabetic individu-als do not easily fit into a single class. Forexample, a person with gestational diabe-tes mellitus (GDM) may continue to behyperglycemic after delivery and may bedetermined to have, in fact, type 2 diabe-tes. Alternatively, a person who acquiresdiabetes because of large doses of exoge-nous steroids may become normoglyce-mic once the glucocorticoids arediscontinued, but then may develop dia-betes many years later after recurrent ep-isodes of pancreatitis. Another examplewould be a person treated with thiazideswho develops diabetes years later. Becausethiazides in themselves seldom cause severehyperglycemia, such individuals probablyhave type 2 diabetes that is exacerbated bythe drug. Thus, for the clinician and patient,it is less important to label the particulartype of diabetes than it is to understand thepathogenesis of the hyperglycemia and totreat it effectively.

Type 1 diabetes (�-cell destruction,usually leading to absolute insulindeficiency)Immune-mediated diabetes. This formof diabetes, which accounts for only5–10% of those with diabetes, previouslyencompassed by the terms insulin-dependent diabetes, type 1 diabetes, orjuvenile-onset diabetes, results from a cel-lular-mediated autoimmune destruction

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

Sections on diagnosis revised Fall 2009.DOI: 10.2337/dc10-S062© 2010 by the American Diabetes Association. Readers may use this article as long as the work is properly

cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

P O S I T I O N S T A T E M E N T

S62 DIABETES CARE, VOLUME 33, SUPPLEMENT 1, JANUARY 2010 care.diabetesjournals.org

of the �-cells of the pancreas. Markers ofthe immune destruction of the �-cell in-clude islet cell autoantibodies, autoanti-bodies to insulin, autoantibodies to GAD(GAD65), and autoantibodies to the ty-rosine phosphatases IA-2 and IA-2�. Oneand usually more of these autoantibodiesare present in 85–90% of individualswhen fasting hyperglycemia is initiallydetected. Also, the disease has strong HLAassociations, with linkage to the DQA andDQB genes, and it is influenced by theDRB genes. These HLA-DR/DQ alleles canbe either predisposing or protective.

In this form of diabetes, the rate of�-cell destruction is quite variable, beingrapid in some individuals (mainly infantsand children) and slow in others (mainlyadults). Some patients, particularly chil-dren and adolescents, may present withketoacidosis as the first manifestation ofthe disease. Others have modest fastinghyperglycemia that can rapidly change tosevere hyperglycemia and/or ketoacidosisin the presence of infection or other stress.Still others, particularly adults, may retainresidual �-cell function sufficient to pre-vent ketoacidosis for many years; such in-dividuals eventually become dependenton insulin for survival and are at risk forketoacidosis. At this latter stage of the dis-ease, there is little or no insulin secretion,as manifested by low or undetectable lev-els of plasma C-peptide. Immune-mediated diabetes commonly occurs in

childhood and adolescence, but it can oc-cur at any age, even in the 8th and 9thdecades of life.

Autoimmune destruction of �-cellshas multiple genetic predispositions andis also related to environmental factorsthat are still poorly defined. Although pa-tients are rarely obese when they presentwith this type of diabetes, the presence ofobesity is not incompatible with the diag-nosis. These patients are also prone toother autoimmune disorders such asGraves’ disease, Hashimoto’s thyroiditis,Addison’s disease, vitiligo, celiac sprue,autoimmune hepatitis, myasthenia gravis,and pernicious anemia.Idiopathic diabetes. Some forms of type1 diabetes have no known etiologies.Some of these patients have permanentinsulinopenia and are prone to ketoacido-sis, but have no evidence of autoimmu-nity. Although only a minority of patientswith type 1 diabetes fall into this category,of those who do, most are of African orAsian ancestry. Individuals with this formof diabetes suffer from episodic ketoaci-dosis and exhibit varying degrees of insu-lin deficiency between episodes. Thisform of diabetes is strongly inherited,lacks immunological evidence for �-cellautoimmunity, and is not HLA associated.An absolute requirement for insulin re-placement therapy in affected patientsmay come and go.

Type 2 diabetes (ranging frompredominantly insulin resistancewith relative insulin deficiency topredominantly an insulin secretorydefect with insulin resistance)This form of diabetes, which accounts for�90–95% of those with diabetes, previ-ously referred to as non–insulin-dependent diabetes, type 2 diabetes, oradult-onset diabetes, encompasses indi-viduals who have insulin resistance andusually have relative (rather than abso-lute) insulin deficiency At least initially,and often throughout their lifetime, theseindividuals do not need insulin treatmentto survive. There are probably many dif-ferent causes of this form of diabetes. Al-though the specific etiologies are notknown, autoimmune destruction of�-cells does not occur, and patients donot have any of the other causes of diabe-tes listed above or below.

Most patients with this form of diabe-tes are obese, and obesity itself causessome degree of insulin resistance. Patientswho are not obese by traditional weightcriteria may have an increased percentageof body fat distributed predominantly inthe abdominal region. Ketoacidosis sel-dom occurs spontaneously in this type ofdiabetes; when seen, it usually arises inassociation with the stress of another ill-ness such as infection. This form of dia-betes frequently goes undiagnosed formany years because the hyperglycemia

Figure 1—Disorders of glycemia: etiologic types and stages. *Even after presenting in ketoacidosis, these patients can briefly return to normogly-cemia without requiring continuous therapy (i.e., “honeymoon” remission); **in rare instances, patients in these categories (e.g., Vacor toxicity, type1 diabetes presenting in pregnancy) may require insulin for survival.

Position Statement

care.diabetesjournals.org DIABETES CARE, VOLUME 33, SUPPLEMENT 1, JANUARY 2010 S63

develops gradually and at earlier stages isoften not severe enough for the patient tonotice any of the classic symptoms of di-abetes. Nevertheless, such patients are atincreased risk of developing macrovascu-lar and microvascular complications.Whereas patients with this form of diabe-tes may have insulin levels that appearnormal or elevated, the higher blood glu-cose levels in these diabetic patientswould be expected to result in evenhigher insulin values had their �-cellfunction been normal. Thus, insulin se-cretion is defective in these patients andinsufficient to compensate for insulin re-sistance. Insulin resistance may improvewith weight reduction and/or pharmaco-logical treatment of hyperglycemia but isseldom restored to normal. The risk ofdeveloping this form of diabetes increaseswith age, obesity, and lack of physical ac-tivity. It occurs more frequently inwomen with prior GDM and in individu-als with hypertension or dyslipidemia,and its frequency varies in different racial/ethnic subgroups. It is often associatedwith a strong genetic predisposition,more so than is the autoimmune form oftype 1 diabetes. However, the genetics ofthis form of diabetes are complex and notclearly defined.

Other specific types of diabetesGenetic defects of the �-cell. Severalforms of diabetes are associated with mo-nogenetic defects in �-cell function.These forms of diabetes are frequentlycharacterized by onset of hyperglycemiaat an early age (generally before age 25years). They are referred to as maturity-onset diabetes of the young (MODY) andare characterized by impaired insulin se-cretion with minimal or no defects in in-sulin action. They are inherited in anautosomal dominant pattern. Abnormali-ties at six genetic loci on different chro-mosomes have been identified to date.The most common form is associatedwith mutations on chromosome 12 in ahepatic transcription factor referred to ashepatocyte nuclear factor (HNF)-1�. Asecond form is associated with mutationsin the glucokinase gene on chromosome7p and results in a defective glucokinasemolecule. Glucokinase converts glucoseto glucose-6-phosphate, the metabolismof which, in turn, stimulates insulin secre-tion by the �-cell. Thus, glucokinaseserves as the “glucose sensor” for the�-cell. Because of defects in the glucoki-nase gene, increased plasma levels of glu-cose are necessary to elicit normal levels

of insulin secretion. The less commonforms result from mutations in other tran-scription factors, including HNF-4�,HNF-1�, insulin promoter factor (IPF)-1,and NeuroD1.

Point mutations in mitochondrialDNA have been found to be associatedwith diabetes and deafness The mostcommon mutation occurs at position3,243 in the tRNA leucine gene, leadingto an A-to-G transition. An identical le-sion occurs in the MELAS syndrome (mi-tochondrial myopathy, encephalopathy,lactic acidosis, and stroke-like syn-drome); however, diabetes is not part ofthis syndrome, suggesting different phe-notypic expressions of this genetic lesion.

Genetic abnormalities that result inthe inability to convert proinsulin to in-sulin have been identified in a few fami-lies, and such traits are inherited in anautosomal dominant pattern. The result-ant glucose intolerance is mild. Similarly,the production of mutant insulin mole-cules with resultant impaired receptorbinding has also been identified in a fewfamilies and is associated with an autoso-mal inheritance and only mildly impairedor even normal glucose metabolism.Genetic defects in insulin action. Thereare unusual causes of diabetes that resultfrom genetically determined abnormali-ties of insulin action. The metabolic ab-normalities associated with mutations ofthe insulin receptor may range from hy-perinsulinemia and modest hyperglyce-mia to severe diabetes. Some individualswith these mutations may have acanthosisnigricans. Women may be virilized andhave enlarged, cystic ovaries. In the past,this syndrome was termed type A insulinresistance. Leprechaunism and the Rabson-Mendenhall syndrome are two pediatricsyndromes that have mutations in the insu-lin receptor gene with subsequent alter-ations in insulin receptor function andextreme insulin resistance. The former hascharacteristic facial features and is usuallyfatal in infancy, while the latter is associatedwith abnormalities of teeth and nails andpineal gland hyperplasia.

Alterations in the structure and func-tion of the insulin receptor cannot be dem-onstrated in patients with insulin-resistantlipoatrophic diabetes. Therefore, it is as-sumed that the lesion(s) must reside in thepostreceptor signal transduction pathways.Diseases of the exocrine pancreas. Anyprocess that diffusely injures the pancreascan cause diabetes. Acquired processesinclude pancreatitis, trauma, infection,pancreatectomy, and pancreatic carci-

noma. With the exception of that causedby cancer, damage to the pancreas mustbe extensive for diabetes to occur; adre-nocarcinomas that involve only a smallportion of the pancreas have been associ-ated with diabetes. This implies a mecha-nism other than simple reduction in�-cell mass. If extensive enough, cysticfibrosis and hemochromatosis will alsodamage �-cells and impair insulin secre-tion. Fibrocalculous pancreatopathy maybe accompanied by abdominal pain radi-ating to the back and pancreatic calcifica-tions identified on X-ray examination.Pancreatic fibrosis and calcium stones inthe exocrine ducts have been found atautopsy.Endocrinopathies. Several hormones(e.g., growth hormone, cortisol, gluca-gon, epinephrine) antagonize insulin ac-tion. Excess amounts of these hormones(e.g., acromegaly, Cushing’s syndrome,glucagonoma, pheochromocytoma, re-spectively) can cause diabetes. This gen-eral ly occurs in individuals withpreexisting defects in insulin secretion,and hyperglycemia typically resolveswhen the hormone excess is resolved.

Somatostatinoma- and aldoster-onoma-induced hypokalemia can causediabetes, at least in part, by inhibiting in-sulin secretion. Hyperglycemia generallyresolves after successful removal of thetumor.Drug- or chemical-induced diabetes.Many drugs can impair insulin secretion.These drugs may not cause diabetes bythemselves, but they may precipitate dia-betes in individuals with insulin resis-tance. In such cases, the classification isunclear because the sequence or relativeimportance of �-cell dysfunction and in-sulin resistance is unknown. Certain tox-ins such as Vacor (a rat poison) andintravenous pentamidine can perma-nently destroy pancreatic �-cells. Suchdrug reactions fortunately are rare. Thereare also many drugs and hormones thatcan impair insulin action. Examples in-clude nicotinic acid and glucocorticoids.Patients receiving �-interferon have beenreported to develop diabetes associatedwith islet cell antibodies and, in certaininstances, severe insulin deficiency. Thelist shown in Table 1 is not all-inclusive,but reflects the more commonly recog-nized drug-, hormone-, or toxin-inducedforms of diabetes.Infections. Certain viruses have been as-sociated with �-cell destruction. Diabetesoccurs in patients with congenital rubella,although most of these patients have HLA

Diagnosis and Classification

S64 DIABETES CARE, VOLUME 33, SUPPLEMENT 1, JANUARY 2010 care.diabetesjournals.org

and immune markers characteristic oftype 1 diabetes. In addition, coxsackievi-rus B, cytomegalovirus, adenovirus, andmumps have been implicated in inducingcertain cases of the disease.Uncommon forms of immune-medi-ated diabetes. In this category, there aretwo known conditions, and others arelikely to occur. The stiff-man syndrome isan autoimmune disorder of the centralnervous system characterized by stiffnessof the axial muscles with painful spasms.Patients usually have high titers of theGAD autoantibodies, and approximatelyone-third will develop diabetes.

Anti-insulin receptor antibodies cancause diabetes by binding to the insulinreceptor, thereby blocking the binding ofinsulin to its receptor in target tissues.However, in some cases, these antibodiescan act as an insulin agonist after bindingto the receptor and can thereby cause hy-poglycemia. Anti-insulin receptor anti-bodies are occasionally found in patientswith systemic lupus erythematosus andother autoimmune diseases. As in otherstates of extreme insulin resistance, pa-tients with anti-insulin receptor antibod-ies often have acanthosis nigricans. In thepast, this syndrome was termed type Binsulin resistance.Other genetic syndromes sometimesassociated with diabetes. Many geneticsyndromes are accompanied by an in-creased incidence of diabetes. These in-clude the chromosomal abnormalities ofDown syndrome, Klinefelter syndrome,and Turner syndrome. Wolfram’s syn-drome is an autosomal recessive disordercharacterized by insulin-deficient diabe-tes and the absence of �-cells at autopsy.Additional manifestations include diabe-tes insipidus, hypogonadism, optic atro-phy, and neural deafness. Othersyndromes are listed in Table 1.

Gestational diabetes mellitusFor many years, GDM has been defined asany degree of glucose intolerance with on-set or first recognition during pregnancy.Although most cases resolve with deliv-ery, the definition applied whether or notthe condition persisted after pregnancyand did not exclude the possibility thatunrecognized glucose intolerance mayhave antedated or begun concomitantlywith the pregnancy. This definition facil-itated a uniform strategy for detection andclassification of GDM, but its limitationswere recognized for many years. As theongoing epidemic of obesity and diabeteshas led to more type 2 diabetes in women

Table 1—Etiologic classification of diabetes mellitus

I. Type 1 diabetes (�-cell destruction, usually leading to absolute insulin deficiency)A. Immune mediatedB. Idiopathic

II. Type 2 diabetes (may range from predominantly insulin resistance with relative insulin deficiencyto a predominantly secretory defect with insulin resistance)

III. Other specific typesA. Genetic defects of �-cell function

1. Chromosome 12, HNF-1� (MODY3)2. Chromosome 7, glucokinase (MODY2)3. Chromosome 20, HNF-4� (MODY1)4. Chromosome 13, insulin promoter factor-1 (IPF-1; MODY4)5. Chromosome 17, HNF-1� (MODY5)6. Chromosome 2, NeuroD1 (MODY6)7. Mitochondrial DNA8. Others

B. Genetic defects in insulin action1. Type A insulin resistance2. Leprechaunism3. Rabson-Mendenhall syndrome4. Lipoatrophic diabetes5. Others

C. Diseases of the exocrine pancreas1. Pancreatitis2. Trauma/pancreatectomy3. Neoplasia4. Cystic fibrosis5. Hemochromatosis6. Fibrocalculous pancreatopathy7. Others

D. Endocrinopathies1. Acromegaly2. Cushing’s syndrome3. Glucagonoma4. Pheochromocytoma5. Hyperthyroidism6. Somatostatinoma7. Aldosteronoma8. Others

E. Drug or chemical induced1. Vacor2. Pentamidine3. Nicotinic acid4. Glucocorticoids5. Thyroid hormone6. Diazoxide7. �-adrenergic agonists8. Thiazides9. Dilantin10. �-Interferon11. Others

F. Infections1. Congenital rubella2. Cytomegalovirus3. Others

G. Uncommon forms of immune-mediated diabetes1. “Stiff-man” syndrome2. Anti-insulin receptor antibodies3. Others

H. Other genetic syndromes sometimes associated with diabetes1. Down syndrome2. Klinefelter syndrome3. Turner syndrome4. Wolfram syndrome5. Friedreich ataxia6. Huntington chorea7. Laurence-Moon-Biedl syndrome8. Myotonic dystrophy9. Porphyria10. Prader-Willi syndrome11. Others

IV. Gestational diabetes mellitus

Patients with any form of diabetes may require insulin treatment at some stage of their disease. Such use ofinsulin does not, of itself, classify the patient.

Position Statement

care.diabetesjournals.org DIABETES CARE, VOLUME 33, SUPPLEMENT 1, JANUARY 2010 S65

of childbearing age, the number of preg-nant women with undiagnosed type 2 di-abetes has increased.

After deliberations in 2008 –2009,the International Association of Diabetesand Pregnancy Study Groups (IADPSG),an international consensus group withrepresentatives from multiple obstetricaland diabetes organizations, including theAmerican Diabetes Association (ADA),recommended that high-risk womenfound to have diabetes at their initial pre-natal visit, using standard criteria (Table3), receive a diagnosis of overt, not gesta-tional, diabetes. Approximately 7% of allpregnancies (ranging from 1 to 14%, de-pending on the population studied andthe diagnostic tests employed) are com-plicated by GDM, resulting in more than200,000 cases annually.

CATEGORIES OFINCREASED RISK FORDIABETES — In 1997 and 2003, TheExpert Committee on Diagnosis and Clas-sification of Diabetes Mellitus (1,2) recog-n ized an in te rmedia te group ofindividuals whose glucose levels do notmeet criteria for diabetes, yet are higherthan those considered normal. These peo-ple were defined as having impaired fast-ing glucose (IFG) [fasting plasma glucose(FPG) levels 100 mg/dl (5.6 mmol/l) to125 mg/dl (6.9 mmol/l)], or impaired glu-cose tolerance (IGT) [2-h values in theoral glucose tolerance test (OGTT) of 140mg/dl (7.8 mmol/l) to 199 mg/dl (11.0mmol/l)].

Individuals with IFG and/or IGT havebeen referred to as having pre-diabetes,indicating the relatively high risk for thefuture development of diabetes. IFG andIGT should not be viewed as clinical en-tities in their own right but rather riskfactors for diabetes as well as cardiovas-cular disease. They can be observed as in-termediate stages in any of the diseaseprocesses listed in Table 1. IFG and IGTare associated with obesity (especially ab-

dominal or visceral obesity), dyslipidemiawith high triglycerides and/or low HDLcholesterol, and hypertension. Structuredlifestyle intervention, aimed at increasingphysical activity and producing 5–10%loss of body weight, and certain pharma-cological agents have been demonstratedto prevent or delay the development ofdiabetes in people with IGT; the potentialimpact of such interventions to reducemortality or the incidence of cardiovascu-lar disease has not been demonstrated todate. It should be noted that the 2003ADA Expert Committee report reducedthe lower FPG cut point to define IFGfrom 110 mg/dl (6.1 mmol/l) to 100mg/dl (5.6 mmol/l), in part to ensure thatprevalence of IFG was similar to that ofIGT. However, the World Health Organi-zation (WHO) and many other diabetesorganizations did not adopt this change inthe definition of IFG.

As A1C is used more commonly todiagnose diabetes in individuals with riskfactors, it will also identify those at higherrisk for developing diabetes in the future.When recommending the use of the A1Cto diagnose diabetes in its 2009 report,the International Expert Committee (3)stressed the continuum of risk for diabe-tes with all glycemic measures and did notformally identify an equivalent intermedi-ate category for A1C. The group did notethat those with A1C levels above the lab-oratory “normal” range but below the di-agnostic cut point for diabetes (6.0 to�6.5%) are at very high risk of develop-ing diabetes. Indeed, incidence of diabe-tes in people with A1C levels in this rangeis more than 10 times that of people withlower levels (4–7). However, the 6.0 to�6.5% range fails to identify a substantialnumber of patients who have IFG and/orIGT. Prospective studies indicate thatpeople within the A1C range of 5.5–6.0%have a 5-year cumulative incidence of di-abetes that ranges from 12 to 25% (4–7),which is appreciably (three- to eightfold)higher than incidence in the U.S. popula-tion as a whole (8). Analyses of nationallyrepresentative data from the NationalHealth and Nutrition Examination Survey(NHANES) indicate that the A1C valuethat most accurately identifies peoplewith IFG or IGT falls between 5.5 and6.0%. In addition, linear regression anal-yses of these data indicate that among thenondiabetic adult population, an FPG of110 mg/dl (6.1 mmol/l) corresponds to anA1C of 5.6%, while an FPG of 100 mg/dl(5.6 mmol/l) corresponds to an A1C of5.4% (R.T. Ackerman, personal commu-

nication). Finally, evidence from the Dia-betes Prevention Program (DPP), whereinthe mean A1C was 5.9% (SD 0.5%), indi-cates that preventive interventions are ef-fective in groups of people with A1Clevels both below and above 5.9% (9). Forthese reasons, the most appropriate A1Clevel above which to initiate preventiveinterventions is likely to be somewhere inthe range of 5.5–6%.

As was the case with FPG and 2-h PG,defining a lower limit of an intermediatecategory of A1C is somewhat arbitrary, asthe risk of diabetes with any measure orsurrogate of glycemia is a continuum, ex-tending well into the normal ranges. Tomaximize equity and efficiency of preven-tive interventions, such an A1C cut pointshould balance the costs of “false nega-tives” (failing to identify those who aregoing to develop diabetes) against thecosts of “false positives” (falsely identify-ing and then spending intervention re-sources on those who were not going todevelop diabetes anyway).

Compared to the fasting glucose cut-point of 100 mg/dl (5.6 mmol/l), an A1Ccutpoint of 5.7% is less sensitive but morespecific and has a higher positive predic-tive value to identify people at risk forlater development of diabetes. A largeprospective study found that a 5.7% cut-point has a sensitivity of 66% and speci-ficity of 88% for the identification ofsubsequent 6-year diabetes incidence(10). Receiver operating curve analysesof nationally representative U.S. data(NHANES 1999-2006) indicate that anA1C value of 5.7% has modest sensitivity(39-45%) but high specificity (81-91%)to identify cases of IFP (FPG �100 mg/dl)(5.6 mmol/l) or IGT (2-h glucose � 140mg/dl) (R.T. Ackerman, personal com-munication). Other analyses suggest thatan A1C of 5.7% is associated with diabe-tes risk similar to the high-risk partici-pants in the DPP (R.T. Ackerman,personal communication). Hence, it isreasonable to consider an A1C range of5.7 to 6.4% as identifying individualswith high risk for future diabetes and towhom the term pre-diabetes may be ap-plied if desired.

Individuals with an A1C of 5.7–6.4%should be informed of their increased riskfor diabetes as well as cardiovascular dis-ease and counseled about effective strate-gies, such as weight loss and physicalactivity, to lower their risks. As with glu-cose measurements, the continuum ofrisk is curvilinear, so that as A1C rises, therisk of diabetes rises disproportionately.

Table 2—Categories of increased risk fordiabetes*

FPG 100 mg/dl (5.6 mmol/l) to 125 mg/dl(6.9 mmol/l) �IFG�

2-h PG in the 75-g OGTT 140 mg/dl (7.8mmol/l) to 199 mg/dl (11.0 mmol/l) �IGT�

A1C 5.7–6.4%

*For all three tests, risk is continuous, extendingbelow the lower limit of the range and becomingdisproportionately greater at higher ends of therange.

Diagnosis and Classification

S66 DIABETES CARE, VOLUME 33, SUPPLEMENT 1, JANUARY 2010 care.diabetesjournals.org

Accordingly, interventions should bemost intensive and follow-up should beparticularly vigilant for those with A1Clevels above 6.0%, who should be consid-ered to be at very high risk. However, justas an individual with a fasting glucose of98 mg/dl (5.4 mmol/l) may not be at neg-ligible risk for diabetes, individuals withA1C levels below 5.7% may still be at risk,depending on level of A1C and presenceof other risk factors, such as obesity andfamily history.

Table 2 summarizes the categories ofincreased risk for diabetes. Evaluation ofpatients at risk should incorporate aglobal risk factor assessment for both di-abetes and cardiovascular disease.Screening for and counseling about risk ofdiabetes should always be in the prag-matic context of the patient’s comorbidi-ties, life expectancy, personal capacity toengage in lifestyle change, and overallhealth goals.

DIAGNOSTIC CRITERIA FORDIABETES MELLITUS — For de-cades, the diagnosis of diabetes has beenbased on glucose criteria, either the FPGor the 75-g OGTT. In 1997, the first Ex-pert Committee on the Diagnosis andClassification of Diabetes Mellitus revisedthe diagnostic criteria, using the observedassociation between FPG levels and pres-ence of retinopathy as the key factor withwhich to identify threshold glucose level.The Committee examined data from threecross-sectional epidemiologic studies thatassessed retinopathy with fundus photog-raphy or direct ophthalmoscopy andmeasured glycemia as FPG, 2-h PG, andA1C. These studies demonstrated glyce-mic levels below which there was littleprevalent retinopathy and above whichthe prevalence of retinopathy increased inan apparently linear fashion. The decilesof the three measures at which retinopa-thy began to increase were the same foreach measure within each population.Moreover, the glycemic values abovewhich retinopathy increased were similaramong the populations. These analyseshelped to inform a new diagnostic cutpoint of �126 mg/dl (7.0 mmol/l) forFPG and confirmed the long-standing di-agnostic 2-h PG value of �200 mg/dl(11.1 mmol/l).

A1C is a widely used marker ofchronic glycemia, reflecting averageblood glucose levels over a 2- to 3-monthperiod of time. The test plays a critical rolein the management of the patient with di-abetes, since it correlates well with both

microvascular and, to a lesser extent, ma-crovascular complications and is widelyused as the standard biomarker for theadequacy of glycemic management. PriorExpert Committees have not recom-mended use of the A1C for diagnosis ofdiabetes, in part due to lack of standard-ization of the assay. However, A1C assaysare now highly standardized so that theirresults can be uniformly applied bothtemporally and across populations. Intheir recent report (3), an InternationalExpert Committee, after an extensive re-view of both established and emerging ep-idemiological evidence, recommendedthe use of the A1C test to diagnose diabe-tes, with a threshold of �6.5%, and ADAaffirms this decision. The diagnostic A1Ccut point of 6.5% is associated with aninflection point for retinopathy preva-lence, as are the diagnostic thresholds forFPG and 2-h PG (3). The diagnostic testshould be performed using a method thatis certified by the National Glycohemo-globin Standardization Program (NGSP)and standardized or traceable to the Dia-betes Control and Complications Trialreference assay. Point-of-care A1C assaysare not sufficiently accurate at this time touse for diagnostic purposes.

There is an inherent logic to using amore chronic versus an acute marker ofdysglycemia, particularly since the A1C isalready widely familiar to clinicians as amarker of glycemic control. Moreover,the A1C has several advantages to theFPG, including greater convenience,since fasting is not required, evidence tosuggest greater preanalytical stability, andless day-to-day perturbations during pe-riods of stress and illness. These advan-tages, however, must be balanced bygreater cost, the limited availability ofA1C testing in certain regions of the de-veloping world, and the incomplete cor-relation between A1C and averageglucose in certain individuals. In addi-

tion, the A1C can be misleading in pa-tients with certain forms of anemia andhemoglobinopathies, which may alsohave unique ethnic or geographic distri-butions. For patients with a hemoglobi-nopathy but normal red cell turnover,such as sickle cell trait, an A1C assaywithout interference from abnormal he-moglobins should be used (an updatedlist is available at www.ngsp.org/prog/index3.html). For conditions with abnor-mal red cell turnover, such as anemiasfrom hemolysis and iron deficiency, thediagnosis of diabetes must employ glu-cose criteria exclusively.

The established glucose criteria forthe diagnosis of diabetes remain valid.These include the FPG and 2-h PG. Addi-tionally, patients with severe hyperglyce-mia such as those who present with severeclassic hyperglycemic symptoms or hy-perglycemic crisis can continue to be di-agnosed when a random (or casual)plasma glucose of �200 mg/dl (11.1mmol/l) is found. It is likely that in suchcases the health care professional wouldalso measure an A1C test as part of theinitial assessment of the severity of the di-abetes and that it would (in most cases) beabove the diagnostic cut point for diabe-tes. However, in rapidly evolving diabe-tes, such as the development of type 1diabetes in some children, A1C may notbe significantly elevated despite frankdiabetes.

Just as there is less than 100% con-cordance between the FPG and 2-h PGtests, there is not full concordance be-tween A1C and either glucose-based test.Analyses of NHANES data indicate that,assuming universal screening of the undi-agnosed, the A1C cut point of �6.5%identifies one-third fewer cases of undiag-nosed diabetes than a fasting glucose cutpoint of �126 mg/dl (7.0 mmol/l) (cdcwebsite tbd). However, in practice, a largeportion of the population with type 2 di-

Table 3—Criteria for the diagnosis of diabetes

1. A1C �6.5%. The test should be performed in a laboratory using a method that is NGSPcertified and standardized to the DCCT assay.*

OR2. FPG �126 mg/dl (7.0 mmol/l). Fasting is defined as no caloric intake for at least 8 h.*

OR3. 2-h plasma glucose �200 mg/dl (11.1 mmol/l) during an OGTT. The test should be

performed as described by the World Health Organization, using a glucose load containingthe equivalent of 75 g anhydrous glucose dissolved in water.*

OR4. In a patient with classic symptoms of hyperglycemia or hyperglycemic crisis, a random

plasma glucose �200 mg/dl (11.1 mmol/l).

*In the absence of unequivocal hyperglycemia, criteria 1–3 should be confirmed by repeat testing.

Position Statement

care.diabetesjournals.org DIABETES CARE, VOLUME 33, SUPPLEMENT 1, JANUARY 2010 S67

abetes remains unaware of their condi-tion. Thus, it is conceivable that the lowersensitivity of A1C at the designated cutpoint will be offset by the test’s greaterpracticality, and that wider application ofa more convenient test (A1C) may actu-ally increase the number of diagnosesmade.

Further research is needed to bettercharacterize those patients whose glyce-mic status might be categorized differ-ently by two different tests (e.g., FPG andA1C), obtained in close temporal approx-imation. Such discordance may arise frommeasurement variability, change overtime, or because A1C, FPG, and postchal-lenge glucose each measure differentphysiological processes. In the setting ofan elevated A1C but “nondiabetic” FPG,the likelihood of greater postprandial glu-cose levels or increased glycation rates fora given degree of hyperglycemia may bepresent. In the opposite scenario (highFPG yet A1C below the diabetes cutpoint), augmented hepatic glucose pro-duction or reduced glycation rates may bepresent.

As with most diagnostic tests, a testresult diagnostic of diabetes should be re-peated to rule out laboratory error, unlessthe diagnosis is clear on clinical grounds,such as a patient with classic symptoms ofhyperglycemia or hyperglycemic crisis. Itis preferable that the same test be repeatedfor confirmation, since there will be agreater likelihood of concurrence in thiscase. For example, if the A1C is 7.0% anda repeat result is 6.8%, the diagnosis ofdiabetes is confirmed. However, there arescenarios in which results of two differenttests (e.g., FPG and A1C) are available forthe same patient. In this situation, if thetwo different tests are both above the di-agnostic thresholds, the diagnosis of dia-betes is confirmed.

On the other hand, when two differ-ent tests are available in an individual andthe results are discordant, the test whoseresult is above the diagnostic cut pointshould be repeated, and the diagnosis ismade on the basis of the confirmed test.That is, if a patient meets the diabetes cri-terion of the A1C (two results �6.5%) butnot the FPG (�126 mg/dl or 7.0 mmol/l),or vice versa, that person should be con-sidered to have diabetes. Admittedly, inmost circumstance the “nondiabetic” testis likely to be in a range very close to thethreshold that defines diabetes.

Since there is preanalytic and analyticvariability of all the tests, it is also possiblethat when a test whose result was above

the diagnostic threshold is repeated, thesecond value will be below the diagnosticcut point. This is least likely for A1C,somewhat more likely for FPG, and mostlikely for the 2-h PG. Barring a laboratoryerror, such patients are likely to have testresults near the margins of the thresholdfor a diagnosis. The healthcare profes-sional might opt to follow the patientclosely and repeat the testing in 3– 6months.

The decision about which test to useto assess a specific patient for diabetesshould be at the discretion of the healthcare professional, taking into account theavailability and practicality of testing anindividual patient or groups of patients.Perhaps more important than which diag-nostic test is used, is that the testing fordiabetes be performed when indicated.There is discouraging evidence indicatingthat many at-risk patients still do not receiveadequate testing and counseling for this in-creasingly common disease, or for its fre-quently accompanying cardiovascular riskfactors. The current diagnostic criteria fordiabetes are summarized in Table 3.

Diagnosis of GDMAt the time of publication of this state-ment, the criteria for abnormal glucosetolerance in pregnancy are those of Car-penter and Coustan (11). Recommenda-tions from ADA’s Fourth InternationalWorkshop-Conference on GestationalDiabetes Mellitus held in March 1997support the use of the Carpenter/Coustandiagnostic criteria as well as the alterna-tive use of a diagnostic 75-g 2-h OGTT.These criteria are summarized below.Testing for gestational diabetes. Previ-ous recommendations included screeningfor GDM performed in all pregnancies.However, there are certain factors thatplace women at lower risk for the devel-opment of glucose intolerance duringpregnancy, and it is likely not cost-effective to screen such patients. Pregnantwomen who fulfill all of these criterianeed not be screened for GDM.

This low-risk group compriseswomen who:

● are �25 years of age● are a normal body weight● have no family history (i.e., first-degree

relative) of diabetes● have no history of abnormal glucose

metabolism● have no history of poor obstetric

outcome● are not members of an ethnic/racial

group with a high prevalence of diabe-tes (e.g., Hispanic American, NativeAmerican, Asian American, AfricanAmerican, Pacific Islander)

Risk assessment for GDM should beundertaken at the first prenatal visit.Women with clinical characteristics con-sistent with a high risk of GDM (markedobesity, personal history of GDM, glyco-suria, or a strong family history of diabe-tes) should undergo glucose testing (seebelow) as soon as feasible. If they arefound not to have GDM at that initialscreening, they should be retested be-tween 24 and 28 weeks of gestation.Women of average risk should have test-ing undertaken at 24 –28 weeks ofgestation.

An FPG level �126 mg/dl (7.0mmol/l) or a casual plasma glucose �200mg/dl (11.1 mmol/l) meets the thresholdfor the diagnosis of diabetes. In the ab-sence of unequivocal hyperglycemia, thediagnosis must be confirmed on a subse-quent day. Confirmation of the diagnosisprecludes the need for any glucose chal-lenge. In the absence of this degree of hy-perglycemia, evaluation for GDM inwomen with average or high-risk charac-teristics should follow one of twoapproaches.One-step approach. Perform a diagnos-tic OGTT without prior plasma or serumglucose screening. The one-step approachmay be cost-effective in high-risk patientsor populations (e.g., some Native-American groups).Two-step approach. Perform an initialscreening by measuring the plasma or se-rum glucose concentration 1 h after a50-g oral glucose load (glucose challengetest [GCT]) and perform a diagnosticOGTT on that subset of women exceedingthe glucose threshold value on the GCT.When the two-step approach is used, aglucose threshold value �140 mg/dl (7.8mmol/l) identifies �80% of women withGDM, and the yield is further increased to90% by using a cutoff of �130 mg/dl (7.2mmol/l).

With either approach, the diagnosisof GDM is based on an OGTT. Diagnosticcriteria for the 100-g OGTT are derivedfrom the original work of O’Sullivan andMahan (12) modified by Carpenter andCoustan (11) and are shown at the top ofTable 4. Alternatively, the diagnosis canbe made using a 75-g glucose load and theglucose threshold values listed for fasting,1 h, and 2 h (Table 4, bottom); however,

Diagnosis and Classification

S68 DIABETES CARE, VOLUME 33, SUPPLEMENT 1, JANUARY 2010 care.diabetesjournals.org

this test is not as well validated as the100-g OGTT.

Results of the Hyperglycemia andAdverse Pregnancy Outcomes study(13), a large-scale (�25,000 pregnantwomen) multinational epidemiologicstudy, demonstrated that risk of adversematernal, fetal, and neonatal outcomescontinuously increased as a function ofmaternal glycemia at 24 –28 weeks,even within ranges previously consid-ered normal for pregnancy. For mostcomplications, there was no thresholdfor risk. These results have led to carefulreconsideration of the diagnostic crite-ria for GDM. The IADPSG recom-mended that all women not known tohave prior diabetes undergo a 75-gOGTT at 24 –28 weeks of gestation. Thegroup developed diagnostic cut pointsfor the fasting, 1-h, and 2-h plasma glu-cose measurements that conveyed anodds ratio for adverse outcomes of atleast 1.75 compared with women withthe mean glucose levels in the HAPOstudy.

At the time of publication of this up-date, ADA is planning to work with U.S.

obstetrical organizations to consideradoption of the IADPSG diagnostic crite-ria and to discuss the implications of thischange. While this change will signifi-cantly increase the prevalence of GDM,there is mounting evidence that treatingeven mild GDM reduces morbidity forboth mother and baby (14).

Acknowledgments— The American Diabe-tes Association thanks the following volunteermembers of the writing group for the updatedsections on diagnosis and categories of in-creased risk: Silvio Inzucchi, MD; Richard Ber-genstal, MD; Vivian Fonseca, MD; EdwardGregg, PhD; Beth Mayer-Davis, MSPH, PhD,RD; Geralyn Spollett, MSN, CDE, ANP; andRichard Wender, MD.

References1. Expert Committee on the Diagnosis and

Classification of Diabetes Mellitus. Report ofthe Expert Committee on the Diagnosis andClassification of Diabetes Mellitus. DiabetesCare 1997;20:1183–1197

2. Genuth S, Alberti KG, Bennett P, Buse J,Defronzo R, Kahn R, Kitzmiller J, KnowlerWC, Lebovitz H, Lernmark A, Nathan D,Palmer J, Rizza R, Saudek C, Shaw J, Ste-ffes M, Stern M, Tuomilehto J, Zimmet P,Expert Committee on the Diagnosis andClassification of Diabetes Mellitus2, theExpert Committee on the Diagnosis andClassification of Diabetes Mellitus. Fol-low-up report on the diagnosis of dia-betes mellitus. Diabetes Care 2003;26:3160–3167

3. International Expert Committee. Interna-tional Expert Committee report on therole of the A1C assay in the diagnosisof diabetes. Diabetes Care 2009;32:1327–1334

4. Edelman D, Olsen MK, Dudley TK, HarrisAC, Oddone EZ. Utility of hemoglobinA1c in predicting diabetes risk. J Gen In-tern Med 2004;19:1175–1180

5. Pradhan AD, Rifai N, Buring JE, Ridker PM.Hemoglobin A1c predicts diabetes but notcardiovascular disease in nondiabeticwomen. Am J Med 2007;120:720–727

6. Sato KK, Hayashi T, Harita N, Yoneda T,Nakamura Y, Endo G, Kambe H. Com-bined measurement of fasting plasma glu-

cose and A1C is effective for theprediction of type 2 diabetes: the KansaiHealthcare Study. Diabetes Care 2009;32:644–646

7. Shimazaki T, Kadowaki T, Ohyama Y,Ohe K, Kubota K. Hemoglobin A1c(HbA1c) predicts future drug treatmentfor diabetes mellitus: a follow-up studyusing routine clinical data in a Japaneseuniversity hospital. Translational Re-search 2007;149:196–204

8. Geiss LS, Pan L, Cadwell B, Gregg EW, Ben-jamin SM, Engelgau MM. Changes in inci-dence of diabetes in U.S. adults, 1997–2003. Am J Prev Med 2006;30:371–377

9. Knowler WC, Barrett-Connor E, FowlerSE, Hamman RF, Lachin JM, Walker EA,Nathan DM, Diabetes Prevention Pro-gram Research Group. Reduction in theincidence of type 2 diabetes with lifestyleintervention or metformin. N Engl J Med2002;346:393–403

10. Droumaguet C, Balkau B, Simon D, CacesE, Tichet J, Charles MA, Eschwege E, theDESIR Study Group. Use of HbA1c in pre-dicting progression to diabetes in Frenchmen and women: data from an Epidemi-ological Study on the Insulin ResistanceSyndrome (DESIR) Diabetes Care 2006;29:1619–1625.

11. Carpenter MW, Coustan DR. Criteria forscreening tests for gestational diabetes. Am JObstet Gynecol 1982;144:768–773

12. O’Sullivan JB, Mahan CM. Criteria for theoral glucose tolerance test in pregnancy.Diabetes 1964;13:278

13. HAPO Study Cooperative ResearchGroup, Metzger BE, Lowe LP, Dyer AR,Trimble ER, Chaovarindr U, Coustan DR,Hadden DR, McCance DR, Hod M, McIn-tyre HD, Oats JJ, Persson B, Rogers MS,Sacks DA. Hyperglycemia and adversepregnancy outcomes. N Engl J Med 2008;358:1991–2002

14. Landon MB, Spong CY, Thom E, Carpen-ter MW, Ramin SM, Casey B, Wapner RJ,Varner MW, Rouse DJ, Thorp JM Jr,Sciscione A, Catalano P, Harper M, SaadeG, Lain KY, Sorokin Y, Peaceman AM, To-losa JE, Anderson GB, Eunice KennedyShriver National Institute of Child Healthand Human Development Maternal-FetalMedicine Units Network. A multicenter,randomized trial of treatment for mildgestational diabetes. N Engl J Med 2009;361:1339–1348

Table 4—Diagnosis of GDM with a 100-g or75-g glucose load

mg/dl mmol/l

100-g glucose loadFasting 95 5.31-h 180 10.02-h 155 8.63-h 140 7.8

75-g glucose loadFasting 95 5.31-h 180 10.02-h 155 8.6

Two or more of the venous plasma concentrationsmust be met or exceeded for a positive diagnosis.The test should be done in the morning after anovernight fast of between 8 and 14 h and after at least3 days of unrestricted diet (�150 g carbohydrate perday) and unlimited physical activity. The subjectshould remain seated and should not smokethroughout the test.

Position Statement

care.diabetesjournals.org DIABETES CARE, VOLUME 33, SUPPLEMENT 1, JANUARY 2010 S69