diagnostic criteria for diabetes mellitus

Diabetes Research and Clinical Practice 55 (2002) 65–85

Discussion

Report of the Committee on the classification anddiagnostic criteria of diabetes mellitus�

The Committee of the Japan Diabetes Society on the diagnostic criteria ofdiabetes mellitus, Takeshi Kuzuya a,*, Shoichi Nakagawa b, Jo Satoh c,

Yasunori Kanazawa d, Yasuhiko Iwamoto e, Masashi Kobayashi f,Kisihio Nanjo g, Akira Sasaki h, Yutaka Seino i, Chikako Ito j, Kenji Shima k,

Kyohei Nonaka l, Takashi Kadowaki m

a Aino Institute for Aging Research, Ibaraki-shi, Osaka-fu, Japanb Hokkaido Uni�ersity, Sapporo, Japan

c Di�ision of Molecular Metabolism and Diabetes, Tohoku Uni�ersity School of Medicine, Sendai, Japand Omiya Medical Center, Jichi Medical School, Omiya, Japan

e Diabetes Center, Tokyo Women’s medical Uni�ersity School of Medicine, Tokyo, Japanf Toyama Medical and Pharmaceutical Uni�ersity, Toyama, Japan

g First Department of Medicine, Wakayama Uni�erstiy of Medical Science, Wakayama, Japanh Department of Epidemiology, Osaka Seijinbyo Center, Medical Institute for Noncommunicable Diseases, Osaka, Japan

i Department of Metabolism and Clinical Nutrition, Graduate School of Medicine, Kyoto Uni�ersity, Kyoto, Japanj Hiroshima A-Bomb Casualty Council Health Management Center, Hiroshima, Japan

k Kawashima Hospital, Tokushima, Japanl Shiroishi Joint Hospital, Shiroishi-cho, Japan

m Department of Metabolic Diseases, Graduate School of Medicine, Uni�ersity of Tokyo, Tokyo, Japan

Accepted 31 August 2001

Abstract

In 1995, the Japan Diabetes Society (JDS) appointed the Committee for the Classification and Diagnosis ofDiabetes Mellitus. The Committee presented a final report in May 1999 in Japanese. This is the English version withminor modifications for readers outside Japan.

Concept of diabetes mellitus: Diabetes mellitus represents a group of diseases of heterogeneous etiology, character-ized by chronic hyperglycemia and other metabolic abnormalities, which are due to deficiency of insulin effect. After

www.elsevier.com/locate/diabres

� This article is the English version of the original report published in Journal of the Japan Diabetes Society 42: 385–404, 1999in Japanese, adapted for readers outside Japan. References are updated.

* Corresponding author. Present address: 9-40 Kyo-Machi, Utsunomiya, Tochigi 320-0842, Japan. Tel./fax: +81-28-634-9070.E-mail address: [email protected] (T. Kuzuya).

0168-8227/02/$ - see front matter © 2002 Elsevier Science Ireland Ltd. All rights reserved.

PII: S 0168 -8227 (01 )00365 -5

mailto:[email protected]

T. Kuzuya et al. / Diabetes Research and Clinical Practice 55 (2002) 65–8566

a long duration of metabolic derangement, specific complications of diabetes (retinopathy, nephropathy, andneuropathy) may occur. Arteriosclerosis is also accelerated. Depending on the severity of the metabolic abnormality,diabetes may be asymptomatic, or may be associated with symptoms (thirst, polyuria, and weight loss), or mayprogress to ketoacidosis and coma.

Classification (cf. Tables 1 and 2 and Fig. 1): Both etiological classification and staging of pathophysiology by thedegree of deficiency of insulin effect need to be considered. The etiological classification of diabetes and relateddisorders of glycemia includes, (1) type 1; (2) type 2; (3) those due to specific mechanisms and diseases; and (4)gestational diabetes mellitus. Type 1 is characterized by destructive lesions of pancreatic � cells either by anautoimmune mechanism or of unknown cause. Type 2 diabetes is characterized by combinations of decreased insulinsecretion and decreased insulin sensitivity (insulin resistance). Category (3) includes two subgroups; subgroup A isdiabetes in which specific mutations have been identified as a cause of genetic susceptibility, while subgroup B isdiabetes associated with other pathologic conditions or diseases. The staging of glucose metabolism includes normal,borderline and diabetic stages. The diabetic stage is further classified into three substages; non-insulin requiring,insulin-requiring for glycemic control, and insulin-dependent (ID) for survival. In each individual, these stages mayvary according to the deterioration or the improvement of the metabolic state, either spontaneously or by treatment.

Diagnosis (cf. Tables 3 and 4): The confirmation of chronic hyperglycemia is a prerequisite for the diagnosis ofdiabetes mellitus. The state of glycemia may be classified within three categories, diabetic type; borderline type; andnormal type. Diabetic type is defined when fasting plasma glucose (FPG) is 7.0 mmol/l (126 mg/dl) or higher, and/orplasma glucose 2 h after 75 g glucose load (2hPG) is 11.1 mmol/l (200 mg/dl) or higher. A casual plasma glucose (PG)�11.1 mmol/l (200 mg/dl) also indicates diabetic type. Normal type is defined when FPG is below 6.1 mmol/l (110mg/dl) and 2hPG below 7.8 mmol/l (140 mg/dl). Borderline type includes those who are neither diabetic nor normaltypes. These cutoff values are for venous PG measurements. The persistence of ‘diabetic type’ in a subject indicatesthat he or she has diabetes. For children, a dose of 1.75 g/kg (maximum, 75 g) is used for oral glucose tolerance test(OGTT). The procedure for clinical diagnosis is as follows.1. Diabetes mellitus is diagnosed when hyperglycemia meeting the criteria for ‘diabetic type’ is shown on two or

more occasions examined on separate days.2. Diabetes can be diagnosed by a single PG test of ‘diabetic type’ if one of the following three conditions co-exists,

(1) typical symptoms of diabetes mellitus; (2) HbA1c �6.5% by a standardized method; or (3) unequivocaldiabetic retinopathy.

3. If the above conditions ((1) or (2)) have been present in the past and well documented, the subject is diagnosedeither to have diabetes or to be suspected of diabetes, even if the present level of glycemia does not reach that of‘diabetic type’.

4. If the diagnosis of diabetes cannot be established by these procedures, re-testing of PG is recommended after anappropriate interval.

5. The physician should assess not only the presence or absence of diabetes, but also its etiology and glycemic stage,and the presence and absence of diabetic complications or associated conditions.

Epidemiological aspects and screening: In order to determine the prevalence of diabetes in a population, ‘diabetictype’ may be regarded as ‘diabetes’. The use of 2hPG cutoff level of �11.1 mmol/l (200 mg/dl) is recommended. Ifthis is difficult, the FPG cutoff level of �7.0 mmol/l (126 mg/dl) can be used, but is likely to lead tounder-ascertainment. For screening, the most important point is not to overlook ‘diabetes’. In addition to parametersof hyperglycemia, clinical information such as family history, obesity etc., should be regarded as indications forfurther testing.

Normal type and borderline type: Only FPG and 2hPG are adopted as cutoff values, but in clinical situations, it isrecommended to measure PG also at 30 and 60 min during 75 g OGTT. Among people with normal type, those with1hPG higher than 10.0 mmol/l (180 mg/dl) are at higher risk to develop diabetes than those with lower 1hPG. WhenOGTT is performed, the borderline type corresponds to the sum of impaired fasting glycemia (IFG) plus impairedglucose tolerance (IGT) according to the new WHO report. Subjects in this category are at higher risk of developingdiabetes than those with ‘normal type’. Those with low insulinogenic index (the ratio of increment of plasma insulinto that of PG at 30 min during OGTT) are at particularly high risk to develop diabetes. Microvascular complicationsare rare but arteriosclerotic complications are fairly frequent in this category.

Gestational diabetes mellitus (GDM): The current definition of GDM is ‘ any glucose intolerance developed ordetected during pregnancy’. We adopt the proposal of the Japan Society of Gynecology and Obstetrics for thediagnosis of GDM (1984). GDM is defined when two or more values during a 75 g OGTT are higher than thefollowing cutoff levels; FPG �5.5 mmol/l (100 mg/dl), 1hPG �10.0 mmol/l (180 mg/dl) and 2hPG �8.3 mmol/l(150 mg/dl). As a screening test, subjects with casual PG �5.5 mmol/l (100 mg/dl) are recommended for further

T. Kuzuya et al. / Diabetes Research and Clinical Practice 55 (2002) 65–85 67

testing. Patients who have had documented glucose intolerance before pregnancy, and who present as ‘diabetic type’should be under closer supervision than those who develop GDM during pregnancy for the first time.

HbA1c : There is a large overlap in the distribution of HbA1c between groups with ‘normal type’ and ‘borderlinetype’ and mild ‘diabetic type’. Therefore, HbA1c is not a suitable parameter to detect mild glucose intolerance. HbA1c

higher than 6.5% suggests diabetes, but HbA1c below 6.5% alone should not be taken as evidence against thediagnosis of diabetes.

Comparison with reports of American Diabetes Association (ADA) in 1997 and WHO in 1999: The present report isunique in the following points when compared with those of the ADA ‘Diabetes Care 20 (1997) 1183’ and WHO‘Report of a WHO Consultation (1999)’. (1) Diabetes due to specific mechanisms and diseases is divided into twosubgroups; diabetes in which genetic susceptibility is clarified at the DNA level and diabetes associated with otherdiseases or conditions. (2) Cutoff PG levels are the same as those of ADA and WHO, but a term ‘type’ is added toeach glycemic category, because a single coding of ‘diabetic type’ hyperglycemia does not define diabetes. Diabetes isdiagnosed when ‘diabetic type’ hyperglycemia is shown on two or more occasions. (3) A single ‘diabetic type’hyperglycemia is considered sufficient for the diagnosis of diabetes, if the patient has typical symptoms, HbA1c

�6.5%, or diabetic retinopathy. (4) OGTT is recommended for those with mild hyperglycemia, because FPG criteriaalone would overlook many subjects with ‘diabetic type’ in Japan. High 1hPG without elevation of FPG and 2hPGis also considered to be a risk factor for future diabetes. (5) Borderline type in the present report corresponds to thesum of IFG and IGT by WHO when OGTT is performed. (6) New criteria for GDM by OGTT are proposed.© 2001 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Diabetes mellitus; JDS; Diagnosis; Classification

1. Introduction

In the past, the Japan Diabetes Society (JDS)has presented two reports on the diagnostic crite-ria for diabetes mellitus, in 1970 and 1982 [1,2].Before 1970, the methods for glucose tolerancetesting and the diagnostic criteria for diabeteswere not standardized. Various criteria were used,using different glucose load, and mutual compari-son of results was almost impossible.

The first Committee of JDS presented cutofflevels of blood glucose for 50 g and 100 g glucosetolerance tests (OGTT). In this 1970 report, asubtitle ‘A recommendation on the criteria ofOGTT to be used for the diagnosis of diabetesmellitus’ was added [1]. The Committee took thestandpoint that OGTT is an important test tohelp diagnosis but not to define diabetes.

In 1979, the National Diabetes Data Group(NDDG) in USA proposed a classification ofdiabetes and diagnostic criteria using a 75 gOGTT [3]. In 1980, the WHO Expert Committeepresented a report similar to that of NDDG [4].These two reports adopted diagnostic criteria us-ing a 75 g OGTT, and classified diabetes intoIDDM (type 1), non-insulin-dependent diabetes(NIDDM) (type 2), and other types. They also

created the concept of impaired glucose tolerance(IGT).

After publication of these reports, JDS orga-nized the second Committee to examine the diag-nosis of diabetes mellitus and to revise the 1970JDS recommendation. This Committee regardedthe WHO report as a basis for international stan-dardization. In addition, the Committee contin-ued the principle of the first JDS Committee,namely that diabetes mellitus is not defined simplyby glycemic cutoff values but is a disease (ordiseases) which possesses other clinical character-istics [2]. The report of the second JDS Commit-tee in 1982 adopted the terms, ‘diabetic type’,‘borderline type’, and ‘normal type’ to describethe OGTT results. The cutoff plasma glucose(PG) values for diabetic type were set as the sameas those for ‘diabetes’ by WHO criteria, whereasthe cutoff values for ‘normal type’ were set lowerthan the lower limits of IGT. The normal typewas defined as a group which would not progressto diabetes after follow-up of several years. Theborderline type was defined as those being neithernormal nor of diabetic type.

The borderline type of the 1982 JDS Committeeincluded not only IGT but also milder degrees ofglucose intolerance than IGT, thus creating two


different categories of mild glucose intolerance(i.e. borderline type and IGT). Some researcherscomplained of the inconvenience of having twosuch categories. The Committee adopted theWHO classification of diabetes without modifica-tion. In 1985, WHO made a small revision to the1980 report [5].

Since that time, many discoveries have beenmade regarding the etiology of diabetes, and epi-demiological data have accumulated. Moves, toseek a new classification and to revise the diagnos-tic criteria for diabetes mellitus, have emerged inJapan as well as internationally. In 1995, JDSassigned a new Committee to reconsider the clas-sification and diagnostic criteria of diabetes.Meanwhile in 1997, the American Diabetes Asso-ciation (ADA) proposed a new classificationbased on etiology, and revised criteria for thediagnosis of diabetes mellitus [6]. WHO also pre-sented a provisional report in 1998 and a finalreport in 1999 [7].

The new Committee of JDS sent inquiries toCouncil members of the JDS, asking about prob-lems in the old 1982 JDS report [8]. The Commit-tee evaluated the 1997 ADA and 1998 WHOreports carefully after their publication, and pro-ceeded to hold an ad-hoc symposium on thediagnosis and classification of diabetes mellitus inTokyo in 28 June 1998 (Proceedings, J. Jpn. Dia-betes Soc. 41 (Suppl. 2), 1999). The Committeemet 16 times, held intensive discussions on pub-lished and unpublished data obtained in Japanand produced the following final report.

2. Concept of diabetes mellitus

Diabetes mellitus is a group of diseases charac-terized by chronic hyperglycemia due to deficiencyof insulin action. The deficiency of insulin action,a common basis of diabetes, leads to characteris-tic abnormalities in the metabolism of carbohy-drate, lipid, protein and so on.

The causes of diabetes are multiple. Both ge-netic and environmental factors play roles in itsetiology. The supply of insulin may be decreasedby a decrease in pancreatic � cell mass and/orfunctional disturbances of � cells. In some cases

of diabetes, the disease can be ascribed solely tothe deficiency of insulin supply, but in other casesthere are both insulin resistance and relative defi-ciency of insulin. The metabolic abnormality canbe improved by various means to ameliorate thedeficiency of insulin effect.

If the metabolic abnormality is mild, patientsmay be asymptomatic, while in the presence ofovert hyperglycemia, characteristic symptomssuch as thirst, polydipsia, polyuria and weight lossoften occur. In severe cases, ketoacidosis or hy-perglycemic–hyperosmolar states may occur andlead to disturbances of consciousness, coma andeven death unless treated appropriately.

With long duration of diabetic metabolism, dia-betes-specific complications, chiefly involvingsmall vessels (retinopathy, nephropathy and neu-ropathy), may ensue, and lead to serious out-comes such as visual disturbance, renal failure,and gangrene. Diabetes accelerates and exacer-bates the occurrence of arteriosclerosis, increasingthe risks for myocardial infarction, cerebral in-farction and occlusive artery disease of the lowerextremities. These complications constitute themajor causes of morbidity and mortality in dia-betic patients.

3. Classification

3.1. Distinction between etiology and themetabolic stages of diabetes mellitus

The 1980 classification by WHO consists ofclinical classes and statistical risk groups [4]. Clin-ical classes included diabetes mellitus, IGT andgestational diabetes. Diabetes was furtherclassified into IDDM (= type 1), NIDDM(= type 2) and other types associated with specificdiseases or syndromes. Statistical risk groups in-cluded previous abnormality of glucose toleranceand potential abnormality of glucose tolerance. Inthe 1985 WHO report, the terms type 1 and type2 were eliminated from the classification, andmalnutrition-related diabetes mellitus (MRDM)was added [5]. In this classification, the degree ofinsulin deficiency and etiology were used in com-bination as means to classify diabetes.


The new ADA report in 1997 and the WHOreport in 1999 adopted a new classification pri-marily based on etiology [6,7]. Etiological classifi-cation of diseases is ideal, but the methods toidentify the pathogenesis of diabetes are not yetsatisfactory, so that we cannot always identify theetiology of diabetes in each patient.

IDDM, in the previous WHO report, wasdefined as encompassing patients who need in-sulin therapy to sustain life. NIDDM was definedby exclusion. Those who were neither IDDM,MRDM nor other types were labeled as NIDDM.The terms IDDM and NIDDM imply states orstages of diabetes representing different degrees ofinsulin deficiency. Evaluation of the degree ofdeficiency of insulin or insulin effect is usuallypossible, and is still useful in clinical practice inaddition to etiological classification.

Etiology and patho-physiological states (orstages) should be assessed separately for eachpatient [9,10]. Before development of overt dia-betes, patients will pass stages at which differentdegrees of deficiency of insulin effect exist. Theabnormal carbohydrate metabolism not only pro-gresses but may regress spontaneously or in re-sponse to treatment. For example, isletautoantibodies are occasionally detected beforerecognition of hyperglycemia, suggesting that theautoimmune process of type 1 diabetes has al-ready begun. Overt type 1 diabetes may some-times be improved temporarily to anon-insulin-requiring state, the so-called ‘honey-moon period’. Obese diabetic patients may beimproved to borderline type or even to normalglucose tolerance after weight reduction with diettherapy. Considering these phenomena, the etiol-ogy (mechanisms) and the degree of deficiency ofinsulin action reflected by glycemic states (stages)can be represented in two-dimensions (2-D) asshown in Fig. 1 [6,7,10].

The horizontal axis represents the degree ofdeficiency of insulin effect or the degree of abnor-mality of glucose metabolism. The patients arejudged to have diabetes when hyperglycemia hasexceeded a certain level, which is presumed toconfer risk for specific complications. The diabeticarea is divided into three stages, (1) in whichinsulin treatment is not needed; (2) in which in-

sulin injections are required for glycemic control;and (3) in which insulin treatment is indispensableto prevent ketosis and to sustain life.

As the terms for etiological classification, type 1and type 2 are used. The terms insulin-dependent(ID) state (stage) and non-insulin-dependent(NID) state (stage) can be used to describe thedegree of deficiency of insulin effect. An ID statemeans that the patient is at immediate risk ofdeveloping ketosis or is unable to survive withoutinsulin injection.

3.2. Etiological classification

Table 1 shows the new classification, whichincludes type 1, type 2, other specific types andgestational diabetes mellitus.

Fig. 1. A scheme of the relationship between etiology (mecha-nism) and patho-physiological states (stages) of diabetes melli-tus The arrows towards the right indicate deterioration ofglucose metabolism, including the development of diabetes.The filled portion of solid and broken lines represents thestates regarded as ‘diabetes’. The broken lines indicate uncom-mon phenomena. For example, a patient with type 2 diabetesusually does not need insulin injection for survival, but maydevelop ketoacidosis in association with severe infection. Thearrows towards the left are filled for their full length, meaningthat patients who have once developed diabetes, should beregarded to have diabetes, even if he or she improves toresume borderline or normal glucose tolerance. NID and IDstates in the diabetic area, correspond to NIDDM and IDDMin the previous classification, respectively (adapted fromKuzuya and Matsuda [10], the Expert Committee of ADA [6]and WHO Consultation [7]).


Table 1Etiological classification of diabetes mellitus and related disorders of glucose tolerance

1. Type 1, (caused by destructive lesions A, Autoimmuneof pancreatic � cells, usually leading to B, Idiopathicabsolute insulin deficiency)2. Type 2, (ranging from predominantly

insulin secretory defect, to apredominantly insulin resistance withvarying degree of insulin secretorydefect)

A, Those in which specific mutations (1) Genetic abnormalities of pancreatic3. Due to other specific mechanisms orhave been identified as a cause of geneticdiseases (details in Table 2) � cell function

(2) Genetic abnormalities of insulinsusceptibilityaction

B, Those associated with other diseases (1) Diseases of exocrine pancreas(2) Endocrine diseasesor conditions(3) Liver diseases(4) Drug- or chemical-induced(5) Infections(6) Rare forms of immune-mediateddiabetes(7) Various genetic syndromes oftenassociating diabetes

4. Gestational diabetes mellitus

The occurrence of diabetes-specific complications has not been confirmed in some of these conditions. This classification differs fromthose of ADA [6] and WHO [7] in two points; category 3 is subdivided into A and B, and liver diseases are added in category 3B.

3.2.1. Type 1 diabetesThis type of diabetes occurs as a result of

deficiency of insulin due to destructive lesions ofpancreatic � cells, and usually progresses to thestage of absolute insulin deficiency. In the major-ity of Japanese patients with type 1 diabetes, anautoimmune process is thought to play a majorrole in � cell destruction. Auto-antibodies to isletcell antigens are detected in many patients (70–90%), particularly at the early period after onset[11]. Typically, it occurs in young subjects withacute onset, but may occur at any age, sometimeswith slow progression [12]. Occasionally, beforethe occurrence of clinical symptoms or hyper-glycemia, auto-antibodies or the loss of acuteinsulin response to intravenous glucose may bedetected. In some patients who progress to an IDstate, auto-antibodies to islet antigens are neverdetected. Such patients are classified as ‘idiopathictype 1 diabetes’. Patients developing an ID stateas a result of mitochondrial mutations or otherknown causes are not included as type 1 diabetes.

In the future, type 1 diabetes may be furtherdivided into subtypes according to the mode ofonset (i.e. acute or slowly-progressive), HLA anti-gens, or epitopes of autoantigens [12,13].

3.2.2. Type 2 diabetesMost patients previously called NIDDM be-

long to this category. In this type, the mass ofpancreatic � cells and their function are preservedto some extent, and insulin injection is seldomneeded to sustain life. Ketoacidosis could occur,however, in the presence of severe infection orother stress. Both decreased insulin secretion anddecreased insulin sensitivity (insulin resistance) areinvolved in its pathogenesis. Insulin resistancemay not always be present. The relative role ofthese two factors varies between patients. Withregard to insulin secretion, the acute insulin re-sponse to a glucose load is characteristically de-fective. The majority of patients are obese or havebeen obese in the past.


Table 2Diabetes and related disorders of glucose tolerance due toother specific mechanisms and diseases

A, Those in which specific (1) Genetic abnormalities ofmutations have been pancreatic � cell functionidentified as a cause of Genes for insulin (abnormalgenetic susceptibility insulinemia, abnormal

proinsulinemia)HNF 4� gene (MODY 1)Glucokinase gene (MODY 2)HNF 1� gene (MODY 3)IPF-1 gene (MODY 4)HNF 1� (MODY 5)Mitochondria gene (MIDD)Amylin geneOthers(2) Genetic abnormalities ofinsulin actionInsulin receptor genemutations (type A insulinresistance, leprechaunism,Rabson–Mendenhallsyndrome etc.)Others(1) Exocrine pancreaticB, Those associated withdiseasesother specific diseases orPancreatitisconditionsTrauma/pancreatectomyNeoplasmHemochromatosisOthers(2) Endocrine diseasesCushing syndromeAcromegalyPheochromocytomaGlucagonomaAldosteroinismHyperthyroidismSomatostatinomaOthers(3) Liver diseasesChronic hepatitisLiver cirrhosisOthers(4) Drug- or chemical-inducedGlucocorticoidsInterferonOthers(5) InfectionsCongenital rubellaCytomegalovirusEpstein–Barr virusCoxackie B virusMumps virusOthers(6) Rare forms ofimmune-mediated diabetes

Table 2 (Continued)

Anti-insulin receptorantibodiesStiffman syndromeInsulin autoimmunesyndromeOthers(7) Various geneticsyndromes often associatingdiabetesDown syndromePrader–Willi syndromeTurner syndromeKlinefelter syndromeWerner syndromeWolfram syndromeCeruloplasmin deficiencyLipoatrophic diabetesMyotonic dystrophyOthers

The occurrence of diabetes-specific complications has not beenconfirmed in some of these conditions.

Typically, this type of diabetes develops aftermiddle age, but may occur in younger people.Screening by urinalysis of large numbers ofschoolchildren has revealed that type 2 diabeteshas been steadily increasing since the 1970s[14,15]. The incidence of type 2 is very low underthe age of 10 years, but exceeds the incidence oftype 1 diabetes in mid-teens. About 80% of themare obese. NID type diabetes with autosomaldominant inheritance in young people has beencalled MODY (maturity-onset diabetes of theyoung) [16]. In many of these, specific DNAabnormalities have been identified [17]. MODYsare now classified into another category separatedfrom the common forms of type 2 diabetes.

Genetic abnormalities have not yet been iden-tified in most patients with type 2 diabetes. Poly-genic inheritance is presumed in these patients. Iffuture investigations reveal new genetic abnormal-ities leading to type 2 diabetes, these would bereclassified into other categories. Type 2 diabetesis not a homogenous entity. There is a possibilityof subclassification according to the presence orabsence of obesity, the varying relative roles ofinsulin secretory and sensitivity defects, but the


method of subclassification has not yet beenestablished.

3.2.3. Other types of diabetes due to specific causesWe classify these cases into two major groups

(Table 2).

3.2.3.1. Diabetes in which specific mutations ha�ebeen identified as a cause of genetic susceptibility.Many susceptibility genes for diabetes have re-cently been identified [18]. They are divided into(1) genetic abnormalities related to � cell function;and (2) genetic abnormalities related to mecha-nisms of insulin action. The first category includesabnormalities of the insulin gene, and MODYcases. Mutations of genes for HNF 4�, glucoki-nase, HNF 1�, IPF-1 (PDX-1), HNF 1� corre-spond to MODYs 1, 2, 3, 4 and 5, respectively.Abnormalities of mitochondrial DNA [19] andthe amylin gene [20] are also included in thiscategory. The second category includes a numberof insulin receptor gene abnormalities [21].

3.2.3.2. Diabetes associated with other pathologicconditions or diseases. Diabetes and glucose intol-erance may occur as a manifestation of variouspathologic conditions, diseases and syndromes.Some have been previously called ‘secondary dia-betes’. These include pancreatic diseases, en-docrinopathies, liver diseases, exposure to certaindrugs and chemicals, viral infections, and variousgenetic syndromes, as shown in Table 2. We haveadded liver diseases to the corresponding tables ofADA [6] and WHO [7], because glucose intoler-ance is common (diabetic type, 12–40%) amongsubjects with hepatitis and liver cirrhosis [22,23].They often have a normal FPG but PG is elevatedafter glucose loading, presumably by decreasedhepatic uptake and increasing shunt of portalblood to systemic circulation.

3.2.4. Gestational diabetes mellitus (GDM)GDM is a state of glucose intolerance occurring

or detected for the first time during pregnancy.Pregnancy is a diabetogenic factor. Etiologically,many GDM patients probably share common ge-netic susceptibilities with type 1 or type 2 dia-betes, and the deterioration of glucose tolerance is

precipitated by the metabolic effect of pregnancy.The reason to treat GDM as an independentcategory is its clinical importance which requiresspecial consideration. During pregnancy, glucoseintolerance milder than ‘diabetic type’ may affectthe infant and mother adversely. Glucose intoler-ance during pregnancy is often normalized afterdelivery, but such cases are at higher risk ofdeveloping diabetes in the future. Problems on thediagnosis of GDM will be discussed later indetail.

3.3. Information to help classify diabetes

For etiological classification of diabetes, thefollowing clinical information is useful.1. Detailed information about the family history

of diabetes.2. Age of onset and the course of diabetes.3. Physical characteristics such as obesity, history

of weight changes in the past, hearing distur-bance (mitochondrial DNA abnormality), andacanthosis nigricans (severe insulin resistance).

4. Anti-GAD antibody, islet cell antibody, in-sulin autoantibody, and anti-IA2 antibody(the presence of any of these autoantibodiessuggests type 1 diabetes).

5. HLA antigens (DR4 and DR9, which are as-sociated with susceptibility to type 1 diabetes,are common among Japanese, and their ab-sence suggests that the patient is unlikely tohave autoimmune type 1 diabetes).

6. Tests for insulin secretion and insulin sensitiv-ity (plasma insulin and C-peptide levels, in-sulin response to glucose loading, euglycemicclamp, minimal model etc).

7. DNA analysis may give a definite diagnosis inspecial cases belonging to A(1) and A(2) inTable 2.

In order to assess the stage of diabetes, clinicalinformation (history of the disease, glycemic leveland its stability, ketosis-proneness, and responseto diet and drug therapy), plasma insulin assays(fasting and after glucose load, and after intra-venous glucagon), and C-peptide assays in plasmaand urine will help evaluate the degree of insulindeficiency.


4. Diagnosis of diabetes mellitus

For the diagnosis of diabetes mellitus, thephysician should evaluate whether the subject fitsto the concept of diabetes as mentioned earlier.The confirmation of chronic hyperglycemia is aprerequisite for diagnosis. The cut-off values offasting plasma glucose (FPG) and 2 h PG follow-ing 75 g oral glucose load (2hPG) are shown inTable 3. Persistence of hyperglycemia of ‘diabetictype’ in Table 3 indicates that the subject hasdiabetes. In order to confirm ‘persistent’ hyper-glycemia, at least two PG measurements on sepa-rate days are necessary. However, diabetes can bediagnosed following a single PG measurementshowing ‘diabetic type’ hyperglycemia, if the pa-tient has typical diabetic symptoms, a high HbA1c

value or definite diabetic retinopathy. If theseconditions have been present in the past and welldocumented, diabetes is diagnosed or should besuspected even though the current degree ofglycemia does not meet the criteria of ‘diabetictype’.

PG is often elevated temporarily in cases ofsevere stress (for example, infections, myocardialinfarction, cerebral hemorrhage and surgery).Therefore, excepting the situation of a severemetabolic disturbance necessitating immediatetreatment, the evaluation of hyperglycemia shouldbe made following resolution of such stressfulconditions.

We will first describe the procedures for diagno-sis in the clinical setting, and then proceed toepidemiological surveys and screening.

4.1. Clinical diagnosis

4.1.1. Procedures for the diagnosis of diabetesmellitus (Table 4)

1. Confirmation of persistent hyperglycemia.(c1) Diabetes is diagnosed when hyper-

glycemia meeting the criteria for diabetic type(FPG�7.0 mmol/l (126 mg/dl), 2hPG�11.1mmol/l (200 mg/dl), and/or casual PG�11.1mmol/l (200 mg/dl)) is recognized on two ormore occasions examined on separate days.These cutoff plasma glucose levels are thesame as those of ‘diabetes’ by ADA [6] andWHO [7] reports.

Casual plasma glucose is the plasma glucosevalue examined at any time regardless of theinterval after meals. FPG means the glucoselevel in the morning before breakfast aftermore than 10 h of fasting since the previousevening. The method for performing OGTTwill be described later.

(c2) When the first plasma glucose mea-surement meets the criteria of ‘diabetic type’,the subject is simply called ‘diabetic type’ untila similar degree of hyperglycemia is confirmedby a second test.

Table 3Criteria of fasting plasma glucose (FPG) and 2 h plasma glucose (2hPG) following 75 g glucose ingestion

Normal range Diabetic range

�6.1 (110) �7.0 (126)FPG2hPG after 75 g glucose �11.1 (200)�7.8 (140)

Normal type:Evaluation of OGTT Diabetic type:If both values belong to normal range If any of the two values falls into diabetic range

Borderline type:Neither normal nor diabetic types

Values are venous plasma in mmol/l. Figures in the brackets are values in mg/dl. Casual plasma glucose �11.1 mmol/l (200 mg/dl)is also regarded as to indicate diabetic type. Subjects with 1hPG �10.0 mmol/l (180 mg/dl) are advised to be followed similarly toborderline type, even if they belong to normal type, because it is known that such individuals are at higher risk of developingdiabetes than those with 1hPG �10.0 mmol/l (180 mg/dl).


Table 4The procedures of clinical diagnosis

1. Diabetes mellitus is diagnosed when hyperglycemiameeting criteria for ‘diabetic type’ in Table 3 is confirmedon 2 or more occasions examined on separate days2. Diabetes can be diagnosed by a single plasma glucose

test meeting criteria for ‘diabetic type’, when either oneof the following three conditions exits

(1) Typical symptoms of diabetes (i.e. thirst, polyuria,polydipsia, weight loss)

(2) HbA1c �6.5% (HbA1c should be determinedaccording to the recommendation of JDSCommittee for Standardization ofGlycohemoglobin)

(3) Unequivolcal diabetic retinopathy3. If the above conditions (1 or 2) existed in the past and

are well documented, the subject is diagnosed either ashaving diabetes, or should be suspected of diabetesregardless of the glycemic status at present

4. If the diagnosis of diabetes cannot be established by1–3., re-testing of plasma glucose after some interval isrecommended

5. At clinical diagnosis, the physicians should assess notonly the presence or absence of diabetes, but also itsetiology and glycemic stage, and the diabeticcomplications

The first and second tests of plasma glucose need not be doneby the same method. When the first test is carried out bycasual plasma glucose, the other methods are preferred as asecond test. If FPG is less than 7.8 mmol/l (140 mg/dl) at thefirst test, OGTT is recommended as the second test.

4. When the diagnosis of diabetes cannot bemade by (1)– (3) as mentioned above, the pa-tient should be followed, and re-testing is rec-ommended after some interval, at leastannually.

5. At clinical diagnosis, the physician shouldevaluate not only the presence or absence ofdiabetes, but also should pay attention to itsetiology, the degree of metabolic disturbanceand the presence of complications.

The methods used for the first and secondtests for glycemia need not be the same. Whenthe first test has been casual plasma glucose,the use of other methods is recommended asthe second test. Confirmation of the fastingcondition is particularly important when FPGis used as a diagnostic test. If FPG is between7.0 and 7.8 mmol/l (126 and 139 mg/dl) at thefirst test, we recommend an OGTT as thesecond test.

4.1.2. Oral glucose tolerance test (OGTT) and thecriteria for OGTT

4.1.2.1. Procedure of OGTT. OGTT evaluates therate of glucose disposal after an oral glucosechallenge, and is the most sensitive test to detect amild disturbance of glucose metabolism. Whencasual and FPG tests give ambiguous results,OGTT helps to establish whether or not the sub-ject has diabetes. In the clinical setting, OGTT isrecommended when FPG is in the borderlinerange (6.1–6.9 mmol/l, 110–125 mg/dl). In addi-tion, when FPG exceeds 7.0 mmol/l (126 mg/dl)but is below 7.8 mmol/l (140 mg/dl) in the dia-betic range, OGTT is useful to clarify thediagnosis.

In order to obtain accurate results, the follow-ing procedures are required. Recipients shouldtake meals including more than 150 g carbohy-drate for 3 days. A 250–350 ml solution of 75 ganhydrous glucose or its equivalent is adminis-tered orally and blood is drawn before and atcertain time intervals after glucose loading. Thetest should be done after an overnight fast for10–14 h. No food or drink except water norsmoking is allowed during the test. Blood shouldbe centrifuged immediately to separate plasma or

2. Diabetes can be diagnosed by a single plasmaglucose measurement which meets the criteriaof ‘diabetic type’, if one of the following threeconditions exists.

(c1) The presence of typical symptoms ofdiabetes (thirst, polydipsia, polyuria, andweight loss).

(c2) HbA1c�6.5%. HbA1c should be de-termined according to the recommendation ofthe JDS Committee for the Standardization ofGlycohemoglobin Assay.

(c3) The presence of definite diabeticretinopathy.

3. If the above conditions [(1) and (2)] existed inthe past and are well documented by the pa-tient’s medical record, diabetes is diagnosed orshould be suspected, even if the presentglycemic tests do not meet the criteria of dia-betic type.


kept cold, preferably with addition of NaF toprevent glycolysis until centrifugation. If a timedurine sample is taken, it can help to estimatethe threshold for glucosuria.

Starvation or low carbohydrate intake mayinduce glucose intolerance. In patients whohave undergone gastrectomy, plasma glucose of-ten reaches abnormally high levels at 30–60min.

At least, fasting and 2 h blood samplesshould be taken. In clinical cases, it is recom-mended also to measure plasma glucose also at30 and 60 min to enhance accurate diagnosis.The measurement of plasma insulin can help topredict the risk for future development of dia-betes, as discussed below.

4.1.2.2. Cutoff le�els for OGTT. Normal rangeand diabetic range are set for FPG and 2hPG(Table 3). Similarly to previous JDS recommen-dations, we classify the OGTT results into dia-betic, borderline and normal types according tocombinations of FPG and 2hPG.

Diabetic type. The cutoff plasma glucose val-ues are the same as those for ‘diabetes’ by 1997ADA and 1999 WHO criteria. In comparisonwith 1982 JDS criteria, the cutoff value of FPGhas been lowered from �7.8 (140) to �7.0mmol/l (126 mg/dl).

Normal type. Normal type is defined as arange of glycemia which is unlikely to developdiabetes after several years of observation. BothFPG and 2hPG should be within the normalrange. Even if FPG and 2hPG are in the nor-mal range, subjects with 1hPG higher than 10.0mmol/l (180 mg/dl) are at higher risk of devel-oping diabetes than those whose 1hPG is lower.Therefore, we recommend that such cases befollowed in the same way as those with border-line type.

Borderline type. This category is defined whenthe pattern of OGTT is neither diabetic nornormal type. Borderline type includes heteroge-neous conditions; a subject in transition to de-velop diabetes, diabetes in remission, andtemporary deterioration of glucose tolerance dueto stress in an essentially healthy individual.Subjects with borderline type are at little risk to

develop diabetes-specific microangiopathy but atincreased risk to progress to diabetes, and todevelop macroangiopathy.

Diabetic patients have a decreased early in-sulin response to glucose with low insulinogenicindex, �IRI/�PG (the ratio of increment ofplasma insulin to that of glucose) at 30 minafter oral glucose load. Borderline type and nor-mal type subjects (particularly borderline type)with a low insulinogenic index (�7.2 �U/ml permmol/l, or �0.4 �U/ml per mg/dl) are athigher risk of diabetes in the future than thosewith a normal insulinogenic index.

4.2. Epidemiological study

The target of epidemiological study of dia-betes is populations rather than individuals. Inpopulation studies, it is usually difficult to ex-amine plasma glucose repeatedly. The reproduci-bility of FPG and OGTT values is not good forindividual subjects, but the mean PG values andtheir distribution curve for each population arefairly reproducible. For these reasons, ‘diabetictype’ hyperglycemia obtained by a single test isto be regarded as ‘diabetes’ for the purpose ofepidemiological study (Table 3). As it is some-times difficult to confirm whether an examinee istested truly at fasting, we recommend to use the2hPG values by OGTT as a basis to estimatethe prevalence of diabetes. In circumstanceswhere OGTT is difficult to perform, FPG crite-ria of �7.0 mmol/l (126 mg/dl) may be usedinstead of 2hPG criteria. The prevalence of dia-betes and individual grouping into glycemic cat-egories differ depending on which criteria areused. Subjects with FPG below 7.0 mmol/l (126mg/dl) often have 2hPG values in the diabeticrange. Thus, the prevalence of ‘diabetes’ byOGTT may be increased by more than 50%compared with that obtained by FPG test alone(Table 7). Therefore, the survey report shoulddescribe what method and criteria are used. Inaddition, it is recommended that the epidemio-logical survey report includes the data of distri-bution of plasma glucose values, as well as theprevalence of ‘diabetes’ and other categories ofglucose intolerance.


4.3. Screening

In health screening for diabetes, the most im-portant point is not to overlook diabetic patientsand subjects with suspect of diabetes. Not onlyglycemic parameters such as plasma glucose,HbA1c and urine glucose, but also information onfamily history, obstetric history, obesity at presentand in the past, hypertension, and other complica-tions should be gathered to select subjects forfurther examination. The diagnosis of subjectsthus screened then follows the same procedure asfor clinical diagnosis.

4.4. Elderly people and children

4.4.1. Elderly peopleThe prevalence of glucose intolerance increases

with age. The criteria for OGTT in the elderly arethe same as in Table 3, but a special considerationshould be paid to treat subjects diagnosed as‘diabetes’. Elderly people (older than 65 years)whose glycemic levels only slightly exceed thecutoff values are to be treated by diet or moderateexercise alone without using medication. In el-derly people, 2hPG tends to be more elevatedcompared with FPG.

4.4.2. ChildrenIn children, type 1 diabetes usually presents

with overt symptoms and with marked hyper-glycemia, leaving little problem in diagnosis. Milddiabetes as detected by urinalysis in school posessome problem in diagnosis. For OGTT, theamount of glucose load is 1.75 g/kg up to amaximum of 75 g. The same glycemic criteria areused as in adults. In childhood diabetes, the dis-tinction of type of diabetes is important. Thosewho develop diabetes under the age of 10 yearsalmost always have type 1 diabetes [14]. Theincidence of type 2 diabetes exceeds that of type 1after 15 years of age. The majority of type 2diabetes in children is associated with obesity.Anti-GAD antibody and other autoantibodies areuseful to identify type 1 diabetes. Detailed familyhistory is helpful not only to detect MODY pa-tients, but also in type 1 and type 2 diabetes.

4.5. Gestational diabetes mellitus (GDM)

The traditional concept of GDM is ‘mild tem-porary glucose intolerance which appears for thefirst time during pregnancy’, implying that glucoseintolerance subsides after delivery [2,24]. Recently,the concept of GDM has undergone modification.The definition by the Fourth International Work-shop on GDM [25] is, ‘carbohydrate intoleranceof varying degrees of severity with onset or firstrecognition during pregnancy. The definition ap-plies regardless of whether insulin is used fortreatment or the condition persists after preg-nancy. It does not exclude the possibility thatunrecognized glucose intolerance may have ante-dated the pregnancy’.

The reason to consider GDM separately fromother types of diabetes is twofold. First, mildglucose intolerance during pregnancy, not meetingthe criteria of ‘diabetes’ in general, may haveharmful effects on mother and infant, needingspecial attention and treatment. Second, such sub-jects are at higher risk to develop diabetes in thefuture, even though glucose intolerance improvesafter delivery.

The definition of the Fourth GDM Workshopincludes the following three different categories.1. Diabetes, which had been present but unno-

ticed before pregnancy, detected incidentallyduring pregnancy.

2. Milder degrees of glucose intolerance whichantedated pregnancy, and progress to ‘diabetictype’ during pregnancy.

3. Glucose intolerance milder than ‘diabetic type’developing during pregnancy for the first time.

The traditional concept of GDM seems closestto this last case [2,24].

Deterioration of glucose tolerance usually de-velops after the second trimester of pregnancy.Therefore, patients who show diabetic type ordiabetic retinopathy during the first trimester arelikely to have had diabetes before pregnancy. It ismore appropriate to regard them as ‘pregnantwomen with diabetes’ rather than as ‘GDM’. Astype 1 diabetes is less frequent in Japanese than inCaucasians, type 2 diabetes is more prevalent thantype 1 diabetes in women of child-bearing age inJapan. Due to the paucity of symptoms of mild


type 2 diabetes, it is common for diabetes to befirst detected during pregnancy. The frequency ofanomalies and perinatal problems are more com-mon among those in whom diabetes or glucoseintolerance antedated pregnancy than in thosewho develop mild glucose intolerance duringpregnancy. The difference between these two situ-ations should be considered during the planningof treatment of ‘GDM’ [26].

In 1984, the Committee for Nutrition andMetabolism of the Japan Society of Gynecologyand Obstetrics proposed cutoff values for thediagnosis of GDM by the use of 75 g OGTT.GDM is diagnosed when any two points amongFPG, 1hPG and 2hPG exceeded 5.5, 10.0 and 8.3mmol/l (100, 180, and 150 mg/dl) respectively(Table 5) [27]. These cutoff values were first se-lected from the mean value +2 S.D. of healthypregnancies, and their validity was also supportedlater by the study of perinatal abnormalities ofpregnant women with varying degree of glucoseintolerance. We have, therefore, decided to adoptthese criteria for the diagnosis of GDM in thepresent report.

As a screening test, a casual plasma glucose testis recommended at the first visit. If it is 5.5mmol/l (100 mg/dl) or higher, then a 75 g OGTTis recommended with evaluation by the criteriashown in Table 5, [26,28]. The frequency of casualPG exceeding 5.5 mmol/l (100 mg/dl) during preg-nancy was reported to be approximately 8% intwo obstetric clinics [29]. The detection of glucoseintolerance during the first trimester suggests thatit was present before pregnancy. Normal glucosetolerance during the first trimester does not ex-clude the possibility that glucose tolerance may

deteriorate after the second trimester. Therefore,screening should be repeated at 24 to 28th weeksof pregnancy.

Patients with GDM should be re-tested 1–3months after delivery to reclassify the glucosetolerance state. The etiological classification isalso re-evaluated if the diabetic state persists.GDM is a group with high risk for future dia-betes. Therefore, periodic follow-up is recom-mended for all cases.

5. Discussion and comments

In preparation of this report, we have paidspecial consideration to the following points.First, we thought it best to adopt the recentreports by ADA and WHO as far as possible.Second, the report should be based on recentclinical and epidemiological data obtained inJapan. Third, we have continued the basic philos-ophy on the diagnosis of diabetes used by the1970 and 1982 JDS Committees. Fourth, we havetaken account of many opinions expressed byCouncil members of JDS. Some discussion andadditional comments not mentioned in the pre-ceding text will be presented below.

5.1. Classification of diabetes mellitus

5.1.1. Diabetes in which specific mutations ha�ebeen identified as a cause of genetic susceptibility

Recently, in a number of families with diabetes,DNA abnormalities have been identified whichexplain genetic susceptibility to specific types ofdiabetes [18]. They include two categories, abnor-malities of insulin supply and abnormalities ofinsulin action. These categories were placedwithin ‘other types of diabetes due to specificcauses’ in the ADA and WHO reports, with pan-creatic diseases, endocrine diseases and so on. Wehave separated the group with known DNA ab-normalities (category A) from other groups of‘secondary diabetes’ (category B).

Pancreatic diseases, endocrinopathies and oth-ers have specific symptoms, and diabetes oftenconstitutes only one of the clinical features. Incontrast, diabetes is a core and sometimes theonly symptom in category A.

Table 5Criteria for gestational diabetes mellitus by 75 g OGTT

Blood glucose mmol/l (mg/dl)

Capillary blood Venous bloodVenous plasma

�5.6 (100) �4.4 (80)FPG �4.4 (80)�8.9 (160)1hPG �10.0 (180) �10.0 (180)

�8.3 (150)2hPG �8.3 (150) �7.8 (140)

Gestational diabetes is diagnosed when two or more pointsexceeded the above cutoff levels. (Sugawa et al. [27]).


At the JDS Committee Meeting, there was dis-cussion that some causes of diabetes in category Awith DNA mutations might better be classified asa subgroup of type 2 diabetes, because many areindistinguishable from common type 2 diabetesclinically, and they have features of insulin secre-tory defect and/or insulin resistance, which arecharacteristic features of type 2 diabetes. Usingsuch a classification, type 2 diabetes would beseparated into two categories, idiopathic type 2diabetes and a group with known DNA abnor-malities. We did not adopt this, however, becausesome DNA abnormalities are also accompaniedspecific physical features different from commontype 2 diabetes, such as deafness with mitochon-drial DNA mutations, and the characteristic fea-tures seen in leprechaunism.

In the future, the cause of genetic susceptibilityof common type 2 diabetes, perhaps polygenic,will probably be gradually clarified. Reconsider-ation of the classification of diabetes will then benecessary, when substantial data on susceptibilitygenes have been accumulated.

5.1.2. Worsening and impro�ing of states ofdiabetes

Each patient with diabetes is expressed in 2-Din Fig. 1, the etiology of diabetes and the degreeof deficiency of insulin effect [9,10]. In Fig. 1,there are arrows directed to right and left. Direc-tion to the right means worsening of themetabolic state, and direction to the left meansimprovement spontaneously or with treatment. Insimilar figures in the ADA and WHO reports[6,7], the arrows are expressed as bi-directionallines. We separate the arrows toward right andleft, and some portions of the arrows are ex-pressed as broken lines to indicate uncommonphenomena. For example, it is not common forestablished diabetes to revert to completely nor-mal glucose tolerance. Nevertheless, such casesare seen in so-called ‘soft-drink ketosis’, in whichpatients presenting with ketoacidosis may be im-proved to normoglycemia [30,31]. The left-di-rected arrow may be regarded as indicating a goalof treatment of diabetes.

Once a patient is diagnosed as diabetes, wethink that he or she should be regarded and

treated as ‘diabetes’ even if the glycemic stateimproves to a borderline type or normoglycemia.This is the reason why the arrows toward the leftare expressed as filled lines throughout theirlength. There are exceptions, however, where thecause of diabetes can be removed and diabetes hasbeen regarded as cured (e.g. the surgical removalof pheochromocytoma etc).

5.2. Diagnostic criteria and rele�ant problems

5.2.1. Rationales for use of the term ‘diabetictype’

The most important purpose of diagnosing dia-betes is to allow early detection and treatment, inorder to prevent complications. The occurrenceand progression of diabetes-specific complicationsare closely related to the degree and duration ofhyperglycemia. The level of hyperglycemia thatcan predispose to diabetic complications is animportant factor in selecting cutoff plasma glu-cose values [8]. Any of three parameters ofglycemia, FPG, 2hPG and HbA1c, can serve topredict the risk of diabetic complications [32].

In order to define ‘diabetic type’ we selected thesame plasma glucose levels as those used to define‘diabetes ‘ in the ADA and WHO reports. Thecutoff level of 2hPG is the same as the previousJDS criteria, but that for FPG is lowered to �7.0mmol/l (126 mg/dl) from �7.8 mmol/l (140 mg/dl). The data in Japan also confirm that a meanFPG value corresponding to 2hPG of 11.1 mmol/l(200 mg/dl) is approximately 6.9 mmol/l (125mg/dl) in subjects younger than 60 years (Table 6)[33].

In the present report, the word ‘type’ is addedto each glucose tolerance category classified byOGTT, similar to earlier JDS recommendations[1,2]. We believe that the classification of OGTTresults is conceptually different from diagnosingdiabetes mellitus. The rationale of the use ofchronic hyperglycemia for diagnosis is that it is amost important feature of diabetes and thestrongest risk factor for specific diabeticcomplications.

We have, therefore, adopted a position thatdiabetes can be diagnosed by confirming persis-tent hyperglycemia according to the criteria of


Table 6Mutual relationships between FPG and 2hPG by 75 g OGTT

FPG mmol: mg/dl � 2hPG mmol/l: mg/dl

6.7: 120 10.4 (7.8–12.9): 187 (140–233)10.9 (8.4–13.6): 197 (151–244)6.9: 12511.6 (9.0–14.2): 208 (162–255)7.2: 13012.2 (9.6–14.8): 219 (173–266)7.5: 135

12.8 (10.2–15.4): 230 (184–277)7.8: 1408.3: 150 14.0 (11.4–16.6): 252 (206–299)

FPG mmol: mg/dl � 2hPG mmol/l: mg/dl

10.0: 1806.6 (5.5–7.6): 118 (99–137)10.6: 1906.7 (5.7–7.8): 121 (102–140)11.1: 2006.9 (5.8–8.0): 125 (105–144)

7.1 (6.1–8.2): 128 (109–147) 11.7: 21012.2: 2207.3 (6.2–8.3): 131 (112–150)12.8: 2307.5 (6.4–8.6): 135 (116–154)13.3: 2407.7 (6.6–8.7): 138 (119–157)13.9: 2507.9 (6.8–8.9): 142 (123–160)

8.1 (7.0–9.1): 145 (126–164) 14.4: 26015.6: 2808.4 (7.4–9.5): 152 (133–171)

Data of 7454 atomic bomb survivors aged 60 years or youngerat the time of OGTT. Figures in the brackets, mean�1 S.D.(Ito et al. [33]).

the chance of misdiagnosis of diabetes on a singleplasma glucose determination would be small (�5%).

Cross-sectional data from epidemiological sur-veys indicate that retinopathy is concentrated inthe upper one to three deciles classified byglycemic parameters [6]. These data show that thefrequency of retinopathy clearly increases whenFPG exceeds 7.8 mmol/l (140 mg/dl) (not 7.0mmol/l (126 mg/dl)), when 2hPG exceeds 12.8–13.3 mmol/l (230–240 mg/dl) (not 11.1 mmol/l(200 mg/dl)), and when HbA1c exceeds 6.5%, re-spectively. A clinical study in Japan indicates thatdiabetic complications seldom occur when HbA1c

is below 6.5% [36]. In view of the definite risk forspecific complications, there was a discussion inthe Committee that the cutoff levels may be betterset as �7.8 mmol/l (140 mg/dl) for FPG and�12.8–13.3 mmol/l (230–240 mg/dl) for 2hPG,namely retaining the old FPG cutoff value whileelevating that for 2hPG. However, we adopted thepresent criteria for two reasons. Firstly, we wereconcerned about inconvenience and confusionwith the use of the different criteria from interna-tional standards. Secondly the above mentioneddata are derived from cross-sectional surveys.With follow-up studies, somewhat lower glycemiclevels may also be turned out to give risk todiabetic complications. In order to prevent com-plications, it would be better to begin treatment ofdiabetes before plasma glucose reaches definitelyrisky levels.

We added casual PG higher than 11.1 mmol/l(200 mg/dl) as indicating ‘diabetic type’. Post-prandial PG �11.1 mmol/l (200 mg/dl) usuallyreflects a more severe degree of glucose intoler-ance than 2hPG �11.1 mmol/l (200 mg/dl) by 75g OGTT. Indeed, patients classified as diabetictype by 75 g OGTT commonly have casual PGvalues below 11.1 mmol/l (200 mg/dl) [37]. Arather high cutoff level was selected for casualPG, to ensure the specificity for abnormallyhigh glycemia, as the amount of meal andtime interval between the meal and blood sam-pling are so variable. Therefore, when the first testhas been a casual PG, we recommend the othermethods of plasma glucose testing for the secondtest.

diabetic type. Repeated measurements of plasmaglucose will be necessary to confirm ‘persistenthyperglycemia’. Nevertheless, it is desirable if thediagnosis of diabetes can be made withoutrepeating the test. We have selected three condi-tions, in the presence of which diabetes can bediagnosed by a single ‘diabetic type’ hyper-glycemia measurement. They are (1) the presenceof typical diabetic symptoms; (2) HbA1c�6.5%;and (3) definite diabetic retinopathy. High HbA1c

indicates that hyperglycemia has continued forsome time.

Variations in plasma glucose can occur due tofluctuations of physical conditions. The standarddeviation of plasma glucose near the cutoff levelsfor diabetic type is about 0.4–0.6 mmol/l (7–11mg/dl) for FPG, and 1.4–2.0 mmol/l (25–35 mg/dl) for 2hPG [34,35]. Taking these variations intoaccount, subjects with FPG �7.8–8.3 mmol/l(140–150 mg/dl), and 2hPG �13.3–13.9 mmol/l(240–250 mg/dl) are likely to remain as ‘diabetictype’ on repeated tests. In such patients withdefinitely higher glycemia than the cutoff levels,


5.2.2. Fasting plasma glucose �ersus 2h plasmaglucose

Diabetic type can be judged either by FPG orby 2hPG. The prevalence of ‘diabetic type’ isusually higher by 2hPG than by FPG in epidemi-ological and screening data in Japan (Table 7)[38]. In some countries FPG criteria �7.0 mmol/l(126 mg/dl) detected a higher prevalence for ‘dia-betes’ than 2hPG criteria �11.1 mmol/l (200mg/dl), or both methods resulted in similar preva-lence of ‘diabetes’ in epidemiological survey [39].

In most studies, however, there are large dis-crepancies in the diagnosis of ‘diabetes’ in individ-ual subjects. Japanese as well as European andAsian data show that ‘diabetes’ classified by FPG�7.0 mmol/l (126 mg/dl) and ‘diabetes’ by 2hPG�11.1 mmol/l (200 mg/dl) agree only partially inmild diabetes (Table 7) [38–41]. In Japan andother Asian countries[38,41], the prevalence of‘diabetes’ identified by 2hPG �11.1 mmol/l is1.5-fold higher than ‘diabetes’ identified by FPG�7.0 mmol/l. From a physiological viewpoint,FPG is primarily governed by glucose outputfrom the liver, while 2hPG is affected by theabsorption rate of glucose from the gut, the uti-lization rate of glucose by muscle and other pe-ripheral tissues and changes in glucose handlingby the liver. It is conceivable that FPG and 2hPGdo not increase in parallel in some patients.

Comparison of the incidence of retinopathybetween groups with similar degree of FPG (ei-ther 6.1–7.0 mmol/l (110–126 mg/dl), or 7.0–7.8mmol/l (126–140 mg/dl)), but with different levels

of 2hPG higher or lower than 11.1 mmol/l (200mg/dl), revealed that groups with higher 2hPGshowed 2–3-fold higher incidence of retinopathythan those with 2hPG lower than 11.1 mmol/l(200 mg/dl) [42].

For these reasons, the measurement of FPGalone is not sufficient and OGTT is necessary todetect diabetes and mild abnormalities of glucosemetabolism.

5.2.3. Cutoff le�els of plasma glucose for ‘normaltype’ and mild glucose intolerance

In the previous JDS reports, ‘normal type’ wasdefined as those who would remain non-diabeticafter follow-up of several years [1,2]. In the 1982report, ‘normal type’ should meet all of the fol-lowing criteria; FPG �6.1 mmol/l (110 mg/dl),1hPG �8.9 mmol/l (160 mg/dl) and 2hPG �6.7mmol/l (120 mg/dl). Subsequent studies haveconfirmed that ‘normal type’ by this definitionseldom deteriorates to ‘diabetic type’ [43,44]. Butas the cutoff levels for ‘normal type’ were setlower than the lower limits of IGT by WHO,claims were raised that too many subjects wereclassified as ‘borderline type’ at the health checkexamination using OGTT [45].

We elevated the cutoff level of 2hPG for ‘nor-mal type’ from �6.7 mmol/l (120 mg/dl) to �7.8mmol/l (140 mg/dl). By this change, the progres-sion rate to ‘diabetic type’ rises a little, but re-mains less than 1.0% per year (Table 8). We thinkthat the merit of conforming with internationalcriteria will outweigh the demerit of a small in-

Table 7Comparison of frequencies of ‘diabetic type’ by FPG and 2hPG criteria

Caucasian*FPG Japanese**2hPG Asian*n=25 219n=5384mmol/l (mg/dl) n=16 739mmol/l (mg/dl)

(a) �7.0 (126) �11.1 (200) 298 (5.5%)** 473 (1.9%)* 546 (3.3%)*(b) �7.0 (126) �11.1 (200) 75 (1.4%)** 613 (2.4%)* 220 (1.3%)*

�11.1 (200) 346 (6.4%)**(c) �7.0 (126) 431 (1.7%)* 449 (2.7%)*719 (13.4%)**(a)+(b)+(c) Sum (diabetic type) 1215 (7.3%)*1517 (6.0%)*

�7.0 mmol/l (126 mg/dl)(b)+(c) FPG 7.8%** 4.2%* 4.0%*�11.1 mmol/l (200 mg/dl) 12.0%** 3.6%* 5.9%*(a)+(c) 2hPG

(Sasaki, A. [35]) (DECODE Study [39]) (DECODA Study [41])

Data of Japanese were taken from results of health check examinees. *, Excludes known diabetes; **, includes known diabetes.


Table 8Deterioration rate to ‘diabetic type’ of health check examinees by the baseline results of 75 g OGTT

(B) Deterioration rate2hPG (A) Deterioration rate1hPGFPGmmol/l (mg/dl)mmol/l (mg/dl) Number of cases (%)/5.1 yearsNumber of cases (%)/4 years

Normal type�6.7 (120) 7/536 (1.3)�6.1 (110) 14/1088 (1.3)–�7.8 (140) 25/778 (3.2)– 27/1441 (1.9)�6.1 (110)

�6.1 (110) �8.9 (160) �7.8 (140) 4/501 (0.8) 5/1137 (0.4)�7.8 (140) 2/125 (1.6)8.9–9.9 (160–179) 6/146 (4.1)�6.1 (110)

10.0–11.0 (180–199)�6.1 (110) �7.8 (140) 8/74 (10.8) 8/92 (8.7)�7.8 (140) 5/51 (9.8)�6.1 (110) 2/40 (5.0)11.1–12.1 (200–219)�7.8 (140) 6/27 (22.2)�12.2 (220) 7/27 (25.9)�6.1 (110)

�6.1 (110) �10.0 (180) �7.8 (140) 19/152 (12.5) 17/159 (10.7)�7.8 (140)�6.1 (110) 11/78 (14.1)�11.1 (200) 9/67 (13.4)

Borderline type* 98/424 (23.1) 64/281 (22.8)–�7.0 (126)** 7.8–11.0 87/333 (25.8) 47/217 (21.7)

7.8–11.0 (140–199)– �7.8 (140) 11/91 (12.1) 17/64 (26.6)6.1–6.9***

(110–125)

Mean follow-up periods, A, 4 years, B, 5.1 years. *, Borderline type, as defined by Table 3, corresponds to the sum of IFG and IGTby WHO, when OGTT is performed; **, corresponding to IGT by ADA and WHO. ***, corresponding to IFG by WHO whenOGTT is performed (A: Sasaki, A., 1998; B: Hara, H. 1999, unpublished data).

crease in the rate of progression to ‘diabetictype’.

The ADA report recommended the use of FPGalone to diagnose diabetes, and defined impairedfasting glucose (IFG) for those with FPG between6.1 and 7.0 mmol/l (110 and 126 mg/dl). Thisrange of FPG corresponds to the intermediaterange between normal and the diabetic ranges inthe present report. Subjects with IFG could beclassified as ‘diabetic type’ or ‘IGT’ if they aretested by OGTT. According to WHO report,when tested by OGTT, the use of term impairedfasting glycemia (IFG) is restricted to those with2hPG less than 7.8 mmol/l (140 mg/dl). BothADA and WHO reports defined IGT as thosewith FPG less than 7.0 mmol/l (126 mg/dl) and2hPG between 7.8 and 11.1 mmol/l (140 and 200mg/dl). The borderline type in the present reportexactly corresponds to the sum of IGT and IFGin the narrow sense by WHO. On an individualbasis, IFG and IGT are highly discordant [38,46].We combined IFG and IGT together as ‘border-

line type’, because we think there is little need tohandle IFG and IGT separately.

We eliminated the cutoff values of 1hPG todefine ‘normal type’, which was set as �8.9mmol/l (160 mg/dl) in the 1982 JDS report. Datashowed, however, that the progression to ‘diabetictype‘ from ‘normal type’ increased with the eleva-tion of 1hPG, even with normal FPG and 2hPG.When 1hPG exceeded 12.2 mmol/l (220 mg/dl),the progression rate was equal to that of IGT(Table 8). Therefore, we recommend that 1hPGbe measured during an OGTT whenever possible,and that subjects with high 1hPG be followedeven though they belong to ‘normal type’ by FPGand 2hPG criteria. One study from USA indicatesthat a high 30–90 min plasma glucose is associ-ated with increased cardiovascular risk factors,even in groups with normal FPG and 2hPG [47].

The low insulin secretory response in the earlyphase after glucose challenge is a characteristicfeature of diabetes mellitus. Among subjects withborderline type, those with a low insulin response


are at higher risk of developing diabetes thanthose with a normal insulin response [48]. A ratioof increment of insulin to that of PG at 30 minafter glucose load lower than 7.2 �U/ml permmol/l (0.4 �U/ml per mg/dl) is a useful parame-ter to predict the future risk of development ofdiabetes. A high fasting insulin level, suggestingthe presence of insulin resistance, is also a riskfactor for type 2 diabetes.

5.2.4. HbA1c

HbA1c is an important and useful parameter forthe treatment of diabetes, but there are severalproblems relating to its use for diagnosis. It variesnot only according to the level of glycemia butalso to the turnover rate of hemoglobin. There isa large overlap of distribution between groupswith ‘normal type’ and ‘borderline type’, makingit difficult to distinguish between these two groups(Fig. 2) [37].

Various methods are now being used to mea-sure HbA1c, yielding somewhat different values ofglycohemoglobins. High performance liquid chro-matography, a standard method of assay, maynot be sufficiently specific for HbA1c. Inter-labo-ratory variation and day-to-day assay errors arestill unsatisfactory. No chemical standard forHbA1c is yet available.

In Japan, thanks to the effort of the Committeefor Standardization of Glycohemoglobin, inter-laboratory variation has been decreased, by mea-

suring only the stable fraction of HbA1c and byinclusion of standard samples in each run toadjust assay values [49,50].

The use of HbA1c should be limited as a supple-ment to plasma glucose tests for the diagnosis ofdiabetes. If HbA1c is definitely high in a hyper-glycemic patient, diabetes can be diagnosed with-out repeating PG. Namely, when FPG, 2hPG orcasual PG is in the diabetic range and HbA1c

exceeds 6.5%, the patient is judged to have dia-betes. To select this value, we put emphasis onspecificity for the diagnosis of diabetes, whileignoring sensitivity. As seen from Fig. 2, subjectswith HbA1c �6.5% are unlikely to be classified as‘normal type’ or ‘borderline type’ by OGTT.However, HbA1c values less than 6.5% should notbe taken as evidence against the diagnosis ofdiabetes, because a significant portion of subjectsof ‘diabetic type’ have HbA1c lower than 6.5%.

In subjects younger than 60 years, the meanHbA1c value, which corresponds to FPG of 7.0mmol/l (126 mg/dl) and 2hPG of 11.1 mmol/l(200 mg/dl), is about 6.1% using the assay methodrecommended by the JDS Committee for Stan-dardization of Glycohemoglobin [33]. This valuecould be used to estimate the prevalence of dia-betes in an epidemiological survey.

The use of HbA1c assay in combination withplasma glucose tests is expected to improve theefficiency of screening for glucose intolerance [51–54]. In Japan, this has already been introducedinto the health check program for the elderlypeople [51]. There is a need for further study toselect the most appropriate criteria for thispurpose.

6. Conclusions: Comparison of this report withthose of American Diabetes Association [6] andWHO [7]

The present report resembles those of ADA andWHO but differs in the following points.

6.1. Classification

1. Diabetes and related disorders of glucose tol-erance, due to specific mechanisms and dis-eases, are divided into two subgroups; 3A, in

Fig. 2. Distribution of HbA1c in groups with varying degree ofglucose intolerance (number of cases in each group of glucosetolerance: Normal type, n=6720; borderline type, n=6296;and diabetic type, n=5040). Diabetic type in which bothFPG�7.0 mmol/l (126 mg/dl) and 2hPG�11.1 mmol/l (200mg/dl), n=2950; (Ito, C. unpublished data).


which genetic susceptibility is clarified at theDNA level, and 3B, which is associated withother diseases or conditions.

2. Liver diseases are added into category 3B.3. The figure showing the relationship of etiology

and stage of diabetes is expressed in a moresophisticated way.

6.2. Diagnosis

1. Cutoff plasma glucose levels for each categoryare the same, but a term ‘type’ is added toeach category.

2. Borderline type is defined as neither diabeticnor normal type, including both IFG and IGTby WHO when OGTT is performed.

3. Diabetes is generally diagnosed when ‘diabetictype’ hyperglycemia is confirmed on two ormore occasions.

4. Diabetes can be diagnosed on a single ‘diabetictype’ hyperglycemia if the patient has typicalsymptoms of diabetes, HbA1c higher than6.5%, or diabetic retinopathy.

5. OGTT is recommended for those with mildhyperglycemia, because FPG criteria alonewould overlook many subjects with diabetictype in Japan. High 1hPG with normal FPGand 2hPG is also a risk factor for futurediabetes.

6. New criteria for gestational diabetes areproposed.

References

[1] N. Kuzuya, M. Abe, H. Ueda, K. Kuzuya, T. Kuzuya, K.Kosaka, Y. Goto, Y. Shigeta, S. Baba, Y. Hirata, A.Horiuchi, K. Yamada, M. Wada, Report of the commit-tee on the diagnostic criteria of the oral glucose tolerancetest for diabetes mellitus, J. Jpn. Diabetes Soc. 13 (1970)1–7 in Japanese.

[2] K. Kosaka, Y. Akanuma, Y. Goto, R. Hagura, Y. Hi-rata, R. Kawate, T. Kuzuya, G. Mimura, H. Nakayama,N. Sakamoto, Y. Shigeta, Report of the committee on thediagnosis of diabetes mellitus, J. Jpn. Diabetes Soc. 25(1982) 859–866 in Japanese; The essence of this reportappears in English; K. Kosaka, Diab. Res. Clin. Pract. 24(Suppl.) (1994) S59–S62.

[3] National Diabetes Data Group, Classification and diag-nosis of diabetes mellitus and other categories of glucoseintolerance, Diabetes 28 (1979) 1039–1057.

[4] WHO Expert Committee on Diabetes Mellitus, SecondReport. World Health Organization, WHO, Geneva,Technical Report Series 646, 1980.

[5] WHO Study Group, Diabetes mellitus. World HealthOrganization, WHO, Geneva, Technical Report Series727, 1985.

[6] The Expert Committee on the Diagnosis and Classifica-tion of Diabetes Mellitus, Report of the Expert Commit-tee on the Diagnosis and Classification of DiabetesMellitus. Diabetes Care 20 (1997) 1183–1197.

[7] Report of a WHO Consultation, Definition, diagnosisand classification of diabetes mellitus and its complica-tions. Part 1 Diagnosis and classification of diabetesmellitus. World Health Organization, Department ofNoncommunicable Disease Surveillance, Geneva, 1999.

[8] T. Kuzuya, S. Nakagawa, J. Satoh, Y. Kanazawa, Y.Iwamoto, M. Kobayashi, K. Nanjo, A. Sasaki, Y. Seino,C. Ito, K. Shima, K. Nonaka, T. Kadowaki, Answers byCouncil members of JDS to a questionnaire on the diag-nostic criteria and classification of diabetes mellitus, J.Jpn. Diabetes Soc. 40 (1997) 203–210 in Japanese.

[9] S. Tarui, Comparison of diabetes mellitus and glycogeno-sis—conceptual and pathophysiological aspects, J. JpnDiabetes Soc. 26 (Suppl. 2) (1983) 1119–1212 inJapanese.

[10] T. Kuzuya, A. Matsuda, Classification of diabetes on thebasis of etiologies versus degree of insulin deficiency,Diabetes Care 20 (1997) 219–220.

[11] E. Kawasaki, S. Nagataki, Japanese type 1 diabetic syn-drome, Diabetes Metab. Rev. 12 (1996) 175–194.

[12] T. Kobayashi, Subtype of insulin-dependent diabetes mel-litus (IDDM) in Japan: Slowly progressive IDDM— theclinical characteristics and pathogenesis of the syndrome,Diabetes Res. Clin. Pract. 24 (Suppl.) (1994) S95–S99.

[13] A. Imagawa, T. Hanafusa, J. Miyagawa, Y. Matsuzawa,A novel subtype of type 1 diabetes melllitus characterizedby a rapid onset and an absence of diabetes-relatedantibodies, New Engl. J. Med. 342 (2000) 301–307.

[14] T. Kitagawa, M. Owada, T. Urakami, N. Tajima, Epi-demiology of type 1 (insulin-dependent) and type 2 (non-insulin-dependent) diabetes mellitus in Japanese children,Diabetes Res. Clin. Pract. 24 (Suppl.) (1994) S7–S13.

[15] T. Kitagawa, M. Owada, T. Urakami, K. Yamauchi,Increased incidence of non-insulin-dependent diabetesmellitus among Japanese schoolchildren correlates withan increased intake of animal protein and fat, Clin.Pediatr. 37 (1998) 111–116.

[16] S.S. Fajans, Scope and heterogeneous nature of MODY,Diabetes Care 13 (1990) 49–64.

[17] P. Froguel, M. Vaxillaire, G. Velho, Genetic andmetabolic heterogeneity of maturity-onset diabetes of theyoung, Diabetes Rev. 5 (1997) 123–130.

[18] K. Nanjo, Y. Oka, T. Kadowaki, A. Kanatsuka, M.Kobayashi, T. Sanke, S. Suzuki, Y. Seino, Recent aspectson diabetes mellitus associated with gene mutation inJapan, J. Jpn. Diabetes Soc. 41 (Suppl. 2) (1998) A29–A31 in Japanese.


[19] J.A. Maassen, T. Kadowaki, Maternally inherited dia-betes and deafness (MIDD), a new diabetes subtype,Diabeteologia 39 (1996) 375–382.

[20] S. Sakagashira, T. Sanke, T. Hanabusa, H. Shimomura,S. Ohagi, K.Y. Kumagae, K. Nakajima, K. Nanjo, Mis-sense mutation of amylin gene (S20G) in JapaneseNIDDM patients, Diabetes 45 (1996) 1279–1281.

[21] M. Kasuga, T. Kadowaki, Insulin receptor abnormalitiesin Japan, Diabetes Res. Clin. Pract. 24 (Suppl.) (1994)S145–S151.

[22] K.W. West, Epidemiology of Diabetes and its VascularLesions, Elsevier, New York, 1978, pp. 336–338.

[23] Y. Aoki, K. Yazaki, K. Kiyosawa, Glucose intolerance inliver disease. Nippon Rinsho No. 728, Diabetes Mellitus 3(1998) 594–599 in Japanese.

[24] J. Pedersen, The Pregnant Diabetic and Her Newborn,second ed., Munksgaard, Copenhagen, 1977, p. 46.

[25] B.E. Metzger, D.R. Coustan, The organizing committeesummary and recommendations of the fourth interna-tional workshop conference on gestational diabetes melli-tus, Diabetes Care 21 (Suppl. 2) (1998) B161–B167.

[26] Y. Omori, T. Hamada, N. Toyoda, S. Minei, M. Sanaka,A proposal on diagnostic criteria for GDM, J. Jpn.Diabetes Soc. 41 (Suppl. 2) (1998) A91–A96 in Japanese.

[27] T. Sugawa, S. Takagi, Report of the Committee forNutrition and Metabolism, the Japan Society of Gynecol-ogy and Obstetrics. A proposal of a guideline for thediagnosis of glucose intolerance particularly of gestationaldiabetes mellitus, J. Jpn. Soc. Gyn. Obst. 36 (1984) 2055–2058.

[28] N. Toyoda, M. Nagata, A guideline for the treatment ofpregnancy complicated with endocrine and metabolic dis-eases. A. Diabetes and gestational diabetes mellitus. In:1997 Report of Study Group for the Prevention of Mor-tality of Pregnant Women (chief investigator: Takeda,Y.), Ministry of Health and Welfare, 1998, pp. 77–83, inJapanese.

[29] Y. Omori, Screening for gestational diabetes mellitus andmeasures to prevent malformations of infants of preg-nancy in diabetes. 1993 report of the study group ondiabetes mellitus. Ministry of Health and Welfare, 1993,pp. 285–289, in Japanese.

[30] K. Yamada, K. Nonaka, Diabetic ketoacidosis in youngobese Japanese men, Diabetes Care 19 (1996) 671.

[31] K. Nonaka, K. Yamada, H. Hayashi, K. Koyama, S.Nakayama, N. Takano-Gyotoku, H. Ishizu, Y. Naka-mura, K. Nunoi, Soft-drink ketosis—a disease entitycharacteristically seen in obese young men in Japan.Diabetes mellitus, in: T. Kaneko, K.H. Huh, M. Shichiri,Y. Oka (Eds.), New Aspects of Pathogenesis and Treat-ment, Nishimura/Smith Gorden, London, 1997, pp. 115–116.

[32] D.R. McCance, R.L. Hanson, M.-A. Charles, L.T.H.Jacobsson, D.J. Pettitt, P.H. Bennett, W.C. Knowler,Comparison of tests for glycated hemoglobin and fastingand 2 h plasma glucose concentrations as diagnosticmethods for diabetes, Br. Med. J. 308 (1994) 1323–1328.

[33] C. Ito, R. Maeda, S. Ishida, H. Sasaki, H. Harada,Correlation among fasting plasma glucose, 2 h plasmaglucose levels in OGTT and HbA1c, Diabetes Res. Clin.Pract. 50 (2000) 225–230.

[34] K. Kosaka, Y. Mizuno, T. Kuzuya, Reproducibility ofthe oral glucose tolerance test and the rice-meal test inmild diabetics, Diabetes 15 (1966) 601–604.

[35] J.M. Mooy, P.A. Grootenhuis, H. DeVries, P.J.Kostense, C. Popp-Snijders, L.M. Bouter, R.J. Heine,Intra-individual variation of glucose, specific insulin andproinsulin concentrations measured by two oral glucosetolerance tests in a general Caucasian population: theHoorn Study, Diabetologia 39 (1996) 298–305.

[36] Y. Ohkubo, H. Kishikawa, E. Araki, T. Miyata, S. Isami,S. Motoyoshi, Y. Kojima, N. Furukoshi, M. Shichiri,Intensive insulin therapy prevents the progression of dia-betic microvascular complications in Japanese patientswith non-insulin-dependent diabetes mellitus: a random-ized prospective 6-year study, Diabetes Res. Clin. Pract.28 (1995) 103–117.

[37] K. Kosaka, Various parameters used for the diagnosis ofdiabetes and for the epidemiological investigation— theircharacteristics, their mutual relationship and their appli-cation, J. Jpn. Diabetes Soc. 41 (Suppl. 2) (1998) A101–A105 in Japanese.

[38] A. Sasaki, T. Shimizu, K. Hasegawa, Study of diagnosticcriteria for diabetes melliltus from a viewpoint of clinicalepidemiology, in: K. Kosaka (Ed.), Diabetology 99, Shin-dan to Chiryo Sha, Tokyo, 1999, pp. 97–105 in Japanese.

[39] DECODE Study Group on behalf of the European Dia-betes Epidemiology Study Group, Will new diagnosticcriteria for diabetes mellitus change phenotype of patientswith diabetes? Reanalysis of European epidemiologicaldata, Br. Med. J. 317 (1998) 371–375.

[40] F. DeVegt, J.M. Dekker, C.D.A. Stenohouwer, G. Ni-jpels, L.M. Bouter, R.J. Heine, The 1997 American Dia-betes Association criteria versus the 1995 World HealthOrganization criteria for the diagnosis of abnormal glu-cose tolerance, Diabetes Care 21 (1998) 1686–1690.

[41] The DECODA Study Group on behalf of the Interna-tional Diabetes Epidemiology Group, Q. Qiao, T. Nak-agami, J. Tuomilehto, K. Bork-Johnsen, B. Balkau, Y.Iwamoto, N. Tajima, Comparison of the fasting and 2-hglucose criteria for diabetes in different Asian cohorts,Diabetologia 43 (2000) 1470–1475.

[42] C. Ito, R. Maeda, S. Ishida, H. Harada, N. Inoue, H.Sasaki, Importance of OGTT for diagnosing diabetesmellitus based on prevalence and incidence of retinopa-thy, Diabetes Res. Clin. Pract. 49 (2000) 181–186.

[43] A. Sasaki, T. Shimizu, K. Hasegawa, Assessment of newcriteria for the diagnosis and classification of diabetesproposed by the American Diabetes Association, J. Jpn.Diabetes Soc. 41 (Suppl. 2) (1998) A37–A39 in Japanese.

[44] H. Hara, Heterogeneity of type 2 diabetes mellitus inJapanese— insulin resistance and insulin secretory defect,J. Jpn. Diabetes Soc. 41 (Suppl. 2) (1998) A21–A25 inJapanese.


[45] Y. Goto, New understanding of the classification anddiagnostic criteria of diabetes mellitus, Endocrinol. Dia-betol. 4 (1997) 19–26 in Japanese.

[46] H. Larsson, G. Berglund, F. Lindgarde, B. Ahren, Com-parison of ADA and WHO criteria for diagnosis ofdiabetes and glucose intolerance, Diabetologia 41 (1998)1124–1125.

[47] M. Modan, M.I. Harris, H. Halkin, Evaluation of WHOand NDDG criteria for impaired glucose tolerance. Re-sults from two national samples, Diabetes 38 (1989)1630–1635.

[48] K. Kosaka, T. Kuzuya, H. Yoshinaga, R. Hagura, Aprospective study of health check examinees for the devel-opment of non-insulin-dependent diabetes mellitus: rela-tionship of the incidence of diabetes with the initialinsulinogenic index and degree of obesity, Diabetes Med.13 (Suppl.) (1996) S120–S126.

[49] K. Shima, J. Endo, M. Oimomi, I. Oshima, Y. Omori, Y.Katayama, T. Kawai, R. Kawamori, T. Kanno, H.Kiyose, K. Nakashima, Y. Nagamine, S. Baba, T.Hoshino, Interlaboratory difference in HbA1c measure-ment in Japan— the interim report of the committee onan interlaboratory standardization of HbA1c determina-tion. The Japan Diabetes Society, J. Jpn. Diabetes Soc. 37(1994) 233–243 in Japanese.

[50] K. Shima, J. Endo, M. Oimomi, I. Oshima, Y. Omori, Y.Katayama, Y. Kanazawa, T. Kawai, R. Kawamori, T.

Kanno, H. Kiyose, M. Kuwajima, K. Nakashima, Y.Nagamine, S. Baba, T. Hoshino, Interlaboratory differ-ence in GHb measurement in Japan— the fourth report ofthe GHb Standardization Committee, the Japan DiabetesSociety, J. Jpn. Diabetes Soc. 40 (1997) 321–326 inJapanese.

[51] Division of Health for the Elderly, Health and WelfareBureau for the Elderly, Ministry of Health and Welfare(Ed.) The manual of checkup for diabetes mellitus accord-ing to the Health Act for the Elderly. Nihon Iji ShinpoSha, Tokyo, 1996, in Japanese.

[52] K. Ishida, A. Okanishi, S. Takashina, A study of screen-ing test of diabetes based on correlation between FPG,HbA1c and 2hPG at 75g OGTT, J. Jpn. Diabetes Soc. 41(Suppl. 2) (1998) A65–A67 in Japanese.

[53] Y. Tanaka, Y. Atsumi, K. Matsuoka, A. Mokubo, T.Asahina, K. Hosokawa, S. Shimada, H. Matsunaga, M.Takagi, O. Ogawa, T. Onuma, R. Kawamori, Usefulnessof stable HbA1c for supportive marker to diagnose dia-betes mellitus in Japanese subjects, Diabetes Res. Clin.Pract. 53 (2001) 41–45.

[54] G.T.C. Ko, J.C.N. Chan, V.T.F. Yeung, C.-C. Chow,L.W.W. Tsang, J.K.Y. Li, W.-Y. So, H.P.S. Wai, C.S.Cockram, Combined use of a fasting plasma glucoseconcentration and HbA1c or fructosamine predicts thelikelihood of having diabetes in high-risk subjects, Dia-betes Care 21 (1998) 1221–1225.

diagnostic criteria for diabetes mellitus

Documents