Definition and Diagnosis 0diabetes

8/9/2019 Definition and Diagnosis 0diabetes

1/50


2/50

definition and diagnosisof diabetes mellitus

and intermediate

hyperglycemia

RepoRt o a WHo/ID ConsultatIon


3/50

WHo librry Cgig-i-pbici D

Denition and diagnosis o diabetes mellits and intermediate hyperglycemia :

report o a WHO/IDF consltation.1.Diabetes mellits diagnosis. 2.Diabetes mellits - classication. 3.Hyperg-

lycemia. 4.Glcose tolerance test. I.World Health Organization. II.International

Diabetes Federation.

ISBN 92 4 159493 4 (NLM classication: WK 810)

ISBN 978 92 4 159493 6

Wrd Hh orgizi 2006

All rights reserved. Pblications o the World Health Organization can be obtained

rom WHO Press, World Health Organization, 20 Avene Appia, 1211 Geneva 27,

Switzerland (tel.: +41 22 791 3264; ax: +41 22 791 4857; e-mail: [email protected]). Reqests or permission to reprodce or translate WHO pblications

whether or sale or or noncommercial distribtion shold be addressed to

WHO Press, at the above address (ax: +41 22 791 4806; e-mail: permissions@

who.int).

The designations employed and the presentation o the material in this pblication

do not imply the expression o any opinion whatsoever on the part o the World

Health Organization concerning the legal stats o any contry, territory, city or

area or o its athorities, or concerning the delimitation o its rontiers or bonda-

ries. Dotted lines on maps represent approximate border lines or which there may

not yet be ll agreement.

The mention o specic companies or o certain manactrers prodcts does notimply that they are endorsed or recommended by the World Health Organization

in preerence to others o a similar natre that are not mentioned. Errors and

omissions excepted, the names o proprietary prodcts are distingished by initial

capital letters.

All reasonable precations have been taken by the World Health Organization

to veriy the inormation contained in this pblication. However, the pblished

material is being distribted withot warranty o any kind, either expressed or

implied. The responsibility or the interpretat ion and se o the material lies with

the reader. In no event shall the World Health Organization be liable or damages

arising rom its se.

Printed by the WHO Docment Prodction Services, Geneva, Switzerland


4/50


5/50

contents

Smmary o technical report and recommendations 1

Introdction 5

Backgrond 7

ISSuE 1:

Shold the crrent diagnostic criteria or diabetes be changed? 9

ISSuE 2:

How shold normal plasma glcose levels be dened? 13

ISSuE 3:

How shold impaired glcose tolerance be dened? 17

ISSuE 4:

How shold impaired asting glcose be dened? 21

ISSuE 5:

What diagnostic tests shold be sed to dene glycaemic stats? 29

Appendices

Reerences 41


6/50


7/50

summary oftechnical report

and recommendations

since 1965 he Wrld Helh orgnizin (WHo) h publihed guide-

li r h digi d clifci dib. th wrl rviwd i 1998 d wr pblihd h gidli r hDefnition, Diagnosis and Classifcation o Diabetes Mellitus and its

Complications. since hen mre inrmin relevn he digni dib h bcm vib. I nvmbr 2005 ji WHo dInerninl Dibee ederin (ID) technicl adviry Grup me

i Gv rviw d d h crr WHo gidi.

Ater consideration o available data and recent recommendations made

by other organisations, the Grop made the ollowing recommendations:

Rcmmdi 1The crrent WHO diagnostic criteria or diabetes shold be maintained asting plasma glcose 7.0mmol/l (126mg/dl) or 2h plasma glcose

11.1mmol/l (200mg/dl).

Despite the limitations with the data rom which the diagnostic criteria or

diabetes are derived, the crrent criteria distingish a grop with signi-

cantly increased prematre mortality and increased risk o microvasclar

and cardiovasclar complications.

Rcmmdi 2

Since there are inscient data to accrately dene normal glcose lev-els, the term normoglycaemia shold be sed or glcose levels associ-

ated with low risk o developing diabetes or cardiovasclar disease, that

is levels below those sed to dene intermediate hyperglycaemia.

Rcmmdi 3

The crrent WHO denition or Impaired Glcose Tolerance (IGT) sholdbe maintained or the present.

Consideration shold be given to replacing this category o intermediate

hyperglycaemia by an overall risk assessment or diabetes, cardiovasclar

disease, or both, which incldes a measre o glcose as a continosvariable.


8/50

Rcmmdi 4

The asting plasma glcose ct-point or Impaired Fasting Glcose (IFG)

shold remain at 6.1mmol/l.

This decision was based on concerns abot the signicant increase in IFG

prevalence which wold occr with lowering the ct-point and the impacton individals and health systems. There is a lack o evidence o anybenet in terms o redcing adverse otcomes or progression to diabetes

and people identied by a lower ct-point eg 5.6mmol/l (100mg/dl) have

a more avorable cardiovasclar risk prole and only hal the risk odeveloping diabetes compared with those above the crrent WHO ct-point. Lowering the ct-point wold increase the proportion o people

with IGT who also have IFG bt decreases the proportion o people withIFG who also have IGT.

Consideration shold be given to replacing this category o intermediate

hyperglycaemia by an overall risk assessment or diabetes, cardiovasclar

disease, or both, which incldes a measre o glcose as a continos

variable.

Rcmmdi 5

1. Venos plasma glcose shold be the standard method or measring

and reporting glcose concentrations in blood. However in recognition

o the widespread se o capillary sampling, especially in nder-re-sorced contries, conversion vales or capillary plasma glcose are

provided or post-load glcose vales. Fasting vales or venos andcapillary plasma glcose are identical.

2. Glcose shold be measred immediately ater collection by near-pa-

tient testing, or i a blood sample is collected, plasma shold be im-mediately separated, or the sample shold be collected into a container

with glycolytic inhibitors and placed in ice-water ntil separated prior

to analysis.

Rcmmdi 6

The oral glcose tolerance test (OGTT) shold be retained as a diagnostic

test or the ollowing reasons:

asting plasma glcose alone ails to diagnose approximately 30% ocases o previosly ndiagnosed diabetes,

an OGTT is the only means o identiying people with IGT,

an OGTT is reqently needed to conrm or exclde an abnormalityo glcose tolerance in asymptomatic people.

An OGTT shold be sed in individals with asting plasma glcose 6.1

6.9mmol/l (110125mg/dl) to determine glcose tolerance stats.

definition and diagnosis of diabetes mellitus and intermediate hyperglycemia


9/50

Rcmmdi 7

Crrently HbA1c is not considered a sitable diagnostic test or diabetes

or intermediate hyperglycaemia.

The ollowing Table smmarises the 2006 WHO recommendations or the

diagnostic criteria or diabetes and intermediate hyperglycaemia.

Dib

Fasting plasma glcose

2h plasma glcose*

7.0mmol/l (126mg/dl)

r11.1mmol/l (200mg/dl)

Imird Gc trc (IGt)

Fasting plasma glcose

2h plasma glcose*


10/50


11/50

introduction

Recen eime indice here were 171 millin peple in he wrld

wih dib i h yr 2000 d hi i prjcd icr 366 milli by 20301. Dib i cdii primrily dfd byh lvl hyprglycmi givig ri rik micrvclrdmg (rihy, hrhy d rhy). I i cidwih rdcd li xpccy, igifc mrbidiy d pcifcdib rld micrvclr cmplici, icrd rik mcrvclr cmplici (ichmic hr di, rk dpriphrl vclr di), d dimiihd qliy li. thamric Dib acii (aDa) imd h il c dib i h usa r 2002 b $us132 billi, icrig $us192 bii i 20202.

Since 1965 the World Health Organization (WHO) has pblished gide-lines or the diagnosis and classication o diabetes. These were lastreviewed in 1998 and were pblished as the gidelines or the Deni-tion, Diagnosis and Classication o Diabetes Mellits3. Since then more

inormation relevant to the diagnosis o diabetes has become available.In addition, in 2003, the ADA reviewed its diagnostic criteria4. While thecriteria or the diagnosis o diabetes and Impaired Glcose Tolerance (IGT)

remained nchanged, the ADA recommended lowering the threshold orImpaired Fasting Glcose (IFG) rom 6.1mmol/l (110mg/dl) to 5.6mmol/l(100mg/dl)4. In view o these developments WHO and the InternationalDiabetes Federation (IDF) decided that it was timely to review its existing

gidelines or the denition and diagnosis o diabetes and intermediate

hyperglycaemia.

Accordingly, WHO appointed an internal Gideline Steering Grop and

convened a Technical Gideline Development Grop. Membership othese committees is shown in Appendix 1. A meeting o the Technical


12/50

Gideline Development Grop was held at WHO headqarters in Geneva

on November 46, 2005. The specic aims o the meeting were to re-view relevant data which addressed and cold inorm the review o thediagnostic criteria or:

DiabetesImpaired glcose tolerance (IGT )

Impaired asting glcose (IFG) (in particlar the 2003 ADA revised

criteria).

The denition o the Metabolic Syndrome was otside the brie o thisreview and was thereore not considered.

In addressing these aims and in ormlating recommendations, the Tech-

nical Gideline Development Grop took into accont the ollowing WHOgiding vales or gideline development:

A poplation perspective, not primarily an individal perspectiveScientic integrity with evidence on ecacy

Feasibility

Cost-eectiveness and opportnity costs

Sensitivity to local contexts

Transparency

A primary adience o health policy makers



13/50

background

WHo h bihd r rvi tchic Rr rig di-

b i 19655, 19806, 19857 d 19993. ovr hi prid hr hvbeen ignifcn chnge in he dignic crieri nd clifcin

dibee nd inermedie hyperglycemi. the dignic crieri

d vr hi rid r mmrid i adix 2.

The crrent diagnostic criteria sed or the diagnosis o diabetes andintermediate hyperglycaemia have been in place globally or almost a

decade and are widely accepted. However, in 2003 the ADA modied its

recommendations reslting in discrepancies between its recommenda-

tions and those o the WHO.

Althogh attention has ocssed on the di erence in asting plasma gl-

cose levels or dening IFG, there are a nmber o important dierences

between the ADA and WHO recommendations which may reslt in dier-

ences in an individals classication o glcose tolerance (see Appendix

3). These inclde:

asting plasma glcose vale sed to dene IFG

inclsion o 2h plasma glcose vale in dening IFG

reqirement or asting plasma glcose level in dening IGT

asting plasma glcose as the recommended method or diagnosingasymptomatic diabetes by ADA whereas WHO recommends the oral gl-

cose tolerance test.

These discrepancies have implications or the individal and or pop-lation prevalence estimates. For example people who all into the ADAcategory o IFG cold inclde people with IGT or diabetes i a 2h plasma

glcose is not measred, and ADA dened IGT cold inclde diabetes i a

asting plasma glcose is not measred.

The objective o the Technical Gideline Development Grop meeting was

to examine the evidence on the ollowing isses related to the diagnosis

o diabetes and intermediate hyperglycaemia:

Crr digiccririCrr digiccriri


14/50

Shold the crrent diagnostic criteria or diabetes be changed?

How shold normal glcose levels be dened?

How shold impaired glcose tolerance be dened?

How shold impaired asting glcose be dened?

What diagnostic tests shold be sed to dene glcose tolerance

stats?

The Technical Gideline Development Grop accepted the 1999 WHOclassication and aetiological ramework or diabetes (Appendix 4).



15/50

issue

1should the currentdiagnostic criteriafor diabetes bechanged?thr r impr dirc bw (i) dfig dib idiy idivid wih dib d h cq ciic dcil implici hi digi d (ii) dfig dib repidemilgicl purpe. In he rmer he digni require cre-

l bii wih rig hr dy l h pri ympmic d h plm glc i qivclly lvd

whr i pidmilgicl di rp ig i rrly pr-rmd. Wh rp ig i prrmd, pprximly 75% ppl wih dib dcd i pidmilgicl di r c-frmd hv ciic dib8,9.

In the absence o a more specic biological marker to dene diabetes,plasma glcose estimation remains the basis o diagnostic criteria. Other

considerations also impact on how a diagnosis o diabetes shold be

made. Does diabetes represent the pper end o a continos distribtiono glcose or a discrete entity? While hyperglycaemia is an importantprognostic parameter, is it the central or most important eatre deter-

mining prognosis in people with hyperglycaemia? In terms o screeningasymptomatic people, how can we best balance the medical, social and

economic benets and costs?

Althogh sch qestions are still debated, knowledge on diagnostic ct-

points or diabetes has been derived rom two sets o inormation:

plasma glcose levels associated with risk o diabetes specic micro-

vasclar complications, particlarly retinopathy

the poplation distribtion o plasma glcose.


16/50

0

The occrrence o diabetes-specic complications has been sed to

derive diagnostic ct-points or diabetes, particlarly sing data romepidemiological stdies which have examined both prevalent and incident

retinopathy across a range o plasma glcose levels.

Figre 1 shows an example o the data typically sed to examine thisrelationship in which deciles o plasma glcose are plotted against preva-

lence o retinopathy. The diagnostic ct-point is determined as the level at

which the risk o retinopathy increases. Few stdies have been ideal or

this prpose and most have limited statistical power. Stdies have diered

in the methods sed to diagnose retinopathy and whether or not peoplewith previosly diagnosed diabetes are inclded in the analysis. I people

with diagnosed diabetes who are receiving blood glcose lowering treat-

ment are inclded, the poplation-based characteristics o the stdy sam-

ple are maintained bt a bias associated with treatment-indced eects

on plasma glcose is introdced. Exclding people with treated diabetesrom analyses eliminates the bias related to treatment eect bt changes

the characteristics o the poplation with diabetes10. Frthermore, thespecic methodology sed to derive ct-points (eg based on decile o the

infection point, or Receiver Operator Characteristic [ROC] crve analyses)

infences the reslt. Figre 1 shows that the prevalence o retinopathyrises markedly in the 9th decile o plasma glcose. Most stdies havesed the lower limit o the decile as the ct-point, bt the median within

the decile might also be an appropriate choice or the ct-point.

Data rom Pima Indians, a stdy in Egypt and npblished NHANES II I data

were cited in the 1997 ADA report11

. These analyses inclded people withmedication-treated diabetes. using the lower limit o the decile in which

prevalent retinopathy increases signicantly, the ct-points rom thesethree stdies are 10.3mmol/l, 8.6mmol/l and 8.5mmol/l respectively or

2h plasma glcose and 6.1mmol/l, 6.0mmol/l and 5.9mmol/l respec-tively or asting plasma glcose. using data which exclded people with

diabetes gives somewhat dierent reslts: 9.0mmol/l, 8.9mmol/l and9.9mmol/l respectively or 2h plasma glcose and 6.0mmol/l, 6.1mmol/l

and 6.4mmol/l respectively or asting plasma glcose.

Figre 1. prvc rihy by dci

ig gc

1. Dibcmici

1. Dibcmici

1 2 3 4 5 6 7 8 9 10

30

25

20

15

10

5

0

Percentwithretinopathy

Decile of fasting glucose

1 2 3 4 5 6 7 8 9 10

30

25

20

15

10

5

0

Percentwithretinopathy

Decile of fasting glucose



17/50

One stdy has reported asting and 2h plasma glcose ct-points andincident retinopathy. Over an 11yr ollow p, development o retinopathy

increased at a baseline asting plasma glcose ct-point o 7.0mmol/l and

at a baseline 2h post-glcose load plasma glcose o 13.3mmol/l12.

In the review o the evidence to prepare this report, at tempts to nd datawhich examined the relationship between plasma glcose and biopsyproven diabetic renal disease were nsccessl. Stdies which haveexamined the relationship between plasma glcose and the less specic

marker o diabetic renal disease, proteinria, have reported some as-sociation bt not as strong as with retinopathy13.

Nmeros stdies have examined the relationship between plasma gl-cose and mortality and cardiovasclar complications bt have ailed toshow a denite threshold which cold be sed to dene diabetes (seeIsses 24 below).

Figre 2 shows two di erent distribtions o plasma glcose a nimodal

distribtion in which the entire poplation is represented by a single crve,

and a bimodal distribtion in which the poplations can be divided intotwo separate bt overlapping grops. With a bimodal distribtion, thepoint at which the two crves intersect is sed to separate abnormal rom

normal. It shold be noted that adopting a ct-point or separating two

components o a bimodal distribtion o plasma glcose does not neces-

sarily have any biological or pathogenic implications or adverse health

otcomes which may be attribtable to diabetes.A bimodal distribtion o plasma glcose concentrations was rst de-scribed in a 1971 stdy in Pima Indians14 and sbseqently in other pop-

lations with a high prevalence o diabetes, inclding Pacic Islanders15,16,

Asian Indians17, Mexican Americans18, Egyptians10 and Malaysians19. Re-

cently bimodality has also been reported in an elderly white poplationliving in Sothern Caliornia20.

Figre 2. Iri imd d bimd

diribi m gc

2. pidiribi m gc

2. pidiribi m gc

Unimodal curve Bimodal curveUnimodal curve Bimodal curve

issue 1 should the current diagnostic criteria for diabetes be changed?


18/50

In poplations with a bimodal distribtion, the plasma glcose ct-point

(dened as the point where the two crves intersect) shows variationbetween stdies. For example the Rancho Bernardo stdy20 and the Ma-

laysian stdy19 reported a similar 2h plasma glcose ct-point range o

11.0 to 13.3mmol/l (depending on age grop). However an analysis o the

DETECT2 data rom 26 dierent contries with plasma glcose meas-red dring an oral glcose tolerance test (OGTT) ond a wide variation

in ct-points. The ct-point or asting plasma glcose or the dierent

contries ranged rom 5.7 to 8.5mmol/l (median 7.1mmol/l) and or 2hplasma glcose ranged rom 9.1 to 17.9mmol/l (median 12.4mmol/l)21.

In smmary there are an abndance o data indicating that hyperglycae-

mia is harml. However there are limitations in the data and the method-

ologies sed to derive ct-points at which this level o harm is specically

increased and which clearly dierentiate diabetes rom non-diabetes.

Rcmmdi 1 The crrent WHO diagnostic criteria or diabetes shold be maintained asting plasma glcose 7.0mmol/l (126mg/dl) or 2h plasma glcose

11.1mmol/l (200mg/dl).

Despite the limitations with the data rom which the diagnostic criteria

or diabetes are derived, the crrent criteria distingish a grop with

signicantly increased prematre mortality and increased risk o micro-

vasclar and cardiovasclar complications.



19/50

issue

how should normalplasma glucoselevels be defined?

o pprch ddrig h i dfig cgri i-rmdi hyprglycmi i df rml glc lrc.Hwvr, hi migy im qi i difc wr. ech h aDa pblici h digic criri r dib hdfd rml plm glc lvl. th 2003 aDa m4dfd rm ig m gc


20/50

Prospective poplation stdies provide inormation on the relationshipbetween plasma glcose and risk o death, cardiovasclar disease andcancer, and o developing diabetes.

Levitan et al22 perormed a meta-analysis o thirty eight prospective std-

ies and conrmed that hyperglycaemia in the non-diabetic range was as-

sociated with increased risk o atal and non-atal cardiovasclar disease,

with a similar relationship between events and each o asting and 2hplasma glcose. From twelve stdies reporting asting plasma glcoselevels and six stdies reporting post-challenge glcose, cardiovasclarevents appeared to increase in a linear ashion with 2h post-challenge

plasma glcose in the range below levels diagnostic o diabetes withot

a threshold, whereas or asting plasma glcose there was a possible

threshold at 5.5mmol/l.

The Paris Prospective stdy showed J-shaped relationships or all-case

mortality with both asting and 2h glcose concentrations, and thelowest observed death rates were in the intervals centred on 5.5mmol/lor asting glcose and 5.0mmol/l or 2h glcose. For ischaemic heartdisease death or asting glcose, the hazards ratio was best modelledby a positive linear relationship bt or 2h glcose, i t was modelled bya J-shaped crve and the lowest observed death rate was in the interval

centred on 6.0mmol/l23.

The DECODE stdy, ater adjsting or other cardiovasclar risk actors,

reported a J-shaped relationship between mortality and glcose withthe lowest rates or a asting plasma glcose o 4.506.09mmol/l. A

J-shaped relationship was also observed between 2h plasma glcoseand all-case mortality and non-cardiovasclar disease mortality withthe lowest rates at a 2h plasma glcose o 4.515.50mmol/l24. Therewas a graded relationship between cardiovasclar mortality and 2h

plasma glcose with the lowest rates at the lowest 2h plasma glcosedistribtion24.

A 33-yr ollow-p o the Whitehall stdy in which a 50g OGTT was per-ormed at baseline in 19671969 in 17869 male civil servants aged 4064

years reported a threshold model with linear slope as best describing the

relationship between 2h post-load blood glcose and mortality risk. The

hazard o coronary mortality rose in a linear ashion rom a threshold 2hblood glcose o 4.6mmol/l. At a 2h blood glcose o 11.1mmol/l, age-

adjsted hazard ratio was 3.6 (95% CI 2.35.6) compared with a level o

4.6mmol/l. The graded relationship persisted bt was at tenated by 45%

ater adjstment or baseline ischaemic heart disease, BMI, systolic blood

pressre, blood cholesterol, smoking, physical activity, lng nction, and

employment grade25.

In the Baltimore Longitdinal stdy, all case mortality increased signi-

cantly rom a asting plasma glcose above 6.1mmol/l bt not lower levels.

For 2h plasma glcose, risk increased signicantly above 7.8mmol/l26.

Plasma glcose levels are also associated with cancer risk. Jee et alreported increased risk or all cancers in a cohort o 1.3 million people

2. Ciic rch rik dvrcm

2. Ciic rch rik dvrcm



21/50

ollowed or pto 10 years with increasing asting serm glcose above6.1mmol/l27.

There is also no level o plasma glcose at which risk o developingdiabetes is clearly increased. In a grop o yong Israeli men, risk o

incident diabetes progressively increased with a asting plasma glcose4.8mmol/l compared with a asting plasma glcose o


22/50


23/50

issue

how should impairedglucose tolerancebe defined?

I 1979 h us nil Dib D Grp31 rcmmdd hcgry Impird Glc tlrc (IGt) d icrd rik prgrig dib, lhgh i w l

d h my wld rvr rml. thi rm w irdcd rmv h igm dib rm h hr rm i h im d h rg bw rml d dib. thicrd rik crdivclr di i ppl wih IGt wl rcgid. thi cgry d dfii l rd i h1980 WHo rr6. IGt i ciic iy b i rik cr rr dib d/r dvr cm. sdi gg h IGti cid wih mcl ili ric d dciv iliecrein, reuling in le efcien dipl he gluce ld dur-

ig h oGtt32.

The prevalence o IGT varies between poplations and across dierentage grops. Prevalence rates in the order o 10% or more are commonand it is typically more common in women than in men. The increasingprevalence with increasing age was illstrated in the DECODE stdy which

showed the prevalence o isolated IGT increasing rom 2.9% in 3039yr

old men to 15.1% in 7079yr old men and rom 4.5% in 3039yr oldwomen to 16.9% in 7079yr old women33. A similar pattern is generally

seen in Asian poplations with prevalence o IGT increasing with age p

to 7089 years. However in the Indian poplation the prevalence o IGT is

higher and does not change mch with age34.


24/50

Data rom Maritis indicate that in people with IGT at baseline, 30%

reverted to normal, 35% remain as IGT, 5% changed to IFG and 30%developed diabetes over the 11-yr ollow p period35.

The reprodcibility o IGT with retesting within 6 weeks is only moder-

ate36

. The proportion o people classied with IGT on the rst OGTT andon retesting ranged rom 33% to 48% with 39% to 46% being reclassied

as normal and 613% as having diabetes on repeat testing.

IGT was initially dened as a category o glycaemia associated with anincreased risk o developing diabetes bt is now increasingly recognised

as being associated with a signicantly increased risk o prematre mor-

tality and cardiovasclar disease. The McMaster review36 reported theollowing:

the annalized relative risk o a person with IGT progressing to diabe-

tes was increased 6-old compared with people with normal glcose

tolerance. This relative risk was even higher in people with both IFGand IGT being increased 12-old.

the relative risk o all-case mortality is 1.48-old higher in peoplewith IGT compared with people with normal glcose tolerance. The

relative risk o a atal cardiovasclar otcome was 1.66-old higher.

The validity o the crrent denition centres on the risk o developingdiabetes or risk o adverse otcomes associated with 2h plasma gl-cose levels. The 2h post-load plasma glcose ct-point o 7.8mmol/l or

dening IGT was derived primarily rom Pima Indian data which examined

the risk o incident diabetes37. The incidence ranged rom less than 2.0percent/yr in those with 2h plasma glcose levels o


25/50

threshold22, or in a J-shaped relationship with the lowest observed death

rates centred on 5.0mmol/l or 2h glcose or all-case mortality and6.0mmol/l or coronary heart disease death23. The DECODE stdy alsoreported a J-shaped relationship between all-case and non-cardiovas-

clar mortality and glcose with the lowest rates at a 2h plasma glcose

o 4.515.50mmol/l, and a graded relationship between cardiovasclarmortality and 2h plasma glcose24. The Whitehall stdy showed thatcoronary mortality rose in a linear ashion rom a 2h blood glcose o

4.6mmol/l25.

In smmary, althogh there are limited data to spport the crrent 2hplasma glcose vale sed to dene IGT, the crrent ct-point seems to

be operationally adeqate. However, it is important to note that the risk o

tre diabetes, prematre mortality and cardiovasclar disease beginsto increase at 2h plasma glcose levels below the IGT range. Since the

rationale or this category is to dene a risk state or tre diabetes

and/or tre cardiovasclar disease and prematre mortality, a risk scorecombining known risk actors which incldes a measre o glcose as a

continos variable, wold seem a more logical approach.

Rcmmdi 3 The crrent WHO denition or Impaired Glcose Tolerance (IGT) sholdbe maintained or the present.

Consideration shold be given to replacing this category o intermedi-ate hyperglycaemia by an overall risk assessment or diabetes, cardio-

vasclar disease, or both, which incldes a measre o glcose as acontinos variable.

issue 3 how should impaired glucose tolerance be defined?


26/50

0


27/50

issue

how should impairedfasting glucose be

defined?I 1997 h aDa expr Cmmi11 irdcd h cgry Im-pired ing Gluce (IG) decribe he zne beween he upper

limi rml ig plm glc d h lwr limi hdibic ig plm glc. thi w blivd h im b lg h z bw h ppr limi rml 2hplm glc d h lwr limi h dibic 2h plm gl-

c dcribd by IGt. thi rcmmdi w dpd by WHoi 19993. IG, wih IGt, i cliicl iy b rhr rikcr r r dib d dvr cm. IG i cidwih impired inulin ecrein nd impired upprein hepic

glc p. I hld b d h ll d i hi ci rr IG dfd by h WHo 1999 rr i.. ig m gc6.16.9mm/ (110125mg/d) iciv, hrwi d.

DECODE data show that o all Eropean people with IFG dened by a

asting plasma glcose o 6.16.9mmol/l alone, 64.8% have isolatedIFG, 28.6% have IGT and 6.6% have diabetes33. Similarly, DECODA data

show that Asian people with IFG dened by a asting plasma glcoseo 6.16.9mmol/l alone, 45.9% have isolated IFG, 35.2% have IGT and

18.9% have diabetes34.

The prevalence o IFG varies between poplations and across dierentage grops within poplations. Overall prevalence rates in the order o5% or more are common. IFG is typically more common in men than in

women. The DECODE stdy showed an increase in prevalence o isolated

IFG rom 5.2% in 3039yr old men pto 10.1% in 5059yr old men and

then a decrease to 3.2% in 8089yr old men, whereas in women preva-lence increased rom 2.6% in 3039yr olds to 5.9% in 7079yr olds 33.


28/50

In Asian poplations prevalence o isolated IFG generally increases withage, except in the Indian poplation where prevalence does not changemch with age34.

Data rom Maritis35 indicate that in people with IFG at baseline, 40%

reverted to normal, 15% remained as IFG, 20% changed to IGT and 25%developed diabetes over an 11-yr ollow p period.

Two stdies which assessed the reprodcibility o IFG with retestingwithin 6 weeks showed that the proportion o people classied as IFGon the rst test and on retesting was 64% and 51% respectively withthe majority being reclassied as normal and less than 10% as having

diabetes on repeat testing36.

The annalized relative risk o people with isolated IFG progressing todiabetes compared with people with normal glcose tolerance showeda 4.7-old increase in the three stdies inclded in the review by the

McMaster grop33

. IFG was associated with increased risk o adverse ot-comes with a relative risk ranging rom 1.191.28 or non-atal myocardial

inarction, non-atal cardiovasclar disease, cardiovasclar mortality and

all-case mortality36.

The adoption o a ct-point or asting plasma glcose o 6.1mmol/l

(110mg/dl) by the ADA11 and WHO3 as the pper limit o normoglycaemia

was based on this being the level above which rst phase inslin secre-

tion is lost in response to intravenos glcose and which is associated

with progressively greater risk o developing micro and macrovasclarcomplications. Since then, there has also been an increased ocs onthe risk o the tre development o diabetes. While the ADA Committee

noted that the ideal method or selecting the ct-point or IFG wold be

by the identication o a threshold o asting plasma glcose at which the

risk o adverse clinical or metabolic otcomes rises sharply, available data

do not show a denite or consistent threshold or asting plasma glcose

and adverse otcomes4.

The risk o developing diabetes is increased even with asting plasma gl-

cose levels which are considered normal. Tirosh and colleages 28 reported

that in a grop o yong Israeli men, risk o incident diabetes progres-sively increased with a asting plasma glcose 4.8mmol/l comparedwith a asting plasma glcose o


29/50

specicity or predicting tre diabetes. However there is a dierence in

the actal incidence o diabetes at dierent asting plasma glcose levels.

In the Maritis stdy the 5-yr incidence o diabetes was in the order o

15% or a asting plasma glcose o 5.55.7mmol/l compared with 30%

or a asting plasma glcose o 6.16.9mmol/l40. In the ARIC stdy aasting plasma glcose 5.6mmol/l had a sensitivity o 70% in predict-

ing incident diabetes compared with 50% or 5.9mmol/l. However this

increased sensitivity occrred at the expense o a marked increase in the

percentage o the poplation identied as being abnormal (40% v 20%respectively) in this middle-aged uS poplation41.

In the Eropean Diabetes Epidemiology Grop position statement on the

threshold or diagnosing IFG, Forohi et al examined the magnitde o the

association between dierent thresholds or IFG and diabetes rom pb-

lished data, or derived it rom the pblished data42. Overall, the magnitde

o the association between diabetes and IFG is greater or IFG dened

as FPG 6.16.9mmol/l than or FPG 5.66.0mmol/l. For example in theData rom an Epidemiological Stdy on the Inslin Resistance Syndrome

(DESIR) stdy, diabetes incidence rates per 1,000 person-years or IFG

categories o


30/50

Since risk o adverse otcomes and tre diabetes is continos across

the asting plasma glcose range, the ct-point chosen to dene IFG will

be somewhat arbitrary. Thereore other considerations shold be takeninto accont in recommending a ct-point. The Grop considered the ol-

lowing points relevant to reaching its recommendation:

Otcomes

Mortality, cardiovasclar disease, microvasclar complications

Incident diabetes

Prevention o

prematre mortality and cardiovasclar disease

progression to diabetes

Impact on prevalence o IFG

Concordance o IFG and IGTRisk prole o individals identied with IFG

Economic considerations and cost implications

Implications or health services and policy

Each o these points is considered below.

As reviewed above, available data do not point to a specic and consistent

ct-point or adverse cardiovasclar otcomes or mortality.

The Grop expressed reservations abot sing incident diabetes as the

only end point or recommending a ct-point or IFG. The Grop also has

reservations abot the method sed by the ADA to derive the ct-pointo 5.6mmol/l (i.e. maximising the sm o sensitivity and specicity). The

Grop was o the opinion that a ct-point shold inclde clinical and pblic

health considerations, and not merely statistical ones.

Actal incidence rates o diabetes are an important consideration. For ex-

ample, Maritis40 and Pima Indian39 data indicate that risk o progressiono IFG in the additional people identied by the recently recommendedADA criteria is hal that o WHO dened IFG (approximately 15% v 30%over 5 years).

unortnately there are no data on clinical otcomes o interventions inpeople with IFG.

shd h crrWHo criri rdfig IG bchgd?

shd h crrWHo criri rdfig IG bchgd?

ocmocm

Icid dibIcid dib

prvi rmrmriy dcrdivcrdi

prvi rmrmriy dcrdivcrdi



31/50

unlike IGT where there is extensive evidence rom well condcted ran-

domised controlled stdies that liestyle and pharmacological interven-tions can prevent or delay progression to diabetes44,45,46,47,48,49,50, there are

crrently only limited data on the preventability o progression o IFG to

diabetes. The recently reported DREAM stdy showed that the thiazoli-dinedione, rosiglitazone, was associated with a similar risk redction oprogression rom WHO dened IFG and IGT to diabetes or death hazards

ratio 030 (019049) or isolated IFG and 045 (036055) or isolated

IGT51. There are no data or prevention o progression to diabetes romADA dened IFG.

A nmber o stdies have reported a 23-old increase in IFG prevalence

sing the new ADA recommended criteria compared with WHO dened

IFG, highlighted by data rom the DETECT2 stdy52. Figre 3 shows the

prevalence o IFG increasing rom 11.8% to 37.6% in Denmark, rom10.6% to 37.6% in India and rom 9.5% to 28.5% in the uS. Sch in-creases have major conseqences in contries sch as in India and China

where the nmber o people with IFG in the 4064-yr age range woldincrease by 13 million and 20 million respectively sing the 2003 ADA

criteria to dene IFG.

Figre 3. Cmri rvc IG digd wih1999 WHo criri d 2003 aDa criri

prvi rgri dib

prvi rgri dib

Imc rvc IGImc rvc IG

Denmark India USA Denmark India USA

0

5

10

15

20

25

30

35

40

45

50

%

WHO criteria New ADA criteria

IFGIFG/GT

Denmark India USA Denmark India USA

0

5

10

15

20

25

30

35

40

45

50

%

WHO criteria New ADA criteria

IFGIFG/GT

issue 4 how should impaired fasting glucose be defined?


32/50

One o the reasons cited by the ADA Committee or lowering the IFG ct-

point was to increase concordance with IGT4. Data rom the Danish INTER-

99 stdy53 show that the percentage o people with IGT who also have IFG

increases rom 25% sing the WHO criteria (FPG 6.16.9mmol/l) to 60%

with the 2003 ADA criteria (FPG 5.66.9mmol/l). However, the overallpercentage o people with IFG who have IGT decreases rom 25% withthe WHO criteria to 20% with the 2003 ADA criteria. While the numbero

people with ADA dened IFG who also have IGT increases, the proportion

o people with ADA dened IFG who have IGT decreases. Thereore opeople identied with IFG by the new ADA denition, there is an increase

in the nmber o people who have IFG bt do not have IGT compared with

WHO dened IFG. Frthermore, intervening in people with ADA denedIFG with prevention strategies which have been shown to benet people

will IGT will inclde intervening in a greater nmber o people who do not

have IGT and or whom there is no evidence o benet.

Stdies have reported that the cardiovasclar risk prole o people withWHO dened IFG is worse than the additional people identied with IFGsing the 2003 ADA criteria52,54,55.

These are important or policy makers, pblic health agencies, insrers,

health care providers and consmers. However there are crrently noanalyses comparing the economic impact o WHO and ADA IFG criteriaor the modelled impact o dierent ct-points as there have been orassessing dierent diabetes screening strategies.

The large increase in prevalence o IFG which wold reslt rom lowering

the ct-point has signicant implications or the health system o manycontries which are strggling to cope with caring or people with estab-

lished diabetes. There is also concern that increasing the ocs on IFG

wold divert prevention resorces rom IGT, where there is more extensiveevidence o benet.

Ccrdc IG d IGtCcrdc IG d IGt

Rik rf idivididifd

Rik rf idivididifd

ecmic

cidri dc imici

ecmic

cidri dc imici

Imc hhrvic d icyImc hhrvic d icy



33/50

Rcmmdi 4 The asting plasma glcose ct-point or Impaired Fasting Glcose (IFG)shold remain at 6.1mmol/l.

This decision was based on concerns abot the signicant increase in

IFG prevalence which wold occr with lowering the ct-point and theimpact on individals and health systems. There is a lack o evidence o

any benet in terms o redcing adverse otcomes or progression to dia-

betes and people identied by a lower ct-point eg 5.6mmol/l (100mg/dl)

have a more avorable cardiovasclar risk prole and only hal the risk

o developing diabetes compared with those above the crrent WHO ct-

point. Lowering the ct-point wold increase the proportion o peoplewith IGT who also have IFG bt decreases the proport ion o people with

IFG who also have IGT.

As discssed in the section on IGT, consideration shold be given to

replacing this category o intermediate hyperglycaemia by an overall riskassessment or diabetes, cardiovasclar disease, or both, which may

inclde a measre o glcose as a continos variable.

The recommendation on IFG threshold in this report is identical to theposition statement o the Eropean Diabetes Epidemiology Grop which

also recommends that the diagnostic threshold or IFG shold remain

at 6.1mmol/l42.

The Grop was mindl o the implications o having dierent WHO andADA criteria or IFG. The ADA recommendations are targeted to health

care providers in one contry compared with the global WHO recom-mendations. It was also noted that other signicant discrepancies already

exist between the ADA and WHO recommendations inclding the method

or diagnosing diabetes eg asting plasma glcose verss oral glcosetolerance test. Althogh these dierent recommendations or IFG maylead to some consion initially it shold stimlate research to provide

data to resolve the discrepancy and other isses associated with dening

ct-points or asting plasma glcose.

Imici chgigh crr WHodigic cririr IG

Imici chgigh crr WHodigic cririr IG

issue 4 how should impaired fasting glucose be defined?


34/50


35/50

issue

what diagnostictests should be

used to defineglycaemic status

Measrement o glcose in blood remains the mainstay o testing orglcose tolerance stats. There are a nmber o important considerations

which can infence this measrement which reqire carel attention in

order to ensre an accrate reslt.Most portable devices measre the glcose concentration directly in the

plasma component o the blood by ltering ot the red blood cells. Thesignal is then calibrated to prodce a readot either as blood or plasmaglcose. Laboratory measres normally now se separated plasma, with

determination o the amont or concentration o glcose in a xed volme.

Only devices which measre ot a xed volme o blood, and then deter-

mine the glcose within that volme, measre tre whole blood glcose

concentration. As glycolysis inhibitors take time to penetrate into redblood cells, only immediate separation o plasma will avoid some lower-

ing o glcose levels in the sample, thogh rapid cooling can redce thisloss. Accordingly modern recommendations are or laboratory plasmameasrements on appropriately handled samples, and matched calibra-

tion o portable devices.

Glcose measred in plasma is approximately 11% higher than glcosemeasred in whole blood. However this dierence is dependent on hae-

matocrit, increasing to 15% at a haematocrit o 0.55 and decreasing to8% at a haematocrit o 0.3056. For this and other reasons the conver-sion o whole blood glcose to plasma glcose is problematic and thepreviosly pblished WHO conversion tables may be inaccrate in some

sitations. It shold also be noted that many portable glcose measr-

ing devices are still calibrated to whole blood despite the International

1. Mrm gc ibd

1. Mrm gc ibd


36/50

0

Federation o Clinical Chemistry (IFCC) recommendation that all glcose

measring devices report in plasma vales57.

Measrement dierences may also arise depending on the site o collec-

tion o the blood sample. Venos and capillary samples will give the same

reslt in the asting state bt in the non-asting state capillary will givehigher reslts than venos samples.

The processing o the sample ater collection is important to ensre ac-

crate measrement o plasma glcose. This reqires rapid separation o

the plasma ater collection (within mintes) bt it is recognised that this

seldom occrs. Collection into a container with glycolytic inhibitors (egNaF) is only partially eective. A minimm reqirement is that the sample

shold be placed immediately in ice-water ater collection and beoreseparating bt even so separation shold be within 30min58.

Rcmmdi 5 1. Venos plasma glcose shold be the standard method or measringand reporting. However in recognition o the widespread se o cap-

illary sampling, especially in nder-resorced contries, conversionvales or capillary plasma glcose are provided or post- load glcose

vales. Fasting vales or venos and capillary plasma glcose are

identical.

2. Glcose shold be measred immediately ater collection by nearpatient testing, or i a blood sample is collected, plasma sholdbe immediately separated, or the sample shold be collected into

a container with glycolytic inhibitors and placed on ice-water ntilseparated prior to analysis.

There is contining debate abot the place o the OGTT or clinical andepidemiological prposes. The test is recommended by the WHO3. Al-thogh ADA acknowledges the OGTT as a valid way to diagnose diabetes,

the se o the test or diagnostic prposes in clinical practice is discor-

aged in avor o asting plasma glcose or several reasons, inclding

inconvenience, greater cost and less reprodcibility11

. Some o this vari-ation can be minimised with attention to dietary preparation and taking

care to collect the 2h sample within 5 min o 120 min59.

Many stdies have reported that asting plasma glcose and 2h post-gl-

cose plasma glcose do not identiy the same people as having diabetes.

In the DECODE stdy60, o the 1517 people with newly diagnosed diabetes,

40% met only the asting plasma glcose criterion, 31% met only the 2h

plasma glcose criterion and 28% met both criteria. Thereore sing only

the asting plasma glcose will ail to diagnose approximately 30% opeople with diabetes. Data rom the NHANES III stdy cited in the 1997

ADA report show similar ndings or newly diagnosed diabetes

11

. Thisdiscrepancy is more obvios in an older poplation. Barrett-Conner et al61

2. or gcrc (oGtt)

2. or gcrc (oGtt)



37/50

reported that 70% o women and 48% o men aged 5089 years had new

diabetes diagnosed solely by an elevated 2h plasma glcose.

Does this matter and are there any dierences in otcomes or peoplediagnosed on the basis o the asting or 2h plasma glcose or both?

Many stdies have docmented increased rates o mortality in peoplewith diabetes. Stdies which have compared these rates in relation todiabetes diagnosed on the basis o asting or 2h plasma glcose haveconsistently shown worse otcomes in those diagnosed on the basis othe 2h plasma glcose reslt.

The Hoorn stdy showed that all-case and cardiovasclar mortality over

an 8yr ollow-p was signicantly elevated in those with 2h plasmaglcose 11.1mmol/l bt not in those with a asting plasma glcose 7.0mmol/l62. In the DECODE stdy, hazard ratios (HR) (95% CI) or diabe-

tes diagnosed on a asting plasma glcose 7.0mmol/l was 1.6 (1.4 1.8)

or all-case mortality, 1.6 (1.31.9) or cardiovasclar mortality, and 1.6

(1.4 1.9) or non-cardiovasclar mortality, respectively. The correspond-

ing HRs or diabetes diagnosed on a 2h plasma glcose 11.1mmol/lwere 2.0 (1.72.3), 1.9 (1.52.4) and 2.1 (1.72.5), respectively. The HR

or previosly ndetected diabetes dened by the 2h plasma glcosewas not signicantly dierent rom that or known diabetes, bt wassignicantly higher than that or ndetected diabetes based on the ast-

ing plasma glcose24. A rther DECODE analysis has shown that whilemortality is increased in people with newly diagnosed diabetes based on

either the asting plasma glcose or 2h plasma glcose, this increased

risk is no longer signicant or asting plasma glcose 7.0mmol/l when

adjsted or 2h plasma glcose bt risk based on 2h plasma glcose 11.1mmol/l remains signicant when adjsted or asting plasma gl-cose63.

Shaw et al64 reported a 2.7-old increased risk o all-case mortality inmen and a 2-old increase in women rom three poplation-based longit-

dinal stdies (in Maritis, Fiji and Nar) in people with newly diagnosed

diabetes on the basis o an elevated 2h plasma glcose compared with

people with normal glcose tolerance. In contrast people with diabetesdiagnosed on the basis o asting plasma glcose alone did not have an

increased risk. However, not all stdies have observed this nding65.

The 2h plasma glcose also seems important or microvasclar compli-cations. Ito et al12 reported that the incidence o retinopathy in people with

newly diagnosed diabetes increased sbstantially only in those with 2h

plasma glcose levels above 11.1mmol/l, even in those with asting plasma

glcose 7.0mmol/l. In people with asting plasma glcose 7.8mmol/lbt 2h plasma glcose


38/50

people can only be achieved with an OGTT. In addition, IGT can only be

diagnosed with an OGTT.

Rcmmdi 6 The oral glcose tolerance test (OGTT) shold be retained as a diagnos-tic test or the ollowing reasons

asting plasma glcose alone ails to diagnose approximately 30% o

cases o previosly ndiagnosed diabetes

an OGTT is the only means o identiying people with IGT

an OGTT is reqently needed to conrm or exclde an abnormality

o glcose tolerance in asymptomatic people

An OGTT shold be sed in individals with asting plasma glcose6.16.9mmol/l (110125mg/dl) to determine glcose tolerance stats.

HbA1c refects average plasma glcose over the previos 23 months in

a single measre which can be perormed at any time o the day and does

not reqire any special preparation sch as asting. These proper ties have

made it the gold standard or assessing glycaemic control in people with

diabetes and have reslted in its consideration as an option or assessing

glcose tolerance in people withot diagnosed diabetes.

The relationship between HbA1c and prevalent retinopathy is similar to

that o plasma glcose as illstrated in Figre 1. This relationship wasoriginally reported in Pima Indians66 and has also been observed in

several poplations inclding Egyptians10, Americans11 and Japanese67.Overall the perormance o HbA1c has been similar to that o asting or2h plasma glcose bt the actal HbA1c threshold vale has dieredbetween stdies.

These avorable aspects o HbA1c need to be balanced against the real-

ity that HbA1c measrement is not widely available in many contriesthroghot the world. Frthermore there are aspects o its measrement

which are problematic. Althogh in reerence laboratories the precision o

HbA1c measrement is similar to that o plasma glcose, global consist-ency remains a problem. Frthermore the HbA1c reslt is infenced byseveral actors inclding anaemia, abnormalities o haemoglobin, preg-nancy and raemia. Some o these actors may be a bigger problem in

nder-resorced contries de to a higher prevalence o anaemia and o

haemoglobinopathies. The precise eect o these actors on the HbA1creslt varies with the laboratory method sed68.

Taking all o these considerations into accont, the Grop conclded that

the role o HbA1c in the diagnosis o diabetes and intermediate hyperg-lycaemia is not established and that it cold not be recommended as adiagnostic test at this time.

3. GycdHmgbi(Hba1c)

3. GycdHmgbi(Hba1c)



39/50

Rcmmdi 7 Crrently HbA1c is not considered a sitable diagnostic test or diabetesor intermediate hyperglycaemia.

The Grop recommends sing the term Intermediate Hyperglycaemia todescribe glycaemic levels between normal glcose tolerance and diabe-

tes. use o pre-diabetes is discoraged to avoid any stigma associated

with the word diabetes and the act that many people do not progress to

diabetes as the term implies. In addition this ocs on diabetes may divert

attention rom the important and signicantly increased cardiovasclarrisk.

Stdies are reqired to gide tre deliberations abot diagnostic criteria

which go beyond plasma glcose considerations and take into accontvarios aspects o benets and costs. uniorm methodologies and ap-proaches to analyses wold assist in prodcing niversally comparablereslts and interpretation o ndings. Pooling o data rom stdies which

individally have inscient power may help to resolve some otstanding

isses eg dening ct-points or diabetes at which risk o retinopathy

increases.

Retaining the risk categories o IFG and IGT in clinical practice shold be

reconsidered. Dening specic levels or intermediate hyperglycaemiamay not be the most appropriate way o dening tre risk o diabetes

or cardiovasclar disease. This risk might be better assessed by the se

o prediction scores which, in addition to plasma glcose, also incldeother risk actors.

trmigytrmigy

r dircir dirci

issue 5 what diagnostic tests should be used to define glycaemic status


40/50


41/50

G Roglic, Department o Chronic Diseases and Health Promotion,Geneva, Switzerland (Chair)

S Resniko, Department o Chronic Diseases and Health Promotion,Geneva, Switzerland

K Strong, Department o Chronic Diseases and Health Promotion,Geneva, Switzerland

N unwin, Department o Chronic Diseases and Health Promotion,

Geneva, Switzerland

KGMM Alberti, university o Newcastle pon Tyne, united Kingdom

PH Bennett, NIDDK, Phoenix, AZ, united States o America

K Borch-Johnsen, Steno Diabetes Center, Gentote, Denmark

S Colagiri, Prince o Wales Hospital, Randwick, Astralia (Chair)

M Engelga, CDC, Atlanta, united States o America

P Home, university o Newcastle pon Tyne, united Kingdom

Z Metelko, university Clinic Vk Vrhovac, Zagreb, Croatia

V Mohan, Madras Diabetes research Fondation, Chennai, India

Chang Y Pan, Chinese PLA General Hospital, Beijing, Peoples

Repblic o China

Qing Qiao, National Pblic Health Institte, Helsinki, Finland (Technical

Advisor)

K Ramaiya, Shree Hind Mandal Hospital, Dar es Salaam, united

Repblic o Tanzania

J Shaw, International Diabetes Institte, Caleld, Astralia

M-I Schmidt, universidade Federal do Rio Grande do Sl, Porto Allegre,

Brazil

J Tomilehto, National Pblic Health Institte, Helsinki, FinlandD Vistisen, Steno Diabetes Center, Gentote, Denmark (StatisticalAdvisor)

P Zimmet, International Diabetes Institte, Caleld, Astralia (cor-

responding member)

K Borch-Johnsen also represented the Eropean Association or the Stdy

o Diabetes. The American Diabetes Association was asked to provide a

representative bt was nable to do so.

G Roglic, Department o Chronic Diseases and Health Promotion,

Geneva, SwitzerlandN unwin, Department o Chronic Diseases and Health Promotion,

Geneva, Switzerland

Gidi srigGrGidi srigGr

tchicGidiDvmGr

tchicGidiDvmGr

WHo scrriWHo scrri

appendix 1membership of guideline

development committees


42/50

appendix 2summary of who

diagnostic criteria fordiabetes and intermediatehyperglycaemia

1965 1980 1985 1999

nrm Not dened Not dened

Fasting glcose Not specied


43/50

appendix 3comparison of 1999 who

and 2003 ada diagnosticcriteria

WHo 1999 aDa 2003

Dib

Fasting glcose 7.0mmol/l

r7.0mmol/l

r

2h glcose* 11.1mmol/l 11.1mmol/l

IGt

Fasting glcose


44/50

appendix 4disorders of glycaemia

aetiological types andclinical stages

ty

sgnrmgycmi Hyrgycmi

nrm gcrgi

Imird gctrc

rImird ig

Gc

Dib Mi

Notinsulin

requiring

Insulin

requ

iring

forcontr

ol

Insulinn

req

uir-

ing

forsurv

ival

Type 1*

Type 2*

Othr spcfc Typ**

Gtatonal Dabt**

Disorders o glycemia: etiologic types and stages. *Even ater prensenting in ketoacidosis, thesepatients can briefy retrn to normoglycemia withot reqiring continos therapy (i.e., honeymoon

remission); **in rare instances, patients in these categories (e.g;, Vacor toxicity, type 1 diabetespresenting in pregnancy) may reqire inslin or srvival.



45/50

references

1. Wild S, Roglic G, Green A, Sicree R, King H. Global Prevalence o Dia-

betes: Estimates or the year 2000 and projections or 2030. DiabetesCare. 2004; 27: 1047-1053.

2. American Diabetes Association. Economic costs o diabetes in the uS

in 2002. Diabetes Care. 2003; 26: 917-932.

3. World Health Organization: Denition, Diagnosis and Classication oDiabetes Mellits and its Complications: Report o a WHO Consltation.

Part 1: Diagnosis and Classication o Diabetes Mellits. Geneva, WorldHealth Org., 1999.

4. The Expert Committee on the Diagnosis and Classication o Diabetes

Mellits. Follow-p Report on the Diagnosis o Diabetes Mellits. Diabetes

Care 2003; 26: 3160-3167.

5. World Health Organization: Diabetes Mellits: Report o a WHO Expert

Committee. Geneva, World Health Org., 1965 (Tech. Rep. Ser., no. 310).

6. World Health Organization: Expert Committee on Diabetes Mellits.

Geneva, World Health Org., 1980 (Tech. Rep. Ser., no. 646)

7. World Health Organization: Diabetes Mellits: Report o a WHO StdyGrop. Geneva, World Health Org., 1985 (Tech. Rep. Ser., no. 727).

8. Mooy JM, Grootenhis PA, de Vries H, Kostene PJ, Popp-Snijders C,Heine RJ. Intra-individal variation o glcose, specic inslin and proin-

slin concentrations measred by two oral glcose tolerance tests in ageneral Cacasian poplation: the Hoorn stdy. Diabetologia 1996; 39:298-305.

9. Christensen JO, Sandbk A, Laritzen T, Borch-Johnsen K. Popla-tion-based stepwise screening or nrecognised Type 2 diabetes is ine-

ective in general practice despite reliable algorithms. Diabetologia 2004;

47: 15661573.10. Engelga MM, Thompson TJ, Herman WH, Boyle JP, Abert RE, Kenny

SJ, Badran A, Sos ES, Ali MA. Comparison o asting and 2-hor glcose

and HbA1c levels or diagnosing diabetes. Diagnostic criteria and perorm-

ance revisited Diabetes Care 1997; 20: 785-791.

11. The Expert Committee on the Diagnosis and Classication o Diabetes

Mellits: Report o the expert committee on the diagnosis and classica-

tion o diabetes mellits. Diabetes Care 1997; 20: 11831197.

12. Ito C, Maeda R, Ishida S, Harada H, Inoe N, Sasaki H. Importanceo OGTT or diagnosing diabetes mellits based on prevalence and in-

cidence o retinopathy. Diabetes Research and Clinical Practice 2000;49: 181186.


46/50

13. Gabir M, Hanson Rl , Dabelea D, Imperatore G, Romain J, Bennett PH,

Knowler WC. Plasma glcose and prediction o microvasclar disease and

mortality. Evalation o 1997 American Diabetes Association and 1999World Health Organization criteria or diagnosis o diabetes. Diabetes Care

2000; 23: 11131118.

14. Rshorth NB, Bennett PH, Steinberg AG, Brch TA, Miller M: Diabetes

in the Pima Indians: evidence o bimodality in glcose tolerance distrib-

tions. Diabetes 1971; 20: 756765.

15. Zimmet P, Whitehose S: Bimodality o asting and two-hor glcose

tolerance distribtions in a Micronesian poplation. Diabetes 1978; 27:

793 800.

16. Raper LR, Taylor R, Zimmet P, Milne B, Balka B: Bimodality in glcose

tolerance distribtions in the rban Polynesian poplation o Western Sa-

moa. Diabetes Res 1984; 1: 1 8.

17. Omar MA, Seedat MA, Dyer RB, Motala AA, Knight LT, Becker PJ.Soth Arican Indians show a high prevalence o NIDDM and bimodality in

plasma glcose distribtion patterns Diabetes Care 1994; 17: 70-73.

18. Rosenthal M, McMahan CA, Stern MP, Eifer CW, Haner SM, Hazda

HP, Franco LJ: Evidence o bimodality o two hor plasma glcose concen-

trations in Mexican Americans: reslts rom the San Antonio Heart stdy.

J Chronic Dis 1985; 38: 516.

19. Lim T-O, Bakri R, Morad Z, Hamid MA Bimodality in blood glcosedistribtion. Is it niversal? Diabetes Care 2002; 25: 22122217.

20. Fan J, May SJ, Zho Y, Barrett-Connor E. Bimodality o 2-h plasmaglcose distribtions in whites. The Rancho Bernardo Stdy. Diabetes

Care 2005; 28: 14511456.

21. Borch-Johnsen K, Colagiri S, Vistisen D on behal o the DETECT-2collaboration. Is there a bimodal distribtion o plasma glcose. American

Diabetes Association Annal Scientic Meeting 2006. Diabetes, 55 (Sppl

1), A19 (Abstract 77).

22. Levitan EB, Song Y, Ford ES, Li S. Is nondiabetic hyperglycemia a risk

actor or cardiovasclar disease? A meta-analysis o prospective stdies.

JAMA 2005; 293: 194-202.

23. Balka B, Bertrais S, Dcimetiere P, Eschwege E. Is there a glycemicthreshold or mortality risk? Diabetes Care 1999; 22: 696699.

24. DECODE Stdy Grop. Is the crrent denition or diabetes relevant to

mortality risk rom all cases and cardiovasclar and non-cardiovasclar

diseases? Diabetes Care 2003; 26: 688696.

25. Brnner EJ, Shipley MJ, Witte DR, Fller JH, Marmot MG. Relation be-

tween blood glcose and coronary mortality over 33 years in the Whitehall

Stdy. Diabetes Care 2006; 29: 26-31.

26. Sorkin, JD, Mller DC, Fleg JL, Andres R. The relation o asting and

2-h postchallenge plasma glcose concentrations to mortality. Data rom

the Baltimore Longitdinal Stdy o Aging with a critical review o theliteratre. Diabetes Care 2005; 28: 26262632



47/50

27. Jee SH, Ohrr H, Sll JW, Yn JE, Ji M, Samet JM. Fasting sermglcose level and cancer risk in Korean men and women. JAMA 2005;293: 194-202.

28. Tirosh A, Shai I, Tekes-Manova D, Israeli E, Pereg D, Shochat T, Ko-chba I, Rdich A or the Israeli Diabetes Research Grop. Normal asting

plasma glcose levels and Type 2 diabetes in yong men. N Engl J Med2005; 353: 1454-62.

29. Godsland IF, Jes JAR, Johnston DG. Loss o beta cell nction asasting glcose increases in the non-diabetic range. Diabetologia 2004;47: 1157-1166.

30. Piche MV, Arcand-Bosse J-F, Despres J-P, Persse L, Lemiex S,

Weisnagel SJ. What is a normal glcose vale? Dierences in indexeso plasma glcose homeostasis in sbjects with normal asting glcose.

Diabetes Care 2004; 27: 24702477.

31. National Diabetes Data Grop: Classication and diagnosis o diabetes

mellits and other categories o glcose intolerance. Diabetes 1979; 28:

10391057.

32. Abdl-Ghani MA, Jenkinson CP, Richardson DK, Tripathy D, DeFronzo

RA. Inslin secretion and action in sbjects with impaired asting glcose

and impaired glcose tolerance. Reslts rom the Veterans Administration

Genetic Epidemiology Stdy. Diabetes 2006; 55: 1430-1435.

33. DECODE Stdy Grop. Age- and sex-specic prevalences o diabetes

and impaired glcose reglation in 13 Eropean Cohorts. Diabetes Care

2003; 26: 6169.

34. DECODA Stdy Grop. Age- and sex-specic prevalence o diabetes

and impaired glcose reglation in 11 Asian Cohorts. Diabetes Care 2003;

26:17701780.

35. Soderberg S, Zimmet P, Tomilehto J, De corten M, Dowse GK,Chitson P, Stenlnd H, Gareeboo H, Alberti KG, Shaw J. High incidenceo type 2 diabetes and increasing conversion rates rom impaired asting

glcose and impaired glcose tolerance to diabetes in Maritis. Jornal

o Internal Medicine 2004; 256: 3747.

36. McMaster university Evidence Based Practice Center. Diagnosis,prognosis and treatment o impaired glcose tolerance and impairedasting glcose. Evidence Report 128. www.ahrq.gov.

37. Bennett PH, Knowler WC, Pett itt DJ, Carraher MJ, Vasqez B. 1982.Longitdinal stdies o the development o diabetes in the Pima Indians.

In Advances in Diabetes Epidemiology. E.Eschwege, editor. Elsevier Bio-

medical Press, Amsterdam, The Netherlands. p65-74.

38. Edelstein SL, Knowler WC, Bain RP, Andres R, Barrett-Connor EL,Dowse GK, Haner SM, Pettit t DJ, Sorkin JD, Mller DC, Collins VR, Ham-

man RF. Predictors o progression rom impaired glcose tolerance to

NIDDM: an analysis o six prospective stdies. Diabetes 1997; 46: 701-710.

references
http://www.ahrq.gov/http://www.ahrq.gov/


48/50

39. Gabir MM, Hanson RL, Dabelea D, Imperatore G, Romain J, Ben-nett PH, Knowler WC. The 1997 American Diabetes Association and 1999

World Health Organization criteria or hyperglycemia in the diagnosis and

prediction o diabetes. Diabetes Care 2000; 23: 11081112.40. Shaw JE, Zimmet PZ, Hodge AM, de Corten M, Dowse GK, Chitson

P, Tomilehto J, Alberti KGMM: Impaired asting glcose: how low shold

it go? Diabetes Care 2000; 23: 3439.

41. Schmidt M, Dncan B, Bang H, Pankow J, Ballantyne C, Golden S,Folsom A, Chambless L or The Atherosclerosis Risk In Commnities Inves-

tigators. Identiying Individals At High Risk For Diabetes: The Atheroscle-

rosis Risk In Commnities Stdy. Diabetes Care 2005; 28: 2013-2018.

42. Forohi NG, Balka B, Borch-Johnsen K, Dekker J, Glmer C, QiaoQ, Spijkerman A, Stolk R, Tabac A, Wareham NJ. The threshold or diag-

nosing impaired asting glcose: a position statement by the EropeanDiabetes Epidemiology Grop. Diabetologia 2006; 49: 822-827.

43. Balka B, Hillier T, Vierron E, DHor A, Lpinay P, Royer B, Born C.Comment to: Borch-Johnsen K, Colagiri S, Balka B et al. Diabetologia

2004; 47:1396-1402. Creating a pandemic o prediabetes: the proposed

new diagnostic criteria or impaired asting glycaemia. Diabetologia 2005;

48: 801802.

44. Pan XR, Li GW, H YH, Wang JX, Yang WY, An ZX, H ZX, Lin J, Xiao

JZ, Cao HB, Li PA, Jiang XG, Jiang YY, Wang JP, Zheng H, Zhang H,Bennett PH, Howard BV. Eects o diet and exercise in preventing NIDDM

in people with impaired glcose tolerance: the Da Qing IGT and DiabetesStdy. Diabetes Care. 1997; 20: 537544.

45. Tomilehto J, Lindstrom J, Eriksson JG, Valle TT, Hamalainen H,Ilanne-Parikka P, Keinanen-Kikaanniemi S, Laakso M, Loheranta A,Rastas M, Salminen V, usitpa M.. Prevention o type 2 diabetes mellits

by changes in liestyle among sbjects with impaired glcose tolerance.

N Engl J Med. 2001; 344: 13431350.

46. The Diabetes Prevention Program Research Grop. Redction in theincidence o type 2 diabetes with liestyle intervention or metormin. NEngl J Med. 2002; 346: 393403.

47. Chiasson J, Josse R, Gomis G, Haneeld R, Karasik M, Laakso A.Acarbose or prevention o type 2 diabetes mellits: the STOP-NIDDMrandomised trial. Lancet. 2002; 359: 20722077.

48. Bchanan TA, Xiang AH, Peters RK, Kjos SL, Marroqin A, Goico J,

Ochoa C, Tan S, Berkowitz K, Hodis HN, Azen SP. Preservation o pancre-

atic beta-cell nction and prevention o type 2 diabetes by pharmacologi-

cal treatment o inslin resistance in high-risk Hispanic women. Diabetes.

2002; 51: 27962803.

49. Torgerson J, Haptman J, Boldrin MN, Sjostrom L. XENical in thePrevention o Diabetes in Obese Sbjects (XENDOS) Stdy. Diabetes Care.

2004; 27: 155161.

http://volumes/workolic/au%20boulot/World%20Health%20Organization/Definition%20and%20Diagnosis%20of%20Diabetes%20Mellitus%20and%20Intermediate%20Hyperglycaemia/content/?Author=A.+D%e2%80%99Hourhttp://volumes/workolic/au%20boulot/World%20Health%20Organization/Definition%20and%20Diagnosis%20of%20Diabetes%20Mellitus%20and%20Intermediate%20Hyperglycaemia/content/?Author=P.+L%c3%a9pinayhttp://volumes/workolic/au%20boulot/World%20Health%20Organization/Definition%20and%20Diagnosis%20of%20Diabetes%20Mellitus%20and%20Intermediate%20Hyperglycaemia/content/?Author=B.+Royerhttp://volumes/workolic/au%20boulot/World%20Health%20Organization/Definition%20and%20Diagnosis%20of%20Diabetes%20Mellitus%20and%20Intermediate%20Hyperglycaemia/content/?Author=C.+Bornhttp://volumes/workolic/au%20boulot/World%20Health%20Organization/Definition%20and%20Diagnosis%20of%20Diabetes%20Mellitus%20and%20Intermediate%20Hyperglycaemia/content/?Author=C.+Bornhttp://volumes/workolic/au%20boulot/World%20Health%20Organization/Definition%20and%20Diagnosis%20of%20Diabetes%20Mellitus%20and%20Intermediate%20Hyperglycaemia/content/?Author=B.+Royerhttp://volumes/workolic/au%20boulot/World%20Health%20Organization/Definition%20and%20Diagnosis%20of%20Diabetes%20Mellitus%20and%20Intermediate%20Hyperglycaemia/content/?Author=P.+L%c3%a9pinayhttp://volumes/workolic/au%20boulot/World%20Health%20Organization/Definition%20and%20Diagnosis%20of%20Diabetes%20Mellitus%20and%20Intermediate%20Hyperglycaemia/content/?Author=A.+D%e2%80%99Hour


49/50

50. Ramachandran A, Snehalatha C, Mary S, Mkesh B, Bhaskar AD,

Vijay V. The Indian Diabetes Prevention Programme shows that liestylemodication and metormin prevent type 2 diabetes in Asian Indian sb-

jects with impaired glcose tolerance (IDPP-1). Diabetologia 2006; 49:289297.

51. The DREAM (Diabetes REdction Assessment with ramipril androsiglitazone Medication) Trial Investigators. Eect o rosiglitazone onthe reqency o diabetes in patients with impaired glcose toleranceor impaired asting glcose: a randomised controlled trial. Lancet 2006;

368: 1096-1105.

52. Borch-Johnsen K, Colagiri S, Balka B, Glmer C, Carstensen B,Ramachandran A, Dong Y, Gao W. Creating a pandemic o prediabetes:the proposed new diagnostic criteria or impaired asting glycaemia Dia-

betologia 2004; 47: 13961402.

53. Glmer C, Jrgensen T, Borch-Johnsen K. Prevalence o diabetes andimpaired glcose reglation in a Danish poplationThe Inter99 stdy.Diabetes Care 2003; 26: 23352340.

54. Blake DR, Meigs JB, Mller DC, Najjar SS, Andres R, Nathan DM.

Impaired glcose tolerance, bt not impaired asting glcose, is associ-ated with increased levels o coronary heart disease risk actors. Reslts

rom the Baltimore Longitdinal Stdy on Aging. Diabetes 2004, 53:20952100.

55. Phillips LS, Weintrab WS, Ziemer DC, Kolm P, Foster JK, VaccarinoV, Rhee MK, Bdhwani RK, Cadle JM. All pre-diabetes is not the same:

metabolic and vasclar risks o impaired asting glcose at 100 verss110 mg/dl: The Screening or Impaired Glcose Tolerance Stdy 1 (SIGT

1). Diabetes Care 2006; 29: 1405-1407.

56. Fogh-Andersen N, Wimberley PD, Thode J, Siggaard-Andersen O,Direct reading glcose electrodes detect the molality o glcose in plasma

and whole blood. Clin Cham Acta 1990; 189: 33-38

57. DOrazio P, Brnett RW, Fogh-Andersen N, Jacobs E, Kwa K, Klp-

mann WR, Larsson L, Lewenstam AJ, Maas AH, Mager G, Naskalski JW,

Okorodd AO. Approved IFCC Recommendation on Reporting Reslts or

Blood Glcose (Abbreviated). Clinical Chemistry 2005; 51: 15731576.

58. Brrin JM, Alberti KGGM. What is blood glcose: can it be measred?Diabetic Medicine 1990; 7: 199-206.

59. Kanedo T, Wang PY, Tawata M, Sato A. Low carbohydrate intake be-

ore oral glcose tolerance tests. Lancet 1998; 352: 289

60. DECODE Stdy Grop. Will new diagnostic criteria or diabetes mel-lits change phenotype o patients with diabetes? Reanalysis o Eropean

epidemiological data. BMJ 1998; 317: 371-375

61. Barrett-Connor E, Ferrara A: Isolated postchallenge hyperglycemia

and the risk o atal cardiovasclar disease in older women and men: the

Rancho Bernardo Stdy. Diabetes Care 1998; 21: 12361239

references


50/50

62. de Vegt F, Dekker JM, Rhe HG, Stehower CDA, Nijpels G, BoterLM, Heine RJ. Hyperglycaemia is associated with all-case and cardio-

vasclar mortality in the Hoorn poplation: the Hoorn Stdy. Diabetologia

1999; 42: 926-931.

63. DECODE stdy grop. Glcose tolerance and cardiovasclar mortality.Comparison o asting and 2-hor diagnostic criteria. Arch Int Med 2001;

161: 397-404.

64. Shaw JE, Hodge AM, de Corten M, Chitson P, Zimmet PZ: Isolatedpost-challenge hyperglycaemia conrmed as a risk actor or mortality.

Diabetologia 1999; 42:10501054.

65. Eschwege E, Charles MA, Simon D, Thiblt N, Balka B: From po-licemen to policies: what is the tre or 2-h glcose? The Kelly WestLectre, 2000. Diabetes Care 2001; 24: 19451950.

66. McCance DR, Hanson RL, Charles MA, Jacobsson LTH, Pettitt DJ,

Bennett PH, Knowler WC: Comparison o tests or glycated haemoglobinand asting and two hor plasma glcose concentrations as diagnosticmethods or diabetes. BMJ 1994, 308:13231328.

67. Miyazaki M, Kbo M, Kiyohara Y, Okbo K, Nakamra H, FjisawalK, Hata1 Y, Toknaga S, Iida M, Nose Y, Ishibashi T. Comparison o di-agnostic methods or diabetes mellits based on prevalence o retinopa-

thy in a Japanese poplation: the Hisayama Stdy. Diabetologia 2004,47:14111415.

68. Goldstein DE, Little RR, Lorenz RA, Malone JI, Nathan D, PetersonCM, Sacks DB. Tests o glycemia in diabetes. Diabetes Care 2004, 27:

1761-1773.


Definition and Diagnosis 0diabetes

Documents