hemoglobina glicosilada diagnostico diabetes julio 2009

International Expert Committee Report onthe Role of the A1C Assay in the Diagnosisof DiabetesTHE INTERNATIONAL EXPERT COMMITTEE*

An International Expert Committee withmembers appointed by the American Diabe-tes Association, the European Association forthe Study of Diabetes, and the InternationalDiabetes Federation was convened in 2008 toconsider the current and future means of di-agnosing diabetes in nonpregnant individu-als. The report of the International ExpertCommittee represents the consensus view ofits members and not necessarily the view ofthe organizations that appointed them. TheInternational Expert Committee hopes thatits report will serve as a stimulus to the inter-national community and professional organi-zations to consider the use of the A1C assayfor the diagnosis of diabetes.

D iabetes is a disease characterizedby abnormal metabolism, mostnotably hyperglycemia, and an

associated heightened risk for relativelyspecific long-term complications af-fecting the eyes, kidney, and nervoussystem. Although diabetes also substan-tially increases the risk for cardiovas-cular disease, cardiovascular disease isnot specific to diabetes and the risk forcardiovascular disease has not been in-corporated into previous definitions orclassifications of diabetes or of subdia-betic hyperglycemia.

BACKGROUND

Diagnosing diabetes based on thedistribution of glucose levelsHistorically, the measurement of glucosehas been the means of diagnosing diabe-tes. Type 1 diabetes has a sufficientlycharacteristic clinical onset, with rela-tively acute, extreme elevations in glucoseconcentrations accompanied by symp-toms, such that specific blood glucose cutpoints are not required for diagnosis in

most clinical settings. On the other hand,type 2 diabetes has a more gradual onset,with slowly rising glucose levels overtime, and its diagnosis has required spec-ified glucose values to distinguish patho-logic glucose concentrations from thedistribution of glucose concentrations inthe nondiabetic population. Virtually ev-ery scheme for the classification and diag-nosis of diabetes in modern times hasrelied on the measurement of plasma (orblood or serum) glucose concentrationsin timed samples, such as fasting glucose;in casual samples independent of prandialstatus; or after a standardized metabolicstress test, such as the 75-g oral glucosetolerance test (OGTT).

Early attempts to standardize the def-inition of diabetes relied on the OGTT,but the performance and interpretation ofthe test were inconsistent and the numberof subjects studied to define abnormalvalues was very small (1–6). Studies inthe high-risk Pima Indian population thatdemonstrated a bimodal distribution ofglucose levels following the OGTT (7,8)helped establish the 2-h value as the diag-nostic value of choice, even though mostpopulations had a unimodal distributionof glucose levels (9). Of note, a bimodaldistribution was also seen in the fastingglucose samples in the Pimas and otherhigh-risk populations (10,11). However,a discrete fasting plasma glucose (FPG) or2-h plasma glucose (2HPG) level that sep-arated the bimodal distributions in the Pi-mas was difficult to identify, withpotential FPG and 2HPG cut points rang-ing from 120 to 160 mg/dl (6.7– 8.9mmol/l) and from 200 to 250 mg/dl(11.1–13.9 mmol/l), respectively.

In 1979, the National Diabetes DataGroup (NDDG) provided the diagnostic

criteria that would serve as the blueprintfor nearly two decades (12). The NDDGrelied on distributions of glucose levels,rather than on the relationship of glucoselevels with complications, to diagnose di-abetes despite emerging evidence that themicrovascular complications of diabeteswere associated with a higher range offasting and OGTT glucose values (11,13–15). The diagnostic glucose values chosenwere based on their association with de-compensation to “overt” or symptomaticdiabetes.

When selecting the threshold glucosevalues, the NDDG acknowledged that“there is no clear division between diabet-ics and nondiabetics in the FPG concen-tration or their response to an oral glucoseload,” and consequently, “an arbitrary de-cision has been made as to what level jus-tifies the diagnosis of diabetes.” Thediagnosis of diabetes was made when 1)classic symptoms were present; 2) the ve-nous FPG was �140 mg/dl (�7.8 mmol/l); or 3) after a 75-g glucose load, thevenous 2HPG and levels from an earliersample before 2 h were �200 mg/dl(�11.1 mmol/l). An intermediate groupwas classified as having “impaired glucosetolerance” (IGT) with FPG �140 mg/dl(7.8 mmol/l) and a 2HPG value between140 and 200 mg/dl (7.8–11.1 mmol/l).IGT was identified on the basis of its rel-atively higher risk of progression to dia-betes compared with that of “normal”glucose tolerance, low frequency of “dia-betic symptoms,” high probability of re-verting to normal glucose tolerance orcontinuing to have IGT, and rarity of“clinically significant” microvascular dis-ease. The NDDG recommendations werealso promulgated by the contemporane-ous report of the World Health Organiza-tion (WHO) (16).

Diagnosing diabetes based on therelationship between glucose levelsand long-term complicationsIn 1997, the Expert Committee on theDiagnosis and Classification of DiabetesMellitus (17) reexamined the basis for di-agnosing diabetes. This committee madetwo seminal contributions: First, they re-focused attention on the relationship be-

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

Corresponding author: David M. Nathan, [email protected].*A list of members of the International Expert Committee can be found in the APPENDIX.DOI: 10.2337/dc09-9033© 2009 by the American Diabetes Association. Readers may use this article as long as the work is properly

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

See accompanying editorial on p. 1344.

R e v i e w s / C o m m e n t a r i e s / A D A S t a t e m e n t sA D A W O R K G R O U P R E P O R T

DIABETES CARE, VOLUME 32, NUMBER 7, JULY 2009 1327

tween glucose levels and the presence oflong-term complications as the basis forthe diagnosis of diabetes. Second, theysummarized data negating the wide-spread hypothesis that the 2HPG was thegold-standard test for diagnosing diabe-tes. The committee examined data fromthree cross-sectional epidemiologi-cal studies that included an Egyptianpopulation (n � 1,018), Pima Indians(n � 960), and the U.S. National Healthand Nutrition Examination Survey(NHANES) population (n � 2,821). Eachassessed retinopathy with fundus photog-raphy or direct ophthalmoscopy andmeasured glycemia as FPG, 2HPG, andA1C. These studies demonstrated glyce-mic levels below which there was littleprevalent retinopathy and above whichthe prevalence of retinopathy increased inan apparently linear fashion (Fig. 1).When the prevalence of retinopathy wasexpressed by deciles of glycemia for eachof the three measures, the deciles at whichretinopathy began to increase were thesame for each measure within each pop-ulation. Moreover, the glycemic valuesabove which retinopathy increased weresimilar among the populations. Thesedata showed a clear relationship betweenglycemia and the risk for retinopathy thatwould supplant the previous notion ofrisk for progression to overt, symptomaticdiabetes as the basis for diagnosingdiabetes.

In comparing the relationship be-tween FPG and 2HPG values and retinop-athy, it was apparent that the previousFPG cut point of �140 mg/dl (7.8mmol/l) was substantially above the glu-cose level at which the prevalence of ret-inopathy began to increase. As a result,the committee recommended that theFPG cut point be lowered to �126 mg/dl(7.0 mmol/l) so that this cut point wouldrepresent a degree of hyperglycemia thatwas “similar” to the 2HPG value and di-agnosis with either measure would resultin a similar prevalence of diabetes in thepopulation. The 1997 committee reportacknowledged that even at the lower FPGcut point, the FPG and OGTT (2HPG)were not perfectly concordant. An indi-vidual could have diabetes using one testbut not the other. This discrepancy hasbeen confirmed in numerous subsequentreports and may be due, in part, to the factthat although both tests are measures ofglycemia, they reflect different physiolog-ical measures of acute glucose metabolism(18). The debate regarding the relativeroles of FPG and 2HPG in the diagnosis of

diabetes in the nonpregnant adult hascontinued (19–21).

The 1997 report also recommendedthat the FPG level, rather than the 2HPG,be the preferred test to diagnose diabetesbecause it was more convenient for pa-tients and less costly and time consumingand the repeat-test reproducibility wassuperior (17). In addition, the committeeintroduced the term “impaired fastingglucose” (IFG) to differentiate the meta-

bolic state between a normal state (FPG�110 mg/dl or �6.1 mmol/l) and diabe-tes (�126 mg/dl or �7.0 mmol/l) whenthe FPG test was used. If an OGTT wasperformed, the intermediate glycemicstate continued to be called IGT, with the2HPG (between 140 and 200 mg/dl [7.8and 11.1 mmol/l]) the same as that as inthe NDDG report. A WHO consultation(22) adopted most of the above recom-mendations except they concluded that,

Figure 1—Prevalence of retinopathy by deciles of the distribution of FPG, 2HPG, and A1C inPima Indians (A), Egyptians (B), and 40- to 74-year-old participants in NHANES III (C). Adaptedwith permission from ref. 17.

Role of the A1C assay in the diagnosis of diabetes

1328 DIABETES CARE, VOLUME 32, NUMBER 7, JULY 2009

whenever feasible, individuals with IFGshould be given an OGTT to exclude thepresence of diabetes that would otherwisebe missed and that the OGTT should re-main the “gold standard.” A 2003 fol-low-up report from the expert committeerefined the fasting glucose value range forIFG from �110 but �126 mg/dl to �100but �126 mg/dl (�6.1 but �7.0 mmol/lto �5.6 but �7.0 mmol/l) to make itmore comparable with the IGT value(21). The WHO did not change its previ-ous recommendations (23).

CAN THE A1C TEST BE USEDTO DIAGNOSE DIABETES? — Ifchronic hyperglycemia sufficient to causediabetes-specific complications is thehallmark of diabetes, common sensewould dictate that laboratory measuresthat capture long-term glycemic exposureshould provide a better marker for thepresence and severity of the disease thansingle measures of glucose concentration.Observational studies that have assessedglycemia with measures that capturelonger-term exposure (i.e., A1C) or withsingle or longitudinal measurements ofglucose levels have consistently demon-strated a strong correlation between reti-nopathy and A1C (24 –26) but a lessconsistent relationship with fasting glu-cose levels (27). In one study that mea-sured both FPG and A1C, there was astronger correlation between A1C andretinopathy than between fasting glucoselevels and retinopathy (25). The correla-tion between A1C levels and complica-tions has also been shown in the setting ofcontrolled clinical trials in type 1 (28) andtype 2 (29) diabetes, and these findingshave been used to establish the widely ac-cepted A1C treatment goals for diabetescare (30).

All of these observations suggest thata reliable measure of chronic glycemiclevels such as A1C, which captures thedegree of glucose exposure over time(31,32) and which is related more inti-mately to the risk of complications thansingle or episodic measures of glucose lev-els, may serve as a better biochemicalmarker of diabetes and should be consid-ered a diagnostic tool. Although the 1997expert committee report considered thisoption, it recommended against usingA1C values for diagnosis in part becauseof the lack of assay standardization (17).The 2003 follow-up report noted that,while the National GlycohemoglobinStandardization Program (33) had suc-ceeded in standardizing the vast majority

of assays used in the U.S., the use of A1Cfor diagnosis still had “disadvantages,”and it reaffirmed the previous recommen-dation that A1C not be used to diagnosediabetes (21).

An updated examination of the labo-ratory measurements of glucose and A1Cby the current International Expert Com-mittee indicates that with advances in in-strumentation and standardization, theaccuracy and precision of A1C assays atleast match those of glucose assays. Themeasurement of glucose itself is less accu-rate and precise than most clinicians real-ize (34). A recent analysis of the per-formance of a variety of clinical laboratoryinstruments and methods that measureglucose revealed that 41% of instrumentshave a significant bias from the referencemethod that would result in potentialmisclassification of �12% of patients(35). There are also potential preana-lytic errors owing to sample handling andthe well-recognized lability of glucose inthe collection tube at room temperature(36,37). Even when whole blood samplesare collected in sodium fluoride to inhibitin vitro glycolysis, storage at room tem-perature for as little as 1 to 4 h beforeanalysis may result in decreases in glucoselevels by 3–10 mg/dl in nondiabetic indi-viduals (36–39).

By contrast, A1C values are relativelystable after collection (40), and the recentintroduction of a new reference methodto calibrate all A1C assay instrumentsshould further improve A1C assay stan-dardization in most of the world (41–43).In addition, between- and within-subjectcoefficients of variation have been shownto be substantially lower for A1C than forglucose measurements (44). The variabil-ity of A1C values is also considerably lessthan that of FPG levels, with day-to-daywithin-person variance of �2% for A1Cbut 12–15% for FPG (45–47). The con-venience for the patient and ease of sam-ple collection for A1C testing (which canbe obtained at any time, requires no pa-tient preparation, and is relatively stableat room temperature) compared with thatof FPG testing (which requires a timedsample after at least an 8-h fast and whichis unstable at room temperature) supportusing the A1C assay to diagnose diabetes.

In summary, compared with the mea-surement of glucose, the A1C assay is atleast as good at defining the level of hy-perglycemia at which retinopathy preva-lence increases; has appreciably superiortechnical attributes, including less pre-analytic instability and less biologic vari-

ability; and is more clinically convenient.A1C is a more stable biological index thanFPG, as would be expected with a mea-sure of chronic glycemia levels comparedwith glucose concentrations that areknown to fluctuate within and betweendays (Table 1).

WHAT IS THE MOSTAPPROPRIATE A1C CUTPOINT FOR THE DIAGNOSISOF DIABETES? — As shown in the1997 committee report, the prevalence ofretinopathy increases substantially atA1C values starting between 6.0 and7.0% (17) (Fig. 1). A recent analysis de-rived from DETECT-2 (48) and includingthe 3 that were included in the 1997report examined the association betweenA1C and retinopathy, objectivelyassessed and graded by fundus photogra-phy (S. Colagiuri, personal communica-tion). This analysis included �28,000subjects from nine countries and showedthat the glycemic level at which the prev-alence of “any” retinopathy begins to riseabove background levels (any retinopathyincludes minor changes that can be due toother conditions, such as hypertension),and for the more diabetes-specific “mod-erate” retinopathy, was 6.5% when thedata were examined in 0.5% increments(Fig. 2). Among the �20,000 subjectswho had A1C values �6.5%, “moderate”retinopathy was virtually nonexistent.The receiver operating characteristiccurve analysis of the same data indicatedthat the optimal cut point for detecting atleast moderate retinopathy was an A1C of6.5%.

In summary, the large volume of datafrom diverse populations has now estab-lished an A1C level associated with an in-crease in the prevalence of moderate

Table 1—Advantages of A1C testing com-pared with FPG or 2HPG for the diagnosis ofdiabetes

● Standardized and aligned to the DCCT/UKPDS; measurement of glucose is lesswell standardized

● Better index of overall glycemic exposureand risk for long-term complications

● Substantially less biologic variability● Substantially less preanalytic instability● No need for fasting or timed samples● Relatively unaffected by acute (e.g., stress

or illness related) perturbations in glucoselevels

● Currently used to guide management andadjust therapy

International Expert Committee


retinopathy and provides strong justifica-tion for assigning an A1C cut point of�6.5% for the diagnosis of diabetes. Arecently published population-basedstudy of 3,190 adults of Malay ethnicityindependently concluded that A1C levels“in the range 6.6 to 7% were optimal fordetecting microvascular complications”(26).

Any suggestion that the relationshipbetween chronic glycemic levels and thelong-term complications of diabetes maybe better expressed as a continuum,rather than as a strictly dichotomous re-lationship, is belied by the retinopathyfindings presented herein. There is a lowprevalence of “any” retinopathy at A1Clevels �6.5% that may reflect a contin-uum of risk; alternatively, retinopathy re-lated to conditions other than diabetes(e.g., hypertension) or inaccurate assess-ment of long-term glycemic levels with asingle A1C measurement may contributeto this observation. However, the sub-stantial increase in the prevalence of mod-erate retinopathy at A1C levels �6.5%supports a threshold level of glycemia thatresults in retinopathy most characteristicof diabetes.

This cut point should not be con-strued as an absolute dividing line be-tween normal glycemia and diabetes;however, the A1C level of 6.5% is suffi-ciently sensitive and specific to identifyindividuals who are at risk for developingretinopathy and who should be diag-nosed as diabetic. The A1C level is at leastas predictive as the current FPG and2HPG values. In selecting a diagnosticA1C level �6.5%, the International Ex-pert Committee balanced the stigma andcosts of mistakenly identifying individu-als as diabetic against the minimal clinical

consequences of delaying the diagnosis insomeone with an A1C level �6.5%. Thecommittee agreed to emphasize specific-ity rather than sensitivity. This decisionwas aided by the parallel decision to rec-ommend effective prevention strategiesfor the highest at-risk group with an A1Cbetween 6.0 and 6.5%. (See below.)

LIMITATIONS OF A1C ASTHE RECOMMENDEDMEANS OF DIAGNOSINGDIABETES — The A1C assay is thetest of choice for the chronic managementof diabetes and is now being recom-mended for its diagnosis; however, thereare parts of the world where the costs ofproviding the assay preclude its routineuse. In such circumstances, cliniciansshould continue to use the previously rec-ommended approaches to diagnose dia-betes based on glucose measurements.The International Expert Committee en-courages clinicians worldwide to move asquickly as possible to A1C testing usingstandardized methods and instrumenta-tion. However, the decision to change toA1C assays as the means of diagnosingdiabetes should take into account the per-formance of local A1C assays and the localprevalence of conditions that may inter-fere with the assay. (See below.)

Although the discussion above arguesfor using the A1C assay for the diagnosisof diabetes in nonpregnant individuals,there are patient conditions that eitherwill require a specific A1C assay methodor will preclude A1C testing. First, somehemoglobin traits, such as HbS, HbC,HbF, and HbE, interfere with some A1Cassay methods (49). Currently, many as-say methods can correct for the presenceof the most common hemoglobin traits

(www.ngsp.org), and affinity assays thatare unaffected by hemoglobin traits maybe used (49). Second, any condition thatchanges red cell turnover, such as hemo-lytic anemia, chronic malaria, majorblood loss, or blood transfusions, willlead to spurious A1C results. Cliniciansmust be aware of these conditions, partic-ularly in populations in which they aremore prevalent. As in the setting whereA1C assays are unavailable, the tradi-tional diagnostic tests (e.g., FPG, 2HPG)must be used in individuals in whom in-terpreting the A1C is problematic. Third,A1C levels appear to increase with age(50), but the extent of the change,whether it relates to factors other thanglucose metabolism, and the effect of theage-related increases on the developmentof complications are not sufficiently clearto adopt age-specific values in a diagnos-tic scheme. Similarly, racial disparities inA1C, based on putative differences in therelationship between glucose levels andA1C, have been suggested (51); however,here too, their etiology and significanceare unclear, and it is premature to estab-lish race-specific diagnostic values. Fi-nally, there are rare clinical settings, suchas rapidly evolving type 1 diabetes, wherethe A1C level will not have had time to“catch up” with the acute elevations inglucose levels; however, in these very rarecases, diabetes should be diagnosablewith typical symptoms and casual glucoselevels �200 mg/dl (11.1 mmol/l) despitea nondiagnostic A1C level.

Notwithstanding the above limita-tions of A1C testing, the assay has numer-ous important advantages compared withthe currently used laboratory measure-ments of glucose (Table 1). The preva-lence of diabetes in some populations

Figure 2—Prevalence of retinopathy by 0.5% intervals and severity of retinopathy in participants aged 20–79 years. NPDR, nonproliferativediabetic retinopathy. Adapted with permission from (S. Colagiuri, personal communication).



may not be the same when diagnosis isbased on A1C compared with diagnosiswith glucose measurements, and onemethod may identify different individualsthan the other. Because the measurementsof glucose levels and A1C reflect differentaspects of glucose metabolism, this is tobe expected. However, establishing iden-tical prevalences should not be the goal indefining a new means of diagnosing dia-betes. The ultimate goal is to identify in-dividuals at risk for diabetes compli-cations so that they can be treated. TheA1C diagnostic level of 6.5% accom-plishes this goal.

CAN A1C MEASUREMENTSDEFINE A SPECIFICSUBDIABETIC “HIGH-RISK” STATE? — The 2003 Inter-national Expert Committee report re-duced the lower bound of IFG from 110mg/dl (6.1 mmol/l) to 100 mg/dl (5.6mmol/l) on the grounds that the lowerlevel optimized the sensitivity and speci-ficity for predicting future diabetes andalso increased the proportion of thosewith IGT who could be identified with anFPG test (21). While previous studieshave shown a powerful effect of IFGand/or IGT on the subsequent develop-ment of diabetes diagnosed with glucosevalues (52–54), recent reports have dem-onstrated a graded risk of diabetes devel-opment at glycemic levels well withinwhat was previously considered “nor-mal,” i.e., FPG �100 mg/dl (5.6 mmol/l)and A1C �6.0% (55,56). In addition,metabolic derangements related to diabe-tes have been documented at similarlylow glycemic levels, increasing in severitywith higher glucose values within thenondiabetic range (57,58).

As with measures of glucose, a con-tinuum of risk for the development of di-abetes based on A1C levels has beendemonstrated (59–61). Thus, while thereappears to be an approximate glycemicthreshold above which the risk for reti-nopathy escalates, there does not appearto be a specific level at which risk for di-abetes clearly begins. A continuum of riskfor the development of diabetes across awide range of subdiabetic A1C levels maymake the classification of individuals intocategories similar to IFG and IGT equallyproblematic for A1C, as it implies that weactually know where risk begins or be-comes clinically important. The contin-uum of risk in the subdiabetic glycemicrange argues for the elimination of dichot-omous subdiabetic classifications, such as

“pre-diabetes,” IFG, and IGT. However,as A1C levels approach the diagnosticlevel for diabetes, the risk of developingdiabetes becomes greatest (59,60,62).

SHOULD A1C TESTING BEUSED TO IDENTIFYINDIVIDUALS AT HIGH RISKFOR DIABETES? — The screeningtests to identify individuals at elevatedrisk for diabetes are the same as the diag-nostic tests; therefore, the technical ad-vantages of A1C testing compared withglucose testing apply to the detection ofindividuals at high risk as they do to thediagnosis of diabetes. Therapeutic deci-sions should be based on how close A1Clevels are to the diagnosis of diabetes. Inthe absence of a specific identifiable lowerthreshold defining when prevention ef-forts should be implemented, and withpotentially limited resources taken intoconsideration, individuals whose A1C val-ues are close to the 6.5% A1C threshold ofdiabetes (i.e., �6.0%) should receive de-monstrably effective interventions (63,64).By identifying this very high-risk subdia-betic group, the International ExpertCommittee is implying not that popula-tions at lower A1C levels are not at riskbut, rather, that they are at lower risk. Allindividuals at risk for diabetes should re-ceive counseling to maintain normalweight, lose weight if necessary, and be-come more physically active.

Other risk factors for diabetes devel-opment in addition to A1C have beenidentified, including elevated levels oftriglycerides, blood pressure, BMI, andfamily history of diabetes (59,60), andthese should be taken into account in de-termining when to initiate interventionsin individuals with A1C �6.0%. The useof well-validated risk assessment toolsmay be valuable in that regard. At thepopulation level, the A1C value at whichprevention services are provided will de-pend on the resources available, the sizeof the target population, and the nature ofthe intervention.

WHAT ARE THE PRACTICALISSUES RELATED TO A1CTESTING? — A1C tests to diagnosediabetes should be performed using clin-ical laboratory equipment. Point-of-careinstruments have not yet been shown tobe sufficiently accurate or precise for di-agnosing diabetes. Although this Interna-tional Expert Committee has concludedthat the attributes of the A1C assay with

regard to diagnosing diabetes and detect-ing individuals at high risk support its useover the FPG or 2HPG tests, the superior-ity of A1C testing does not invalidate thediagnostic criteria based on glucose test-ing. In circumstances when A1C testingcannot be performed, the diagnostic glu-cose tests are acceptable alternatives.

Whichever of the three different testsnow available to diagnose diabetes (A1C,FPG, and 2HPG) is used, both initial andconfirmatory testing should be performedwith the same test. As the three tests arenot completely concordant, using differ-ent tests could easily lead to confusion.The only exception to the need to confirmthe diagnosis of diabetes with the sametest would be the presence of clinicalsymptoms characteristic of diabetes andglucose levels �200 mg/dl (�11.1 mmol/l). Confirmatory testing is also not requir-ed to establish risk status in individualsidentified as in the highest-risk group fordiabetes (A1C of 6.0 to �6.5%).

Most cases of type 1 diabetes, partic-ularly in children and adolescents, are di-agnosed by the classical symptoms ofpolyuria, polydipsia, polyphagia, unex-plained weight loss, and a casual glucose�200 mg/dl. If diabetes is suspected inthe absence of those conditions, A1C test-ing is warranted.

RECOMMENDATIONS ANDCONCLUSIONS — Based on the abovediscussion, the International Expert Com-mittee has concluded that the best currentevidence supports the following recom-mendations, summarized in Table 2.

For the diagnosis of diabetes● There is no single assay related to hy-

perglycemia that can be consideredthe gold standard, as it relates to therisk for microvascular or macrovascularcomplications.

● A measure that captures chronic glu-cose exposure is more likely to be infor-mative regarding the presence ofdiabetes than is a single measure ofglucose.

● The A1C assay provides a reliable mea-sure of chronic glycemia and correlateswell with the risk of long-term diabetescomplications.

● The A1C assay (standardized and alignedwith the Diabetes Control and Complica-tions Trial/UK Prospective DiabetesStudy assay) has several technical, in-cluding preanalytic and analytic, advan-



tages over the currently used laboratorymeasurements of glucose.

● For the reasons above, the A1C assaymay be a better means of diagnosingdiabetes than measures of glucoselevels.

● The diagnosis of diabetes is made if theA1C level is �6.5%. Diagnosis shouldbe confirmed with a repeat A1C test un-less clinical symptoms and glucose lev-els �200 mg/dl (�11.1 mmol/l) arepresent.

● If A1C testing is not possible owing topatient factors that preclude its inter-pretation (e.g., hemoglobinopathy orabnormal erythrocyte turnover) or tounavailability of the assay, previouslyrecommended diagnostic measures(e.g., FPG and 2HPG) and criteriashould be used. Mixing different meth-ods to diagnose diabetes should beavoided.

● In children and adolescents, A1C test-ing is indicated when diabetes is sus-pected in the absence of the classi-cal symptoms or a plasma glucoseconcentration �200 mg/dl (�11.1mmol/l).

● The diagnosis of diabetes during preg-nancy, when changes in red cell turn-over make the A1C assay problema-tic, will continue to require glucosemeasurements.

For the identification of individualsat high risk for diabetes● Individuals with an A1C level �6% but

�6.5% are likely at the highest risk forprogression to diabetes, but this rangeshould not be considered an absolutethreshold at which preventative mea-sures are initiated.

● The classification of subdiabetic hyper-glycemia as pre-diabetes is problematicbecause it suggests that all individualsso classified will develop diabetes andthat individuals who do not meet theseglycemia-driven criteria (regardless ofother risk factor values) are unlikely todevelop diabetes—neither of which isthe case. Moreover, the categorical clas-sification of individuals as high risk(e.g., IFG or IGT) or low risk, based onany measure of glycemia, is less thanideal because the risk for progression todiabetes appears to be a continuum.The glucose-related terms describingsubdiabetic hyperglycemia will bephased out of use as clinical diagnosticstates as A1C measurements replaceglucose measurements for the diagnosisof diabetes.

● When assessing risk, implementingprevention strategies, or initiating apopulation-based prevention program,other diabetes risk factors should betaken into account. In addition, theA1C level at which to begin preventa-

tive measures should reflect the re-sources available, the size of thepopulation affected, and the antici-pated degree of success of the interven-tion. Further analyses of cost-benefitshould guide the selection of high-riskgroups targeted for intervention withinspecific populations.

Acknowledgments— No potential conflictsof interest relevant to this article werereported.

APPENDIX — International ExpertCommittee members: David M. Nathan,MD (Chair); Beverly Balkau, PhD; EnzoBonora, MD, PhD; Knut Borch-Johnsen,MD, DMSc; John B. Buse, MD, PhD; Ste-phen Colagiuri, MD; Mayer B. Davidson,MD; Ralph DeFronzo, MD; Saul Genuth,MD; Rury R. Holman, FRCP; Linong Ji,MD; Sue Kirkman, MD; William C.Knowler, MD, Dr PH; Desmond Schatz,MD; Jonathan Shaw, MD; EugeneSobngwi, MD; Michael Steffes, MD, PhD;Olga Vaccaro, MD; Nick Wareham, MD;Bernard Zinman, MD; and Richard Kahn,PhD.

References1. Moyer JH, Womack CR. Glucose toler-

ance: comparison of four types of diag-nostic tests in 103 control subjects and 26patients with diabetes. Am J Med 1950;219:161–173

2. Mosenthal HO, Barre E. Criteria for and in-terpretation of normal glucose tolerancetest. Ann Intern Med 1950;33:1175–1194

3. Fajans SS, Conn JW. The early recogni-tion of diabetes mellitus. Ann N Y AcadSci 1959;82:208–218

4. Hayner NS, Kjelsberg MD, Epstein FH,Francis T. Carbohydrate tolerance and di-abetes in a total community, Tecumseh,Michigan: effects of age, sex, and test con-ditions on one-hour glucose tolerance inadults. Diabetes 1965;14:413–423

5. Sisk CW, Burnham CE, Steward J, Mc-Donald GW. Comparison of the 50 and100 gram oral glucose tolerance test. Di-abetes 1970;19:852–862

6. West KM. Substantial differences in thediagnostic criteria used by diabetes ex-perts. Diabetes 1975;24:641–644

7. Rushforth NB, Bennett PH, Steinberg AG,Burch TA, Miller M. Diabetes in the PimaIndians: evidence of bimodality in glucosetolerance distributions. Diabetes 1971;20:756–765

8. Rushforth NB, Bennett PH, Steinberg AG,Miller M. Comparison of the value of thetwo- and one-hour glucose levels of theoral GTT in the diagnosis of diabetes inPima Indians. Diabetes 1975;24:538–546

Table 2—Recommendation of the International Expert Committee

For the diagnosis of diabetes:● The A1C assay is an accurate, precise measure of chronic glycemic levels and correlates well

with the risk of diabetes complications.● The A1C assay has several advantages over laboratory measures of glucose.● Diabetes should be diagnosed when A1C is �6.5%. Diagnosis should be confirmed with a

repeat A1C test. Confirmation is not required in symptomatic subjects with plasma glucoselevels �200 mg/dl (�11.1 mmol/l).

● If A1C testing is not possible, previously recommended diagnostic methods (e.g., FPG or2HPG, with confirmation) are acceptable.

● A1C testing is indicated in children in whom diabetes is suspected but the classic symptomsand a casual plasma glucose �200 mg/dl (�11.1 mmol/l) are not found.

For the identification of those at high risk for diabetes:● The risk for diabetes based on levels of glycemia is a continuum; therefore, there is no lower

glycemic threshold at which risk clearly begins.● The categorical clinical states pre-diabetes, IFG, and IGT fail to capture the continuum of

risk and will be phased out of use as A1C measurements replace glucose measurements.● As for the diagnosis of diabetes, the A1C assay has several advantages over laboratory

measures of glucose in identifying individuals at high risk for developing diabetes.● Those with A1C levels below the threshold for diabetes but �6.0% should receive

demonstrably effective preventive interventions. Those with A1C below this range may stillbe at risk and, depending on the presence of other diabetes risk factors, may also benefitfrom prevention efforts.

● The A1C level at which population-based prevention services begin should be based on thenature of the intervention, the resources available, and the size of the affected population.



9. Gorden T. Glucose Tolerance of Adults. Se-ries 11, No. 2. National Center for HealthStatistics, U.S. Public Health Service,1964

10. Zimmet P, Whitehouse S. Bimodality offasting and two-hour glucose tolerancedistributed in a Micronesian population.Diabetes 1978;27:793–800

11. Rushforth NB, Miller M, Bennett PH.Fasting and two-hour post-load glucoselevels for the diagnosis of diabetes: therelationship between glucose levels andcomplications of diabetes in the Pima In-dians. Diabetologia 1979;16:373–379

12. National Diabetes Data Group. Classifica-tion and diagnosis of diabetes mellitusand other categories of glucose intoler-ance. Diabetes 1979;28:1039–1057

13. Jarrett RJ, Keen H. Hyperglycemia and dia-betes mellitus. Lancet 1976;1:1009–1011

14. Sayegh HAI, Jarrett RJ. Oral glucose toler-ance tests and the diagnosis of diabetes: re-sults of a prospective study based on theWhitehall Survey. Lancet 1979;2:431–433

15. Pettitt DJ, Knowler WC, Lisse, BennettPH. Development of retinopathy and pro-teinuria in relation to plasma-glucoseconcentrations in Pima Indians. Lancet1980;2:1050–1052

16. World Health Organization. World HealthOrganization Expert Committee on DiabetesMellitus: Second Report. Geneva, WorldHealth Org., 1980 (Tech. Rep. Ser., no.646)

17. The Expert Committee on the Diagnosisand Classification of Diabetes Mellitus.Report of the Expert Committee on theDiagnosis and Classification of DiabetesMellitus. Diabetes Care 1997;20:1183–1197

18. Abdul-Ghani MA, Jenkinson C, Richard-son D, Tripathy D, DeFronzo RA. Insulinsecretion and insulin action in subjectswith impaired fasting glucose and im-paired glucose tolerance: results from theVeterans Administration Genetic Epide-miology Study. Diabetes 2006;55:1430–1435

19. Tuomilehto J. Point: a glucose tolerancetest is important for clinical practice. Di-abetes Care 2002;25:1880–1882

20. Davidson MB. Counterpoint: the oral glu-cose tolerance test is superfluous. Diabe-tes Care 2002;25:1883–1885

21. The Expert Committee on the Diagnosisand Classification of Diabetes Mellitus.Follow-up report on the diagnosis of dia-betes mellitus. Diabetes Care 2003;26:3160–3167

22. World Health Organization. Definition,Diagnosis and Classification of Diabetes Mel-litus and its Complications: Report of a WHOConsultation. Part 1. Diagnosis and Classifi-cation of Diabetes Mellitus. Geneva, WorldHealth Org., 1999

23. World Health Organization. Definitionand Diagnosis of Diabetes Mellitus and Inter-mediate Hyperglycemia: Report of a WHO/

IDF Consultation. Geneva, World HealthOrg., 2006

24. van Leiden HA, Dekker JM, Moll AC. Riskfactors for incident retinopathy in a dia-betic and nondiabetic population: theHoorn study. Arch Ophthalmol 2003;121:245–251

25. Tapp RJ, Tikellis G, Wong TY, Harper C,Zimmet PZ, Shaw JE. Longitudinal asso-ciation of glucose metabolism with reti-nopathy. Diabetes Care 2008;31:1349–1354

26. Sabanayagam C, Liew G, Tai ES, ShankarA, Lim SC, Subramaniam T, Wong TY.Relationship between glycated hemoglo-bin and microvascular complications: isthere a natural cut-off point for the diag-nosis of diabetes. Diabetologia. 22 April2009 [Epub ahead of print]

27. Wong TY, Liew G, Tapp RJ, Schmidt MI,Wang JJ, Mitchell P, Klein R, Klein BEK,Zimmet P, Shaw J. Relation between fastingglucose and retinopathy for diagnosis ofdiabetes: three population based cross-sectional studies. Lancet 2008;371:736–743

28. DCCT Research Group. The associationbetween glycemic exposure and long-term diabetes complications in the Diabe-tes Control and Complications Trial.Diabetes 1995;44:968–983

29. Stratton IM, Adler AI, Neil HA, MatthewsDR, Manley SE, Cull CA, Hadden D,Turner RC, Holman RR. Association ofglycaemia with macrovascular and micro-vascular complications of type 2 diabetes:prospective observational study (UKPDS35). BMJ 2000;321:405–412

30. Nathan DM, Buse JB, Davidson MB, HeineRJ, Holman RR, Sherwin R, Zinman B.Management of hyperglycemia in type 2diabetes: a consensus algorithm for theinitiation and adjustment of therapy. Dia-betologia 2009;52:17–30

31. Nathan DM, Turgeon H, Regan S. Rela-tionship between glycated haemoglobinlevels and mean glucose levels over time.Diabetologia 2007;50:2239–2244

32. Nathan DM, Kuenen J, Borg R, Zheng H,Schoenfeld D, Heine RJ, the A1c-DerivedAverage Glucose (ADAG) Study Group.Translating the A1C assay into estimatedaverage glucose values. Diabetes Care2008;31:1473–1478

33. Little RR, Rohlfing CL, Wiedmeyer HM,Myers GL, Sacks DB, Goldstein DE. Thenational glycohemoglobin standardiza-tion program: a five year progress report.Clin Chem 2001;47:1985–1992

34. Gambino R. Glucose: a simple moleculethat is not simple to quantify. Clin Chem2007;53:2040–2041

35. Miller WG, Myers GL, Ashwood ER,Killeen AA, Wang E, Ehlers GW, Hasse-mer D, Lo SF, Secombe D, Siekmann L,Thienpont LM. State of the art in truenessand interlaboratory harmonization for 10

analytes in general clinical chemistry.Arch Pathol Lab Med 2008;132:838–846

36. Lin YL, Smith CH, Dietzler DN. Stabiliza-tion of blood glucose by cooling with ice:an effective procedure for preservation ofsamples from adults and newborns. ClinChem 1976;22:2031–2033

37. Murphy JM, Browne RW, Hill L, BolelliGF, Abagnato C, Berrino F, FreudenheimJ, Trevisan M, Muti P. Effects of transpor-tation and delay in processing on thestability of nutritional and metabolic bi-omarkers. Nutr Cancer 2000;37:155–160

38. Gambino R, Piscitelli J, Ackattupathil TA,Theriault JL, Andrin RD, Sanfilippo ML,Etienne M. Acidification of blood issuperior to sodium fluoride alone as aninhibitor of glycolysis. Clin Chem 2009;55:1019–1021

39. Bruns DE, Knowler WC. Stabilization ofglucose in blood samples: why it matters.Clin Chem 2009;55:850–852

40. Little RR, Rohlfing CL, Tennill AL, Con-nolly S, Hanson S. Effects of sample stor-age conditions on glycated hemoglobinmeasurement: evaluation of five differenthigh performance liquid chromatographymethods. Diabetes Technol Ther 2007;9:36–42

41. Hoelzel W, Weykamp C, Jeppsson JO,Miedema K, Barr JR, Goodall I, HoshinoT, John WG, Kobold U, Little R, Mosca A,Mauri P, Paroni R, Susanto F, Takei I,Theinpont L, Umemoto M, WiedmeyerHM, the IFCC Working Group on HbA1cStandardization. IFCC reference systemfor measurement of hemoglobin A1C inhuman blood and the national standard-ization schemes in the United States, Ja-pan, and Sweden: a method-comparisonstudy. Clin Chem 2004;50:166–174

42. Consensus Committee. Consensus state-ment on the worldwide standardizationof the hemoglobin A1C measurement:American Diabetes Association, EuropeanAssociation for the Study of Diabetes, In-ternational Federation of Clinical Chem-istry and Laboratory Medicine, and theInternational Diabetes Federation. Diabe-tes Care 2007;30:2399–2400

43. Weykamp C, John WG, Mosca A,Hoshino T, Little R, Jeppsson J-O, Good-all I, Miedema K, Myers G, Reinauer H,Sacks DB, Slingerland R, Siebelder C. TheIFCC reference measurement system forHbA1c: a 6-year progress report. ClinChem 2008;54:240–248

44. Rohlfing C, Wiedmeyer HM, Little R,Grotz VL, Tennill A, England J, Madsen R,Goldstein D. Biological variation of glyco-hemoglobin. Clin Chem 2002;48:1116–1118

45. Ollerton RL, Playle R, Ahmed K, DunstanFD, Luzio SD, Owens DR. Day-to-dayvariability of fasting plasma glucose innewly diagnosed type 2 diabetic subjects.Diabetes Care 1999;22:394–398



46. Sacks DB, Bruns DE, Goldstein DE, Ma-claren NK, McDonald JM, Parrott M.Guidelines and recommendations for lab-oratory analysis in the diagnosis and man-agement of diabetes mellitus. Clin Chem2002;48:436–472

47. Petersen PH, Jorgensen LG, Brandslund I,Olivarius DF, Stahl M. Consequences ofbias and imprecision in measurements ofglucose and HbA1c for the diagnosis andprognosis of diabetes mellitus. ScandJ Clin Lab Invest Suppl 2005;240:51–60

48. Colagiuri S, Borch-Johnsen K. DETECT-2:early detection of type 2 diabetes and IGT.Diabetes Voice 2003;48:11–13

49. Roberts WL, Safa-Pour S, De BK, RohlfingCL, Weykamp CW, Little RR. Effects ofhemoglobin C and S traits on glycohemo-globin measurements by eleven methods.Clin Chem 2005;51:776–778

50. Pani LN, Korenda L, Meigs JB, Driver C,Chamany S, Fox CS, Sullivan L, D’AgostinoRB, Nathan DM. Effects of aging on A1Clevels in individuals without diabetes. Dia-betes Care 2008;31:1991–1996

51. Herman WH, Ma Y, Uwaifo G, Haffner S,Kahn SE, Horton ES, Lachin JM, MontezMG, Brenneman T, Barrett-Conner E, theDiabetes Prevention Program ResearchGroup. Differences in A1C by race andethnicity among patients with impairedglucose tolerance in the Diabetes Preven-tion Program. Diabetes Care 2007;30:2756–2758

52. Gabir M, Hanson RL, Dabelea D, Impera-tore G, Roumain J, Bennett PH, KnowlerWC. The 1997 American Diabetes Asso-ciation and 1999 World Health Organiza-

tion criteria for hyperglycemia in thediagnosis and prediction of diabetes mel-litus. Diabetes Care 2000;23:1108–1112

53. Meigs JB, Muller DC, Nathan DM, BlakeDR, Andres R. The natural history of pro-gression from normal glucose tolerance totype 2 diabetes in the Baltimore Longitu-dinal Study of Aging. Diabetes 2003;52:1475–1484

54. Bonora E, Kiechl S, Willeit J, Oberhollen-zer F, Egger G, Meigs JB, Bonadonna RC,Muggeo M. Population-based incidencerates and risk factors for type 2 diabetes incaucasians: the Bruneck Study. Diabetes2004;53:1782–1789

55. Tirosh A, Shai I, Tekes-Manova DT, Is-raeli E, Pereg D, Shochat T, Kochba I, Ru-dich A. Normal fasting plasma glucoselevels and type 2 diabetes in young men.N Engl J Med 2005;353:1454–1462

56. Nichols GA, Hillier TA, Brown JB. Normalfasting plasma glucose and risk of type 2diabetes diagnosis. Am J Med 2008;121:519–524

57. Piche M-E, Arcand-Bosse J-F, Despres J-P,Peruse L, Lemieux S, Weisnagel SJ. Whatis a normal glucose value? Differences inindexes of plasma glucose homeostasis insubjects with normal fasting glucose. Di-abetes Care 2004;27:2470–2477

58. Meigs JB, Nathan DM, Wilson PWF,Cupples LA, Singer DE. Metabolic riskfactors worsen continuously across thespectrum of nondiabetic glucose toler-ance. The Framingham Offspring Study.Ann Intern Med 1998;128:524–533

59. Droumaguet C, Balkau B, Simon D, CacesE, Tichet J, Charles MA, Eschwege E, the

DESIR Study Group. Use of HbA1c in pre-dicting progression to diabetes in Frenchmen and women: data from an Epidemi-ological Study on the Insulin ResistanceSyndrome (DESIR). Diabetes Care 2006;29:1619–1625

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

61. Little RR, England JD, Wiedemeyer HM,Madsen RW, Pettitt DJ, Knowler WC,Goldstein DE. Glycated haemoglobinpredicts progression to diabetes melli-tus in Pima Indians with impaired glu-cose tolerance. Diabetologia 1994;37:252–256

62. Ko GT, Chan JC, Tsang LW, Cockram CS.Combined use of fasting plasma glucoseand HbA1c predicts the progression to di-abetes in Chinese subjects. Diabetes Care2000;23:1770–1773

63. Tuomilehto J, Lindstrom J, Eriksson JG,Valle TT, Hamalainen H, Ilanne-ParikkaP, Keinanen-Kiukaanniemi S, Laakso M,Louheranta A, Rastas M, Salminen V,Uusitupa M; Finnish Diabetes PreventionStudy Group. Prevention of type 2 diabe-tes mellitus by changes in lifestyle amongsubjects with impaired glucose tolerance.N Engl J Med 2001;344:1343–1350

64. Knowler WC, Barrett-Connor E, FowlerS, Hamman R, Lachin J, Walker E, NathanDM, the DPP Research Group. Reductionin incidence of type 2 diabetes with life-style intervention or metformin. N EnglJ Med 2002;346:393–403



hemoglobina glicosilada diagnostico diabetes julio 2009

Documents