The Effect of Glucose Variability on theRisk of Microvascular Complications inType 1 DiabetesERIC S. KILPATRICK, MD, FRCPATH1

ALAN S. RIGBY, MSC2

STEPHEN L. ATKIN, PHD, FRCP3

OBJECTIVE It is not known whether glycemic instability may confer a risk of microvas-cular complications that is in addition to that predicted by the mean blood glucose (MBG) valuealone. This study has analyzed data from the Diabetes Control and Complications Trial (DCCT)to assess the effect of glucose variability on the risk of retinopathy and nephropathy in patientswith type 1 diabetes.

RESEARCH DESIGN AND METHODS Pre- and postprandial seven-point glucoseprofiles were collected quarterly during the DCCT in 1,441 individuals. The mean area under thecurve glucose and the SD of glucose variability within 24 h and between visits were comparedwith the risk of retinopathy and nephropathy, having adjusted for age, sex, disease duration,treatment group, prevention cohort, and phase of treatment.

RESULTS Multivariate Cox regression showed that within-day and between-day variabilityin blood glucose around a patients mean value has no influence on the development or pro-gression of either retinopathy (P 0.18 and P 0.72, respectively) or nephropathy (P 0.32and P 0.57). Neither preprandial (P 0.18) nor postprandial (P 0.31) glucose concentra-tions preferentially contribute to the probability of retinopathy.

CONCLUSIONS This study has shown that blood glucose variability does not appear tobe an additional factor in the development of microvascular complications. Also, pre- andpostprandial glucose values are equally predictive of the small-vessel complications of type 1diabetes.

Diabetes Care 29:14861490, 2006

I t is well established that the risk of de-veloping the microvascular complica-tions of diabetes is intimately related tothe glycemic control of an individual.Having determined that HbA1c (A1C)could be used as a surrogate marker forglycemia (1), both the Diabetes Controland Complications Trial (DCCT) in type1 diabetes and the U.K. Prospective Dia-betes Study in type 2 diabetes confirmedan exponential relationship between ris-ing blood glucose and the risk of either

developing or worsening retinopathy, ne-phropathy, and neuropathy (25). Whatis less clear is whether glycemic instabilitymay confer a risk to complications that isin addition to that predicted by the meanglucose value alone. It is therefore un-known if two individuals with the samemean blood glucose (MBG), but extremesof glucose variability, might be expectedto have the same or different complicationrisks.

Circumstantial evidence from differ-

ent studies gives conflicting conclusionsas to whether variability in glucose valuesadds to the likelihood of complications.In favor of this association is the fact thatin the DCCT, the rate of complications ata given value of A1C was higher in theconventionally treated patients than inthose intensively treated (3). It was sug-gested that this may be a consequence oflarger glycemic excursions in the formergroup of patients since they were on fewerinjections of insulin per day. Also in sup-port is another study where the incidenceof retinopathy in a group of adolescentswith type 1 diabetes appeared to fall sub-stantially between 1990 and 2002, de-sp i te A1C leve l s changing l i t t l ethroughout the study period (6). It wasagain felt that the move to multiple injec-tion regimes over the time period mayhave contributed to this improvement byreducing glycemic fluctuations ratherthan the mean glucose concentration. Ithas therefore been proposed that beyondsimply avoiding short-term complica-tions such as hypoglycemia and diabeticketoacidosis, minimizing variability inblood glucose control should be a thera-peutic goal for patients to help avoid anyexcess risk of long-term complications aswell (7). More recently, however, it hasbeen shown that the variability of bloodglucose seems to have little influence onthe A1C of a patient over the mean glu-cose value, but it is not known if thistranslates to a similar risk of complica-tions (8).

In a related issue, there is firm evi-dence that postprandial glycemia is astronger risk marker for the large-vessel(macrovascular) complications of diabe-tes than fasting or preprandial concentra-tions (9,10). Indeed, many treatments fordiabetes now focus on reducing glycemicexcursions following meals in the beliefthat they may preferentially help reducethe incidence of cardiovascular disease inthis high-risk group of patients (11).However, little is known about the rela-tive contribution of pre- and postprandialglycemia to the likelihood of developingmicrovascular complications. As a conse-quence, this gap in knowledge has beenseen as a priority for further research (12).

From the 1Department of Clinical Biochemistry, Hull Royal Infirmary, Hull, U.K.; the 2Academic Depart-ment of Cardiology, University of Hull, Hull, U.K.; and the 3Department of Diabetes, Hull York MedicalSchool, Hull, U.K.

Address correspondence and reprint requests to Dr. Eric S. Kilpatrick, Department of Clinical Biochem-istry, Hull Royal Infirmary, Anlaby Road, Hull HU3 2JZ. E-mail: eric.kilpatrick@hey.nhs.uk.

Received for publication 4 February 2006 and accepted in revised form 9 April 2006.Abbreviations: AER, albumin excretion rate; AUC, area under the curve; DCCT, Diabetes Control and

Complications Trial; MBG, mean blood glucose.A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion

factors for many substances.DOI: 10.2337/dc06-0293 2006 by the American Diabetes Association.The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby

marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

See accompanying editorial, p. 1707.

C l i n i c a l C a r e / E d u c a t i o n / N u t r i t i o nO R I G I N A L A R T I C L E

1486 DIABETES CARE, VOLUME 29, NUMBER 7, JULY 2006

The publicly available DCCT data-base has provided a means for investigat-ing these questions more closely, sincethe 1,441 participants with type 1 diabe-tes had 1-day glucose profiles (includingpre- and postprandial measurements)performed quarterly by a laboratory andhad assessments closely detailing any de-velopment or progression of small-vesselcomplications. This current study hastherefore analyzed the DCCT data to de-termine the relative contribution of pre-and postprandial glycemia to the develop-ment of microvascular complications andalso to establish whether the degree of in-stability of glucose control has any addi-tional influence on this risk.

RESEARCH DESIGN ANDMETHODS

The datasetsWe used the publicly accessible datasetscollected by the DCCT, which werestored in SAS format (available at www.gcrc.umn.edu). The DCCT was a 9-yearfollow-up study of 1,441 participantswith type 1 diabetes comparing the effectof intensive versus conventional bloodglucose management on the developmentof microvascular complications of diabe-tes. At randomization, patients were strat-ified into one of two cohorts. The primaryprevention cohort (n 726) had no evi-dence of retinopathy by fundus photogra-phy and a urinary albumin excretion rate(AER) 40 mg/24 h (28 g/min). Thesecondary prevention cohort (n 715)had only minimal retinopathy and a AER200 mg/24 h (140 g/min). The studyparticipants were randomized into inten-sive (n 711) and conventional (n 730) treatment groups.

Definition of eventsSeverity of retinopathy was determinedby the 25-point Early Diabetic Retinopa-thy Treatment Study interim score (2).The development and progression of sus-

tained retinopathy was defined as achange from baseline of three or moreunits on the Early Diabetic RetinopathyTreatment Study score on any two succes-sive annual evaluations. During the 9years of follow-up, 242 people developedsustained retinopathy, 67 of whom werein the intensive treatment group. Ne-phropathy was defined as an increase inAER 40 mg/24 h (28 g/min) on anyannual evaluation providing that thebaseline AER was40 mg/dl (28g/min)(13). The mean age was 27 years (range1339). Just over half (n 761, 52.8%)were men. Average BMI was 23.4 kg/m2;2% had a BMI 30 kg/m2. Nearly allparticipants were Caucasian. The mediandisease duration was 4 years. Approxi-mately one-fifth declared themselves ascurrent smokers.

To have as complete a dataset on eachpatient as possible, we based our analyseson the data up to 5 years since the major-ity of those lost to follow-up defaulted af-ter this time. At 5 years, 94 participantshad retinopathy, 126 had nephropathy,and 189 had either retinopathy ornephropathy.

Glycemic variables and statisticalmethodsA blood glucose profile was taken at3-monthly intervals. Blood glucose wasassessed at seven points throughout theday, namely, prebreakfast (we assumed atime of 0700), postbreakfast (0830), pre-lunch (1200), postlunch (0130), presup-per (1800), postsupper (1930), andbedtime (2200) on 24,652 occasions. Anadditional data point was collected at0300, but since this was only measured in1% of subjects, it is not considered fur-ther. Mean blood glucose was calculatedby the area under the curve (AUC) usingthe trapezoidal rule in accordance withRohlfing et al. (14). Instability of bloodglucose (within-day SD) was calculated asthe SD of daily blood glucose around themean from each quarterly visit (15). Vari-

ability in MBG over time was estimated asthe SD of the MBG (AUC) measurementsmeasured at each quarter. Mean prepran-dial glucose was taken as the average ofprebreakfast, prelunch, and presupperand mean postprandial glucose as the av-erage of postbreakfast, postlunch, andpostsupper.

The relationship between each riskfactor and the development of diabetescomplications was assessed by Cox re-gression from which hazard ratios (HRs)and 95% CIs were calculated. The Coxregression model is semiparametric in thesense that no assumption concerningevent-free survival times is necessary. TheCox regression model is based on the as-sumption that the effect of a risk factor,expressed as an HR, is constant over time.The assumption of proportionality of theCox model covariates was tested by plot-ting Schoenfeld residuals (16,17). Wealso fitted a separate Cox model for MBG(AUC) assuming that MBG was a seg-mented time-dependent covariate withinthe Cox model structure. This modeltakes into account the different measure-ments for MBG over time. MBG was mea-sured at quarterly intervals throughoutthe study period. In the segmented time-dependent model, if MBG was measuredat more than baseline but less than thefirst quarter, the baseline value is used. IfMBG was measured at more than the firstquarter but less than the second, the valueat first quarter is used and so on. Hence,the mean is continuously updated. AllCox regression models were adjusted forthe following baseline covariates: age(years), sex, disease duration (years), ran-domization treatment (conventional ver-sus intensive), prevention cohort(primary versus secondary), and studyphase (first or second). The GLIM4 andSPSS statistical computer packages wereused to analyze the data. An arbitrary levelof 5% statistical significance (two tailed)was assumed.

RESULTS A residual plot for meanMBG (AUC) versus survival time showedthe residuals centered around zero, indi-cating no departure from proportionalityof hazards. Univariate Cox regressionmodels are presented in Table 1. Usingthe MBG profile, larger HRs were seenwith retinopathy (1.16) as an outcomemeasure when compared with nephropa-thy (1.04). For the time-dependentmodel, the HRs were similar for bothcomplications (retinopathy 1.12, ne-phropathy 1.07). Given similar estimates

Table 1Univariate Cox regression models relating glycemia to microvascular complications

Variable Nephropathy P value Retinopathy P value

MBG (AUC) 1.04 (0.971.11) 0.26 1.16 (1.081.25) 0.0001SD MBG (AUC) 0.97 (0.831.13) 0.70 0.99 (0.831.18) 0.92SD (within day) 1.01 (0.991.03) 0.20 1.03 (1.011.05) 0.02Preprandial mean 1.07 (0.991.45) 0.09 1.19 (1.101.29) 0.0001Postprandial mean 1.05 (0.981.12) 0.19 1.17 (1.091.26) 0.0001

Data are HR (95% CI). MBG (AUC), AUC of seven-point blood glucose profile (mmol/l); SD MBG (AUC), SDof between-visit AUC; SD (within-day), SD of glucose profile.

Kilpatrick, Rigby, and Atkin

DIABETES CARE, VOLUME 29, NUMBER 7, JULY 2006 1487

of effect size and residual plots showingno departure from proportionality, therest of our Cox models did not assume atime-dependent relationship. Two multi-variable Cox models are presented in Ta-ble 2. Complication risk was notsignificantly related to variability in bloodglucose after adjusting for MBG (AUC)(model 1, Table 2). Figure 1 shows therelative contribution of rises in both MBGand within-day glucose variability to ret-inopathy risk as derived from the multi-variate model. Considering both pre- andpostprandial MBG together in the sameCox regression model, neither was signif-icantly associated with complication out-come (model 2, Table 2).

There was a nonsignificant (P 0.09)trend to less within-day variability amongpatients using an insulin pump 99100%of the time (average within-day SD 2.75mmol/l) compared with intermittent us-ers (3.00 mmol/l) and those only takingmultiple injections (3.11 mmol/l).

CONCLUSIONS This study hasshown that the variability in blood glu-cose around a patients mean value has noinfluence on the development or progres-sion of either retinopathy or nephropathyin type 1 diabetes. Thus, on average, onepatient with wildly fluctuant glucose con-centrations is likely to have the same riskof these complications as one whose gly-cemia varies little throughout the day, solong as their mean (AUC) glucose valuesremain similar. We have also found thatneither pre- nor postprandial glucoseconcentrations preferentially contributeto the risk of microvascular disease.

There are few studies other than theDCCT that have been able to directly re-late glycemia to microvascular complica-tion risk. Although the U.K. ProspectiveDiabetes Study showed similar findings ina group with type 2 diabetes (4), only fast-

ing glucose measurements were recordedand not the detailed profiles found here.However, by comparing patients on insu-lin with those on oral agents, the U.K.Prospective Diabetes Study probably un-wittingly gave an insight into the effect ofglucose variability on complication risk.As neither of these groups appeared at arisk any different to that expected for theirA1C, the relative glycemic instability ofpatients on insulin did not seem to confera higher probability of microvascular dis-ease. Our findings are also consistent witha recent study using DCCT data thatfound that glycemic instability had littleinfluence on the A1C value of a patient(8).

Notwithstanding, it has certainlybeen proposed that the reason conven-tionally treated patients in the DCCT wereat apparently higher chance of developingretinopathy at any given A1C value thanthose treated intensively is because oftheir relatively greater glycemic instability(3). A similar explanation has been givenfor the reduction in complications seen inadolescents with type 1 diabetes between1990 and 2002 (6). By showing in thecurrent study that glucose variability doesnot influence the relationship betweenAUC glucose and retinopathy, it wouldsuggest that there must be other reasonsfor these findings, which may or may notbe related to the glycemia of the patientsinvolved. With particular regard to theDCCT, our data analysis has also shownthat the lack of influence of glycemic sta-bility is independent of whether the pa-tient was in the conventional or intensivetreatment group.

Postprandial glucose measurementsare only one component of the variabilityin blood glucose within an individual buthave received considerable attention as apredictor of cardiovascular disease, espe-cially in subjects not already known to

have diabetes (1823). The role of post-prandial glucose in the development ofmicrovascular complications is less wellestablished and has seldom extended be-yond assuming reductions in risk on thebasis of changes in A1C (24 26). Re-cently, a cross-sectional study of subjectswith type 2 diabetes, but not insulintreated, has suggested that postprandialglucose may be a better predictor of reti-nopathy (though not nephropathy orneuropathy) in this group of patients(27,28). In contrast, our data analysis hasfound that while both pre- and postpran-dial glucose measurements are individu-ally predictive of retinopathy in aunivariate analysis, there is no evidencethat either measurement is superior to theother in establishing risk in the DCCTgroup of patients.

The fact that MBG was predictive ofretinopathy in both univariate and multi-variate analysis, while it was a relativelypoor predictor of microalbuminuria pro-gression, is likely to be partly a consequenceof the degree of random within-patientvariability in this latter measurement(29). It is also possible that MBG is actu-ally a poorer predictor of this complica-tion than A1C measurement.

The weight that can be placed on thefindings of this study directly relate to thestrengths inherent in the DCCT databasedue to its size and the rigor of its protocol.This is reflected in the large (n 24,652)number of seven-point glucose profilescollected. These were performed every 3months during the study, were from sam-ples collected throughout most of the day(including both pre- and postprandially),and were then analyzed by a laboratory.Crucially, the study was also benefited bybeing performed before other potentialconfounding factors such as antihyper-tensives and lipid-lowering agents camein to common use.

These findings are of direct relevanceto the clinical management of patientswith type 1 diabetes. Patients with thegreatest variability in blood glucose areoften those most likely to face quality-of-life issues related to hypo- and hypergly-cemia (30). It is therefore natural toconsider that they could be at especiallyhigh risk of the microvascular compli-cations of diabetes as well. However,this study would indicate that these in-dividuals, and their caring physicians,can be reassured that they are at nogreater risk of retinopathy than theirglucose AUC suggests.

This study also gives guidance to pa-

Table 2Multivariate Cox regression models relating glycemia to microvascular complica-tions

Variable Nephropathy P value Retinopathy P value

Model 1MBG (AUC) 1.03 (0.961.11) 0.42 1.15 (1.061.25) 0.0001SD MBG (AUC) 0.96 (0.821.11) 0.57 0.97 (0.811.16) 0.72SD (within day) 1.01 (0.991.03) 0.32 1.02 (0.991.04) 0.18

Model 2Preprandial 1.08 (0.941.24) 0.28 1.11 (0.951.30) 0.18Postprandial 0.99 (0.891.12) 0.83 1.08 (0.941.24) 0.31

Data are HR (95% CI). Both models adjusted for baseline covariates. MBG (AUC), AUC of seven-point bloodglucose profile (mmol/l); SD MBG (AUC), SD of between-visit AUC; SD (within-day), SD of glucose profile.

Glucose variability and complications

1488 DIABETES CARE, VOLUME 29, NUMBER 7, JULY 2006

tients who self-monitor their blood glu-cose. Current recommendations do notstipulate a need to measure postprandialblood glucose in preference to basal val-ues (31), and the data presented here sug-gest that there is no specific need for thisto change in order to reduce the likeli-hood of small-vessel complications. Astreatments move toward the increasinguse of analog insulins with both long- and

short-term modes of action (30), thesedata also infer that no preference in tar-geting either pre- or postprandial glucoseis required if the aim is purely to reducemicrovascular risk.

In summary, this study has shownthat pre- and postprandial glucose valuesare equally predictive of the microvascu-lar complications of type 1 diabetes andthat blood glucose variability does not ap-

pear to be an additional factor in theirdevelopment.

References1. The DCCT Research Group: Diabetes

Control and Complications Trial (DCCT):results of feasibility study. Diabetes Care10:119, 1987

2. The Diabetes Control and ComplicationsTrial Research Group: The effect of inten-sive treatment of diabetes on the develop-ment and progression of long-termcomplications in insulin-dependent dia-betes mellitus.NEngl JMed 329:977986,1993

3. The relationship of glycemic exposure(HbA1c) to the risk of development andprogression of retinopathy in the DiabetesControl and Complications Trial.Diabetes44:968983, 1995

4. UK Prospective Diabetes Study (UKPDS)Group: Intensive blood-glucose controlwith sulphonylureas or insulin comparedwith conventional treatment and risk ofcomplications in patients with type 2 di-abetes (UKPDS 33). Lancet 352:837853,1998

5. 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(UKPDS 35): prospective observationalstudy. BMJ 321:405412, 2000

6. Mohsin F, Craig ME, Cusumano J, ChanAKF, Hing S, Lee JW, Silink M, HowardNJ, Donaghue KC: Discordant trends inmicrovascular complications in adoles-cents with type 1 diabetes from 1990 to2002.Diabetes Care 28:19741980, 2005

7. Hirsch IB, Brownlee M: Should minimalblood glucose variability become the goldstandard of glycemic control? J DiabetesComplications 19:178181, 2005

8. McCarter RJ, Hempe JM, Chalew SA:Mean blood glucose and biological varia-tion have greater influence on HbA1c lev-els than glucose instability: an analysis ofdata from the Diabetes Control and Com-plications Trial. Diabetes Care 29:352355, 2006

9. Home P: Contributions of basal and post-prandial hyperglycaemia to micro- andmacrovascular complications in peoplewith type 2 diabetes. Curr Med Res Opin21:989998, 2005

10. Ceriello A: Postprandial hyperglycemiaand diabetes complications: is it time totreat? Diabetes 54:17, 2005

11. Esposito K, Giugliano D, Nappo F,Marfella R: Regression of carotid athero-sclerosis by control of postprandial hy-perglycemia in type 2 diabetes mellitus.Circulation 110:214219, 2004

12. American Diabetes Association: Postpran-dial blood glucose. Diabetes Care 24:775778, 2001

Figure 1Relative risk of retinopathy as a function of increasing MBG (P 0.0001) (A) andincreasing variability (SD) (B) of the within-day blood glucose profile (NS).

Kilpatrick, Rigby, and Atkin

DIABETES CARE, VOLUME 29, NUMBER 7, JULY 2006 1489

13. The Diabetes Control and Complications(DCCT) Research Group: Effect of inten-sive therapy on the development and pro-gression of diabetic nephropathy in theDiabetes Control and ComplicationsTrial. Kidney Int 47:17031720, 1995

14. Rohlfing CL, Wiedmeyer HM, Little RR,England JD, Tennill A, Goldstein DE: De-fining the relationship between plasmaglucose and HbA1c: analysis of glucoseprofiles and HbA1c in the Diabetes Con-trol and Complications Trial. DiabetesCare 25:275278, 2002

15. Moberg E, Kollind M, Lins PE, AdamsonU: Estimation of blood-glucose variabilityin patients with insulin-dependent diabe-tes mellitus. Scand J Clin Lab Invest 53:507514, 1993

16. Schoenfeld D: Partial residual estimationfor the proportional hazards regressionmodel. Biometrika 69:239241, 1982

17. Grambsch P, Therneau TM: Proportionalhazards tests and diagnostics based onweighted residuals. Biometrika 81:515526, 1994

18. de Vegt F, Dekker JM, Ruhe` HG, Stehou-wer CD, Nijpels G, Bouter LM, Heine RJ:Hyperglycaemia is associated with all-cause and cardiovascular mortality in theHoorn population: the Hoorn Study. Dia-betologia 42:926931, 1999

19. Donahue R, Abbott R, Reed D, Yano K:Postchallenge glucose concentration andcoronary heart disease in men of Japaneseancestry: Honolulu Heart Program. Dia-betes 36:689692, 1987

20. Lowe L, Liu K, Greenland P, Metzger B,

Dyer A, Stamler J: Diabetes, asymptom-atic hyperglycemia, and 22-year mortalityin black and white men: the ChicagoHeart Association Detection Project in In-dustry Study. Diabetes Care 20:163169,1997

21. Balkau B, Shipley M, Jarrett R, Pyorala K,Pyorala M, Forhan A, Eschwege E: Highblood glucose concentration is a risk fac-tor for mortality in middle-aged nondia-betic men: 20-year follow-up in theWhitehall Study, the Paris ProspectiveStudy, and the Helsinki Policemen Study.Diabetes Care 21:360367, 1998

22. Coutinho M, Gerstein H, Wang Y, YusufS: The relationship between glucose andincident cardiovascular events: a metare-gression analysis of published data from20 studies of 95,783 individuals followedfor 12.4 years. Diabetes Care 22:233240,1999

23. DECODE Study Group, the European Di-abetes Epidemiology Group: Glucose tol-erance and cardiovascular mortality:comparison of fasting and 2-hour diag-nostic criteria. Arch Intern Med 161:397405, 2001

24. Bastyr E, Stuart C, Brodows R, Schwartz S,Graf C, Zagar A, Robertson K: Therapyfocused on lowering postprandial glu-cose, not fasting glucose, may be superiorfor lowering HbA1c: IOEZ Study Group.Diabetes Care 23:12361241, 2000

25. Avignon A, Radauceanu A, Monnier L:Nonfasting plasma glucose is a bettermarker of diabetic control than fastingplasma glucose in type 2 diabetes. Diabe-

tes Care 20:18221826, 199726. Monnier L, Lapinski H, Colette C: Contri-

butions of fasting and postprandialplasma glucose increments to the overalldiurnal hyperglycemia of type 2 diabeticpatients: variations with increasing levelsof HbA1c. Diabetes Care 26:881885,2003

27. Shiraiwa T, Kaneto H, Miyatsuka T, KatoK, Yamamoto K, Kawashima A, Kanda T,Suzuki M, Imano E, Matsuhisa M, Hori M,Yamasaki Y: Post-prandial hyperglycemiais an important predictor of the incidenceof diabetic microangiopathy in Japanesetype 2 diabetic patients. Biochem BiophysRes Commun 336:339345, 2005

28. Shiraiwa T, Kaneto H, Miyatsuka T, KatoK, Yamamoto K, Kawashima A, Kanda T,Suzuki M, Imano E, Matsuhisa M, Hori M,Yamasaki Y: Postprandial hyperglycemiais a better predictor of the progression ofdiabetic retinopathy than HbA1c in Japa-nese type 2 diabetic patients (Letter). Di-abetes Care 28:28062807, 2005

29. The absence of a glycemic threshold forthe development of long-term complica-tions: the perspective of the DiabetesControl and Complications Trial.Diabetes45:12891298, 1996

30. DeWitt DE, Hirsch IB: Outpatient insulintherapy in type 1 and type 2 diabetes mel-litus: scientific review. JAMA 289:22542264, 2003

31. American Diabetes Association: Stan-dards of medical care in diabetes, 2006(Position Statement). Diabetes Care 29(Suppl. 1):S4S42, 2006

Glucose variability and complications

1490 DIABETES CARE, VOLUME 29, NUMBER 7, JULY 2006