Effect of Early Multifactorial Therapy Compared With Routine Care on Microvascular Outcomes at 5 Years in People With Screen-Detected Diabetes: A Randomized Controlled Trial: The ADDITION-Europe Study

Type 2 diabetes is a common chronic condition associated with a substantial burden of microvascular and macrovascular morbidity. Treatment of individual risk factors, such as blood pressure and glucose, reduces the risk of microvascular complications (1–4). Intensive treatment (IT) of multiple risk factors can halve the risk of retinopathy, nephropathy, and neuropathy among individuals with long-standing diabetes and microalbuminuria (5). However, the effects of starting multifactorial treatment earlier in the course of the disease are uncertain.

Diabetes is frequently asymptomatic, with the true onset occurring several years before diagnosis. When diagnosed, many patients exhibit evidence of microvascular complications (2,3,6). Early detection by screening is not associated with psychological harms (7); therefore, diabetes meets many of the criteria for screening. However, uncertainties still exist about the magnitude of the benefits of early detection and subsequent intensive management. The ADDITION (Anglo-Danish-Dutch Study of Intensive Treatment in People With Screen Detected Diabetes in Primary Care)-Europe trial was set up to investigate whether intensive multifactorial treatment improved outcomes compared with routine care (RC) when commenced in the lead time between detection by screening and clinical diagnosis. IT was associated with a nonsignificant 17% reduction in the composite cardiovascular end point over 5.3 years compared with RC (8). Data from trials of IT of hyperglycemia suggest that beneficial effects can be seen for microvascular outcomes in the short term (9–11), whereas cardiovascular benefits are only evident with longer follow-up (5,12). We report the effect of early intensive multifactorial treatment compared with RC on risk of microvascular complications, including retinopathy, nephropathy, and peripheral neuropathy, at 5 years following screening for diabetes.

The study design and rationale have been reported previously (8). In brief, ADDITION-Europe comprises two phases: 1) a screening phase and 2) a pragmatic cluster randomized parallel group trial in four centers (Denmark; Cambridge, U.K.; the Netherlands; and Leicester, U.K.). The study was approved by the local ethics committee of each center. All participants provided informed consent.

Of 1,312 general practices invited to participate, 379 (29%) agreed, and 343 (26%) were independently randomized to screening plus RC for diabetes or screening followed by intensive multifactorial treatment. Population-based stepwise screening programs among people aged 40–69 years (50–69 years in the Netherlands) without known diabetes were completed between April 2001 and December 2006 (8). The screening phase included an oral glucose tolerance test for all individuals in Leicester and a stepwise screening program using random glucose measurements and HbA_1c followed by fasting glucose and oral glucose tolerance test in all other centers. Individuals were given a diagnosis of diabetes based on World Health Organization criteria (13). All patients with newly diagnosed type 2 diabetes were eligible to participate in the treatment study unless their family physician indicated that they had contraindications to the proposed study medication, an illness with a life expectancy of <12 months, or psychological or psychiatric problems that were likely to invalidate informed consent.

The practices were randomly assigned by statisticians independent of measurement teams to provide intensive multifactorial treatment or routine diabetes care according to national guidelines in a 1:1 ratio. Randomization included stratification by county, number of full-time family physicians in Denmark, and single-handed or group practice in the Netherlands. In Cambridge, randomization included minimization for the local district hospital and the number of patients with diabetes per practice. In Leicester, randomization included minimization for practice demographics, deprivation status, and prevalence of type 2 diabetes. Participants were unaware of study group allocation.

The characteristics of the interventions to promote target-driven treatment in each center have been described previously (8,14). Further details are available on the study website (www.addition.au.dk). We aimed to educate and support family physicians, practice nurses, and participants in target-driven management (using medication and promotion of healthy lifestyles) of hyperglycemia, blood pressure, and cholesterol based on the stepwise regimen used in the Steno-2 study (15). Treatment targets and algorithms were identical for the IT group in all centers and were based on trial data demonstrating benefits of IT of cardiovascular risk factors in people with type 2 diabetes (4,16–18). Targets included HbA_1c <7.0% (53 mmol/mol), blood pressure ≤135/85 mmHg, and cholesterol <5 mmol/L in patients without ischemic heart disease or <4.5 mmol/L in patients with ischemic heart disease. Family physicians were encouraged to consider prescribing an ACE inhibitor for patients with blood pressure ≥120/80 mmHg or a previous cardiovascular event (17), and 75–80 mg of aspirin daily to patients with controlled hypertension and without specific contraindications. Following publication of the Heart Protection Study (19), the treatment algorithm included a recommendation to prescribe a statin to all patients with a cholesterol level ≥3.5 mmol/L. Although targets for treatment were specified and classes of drugs recommended, the choice of therapy was made by prescribing clinicians and by patients. In the RC group, patients received standard diabetes care according to the current recommendations applicable in each center (20–22).

Prespecified secondary outcomes included measures of kidney function (microalbuminuria, macroalbuminuria, and estimated glomerular filtration rate [eGFR]), retinopathy, and peripheral neuropathy. Health assessments at baseline and follow-up included biochemical, anthropometric, and questionnaire measures and were undertaken by centrally trained staff following standard operating procedures and unaware of study group allocation. Follow-up examinations took place from September 2008 to the end of December 2009, which gives a mean (SD) follow-up period of 5.3 (1.6) years. All biochemical measures were analyzed in five regional laboratories at baseline and follow-up. Standardized self-report questionnaires were used to collect information on sociodemographic characteristics (age, sex, ethnicity), smoking status, and prescribed medication.

Nephropathy was assessed by the urinary albumin-to-creatinine ratio (ACR) and eGFR. Spot urine ACR was measured with a Roche Hitachi 912 chemistry analyzer at Aarhus Hospital (Aarhus, Denmark) and the Steno Diabetes Centre (Gentofte, Denmark), an Olympus AU400 analyzer at Addenbrooke’s Hospital (Cambridge, U.K.) and the Royal Infirmary (Leicester, U.K.), and a Roche/Hitachi Modular P analyzer at the SHL Centre for Diagnostic Support in Primary Care (Etten-Leur, the Netherlands). Repeated analyses of standardized trial control samples for urine creatinine levels during follow-up confirmed reliability and precision of laboratory methods with coefficients of variation (CVs) <3.4% in all laboratories. Analyses of trial and external quality control samples of urine albumin revealed CVs between 2.0% and 9.8% in Etten-Leur, Leicester, and Gentofte and 4.9% and 3.4% for low and high concentrations, respectively, in Cambridge during the trial testing period. Microalbuminuria was defined as an ACR ≥2.5 mg/mmol for males and ≥3.5 mg/mmol for females, and macroalbuminuria was defined as an ACR ≥25 mg/mmol.

Nephropathy was defined as the presence of either microalbuminuria or macroalbuminuria. eGFR was calculated using data on serum creatinine level, age, sex, and ethnicity for each individual using the Modification of Diet in Renal Disease equation (23) at baseline and follow-up. Change between the two time points was analyzed as a continuous variable. Plasma creatinine level was analyzed with kinetic colorimetric methods at all laboratories at baseline and follow-up except in the Netherlands, where an enzymatic method was used at follow-up. Repeated analyses of standardized control samples for creatinine level during follow-up confirmed reliability and precision of laboratory methods, with CVs between 1.3% and 6.4%.

Retinopathy was assessed using gradable digital images (two from each eye: one with the fovea in the center and one with the macula in the center). In the Netherlands and Leicester, all retinal images were taken as part of the follow-up examination. In Denmark, 81% of the images were taken as part of study follow-up, whereas the remainder were obtained from routine health service records. All retinal images in Cambridge were retrieved from routine medical records. Only images taken in the 2 years preceding the follow-up visit were included in this analysis. Information on retinal photography devices used at the four centers is available on the study website (www.addition.au.dk). Three certified graders unaware of the participants’ study group allocation used a quantitative method to grade retinal images, which were subsequently categorized according to the Early Treatment Diabetic Retinopathy Study (ETDRS) semiquantitative scale (24). Two binary end points were then defined: 1) any retinopathy versus no retinopathy and 2) severe or proliferative retinopathy versus none, mild, or moderate retinopathy.

Peripheral neuropathy was assessed using the self-administered Michigan Neuropathy Screening Instrument (MNSI), which includes 13 questions about neuropathic symptoms (25). Responses to the questions were summed to calculate the total score. A yes response to items 1–3, 5–6, 8–9, 11–12, 14–15 is counted as 1 point. A no response to items 7 and 13 is counted as 1 point. Participants were defined as having peripheral neuropathy if they had a score of ≥7. The peripheral neuropathy scores were considered missing if categorization was not possible due to unanswered items (26).

The sample size for the trial was estimated based on the composite cardiovascular event end point (the primary end point for the trial) as previously described (8). The analysis plan for microvascular end points was finalized and published (www.addition.au.dk) before the final analysis. The analysis was by intention to treat. Participant demographic and clinical characteristics at baseline were summarized by group using means and SDs, medians and interquartile ranges, or frequencies and percentages, as appropriate.

Binary end points (any albuminuria and any retinopathy) were analyzed using a logistic regression model estimating odds ratios (ORs) and 95% CIs for the comparison of IT with RC separately within each center. The continuous end points (ACR, eGFR) were analyzed using a normal error regression model, with adjustment for baseline.

In both logistic and normal error regression models, given the cluster randomized design, SEs were adjusted to allow for correlation between patients within practices using the cluster option in Stata. The estimated ORs and differences in means from the four centers were pooled using fixed-effects meta-analysis, and a forest plot was used to display the results. The I² statistic, representing the proportion of variability between centers due to heterogeneity, was calculated. Intracluster correlation coefficient values were estimated for all the microvascular outcomes.

Prespecified analyses were done for potential interactions between randomized groups and subgroups defined by the following baseline variables: age (<60 and ≥60 years), sex, HbA_1c (<6.6% [49 mmol/mol] or ≥6.6% [49 mmol/mol]), and presence of albuminuria. Participants with missing baseline values for ACR and eGFR were included in the analysis using the missing indicator method (27). The impact of missing data for the retinopathy and peripheral neuropathy end points was investigated with a sensitivity analysis using multiple imputation.

Changes in mean values or percentages of other clinical and medication variables from baseline to follow-up were summarized in each randomized group, and intervention effects and 95% CIs of these changes were estimated using the methods described. ORs of meeting each of three treatment targets (blood pressure, cholesterol, and HbA_1c) comparing IT with RC with were also estimated using the same methods.

Screening identified 3,233 patients with type 2 diabetes, of whom 3,057 agreed to take part in the randomized controlled trial (Denmark, n = 1,533; Cambridge, n = 867; the Netherlands, n = 498; Leicester, n = 159); 196 participants died before follow-up. The effect of the intervention on change in clinical and prescribed medication variables from baseline to follow-up is shown in Table 1 and has been reported previously (8). Clinical, biochemical, and treatment characteristics of ADDITION-Europe participants at baseline are presented in Supplementary Table 1. In summary, small, but significant differences between groups in favor of the IT group were seen in the changes for systolic and diastolic blood pressure, HbA_1c, and total and LDL cholesterol levels. At follow-up, more patients in the IT group reported using glucose-lowering, antihypertensive, and lipid-lowering medication and aspirin than in the RC group. However, in both groups, the use of these drugs significantly increased. Furthermore, a significantly higher proportion of patients in the IT group met the targets compared with the RC group at follow-up (Table 2).

Of the 2,861 patients still alive at 5 years, 2,493 (87.1%), 2,710 (94.7%), and 2,312 (80.9%) had data for urine ACR, eGFR, and peripheral neuropathy, respectively. Retinal photographs were obtained for 2,190 (76.6%) participants (Fig. 1).

At follow-up, any albuminuria was present in 316 (22.7%) participants in the IT group and 269 (24.4%) in the RC group, whereas macroalbuminuria was present in 56 (4.0%) and 37 (3.4%) patients, respectively. Center-specific ORs for any albuminuria favored the IT group, but the pooled OR was not statistically significant at 0.87 (95% CI 0.72–1.07) (Fig. 2). The pooled OR for macroalbuminuria was 1.15 (0.76–1.74). In both groups ACR increased between baseline and follow-up. In the IT group, the mean (SD) increase was 1.45 (0.60) mg/mmol and in the RC group, 1.30 (0.66) mg/mmol. The overall difference in means was −0.02 (95% CI −0.96 to 0.91 mg/mmol). There were no significant interactions between study group and any of the subgroups. Mean eGFR increased between baseline and follow-up in both the IT (4.31 [0.49] mL/min) and RC (6.44 [0.90] mL/min) groups, with an overall difference of −1.39 (−2.97 to 0.19). There were no significant interactions between study group and any of the subgroups regarding eGFR. The number of missing values was equally distributed between groups.

Retinopathy was present in 125 (10.2%) patients in the IT group and 116 (12.1%) in the RC group. Center-specific ORs favored the IT group, but the pooled OR was not statistically significant (0.84 [0.64–1.10]) (Fig. 2). Imputation of missing values did not affect the estimates.

Participants without retinal images at follow-up had significantly higher mean HbA_1c levels at baseline than those with retinal images (7.18% [55 mmol/mol] and 6.99% [53 mmol/mol], respectively, P = 0.044), but there was no difference between groups (interaction P = 0.78). We found a significant interaction among retinopathy, randomized group, and baseline HbA_1c (P = 0.007). IT appeared to be more effective among participants with HbA_1c ≥6.6% (49 mmol/mol) at baseline (OR 0.65 [0.45–0.93]) than among those with HbA_1c <6.6% (49 mmol/mol) (1.17 [0.75–1.82]). There was no evidence of an interaction with either age or sex. Severe retinopathy was present in one participant in the IT group and seven in the RC group.

Peripheral neuropathy was present in 63 (4.9%) participants in the IT group and 60 (5.9%) in the RC group (pooled OR 0.95 [0.68–1.34]) (Fig. 2). Nonresponders had higher BMI values (P = 0.055) and were more likely to be from a minority ethnic group (P = 0.004) than responders. Imputation of missing values did not affect the estimates. The overall intracluster correlation coefficient values were 0.024 (95% CI 0.0060–0.095), 0.014 (0.00017–0.55), and 0.011 (4.6 × 10⁻⁷ to 1) for albuminuria, retinopathy, and neuropathy, respectively.

An intervention achieving modest changes in prescribed treatment and improvements in cardiovascular risk factors in patients with screen-detected type 2 diabetes was not associated with significant reductions in the frequency of microvascular events at 5 years. Differences between study groups for microvascular end points tended to favor the IT group; differences were greatest for retinopathy and smallest for neuropathy. However, we cannot rule out the possibility that these findings were due to chance. The frequency of microvascular complications 5 years after diagnosis was lower than expected and lower than that reported among individuals who have had diabetes for a similar length of time. This finding is likely due to early detection and close-to-optimal treatment in both study groups.

Any albuminuria was common at diagnosis in participants in the ADDITION study (19.9%) (28). The finding of 7.2% with microalbuminuria at baseline in the UK Prospective Diabetes Study (UKPDS) is not directly comparable because a higher level of urinary albumin excretion was used to define microalbuminuria in the UKPDS compared with the ADDITION study (3). Although ADDITION participants were older (mean age 60 years vs. 52 years in UKPDS) and more likely to have hypertension at diagnosis, the rate of progression of albuminuria from 19.9% at baseline to 23.5% (0.7% per year) was one-half that observed in the UKPDS (3). Higher progression rates were also reported in the ACCORD (Action to Control Cardiovascular Risk in Diabetes) (10), ADVANCE (Action in Diabetes and Vascular Disease—Preterax and Diamicron Modified Release Controlled Evaluation) (9), and VADT (Veterans Affairs Diabetes Trial) (11) trials; however, these trials recruited older people with long-standing diabetes.

The increase in eGFR observed in both groups represents a benefit for the patients and is probably a result of the complex treatment regimen, including ACE inhibitors and angiotensin receptor blockers, lipid-lowering treatment, and glucose-lowering treatment.

The frequency of retinopathy was lower than expected. In the UKPDS, 37% of patients had retinopathy at diagnosis, and retinopathy developed in a further 22% within 6 years (2). In contrast, retinopathy developed in only 12% of patients in the ADDITION RC group within 5 years of diagnosis. Again, this finding is likely to reflect treatment in the lead time between detection by screening and clinical diagnosis.

IT appeared to be more effective among participants with HbA_1c ≥6.6% (49 mmol/mol) at baseline than among those with HbA_1c <6.6% (49 mmol/mol). This finding is likely to be explained by the increased risk of retinopathy at baseline in the former group, which, therefore, had a larger potential for risk reduction by IT.

The prevalence of peripheral neuropathy was ∼5%, with no difference between study groups and substantially lower than the 11% reported in a population having had type 2 diabetes for 16 years (26). However, the MNSI may not be sufficiently sensitive (29) or reliable to detect early features of neuropathy and differences between study groups (30). The heterogeneity in effect size estimates for this outcome across centers supports this view.

Participants were drawn from large, population-based samples in three European countries. Although recruitment was nonrandom, a large geographical area was covered in each country. The practices randomized (26% of those invited) were nationally representative for key sociodemographic and clinical characteristics. Participants were identified using a range of screening procedures, but all were diagnosed based on World Health Organization criteria. The intervention targeted both patient and practitioner behavior, and treatment algorithms and targets were based on robust trial data.

To minimize the risk of contamination, we randomized general practices rather than individual patients, and the analyses allowed for this cluster design. The intracluster correlation coefficients are small (31) and did not adversely affect study power. They have been reported here for future designs of cluster trials. Both patient and practice characteristics were well matched at baseline. Clinically important outcomes were assessed using standardized equipment and protocols by trained staff unaware of study group allocation. Although screening procedures and intervention delivery varied among centers, there was little heterogeneity in effect size estimates for nephropathy and retinopathy.

There were differences in the proportion of participants with data for the various microvascular end points at 5 years. These ranged from 77% for retinopathy to 95% for eGFR. Examination of characteristics of ADDITION-Europe participants with and without outcome data suggests that some healthy volunteer bias may have been present, although absolute differences were small (e.g., 0.19% [2 mmol/mol] for HbA_1c). There were no between-group differences in the proportion of patients with missing data. However, we may have underestimated the prevalence of retinopathy and neuropathy given the adverse risk profile of those with missing data for these outcomes.

Furthermore, the microvascular end points reported here were predefined secondary end points in the original trial; therefore, a formal power calculation was not performed for these end points. However, it is still legitimate to report the effects of the intervention on these end points. The CIs show the range of effect sizes, which are compatible with the data and, therefore, are more informative than considerations of power after a trial has been completed, as discussed in the CONSORT (Consolidated Standards of Reporting Trials) sample size reporting guidelines (32).

In this pragmatic trial, outcome assessment and laboratory testing were standardized across participants from both study groups in each center but differed between centers. For example, there was no standardized examination procedure for retinal photography. Some centers collected retinal data as part of the 5-year follow-up examination, whereas others used routine data sources. This variation may have led to differential precision of outcome assessment among centers but not between study groups. Furthermore, only gradable photographs were assessed, and these were coded by three experienced ophthalmologists using a standard scale (Early Treatment of Diabetic Retinopathy Study) while unaware of study group allocation. This may have contributed to some heterogeneity in the results.

Outcomes tended to favor more IT; however, the low frequency of microvascular events means that the 5-year duration of follow-up may have been insufficient to detect potentially clinically important differences between the RC and the IT groups. The trial was conducted during a period in which targets for blood pressure and cholesterol levels became stricter for diabetic patients, which resulted in smaller-than-expected differences between the study groups in terms of cardiovascular risk factors, prescribed medications, and cardiovascular disease outcomes (8). Further follow-up of the trial cohort may be justified to examine whether early intensive multifactorial treatment reduces microvascular risk in the long term as seen in the UKPDS and Steno-2 studies (5,12).

When compared with RC, an intervention to promote target-driven, intensive management of patients with type 2 diabetes detected by screening was associated with modest differences in prescribed treatment and levels of cardiovascular risk factors. However, the intervention was not associated with significant reductions in the frequency of microvascular events at 5 years.

Acknowledgments. The authors thank Toke Bek, Eye Department, Aarhus University Hospital, and Henrik Lund Andersen, Eye Department, Glostrup Hospital, for contributions to the manual operating procedures of grading retinal photos.

Funding. ADDITION-Denmark was supported by the National Health Services in the counties of Copenhagen, Aarhus, Ringkøbing, Ribe, and South Jutland; the Danish Council for Strategic Research; the Danish Research Foundation for General Practice; Novo Nordisk Foundation; the Danish Centre for Evaluation and Health Technology Assessment; the diabetes fund of the National Board of Health; the Danish Medical Research Council; and the Aarhus University Research Foundation. Parts of the grants from the Novo Nordisk Foundation and the Danish Council for Strategic Research were transferred to the other centers. ADDITION-Cambridge was supported by the Wellcome Trust (grant reference no: G061895), the Medical Research Council (grant reference no: G0001164), the National Institute for Health Research (NIHR) Health Technology Assessment Programme (grant reference no: 08/116/300), and National Health Service research and development support funding (including the Primary Care Research and Diabetes Research Networks), and the NIHR under its Programme Grants for Applied Research scheme (RP-PG-0606-1259 to M.J.D. and K.K.). Bio-Rad provided ADDITION-Cambridge with the equipment for HbA_1c testing during the screening phase. ADDITION-Netherlands was supported by the Julius Centre for Health Sciences and Primary Care, University Medical Centre, Utrecht. ADDITION-Leicester was supported by the Department of Health and ad hoc Support Sciences, the NIHR Health Technology Assessment Programme (grant reference no: 08/116/300), and National Health Service research and development support funding (including the Primary Care Research and Diabetes Research Networks Leicestershire, Northamptonshire, and Rutland Collaborative for Leadership in Applied Health Research and Care).

The funding sponsors of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.

Duality of Interest. ADDITION-Denmark has been given unrestricted grants from Novo Nordisk AS, Novo Nordisk Scandinavia AB, Novo Nordisk U.K., AstraZeneca Denmark, Pfizer Denmark, GlaxoSmithKline Pharma Denmark, Servier Denmark A/S, and HemoCue Denmark A/S. Part of the grant from Novo Nordisk was transferred to the other centers. ADDITION-Netherlands was supported by unrestricted grants from Novo Nordisk, GlaxoSmithKline, and Merck.

All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare the following conflicts of interest. A.S. has received honoraria for speaking and travel expenses from Novo Nordisk Denmark. S.J.G. has attended an advisory board for Colgate Palmolive; has received honoraria for speaking from Unilever, Eli Lilly, GlaxoSmithKline, Merck, Sharp & Dohme (MSD), and Novo Nordisk; and has received travel expenses from Eli Lilly. K.B.-J. was the director of the Steno Diabetes Centre, which is owned by Novo Nordisk, and holds stock in Novo Nordisk. G.E.H.M.R. has served as a consultant and participated on advisory boards for Novo Nordisk and MSD and has received honoraria for speaking from Novo Nordisk. M.v.d.D. has received travel expenses from Eli Lilly. T.L. has received research funding from Novo Nordisk, AstraZeneca, Pfizer, GlaxoSmithKline, Servier, and HemoCue; has received honoraria for speaking from various pharmaceutical companies; and holds stock in Novo Nordisk. M.J.D. has served on advisory boards for Novo Nordisk, Eli Lilly, MSD, Bristol-Myers Squibb (BMS), and Roche; and has received honoraria for speaking from Novo Nordisk, Eli Lilly, Sanofi, Novartis, and MSD. M.J.D. and K.K. have received funds for research from Novo Nordisk, Eli Lilly, MSD, Boehringer Ingelheim, Novartis, Roche, and BMS-AstraZeneca. K.K. has participated on advisory boards for Novo Nordisk, Eli Lilly, MSD, Boehringer Ingelheim, Sanofi-Aventis, Novartis, and BMS-AstraZeneca and has received honoraria for speaking from Novo Nordisk, Eli Lilly, Boehringer Ingelheim, Novartis, BMS-AstraZeneca, and MSD. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. A.S. is a principal investigator of the trial and contributed to the study design; data acquisition, analysis, and interpretation; drafting of the manuscript; and critical revision of the manuscript for intellectual content. S.J.G., G.E.H.M.R., N.J.W., M.J.D., and K.K. are principal investigators of the trial and contributed to the study design; data acquisition, analysis, and interpretation; and critical revision of the manuscript for intellectual content. S.J.S. contributed to the data analysis and interpretation and critical revision of the manuscript for intellectual content. R.K.S. and M.v.d.D. contributed to the data acquisition, analysis, and interpretation and critical revision of the manuscript for intellectual content. K.B.-J. is vice-chair of the steering committee and a principal investigator for the trial and contributed to the study design, data analysis and interpretation, and critical revision of the manuscript for intellectual content. T.L. is chair of the steering committee and a principal investigator of the trial and contributed to the study design; data acquisition, analysis, and interpretation; and critical revision of the manuscript for intellectual content. A.S. and S.J.S. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the International Diabetes Federation World Congress, Dubai, United Arab Emirates, 4–8 December 2011.