Key landmark studies in the clinical management of type 2 diabetes: evolution or revolution?

Oxford Centre for Diabetes, Endocrinology and Metabolism
Harris Manchester College Oxford; National Institute for Health Research (Senior Research Fellow);
and Oxford Biomedical Research Centre

D. R. Matthews , United Kingdom

Many decades ago, type 2 diabetes was already known to be associated with damaging complications. Although the symptoms of hyperglycemia could be prevented, there was continuing debate about how best to avoid the damaging long-term effects. Early observational data suggested that optimizing metabolic control could be advantageous, but findings from the Universities Group Diabetes Program (UGDP) trials suggested that the solution might not be straightforward. Was good control improving outcome or increasing dangers to the patient? A series of subsequent trials have thrown light on the question—suggesting that the answer depends on the selection of patients for different regimens. Taking care in the clinical appraisal of a patient allows the selection of optimum therapy, which the synthesis of trials suggests will reduce bothmicrovascular andmacrovascular disease. It is now apparent that early treatment of both glycemia and hypertension is beneficial, and the trials of lipid lowering suggest that the risk of cardiovascular disease can be significantly reduced. All the trial data suggest that hyperglycemia is a risk for cardiovascular disease and should be lowered if possible. The target for glycemia is for a glycated hemoglobin (HbA1c) level lower than 7.5%, and this may be nearer 6.5% if achieved slowly and without dangerous hypoglycemia. Early intervention is clearly beneficial. Late intervention to strict targets requires a careful incremental approach.

Medicographia. 2011;33:22-28 (see French abstract on page 28)


Estimates from current epidemiology suggest that the number of those with type 2 diabetes will increase to 285 million people by the year 2010, and to more than 400 million by 2030.1 The problem is not confined to country, race, or geographical location, and so an unprecedented challenge is one of provision of appropriate health care. Many decades ago, type 2 diabetes was already known to be associated with damaging complications.2 Although the symptoms of hyperglycemia could be prevented, there was a continuing debate about how best to avoid the damaging long-term effects. Early observational data suggested that optimizing metabolic control could be advantageous, but until the Universities Group Diabetes Program (UGDP) trial3 no systematic trial evidence had been published. The UGDP results were not clear,4 however, and so, in the following decades, a variety of trials were undertaken. Astonishingly, not a single one of these trials was without controversy over its design, results, or interpretation. Nevertheless, the synthesis of the knowledge adduced from them all allows us a clear vision of the advantages and disadvantages of the pursuit of normoglycemia. Our understanding has been evolutionary—though perhaps a physician from 1985 transported a quarter of a century ahead in time would feel that there had been a revolution in attitudes and therapeutics.

Trials of glycemic control

Throughout the 1950s and 1960s, there had been a growing awareness that diabetes complications—both of microvascular and macrovascular origin—were presenting the greatest challenge to the quality of life and longevity of type 2 diabetes patients. Clinical acumen and observation had demonstrated that patients with glycosuria and very high blood glucose levels had a poor quality of life, but some physicians thought that moderate glycosuria might be advantageous in terms of weight loss—a view now resurgent with the development of sodium-glucose cotransporter 2 (SGLT-2) inhibitors.5

UGDP3 was the first trial to attempt to optimize glycemia using a controlled trial approach. Launched in 1960, this placebocontrolled, multicenter clinical trial aimed to determine which, if any, of the treatments for type 2 diabetes was efficacious. Although the differences seen in cumulative total mortality were not statistically significant, a subgroup analysis suggested that cardiac deaths occurred more frequently in the tolbutamide group. The investigators terminated this limb of the study. However, the randomization was significantly skewed at baseline—there was 30% more electrocardiographic abnormality, 40% more angina, and 90% more hypercholesterolemia in the tolbutamide group.4 Randomization had failed to deliver equipoise in the outcome.

The United Kingdom Prospective Diabetes Study (UKPDS) was established to definitively answer the glycemic control controversy as well as to attempt to answer important questions about the class of agents used to achieve control.6-8 UKPDS was one of very few trials that recruited newly diagnosed type 2 diabetes patients (5012 in total)—an important point, as it transpired, since only UKPDS had the capacity to answer the question of the suitability of early treatment before the onset of serious complications. Despite this criterion, it was nevertheless apparent that many had early signs of trouble ahead—background retinopathy and ECG abnormalities, in particular. The evolution of complications was meticulously recorded—the trial lasted a median of 10 years, with some patients having been followed for 20 years at closeout.

UKPDS had stringent aims for euglycemia on monotherapy, but allowed fasting glucose to rise to 15 mmol/L before adding additional agents. Because one of the aims was to address the question of which monotherapy should be used, glycemia rose progressively throughout the trial. In contrast with later trials, the aim was to persist with monotherapy for as long as possible rather than to achieve a predominant glycemic target. At closeout, the results showed that intensive glucose control was efficacious in reducing many complications. Metformin, used only in the overweight, reduced diabetes-related deaths (risk reduction [RR], 0.58; P=0.017) and myocardial infarction (RR, 0.61; P=0.01) compared with conventional treatment. This is the prime evidence base for the use of metformin. It has been criticized as being based on a UKPDS subset of low statistical power, but one should note that the effect demonstrated in small numbers increases our certainty that this is clinically, as well as statistically, useful. In the main study of sulfonylurea or insulin use, there were clear reductions in relative risk in the intensively treated group: a 12% risk reduction for any diabetes related end point (P=0.029); a 25% risk reduction for microvascular end points (P=0.0099); a 21%risk reduction for retinopathy at twelve years (P=0.015); and a 33% risk reduction for albuminuria at twelve years (P=0.000054). The 16% risk reduction for myocardial infarction had borderline significance (P=0.052).7,8

The study compared intensive versus conventional treatment for blood glucose control and achieved a glycated hemoglobin (HbA1c) level of 7% in the intensive groups of the study population compared with 7.9% in the conventional group (Figure 1). Nevertheless, questions remained—especially the question of how low an HbA1c level one should aim for in glycemic control. Would more aggressive glucose control decrease macrovascular or microvascular disease further?

Figure 1
Figure 1. Diagram of the glycemic control achieved in UKPDS,
ADVANCE, and ACCORD showing differences in duration, duration
of diabetes at recruitment, and glycemic control achieved.

Grey lines show the control groups. Other groups are shown in color: UKPDS:
green = glibenclamide, cyan = chlorpropamide, yellow = insulin; ACCORD: red = intensive control group; ADVANCE: blue = intensive control group.
Abbreviations: ACCORD, Action to Control CardiOvascular Risk in Diabetes; ADVANCE, Action in Diabetes and Vascular disease: PreterAx and DiamicroN MR Controlled Evaluation; HbA1c, glycated hemoglobin; UKPDS, United Kingdom Prospective Diabetes Study.

In 2005, PROACTIVE (PROspective pioglitAzone Clinical Trial In macroVascular Events)9 reported its results. PROACTIVE was a prospective, randomized controlled trial of 5238 patients with type 2 diabetes who had evidence of macrovascular disease. The median follow-up was just under 3 years. Patients were assigned to pioglitazone or placebo taken in addition to their glucose-lowering drugs and other medications. The primary end point was a composite of cardiovascular disease, including surgical intervention in the coronary or leg arteries and amputation above the ankle. The outcome of this was not significant (P=0.095). However, the main secondary end point—the composite of all-cause mortality, nonfatal myocardial infarction, and stroke—was, with a significant, favorable response to pioglitazone (P=0.027).

The trial was marred by the problem of the selection of a primary combined outcome that involved not only the onset of new pathology, but also surgical interventions relating to pathology. Surgical interventions are not emerging pathology— they are a response to emerging pathology and, as such, have many constraints on their timing. A decision about when an amputation is undertaken is as much a clinical decision as it is an emergent complication of diabetes. By contrast, the timing of a myocardial infraction is a direct measure of an agonal point in underlying pathology. A multiplicity of outcomes increases the event count, but can do so at the expense of specificity. Nevertheless, the secondary analyses in PROACTIVE were highly significant—a risk reduction of 28% for myocardial infarction (P=0.045) and 47% for fatal and nonfatal stroke (P=0.009).9

In a remarkable coup-de-théâtre, Nissen et al10 produced a meta-analysis that seemed to demonstrate that rosiglitazone might have an adverse effect on cardiovascular outcome. This meta-analysis has been criticized,11 especially on the grounds that it was not based on a comprehensive search of all the studies that might yield evidence of rosiglitazone’s cardiovascular effects and that the studies were combined on the basis of a lack of statistical heterogeneity, despite variability in study design and outcome assessment.11 Then, in 2009, the Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of glycemia in Diabetes (RECORD) trial reported.12 This trial featured 4447 patients with type 2 diabetes on metformin or sulfonylurea monotherapy with a mean HbA1c of 7.9%. They were randomly assigned to take additional rosiglitazone or a combination of metformin and sulfonylurea. In a 5.5 year follow-up, there was no difference in the primary outcome. There was an increase in heart failure causing admission to hospital or death in the rosiglitazone group (hazard ratio [HR], 2.10; 95% confidence interval [CI], 1.35 to 3.27), and upper and distal lower limb fracture rates increased, mainly in women. So, although rosiglitazone lowers glycemia, it seems that there is a significant increase in complications.

In 2008, three cardiovascular disease trials reported at the American Diabetes Association. These were ADVANCE (Action in Diabetes and Vascular disease: PreterAx and DiamicroN MR Controlled Evaluation), ACCORD (Action to Control CardiOvascular Risk in Diabetes), and VADT (Veterans Affairs Diabetes Trial). ACCORD13 produced a startling headline result that mortality was worse in the group that was intensively treated to lower HbA1c toward 6%. At one year, stable median HbA1c levels of 6.4% and 7.5% were achieved in the intensive- therapy group and the standard-therapy group, respectively. During follow-up, the primary outcome (a composite of nonfatal myocardial infarction, stroke, or cardiovascular death) occurred in 352 patients in the intensive-therapy group, compared with 371 in the standard-therapy group (HR, 0.90; 95% CI, 0.78 to 1.04; P=0.16). However, 257 patients in the intensive-therapy group died, compared with 203 patients in the standard-therapy group (HR, 1.22; 95% CI, 1.01 to 1.46; P=0.04).13 This finding brought the main result from UKPDS into question again. Is intensive glucose lowering harmful? Here, however, the significant differences between ACCORD and UKPDS should be noted.

UKPDS recruited “healthy” new-onset type 2 diabetes patients (serious disease of any kind was a contraindication). In ACCORD, those recruited had been diagnosed a median of 10 years previously and were selected for preexisting cardiovascular disease or specific risk factors. Sudden changes in glycemia in such patients may not be advisable. In this trial, the reports show that the majority of the glucose-lowering effect had already been achieved within the first 4 months, by which time median HbA1c was 6.6%. Although there was no explicit evidence that hypoglycemia was the precipitating cause of death, it remains the number one suspect for the increased death rate. Hypoglycemia rates were three times higher in the intensively treated group (death precludes con- temporaneous measurement of blood glucose). Many of the patients were receiving rosiglitazone (91% in the intensive arm and 57% in the standard therapy arm). The excess mortality was not simply cardiovascular; hypoglycemia can cause falls or nocturnal aspiration that leads to pneumonia. In the elderly, any significant medical event may be seriously life-threatening.

ADVANCE14 is the largest trial of cardiovascular disease in type 2 diabetes to date, recruiting 11 140 patients with type 2 diabetes randomized to standard or intensive glucose control with the aim of using gliclazide (modified release) plus other drugs, as required, to achieve an HbA1c value of 6.5% or less. After a median of 5 years’ follow-up, mean HbA1c in the intensive-control group was 6.5% compared to 7.3% in the standard-control group. In the intensive-control group, there was a reduced incidence in the combined end point of major macrovascular and microvascular events (HR, 0.90; 95% CI, 0.82 to 0.98; P=0.01), as well as that of major microvascular events (9.4%vs 10.9%; HR, 0.86; 95%CI, 0.77 to 0.97; P=0.01), primarily because of a reduction in the incidence of nephropathy (HR, 0.79; 95%CI, 0.66 to 0.93; P=0.006). However, the type of glucose control had no significant effect on major macrovascular events (HR with intensive control, 0.94; 95% CI, 0.84 to 1.06; P=0.32), death from cardiovascular causes (HR with intensive control, 0.88; 95% CI, 0.74 to 1.04; P=0.12), or death from any cause (HR with intensive control, 0.93; 95% CI, 0.83 to 1.06; P=0.28), although the 12% decrease in cardiovascular death is worth noting given the significant 35% increase in ACCORD.

This trial15 randomized 1791 predominantly male military veterans (mean age, 60.4 years) to intensive or standard glucose control, achieving about a 1.5% HbA1c difference over a median duration of 5.6 years. There was no significant difference between the two groups in any component of the primary outcome or in the rate of death from any cause—a finding unremarkable in that the trial was essentially underpowered (both in terms of numbers of subjects and duration). There was, however, a lessening of progression of albuminuria (P=0.01).

Table I
Table I. Relative risk reduction with sulfonylurea/insulin after 10
years’ follow-up in UKPDS.

Abbreviations: MI, myocardial infarction; UKPDS, United Kingdom Prospective Diabetes Study.
Modified from reference 16: Holman et al. N Engl J Med. 2008;359:1577-
1589. © 2008, Massachusetts Medical Society.

_ UKPDS Post-Trial Monitoring
UKPDS monitored its patients for outcome after the study for a median of ten years— with biochemical indices for five of these. The study, published as UKPDS-PTM (UKPDS Post- Trial Monitoring),16 examined whether the effects of being in the intensively controlled group would dissipate with time. After the trial, everyone was given advice about intensive control. The 3277 patients remaining in the trial were asked to attend annual UKPDS clinics for 5 years, but no attempts were made to maintain their previously assigned therapies. Annual questionnaires were used to survey patients who were unable to attend the clinics, and all patients in years 6 to 10 were assessed using questionnaires. In the years that followed their inclusion in the trial, no glycemic differences were strived for, nor seen. The null hypothesis was that with no differences in treatment, the differences in outcome would be lost. But far from there being a diminution of the glycemic trial effect over the ten years, the lower incidence of pathological effects was maintained, and with the advent of more events the statistical probabilities of error declined. In the sulfonylurea-insulin group, relative reductions in risk persisted at 10 years for any diabetes-related end point (9%, P=0.04) and microvascular disease (24%, P=0.001), and risk reductions for myocardial infarction (15%, P=0.01) and death from any cause (13%, P=0.007) emerged over time as more events occurred (Table I).16 In the metformin group, significant risk reductions persisted for any diabetes-related end point (21%, P=0.01), myocardial infarction (33%, P=0.005), and death from any cause (27%, P=0.002). So, despite there being no glycemic differences, a continued reduction in microvascular risk and emergent risk reductions for myocardial infarction and death from any cause were observed during the 10 years of posttrial follow-up.

Evolution or revolution?

It has taken many years for a clear picture to emerge from the glycemic trials, and our understanding has evolved. Nevertheless, looking back on what we knew in 1997 and what we know now, the change in knowledge is a revolution. How do all the trials lock together into one cohesive pattern? The lessons learnt are summarized in Table II (page 26). Interestingly, it turns out that UKPDS7 and UKPDS-PTM16 hold the important core of what we need to know; the other trials color in the details. UKPDS established beyond any reasonable doubt that outcomes in those whose blood pressure and glycemic control were near normal were better, and it provided the major evidence base for the use of metformin.17 It laid to rest the old canard that it was somehow the “diabetes” causing the problems—perhaps by insulin resistance or some other arcane process. UKPDS-PTM16 went further. It established that the first ten years of treatment were crucial to outcome—that there was a glycemic legacy effect. There has been no suggestion from the authors that this was a “metabolic memory” effect— a term used by the Diabetes Control and Complications Trial (DCCT). The effects are most likely to be simply related to atherosclerosis. Higher blood glucose over ten years leads to more vascular damage, and the effect is permanent.

ACCORD taught us all a sharp lesson. Taking high-risk patients and imposing very tight glycemic control led to the perverse outcome of greater mortality in the intensive group. This shows that we need to use clinical care in those in whom hypoglycemia (the major suspect for the adverse outcome) may pose a problem. These patients were identified as general groups in the trial by the presence of preexisting high HbA1c or by having been diagnosed at a younger age. Extra caution is needed in those in whom established pathology can be detected. The effects shown in the ADVANCE trial,18 whose duration of diabetes was similar to that of ACCORD (8 years), were mainly attributable to a 21% relative reduction in nephropathy, but unlike the ACCORD trial, there was no indication that achieving the target of 6.5% gradually over four years had any detrimental cardiovascular effects nor did it cause increased mortality. How can one explain the differences between these outcomes? ADVANCE used gliclazide (mainly gliclazide modified release) and metformin to lower glycemia in the intensive control group, which contrasts with ACCORD where rosiglitazone was used extensively (in both arms), as was insulin and sulfonylurea in combination. In ACCORD, glycemic targets were aggressively pursued and control of glycemia over time did not deteriorate (Figure 2).13 By contrast, in ADVANCE, the target HbA1c of below 6.5% was achieved progressively over a period of 4 years—a much slower rate than that of the ACCORD patients, and the totality of the updated mean difference was much less. So, the differences between the two trials were a marked difference in the rate of achievement of target glycemia, a very high hypoglycemia rate in ACCORD (nearly 4 times greater than the rate in ADVANCE), and a clear difference in the choice of agents for the two trials. ACCORD suggests that intensive glycemic control achieved fast and late in diabetes using multiple agents might not be wise. ADVANCE suggests that the achievement of such targets over a period of several years should not be contraindicated and that there may be gains to be achieved in the prevention of renal disease. ACCORD teaches us that very sudden changes in glycemia in the elderly may do more harm than good, while ADVANCE suggests that even if cardiovascular disease cannot be diminished in the short term, one can at least gain from less renal pathology in the long term.

Table II
Table II. Lessons learnt from trials of glycemia in diabetes.

Abbreviations: ACCORD, Action to Control CardiOvascular Risk in Diabetes; ADVANCE, Action in Diabetes and Vascular disease: PreterAx and DiamicroN MR
Controlled Evaluation; HbA1c , glycated hemoglobin; MI, myocardial infarction; UKPDS, United Kingdom Prospective Diabetes Study; UKPDS-PTM, United Kingdom Prospective Diabetes Study Post-Trial Monitoring.

Figure 2
Figure 2. Glucose control at baseline and during follow-up in the ACCORD study.

The median glycated hemoglobin levels at each study visit are shown. The bars denote interquartile ranges.
Abbreviation: ACCORD, Action to Control CardiOvascular Risk in Diabetes.
Modified from reference 13: Gerstein et al. N Engl J Med. 2008;358:2545-2559. © 2008, Massachusetts
Medical Society.

What agents should one use? The peroxisome proliferator-activated receptor γ (PPAR-γ) agonist pioglitazone has emerged as being efficacious in the light of the PROACTIVE trial, but rosiglitazone still has to prove its effectiveness on cardiovascular outcomes. The evidence tends to point in the other direction. And what about sulfonylureas? Despite debates dating back to the days of UGDP and the fear that â-cell failure might be affected, sulfonylureas have continued to stand the test of time. There were no detrimental signals from UKPDS or from UKPDS-PTM,16 and ADVANCE shows that gliclazide modified release can be an effective late intervention. And what about insulin? UKPDS used insulin as one of its randomized interventions, and there was no suggestion that this policy was unsafe or had detrimental outcomes.

What about hypoglycemia? Here, I think we are closer to real answers. ACCORD had a very high rate of hypoglycemia, and there are many rational reasons to suppose this to be dangerous in the elderly. So, we need to take new stock of this as a real life-threatening risk as well as a threat to quality of life. Hypoglycemia in the elderly threatens events related to falls, aspiration pneumonia, accidents, forgetfulness, and other significant risks.

The trials reported here have focused on glycemic control, but type 2 diabetes cannot simply be treated as a disease of abnormal glucose. The trial data are strongly indicative that lipids and blood pressure19,20 should be treated in parallel, and the Steno 221,22 trial—a trial of multiple intervention care-packages— suggests real benefits from this approach.


All the trial data suggest that hyperglycemia is a risk for cardiovascular disease and should be lowered if possible. The targets for glycemia are an HbA1c lower than 7.5%, and this may be nearer 6.5% if achieved slowly and without dangerous hypoglycemia. Early intervention is clearly beneficial. Late intervention to strict targets requires a careful incremental approach. _

1. The Global Burden. IDF Diabetes Atlas. 4th ed. International Diabetes Federation: Brussels, Belgium; 2009:21-27.
2. Moss SE, Klein R, Klein BE. Cause-specific mortality in a population-based study of diabetes. Am J Public Health. 1991;81:1158-1162.
3. Meinert CL, Knatterud GL, Prout TE, Klimt CR. A study of the effects of hypoglycemic agents on vascular complications in patients with adult-onset diabetes. II. Mortality results. Diabetes. 1970;19(suppl):789-830.
4. Leibel B. An analysis of the University Group Diabetes Study Program: data results and conslusions. Can Med Assoc J. 1971;105:292-294.
5. Idris I, Donnelly R. Sodium-glucose co-transporter-2 inhibitors: an emerging new class of oral antidiabetic inhibitors. Diabetes Obesity Metabol. 2009;11:79-88.
6. UKPDS Group. UK Prospective Diabetes Study VIII: Study design, progress and performance. Diabetologia. 1991;34:877-890.
7. UKPDS Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet. 1998;352:837-853.
8. Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovascu- lar and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ. 2000;321:405-412.
9. Dormandy JA, Charbonnel B, Eckland DJ, et al. Secondary prevention of macrovascular events in patients with type 2 diabetes in the PROactive Study (PROspective pioglitAzone Clinical Trial In macroVascular Events): a randomised controlled trial. Lancet. 2005;366:1279-1289.
10. Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med. 2007;356:2457-2471.
11. Diamond GA, Bax L, Kaul S. Uncertain effects of rosiglitazone on the risk for myocardial infarction and cardiovascular death. Ann Intern Med. 2007;147:578- 581.
12. Home PD, Pocock SJ, Beck-Nielsen H, et al. Rosiglitazone evaluated for cardiovascular outcomes in oral agent combination therapy for type 2 diabetes (RECORD): a multicentre, randomised, open-label trial. Lancet. 2009;373:2125- 2135.
13. Gerstein HC, Miller ME, Byington RP, et al. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med. 2008;358:2545-2559.
14. Patel A, MacMahon S, Chalmers J, et al. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med. 2008;358: 2560-2572.
15. Duckworth W, Abraira C, Moritz T, et al. Glucose control and vascular complications in veterans with type 2 diabetes. N Engl J Med. 2009;360:129-139.
16. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med. 2008;359: 1577-1589.
17. UKPDS Group. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). UK Prospective Diabetes Study (UKPDS) Group. Lancet. 1998;352:854-865.
18. Patel A, MacMahon S, Chalmers J, et al. Effects of a fixed combination of perindopril and indapamide on macrovascular and microvascular outcomes in patients with type 2 diabetes mellitus (the ADVANCE trial): a randomised controlled trial. Lancet. 2007;370:829-840.
19. UKPDS Group. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). Lancet. 1998;352:854-865.
20. Adler AI, Stratton IM, Neil HA, et al. Association of systolic blood pressure with macrovascular and microvascular complications of type 2 diabetes (UKPDS 36): prospective observational study. BMJ. 2000;321:412-419.
21. Gaede P, Valentine WJ, Palmer AJ, et al. Cost-effectiveness of intensified versus conventional multifactorial intervention in type 2 diabetes: results and projections from the Steno-2 study. Diabetes Care. 2008;31:1510-1515.
22. Gaede P, Pedersen O. Multi-targeted and aggressive treatment of patients with type 2 diabetes at high risk: what are we waiting for? Horm Metab Res. 2005;37(suppl 1):76-82.

Keywords: trials; type 2 diabetes; cardiovascular disease; gliclazide; insulin; rosiglitazone; pioglitazone; ADVANCE; ACCORD; UGDP; UKPDS; PROACTIVE; RECORD; VADT; Steno 2