The UK Prospective Diabetes Study (UKPDS)¹ was a landmark randomised, multicentre trial of glycaemic therapies in 5,102 patients with newly diagnosed type 2 diabetes (T2D). It ran for 20 years (1977 to 1997) in 23 UK clinical sites and showed conclusively that the complications T2D, previously often regarded as inevitable, could be reduced by improving blood glucose and/or blood pressure control.

The UKPDS was designed and run by the late Prof. Robert Turner and Prof. Rury Holman. When the intervention trial finished in September 1997, all surviving UKPDS patients were entered into a 10-year, post-trial monitoring programme. Since that time, 116 manuscripts have been published using data from UKPDS. One of the initial, landmark findings from UKPDS was that progressive hyperglycaemia is a feature of T2D, despite continued treatment with metformin, sulphonylurea, or insulin, and that β-cell function continues to decline over time. The trial assessed the relationship between glycaemic exposure and the rate of microvascular and myocardial infarction (MI) complications, documenting a significant increase over time. In 1998, UKPDS had 5 simultaneous publications that demonstrated the benefits of intensive or tight glycaemic control on complications as well as a cost analysis of improved blood pressure control in hypertensive patients with T2D. Reducing glucose exposure (HbA_1c 7.0 % versus 7.9 % over median 10.0 years), with sulphonylurea or insulin therapy, reduced the risk of “any diabetes-related endpoint” by 12% and microvascular disease by 25%, with a 16% trend to a reduced risk of MI. In the metformin study, a 0.6% difference in HbA_1c was found to correlate with a 32% reduction in any-diabetes related endpoint, while the blood pressure study showed that a 10/5 mmHg mean difference in blood pressure reduced the risk of any diabetes-related endpoint by 24%, fatal and non-fatal stroke by 44%, microvascular disease by 37% and retinopathy progression by 34%. Following publication of this data from the UKPDS, due to changes in the management of patients, age-standardised diabetes complication rates for stroke and acute MI decreased dramatically.

In later studies, UKPDS reported on a ‘legacy’ effect that early glucose control has on all-cause mortality. Indeed, despite an early loss of glycaemic differences, a continued reduction in microvascular risk and emergent risk reductions for MI and death from any cause were observed during 10 years of post-trial follow-up. Today, it is taken for granted that increased survival is associated with earlier glycaemic reduction. The results of the UKPDS have highlighted that complications in T2D need to be identified and treated early, and that hyperglycaemia is an independent risk factor for coronary heart disease, progressive and secondary to declining β-cell function. Importantly, improved glucose control can substantially reduce the risk of microvascular disease. Given the legacy effect of glycaemic control, glucose-lowering therapies need to be introduced as early as possible to maximise their benefit. Foundation therapy with metformin can substantially reduce cardiovascular risk and all-cause mortality. These findings paved the way for later trials on the benefits of early glucose control and earlier introduction of combination therapy.

From the earliest involvement of health economists in UKPDS, it was evident that some form of diabetes model would be required in order to fully capture all the relevant costs and effects of treatment and to be able to generalise what was learned in UKPDS and other relevant studies in different populations and interventions. It was also clear that the existing models at the time could be improved upon using rich and reliable data from UKPDS.

The objectives of the original UKPDS Outcomes Model were to provide a comprehensive outcomes model for people with T2D in order to predict the risk of major diabetes-related complications and to capture time varying risk factors such as HbA_1c and history of events. It also attempted to estimate lifetime health outcomes and costs of alternative management strategies through their impact on risk factors. The first version of the UKPDS Outcomes Model had good temporal validity in predicting post-trial events and good internal validity. The model also performed well in external predictions using data from PROactive, NHANES and CARDS. In 2013, a major upgrade was made to the model by taking into account more patient years of data of post-trial monitoring, with more events and a longer duration of diabetes, thereby increasing the statistical power. New endpoints were also added such as second MI and second stroke as well as second amputation and ulcer. New risk factors were also added, i.e. eGFR, microalbuminuria, white blood cell count, heart rate and haemoglobin. This resulted in new sets of utility values and costs for each complication. Some of the insights of the UKPDS Outcomes Model have demonstrated that even small changes in risk factors can have worthwhile effects on lifetime outcomes, and that a model which starts at diagnosis is not sufficient.

UKPDS Outcomes Model have helped drive forward diabetes modelling globally. As of September 2018, 278 non-commercial licenses have been issued worldwide together with 33 commercial licenses, including almost all major pharma players in the therapeutic area. Manuscripts from the UKPDS Outcomes Model continue to be the most heavily cited in the field, and most other diabetes models are heavily reliant on UKPDS equations. It has set the standard for transparency, and all equations are fully in the public domain.

In the effort to put the UKPDS into perspective, this session was focused on debunking four common misconceptions about the UKPDS. The first misconception is that the results related to intensive glucose control should not be trusted given that another trial (i.e., ACCORD) demonstrated an increase in death rates upon lowering HbA_1c to normal values. However, to put this into perspective,the differences in trial designs between UKPDS and ACCORD were detailed, and in particular the initiation of glycaemic interventions at diagnosis in patients with no history of a CV event in the former render the two trials different. A second misconception is that UKPDS used old glucose-lowering techniques that are not relevant in today’s interventions. However, it was argued that the UKPDS was about glycaemic ‘policy’ and advocating early, tighter glucose control and the trial was not centered on choice of an individual drug. The third misconception is that the metformin arm was underpowered in UKPDS, and thus it is unclear if the result can be considered valid. To debunk this misapprehension, it was suggested that since patients had to be randomised to metformin, the real number of patients was double that cited by many, and that statistically speaking the results are nonetheless significant. The last misconception is that many recent trials of newer agents (e.g. DPP-4 inhibitors) show no difference in cardiovascular outcomes and therefore lowering glucose is thus not of importance. However, in reality it was pointed out that the majority of the cardiovascular trials for these newer agents were actually safety trials, predicated on glycemic equipoise. As such, such a comparison cannot be justified.