Type 2 Diabetes

by Stephen Fenerty

As the diagnosis rates continue to rise, a world-renowned team analyzes masses of clinical data and gains valuable insight to change treatment around the globe.

Recent years have seen a sharp rise in the incidence of Type 2 diabetes, also called non-insulin dependent diabetes, and the serious complications that often result. Hence the importance of work carried out by the Diabetes Trials Unit (DTU).

Headed by Professor Rury Holman at the Oxford Centre for Diabetes, Endocrinology and Metabolism at the University of Oxford, the DTU is one of Europe’s largest clinical diabetes research groups. Part of the Nuffield Department of Clinical Medicine, it has two main activities: running clinical trials involving different treatments for Type 2 and analyzing results to understand the condition better. A key focus is finding ways to avoid complications that can include blindness, kidney failure, heart disease, stroke and nerve damage resulting in amputation.

"SAS is our main tool for research and analysis," says Dr. Carole Cull, research lecturer for DTU. DTU researchers have gathered data from thousands of patients over 20 years, building a unique data store that has international significance. "Results provided through our use of SAS have helped to change the way Type 2 diabetes is treated worldwide," Cull adds. Medical practitioners, researchers and healthcare planners have been able to improve their treatment and approaches.

Enhancing quality of life
Type 2 diabetes develops when the pancreas fails to produce enough insulin, or when the insulin produced doesn’t work properly. Along with diet and exercise, treatments aim to get patients’ blood glucose and blood pressure levels as close to normal as possible, to improve their well-being and, critically, to avoid complications. (When treated, Type 2 alone is not usually the primary cause of mortality.) Previously, the condition tended to affect people over age 40; however, children, teenagers and people in certain ethnic groups are now being diagnosed, and the condition has been linked to such factors as obesity. The research is therefore vitally important in helping alleviate the misery caused by complications for patients, their families and caregivers.

The "output" of the DTU’s work is twofold: publishing trial reports on the efficacy of drugs and providing more descriptive analyses. "We’re particularly interested in how the condition progresses and the various intervention treatments available," says Cull.

"Through our research we want to improve quality of life in a general sense. This means exploring different aspects and treatments to see how complications can be prevented. Something like a stroke or deteriorating eyesight will seriously affect someone’s quality of life. We also look at side effects that can result from treatments."

The DTU’s use of SAS, which began in 1988, is centered on a unique store of data gathered in the UK Prospective Diabetes Study (UKPDS), a 20-year project involving more than 5,000 patients. "We wanted to see whether more intensive treatment would prevent the onset of complications," says Cull. "This hypothesis was shown to be true, with patients receiving more intense therapy returning to more normal blood glucose levels and experiencing lower rates of complications, particularly eye and renal problems." The results, first published in 1998, have been widely reported, with more than 60 reports appearing in journals such as The Lancet and the BMJ (British Medical Journal). "Almost every number in every paper is generated using SAS," says Cull.

Massive amounts of data
A key reason for using SAS is the large data volumes – some 20 gigabytes representing over 85 million records – and SAS’ capabilities in data handling and cleansing. While other tools could have supported analysis, few could handle such volumes. Allied with this is the quality of both the data and results: With SAS, the DTU has confidence in its output.

Cull continues, "In the UKPDS, we used SAS in two main ways. First, we based standard analyses and modeling of trials data on, for example, predefined clinical end points or the efficacy of certain treatments, using approaches like linear regression and multivariate regression models. Second, SAS was great for the more sophisticated approaches like Cox modeling; it’s particularly good at time-dependent co-variables, for example. You can ask also more descriptive questions, like ‘What happens to a specific group of people as Type 2 progresses?’

"We’re increasingly moving into genetics, to see if genes are connected with the onset of complications," Cull adds. "All of this requires the sophisticated computing SAS provides." Another factor, she says, is that "the FDA prefers to see analyses carried out using SAS, and that helps when it comes to getting regulatory approval for products.

"SAS is widely used in academia. In the journals you only have to count the number of times you see 'work was done using SAS.' I have more confidence that the researchers have produced the correct analyses if they’re carried out with SAS. I’m reassured because they’ve used SAS."

Changing the clinical landscape
"After we published, we saw a material phase shift in the clinical community in how Type 2 diabetes was treated," says Cull. For instance, the National Institute for Clinical Excellence (NICE) now recommends intensive therapy from the start, while consultants at all levels now follow guidelines put in place following the team’s work. "This was a big trial and the results were really quite clear, so most people believed us," says Cull.

SAS is particularly good at handling different data types, and provides many of the facilities the DTU team needs. "In academia, as opposed to commercial research," says Cull, "you need the flexibility to ask nonstandard questions and try out different approaches; SAS provides this." Researchers can work in a more discursive fashion as well as doing straightforward analyses. "And once you know SAS," she adds, "it’s very simple to do something – you don’t have to write lots of code. This is especially useful if, like me, you’re an applied statistician."

Another valuable feature is error checking. "If you do something wrong – if, say, the data selected doesn’t fit the requirements for a particular analysis – SAS usually tells you," says Cull. "This is also useful when training people." Speed is another factor. The university has decentralized its computing so users have major data sets on their desktop PCs. In a recent job, each run took only 10 minutes using SAS – other packages might take three days. As for the future, "we’re hoping SAS Enterprise Guide will enable us to customize our output and produce graphics in a far easier way, something that’s very important in an academic environment," says Cull.

She concludes, "Across the world, people involved in treating diabetes all talk about the UKPDS and our work, and we really believe it has resulted in a far better outlook for patients."