Making the case for big data: Swedish study offers real evidence of real world data’s impact
22.01.14
A lot has been said about big data, and the opportunities and challenges it poses – but people sometimes wonder what lies behind the buzzword. We believe that if we harness the power of big data in the long run, we will benefit – but what’s so convincing about the case for big data anyway?
One concrete example comes from diabetes care, and how this was improved thanks to knowledge creation in real patients. A study released late last year in my native Sweden shows how big data can drastically enhance knowledge about a disease and work towards improved care solutions. Diabetes affects a huge amount of people – some 55.4 million Europeans had diabetes in 2010[1]alone. The disease leads to major complication, causing eye, kidney, and nerve damage. Patients tend to die earlier than the general population because of complications, mostly from cardiovascular disease[2].
Major shifts in diabetes treatment occurred in 1922 with the introduction of insulin, and in the late 1950s with the introduction of metformin[3] and sulfonylureas[4]. Based on a historical analysis of diabetic patients in Sweden, the study I’m talking about clearly shows the tremendous impact these treatments had in decreasing mortality rates. But looking at the two graphs shown below, something struck me. (Note the blue line represents the general population; the red line the population with Type 1 or Type 2 Diabetes, depending on the graph). The introduction of insulin in 1922 clearly brought a striking improvement in life expectancy for type-1 diabetes – even though the gap with the general population has not been caught up until today. In contrast, and looking at the curve for diabetes type II, the introduction of metformin and sulfonylureas in the late 50s does not seem to have created a major shift – on the other hand the gap with the general population is so narrow today that people can expect a similar life expectancy than the general population.
Despite this evidence, the gaps in life expectancy between patients and the general population were not significantly closed until today[E1] . Why is this the case? Simply put, the full value of these treatments wasn’t immediately recognised when they were first launched. Although these treatments were available for a long time, already in the 50s, their real value was only realised after examining data generated from thousands of patients diagnosed with diabetes. It wasn’t until the early 1990s that sufficient evidence was available to demonstrate that diabetes management was helping to reduce cardiovascular events – a major factor in mortality. It took real world information to inspire the international UKPDS study, which in 1998 demonstrated that, any improvement in glycaemic control and blood pressure reduces diabetes-related complications. The UKPDS – together with the DCCT, a prospective clinical trial that later confirmed its findings – was critical in revising the guidelines to diabetes care.
Researchers did not realise the full impact these treatments were having until they could examine mass amounts of “real world” data gathered from patients – this resulted in a revelation that improved our knowledge of which treatments work best. That’s exactly what this type of data mining is about:Enhancing knowledge based on significant amounts of information. In the world of diseases, this knowledge is what we need to improve care and, ultimately, benefit patients.
Since then, many more innovative treatments for diabetes have been developed and put on the market, and we can be hopeful that the one-year gap between the life expectancy of patients with diabetes and the general population will be closed one day. Big data will also be helpful there: The authors note the promising work being done in Sweden towards developing methods for utilising observational data to research new diabetes drugs. This will hopefully again help influence treatment patterns for the benefit of patients in the future.
Remaining life expectancy for a 10-year-old with type 1 diabetes compared with a 10-year-old in the population for 100 years
Remaining life expectancy for persons 50 years of age with type 2 diabetes compared to 50-year-olds in the population for 100 years
________________________________________
[1] http://www.idf.org/diabetesatlas/europe#footnote-2
[2] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2014359/
[3] which works by suppressing glucose production by the liver
[4] which works by increasing insulin release from the beta cells in the pancreas
