Real world data: a gain for benefit assessments? This question was the focus of the Autumn Symposium of the German Institute for Quality and Efficiency in Health Care (IQWiG), which was held in Cologne at the end of November 2015. Speakers from Germany and abroad presented their different points of view on this topic. The presentations are now available on iqwig.de as multimedia presentations or PDF documents for downloading.
Where's the beef?
Besides results from allegedly "artificial" randomized controlled trials (RCTs), does one also need "real world" health care data for benefit assessments? Does the reference to later data from the real world suffice to accelerate drug approval – and thus base approval on weaker evidence? What actually characterizes such data obtained from registries or routine health care? Where's the beef in real world data? These were the questions that the Institute's Director Jürgen Windeler asked at this year's Autumn Symposium and that were in part discussed controversially by an enthusiastic audience.
The term "real world data" is dispensable
However, despite all the differing points of view, a consensus existed that the term "real world data" is inaccurate and incorrect. This is because, on the one hand, the data obtained in this way by no means always reflect routine health care. On the other, data obtained from RCTs indeed originate from real life. As the discussion showed, the term "real world data" is ultimately a synonym for everything that is not an RCT, i.e. for "non-RCTs". But then the term "real world data" is dispensable.
There was also agreement that RCTs are best suited to draw valid conclusions on the benefit of a medical intervention: Non-RCTs can have a supplementary function – whereby after the discussion it still remained unclear precisely what this function is.
Standards should not be lowered
In his closing speech at the Autumn Symposium Jürgen Windeler summarized: "Fair comparisons are the crucial principle of evidence-based medicine. If non-RCTs fulfil this requirement, we can also use them for benefit assessments. However, this will only be the case in very specific situations and constellations. But what I currently observe is that the standards achieved are being abandoned and without any need, an increased uncertainty is supposed to be accepted. I see this tendency not only in sections of research, but also in politics. We have to counteract this."