Cancer research papers with industry ties more likely to report favourable results

The paper by Jagsi in Cancer is possibly an example of classical
confounding. The industry-funded trials may differ from non-industry
funded trials in a way (e.g the treatments used) that is associated with
the outcome. The treatments they compare are actually more effective.

The authors themselves note this, though not very clearly. Industry
funded trials may not distort results at all (which is not the impression
left in the reader's mind after reading the BMJ), but they address
different questions.

They may choose to investigate areas of cancer where success is
likely to be greater. The questions they address and the designs of the
studies may be different. This is bias, but of a very different nature to
the idea that they distort results. The evidence is they tend to interpret
similar results with a more positive spin, but the results themselves are
not distorted.

Had Jagsi et al compared like with like in terms of the types of
trials they may (or may not) have found similar results. The paper in
Cancer has not answered the correct question.

Competing interests:
None declared