We disagree with Joshua Nicholson and John Ioannidis' claim that the peer-review system of the US National Institutes of Health (NIH) works against genuinely innovative research, because we believe that their analysis is flawed (Nature 492, 34–36; 2012).

They use 1,000 or more citations as a proxy for identifying breakthrough discoveries. However, more than 60% of the 158 highly cited articles they analyse cannot be categorized as innovative primary research: 34% are reviews, reports, clinical guidelines or descriptions of resources; 18% are clinical trials of the type primarily funded by industry; and 11% fall outside the NIH mandate that research should have the potential to improve human health (for our analysis, see http://dpcpsi.nih.gov/opa/natcorr).

Excluding these articles leaves 58 of the original 158: of these, 83% were funded by the NIH and 17% were funded by private industry.

The NIH welcomes further innovation in biomedical research, as evidenced by funding mechanisms such as the NIH Director's Pioneer, Transformative Research, Early Independence and New Innovator awards.