The largest study yet assessing the success rates of clinical trials for drugs has yielded more optimistic data than previous studies.

Andrew W. Lo and colleagues at the MIT Laboratory for Financial Engineering used an automated algorithm to draw data from nearly 186,000 unique trials of over 21,000 compounds, from January 2000 through October 2015. The team found that 13.8% of drug programs make it from Phase I to approval, a higher figure than the often-touted estimates of 5% or 10% (Biostatistics 2018, DOI: 10.1093/biostatistics/kxx069).

The figure is even higher—20.9%—if cancer trials are excluded. That’s because cancer drugs had the lowest success rate, a mere 3.4%.

Cancer trials that used biomarkers had higher success rates, though, and the overall oncology drug approval rate for 2015 was 8.3%, which the authors surmise was due to the booming immuno-oncology field.

Vaccines for infectious disease and ophthalmology drugs had the highest success rates, with 33.4% and 32.6% of Phase I compounds, respectively, eventually approved.

“It is encouraging to see that it confirms our recent observations of improvement in the overall probability of success,” says Katarzyna Smietana of McKinsey & Co. She was an author of an analysis finding that the success rate of moving from a Phase I trial to product launch was 11.6% from 2012 to 2014, up from 7.5% between 2008 and 2011. In both periods, the success rate for biologics was twice as high as that for small molecules (Nat. Rev. Drug Discovery 2016, DOI: 10.1038/nrd.2016.85).

A 2016 report from the Biotechnology Innovation Organization and Biomedtracker calculated a 9.6% success rate for drug development programs between 2006 and 2015.

Joseph A. DiMasi, director of economic analysis at the Tufts Center for the Study of Drug Development, cautions that “making precise comparisons” between studies of drug development success is impossible due to differences in how samples were included in the studies.