I recently read a paper by some colleagues at SPRU on the publication patterns of the pharmaceutical industry. I thought I’d share a short write up of my notes. A version of the paper is here (pdf) or full citation:
- Rafols, Ismael, Hopkins, Michael M, Hoekman, Jarno, Siepel, Josh, O’Hare, Alice, Perianes-Rodríguez, Antonio and Nightingale, Paul (2012) Big pharma, little science? A bibliometric perspective on big pharma’s R&D decline. Technological Forecasting and Social Change.
You might be forgiven for thinking the pharmaceutical industry doesn’t publish research. It does. Quite a bit. “Big Pharma” (the world’s 15 largest firms) have around 10,000 publications. It’s still only about 4% of total publications in the field (and to put that 4% in some context, Big Pharma firms invested £46 billion per year in research and development between 2004-07, compared to the National Institute of Health’s £15.25b). But looking at some of the whats, whens and wheres of Big Pharma publishing can be interesting. They seem to be publishing around 9% less than they did 15 years ago, in slightly different areas, with evidence of different collaborations in ways which suggest they are a bit more of a follower than a leader when it comes to science. Although the issue is very complex, as the paper notes in its conclusion, is Big Pharma is not doing the research itself, will they be able to continue to be able to justify their large profits?
The researchers used Web of Science to download publications – “article” “letter” “note” “proceeding paper” and ”review” – from the largest 15 European and American pharmaceutical firms, amounting to a total of 160,841 records for the 1995-2009 period. Note: Here “European” meant they had an affiliation to at least one European Free Trade Association country (EU + Switzerland, Norway, Lichtenstein and Iceland). Though I’m personally taken by the idea of applying a Eurovision view of “European” in research policy data collection, I guess the politics of Israeli publishing might be a complicating factor. Also it’d just lead to really bad null points jokes in seminars. The data was then analysed with VantagePoint software, and they used Pajek and VOSviewer to create visualisations.
There are different motivations to publish in the drug discovery and drug development stages. In the discovery phase, publications can signal to investors but participating in the more traditional exchanges of “open science” (open in that they are published, even if there’s a paywall) also allows industrially based scientists access to the resources of the scientific community. In drug development, a degree of openness is often required due to the need for clinical evidence for the uptake of innovations. The authors argue that the evidence-based medicine paradigm has encouraged this (though some might argue not enough). They also suggest that scientific publications are sometimes written to diffuse information about the effectiveness and safety of pharmaceuticals; acting as a marketing tool, to win support in regulatory/ policy areas.
So, what’s Big Pharma publishing? More in rheumatology and ophthalmology (up 190% and 122% respectively), perhaps reflecting areas of increased therapeutic interest. Interestingly, also up was health policy and services (151%) reflecting greater role in fields closer to patient (or perhaps closer to the market) such as health services and clinical research. The industry’s shift away from agrichemicals and materials explains drop in plant (down 62%) and polymer science (down 71%). Chemistry, in comparison, was relaivly stable, but biomedicine was down 40% and cell biology down 46.5%. Put this data through visualization, and you get a pretty map of threads and bubbles lets you see a larger picture of a diversification of research interests. Also, the drop in biomed sciences can be seen in context of increase in peripheral areas related to new techniques (e.g. computation bio) and a marked shift towards disciplines more orientated to clinical applications. As the conclusion put it; we can see an apparent shift from bench to bedside.
There also seems to be some evidence of the organisational shifts many others have noted of Big Pharma in recent years. E.g. there’s a lot of talk of a shift to “team science” of collaboration, especially in biological sciences, medicine and neuroscience. This data suggests it’s even stronger in big pharma. However, interestingly it seems to be external partners which seem to be taking the intellectual lead: there’s a distinctive decrease in the number of big pharma first authors in collaborative publications (from 43% to 35%). The researchers conclude that the picture of Big Pharma their data presents is that of an intellectual follower, not leader.
We should be careful of drawing conclusions on this data as a simple proxy for what’s happen in Big Pharma. For example, there was no obvious sign of off-shoring of research to countries outside US and Europe (e.g. Singapore, India and China). However, the researchers suggest that due to division of labour in this particular collaborations and which part of that labor gets author credit, this is one of the areas where publication data underestimates actual activity. Also interestingly, although the oncology publishing increase is consistent with other data on pharmaceutical projects, there was increase in cardiovascular publishing despite decrease in number of projects. As Brian Balmer’s book on secrecy in science tells us, there are complex, often inter-mingling reasons for opening and closing scientific research. Similarly, there are a complex of reasons for publishing and not publishing. Still, it tells us something things and suggests others and if you’re interested it’s worth having a look at the full paper, including appendix.