From bedside to bench: Comroe and dripps revisited

Abstract

Twenty-five years ago a paper published in Science by Julius Comroe and Robert Dripps purported to demonstrate that 41 per cent of all articles judged to be essential for later clinical advances were not clinically oriented at the time of the study and 62 per cent of key articles were the result of basic research.

Since that analysis, support for basic research has increased in the G7 countries. In the UK, Research Council expenditure on basic research has increased from a low of £444 million (or 42 per cent of total civil R&D) in 1991/92 to £769 million (or 61 per cent of total civil R&D) in 1998/99. Although it would be difficult to argue that Comroe and Dripps were directly responsible for a strategic shift (or drift) in the type of science supported by research funders, their arguments are often cited (albeit at times implicitly) in support of the increased funding for basic biomedical research.

In 1987 Richard Smith wrote a critical paper reassessing Comroe and Dripps. His main argument was that the original study was in itself ‘unscientific’ and that it should be “followed by bigger and better studies”. This study is, in part, an answer to that challenge.

Given the increased support for basic research, and the apparent importance based on the work of Comroe and Dripps, we felt it was important to investigate Smith’s comments by replicating Comroe and Dripps’s study and at the same time try to improve upon the methodology. The current project had two objectives: 1. To see if the original Comroe and Dripps’s methodology was ‘replicable’. 2. To validate the key findings of Comroe and Dripps.

By looking at neonatal intensive care (NIC), we concluded that Comroe and Dripps’ study – as reported – is not repeatable, reliable or valid, and thus is an insufficient evidence base for increased expenditure on basic biomedical research. We did, however, develop an alternative methodology which used bibliographic databases and bibliometric techniques to describe the research underpinning five of the most important clinical advances in NIC, as identified through a Delphi survey.

Using the revised bibliometric protocol, we demonstrated that after a time-lag of about 17 years, between 2 and 21 per cent of research underpinning the clinical advances could be described as basic. This observation is at odds with Comroe and Dripps’s finding that 62 per cent of key research articles judged to be essential for latter clinical advance were the result of basic research.

In reaching this conclusion we are acutely aware of the significant limitations to the revised methodology and, therefore, we caution against the over-interpretation of our results. However, we would argue that there needs to be a greater understanding of how basic research supports healthcare and hope this report will inform part of this wider debate.