PRESIDENTIAL ADDRESS: XXI International Biometric Conference, Freiburg, Germany, July 2002
Are Statistical Contributions to Medicine Undervalued?
Article first published online: 24 MAR 2003
Volume 59, Issue 1, pages 1–8, March 2003
How to Cite
Breslow, N. E. (2003), Are Statistical Contributions to Medicine Undervalued?. Biometrics, 59: 1–8. doi: 10.1111/1541-0420.00001
- Issue published online: 24 MAR 2003
- Article first published online: 24 MAR 2003
- Received October 2002. Revised October 2002. Accepted October 2002.
- Black box;
- Nobel Prize;
- Observational studies;
- Ronald Ross
Summary. Econometricians Daniel McFadden and James Heckman won the 2000 Nobel Prize in economics for their work on discrete choice models and selection bias. Statisticians and epidemiologists have made similar contributions to medicine with their work on case-control studies, analysis of incomplete data, and causal inference. In spite of repeated nominations of such eminent figures as Bradford Hill and Richard Doll, however, the Nobel Prize in physiology and medicine has never been awarded for work in biostatistics or epidemiology. (The “exception who proves the rule” is Ronald Ross, who, in 1902, won the second medical Nobel for his discovery that the mosquito was the vector for malaria. Ross then went on to develop the mathematics of epidemic theory—which he considered his most important scientific contribution—and applied his insights to malaria control programs.) The low esteem accorded epidemiology and biostatistics in some medical circles, and increasingly among the public, correlates highly with the contradictory results from observational studies that are displayed so prominently in the lay press. In spite of its demonstrated efficacy in saving lives, the “black box” approach of risk factor epidemiology is not well respected. To correct these unfortunate perceptions, statisticians would do well to follow more closely their own teachings: conduct larger, fewer studies designed to test specific hypotheses, follow strict protocols for study design and analysis, better integrate statistical findings with those from the laboratory, and exercise greater caution in promoting apparently positive results.