Article

Statistics in medicine

GULLER, U, BUEHLER, Leo Hans, CLAVIEN, P A

Reference

GULLER, U, BUEHLER, Leo Hans, CLAVIEN, P A. Statistics in medicine. Swiss Medical Weekly, 2003, vol. 133, no. 39-40, p. 521

DOI : 10.2003/39/smw-10504 PMID : 14655051

Available at: http://archive-ouverte.unige.ch/unige:45061

Disclaimer: layout of this document may differ from the published version.

1 / 1 Editorial SWISS MED WKLY 2003;133:521 · www.smw.ch 521

Statistics in medicine

U. Guller, L. Buhler, P. A. Clavien

For most of its history, the practice of medi- as the difference between standard deviation and cine has been a profoundly empirical enterprise. standard error of the mean, over-analysis of data While this empiricism continues by necessity to by doing multiple comparisons, and putative dis- exist in the clinical environment, the advent of crepancies between statistical significance and scientifically rigorous epidemiology has tremen- clinical importance are emphasised. A simple and dously changed medical research in the 20th cen- intuitive hypothetical example of a randomised tury. controlled trial comparing post-operative compli- The driving force behind the maturation of an cations after two surgical procedures highlights epidemiological approach to medicine has been the tremendous importance of the appropriate as- the incorporation of statistics in modern medical sumption of whether or not the data are normally research, a practice that has become mandatory in distributed. past decades [1, 2]. Sound statistical methods are Dr. Sprent’s manuscript represents an expan- essential to medical science, as they transform un- sion on his statistical lectures during the first “Sci- interpretable raw data into meaningful results [3]. ence Writing Course” organised by the Research However, trends toward evidence-based medicine Section of the Swiss Surgical Society (RS-SSS) in can only flourish in a culture of statistical literacy. May 2003. The primary objective of this course Such a culture requires physicians who are was to teach young physicians with a potential in- equipped with the knowledge and skills to critically terest in research and an academic career how to and accurately interpret statistics [2, 4–6]. write a scientific manuscript as well as to provide Unfortunately, there is ample evidence that some basic statistical knowledge. The first course many physicians are poorly prepared for accurately enjoyed an overwhelmingly positive feed-back interpreting statistical computations in medical lit- from the participants, which has stimulated the erature [2, 7, 8]. The vast majority of medical RS-SSS to organise further courses. From 2004, schools and residency programs in Switzerland do we plan to organise two courses per year, one fo- not provide appropriate training in the under- cusing on basic knowledge of statistics and epi- standing and application of statistics in medicine. demiology, mainly intended for junior physicians Moreover, it has been reported that the little sta- interested in clinical or basic research, and a sec- tistical knowledge acquired during medical school ond one for more advanced investigators covering is rapidly lost in the years thereafter [7]. Given the other more sophisticated aspects of statistics and ever-increasing importance of evidence-based grant writing. practices, such a lack has potentially grave impli- Basic knowledge about statistical computa- cations for the medical community. tions in medical literature is invaluable for the crit- Dr Sprent’s manuscript “Statistics in Medical ical assessment of scientific findings and their im- Research” is an excellent start to compensate at plementation in clinical practice. The learning least partially for the unfortunate lack of statistical curve for the appropriate interpretation of statis- training during medical school or residency. It rep- tics is, however, steep and the process highly iter- resents a concise and very helpful summary of the ative. Dr. Sprent’s manuscript only scratches the most relevant basic concepts of statistical compu- surface of statistics in medicine. Nonetheless, it is tations in medical research. Simple, yet frequently certain that this publication will facilitate the crit- misunderstood notions such as p-values, 95% con- ical appraisal of medical literature and hopefully fidence intervals, error of first and second kind, represents a stimulus for physicians to further and power computations are explained in easily deepen their ability in the interpretation of statis- understandable language. Important caveats such tical computations.

References

1 Altman DG. Statistics in medical journals: developments in the tice of medicine. Evidence-Based Medicine Working Group. 1980s. Stat Med 1991;10:1897–913. JAMA 1992;268:2420–5. 2 Hayden GF. Biostatistical trends in Pediatrics: implications for 6 Guyatt GH, Rennie D. Users’ guides to the medical literature. the future. Pediatrics 1983;72:84–7. JAMA 1993;270:2096–7. 3 Rothman KJ. A show of confidence. N Engl J Med 1978;299: 7 Berwick DM, Fineberg HV, Weinstein MC. When doctors meet 1362–3. numbers. Am J Med 1981;71:991–8. 4 How to read clinical journals: I. why to read them and how to 8 Friedman SB, Phillips S. What’s the difference? Pediatric resi- start reading them critically. Can Med Assoc J 1981;124:555–8. dents and their inaccurate concepts regarding statistics. Pedi- 5 Evidence-based medicine. A new approach to teaching the prac- atrics 1981;68:644–6. Swiss Swiss Medical Weekly: Call for papers Medical Weekly

Official journal of the Swiss Society of Infectious disease the Swiss Society of Internal Medicine the Swiss Respiratory Society

The many reasons why you should choose SMW to publish your research

What Swiss Medical Weekly has to offer: Editorial Board Prof. Jean-Michel Dayer, Geneva • SMW’s has been steadily Prof. Peter Gehr, Berne rising, to the current 1.537 Prof. André P. Perruchoud, Basel • to the publication via Prof. Andreas Schaffner, Zurich the Internet, therefore wide audience (Editor in chief) and impact Prof. Werner Straub, Berne • Rapid listing in Medline Prof. Ludwig von Segesser, Lausanne • LinkOut-button from PubMed with link to the full text International Advisory Committee website http://www.smw.ch (direct link Prof. K. E. Juhani Airaksinen, Turku, Finland from each SMW record in PubMed) Prof. Anthony Bayes de Luna, Barcelona, Spain Prof. Hubert E. Blum, Freiburg, Germany • No-nonsense submission – you submit Prof. Walter E. Haefeli, Heidelberg, Germany a single copy of your manuscript by Prof. Nino Kuenzli, Los Angeles, USA e-mail attachment Prof. René Lutter, Amsterdam, • based on a broad spectrum The Netherlands of international academic referees Prof. Claude Martin, Marseille, France • Assistance of our professional statistician Prof. Josef Patsch, Innsbruck, Austria for every article with statistical analyses Prof. Luigi Tavazzi, Pavia, Italy

• Fast peer review, by e-mail exchange with We evaluate manuscripts of broad clinical the referees interest from all specialities, including experi- • Prompt decisions based on weekly confer- mental medicine and clinical investigation. ences of the Editorial Board • Prompt notification on the status of your We look forward to receiving your paper! manuscript by e-mail • Professional English copy editing Guidelines for authors: • No page charges and attractive colour http://www.smw.ch/set_authors.html offprints at no extra cost

Impact factor Swiss Medical Weekly 2 1.8 1.537 Editores Medicorum Helveticorum 1.6 1.4 1.2 1.162 1 All manuscripts should be sent in electronic form, to: 0.8 0.770 0.6 EMH Swiss Medical Publishers Ltd. 0.4 SMW Editorial Secretariat 0.2 Farnsburgerstrasse 8 0 CH-4132 Muttenz

1995 1996 1997 1998 1999 2000 2002 2003 2004 Manuscripts: [email protected] Letters to the editor: [email protected] Schweiz Med Wochenschr (1871–2000) Editorial Board: [email protected] Swiss Med Wkly (continues Schweiz Med Wochenschr from 2001) Internet: http://www.smw.ch