THE NATURE OF STATISTICAL INPUT IN
MEDICAL RESEARCH
Douglas G Altman,1 Steven N Goodman,2 Frank Davidoff,3 and Richard Smith4
1ICRF Medical Statistics Group, Centre for Statistics in Medicine, Institute of Health Sciences, Old Road, Headington, Oxford OX3 7LF, UK; 2Division of Biostatistics, Johns Hopkins University, Baltimore, USA; 3Annals of Internal Medicine, Philadelphia, USA;4 and British Medical Journal, London, UK
Objective: To investigate the nature of statistical input in medical research projects and its relation to the statistical methods used.
Design: As an initial study of the nature of statistical input in medical research, we are conducting a survey of authors who submit papers to the British Medical Journal and Annals of Internal Medicine. Authors of papers that report the results of original research are being asked to complete a short questionnaire about the statistical input to their study. They are assured that completion of this will not affect the editorial processing of their paper. Aspects being investigated include the availability of a statistical consultant, affiliation, the stage of the research project when the statistical consultant became involved, and their contribution to the paper, or the reasons why no statistical consultant was brought in at any stage. We intend to include several hundred papers.
Results: Answers to the questionnaire will be tabulated, and studied in relation to which statistical methods were used in the paper. Different types of study (eg, randomized trials) will be investigated separately. The data are being collected during the first part of 1997.
Conclusions: This study will provide important information about the nature of statistical input into medical research projects. The study is a prelude to a fuller investigation of the relation between the nature of statistical input and both the quality of the resulting papers and their chance of acceptance for publication.
Return to Session Information
