THE ROLE OF STATISTICAL REVIEW
IN A NEW JOURNAL
Julia Brown,1 Janet Dunn,2 William Jones,3 David Machin,4 and Patricia Shevlin1
1Yorkshire Clinical Trials and Research Unit, Arlington House, Hospital, Leeds LS16 6QB, UK; 2CRC Clinical Trials Unit, University of Birmingham, UK; 3United Leeds Teaching Hospitals, NHS Trust, UK; 4MRC Clinical Trials and Epidemiology Research Unit, Singapore
Clinical Oncology is a new journal (approximately 8 years old) covering all aspects of the management of patients with cancer. It is an international journal with contributions and subscriptions from many countries and an institutional subscription base of 202. It began a policy of statistical review of manuscripts in 1993. Three statisticians with experience in cancer research have led the review process. Articles that require epidemiologic input are referred to appropriate experts. All 3 statisticians are on the editorial board. All manuscripts are reviewed. There have been difficulties. One problem was how to decide at what level to set the statistical standards in the early days when the journal needed to have material to publish. A gradual process of raising standards was adopted. The review process has meant large volume of work, frustration at lack of time for detailed review, lack of recognition by employers of this contribution, and slow progress in achieving higher standards. However, over the 3 years improvements have been seen. In 1993, 35% of articles were either rejected or required revision. In 1996, this had risen to 70% mainly as a result of the statistical input. In 1994, statisticians accepted 65% of the manuscripts without revision and returned 29% for revision to be rereviewed. The corresponding figures in 1996 were 21% and 59%, respectively. The editor has final say but few statistical recommendations have been overruled. Statistical guidelines have been published.
Return to Session Information
