Appointment of statistical editor and quality of statistics in a small medical journal.

AIM: To test if the appointment of a statistical editor improves the quality of manuscripts published in a small general medical journal.

METHODS: Retrospective review of all manuscripts containing statistical data published in the Croatian Medical Journal between 1992 and 2000 (n=241). Statistical analysis and its presentation were assessed by a single observer.

RESULTS: Before the appointment of statistical editor in 1996, 97 manuscripts with statistical data were published. Statistics was not satisfactory in 52 (54%) of them, including 26 definite errors in analysis and 43 in presentation. After the appointment of statistical editor, 144 manuscripts containing statistical data were published. Statistics was not satisfactory in 91 (63%) of them, with 51 definite errors in analysis and 69 in presentation. Out of 144 manuscripts, the editor-in-chief sent out 30 (21%) for statistical review. Statistics was not satisfactory in 25 of them, including 11 definite errors in analysis and 17 in presentation. Statistical editors comments improved three manuscripts. If the authors had acknowledged all statistical editors suggestions, 9 more manuscripts would have been improved. Statistical editor had a total of 195 comments on 30 published manuscripts. Most numerous were the comments concerning the presentation of the statistical analysis (51%), followed by the general comments (26%), comments on analysis (11%), study design (8%), and interpretation (4%).

CONCLUSION: Appointment of a statistical editor is not a guarantee of improvement of statistics in small journals. Other measures are necessary, including strict editorial policy on statistical review, monitoring of revised manuscript versions, and enrollment of formally trained biostatisticians.