Search

Literature references contain a considerable number of inaccuracies, both in citations and in quotations. We investigated the accuracy of quotations and references in Nederlands Tijdschrift voor Geneeskunde. We checked at random 100 references of articles presented for publication during a period of 4 weeks using Index Medicus. The citation and quotation of another 100 references (50 with a reading list of less than 13, and 50 with a reading list of more than 24 articles) of published original articles were checked, using the original article. In the references of articles presented for publication 70% contained one or more inaccuracies. In the published articles 31% was not completely correct. In 5% these errors prevented immediate identification of the cited article. In one of every three articles a citation error was made: 5 trivial errors, 16 slightly misleading and 12 seriously misleading errors were found. We could not find a relation between the number of citation errors and the length of the reading list. The original source was not cited in 8 articles. Our conclusion is that the number of inaccuracies, in references as well as in citations, is high; authors should check this part of their articles more carefully. Notably the use of multiple references after a sentence with a number of statements is to be avoided.

To deduce recommendations for authors which decrease the rejection probability we investigated retrospectively which reasons were mentioned in 1990 by the editorial board of the 'Nederlands Tijdschrift voor Geneeskunde' when it rejected original articles, reviews, clinical lessons and case reports, and whether length of the article, professional status of the first author, most relevant specialism and origin of the article influenced the rejection probability. Out of 547 articles 38% (208) were rejected; case reports were rejected most often (49%), reviews least often (33%). Peer review of original articles was the most thorough, of case studies the least. The most frequent reason for rejection in all categories was 'substantial shortcomings', particularly in the original articles (80%). Rejection of clinical lessons and case studies was relatively often due to criteria concerning the clinical message. Both rejected and accepted articles on average exceeded the length limit. Professional status and specialism were associated with a difference in rejection probability (chi 2, p less than 0.05). Specialists had the lowest rejection probability (31%). Nonspecialists benefited from specialist supervision. Articles originating from non-academic institutions were more often rejected than articles from academic centres. We advise authors to judge their articles by means of the criteria mentioned. Consultation of (methodological) experts, literature data bases, and (or) the editorial board can improve their chances. Non-specialists can benefit from the experience and expertise of specialists. In general, scientific attitude and willingness to alter the article according to editorial advice appear to have beneficial results.

Objective: To determine and to compare the information contained in structured abstracts in original articles in Annals of Internal Medicine (AIM), British Medical Journal (BMJ), New England Journal of Medicine (NEJM) and of abstracts that have been published in Nederlands Tijdschrift voor Geneeskunde (NTvG)., Setting: Editorial office NTvG., Design: Descriptive., Material and Method: Critical reading of 15 original articles of AIM (5), BMJ (5) and NEJM (5) in 5 consecutive issues starting from 1 March 1990 and all original articles of 6 consecutive issues of NTvG starting from 7 april 1990. According to the criteria of the 'ad hoc working group for critical appraisal of the medical literature' the amount of information in all foreign abstracts and 10 selected abstracts of NTvG were determined. The authors made structured abstracts of all read articles of NTvG., Results: Structured abstracts of AIM, BMJ and NEJM are clear and detailed. However information about key sociodemographic features of patients, their selection and the way of statistical analysis of the results was often missing. In NTvG articles essential information was lacking with respect to objective, design, setting of the study, sociodemographic features of patients, and patients and methods. Structured abstracts of articles in NTvG take more space: an average of 266 instead of 164 words., Conclusion: Structured abstracts take more space but are more informative. Authors of original articles in NTvG are advised to prepare structured abstracts. Editors and peer reviewers should ascertain that no essential information is lacking.