Background: Academic biomedical journals use peer review and editing to help to select and improve the quality of articles. We have investigated whether articles accepted by the Nederlands Tijdschrift voor Geneeskunde, the Dutch Journal of Medicine, were improved after peer review and editing (post-acceptance scientific and copy editing).
Methods: 400 readers of the journal (100 each of medical students, recent medical graduates, general practitioners, and specialists) were invited to participate in a questionnaire survey. The first 25 from each group who agreed to participate were included. We posted a pack containing a set of identically appearing typescripts (ie, blinding) of the submitted, accepted, and published versions of 50 articles that had been published in Ned Tijdschr Geneeskd. Each evaluator received two of the sets of versions, and each set was evaluated by one person from each group. The package also included two questionnaires: the first was used to compare the submitted with the accepted version (25 questions), the second compared the accepted with the published version (17 questions). The questions were answered on five-point scales, and were about the quality of the articles or were general/overall scores. We analysed the data as scores of 3-5 (ie, improvement) versus 1-2.
Findings: After peer review, the quality in 14 of 23 questions (61%) was significantly improved (p = 0.03 or smaller). In particular, the overall score and general medical value were significantly improved (p = 0.00001 for each). Editing led to significant improvement in 11 of 16 questions (69%, p = 0.017 or smaller), and especially in style and readability (p = 0.001 and p = 0.004). Generally, we found no differences between the scores of the four categories of evaluators. 72% of the evaluators correctly identified which version was which.
Interpretation: Evaluations by readers of the Ned Tijdschr Geneeskd indicated significant improvement of published articles after both peer review and editing. We think that peer review and editing are worthwhile tasks. We also think that possible biases would have had a negligible effect on our results (including the fact that we selected the first 25 evaluators who responded, that some evaluators may have read the published version, and that one questionnaire may have looked more scientific than the other, more editorial one).