J.R. Davids and D.M. Weigl reply:
We thank Mr. Barnes for his careful review of our study and his thoughtful comments. To our embarrassment, and despite our best efforts, our article on reference accuracy contained a citation error. In the classification scheme employed in our study, this would be considered a major citation error. The reference in question was itself a letter to the editor of the BMJ concerning reference accuracy. This letter was not accessible through PubMed1 but was identified through our review of the literature and retrieved manually. The citation error is therefore attributable to human error on the part of the senior author.
With respect to citation errors, it is our opinion that the impact of literature search engines and bibliographic software is generally positive. Recurrent manual collection and documentation of literature references creates repeated opportunity for new errors of citation. As noted in our study, the importance of citation errors is minimized by the strength of these technologies.
Mr. Barnes also raises the issue of quotation errors, and we agree that they are more insidious and present a more difficult problem to solve. Stakeholders in addressing the problem of reference accuracy include the authors, the journals, and the readers. Direct verification of references for citation and quotation accuracy is certainly the responsibility of the authors. Responsible parties at the journal level include the editors and reviewers. Our study confirmed that post-submission, prepublication technical editing can improve citation accuracy. It is our opinion that clarification of the role of references, limiting the number and type of references permitted (e.g., to primary sources only), and republication of classic articles (to promote review of primary sources, in individual or journal club settings) could have a positive impact on quotation accuracy as well. Peer reviewers generally serve on a voluntary basis and may be motivated to do so for a variety of reasons, including professionalism, prestige, and academic career advancement. To expect peer reviewers to monitor reference accuracy seems unrealistic given the additional time that would be required. Finally, the readers should be aware of the issues and challenges related to reference accuracy. Residency training programs should support journal club activities (e.g., current, topical, and classic literature formats), where critical review of the literature by experienced mentors establishes a "caveat emptor" approach for the physicians in training.