| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters |
Department of Medicine, Wellington School of Medicine, PO Box 7343, Wellington South, New Zealand, Fax 64-4-389-5427, E-mail rob{at}wnmeds.ac.nz
To the Editor:
Readers of articles in biomedical journals may wish to follow up a cited reference to further their knowledge, to verify claims made by the authors, or to obtain details of an analytical method. To easily retrieve the cited article, it is essential that the reference be accurate, especially in regard to the journal title, the year of publication, the volume number, and the first page number.
Previous studies have shown that references in published articles in general medical journals (1)(2) and in specialist biomedical journals (3)(4)(5)(6) frequently contain errors. Reference error rates of 8% to >50% have been found in biomedical journals (1)(3). To our knowledge, no study has been done to assess the accuracy of references in clinical chemistry biomedical journals. The aim of this study was to determine the reference error rate in Clinical Chemistry.
The December 1999 issue of the Journal was selected for the study. It contained 2 Editorials, 1 Review, 23 Articles, 1 Case Conference, 12 Technical Briefs, 5 Letters, and 1 Reply. All of the references cited in these were checked for accuracy with Medline from the National Library of Medicine (www.ncbi.nlm.nih.gov/PubMed). References excluded were those not listed in Medline, such as book chapters, newspapers, website articles, theses, articles in press, and articles before 1966. Errors in the references were categorized as belonging to one of the six main elements of a reference, these being (a) author(s), (b) title, (c) journal, (d) year, (e) volume, and (f) pages. An error was deemed major if the first author’s name was misspelled or missing, the wrong journal was cited, or if the year of publication, the volume (or supplement) number, or the first page number was wrong. The frequency of errors in each of the main reference elements was recorded, and the percentage of references containing any error was calculated.
Of the 1063 cited references in the December 1999 issue of
Clinical Chemistry, 892 references were Medline listed and
checked for accuracy. There were a total of 226 erroneous references,
giving a reference error rate of 25.3% (Table 1
). The most frequently occurring errors were in the author
element, followed by the title element. Two references cited the wrong
journal, 12 had misspellings of the first author’s name or had the
wrong first author listed, 3 listed an incorrect year of publication,
19 listed an incorrect volume number, and 8 listed an incorrect first
page number. Seventy references contained multiple (two to six) errors.
|
This study, the first to determine reference accuracy in the clinical chemistry literature, found a reference error rate of 25.3%. Our study of three general medical journals in Australia and New Zealand found error rates of 34–49%, and reference error rates as high as 66.7% have been reported (3). A limitation of this study is that only one issue of Clinical Chemistry was studied and other issues may contain much lower (or higher) reference error rates. However, 38 of the 45 articles studied contained erroneous references, ranging from a small proportion to virtually all the cited references. It remains the authors’ responsibility to check proofs for errors before publication.
The majority of errors were spelling mistakes in authors’ names or in the title. These errors would not make it difficult to retrieve the cited article. However, if the cited reference contains an error in a critical element of the reference, this would make it more difficult for the reader or a librarian to retrieve the article. In this study, 32 cited references contained a major error that would make it difficult and frustrating to retrieve the article, unless first checked against Medline.
The majority of biomedical journals, including Clinical Chemistry, instruct authors to adhere to the "Uniform Requirements for Manuscripts Submitted to Biomedical Journals" (7). One such requirement is that authors must verify cited references against the original document. This should theoretically ensure that the cited reference is correct. It should also ensure that the cited article is accurately quoted, although this is not always the case (1)(3)(4). The problem of errors in published articles is not confined to cited references, as abstracts have also been shown to frequently contain data errors (8). The majority of biomedical journals place the responsibility of accurate reference lists on the authors, although some American medical journals check references in-house. Other biomedical journals require authors to either submit a photocopy of the first page of cited articles or sign in a covering letter that all cited references have been checked against the original or appropriate data bases (6). Through the use of modern bibliographical aids such as EndnoteTM and Reference ManagerTM, in which references can be downloaded directly from Medline in the required format, accurate reference lists should be possible.
In conclusion, many articles in Clinical Chemistry contain errors in cited references. Emphasis should be directed to authors in the first instance to ensure accuracy of references in their submitted articles. Cited references in published articles can be error free if greater care is taken. Indeed, one Editorial, one Article, one Letter, and four Technical Briefs in the December 1999 issue of Clinical Chemistry had error-free reference lists (9)(10)(11)(12)(13)(14)(15).
Footnotes
Editor’s Note: The accuracy of reference listings is important for investigators and clinicians, and no less so in the online era. In Clinical Chemistry Online, references are linked to the full text of cited articles or to their abstracts at Medline. This linking requires accurate citations.
I examined the reference linking in the first 21 pieces in the December 1999 issue of Clinical Chemistry (the issue studied by Siebers) at www.clinchem.org. Among 440 references to articles in journals that are indexed at Medline, 409 (93%) were linked to full text of the articles or to Medline entries. The remaining 7% that were not linked presumably represent a subset of the 25% of articles in which Siebers found some errors in the citation.
The author of the Letter above examined the same online issue for us. He reports that references were not linked when they had errors in the year, volume number, first page number, journal name (or its abbreviation), or name of the first author. References that were linked included references with errors in co-authors’ names or ending page numbers, spelling errors in the title, simple spelling errors of the first author’s name, transpositions of authors’ names, and omissions of authors’ initials. This information sheds additional light on the types of errors that were the most common.
Authors’ errors in references should become exceedingly rare with the current availability of programs that import citations directly from Medline. We encourage authors to avail themselves of these tools (which also save hours in manuscript preparation). Citation errors are glaringly obvious in the online journal. In the near future, we hope to provide additional electronic tools to help authors to identify errors in their reference listings. It will remain the author’s responsibility to the community to check the accuracy of the references. We cannot know the references that the author has in mind.
—DB
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  D. Aronsky, J. Ransom, and K. Robinson Accuracy of References in Five Biomedical Informatics Journals JAMIA, March 1, 2005; 12(2): 225 - 228. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
