Endorsement of the STARD Statement by Biomedical Journals: Survey of Instructions for Authors

To the Editor:

The Standards for the Reporting of Diagnostic accuracy studies (STARD) statement was first published in 2003 (1)(2). In February 2006, the International Committee of Medical Journal Editors decided to encourage the use of the STARD statement by including it in their Uniform Requirements for Manuscripts Submitted to Biomedical journals (www.icmje.com).

The STARD statement contains a checklist of 25 recommended items for the reporting of diagnostic accuracy studies and encourages the use of a flow diagram to represent the design of the study and the flow of patients through the study (1)(2).

Early evaluation of the STARD statement showed that the quality of diagnostic research reports had improved slightly over time, without a more pronounced effect in adopting journals (i.e., journals mentioning STARD in their Instructions for Authors) (3). We determined to what extent journals have adopted the STARD statement and incorporated the recommendations in their Instructions for Authors.

We identified the top 50 journals that frequently publish diagnostic accuracy studies, including general and internal medical journals (26%) and specialty journals (74%) (4). In February 2007, 1 reviewer (N.S.) searched for Instructions for Authors on each journal’s website and extracted any text mentioning STARD. Editors of 8 biomedical journals who were included in the STARD evaluation study(3) were asked to participate in a discussion about the quality of diagnostic research reports, factors affecting the use of the STARD statement within the editorial process, and how to improve the quality of diagnostic accuracy studies. A survey of editors of nonadopting journals was also carried out to evaluate their interest in adopting the STARD statement.

The STARD statement was mentioned in the Instructions for Authors in 19 of the 50 journals (38%). STARD was mentioned more often in general and internal medicine journals (6 of 13, 46%) than in specialty journals (13 of 37, 35%). The 19 adopting journals referred to the STARD initiative paper (11 of 19, 58%) and/or the STARD statement on the CONSORT website (www.consort-statement.org) (8 of 19, 42%). Three journals referred to the STARD background document. Only 2 journals referred directly to the STARD website (www.stard-statement.org).

In the adopting journals, we observed a broad variation in the language explaining how to use the STARD statement. Five journals clearly described their expectation regarding the use of the STARD statement, asking authors to submit the STARD checklist and/or to include a flow diagram in the manuscript. Most adopting journals, however, only referred to the STARD statement, without describing their expectations regarding use of the statement (Table 1¹ ).

Editors of nonadopting journals indicated that they do not need these guidelines because their reviewers and editors are well attuned to the items listed on the STARD checklist, and they publish relatively few diagnostic accuracy studies. Only 1 nonadopting journal became an adopter after our survey (Table 1¹ ). Editors of adopting journals indicated that examples of flow diagrams and proper references to statistical papers would be helpful for authors.

The STARD statement was developed to improve the quality of diagnostic research reports and to facilitate adequate appraisal of the internal and external validity of the study being reported. Our results show that not all journals that publish diagnostic accuracy studies refer to the STARD statement in their Instructions for Authors, and that those that do refer to STARD use varying language in their instructions. This variability may be a reason for the incompleteness of reporting still observed in major journals (3).

Editors have suggested that it takes time before authors and reviewers actually use guidelines such as STARD. Dissemination takes time, but we question whether implementation is a gradual and automatic process. We are aware that editors must make decisions and compromises in what they ask from authors and reviewers. Some editors believe that they ask too much from authors if they require a STARD checklist and flow diagram, whereas others feel that such a requirement may be the solution to improving the completeness of reporting of diagnostic accuracy studies.

In addition to including the STARD statement in the Instructions for Authors, journals could improve the quality of diagnostic research reports by more clearly communicating their expectations regarding use of the STARD statement and providing reader- and printer-friendly versions of these instructions on the journal website. The STARD checklist can be used by reviewers as a tool during the evaluation process and by authors as a means to assess the completeness of the reporting in their manuscripts. We believe that continuous attention of all parties involved is needed to improve the quality of diagnostic research reports.

Table 1.

The citations of STARD statement and the strength of direction regarding the use of the STARD statement by the 19 adopting journals.

Adopting journals	Citations of the STARD statement											Strength of direction regarding the use of the STARD statement in the Instructions for Authors
		STARD website1	CONSORT website2	STARD initiative paper	STARD background document	Other reference	Date of last update of the Instructions for Authors3	STARD statement	STARD checklist	Flow diagram
Annals of Emergency Medicine	No	Yes	Yes	No	Yes	NR	Should reflect
Annals of Internal Medicine45	No	No	Yes	Yes	No	NR	Consult	Consult	Include
Annals of the Rheumatic Diseases	No	No	Yes	No	No	NR	Should conform
Archives of Disease in Childhood	No	No	Yes	No	No	NR	Should conform
British Journal of Ophthalmology	No	No	Yes	No	No	NR	Should conform
British Journal of General Practice	No	Yes	No	No	No	NR	Should follow
British Medical Journal4	No	Yes	Yes	No	No	NR	Please follow/report
Cancer	No	No	Yes	No	No	July 2007		Encourage to submit	Encourage to submit
Circulation	No	No	No	Yes	No	NR	See
Clinical Chemistry456	No	No	Yes	Yes	Yes	April 2006		Complete	Strongly recommended
Clinical Radiology46	No	Yes	No	No	No	NR	Please refer to
Gut	No	No	Yes	No	No	NR	Should conform
The Journal of the American Medical Association6	No	Yes	No	No	No	January 2007		Encourage to submit	Encourage to submit
Journal of Clinical Microbiology	No	Yes	No	No	No	January 2007	Should refer to
Lancet6	Yes	No	No	No	No	NR	Must be reported
Neurology6	No	Yes	No	No	No	December 2006		Requires	Requires
Obstetrics and Gynaecology	No	No	Yes	No	Yes	July 2007		As resource
Radiology4	No	No	Yes	No	No	NR	Please refer
American Journal of Roentgenology47	Yes	Yes	No	No	No	July 2007	Please refer

1

www.stard-statement.org.

2

www.consort-statement.org.

3

Date of last update of Instructions for Authors on journals’ website (search was carried out on 11 July 2007); NR, not reported.

4

Published the STARD initiative paper.

5

Published the background document.

6

Published an editorial or commentary about the STARD statement.

7

Became an adopter after the survey in May 2007.

Grant/funding support: This project was funded by grants from The Netherlands Organisation for Health Research and Development, International Federation of Clinical Chemistry and Laboratory Medicine, and the NHS National Knowledge Service (as part of EQUATOR project).

Financial disclosures: None declared.

Acknowledgments: We thank the editors of the British Medical Journal, the Lancet, The Journal of the American Medical Association, New England Journal of Medicine, Annals of Internal Medicine, Clinical Chemistry, Gut, and Radiology and their collaborators for allowing us to have meetings to discuss measures to improve the quality of diagnostic research papers.