Is Wikipedia a Legitimate Reference Source for Health…

ABSTRACT & COMMENTARY

Is Wikipedia a Legitimate Reference Source for Health Care Professionals?

By David J. Pierson, MD, Editor

SYNOPSIS: An increasing number of peer-reviewed academic papers in the health sciences are citing Wikipedia. Potential problems with this for the clinical context are that articles in Wikipedia are tertiary sources (rather than primary or even secondary sources of the information cited), they can be written and edited by anyone, and because they are altered frequently, precise and accurate source citations are challenging.

SOURCE: Bould MD, et al. References that anyone can edit: Review of Wikipedia citations in peer reviewed health science literature.
BMJ 2014;348:g1585.

This study by a group of Canadian investigators sought to determine the prevalence of citations to Wikipedia in indexed health science journals, to identify the journals citing Wikipedia in the articles they publish, and to assess the way in which this source is being cited in those articles. They examined English-language journals that had cited Wikipedia since its inception in 2001, using the online databases Scopus and Web of Science, and focusing on health science journals with at least one citation in PubMed, Medline, or Embase according to Ulrich’s Periodicals Directory. The authors retrieved full-text copies of all identified articles through the University of Ottawa library and other Canadian library facilities, and ascertained the 2011 impact factor for each identified journal from Thomson Reuters Journal Citation Reports.

After excluding duplicates, non-English language journals, and articles that did not use Wikipedia as a direct citation, the authors identified 1433 articles published in 1008 indexed journals that included 2049 individual Wikipedia citations. Impact factors for 2011 were available for 1420 citations, in 980 articles from 650 indexed journals.

The rate of use of Wikipedia as a cited reference increased dramatically after December 2010, and was still increasing as of November 2013. Citations varied by category, with definitions (648 citations, 31.6% of the total) and statements describing a process, system, or event (482 citations, 23.5%) being most common. Other types of citation identified by Bould et al included descriptions of historical events (277 citations, 13.5%), statistics (161 citations, 7.9%), and biographical information (91 citations, 4.4%), along with seven other categories at lesser frequencies. However, 97 citations (4.8%) were used by authors in place of original research studies —that is, the Wikipedia article referred to an original publication but the former was cited rather than the latter.

Journals citing Wikipedia since 2001 had a median impact factor of 2.0, which has remained constant throughout that interval. However, 16 of the journals had 2011 impact factors exceeding 10, including Nature, Science, Nature Medicine, the Annals of Internal Medicine, PLoS Medicine, Circulation, and the BMJ. Some citations to Wikipedia were deemed appropriate by the authors, such as when referring to Wikipedia itself or when data on Wikipedia contents were used in the study, but the top 25 journals by impact factor accounted for only 2.2% of such citations. The authors conclude that although only a very small proportion of all journal citations are to Wikipedia articles, "the possibility for the spread of misinformation from an unverified source is at odds with the principles of robust scientific methodology and could potentially affect care of patients."

COMMENTARY

Bould and colleagues identify three potential problems with the use of Wikipedia as a reference in academic publications in science and health care. First, as stated in Wikipedia’s main article on itself, "Almost anyone who can access the site can edit almost any of its articles."¹ The fact that the articles are anonymous is also not reassuring in this respect. Second, the changing nature of Wikipedia makes precise citation difficult, in the context that reference citation in science attempts to identify permanent sources for data or statements. It is possible to identify the date and time a particular Wikipedia page was established, although this is not widely known and was almost never found in the citations in the Bould study. And, third, Wikipedia is a tertiary source (that is, one that compiles or summarizes other sources), as opposed to primary sources (which present original data or documents) and secondary sources (such as reviews, books, or unpublished literature interpreting the contents of primary sources). International guidelines for publication in science² instruct authors to provide direct references to original research sources; thus, as Bould et al state, "citing Wikipedia or any other tertiary source in the academic literature opposes literary practice."

To place this study’s findings and its potential implications for readers of Critical Care Alert into perspective, I looked up the topics addressed in the three preceding abstract-commentaries on Wikipedia. In the entry about sepsis, under prognosis,³ it says that "some people may experience severe long-term cognitive decline following an episode of severe sepsis, but the absence of baseline neuropsychological data in most sepsis patients makes the incidence of this difficult to quantify or to study." This statement, which is accurate, references a 2009 case study and review. However, the article discussed by Dr. Tran in this issue cites at least 10 articles published since then that pertain to this issue, several of them directly on the topic of cognitive decline and neuropsychological function in survivors of severe sepsis.

More worrisome is the following statement, in the Wikipedia article on high-flow oxygen therapy, the subject of the article reviewed by Dr. Hoffman in this issue, "Some patients requiring respiratory support for bronchospasm benefit using air delivered by HFT [high-flow therapy] without additional oxygen."⁴ The cited reference is to a bench study of two devices for providing high-flow therapy, which did not deal with patients. In the "ICU Sedation" section of the Wikipedia article on dexmedetomidine, it says the following: "Compared to midazolam, dexmedetomidine is superior due to reduced intensive care costs. The reduced costs are due to a reduction in intensive care unit stay as well as reduced mechanical ventilation." This statement, which is contradicted by the study described in this issue by Dr. Walter, cites as its source a survey of opinions and current practice of critical care professionals in Australia and New Zealand, which reported that dexmedetomidine was "rarely used by most respondents."⁶

The sepsis sequelae example above simply shows that the Wikipedia entry is not up to date. More troubling is the possibility of bias, as suggested by the other two examples. Bias exists in all aspects of critical care, but I think it is a special problem when topics have direct commercial implications, as with medical devices (such as high-flow oxygen systems) and pharmaceutical agents (such as dexmedetomidine). This is not to say that mainstream, indexed, peer-reviewed journals are free from bias. Critical reading and knowledge of the principles of good clinical research are irreplaceable, and the bottom line with assessing all resources in managing patients is always caveat lector — let the reader beware.

REFERENCES

http://en.wikipedia.org/wiki/Wikipedia. Accessed September 9, 2014.
International Committee of Medical Journal Editors. Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals. December, 2013. http://www.icmje.org/icmje-recommendations.pdf. Accessed September 9, 2014.
http://en.wikipedia.org/wiki/Sepsis#Prognosis. Accessed September 9, 2014.
http://en.wikipedia.org/wiki/High_Flow_Therapy. Accessed September 9, 2014.
http://en.wikipedia.org/wiki/Dexmedetomidine. Accessed September 9, 2014.
O’Connor M, et al. Sedation management in Australian and New Zealand intensive care units: Doctors’ and nurses’ practices and opinions. Am J Crit Care 2009;19:285-295.