Wikipedia:Identifying reliable sources (medicine)

Wikipedia's articles are not meant to provide medical advice. Nevertheless, they are widely used among those seeking health information.[1] For this reason, all biomedical information must be based on reliable, third-party published secondary sources, and must accurately reflect current knowledge. This guideline supports the general sourcing policy with specific attention to what is appropriate for medical content in any Wikipedia article, including those on alternative medicine. Sourcing for all other types of content – including non-medical information in medicine-articles – is covered by the general guideline on identifying reliable sources.

Ideal sources for biomedical information include: review articles (especially systematic reviews) published in reputable medical journals; academic and professional books written by experts in the relevant fields and from respected publishers; and guidelines or position statements from national or international expert bodies. Primary sources should generally not be used for medical content – as such sources often include unreliable or preliminary information, for example early lab results which don't hold in later clinical trials.

See the reliable sources noticeboard for questions about reliability of specific sources, and feel free to ask at WikiProjects such as WikiProject Medicine and WikiProject Pharmacology.

Per the policies of neutral point of view, no original research, and verifiability, Wikipedia articles should be based on reliable, independent, published secondary or tertiary sources. For biomedical content, primary sources should generally not be used. This is because primary biomedical literature is exploratory and often not reliable, and any given primary source may be contradicted by another. The Wikipedia community relies on guidance of expert reviews, and statements by major medical and scientific bodies. Text that relies on primary sources should usually have minimal weight, only be used to describe conclusions made by the source, and must describe these findings clearly so that all editors even those without specialist knowledge can check sources. Primary sources should never be cited in support of a conclusion that is not clearly made by the authors (see WP:Synthesis).

Sources about health in the general news media should, in general, not be used to source content about health in Wikipedia articles but may be useful for "society and culture" content. Please see Popular press below.

Biomedical information requires sources complying with this guideline, whereas general information in the same article may not. For example, an article on Dr Foster's Magic Purple Pills could contain the following:

Primary sources should not be cited with intent of "debunking", contradicting, or countering any conclusions made by secondary sources. is original research, and Wikipedia is not a venue for open research. Controversies or uncertainties in medicine should be supported by reliable secondary sources describing the varying viewpoints. Primary sources should not be aggregated or presented without context in order to undermine proportionate representation of opinion in a field. If material can be supported by either primary or secondary sources – the secondary sources should be used. Primary sources may be presented together with secondary sources.

Findings are often touted in the popular press as soon as original, primary research is reported, before the scientific community has analyzed and commented on the results. Therefore, such sources should generally be entirely omitted (see recentism). Determining weight of studies generally requires reliable secondary sources (not press releases or newspaper articles based on such sources). If conclusions are worth mentioning (such as large randomized clinical trials with surprising results), they should be described appropriately as from a single study:

"A large, NIH-funded study published in 2010 found that selenium and Vitamin E supplements, separately as well as together, did not decrease the risk of getting prostate cancer and that vitamin E may increase the risk; they were previously thought to prevent prostate cancer." (citing PMID )

Given time a review will be published, and the primary sources should preferably be replaced with the review. Using secondary sources then allows facts to be stated with greater reliability:

"Neither vitamin E nor selenium decreases the risk of prostate cancer and vitamin E may increase it." (citing PMID PMID )

If no reviews on the subject are published in a reasonable amount of time, then the content and primary source should be removed.

A reason to avoid primary sources in the biomedical field – especially papers reporting results of in vitro experiments – is that they are often not replicable[2][3][4] and are therefore unsuitable for use in generating encyclopedic, reliable biomedical content. Drug discovery scientists at Bayer in 2011 reported that they were able to replicate results in only ~20 to 25% of the prominent studies they examined;[5] scientists from Amgen followed with a publication in 2012 showing that they were only able to replicate 6 (11%) of 53 high-impact publications and called for higher standards in scientific publishing.[6] The journal Nature announced in April 2013 that in response to these and other articles showing a widespread problem with reproducibility, it was taking measures to raise its standards.[7] Further, the fact that a claim is published in a refereed journal need not make it true. Even well-designed randomized experiments will occasionally (with low probability) produce spurious results. Experiments and studies can produce flawed results or even fall victim to deliberate fraud (e.g. the and the Schön scandal.)

Scientific journals are the best place to find both primary source articles about experiments, including medical studies, and secondary sources. Every rigorous scientific journal is peer reviewed. Be careful of material published in journals lacking peer review or which reports material mainly in other fields. (See: Martin Rimm.) Be careful of material published in disreputable journals or disreputable fields. (See: Sokal affair.)

Wikipedia policies on the neutral point of view and not publishing original research demand that we present prevailing medical or scientific consensus, which can be found in recent, authoritative review articles, in statements and practice guidelines issued by major professional medical or scientific societies (for example, the European Society of Cardiology or the Infectious Disease Society of America) and widely respected governmental and quasi-governmental health authorities (for example, AHRQ, USPSTF, NICE, and WHO), in textbooks, or in some forms of monographs. Although significant-minority views are welcome in Wikipedia, such views must be presented in the context of their acceptance by experts in the field. Additionally, the views of tiny minorities need not be reported.

Finally, make readers aware of controversies that are stated in reliable sources. A well-referenced article will point to specific journal articles or specific theories proposed by specific researchers.

When writing about treatment efficacy, knowledge about the quality of the evidence helps distinguish between minor and major views, determine due weight, and identify accepted evidence-based information. Even in reputable medical journals, different papers are not treated as of equal value. Studies can be categorized into levels of evidence,[8] and editors should rely on high-quality evidence, such as systematic reviews. Lower quality evidence (such as case reports or series) or non-evidence (such as anecdotes or conventional wisdom) are avoided. Medical guidelines or position statements by nationally or internationally recognized expert bodies also often contain assessments of underlying evidence. This guideline is not general in nature, but specifically concerns quality when used as a source for encyclopedic articles on Wikipedia. Sources that are ruled out for use on Wikipedia may still be very useful for other purposes.

There are different ways to rank level of evidence in medicine, but they similarly put high level reviews and practice guidelines at the top.

The best evidence for treatment efficacy is mainly from meta-analyses of randomized controlled trials (RCTs).[12] Systematic reviews of literature of overall good quality and consistency, addressing specific recommendation, have less reliability when they include non-randomized studies.[13] Narrative reviews can help establish the context of evidence quality. Roughly in descending order of quality, lower-quality evidence in medical research comes from individual RCTs; quasi-experimental studies; prospective observational (non-experimental) studies, such as prospective cohort studies; case control studies; cross-sectional studies (surveys), and other correlation studies such as ecological studies; other retrospective analyses (including retrospective cohort studies); and non-evidence-based expert opinion or clinical experience. Case reports, even when published in peer-reviewed medical journals, are normally considered a type of anecdotal evidence, and they generally fall below the minimum requirements of reliable medical sources.

Speculative proposals and early-stage research should not be cited to imply wide acceptance. For example, results of an early-stage clinical trial would not be appropriate in the Treatment section on a disease because future treatments have little bearing on current practice. The results might – in some cases – be appropriate for inclusion in an article specifically dedicated to the treatment in question or to the researchers or businesses involved in it. Such information, particularly when citing secondary sources, may be appropriate in research sections of disease articles. To prevent misunderstanding, the text should clearly identify the level of research cited (e.g., "first-in-human safety testing").

Several formal systems exist for assessing the quality of available evidence on medical subjects.[14][15] "Assessing evidence quality" means editors should determine quality of the type of study. Respect the levels of evidence: Do not reject a high-quality type of study (e.g., a meta-analysis) in favor of a source from lower levels of evidence (e.g., any primary source) because of personal objections to the inclusion criteria, references, funding sources, or conclusions in the higher-level source.

"Assessing evidence quality" means editors should determine quality of the type of source and quality of the publication. Editors should not perform detailed academic peer review.

Avoid over-emphasizing single studies, particularly in vitro or animal studies

In vitro studies and animal models serve a central role in research, and are invaluable in determining mechanistic pathways and generating hypotheses. However, in vitro and animal-model findings do not translate consistently into clinical effects in human beings. Where in vitro and animal-model data are cited on Wikipedia, it should be clear to the reader that the data are pre-clinical, and the article text should avoid stating or implying that reported findings hold true in humans. The level of support for a hypothesis should be evident to a reader.

Using small-scale, single studies makes for weak evidence, and allows for cherry picking of data. Studies cited or mentioned in Wikipedia should be put in context by using high-quality secondary sources rather than by using the primary sources.

Keeping an article up-to-date while maintaining the more-important goal of reliability is important. These instructions are appropriate for actively researched areas with many primary sources and several reviews and may need to be relaxed in areas where little progress is being made or where few reviews are published.

Many treatments or proposed treatments lack good research into their efficacy and safety. In such cases, reliable sources may be difficult to find, while unreliable sources are readily available. When writing about medical claims not supported by mainstream research, it is vital that third-party, independent sources be used. Sources written and reviewed by the advocates of marginal ideas may be used to describe personal opinions, but extreme care should be taken when using such sources lest more controversial opinions be taken at face value or, worse, asserted as fact. If independent sources discussing a medical subject are of low quality, then it is likely that the subject itself is not notable enough to have its own article or relevant for mention in other articles.

Symposia and supplements to academic journals are commonly sponsored by industry groups with a financial interest in the outcome of the research reported. They may lack independent editorial oversight and peer review with no supervision of content by the parent journal.[16] Such articles do not share the reliability of their parent journal.[17] Indications that an article was published in a supplement may be fairly subtle; for instance, a letter "s" added to a page number,[18] or "Suppl." in a reference.

Bias caused by conflicts of interest is an important issue in medical research. It arises in part due to financial interests that compete within medicine. Disclosure of conflicts of interest is mandated, but isn't always done – and even when it is may not be helpful. A source can also simply be bad, where biases in criteria make it less than ideal. Claims of bias should not be made lightly – if you simply call out results as biased, you may introduce your own bias. Claims of bias should be sourced to reliable secondary sources, and are not reason to omit sources without consensus – instead, qualify sources with information of why a source may be biased, and who is calling it biased.

Obvious or overt bias in a source is a difficult problem for Wikipedia. If there is consensus on an article that a certain source should be omitted for bias, it may be excluded. It may be simpler to find a "better" source – either a higher quality study type or a more specific source instead (see WP:MEDASSESS). If no high-quality source exists for a controversial statement it is best to leave it out; this is not bias.

People most interested in improving an article may have a connection to its subject.

Use your best judgement when writing about topics where you may have a conflict of interest: citing yourself on Wikipedia is problematic. Citing your own organization, such as a governmental health agency or an NGO producing high-quality systematic reviews is generally acceptable – if it is done to improve coverage of a topic, and not with the sole purpose of driving traffic to your site. All edits should improve neutral encyclopedic coverage; anything else, such as promoting an organization is not allowed.

According to the conflict of interest guideline – conflicts of interest (COI) must be disclosed. Editing on topics where one is involved or closely related, especially when there is potential financial gain, is discouraged. Medicine is not an exception. One way to contribute with a COI is to post on talk-pages, suggesting edits. Another alternative is the articles for creation pathway. These methods are often best when writing about oneself, one's organization or company– but may be less so when there is a potential conflict of interest in a research field. For example, one may legitimately be an authority on a certain topic – a volunteer who reads the talk-page will not always have the knowledge to assess the sources properly. Then it is better to follow ordinary editing protocol, disclosing any COI and to be careful not to overemphasize your own sources.

A Wikipedia article should cite high-quality reliable sources . Some high-quality journals, such as JAMA, publish a few freely readable articles even though most are not free. A few high-quality journals, such as PLoS Medicine, publish only freely readable sources. Also, a few sources are in the public domain; these include many U.S. government publications, such as the Morbidity and Mortality Weekly Report of the Centers for Disease Control and Prevention.

When searching for biomedical sources, it is wise to skim-read everything available, including abstracts of papers that are not freely readable, and use that to get a feel for what reliable sources are saying. However, when it comes to actually writing a Wikipedia article, it is misleading to give a full citation for a source after reading only its abstract; the abstract necessarily presents a stripped-down version of the conclusions and omits the background that can be crucial for understanding exactly what the source says, and may not represent the article's actual conclusions.[19] To access the full text, the editor may need to visit a medical library or ask someone at the WikiProject Resource Exchange or WikiProject Medicine's talk page to either provide an electronic copy or read the source and summarize what it says; if neither is possible, the editor may need to settle for using a lower-impact source.

Peer-reviewed medical journals are a natural choice as a source for up-to-date medical information in Wikipedia articles. Journal articles come in many different types, and are a mixture of primary and secondary sources. It is normally best to use reviews and meta-analyses where possible. Reviews in particular give a balanced and general perspective of a topic, and are usually easier to understand.

Primary publications describe novel research for the first time, while review articles summarize and integrate a topic of research into an overall view. In medicine, primary sources include clinical trials, which test new treatments. Broadly speaking, reviews may be narrative or systematic (and sometimes both). Narrative reviews often set out to provide a general summary of a topic based on a survey of the literature, which can be useful when outlining a topic. Systematic reviews use sophisticated methodology to address a particular clinical question in as balanced (unbiased) a way as possible. Some systematic reviews also include a statistical meta-analysis to combine the results of several clinical trials to provide stronger quantitative evidence about how well a treatment works for a particular purpose. Systematic reviews and meta-analyses of randomized controlled trials can provide strong evidence of the clinical efficacy of particular treatments in given scenarios, which may in turn be incorporated into medical guidelines or institutional position papers (ideal sources for clinical recommendations).

Research papers that describe original experiments are primary sources. However, they normally contain introductory, background, or review sections that place their research in the context of previous work; these sections may be cited in Wikipedia with care: they are often incomplete[20] and typically less useful or reliable than reviews or other sources, such as textbooks, which are intended to be reasonably comprehensive. If challenged by another editor in good faith, the primary source should be supplemented with a more appropriate source. A general narrative review of a subject by an expert in the field can make a good secondary source covering various aspects of a subject within a Wikipedia article. Such reviews typically do not contain primary research, but can make interpretations and draw conclusions from primary sources that no Wikipedia editor would be allowed to do. A systematic review uses a reproducible methodology to select primary (or sometimes secondary) studies meeting explicit criteria to address a specific question. Such reviews should be more reliable and accurate and less prone to bias than a narrative review.[14] However, whereas a narrative review may give a panorama of current knowledge on a particular topic, a systematic review tends to have a narrower focus.

Journals may specialize in particular article types. A few, such as Evidence-based Dentistry (ISSN ), publish third-party summaries of reviews and guidelines published elsewhere. If an editor has access to both the original source and the summary, and finds both helpful, it is good practice to cite both sources together (see: Citing medical sources for details). Others, such as Journal of Medical Biography, publish historical material that can be valuable for History sections, but is rarely useful for current medicine. Still others, such as Medical Hypotheses, publish speculative proposals that are not reliable sources for biomedical topics.

The Abridged Index Medicus provides a list of 114 selected "core clinical journals".[21] Another useful grouping of core medical journals is the 2003 Brandon/Hill list, which includes 141 publications selected for a small medical library[22] (although this list is no longer maintained, the listed journals are of high quality). Core general medical journals include the New England Journal of Medicine, The Lancet, the Journal of the American Medical Association (JAMA), the Annals of Internal Medicine, The BMJ (British Medical Journal), and the Canadian Medical Association Journal. Core basic science and biology journals include Nature, Science and Cell.

An integral part of finding high-quality sources is avoiding articles from journals without a reputation for fact-checking and accuracy. A red flag that a journal article is probably not reliable for biomedical claims might be publication by a publisher that has a reputation for exhibiting "predatory" behavior, which includes questionable business practices and/or peer-review processes that raise concerns about the reliability of their journal articles. (See also WP:RS#Predatory journals and the #References section below for examples of such publishers.[23][24]) Other indications that a biomedical journal article may not be reliable are its publication in a journal that is not indexed in the bibliographic database MEDLINE,[25] or its content being outside the journal's normal scope (for instance, an article on the efficacy of a new cancer treatment in a psychiatric journal or the surgical techniques for hip replacement in a urology journal). Determining the reliability of any individual journal article may also take into account whether the article has garnered significant positive citations in sources of undisputed reliability, suggesting wider acceptance in the medical literature despite any red flags suggested here.

An archive of Beall's List, an early list of potentially predatory journals, can be found at ; updates are added separately by an anonymous post-doctoral researcher. On Wikipedia, the CiteWatch compilation (updated twice monthly) and the Unreliable/Predatory Source Detector script can be leveraged to facilitate the detection of predatory journals.

Symposia and supplements to academic journals are often (but far from always) unacceptable sources. They are commonly sponsored by industry groups with a financial interest in the outcome of the research reported. They may lack independent editorial oversight and peer review, with no supervision of content by the parent journal.[16] Such shill articles do not share the reliability of their parent journal,[17] being essentially paid ads disguised as academic articles. Such supplements, and those that do not clearly declare their editorial policy and conflicts of interest, should not be cited.

Indications that an article was published in a supplement may be fairly subtle; for instance, a letter "S" added to a page number,[26] or "Suppl." in a reference.[27] However, note that merely being published in a supplement is not prima facie evidence of being published in a sponsored supplement. Many, if not most, supplements are perfectly legitimate sources, such as the Astronomy & Astrophysics Supplement Series, Nuclear Physics B: Proceedings Supplements, Supplement to the London Gazette, or The Times Higher Education Supplement. A sponsored supplement need not necessarily have a COI with its medical content; for instance, public health agencies may also sponsor supplements. However, groups that do have a COI may hide behind layers of front organizations with innocuous names, so the ultimate funding sources should always be ascertained.

High-quality textbooks can be a good source to start an article, and often include general overviews of a field or subject. However, books generally move slower than journal sources, and are often several years behind the current state of evidence. This makes using up-to-date books even more important. Medical textbooks published by academic publishers are often excellent secondary sources. If a textbook is intended for students, it may not be as thorough as a monograph or chapter in a textbook intended for professionals or postgraduates. Ensure that the book is up to date, unless a historical perspective is required. Doody's maintains a list of core health sciences books, which is available only to subscribers.[28] Major academic publishers (e.g., Elsevier, Springer Verlag, Wolters Kluwer, and Informa) publish specialized medical book series with good editorial oversight; volumes in these series summarize the latest research in narrow areas, usually in a more extensive format than journal reviews. Specialized biomedical encyclopaedias published by these established publishers are often of good quality, but as a tertiary source, the information may be too terse for detailed articles.

Additionally, popular science books are useful sources, but generally should not be referenced on Wikipedia to support medical statements (see #Popular press). In addition, most self-published books or books published by vanity presses undergo no independent fact-checking or peer review and, consequently, are not reliable sources. However, books published by university presses or the National Academy of Sciences tend to be well-researched and useful for most purposes.

Guidelines and position statements provided by major medical and scientific organizations are important on Wikipedia because they present recommendations and opinions that many caregivers rely upon (or may even be legally obliged to follow).

Guidelines by major medical and scientific organizations sometimes clash with one another (for example, the World Health Organization and American Heart Association on salt intake), which should be resolved in accordance with WP:WEIGHT. Guidelines do not always correspond to best evidence, but instead of omitting them, reference the scientific literature and explain how it may differ from the guidelines. Remember to avoid WP:original research by only using the best possible sources, and avoid weasel words and phrases by tying together separate statements with "however", "this is not supported by", etc. The image below attempts to clarify some internal ranking of statements from different organizations in the weight they are given on Wikipedia.

Statements and information from reputable major medical and scientific bodies may be valuable encyclopedic sources. These bodies include the U.S. National Academies (including the National Academy of Medicine and the National Academy of Sciences), the British National Health Service, the U.S. National Institutes of Health and Centers for Disease Control and Prevention, and the World Health Organization. The reliability of these sources ranges from formal scientific reports, which can be the equal of the best reviews published in medical journals, through public guides and service announcements, which have the advantage of being freely readable, but are generally less authoritative than the underlying medical literature.

The popular press is generally not a reliable source for scientific and medical information in articles. Most medical news articles fail to discuss important issues such as evidence quality,[29] costs, and risks versus benefits,[30] and news articles too often convey wrong or misleading information about health care.[31] Articles in newspapers and popular magazines generally lack the context to judge experimental results. They tend to overemphasize the certainty of any result, for instance, presenting a new and experimental treatment as "the cure" for a disease or an every-day substance as "the cause" of a disease. Newspapers and magazines may also publish articles about scientific results before those results have been published in a peer-reviewed journal or reproduced by other experimenters. Such articles may be based uncritically on a press release, which themselves promote research with uncertain relevance to human health and do not acknowledge important limitations, even when issued by an academic medical center.[32] News articles also tend neither to report adequately on the scientific methodology and the experimental error, nor to express risk in meaningful terms. For Wikipedia's purposes, articles in the popular press are generally considered independent, primary sources.

Conversely, the high-quality popular press can be a good source for social, biographical, current-affairs, financial, and historical information in a medical article. For example, popular science magazines such as New Scientist and Scientific American are not peer reviewed, but sometimes feature articles that explain medical subjects in plain English. As the quality of press coverage of medicine ranges from excellent to irresponsible, use common sense, and see how well the source fits the verifiability policy and general reliable sources guidelines. Sources for evaluating health-care media coverage include the review websites or along with specialized academic journals, such as the Journal of Health Communication; reviews can also appear in the American Journal of Public Health, the Columbia Journalism Review, Ben Goldacre's and others. Health News Review's criteria for rating news stories[33] can help to get a general idea of the quality of a medical news article.

Reliable sources must be strong enough to support the claim. A lightweight source may sometimes be acceptable for a lightweight claim, but never for an extraordinary claim.

Press releases, newsletters, advocacy and self-help publications, blogs and other websites, and other sources contain a wide range of biomedical information ranging from factual to fraudulent, with a high percentage being of low quality. Conference abstracts present incomplete and unpublished data and undergo varying levels of review; they are often unreviewed and their initial conclusions may have changed dramatically if and when the data are finally ready for publication.[34] Consequently, they are usually poor sources and should always be used with caution, never used to support surprising claims, and carefully identified in the text as preliminary work. Medical information resources such as WebMD and eMedicine are usually acceptable sources for uncontroversial information; however, as much as possible Wikipedia articles should cite the more established literature directly. UpToDate is less preferred as it is not possible to reference specific versions of their articles, archives do not exist, and it can be difficult to access.

Search engines are commonly used to find biomedical sources. Each engine has quirks, advantages, and disadvantages, and may not return the results that the editor needs unless used carefully. It typically takes experience and practice to recognize when a search has not been effective; even if an editor finds useful sources, they may have missed other sources that would have been more useful or they may generate pages and pages of less-than-useful material. A good strategy for avoiding sole reliance on search engines is to find a few recent high-quality sources and follow their citations to see what the search engine missed. It can also be helpful to perform a plain web search rather than one of scholarly articles only.

PubMed is an excellent starting point for locating peer-reviewed medical . It offers a free search engine for accessing the MEDLINE database of biomedical research articles offered by the National Library of Medicine at the U.S. National Institutes of Health.[35] There are basic and advanced options for searching PubMed.[36] For example, clicking on the "Review" tab will help narrow the search to review articles. The "Filters" options can further narrow the search, for example, to meta-analyses, to practice guidelines, and/or to freely readable sources. Although PubMed is a comprehensive database, many of its indexed journals restrict online access. Another website, PubMed Central, provides free access to full texts. While it is often not the official published version, it is a peer-reviewed manuscript that is substantially the same, but lacks minor copy-editing by the publisher.[37]

When looking at an individual abstract on the PubMed website, an editor can consult "Publication Types", "MeSH Terms", etc. at the bottom of the page to see how the document has been classified in PubMed. For example, a page that is tagged as "Comment" or "Letter" is a letter to the editor (often not peer-reviewed). The classification scheme includes about 70 types of documents.[38] For medical information, the most useful types of articles are typically labeled "Guideline", "Meta-analysis", "Practice guideline", or "Review". Even when an article is one of the most useful types and recently published, it can be helpful to check the journal on DOAJ and other databases as well as the status and publishing track of authors if they make extraordinary claims. There is no magic number, but it is useful to compare the authors to others' in the same field of study.