Health and Biomedical Information(2)

2.6.1  Benefits and challenges of electronic scholarly publication

(4/3/05)  The growing effort devoted to unimpeded access to scientific information has taken on the name "open access" publishing.  It is guided by the philosophy that access to scientific archives should be free and unimpeded, with other means used to finance the cost of scientific publishing.  There is considerable debate over the merits of this approach to publishing, including discussion of its financial feasibility as well as the issue of who controls scientific literature.

A big event in the debate occurred on July 14, 2004, when the US House of Representatives mandated that the National Institutes of Health (NIH) develop a plan to increase access to reports of US taxpayer-funded research.  It instructed the NIH to develop a plan by December 1, 2004 that would become effective by July 1, 2005.  The legislation espoused the principle that all articles based on NIH-funded research would be deposited in the PubMed Central repository, with articles produced by any amount of NIH funding entering immediately and all others within six months.  The legislation was based on a report by NIH Director Elias Zerhouni published in May, 2004 (Zerhouni, 2004a).  An overview of this process has been maintained by Suber (2005).

数据挖掘研究院

This legislation reflected other sentiments from around the world.  A summary of various worldwide efforts to promote open access publishing was published earlier in 2004 by the European Molecular Biology Organization (Grivell, 2004).  A report to the United Kingdom House of Commons promoted similar access (Anonymous, 2004a; Clery, 2004).  Also in 2004, the Washington DC Principles for Free Access to Science (Anonymous, 2004b) was established, led by non-profit publishers representing scholarly socieities.  These publishers reaffirmed their view that they maintain the copyright on their publications, but advocated a number of principles:

• Selected important articles available free on-line when published
• Complete contents freely available within months of publication
• Complete contents available for free to scientists in low-income countries
• Content available through on-line reference linking and major search engines 数据挖掘研究院
  

As a result of the legislation, the NIH released a request for comments (RFC) on the principles (Anonymous, 2004c).  Zerhouni (2004b) summarized the process in the fall of 2004 and put forth an analysis of the cost.  For the latter, he based his work on an estimate that 0.32% of grant funding is devoted to publication costs.  Based on this estimate, the NIH already provides about $30 million in direct costs for publications through its funded research grants.  Zerhouni asserted that adopting a plan to archive all publications in PubMed Central would add only another $2-4 million per year, since it would be built on top of the existing NIH information technology infrastructure.  The RFC drew over 6,000 responses.  Many were in favor, but both commercial and non-profit publishers raised concerns about the plan.  Among the latter were public documents from the American College of Physicians (Tooker, 2004) and the New England Journal of Medicine (Drazen and Curfman, 2004). 数据挖掘研究院

In early 2005, the NIH released its policy on archiving scientific publications, adopting a voluntary policy whereby NIH-funded were strongly encouraged either themselves or via their publishers to deposit reports of their research into PubMed Central as soon as possible, and within 12 months after final publication.  Not surprisingly, this political compromise pleased few participants in the debate, e.g., Washington DC Principles for Free Access to Science (Anonymous, 2005c).  Whether the new policy will end the debate on this is unclear.

In the meantime, there have been other developments related to open-access publishing.  Biomed Central (BMC), described in the text, continues to grow and evolve.  One new innovation has been the notion of institutional memberships, where institutions such as universities pay a fee to join, which allows any member of that institution to publish without paying the usual publishing charge.  In the UK, the government recently reached an agreement to pay $170,000 to BMC that allows any UK research to publish in BMC without paying the up-front charge.  Now that several BMC journals have been in existence for more than two years, they are receiving IFs.  Several BMC journals have IFs > 5.0, and most IFs have increased over time for longer-established journals (Anonymous, 2005c). 数据挖掘实验室

Another new journal taking the open-access approach is the Public Library of Science (PLoS, http://publiclibraryofscience.org/ or http://www.plos.org/ ).  The text notes that PLoS began as effort to gather signatures advocating open access to scientific archives.  Under the leadership of Harold Varmus, former director of the National Institutes of Health, PLoS has developed two journals using the open access model.  Its first journal, PLoS Biology, began publication in 2003, while PLoS Medicine launched in 2004.  Both journals aim to publish top-flight articles of crucial impact and additional journals are planned.

Lancet has addressed the open access issue in its pages, noting that the up-front page charges may limit venues of publishing for resource-poor scientists (although BMC and PLoS pledge to waive fees for such scientists) and that open access threatens the survival of non-profit presses, such as university presses (Horton, 2003).  The editor of JAMA also took a negative view, noting that the charges of BMC and PLoS ($500-$1500) may not cover the costs of the publishing process, meaning that their business models are not sustainable, especially for journals with low acceptance rates like JAMA at 8% (DeAngelis and Musacchio, 2004).  (She noted in particular that PLoS is funded by a one-time grant.)  She also expressed concern that this model may provide incentive for journals to publish more and, as a result, lower their quality.  The publisher of BMC, on the other hand, retorted that journals still need to maintain their quality if they want to provide incentive for scientists to publish there (J Velterop, personal communication).

Another development was the decision of the British Medical Journal to cease providing free access to its entire content on its Web site (Delamothe and Smith, 2003).  Beginning in 2005, portions of the Web site were no longer be freely available.  The portions residing behind access controls are the abridged versions of articles that appear in the print journal.  The journal notes undertook this action because subscriptions to BMJ had fallen off, resulting in decreased revenue for the production of the journal.

Anonymous (2004a). Scientific Publications:  Free for All? London, UK, Science and Technology Committee, House of Commons. http://www.publications.parliament.uk/pa/cm200304/cmselect/cmsctech/399/399.pdf.
Anonymous (2004b). Washington D.C. Principles For Free Access to Science - A Statement from Not-for-Profit Publishers. Washington, DC, Washington D.C. Principles For Free Access to Science. http://www.dcprinciples.org/statement.pdf. 数据挖掘研究院
Anonymous (2004c). Enhanced Public Access to NIH Research Information. Washington, DC, National Institutes of Health. http://grants1.nih.gov/grants/guide/notice-files/NOT-OD-04-064.html.
Anonymous (2004d). Open access journals proven to compete on quality. London, UK, Biomed Central.  http://www.biomedcentral.com/info/about/pr-releases?pr=20040624.
Anonymous (2005a). Policy on Enhancing Public Access to Archived Publications Resulting from NIH-Funded Research. Washington, DC, National Institutes of Health. http://grants.nih.gov/grants/guide/notice-files/NOT-OD-05-022.html.
Anonymous (2005b). NIH Calls on Scientists to Speed Public Release of Research Publications. Washington, DC, National Institues of Health. http://www.nih.gov/news/pr/feb2005/od-03.htm. 数据挖掘研究院
Anonymous (2005c). Not-for-Profit Publishers Call New NIH Rule a Missed Opportunity. Washington, DC, Washington DC Principles for Free Access to Science. http://www.dcprinciples.org/nih_rule.htm.
Clery, D. (2004). U.K. lawmakers urge prompt access to published papers. Science, 305: 458-459.
DeAngelis, C. and Musacchio, R. (2004). Access to JAMA. Journal of the American Medical Association, 291: 370-371.
Delamothe, T. and Smith, R. (2003). Paying for bmj.com. British Medical Journal, 327: 241-242.
Drazen, J. and Curfman, G. (2004). Public access to biomedical research. New England Journal of Medicine, 351: 1343.
Grivell, L. (2004). Access for all? EMBO Reports, 5: 222-225.
Horton, R. (2003). 21st-century biomedical journals:  failures and futures. Lancet, 362: 1510-1512.

数据挖掘研究院

Suber, P. (2004). NIH Public-Access Policy Frequently Asked Questions. Earlham College. http://www.earlham.edu/~peters/fos/nihfaq.htm. Accessed: April 3, 2005.
Tooker, J. (2004). ACP Comments on Proposed NIH Public Access Policy. Philadelphia, PA, American College of Physicians. http://www.acponline.org/hpp/nih_open.htm.
Zerhouni, E. (2004a). Access to Biomedical Research Information. Bethesda, MD, National Institutes of Health. http://library.cpmc.columbia.edu/hsl/eres/countact.cfm?resourceid=5673.
Zerhouni, E. (2004b). NIH public access policy. Science, 306: 1895.

2.6.2  Quality of health information on the Web

(4/4/05)  Despite the growing ubiquity of the Web, there are still concerns about the quality of information, yet there appears to be little that can be done outside of user education.  Nonetheless, there have been a number of new research findings.

Eysenbach and Kohler attempted to measure how often consumers use the Web for health-related searches.  While previous research looked at how often people searched for health information one or more time, this study attempted to measure how many search-engine queries were actually health-related.  They found about 4.5% of queries to the Metacrawler search engine were on health topics.  The most common topics included

• Health care services and organizations - 9.6%
• Medications - 8.1%
• Diet, nutrition, and weight loss - 6.7%
• Infectious diseases - 5.9%
• Cancer - 5.2%
• Rheumatology - 5.2%
• Women′s health (excluding pregnancy) - 5.2%
• Addiction - 4.4%
• Mental health - 4.4%
• Cardiovascular - 3.7%
• Pregnancy and fertility - 3.7%
• Dermatology - 3.7%
• Natural/alternative medicine - 3.7%
  

A number of publications continue to promote frameworks and instruments for assessing quality of information on the Web (Wilson, 2002; Risk, 2002).  Some have suggested that well-known tools for documenting quality, e.g., the voluntary HON Code, is subject to manipulation (Wanjek, 2004) 数据挖掘实验室

A group of studies looked at the quality of information related to complementary and alternative medicine (CAM).  Walji et al. (2004) assessed 150 sites for information related to three CAM treatments:  ginseng, gingko, and St. John′s Wort.  They found that one-quarter of the sites contained statements that could lead to physical harm.  Almost all sites had omitted information about potential harm of the treatments.  There was no association between common measures of technical quality (the Eysenbach, 2002 systematic review cited in the book) and potentially harmful information.  Bernstam et al. (2005) have identified 273 distinct tools pruporting to help consumers assess the quality of health information on the Web, yet only seven met the usablity criteria of having evaluation criteria publicly available, having 10 or few items, and having elements that could be objectively evaluated.

Bernstam, E., Shelton, D., et al. (2005). Instruments to assess the quality of health information on the World Wide Web:  what can our patients actually use? International Journal of Medical Informatics, 74: 13-20. 数据挖掘研究院
Eysenbach, G. and Kohler, C. (2004). Health-related searches on the Internet. Journal of the American Medical Association, 291: 2946.
Risk, A. (2002). Commentary:  on the way to quality. British Medical Journal, 324: 601-602.
Walji, M., Sagaram, S., et al. (2004). Efficacy of quality criteria to identify potentially harmful information:  a cross-sectional survey of complementary and alternative Web sites. Journal of Medical Internet Research, 6(2): e21. http://www.jmir.org/2004/2/e21/.
Wanjek, C. (2004). Attacking Their HONor. Washington Post. April 20, 2004. HE01. http://www.washingtonpost.com/wp-dyn/articles/A25556-2004Apr19.html.
Wilson, P. (2002). How to find the good and avoid the bad or ugly:  a short guide to tools for rating quality of health information on the internet. British Medical Journal, 324: 598-602. 数据挖掘实验室

(4/5/03)  An overview of the major quality initiatives was published by Risk and Dzenowagis (2001).   These authors reviewed the various efforts and their specific approaches.  They also came to ten conclusions:

1. Internet health quality initiatives generally follow one of three mechanisms:  codes of conduct or ethics, third-party certification of compliance (accreditation), and tool-based evaluation of quality
2. Based on the analysis of quality initiatives, successful quality programs should have three essential elements:  a set of health information quality criteria, an educated, interested, and active citizen, and credible enforcement instrument(s).  None of the initiatives discussed in the paper comprise all three completely.
3. The current quality initiatives have gaps that need to be addressed (as discussed and listed in the paper). The most serious of these are the excessive burdens placed upon citizens and the cost of implementing credible programs providing accreditation and enforcement.  Further examination and the addressing of these gaps is necessary for further development work on Internet health information governance.
4. Additional research is needed to further clarify the complexities of Internet health information.  Most important are "the governance mechanisms that need to address quality of information content and information value, the context and relevance of the content of information, the educated interested citizen, and the desired instruments that would strengthen any envisaged enforcement provision."
5. An urgent requirement is the examination of the needs of the developing world and the info-poor in relation to quality of Internet health information, products, and services. The current batch of quality initiatives do not address those needs well.
6. There are currently no mechanisms for ensuring the quality of Internet health information in relation to the "pseudo-health" sector. This sector will probably remain outside the efforts to implement governance of health information quality.
7. The existing quality initiatives do not address issues in complementary and alternative medicine.  ("These disciplines differ from those of the pseudo-health sector in that they have a legitimate place in health care, whereas the pseudo-health sector is essentially about fraud and quackery.")
8. Language remains "a major obstacle to the dissemination of good practices and the education of citizens and information providers alike."
9. There is a need for "coordination and harmonization of the efforts striving towards quality health information on the Internet. This extends to the key players in both the self-regulation and the mainstream and regulatory camps, and includes regional and international bodies, the health care products industry, foundations with an explicit interest in Internet health information, private and corporate interests, and citizen and country representation and participation."
10. There are concerns and criticisms directed against that must be addressed regarding models of governance for Internet health information quality. These arguments come from different perspectives and take different routes but arrive at the same destination.  Among these concerns are that users are ambivalent or indifferent about quality, quality programs that are not rigorously enforced and validated might produce a false sense of security, traditional media did not require quality standards; therefore neither should the new media, brand loyalty is more important than quality seals, and "kitemarking (referring to the application of a kite-shaped mark granted for use on goods approved for use by the British Standards Institution) the Internet is like ′kitemarking the west wind,′" freedom of speech, free market forces, and the practical and logistical difficulties associated with implementing quality programs.

Risk, A. and Dzenowagis, J. (2001). Review of internet health information quality initiatives. Journal of Medical Internet Research , 3: e28. http://www.jmir.org/2001/4/e28/ .

(3/12/03)  Another study attempted to characterize the sites found by five Web search engines (Yahoo, America Online, Micosoft Network, Lycos, and Go; it is interesting that they did not look at Google) for eight terms from each of three medical conditions (heart disease, cancer, and weight loss) (Salter and Zimmerman, 2002).  The top 10 sites were analyzed from each of the 5 portals for the 24 terms, yielding a total of 1,200 sites.  Sites listed were categorized as having content based on scientific or unproven claims.  They were also classified as to whether they offered products or services.  The largest number of sites were non-promotional science-based (35.3%), followed by promotional non-science-based (21.8%), promotional science-based (11.1%), and non-promotional non-science-based (0.8%).  The remaining 31.0% were book catalog sites, personal pages, unavailable sites, and so forth.

Slater, M. and Zimmerman, D. (2002). Characteristics of health-related Web sites identified by common Internet portals. Journal of the American Medical Association, 288: 316-317.

数据挖掘研究院

(4/5/03)  An additional study evaluated over 460 sites for factors related to their credibility (Anonymous, 2002).  Quoting from their results:

• 49% of health and financial sites failed to give warnings about the appropriate use of their information. For example, they did not warn consumers searching for health or financial advice that they should consult a professional before acting on advice given.
• At least 50% of sites giving advice on medical and financial matters failed to provide full information about the authority and credentials of the people behind that advice.
• Only 57% of general advice sites gave sources for that advice.
• 39% of sites that collected personal information did not have a privacy policy.
• 62% of sites contained claims that were vague and unspecific.
• 55% of sites said nothing about how up-to-date their content was.
• 30% of sites provided no address or telephone number.
• Only 41% of the sites that recommended products gave sources for their prices.
• 26% of sites gave no clear information about who owned them.
• 60% of sites provided no information that indicated whether or not their content was influenced by commercial interests (e.g., partners, sponsors or advertisers).

Anonymous (2002). Credibility on the Web. Consumers International. http://www.consumersinternational.org/newsdocs/{FAC5B4E3-2CC7-483A-9C46-92F8A0305602}.doc . 数据挖掘研究院

(4/5/03)  The Federal Trade Commission (FTC) project to combat promulgation of false or unsubstantiated information on Internet is now called Operation Cure.All:
http://www.ftc.gov/bcp/conline/edcams/cureall/

(4/5/03)  A book covering misinformation of the Internet in general (including health misinformation) has been published:
Mintz, A., ed. (2002). Web of Deception:  Misinformation on the Internet . Medford, NJ. Information Today.

(4/3/04)  In a follow-up to their study on "too much, too soon" reporting of findings presented as abstracts at medical meetings, Schwartz and Woloshin (2004) recommend the following principles to journalists:

• In general, do not report preliminary findings.
• Communicate absolute (not relative) magnitudes of differences.
• Include caveats, i.e., limitations of studies.
  

They also advocate that the medical community also act responsibly by stopping courting of coverage of preliminary work and writing press releases to communicate the science and not generate press coverage.  The latter should apply to biomedical journals, who also jockey for coverage in the general media. 数据挖掘研究院

Schwartz, L. and Woloshin, S. (2004). The media matter:  a call for straightforward medical reporting. Annals of Internal Medicine , 140: 226-228.

2.7  Use of knowledge-based health information

(4/3/05)  Here is an interesting use of data from IR systems:  biosurveillance.  Johnson et al. (2004) assessed Web logs from the standpoint of surveillance for disease, something that could have great value in the early detection of bioterrorism.  They found a modest correlation between information seeking of influenze-related articles during epidemic periods.  The unanswered question is whether Web search logs might provide clues to early detection of other diseases, in particular those associated with bioterrorism.

Johnson, H., Wagner, M., et al. (2004). Analysis of Web access logs for surveillance of influenza. MEDINFO 2004 - Proceedings of the Eleventh World Congress on Medical Informatics, San Francisco, CA. IOS Press. 1202-1206.

数据挖掘研究院

2.7.1  Models of physician thinking

2.7.2  Physician information needs

(4/3/05)  Bryant (2004) recently published a quantitative and qualitative analysis of information needs of family physicians in Aylesbury Vale region of the United Kingdom.  The most common perceived information needs were for clinical care, keeping up to date, and providing information for patients.  The most commonly used resource was the physician′s personal collection, followed by electronic resources.  Medical library use was found to be small and declining.  Physicians in practices that had medical trainees were somewhat more likely to use electronic resources as well as the library.

Bryant, S. (2004). The information needs and information seeking behaviour of family doctors. Health Information and Libraries Journal, 21: 84-93.

(2/7/03)  Another resource on physician information-seeking is, Verhoeven, A. (1999). Information-Seeking by General Practitioners . Groningen. Rijksuniversiteit Groningen.  The book is available on-line at http://www.ub.rug.nl/eldoc/dis/medicine/a.a.h.verhoeven/ .

数据挖掘实验室

(2/12/03)  A book information seeking in general was recently published by Case (2002).

Case, D. (2002). Looking for Information:  A Survey of Research on Information Seeking Needs and Behavior. San Diego. Academic Press.

2.7.2.1  Unrecognized needs

(3/10/03)  The updated taxonomy of clinical questions from Ely, Osheroff et al. (2000) includes 64 generic question types and is available at:
http://bmj.com/cgi/content/full/321/7258/429/DC1

2.7.2.2  Recognized needs

(4/1/03)  Another study of questions generated in practices comes from New Zealand, where a group of 50 family physicians was observed (Arrol et al., 2002).  These physicians generated questions at a slightly lower rate than found by Covell et al., Gorman et al., and Ely et al., with an average of one question for every 3.4 patients.  The types of questions and their proportions were comparable to the other studies.

数据挖掘实验室

Arroll, B., Pandit, S., et al. (2002). Use of information sources among New Zealand family physicians with high access to computers. Journal of Family Practice, 51: 8. http://www.jfponline.com/content/2002/08/jfp_0802_0706a.asp .

2.7.2.3  Pursued needs

(4/1/03)  The Arroll et al. (2002) study described above had a similar finding with other studies that computers were infrequently used to pursue answers.  Only 5% of questions were answered using a computer.  One difference from earlier studies was that 78% of these physicians had computers in their offices, and most could access well-known clinical information resources.  Thus, even though computer-based information resources were available in these physicians′ offices, their use was still modest.

(3/14/03)  Ely and colleagues (2002) have also assessed obstacles to obtaining answers to clinical questions.  They identified a total of 59 obstacles to accessing information.  These were organized into five steps in asking and answering questions:  recognizing a gap in knowledge, formulating a question, searching for relevant information, formulating an answer, and using the answer to direct patient care.  They noted six obstacles that were particularly prominent to themselves and the clinicians they observed, quoted as follows:

• The excessive time required to find information
• Difficulty modifying the original question, which could be vague and open to interpretation
• Difficulty selecting an optimal strategy to search for information
• Failure of a seemingly appropriate resource to cover the topic
• Uncertainty about how to know when all the relevant evidence has been found so that the search can stop
• Inadequate synthesis of multiple bits of evidence into a clinically useful statement

Ely, J., Osheroff, J., et al. (2002). Obstacles to answering doctors′ questions about patient care with evidence:  qualitative study. British Medical Journal, 324: 710-713.

(4/7/05)  Ely et al. (2005) recently continued their work assessing barriers to information for physicians.  They followed 48 primary care physicians from small towns in Iowa, prompting them for information needs after they saw each patient during a half day.  They later interviewed these physicians to ask them what recommendations they would have to improve knowledge resources. 数据挖掘实验室

The 48 physicians generated 1,062 questions, an average of 5.5 per half day.  Of these questions, 441 arose during the observation period (with the remainder arising before then.)  Of the 1,062 questions, 55% were pursued.  Of those pursued, 41% were answered without difficulty, 31% were answered with difficulty, and 28% were not answered.  The most common categories of resources used to answer questions were:

• Single textbook - 31%
• Another human (informal consultation) - 18%
• Desktop computer application - 12%
• Mutiple textbooks - 8%
• Human plus textbook - 6%
• Single journal article - 4%
• Handheld computer - 4%

No non-human resource was used for more than 7% of the answers, with the most common being the desktop computer application Up to Date (7%), the textbook Nelson Textbook of Pediatrics (6%) and the handheld drug reference ePocrates (4%). 数据挖掘研究院

The reasons for not pursuing an answer were identified for 212 questions (the physicians were too rushed or otherwise busy to provide reasons for the remainder).  The most common reasons for not pursuing an answer were:

• Doubted existence of relevant information - 25%
• Readily available consultation leading to referral rather than pursuit - 22%
• Lack of time to pursue - 19%
• Not important enough to pursue answer - 15%
• Uncertain where to look for answer - 8%

After the observation period, each of the 48 physicians was interviewed to provide recommendations for the knowledge resource developers.  A qualitative analysis broke the recommendations down into content and access issues.  The key content issues recommended were for comprehensive and trustworthy information.  The important access issues were user-friendly and intuitive search function, rapid access to information, concise operation, and available anywhere all the time.

Ely, J., Osheroff, J., et al. (2005). Answering physicians′ clinical questions:  obstacles and potential solutions. Journal of the American Medical Informatics Association, 12: 217-224.

2.7.2.4  Satisfied needs

2.7.3  Information needs of other health care professionals

(4/6/04)  As noted in the text, studies of the information needs of non-physician clinicians are relativly rare.  One recent study looked at information needs and seeking of nurse pracitioners (NPs) (Cogdill, 2003).  The findings were not much different than those noted for physicians.  The most frequent information needs reported were somewhat more focused than physicians and on two specific areas:  drug therapy (43%) and diagnosis (41%).  Most commonly used information sources were supervising physicians, durg reference manuals, textbooks, journal articles, and other NPs.  The generalness of a need was found to be a negative predictor of information seeking. 数据挖掘研究院

Cogdill, K. (2003). Information needs and information seeking in primary care:  a study of nurse practitioners. Journal of the Medical Library Association, 91: 203-214.

(4/3/05)  One group in biomedicine much less studied than clinicians is researchers.  In recent years, there has been some effort to look at the information needs of genomics researchers.  This group would presumably have great information needs due to the rapid growth of new "high throughput" biotechnologies that generate vast amounts of data, which in turn require researchers to need information about new genes, proteins, etc. involved in the biological systems they study.  A prototype biotechnology is the gene microarray, which tests the expression of tens of thousands of genes (via their messenger RNA) in a given biological sample (Moshaberi, 2004).  A researcher using a microarray might find dozens or more genes differentially expressed and now have the need to find out information about those genes or their protein products rapidly.

数据挖掘研究院

Very little attention has been paid to the information tasks of biomedical researchers generally.  Stevens et al. (2001) were the first to assess the information tasks in bioinformatics.  Most tasks involved sequence similarity searching, with literature searching representing a tiny fraction of their work.  Tran et al. (2004) carried out a task analysis on a smaller sample of researchers but in more detail.  There were too few researchers to generate reliable quantitative assessment of IR tasks, but several themes emerged from the analysis, including that laboratores lack procedural documentation of their information-related work, use and stick to "home-grown" strategies, and unaware of many tools that are available.  In a proprietary report, Strouse (2004) notes that biomedical researchers spend about nine hours (18%) of their work week gathering and reviewing information, a figure verified by bioinformatician William Hayes of AstraZeneca (personal communication).

Mobasheri, A., Airley, R., et al. (2004). Post-genomic applications of tissue microarrays: basic research, prognostic oncology, clinical genomics and drug discovery. Histology and Histopathology, 19: 325-335. 数据挖掘研究院
Stevens, R., Goble, C., et al. (2001). A classification of tasks in bioinformatics. Bioinformatics, 17: 180-188.
Strouse, R. (2004). Content User Profile:  Update On Scientists. Burlingame, CA, Outsell Inc. http://content.outsellinc.com/coms2/summary_0245-888_ITM.
Tran, D., Dubay, C., et al. (2004). Applying task analysis to describe and facilitate bioinformatics tasks. MEDINFO 2004 - Proceedings of the Eleventh World Congress on Medical Informatics, San Francisco, CA. IOS Press. 818-822.

2.8  Evidence-based medicine

(4/1/03)  The Oxford University Centre for Evidence-Based Medicine has changed its URL twice since the book was written.  Its official URL is now http://www.cebm.net/ .

(4/5/04)  Bakken (2001) has noted that an informatics infrastructure, which includes standards for clinical data interchange and clinical terminologies, electronic sources of evidence for practitioners, informatics processes that support the application of evidence to specific situations, and informatics competencies of clinicians, is essential to achieve evidence-based practice.

Bakken, S. (2001). An informatics infrastructure is essential for evidence-based practice. Journal of the American Medical Informatics Association , 8: 199-201.

(4/6/04)  Chapter 3 of the texbook discusses the difference between statistical and clinical significance of experimental results.  Statistical significance means the results are not due to chance whereas clinical significance means they are truly meaningful.  Alderson (2004) describes this difference in the context of EBM and also notes that "negative" studies without statistical significance must be carefully reported to not imply that they necessarily "prove" no effect.

Alderson, P. (2004). Absence of evidence is not evidence of absence. British Medical Journal, 328: 476-477.

2.8.1  First-generation EBM

(4/6/04)  The first RCT is usually ascribed to a study carried out by the Medical Research Council of the United Kingdom and published in the British Medical Journal in 1948 (Anonymous, 1948).  However, the earliest trials may have been performed by James Lind, M.D. in the mid-1700s, where he detemrined that lack of vitamin C was the cause of scurvy in sailors on British ships.  A Web site devoted to Lind′s work has been developed at http://www.jameslindlibrary.org . 数据挖掘实验室

Anonymous (1948). MRC Streptomycin in Tuberculosis Trials Committee.  Streptomycin treatment of pulmonary tuberculosis. British Medical Journal , ii: 769-783.

2.8.2  Next-generation EBM

(4/6/04)  As noted in the text, a key motivation for "next-generation" EBM is its better usefulness for clinicians.  Ebell (2003) notes we are making progress but still have a ways to go in getting there.

Ebell, M. (2003). Point-of-care information that changes practice:  it′s closer than we think. Family Medicine, 35: 261-263.

2.8.3  Limitations of EBM

(4/5/04)  Cohen et al. (2004) have published a categorization of the major criticisms of EBM.  In particular, they note the following problems with EBM from a philosophical standpoint:

• It relies solely on empiricism and not other forms of scientific investigation and analysis, such as physiological processes and reasoning.
• Its definition of evidence is narrow and excludes other types of scientific studies, e.g., cohort studies
• It is not evidence-based, i.e., the practice of EBM itself has not been demonstrated to lead to improved clinical outcomes.
• Its usefulness to individual patients is limited, e.g., studies of highest quality evidence are not necessarily applicable to individual patients and tend to be limited to the most common diseases.
• It may threaten the autonomy of the patient-physician relationship by limiting diagnostic and therapeutic options. 数据挖掘研究院
  

A tounge-in-cheek criticism (with many inside jokes) has also been published (Anonymous, 2002).

Cohen, A., Stavri, P. and Hersh, W. (2004). A categorization and analysis of the criticisms of evidence-based medicine. International Journal of Medical Informatics, 73: 35-43.  (Note:  this paper began as a term paper in my Introduction to Biomedical Informatics course!)
Anonymous (2002). EBM:  unmasking the ugly truth. British Medical Journal, 325: 1496-1498.

(3/1/03)  As noted in section 2.3.2 (see above), the "truth survival" for conclusions in the domains of cirrhosis and hepatitis is not any higher for those obtained using more rigorous evidence-based methods.

(4/1/03)  R. Brian Haynes, one of the founders of the concept of EBM, wrote a historical perspective on EBM in 2002.  He noted in retrospect that the originators of EBM should probably not have touted it as an alternative paradigm for medical practice, but instead should have promoted it augment (rather than replace) individual clinical experience and understanding of basic disease mechanisms.  He advocated that EBM must evolve to not only respond to theoretical and moral challenges in its use but also to address practical concerns such as dissemination and generalizability.

数据挖掘研究院

Haynes, R. (2002). What kind of evidence is it that Evidence-Based Medicine advocates want health care providers and consumers to pay attention to? BMC Health Services Research, 2: 3. http://www.biomedcentral.com/1472-6963/2/3 .