Evidence-Based Medicine Working Group. (1992). Evidence-Based medicine: A new approach to teaching the practice of medicine. JAMA, 268 (17), 2420-2425.
Evidence-based medicine is a new paradigm that places emphasis on new skills for physicians that include: performing efficient in performing literature searches and applying formal rules of evidence in examining clinical literature (critical appraisal exercise, which applies when authority is not trusted, the answer unknown, or there are divergent opinions). This is in addition to traditional clinical skills, understanding patients' emotional needs
This represents a shift from old processes used by physicians such as intuition, unsystematic clinical experience, and pathophysiologic rationale. Discusses Kuhn's notions of paradigms and paradigm shifts: paradigms are ways of viewing the world that define the problems addressed and the range of admissible evidence that can be used to solve them. Paradigm shifts occur when the old paradigm does not answer problems and a new paradigm in line with the evidence replaces it.
Paradigm shifts in medicine developed due to randomized controlled trials (RCTs; rare up until 30 years before this article), applications of RCTs in diagnostic tests and surgical therapies, meta-analysis usage, profusion of articles published and their structure, practice guidelines, updated textbooks that review the literature.
The former paradigm is characterized by: 1) Using unsystematic clinical observations to build knowledge, 2) believing that understanding the basic pathology of disease is a guide for practice, 3) medical training and common sense allow the physician to evaluate the efficacy of medical treatments, 4) clinical experience and content expertise are sufficient to generate valid guidelines for clinical practice, 5) authority and expert opinions.
The new paradigm is characterized by: 1) understands the value of clinical experience but experiences and observations should be recorded systematically and evaluated for biases and be replicable in other settings, 2) studying pathophysiology and disease are necessary but not sufficient to guides and predictions made without empirical evidence may be flawed, 3) understanding the rules of evidence is required to evaluate the literature concerning causes, prognoses, tests, and treatment strategies, 3) regularly updating the literature and reading it keeps the process current, 4) on evidence.
Methodological criteria 1) diagnosis test applied to appropriate sample, 2) random assignment to treatment conditions and dropouts accounted for, 3) review articles determined explicitly.
Misinterpretations: 1) Ignores clinical intuition and experts- answer no, and expertise is developed by this method, intuitions are tested, 2) understanding pathophysiology plays no part- answer, it is necessary for interpretation of observations and evidence, 3) ignores standard aspects of training such as physical exam- no these provide much information and are empirically evaluated.
Barriers to teaching 1) topic threatening 2) it's work! 3) lack of evidence for some questions 4) people find change hard.
Barriers to practice 1) lit not there 2) economic issues 3) time constraints.
Evidence for effectiveness -at time of paper no long-term studies; short-term studies indicate evidence-based trainees more up-to-date on HTN guidelines than traditional
Smith, C.G., Herzka, A.S., & Wenz, J.F. (2004). Searching the medical literature. Clinical Orthopaedics and Related Research 421, 43-49.
MEDLINE
MEDLINE from the National Library of Medicine (NLM) indexes 4498 journals, whereas there are 40, 000 medical journals world-wide. Medline became free (6-26-97) and more leading to the misconception that searching was easy -- the authors report that this is a skill that takes practice and offer advice.
MEDLINE interfaces are Gateway (access to MEDLINE [1966-present]; oldMEDLINE [1958-1965]; and PreMEDLINE [still in process]) and PUBMED (MEDLINE and PreMEDLINE). These NLM indexes use MeSH (medical subject headings) vocabulary. Using more descriptors reduces error rates. NLM interfaces are not only access points; commercial sources also provide interfaces with own search methods and tutorials
Text word searching looks for text anywhere in articles, but matches letter sequence in the search and does not search by content. Can use all possible terms separated by Boolean "OR" operator will reduce this issue.
Subject word searchers are arranged in a MeSH tree from greatest generality to greatest specificity. You do not need to specify all forms for a subject, spellings, or synonyms. A problem is that articles indexed before a term was introduced are indexed under the closest related topic and not the term.
Search subject terms involves making a question, breaking it down in to conceptual parts, accessing MeSH database, entering in each concept, selecting LINKS to the right of the term wanted. Links choices PubMed (will run search) and NLM MeSH database (gives additional indexing info and displays tree)
Automatic explosion occurs when a broad term is used narrower related terms will be searched. Can turn off by clicking "Restrict Search to Major Topic Headings" and "Do not explode this term." Other techniques can be found in tutorials listed in Table One
By entering a term and selecting LIMITs link one can search select fields,...
Group names can be searched in all fields or the Collective Name Field that became available in 2003 (e.g., corporate authorship). Collective name appears after other names
Abbreviations can be used as search terms and may be required for certain concepts in addition to the concept
The Single Citation Matcher at left margin of PubMed page can be searched to locate a paper with any combination of information. Another way is to enter info without connectors into main search box. The more elements (at least 3) will result in short list. Use least common search terms to narrow down.
Automatic Term Mapping feature searches for unqualified terms looking for a match in several lists: Subject, if it doesn't find a match, it looks as a journal, then author and investigators. When any match, the mapping stops, if no match occurs it breaks apart the phrase and repeats until a match is found.
Can filter searches by study design or type. Table two lists popular search services
Other resources (Table three compares selected resources for locating literature citations)
Recent study- 85% of family dr's clinical questions could be answered by STAT!Ref or MDConsult
Cochrane CENTRAL register has >300, 0000 clinical trial reports
Science Citation Index - costs $ to use. Indexes 5700 journals. Has tow searchers: General Search is for topic searchers and Science Citation Search to use a given work as a subject term to find more articles on that subject.
Current Contents- accesses 8000 journals and 2000 books. Useful for latest lit on a topic
EMBASE- is expensive and emphasizes pharmacology and toxicology
CINAHIL- primarily for nursing and allied health practitioners and alternative techniques
InfoRetriever- literature with greatest validity to a clinical problem
National Guideline Clearinghouse- 19 guidelines for orthopedic procedures
TRIP -- evidence-based medicine (British search engine)
Index Medicus -- older publications prior to 1950s
Librarians-can be useful in helping to navigate search engines
PubMed -loansome.Doc allows ordering of full text articles
Laupacis, S. & Straus, A. article
Well-done randomized trials (RTs) give the most valid estimate of health interventions as they minimize bias. Systematic reviews identify all studies addressing a particular question; meta-analyses combine study results and minimize bias; systematic reviews are considered the best source of information for making clinical and policy decisions.
Groups like the Cochrane Collaboration established standards for the conduct and reporting of systematic reviews
Shojania and colleagues describe 100 systematic reviews in ACP Journal Club quickly became outdated when new RTs were added.
50% were 5.5 years after publication and 23% were out of date within 2 years; most likely to change were those for cardiovascular interventions
Researchers should update a literature search annually which should take little time as same search strategy should be used.
It is time make the content and format of reviews useful to a variety of decision makers.
Current evidence suggests they are used infrequently by clinicians, health care workers policymakers and patients
Physicians and other use textbooks first, followed by advice from colleagues; reasons for not using reviews include:
Most address highly specific questions that interest the author and not policymakers, clinicians, patients, or health care managers about what's important to them.
Clinicians questions are broad; researchers in reviews more narrow
Reviews are dependent on the relevance of the existing randomized trials
Trials that most reviews use selected patients and physicians and raising questions about their generalizability.
RTs underreport adverse events; often do not study patients long enough to detect important side effects.
Systematic reviews are lengthy (often longer than 30 pages), appear complicated, and take a long time to read.
Authors focus methodological rigor failing to describe the policy or clinical context, thus not providing information for policymakers
Glasziou and Shepperd found < 15% of reviews published in Evidence-Based Medicine provide sufficient information to allow clinicians or policymakers to implement it.
Reviews regarding the cost-effectiveness, budget impact, financial, and health care delivery strategies can be found but there are not many of them for policymakers.
Ways to encourage the use of reviews:
Provide more exposure to systematic reviews during clinical training.
More interaction among researchers, policymakers, and clinicians, so studies can be aimed at answering questions of interest to policymakers and clinicians.
Policymakers should commission RTs of policy-relevant questions
Authors of reviews should provide more contextual information or refer readers to easy-to-find articles
Authors of reviews should develop user-friendly formats: this article suggests 3 formats: a 20-second version of 1-paragraph…
