Stumped? 5 steps to find the latest evidence

Evidence hierarchies (Table 1) ^5,6,12,14 indicate that certain types of evidence are considered more credible than others. The higher the level of evidence, the more likely it is to provide valid, unbiased estimates of an intervention’s effect.

Table 1

Hierarchy for studies of therapy or harm

For example, observational studies may give misleading results, compared with randomized, controlled, clinical trials.¹⁵ Expert opinion—which does not necessarily reflect the best evidence in the literature—is considered the lowest evidence level.¹⁶

Beginning the literature search with Medline—as many clinicians do—is relatively inefficient. Searching the massive National Library of Medicine database often identifies a large number of articles, which you then must appraise for validity.

More efficient is Haynes’ “4S” strategy of:

• systems
• synopses
• syntheses
• studies.^12,17

It gives priority to sources of high-quality, pre-appraised information ( Table 2), so that you may omit Step 3—appraisal—in the EBM process.

Table 2

“4S” approach to answering clinical questions

Systems. A system is an information source that covers a variety of diagnoses, summarizes the results of high-quality systematic reviews, is frequently updated, and is linked to the original studies. Three examples are:

• Clinical Evidence, a journal published semi-annually in the United Kingdom that is also available online (www.clinicalevidence.com). Its ease of use, frequent updates, and clear links to the best evidence make it a good starting point.
• National electronic Library for Mental Health (NeLMH) (www.nelmh.org), created by the U.K.’s Centre for Evidence-Based Mental Health.¹⁸ It provides information only about depression, schizophrenia, and suicide but may become the model for evidence-based psychiatry.
• Evidence-based clinical practice guidelines developed by the American Psychiatric Association⁹ and other organizations. Many of these are included in the National Guideline Clearinghouse (www.guideline.gov), produced by the U.S. Department of Health and Human Service’s Agency for Healthcare Research and Quality.

Synopses. If a system does not answer a clinical question, look for synopses—structured abstracts of high-quality systematic reviews or original articles. Synopses are brief and pre-appraised for quality, allowing you to quickly get the point without reading a lengthy article. For psychiatry, Evidence-Based Mental Health and ACP Journal Club are the best synopses sources.

Syntheses. If you don’t find a synopsis to answer your question, then search for a synthesis—a high-quality systematic review. The best source is the Cochrane Database of Systematic Reviews;²⁰ others include:

• Database of Abstracts of Reviews of Effectiveness (DARE)
• Health Technology Assessment (HTA) database.

Table 3

Web sites for learning more about EBM

Studies. Only if the first three “S’s” fail to yield an answer would you then search Medline or similar databases for applicable studies.

The PubMed Clinical Queries interface lets you specify the type of question (therapy, diagnosis, etiology, or prognosis) and whether the search is to be sensitive or specific. You can also use filters built into this interface when searching for systematic reviews.

Step 3: Appraise the evidence. After you find an article, appraise its validity and importance. Checklists for appraising studies may be found in EBM texts^4,6,12 or downloaded from Web sites (Table 3). Specific appraisal questions to ask depend on the study design and your clinical question.

Step 4: Apply the results to your patient. Assuming the evidence is valid, important, and feasible in your setting, consider your patient’s preferences²¹ and apply the results.

Step 5: Assess the outcome. Evaluate your performance in searching the literature, and assess the patient’s response.

STREAMLINING THE EBM PROCESS

Studies in academic settings have shown that the full 5-step model can be incorporated into daily practice.^6,12 In nonacademic settings, however, practitioners complain about lack of time and information resources, as well as inadequate EBM skills.^11,12 Two shortcuts can streamline the process:

Use the 5-step process selectively; it is not required for every patient encounter.^6,12 After you research a question for one patient with a particular diagnosis, you can apply the answer to similar patients. Because most psychiatrists’ patients fall into relatively few diagnostic categories, only the exceptional patient would trigger the full 5-step process.