What is EBM

After you have a decent understanding of the basics of biostatistics and epidemiology, we can now move to using these to read the medical literature. Evidence-based medicine is actually more than evaluating articles.

Using evidence in the practice of medicine requires minding three areas:

  1. the best available evidence,
  2. the clinical scenario and
  3. the patient’s values.

There are four steps to carry out this process.

  1. ASK a clinical question (PICO).  During the care of your patients, you’ll come across something to which you don’t know the answer. This uncertainty is the start of your EBM journey. If the answer is well established can easily be found in a textbook, it’s considered background information. If the answer is being newly discovered and found in the literature, it’s considered foreground information.You’ll need to shape that question into one which can be fit to the literature. This most commonly used format is PICO.
    • P = patient – describe your patient, demographics, co-morbidities, etc so you can find as good a match of patients in the literature
    • I = intervention – what is the intervention about which you have a question? Are wondering about the effectiveness of a new drug, the potential harm of continued exposure to a toxin or the usefulness of a new blood test in diagnosing your patient.
    • C = comparison – you need to compare your intervention against something. Usually, this will be against what a placebo drug (for new treatments), unexposed people (for patients exposed to a toxin) or a gold standard test (for new diagnostic tests).
    • O = outcome – what is the outcome you’re looking at? Will the patient’s disease improve? Will the exposure cause cancer or death? Or will the new blood test have a high sensitivity in picking up the disease?
  2. ACQUIRE the evidence. Next, you need to acquire the evidence. This is easier said than done. If you put the question into a search engine you may get thousands of irrelevant hits. Even specialized search engines such as PubMed or Google Scholar can lead to unrelated articles. Using the proper search terms and tools helps narrow down the results.
  3. APPRAISE the evidence. Once you have the article, you need to read it and appraise it’s validity. Is this paper done well? Can you trust the results? Can you use the results? The famous Users Guide to the Medical Literature (from JAMA in the 1990’s by Guyatt and Leavitt) provides a framework for how to evaluate papers based on the type of question.
  4. APPLY the evidence to your patient. If you’ve determined the paper is usable, then you need to apply it to your patient. Part of this should include an audit of how well that evidence worked.

Ask a Question (PICO)

  • P: Population: Which factors, demographic and otherwise, best describe your patient scenario? These may include age, gender, diagnosis, severity of diagnosis, clinical setting, and so on. Keep in mind that the more specific this part of the question becomes, the more difficult it may be to find relevant studies.
  • I: Intervention: What is the intervention of interest? This may be a diagnostic test, a new medication, an interventional program, or an exposure, to name a few. Again you want to best describe the intervention that you are considering for your patient, and this may include factors such as timing, duration, or dose of the intervention. Keep in mind that at times you may want to know if an intervention is better than the standard intervention, and at other times you may want to know if an intervention is as good as the standard intervention.
  • C: Comparison: What is the comparison of interest? If you are looking for studies regarding a diagnostic test, the comparison should be to the “gold standard” diagnostic test. If you are looking for studies regarding a new medication or therapy, the comparison should either be to placebo or to the “standard of care” medication or therapy. If you are looking for studies regarding exposures, the comparison should be to no exposure. If you are looking for studies regarding prognosis, there usually is no comparison group.
  • O: Outcome: What is the outcome or outcomes of interest? The outcome of interest should be measurable and should be patient-oriented. This often is described simply as “improved outcome for the patient”, but this may be measured in a number of ways. Be cognizant of whether a measured outcome is clinically relevant.

So instead of asking “Hmm, is this new medication, Rivaroxaban, any good for treating pulmonary embolus?” You would ask:

  • P: Population: In otherwise healthy patients between the ages of 18 to 65 who present to the emergency department with a pulmonary embolus…
  • I: Intervention: …does outpatient treatment with the medication Rivaroxaban…
  • C: Comparison: …when compared with the usual treatment of three days of admission with heparin and subsequent treatment of oral warfarin…
  • O: Outcome: …result in similar 30 day mortality or hospital re-admission rates? 

Acquire the Evidence

This used to be much tougher prior to the days of Google Scholar and Pub Med. Now AI engines such as perplexity.ai can help you find articles much more easily.

Appraise the Evidence

Use the Sackett and Guyatt rubrics to evaluate the article you found. Every student will present their article to the group. No slides. Less than 5 minutes each. Then as a group come to a consensus opinion based on all your articles, regarding the recommendation to follow.

Here are the rubrics to evaluate articles based on question-type.

Apply To My Patient

These we’ll go through in the separate pages related to the different article types.