OmanDocSign in

Curriculum · Public Health and Communicable Diseases

Study design

What it is

EBM is the best available evidence for diagnosis, prognosis, or treatment. Clinical decision making requires three principles: 1. awareness of the best available evidence, 2. level of evidence, and 3. benefits and risks, burden, and costs. The GRADE approach, Grading of Recommendations Assessment, Development and Evaluation, allows one to rate confidence in estimates of effects. That confidence is rated on four levels: 1. high, 2. moderate, 3. low, 4. very low. The study design it is read against is either randomized or observational. It goes lower with 1. risk of bias, 2. inconsistency, 3. indirectness, 4. imprecision, 5. publication bias; and higher with 1. a large effect, 2. a very large effect, 3. a dose response gradient. The five steps to practice it are 1. ask, 2. acquire, 3. appraise, 4. apply, and 5. act or assess. Worked as a scenario: asking whether a 3-day course of trimethoprim-sulfamethoxazole in adolescents with acute maxillary sinusitis yields the same cure rates as a 10-day course with fewer adverse effects and costs is a therapy question, and its best level of evidence is a systemic review of randomized clinical trial. In practice a recommendation is written together with the quality of the evidence behind it, so that the same guideline carries a strong recommendation on low-quality evidence and a strong recommendation on moderate-quality evidence, each with values and preferences given beside it. Evidence itself is also given a level, as level 2+ or level 2-.

How the question picks the design

A background question is general knowledge, from a textbook or clinical database. A foreground question is specific knowledge and requires a search of the primary medical literature; it is put in the PICO format, where P is the population or patient or problem, I is the intervention or exposure such as diagnostic tests, foods, drugs, surgical procedures, time, risk factors or prevention, C is the comparison or control, and O is the outcome.

Five types of question are named: therapeutic, which treatment is the most effective; diagnostic, how to select a test or interpret its results; prognostic, what is the patient's likely course of disease; prevention, how the risk factors can be adjusted to reduce the risk of disease; and harm, whether a treatment has harmful effects and how they can be avoided. For harm and for prevention it is better to use a cohort.