Thorp, John M., Jr. (2018). BJOG Editor's Choice: Observations often Trump Experiments. BJOG: An International Journal of Obstetrics & Gynaecology, 125(6)
When clinicians in practice open a journal or click on a link to a BJOG article they do so with the hope that they will find information to help them care for their patients. If steeped in evidence-based medicine (EBM) which is the touchstone of epidemiologic instruction in most schools of medicine and nursing, they are biased to believe that randomised clinical trials (RCT) will shed the most light on any topic and deserve the highest authority being ‘Grade A’ or ‘Level One’ evidence on the EBM totem. That approach neglects important facts about RCTs and EBM that are often lost on us who practice clinical medicine. First, it is not ethically or practically possible to randomly assign an exposure that really matters to a research subject. She might accept random assignment of a pill knowing it may be a placebo but won't tolerate random assignment of her dinner menu or an after dinner libation, to name just a few exposures. Second, investigators lack the knowledge to create most exposures and if we did it would be dangerous to do so. Think of placental abruption as one example. Third, properly powered trials looking at the rare world of poor perinatal outcomes are large and expensive. Finally, subjects who choose to enroll in trials differ substantively from those who decline, limiting generalisability.
BJOG: An International Journal of Obstetrics & Gynaecology
Thorp, John M., Jr.