Statistics Corner. August 2023
Statistics Corner August
RCT (Randomized Control Trial)- A prospective experimental design where eligible populations for a study are divided into study groups at random. One of the most effective ways to establish a causal relationship and a gold standard for experimental design developed to limit confounding variables.
Intention to treat- A way of analyzing data in a prospective RCT in which after randomization each subject is kept in their respective randomized group regardless of if they adhered to the guidelines or were not given proper study protocol interventions.
Mitigates the healthy adherer bias- a sociological observation that study participants who adhere to prescribed protocols even if they are placebo arms tend to do better than non-adherers likely a surrogate for more disciplined and healthy choices in general.
Also mitigates other sources of bias that come from excluding study participants either intentional or not.
Makes null hypothesis more likely.
Superiority trials use a predetermined factor of clinical significance as a statistical goal to achieve in order to show a positive result. A factor of clinical significance is usually determined by risk to life. For example clinical significance would be very large for dermatologic condition and smaller for something like death by suicide.
Shows that given treatment is comparable to a standard therapy. Often the treatment seeking non-inferiority has other advantages. For example it could be less expensive or less days of antibiotics. It can be easier to achieve non-inferiority as compared to well designed superiority trials.
Tobias Krussig, MD
Emergency Medicine Resident, PGY2