Fig. 3.

Types of bias met in non-randomized studies (left column) based on the ROBINS-I tool and randomized studies (right column) based on the RoB 2.0 tool, with the situations or actions that, when properly performed (center column), protects against these biases in RCTs and would prevent bias in NRSI if we were able to do a random assignment of participants; this is the hallmark feature of the ‘target trial’. To the right, in parentheses, are depicted other nomenclatures for biases, which are based on the classic (previous) risk of bias tool from Cochrane.