Fig. 1From: Statistical methods for elimination of guarantee-time bias in cohort studies: a simulation studyGuarantee-Time Bias Considering four people, time to initiation of drug use w i and a time to event t i were randomly generated for each person. The person whose value of T 0 is smaller than W is allocated to the non-user group (person #1). Otherwise, individuals are allocated to the drug-user group, according to cumulative drug exposure (persons #2 to 4)Back to article page