Lucy’s MSc in Statistics thesis analyzed time to ART initiation, exploring the effect of factors that predict how soon a patient initiates ART and the causal effect of ART initiation on the survival of patients. She used observational data on patients co-infected with HIV/TB from three clinics in the Western Cape. Using observational data presents a number of statistical challenges: measurements on patients are not always recorded within time intervals of interest, nor are they always recorded regularly; treatment of patients is not assigned at random, so that patients who have a worse prognosis a-priori will initiate ART sooner than those who have a better prognosis, which could bias the estimated effect of ART initiation on survival. Lucy therefore used methods to reduce potential biases, as well as to explore the sensitivity of assumptions that she makes where possible.