OBJECTIVES: To evaluate how variation in duration of follow-up affects the predicted probability of initiating multiple myeloma (MM) lines of therapy (LOTs) when using incidence density ratios (IDRs). METHODS: A U.S. administrative claims database was used to identify adult patients with ≥2 MM diagnoses who received ≥1 MM treatment ≤90 days of their MM diagnosis (first MM treatment = index) with insurance eligibility 24-months pre-and ≥3-months post-index. Patients were excluded if they had MM treatment pre-index, non-MM chemotherapy, pregnancy-related codes or an HIV diagnosis. LOTs were defined as each occurrence of: stem cell transplant, a ≥60-day gap in treatment, retreatment, or change in MM therapy. Generalized linear Poisson models were used to estimate IDRs (per person-year of observation), predicting the percentage of patients with 2, 3 and 4 LOTs at 6-month observation intervals for 5 years. RESULTS: A total of 4,987 MM patients were included in this analysis. At six months of observation, the estimated percentage of patients (i.e. the IDRs) with 2, 3 and 4 LOTs per person-year of observation was 40.4%, 4.6% and 0.5%, respectively. At five years of observation, the IDRs for patients with 2, 3 and 4 LOTs per person-year of observation were 19.9%, 9.4% and 4.5%, respectively. If only 6-months of observation is used to calculate IDRs, the probability of having a second LOT is 203% (40.4%/19.9%) greater than calculation using 5-years of observation. Stability of estimates did not occur until approximately 3.5 years after initiation of LOT 1. CONCLUSIONS: Economic models using treatment probabilities based on IDRs should be evaluated to ensure the duration of time over which the values are calculated is consistent with the characteristics of the disease and population being modeled. Failure to utilize an appropriate length of follow-up may result in significant estimation error.