A retrospective analysis to examine factors associated with mortality in Medicare beneficiaries newly diagnosed with multiple myeloma.

Objective: Population-based data on mortality and associated factors in patients with multiple myeloma (MM) are limited. We examined the association between all-cause mortality and demographic and clinical characteristics in newly diagnosed MM patients treated with guideline-recommended chemotherapeutic agents.
Research Design and Methods: This retrospective cohort analysis used Medicare 20% data to create a cohort of adult (aged ≥18 years) newly diagnosed MM patients who received chemotherapy 2008-2011 and had no MM diagnosis in the 12 months before the disease index date. Patients were followed from treatment initiation through the earliest of death, loss of insurance coverage, or study end (December 2011). Modified Charlson Comorbidity Index scores and MM-related comorbid conditions (anemia, hypercalcemia, skeletal-related events) were identified in the 6 month pre-index-date period. All-cause mortality and associated factors were examined using multivariable Cox proportional hazard models.
Results: We identified 2419 newly diagnosed patients who received MM therapy during follow-up. Mean (SD) and median follow-up were 1.51 (1.0) and 1.37 years. Of the cohort, 55% were female, 78% white, and 92% aged ≥65 years. Pre-index, 54%, 9%, and 5% were diagnosed with anemia, hypercalcemia, and skeletal-related events. Overall, 942 (39%) patients died during follow-up. Factors associated with increased risk of death were older age (≥65 vs. 18-64 years; hazard ratio 1.49, 95% confidence interval 1.13-1.99), higher comorbidity score (≥4 vs. 0; 1.78, 1.43-2.21), anemia (1.23, 1.06-1.42), and hypercalcemia (1.45, 1.19-1.76); female sex (0.86, 0.75-0.98) was associated with decreased risk.
Conclusions: Older age, male sex, high comorbidity burden, anemia, and hypercalcemia were risk factors for death in newly diagnosed Medicare MM patients. Study limitations included non-causal observational design, non-validated MM algorithm, potential treatment misclassification, and non-availability of prognostic factors including disease staging information, biomarkers, and other laboratory variables. Additional analyses are warranted to understand the relationship between treatments and death.