Primary Submission Category: Weighting
Physical Function Decline in Aging Cancer Survivors and Cancer-Free Controls: Accounting for Bias Due to Selective Attrition
Authors: Sophia Fuller, Sowmya Vasan, Hailey Banack, Alexandra Binder, Elizabeth Feliciano,
Presenting Author: Sophia Fuller*
With improvements in screening and treatment, the number of cancer survivors is growing, increasing the need for research into long-term health and well-being after diagnosis. Investigations of how cancer and its treatment influence trajectories of aging is complicated by differential loss to follow-up and death between survivors and cancer-free individuals. Those with more severe disease, and correspondingly, more cytotoxic treatment, are also more likely to die or drop out. The confluence of cancer, stage, and treatment on censoring obscure our understanding of when and which survivors are at risk for accelerated aging. To demonstrate the value of accounting for censoring, we use longitudinal data from the Women’s Health Initiative to compare trajectories of physical function between women with cancer to non-cancer controls. We estimate inverse probability of censoring weights due to loss to follow up and, separately, death, using ensemble machine learning for each wave of the study. Then we fit GEE models incorporating these weights to capture the yearly decline in physical function among cancer survivors, subset by cancer type and stage at diagnosis, and their non-cancer controls. We hypothesize that weighting to account for attrition will yield estimates of cancer survivors’ yearly rates of decline that are larger in magnitude, and this decline will be greatest for women with more aggressive cancer types diagnosed at more advanced stages.