Predictors of falls in older survivors of breast and prostate cancer: A retrospective cohort study of surveillance, epidemiology and end results-Medicare health outcomes survey linkage.

Objectives: To identify predictors of falls in older breast and prostate cancer survivors.
Methods: This retrospective cohort study analyzed population-based Surveillance, Epidemiology and End Results-Medicare Health Outcomes Survey (SEER-MHOS) linkage. Inclusion criteria were age >65 years at cancer diagnosis, first primary female breast or prostate cancer, cancer staging information available, completion of baseline MHOS during years 2-3 and follow-up MHOS during years 4-5 post-diagnosis, and falls information available. Data from 437 breast and 660 prostate cancer survivors were analyzed. Multivariable logistic regression was constructed to evaluate variables from baseline MHOS with relation to falls from follow-up MHOS. Model accuracy was assessed using area under receiver-operating-characteristic curve (AUC).
Results: At follow-up MHOS, 26% of breast and 22% of prostate cancer survivors reported falls in the past 12 months. In breast cancer, a history of falls (odds ratio (OR) = 4.95, 95% confidence interval (CI) = 2.44-10.04) and sensory impairment in feet (OR = 3.33, 95%CI = 1.51-7.32) were significant predictors of falls. In prostate cancer, a history of falls (OR = 3.04, 95%CI = 1.79-5.15), unmarried (OR = 1.82, 95%CI = 1.12-2.95), lower physical summary score of quality-of-life(OR = 0.96, 95%CI = 0.94-0.98), urinary incontinence (OR = 1.69, 95%CI = 1.08-2.65), older age at diagnosis (OR = 1.05, 95%CI = 1.01-1.09), and shorter time post-diagnosis (OR = 0.96, 95%CI = 0.93-0.99) were significant predictors of falls. AUC was 0.67 and 0.77 for breast and prostate cancer, respectively, indicating moderate accuracy of models in detecting fallers.
Conclusions: Asking older breast and prostate cancer survivors about falls in the past 12 months is imperative in fall prevention. Further examination of deficits specific to each cancer is necessary to assess fall risks.
(Copyright © 2018 Elsevier Ltd. All rights reserved.)