Purpose: Little is known about the fall risk of older adults with overactive bladder, especially in the absence of urgency incontinence. We evaluated the impacts of overactive bladder with and without urgency incontinence (overactive bladder wet and overactive bladder dry) on the fall risk in older adults, and investigated the importance of overactive bladder as a predictor of falls by using tree based models.
Materials and methods: This prospective cohort study included 630 community dwelling, independent older adults 75 years old or older who attended a health checkup in 2017 with a 1-year followup. The associations of overactive bladder dry and overactive bladder wet with a fall history, and future fall risk compared to no overactive bladder were assessed using logistic regression models. The contribution of overactive bladder as a predictor of falls was examined using a random forest and decision tree approach.
Results: Of the 577 analyzed participants (median age 79 years), 273 (47%) were men. The prevalence of overactive bladder dry and overactive bladder wet at baseline was 15% and 14%, respectively. Multivariable logistic regression analysis revealed that both overactive bladder dry and overactive bladder wet were associated with a higher likelihood of prior falls (adjusted ORs vs no overactive bladder 2.03 and 2.21, respectively; 95% CI 1.23-3.37 and 1.29-3.78, respectively). Among the 363 participants without a fall history, the adjusted ORs (95% CIs) of overactive bladder dry and overactive bladder wet for the occurrence of falls during the 1-year followup were 2.74 (1.19-6.29) and 1.35 (0.47-3.87), respectively. The tree based approach used for all participants showed that overactive bladder was an important predictor of falls in adults without a fall history, and the model had 83.6% accuracy and 81.8% AUC.
Conclusions: Overactive bladder, even in the absence of urgency incontinence, is an important predictor of falls in older adults with a low absolute fall risk.
Keywords: accidental falls; decision trees; machine learning; overactive; urinary bladder; urinary incontinence.