Many studies collect a wealth of information that, although not integral to achieving their primary purpose, proves to be extremely valuable in answering subsequent research questions. Even though extracting and analyzing existing data may be efficient and cost effective, identifying meaningful relationships within the volumes of data often contained in existing datasets can be challenging. Developing reliable and effective statistical models to aid in mining data enhances the ability to utilize the data to answer novel research questions with significant potential to impact troubling conditions such as urinary incontinence (UI). Beaumont, Royal Oak investigator and urologist, Dr. Ananias Diokno, has accumulated a wealth of data over the past two decades from in depth population-based studies of UI. Dr. Diokno's significant productivity in UI research contributed to earning a coveted NIH MERIT award. These data are now contained in two databases. The first contains information gathered during the original Medical, Epidemiologic, and Social Aspects of Aging (MESA) project. Funded by the National Institutes on Aging (NIA), this multi-stage longitudinal study of senior men and women began in 1983 and focused on factors related to the epidemiology of UI (MESA 1 database). The second database (MESA 2 database) contains data collected during a subsequent collaborative study with Dr. Carolyn Sampselle at the University of Michigan.
The aim of this multiphase prospective, randomized controlled "MESA 2" trial was to prevent UI in older women. Continent women 55 years and older were recruited from mailings and clinic visits and were randomized to a behavior modification program (BMP) treatment or control group. Many of the interview questions and other data collected during this prevention trial were the same, or similar, to those collected in the original MESA 1 database study. In this proposed project, Dr. Diokno and his team at Beaumont, Royal Oak, MI will build upon established collaborations with internationally recognized data mining experts Dr. Mohammad Siadat and colleagues at Oakland University in nearby Rochester, Michigan, and Dr. Sampselle at the University of Michigan. Dr. Diokno's domain knowledge and possession of the MESA data provides a unique opportunity to create a robust "Urinary Continence Index" to identify women who are likely to become incontinent over time. Since UI is more prevalent in women than men and the underlying causes differ, this project will focus on developing a UI index for use in women. However, the statistical modeling methods developed might be applied in future projects using male MESA subjects'data or other data to identify predictive indices for UI or perhaps other chronic and costly conditions. We propose a novel application of data mining strategies to identify salient predictors of UI in women and create and test a predictive index. We are in an extraordinary position to further utilize MESA data and take a major step towards developing a predictive index for widespread use. Since UI is a costly condition contributing to social isolation and poor quality of life, establishing a clinically useful index would create significant opportunities for focused prevention and early intervention strategies.
Urinary incontinence (UI) is a costly condition that causes social isolation and poorer quality of life. Data collected from years of productive population-based UI research are now contained in two databases (MESA 1 and 2). In collaboration with data mining experts, we propose to develop a urinary continence index that will help predict UI in older women in order to promote early prevention and treatment strategies.