Faculty Advisor
Nicole Dawson, PT, PhD, GCS
Publication Date
Spring 2020
Abstract
The majority of falls research has been focused on investigating causes linked to falls in community-dwelling older adults. However, falls research utilizes a myriad of ways to categorize fall group variables, leading to incomparable and incompatible data, causing a loss of information in translation.
PURPOSE: The purpose of the current study was to examine if the categorization of falls, as the dependent variable, impacted the results of risk factors identified.
METHOD: Retrospective data analysis from a study which recorded independent variables linked to falls in 80 community-dwelling older adults (e.g., age, medical comorbidities, physical performance tests, cognitive function). Data were examined using bivariate correlation analysis to assess outcome variance after altering fall categorization of non-faller vs. faller: (0, 1+); (0-1, 2+); (continuous data).
RESULTS: Different correlations emerged depending on the categorization of falls used as the outcome variable. Falls dichotomized into: non-faller or faller (0,1+) significantly correlated with self-reported depression (p=0.04) and postural stability (p=0.003), and into non-faller and recurrent faller (0-1, 2+) which only correlated significantly with postural stability (p=0.004). Falls trichotomized into non, single, and recurrent faller (0, 1, 2+) significantly correlated with self-reported depression (p=0.05) and postural stability (p=0.007). Finally, when analyzed falls as a continuous variable, falls were associated with high-risk prescription medications (p=0.03).
CONCLUSION: Currently, there are no standards for fall categorization, allowing researchers to potentially choose the best fitting results to confirm their hypothesis. Our research demonstrates how using different methods of categorization clouds the pool of fall research, slowing progress to determine which risk factors should be focused on to prevent future falls. Creating a standard method to categorize falls may lead the field to finding a solution to better assist our aging population across multiple disciplines.
Recommended Citation
Carrion, Marta; Jahan-Flynn, Fahima; and Sarto, Andrea, "Is How We Group Data Important? Statistical Differences in Analyzing Independent Variables for Categorizing Fall Group" (2020). UCF DPT Research Capstone. 18.
https://stars.library.ucf.edu/dpt-capstone/18
Access Status
UCF Only