Perceived self-efficacy and body mass index to identify older adults with and without fall history.

Objective: The study aimed to analyze the relationship between a combined set of variables of several domains to identify older adults with and without a fall history. It was also aimed to identify the contribution of the variables included in the model.

Methods: This cross-sectional study was conducted with 191 older adults who performed the Timed-up-and-go Test (TUG), an imaginary TUG (iTUG), and were assessed for physical characteristics. The difference between the actual and imagined performances was calculated to represent perceived self-efficacy (TUG_diff). A decision tree model was applied to indicate the relevant variables to identify fallers and non-fallers, and a logistic regression was also used to determine the contribution of such variables.

Results: The decision tree identified body mass index (BMI) and TUG_diff as related to fall risk, with an accuracy of 71.4 %. Individuals with a BMI < 26.4 kg.m^-2 were identified as non-fallers, while those with a BMI > 26.4 kg.m^-2 and a TUG_diff lower than 3.89 s were also identified as non-fallers. The sensitivity of 0.91 and specificity of 0.26 showed good model accuracy in identifying non-fallers. The regression included BMI and TUG_diff as the main explanatory factors, while age, sex, cognition, TUG, and iTUG were not included as covariates because they did not reach the relative importance threshold of 15% in the decision tree analysis.

Conclusion: Older adults with decreased BMI were less prone to experience a fall. Older adults with increased BMI may experience a protective benefit of more accurate self-efficacy and become less prone to falls.