Health-Related Physical Fitness as a Risk Factor for Falls in Elderly People Living in the Community: A Prospective Study in China

Abstract

Objectives: Health-related physical-fitness (HRPF) involves multi-components of physical functional tests and is reported to be associated with the risk of fall. The study sought to determine whether specific physical fitness components were stronger predictors of falls among elderly people.

Methods: This prospective cohort study involved 299 community residents age ≥60 years from Shanghai, China. The baseline data included comprehensive assessment of sociodemographic, clinical, and HRPF test. Subjects were followed for 1 year and were contacted by telephone to report falls. LASSO regression and Multivariate regression analysis were used to identify risk predictors of fall. In addition, we used receiver operating characteristic (ROC) curve analyses to determine whether the predictors have diagnostic.

Results: During the 1-year prospective fall assessment, 11.7% of these subjects experienced one or frequent falls. LASSO models revealed that age (=0.01) and 8-ft up-and-go test score (=0.06) were positively associated with falls, while activity-specific balance confidence (ABC; = -0.007) and 2-min step test score (= -0.005) were inversely related. The Area Under roc Curve (AUC) for a linear combination of age, ABC scale score, 2-min step test and 8-ft up-and-go test was 0.778 (95% confidence interval: 0. 700-0.857), which was superior to any of the variables taken alone.

Conclusion: Age, activity-specific balance confidence and fitness abnormalities were determined to contribute to the incident of falls. The value of 2-min step test score, and 8-ft up-and-go test score were the key HRPF components in predicting falls among elderly people.

Keywords: LASSO; community; elderly; falls; health-related physical fitness; prospective study.

Figure 1

Flowchart of the study.

Figure 2

Texture feature selection using the least absolute shrinkage and selection operator (LASSO) binary logistic regression model. (A) Tuning parameter (λ) selection in the LASSO model used 10-fold cross-validation via minimum criteria. Dotted vertical lines were drawn at the optimal values by using the minimum criteria and the 1 standard error of the minimum criteria (the 1-SE criteria). A λ value of 0.0111, with log (λ),−4.4982 was chosen (1-SE criteria) according to 10-fold cross-validation. (B) LASSO coefficient profiles of the 12 texture features. A coefficient profile plot was produced against the log (λ) sequence. Vertical line was drawn at the value selected using 10-fold cross-validation, where optimal one resulted in 4 nonzero coefficients.

Figure 3

Colored lines represent receiver operating characteristic (ROC) curves of each variable for the prediction of fall events. 8-ft up-and-go test (AUC = 0.728, 95% confidence interval (CI): 0.636–0.820), Age (AUC = 0.718, 95% CI: 0.628–0.808), ABC (AUC = 0.689, 95% CI: 0.595–0.784), 2-min step test (AUC = 0.632, 95% CI: 0.521–0.744). Black line represents ROC curves of LASSO regression with the above four variables for the prediction of fall events. AUC = 0.778, 95% CI: 0.700–0.857.