Background
Physical activity is a crucial component of the treatment and management of diabetes, yet older individuals with diabetes generally exhibit insufficient physical activity levels. Life space mobility (LSM), which reflects the physical activity and psychological well-being of older individuals within their actual environment, poses challenges to the management of diabetes in this age group, significantly jeopardizing their physical and mental health.
Aim
The objective of this study is twofold: first, to investigate the risk factors associated with mobility restriction among older adults with diabetes mellitus; and second, to develop and verify a restricted life-space risk nomogram prediction model for this population by nomogram.
Methods
The convenient sampling method was employed to recruit participators aged 60 years and above from both the endocrinology clinic and health examination center situated in a Grade A hospital located in Anhui. These participators were then categorized into two groups: the Restricted Life-Space(RLS) group and the non-Restricted Life-Space group. The Least absolute shrinkage and selection operator (LASSO) regression was used to decrease data dimensionality and select features, while the C index and Brier score were employed to assess the model's discrimination and calibration, respectively. Receiver operating characteristic (ROC) curves and calibration curve were generated to visualize the performance of the model.Decision curve analysis (DCA) and clinical impact curve (CIC) were conducted to evaluate the clinical value of the model. The internal validity of the model was confirmed using the bootstrapping method, while external validation was performed to test its generalizability.
Results
A risk nomogram prediction model was developed using four predictors: fear of falling, decline of activities of daily living (ADL), lower limb hypofunction, and decline of vision.Internal validation of the model yielded the following results: a C index of 0.936, a goodness-of-fit test χ2 value of 1.21 with a corresponding P-value of 0.75, and a Brier score of 0.081. In the DCA, the threshold ranged from 0.1 to 0.95, and the clinical net benefit was consistently greater than 0. The CIC demonstrated that the predicted risk generated via the model was highly consistent with the actual risk. External validation of the model resulted in a C index of 0.932, a goodness-of-fit test χ2 value of 3.11 with a corresponding P-value of 0.60, and a Brier score of 0.106.
Conclusion
The risk nomogram prediction model constructed in this study based on the above four independent risk factors is noninvasive, inexpensive, and easily accessible, with high sensitivity, specificity, and utility, and provides a reference for the assessment and intervention of life-space mobility levels in older adults with diabetes.