Search for a command to run...
Background Diabetes-related gait disorders are important drivers of falls and functional decline in older adults. Gait variability, as an indicator highly sensitive to fluctuations between steps, remains underexplored in diabetic populations. Compared with the average gait parameters, gait variability may better reflect impaired neuromuscular control and the risk of falling. This study aimed to evaluate the diagnostic accuracy of gait variability parameters, and test whether cognitive function plays an intermediary role between type 2 diabetes mellitus (T2DM) and gait variability. Methods A total of 56 non-diabetic older participants and 37 T2DM patients (aged 60 years or older) were enrolled in this study. We used wearable inertial sensors to evaluate the gait parameters (including the variability in stance time, gait speed, stride length and turn duration) during straight walking and turning tasks. Standardized tools were used to evaluate cognitive functions, including the Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) and other verified measures. Results Our results showed that T2DM patients exhibited significantly higher gait variability across all indicators. All gait variability indices were significantly negatively correlated with cognitive function (r = −0.20 to −0.53, P < 0.05). After adjusting for demographic characteristics and cognitive functions, the T2DM status was still an independent predictor of gait variability. The receiver operating characteristic (ROC) curve analysis showed that the stance time variability had good diagnostic accuracy (area under the curve [AUC] = 0.813, 95% CI 0.727–0.898, p < 0.001, sensitivity 94.6%), and gait speed variability also demonstrated good diagnostic performance (AUC = 0.801, 95% CI 0.705–0.897, p < 0.001). Mediation analysis showed that cognitive function mediated the effect of T2DM on stance time variability, and the mediated effect accounted for 31.9% of the total effect. Conclusion This study showed that T2DM patients demonstrated a significant increase in gait variability. This variation was closely associated with cognitive decline. Stance time and gait speed variability could be used as a sensitive and non-invasive screening method to identify gait dysfunction related to diabetes. T2DM may affect gait stability through dual pathways, involving both cognitive decline and non-cognitive mechanisms. Comprehensive intervention strategies (including blood sugar control, neuropathy management and cognitive training) could improve the gait stability of T2DM elderly people and mitigate the risk of falling.