Abnormal vibration perception threshold alters the gait features in type 2 diabetes mellitus patients

Abstract

Objective: It is generally believed that gait characteristics of diabetic neuropathic patients differ from those of non-diabetic ones. However, it is still unclear how the abnormal foot sensation influences the gait during walking in type 2 diabetes mellitus (T2DM). For the purpose of gaining a better insight into the alterations of detailed gait parameters and figuring out important aspects in the gait indexes by peripheral neuropathy in elder T2DM patients, we compared the gait features in participants with normal glucose tolerance (NGT) controls and diabetic individuals complicated by peripheral neuropathy or not.

Subjects and methods: Gait parameters were observed during the 10-m walk on flat land among different conditions of diabetes in 1,741 participants from three clinical centers. Subjects were divided into four groups: persons with NGT were taken as the control group; patients with T2DM included three subgroups: DM control (no chronic complications), DM-DPN (DM complicated by only peripheral neuropathy), and DM-DPN+LEAD (DM complicated by both neuropathy and artery disease). The clinical characteristics and gait parameters were assessed and compared among these four groups. Analyses of variance were employed to verify possible differences of gait parameters between groups and conditions. Stepwise multivariate regression analysis was performed to reveal possible predictors of gait deficits. Receiver operating characteristic (ROC) curve analysis was employed to find any discriminatory power of diabetic peripheral neuropathy (DPN) for the step time.

Results: In participants burdened with DPN, whether complicated by lower extremity arterial disease (LEAD) or not, step time increased sharply (p < 0.05). Stepwise multivariate regression models showed that independent variables of gait abnormality were sex, age, leg length, vibration perception threshold (VPT), and ankle-brachial index (ABI) (p < 0.01). Meanwhile, VPT was a significant independent predictor of step time, spatiotemporal variability (SD_A), and temporal variability (SD_B) (p < 0.05). ROC curve analysis was explored to find the discriminatory power of DPN for the occurrence of increased step time. The area under the curve (AUC) value was 0.608 (95% CI: 0.562-0.654, p < 0.01), and the cutoff point was 538.41 ms accompanied by a higher VPT. A significant positive association was observed between increased step time and the highest VPT group [odds ratio (OR) = 1.83, 95% CI: 1.32-2.55, p< 0.01]. In female patients, this OR value elevated to 2.16 (95% CI: 1.25-3.73, p< 0.01).

Conclusions: In addition to sex, age, and leg length, VPT was a distinct factor that associated with altered gait parameters. DPN is associated with increased step time, and the step time increases with worsening VPT in type 2 diabetes.

Keywords: complication; diabetic; diabetic peripheral neuropathy; gait; type 2 diabetes; vibrating perception threshold.

Figure 1

The comparison of gait spatiotemporal variables among different groups. (A–C) Duty-factor_double stance-CoV was different between the non-diabetic group and the diabetic subgroups, and the step time was different between the non-diabetic group and the complication groups. Variables were cadence, stride length, walking speed, walk ratio, duty-factor_double stance, step time, and CoVs (CoV = SD/mean x 100). The difference of all gait variables between groups was significant (p < 0.05). Subjects with abnormal blood glucose control showed a considerably higher duty-factor_double stance-CoV (52.50% vs. 46.30%, p < 0.05). In participants burdened with DPN, whether complicated by LEAD or not, step time increased sharply (p < 0.05).

Figure 2

The comparison of gait phase plot variables among the different groups. Variables were SD_A, SD_B, A ratio, and Δangleβ. The difference of all gait variables between groups was significant (p < 0.05). Trend analysis. The results displayed lower SD_A (1.32 vs. 1.57, p < 0.01) and SD_B (0.38 vs. 0.51, p < 0.01) in subjects with both DPN and LEAD. Moreover, trend analysis revealed gradually decreased SD_A and SD_B in the four DM groups (p < 0.05).

Figure 3

The gender and age difference of gait variables among type 2 diabetics. Panels (A, B) showed the comparison of spatiotemporal variables between different age and sex groups, and panel (C) showed the comparison of phase plot variables. A total of 1,420 individuals were selected and divided into four groups according to their sex and age and showed an obvious difference between men and women, young and elder patients. * p < 0.05, comparison between men of different age ranges; # p < 0.05, comparison between women of different age ranges; p < 0.05, comparison between different sex groups at the same age range. In the age range ≤65 years and ≥65 years, male participants were found to have fewer steps/min than female participants, but step length was significantly increased, walking faster, spent more time per walk, and showed greater SD_A and SD_B. A larger part of distinction was observed only in female subjects, such as less steps/min, shorter stride length, more time spent per walk, and smaller SD_A, SD_B, and A ratio. Men<65 means men aged <65 years (56.1 ± 5.2 years), n = 372; Men≥65 means men aged ≥65 years (69.7 ± 3.2 years), n = 190; Women<65 means women aged <65 years (56.5 ± 5.8 years), n = 571; Women≥65 means women aged ≥65 years (69.7 ± 3.3 years), n = 287.

Figure 4

ROC curve analysis. ROC curve analysis was explored to find any discriminatory power of DPN for step time. The AUC value was 0.608 (95% CI: 0.562–0.654, p < 0.01). Cutoff point was 538.41 ms. The Youden index was 0.184; its sensitivity was 58.20% and the specificity was 39.81%.