Characteristics of the patients
Table 1
Demographic characteristics of the study participants.
Characteristic
|
All patients
(n = 131)
|
Non-hemodialysis
(n = 91)
|
Hemodialysis
(n = 40)
|
P-value
|
Male, n (%)
|
93 (71.0%)
|
75 (82.4%)
|
18 (45.0%)
|
0.017
|
Numbers of comorbidities
|
9 (6–10)
|
9 (6–10)
|
8 (6–10)
|
0.901
|
Education level (years)
|
15 (12–16)
|
16 (12–16)
|
15 (12–16)
|
0.183
|
Death, n (%)
|
12 (9.2%)
|
3 (3.3%)
|
9 (22.5%)
|
<0.001
|
Age (years)
|
81 (69–88)
|
80 (67–87.5)
|
84 (73–89)
|
0.200
|
BMI (kg/m²)
|
23.5 (20.9–25.7)
|
24.4 (21.5–26.1)
|
21.0 (18.8–24.2)
|
<0.001
|
SMI (kg/m²)
|
7 (5.7–8.0)
|
7.5 (6.6–8.2)
|
5.6 (5.2–6.3)
|
<0.001
|
BMI/SMI ratio
|
3.33 (3.05–3.87)
|
3.26 (2.98–3.61)
|
3.75 (3.26–4.30)
|
<0.001
|
Cr (µmol/L)
|
104.0 (71.3–181.2)
|
84.1 (66.0–116.5)
|
248.1 (181.2–359.6)
|
<0.001
|
CysC (mg/L)
|
1.69 (1.12–3.42)
|
1.29 (1.05–2.01)
|
5.67 (4.32–7.07)
|
<0.001
|
Cr/CysC ratio
|
63.25 (51.49–73.93)
|
64.30 (54.86–75.43)
|
53.19 (32.2–67.21)
|
0.001
|
WBC (´109/L)
|
6.53 (5.30–8.81)
|
6.23 (5.22–8.42)
|
7.06 (5.67–8.87)
|
0.202
|
ALB (g/L)
|
35.6 (31.6–39.5)
|
36.3 (32.7–40.1)
|
34.1 (30.1–37.2)
|
0.031
|
Hemoglobin (g/L)
|
119 (101–135)
|
125 (112–140)
|
96 (81–110)
|
<0.001
|
TG (mmol/L)
|
1.17 (0.85–1.62)
|
1.12 (0.84–1.56)
|
1.19 (0.96–1.70)
|
0.251
|
TC (mmol/L)
|
3.72 (3.20–4.64)
|
3.74 (3.22–4.78)
|
3.68 (2.76–4.48)
|
0.225
|
CRP (mg/L)
|
5.96 (2.37–21.65)
|
5.00 (2.20–13.20)
|
9.35 (5.67–29.60)
|
0.005
|
NRS-2002 score
|
3 (2–4)
|
3 (1–4)
|
3 (2–5)
|
0.009
|
Modified Thomas score
|
2 (1.25–3.33)
|
2 (1–3)
|
3 (2–4)
|
0.001
|
Barthel index
|
83.7 (59.0–100.0)
|
89.2 (72.0–100.0)
|
63.40 (27.40–92.00)
|
<0.001
|
Data are shown as median (interquartile range) unless otherwise stated. ALB: Albumin; BMI: body-mass index; Cr: serum creatinine; CRP: C-reactive protein; CysC: cystatin C; Hb: hemoglobin; NRS-2002: Nutritional Risk Assessment Scale 2002; SMI: skeletal muscle index; TC: total cholesterol; TG: triglyceride; WBC: white blood cell.
The final analysis included 131 patients (93 males, 71.0%) with a median age of 81 (IQR, 69–88) years. The non-hemodialysis group included 91 patients (75 males, 82.4%) with a median age of 80 (67–87.5) years, and the hemodialysis group included 40 patients (18 males, 45.0%) with a median age of 84 (73–89) years. The demographic and clinical characteristics of the patients are presented in Table 1. There were no significant differences between the two groups in age, education level, number of comorbid diseases, serum TC level, serum TG level or WBC (Table 1). However, the hemodialysis group contained a higher proportion of female patients than the non-hemodialysis group (P < 0.05; Table 1). The hemodialysis group also had a significantly lower BMI and SMI and a significantly higher BMI/SMI ratio than the non-hemodialysis group (all P < 0.05; Table 1), suggesting that the hemodialysis group had lower muscle mass and strength. In agreement with this finding, the hemodialysis group had significantly higher serum levels of Cr and CysC but a significantly lower Cr/CysC ratio than the non-hemodialysis group (all P < 0.05; Table 1). Additionally, when compared with the non-hemodialysis group, the hemodialysis group had significantly lower blood levels of ALB and hemoglobin and a significantly higher serum CRP level (all P < 0.05; Table 1). The hemodialysis group also had a significantly higher NRS-2002 score (indicating a higher risk of malnutrition), a significantly higher modified Thomas fall risk score (indicating a higher risk of falling) and a significantly lower Barthel index score (indicating lower performance in ADL) than the non-hemodialysis group (all P < 0.05; Table 1). Mortality rate was significantly higher in the hemodialysis group than in the non-hemodialysis group (22.5% vs. 3.3%, P < 0.001).
Multivariate linear regression analysis of factors associated with risk of falling
The multivariate linear regression analysis revealed that older age (b = 0.07, 95%CI = 0.06–0.09, P < 0.001) and hemodialysis (b = 0.67, 95%CI = 0.27–1.06, P = 0.001) were associated with a higher modified Thomas fall risk score. Gender, education level and number of comorbidities were not related to fall risk (Table 2).
Table 2
Linear regression analyses of factors associated with fall risk (modified Thomas scale score).
Characteristic
|
b (95% confidence interval)
|
P-value
|
Hemodialysis
|
0.67 (0.27, 1.06)
|
0.001
|
Age
|
0.07 (0.06, 0.09)
|
<0.001
|
Gender (male vs. female)
|
-0.26 (-0.63, 0.12)
|
0.174
|
Education level (years)
|
0.04 (-0.08, 0.16)
|
0.521
|
Number of comorbid diseases
|
0.05 (-0.02, 0.12)
|
0.130
|
Path analysis was used to explore the direct effects of hemodialysis on fall risk (modified Thomas scale score) as well as the indirect effects of hemodialysis on fall risk mediated through the Barthel index and Cr/CysC ratio. Modified Thomas fall risk score was used as the dependent variable, spasticity and strength were used as exogenous independent variables, and the Barthel index and Cr/CysC ratio were used as the endogenous independent (mediating) variables. The proposed path models (a crude model and a model adjusted for age, gender, education level and number of comorbid diseases) are shown in Figure 1.
In the crude model, hemodialysis had no significant direct effect on the modified Thomas fall risk score and no significant indirect effect on the modified Thomas fall risk score via the Cr/CysC ratio. However, hemodialysis exerted a significant indirect effect on the modified Thomas fall risk score via the Barthel index (b = 0.402, 95%CI = 0.149–0.655, P = 0.002). Furthermore, there was a significant overall indirect effect (b = 0.554, 95%CI = 0.257–0.851, P < 0.001) and total effect (b = 0.993, 95%CI = 0.432–1.433, P < 0.001) of hemodialysis on the modified Thomas fall risk score (Table 3).
Table 3
Path analysis of the direct and indirect effects of hemodialysis on fall risk.
Effect
|
Crude model
|
Adjusted model
|
β
|
95%CI
|
P
|
β
|
95%CI
|
P
|
Effect of hemodialysis on Cr/CysC ratio (a1)
|
-12.679
|
-19.033, -6.324
|
<0.001
|
-12.679
|
-19.033, -6.324
|
<0.001
|
Effect of Cr/CysC ratio on Thomas score (b1)
|
-0.012
|
-0.024, 0
|
0.056
|
-0.004
|
-0.014, 0.007
|
0.510
|
Indirect effect of hemodialysis on Thomas score via Cr/CysC ratio (a1*b1)
|
0.152
|
-0.021, 0.325
|
0.086
|
0.045
|
-0.091, 0.181
|
0.516
|
Effect of hemodialysis on Barthel index (a2)
|
-22.006
|
-32.723, -11.29
|
<0.001
|
-22.006
|
-32.723, -11.29
|
<0.001
|
Effect of Barthel index on Thomas score (b2)
|
-0.018
|
-0.026, -0.011
|
<0.001
|
-0.008
|
-0.014, -0.002
|
0.015
|
Indirect effect of hemodialysis on Thomas score via Barthel index (a2*b2)
|
0.402
|
0.149, 0.655
|
0.002
|
0.172
|
0.010, 0.333
|
0.037
|
Indirect effect of hemodialysis on Thomas score via Cr/CysC ratio and Barthel index (a1*b1+a2*b2)
|
0.554
|
0.257, 0.851
|
<0.001
|
0.217
|
0.029, 0.404
|
0.023
|
Direct effect of hemodialysis on Thomas score (c)
|
0.379
|
-0.105, 0.862
|
0.125
|
0.526
|
0.101, 0.952
|
0.015
|
Total effect of hemodialysis on Thomas score (c+a1*b1+a2*b2)
|
0.933
|
0.432, 1.433
|
<0.001
|
0.743
|
0.339, 1.147
|
<0.001
|
The adjusted model was adjusted for age, gender, education level and number of comorbid diseases. Cr/CysC ratio: ratio of serum creatinine level to serum cystatin C level; 95%CI: 95% confidence interval.
In the adjusted model, hemodialysis exerted a significant direct effect on the modified Thomas fall risk score (b = 0.526, 95%CI = 0.432–1.433, P = 0.015) as well as a significant indirect effect via the Barthel index (b = 0.172, 95%CI = 0.010–0.333, P = 0.037), but there was no significant indirect effect via the Cr/CysC ratio. The overall indirect effect (b = 0.217, 95%CI = 0.029–0.404, P = 0.023) and total effect (b = 0.743, 95%CI = 0.339–1.147, P < 0.001) of hemodialysis on the modified Thomas fall risk score were significant (Table 3).