Clinicopathological characteristics
In the primary cohort, there were initially 228 patients, excluding those who lost contact, died and had missing data All patients were evaluated by MNA-SF, NRS-2002 and GLIM criteria, people who are defined as malnutrition are included in the follow-up study. A total of 161 patients who met the inclusion criteria were finally enrolled in the study (Fig. 1). There were 107 women (66%) and 54 men (34%), with an average age of 77.90 ± 8.17 years. The demographic features and clinical characteristics of the primary and validation cohorts are presented in Table 1.
GLIM-defined malnutrition
Among the 161 older patients with hip fracture, according to the GLIM criteria, there were 67 patients (41.6%) with malnutrition, including 43 (26.7%) with moderate malnutrition and 24 (14.9%) with severe malnutrition (Table 1).
Table 1 Clinical characteristics of hip fracture patients
GLIM criteria
|
Normal
|
Moderate
|
Severe
|
Overall
|
n
|
94
|
43
|
24
|
161
|
Age
|
75.71 ±7.40
|
79.26 ± 8.21
|
84.04 ± 7.62
|
77.90 ± 8.17
|
Gender
|
|
|
|
|
Male
|
37 (39.4%)
|
14 (32.6%)
|
3 (12.6%)
|
54 (33.5%)
|
Female
|
75 (60.6%)
|
29 (67.4%)
|
21 (87.4%)
|
107 (66.5%)
|
Height
|
1.60 ± 0.08
|
1.61 ± 0.08
|
1.55 ± 0.08
|
1.60 ± 0.08
|
Weight
|
62.59 ± 9.23
|
52.45± 5.76
|
43.08 ± 5.54
|
56.98 ± 10.75
|
BMI
|
24.26 ± 2.70
|
20.22 ± 1.88
|
17.80 ± 1.38
|
22.22 ± 3.45
|
Fracture type
|
|
|
|
|
ITF
|
39 (41.5%)
|
13 (30.2%)
|
10 (41.7%)
|
62 (38.5%)
|
FNF
|
55 (58.5%)
|
30 (69.8%)
|
14 (58.3%)
|
91 (61.5%)
|
Surgery type
|
|
|
|
|
Replacement
|
55(58.5%)
|
29 (67.4%)
|
11 (45.8%)
|
95
|
Fixation
|
39 (41.5%)
|
14 (32.6%)
|
13 (54.2%)
|
66
|
NRS-2002 Risk (%)
|
17 (18.1%)
|
43 (100.0%)
|
24 (100.0%)
|
84 (54.7%)
|
MNA-SF Risk (%)
|
48 (51.1%)
|
41 (100.0%)
|
24 (100.0%)
|
113 (70.7%)
|
Abbreviations: BMI, body mass index; ITF, Intertrochanteric fracture; FNF, femoral neck fracture; NRS-2002, nutrition risk screening 2002; MNA-SF, mini nutrition Assessment
Comparison of functional recovery of patients with GLIM, MNA-SF and NRS-2002 criteria
According to HHS, the patients were divided into better recovery of postoperative motor function (HHS>75) and worse recovery of postoperative motor function (HHS≤75). According to the univariable model, the postoperative motor function of patients was significantly associated with age, GLIM criteria, MNA-SF and NRS-2002 (P<0.05) (Table 2). After controlling for gender and age, multivariable logistic regression models were performed on the three nutritional screening tools. The results showed that the three nutritional screening methods were significant (P<0.05) and could effectively predict the recovery of postoperative motor function (Table 2).
Table 2 Univariable and multivariable logistic regression models
|
Univariable Model
|
Multivariable Models
|
|
|
NRS-2002
|
MNA-SF
|
GLIM
|
Gender (Male)
|
2.13
[0.096, 5.12]
|
|
|
|
Age
|
0.93
[0.89, 0.97] *
|
0.96
[0.91, 1.01]
|
0.95
[0.90, 0.99] *
|
0.98
[0.93, 1.04]
|
Fracture type
|
|
|
|
|
FNF
|
Reference
|
|
|
|
ITF
|
0.85
[0.41, 1.76]
|
|
|
|
Surgery type
|
|
|
|
|
Replacement
|
Reference
|
|
|
|
Fixation
|
0.85
[0.42, 1.76]
|
|
|
|
NRS-2002
|
0.05
[0.01, 0.15] *
|
0.06
[0.01, 0.17] **
|
|
|
MNA-SF
|
0.04
[0.00, 0.19] *
|
|
0.05
[0.00, 0.23] *
|
|
GLIM
|
|
|
|
|
Normal
|
Reference
|
|
|
|
Moderate
|
0.03
[0.01, 0.10] **
|
|
|
0.03
[0.01, 0.11] **
|
Severe
|
0.02
[0.00, 0.06] **
|
|
|
0.02
[0.00, 0.07] **
|
Abbreviations: ITF, Intertrochanteric fracture; FNF, femoral neck fracture; NRS-2002, nutrition risk screening 2002; MNA-SF, mini nutrition Assessment; GLIM, Global Leadership Initiative on Malnutrition. *: p<0.05; **: p<0.01;
AUC of the ROC curve of the GLIM model was 86.2, which was larger than that of the NRS-2002 (AUC=81.2) and MNA-SF (AUC=76.3) (Figure 2a). We found that the GLIM criteria were better able to predict prognosis of older hip fracture patients. According to the decision curve analysis, the three nutritional screening models all showed predictive effect on postoperative recovery (Figure 2b). However, GLIM model has higher net benefit than that of the other two models. The calibration curve was constructed by bootstrap method (Figure 2c), in which the apparent and biased-corrected curve were both close to the ideal line. The Hosmer–Lemeshow test also showed acceptable fitness of the model (χ2 = 5.0952, P=0.7474).
Validation of the predictive accuracy of the nomogram
The GLIM model was selected as the best to predict functional recovery, and the scores of the predictors were coded and shown in the nomogram in Figure 3a. As shown in the Forest plot (Figure 3b), GLIM criteria defined malnutrition severity revealed affecting postoperative functional recovery significantly, in which OR of moderate level and severe level were 0.03(0.01,0.11) and 0.02(0.00,0.07) respectively.