Patient demographics
The patients included 35 men and 25 women. The mean age at surgery was 66.5 (standard deviation [SD], 14.5) years, and the mean BMI was 22.7 (SD, 4.8) kg/m2. The mean duration of hospitalization was 69 (SD, 62) days (Tables 1 and 2). Inflammatory values such as CRP, white blood cell count, and JOA score were significantly better at discharge than at admission (Table 3). No patient in this study group required rehospitalization for recurrent pyogenic spondylitis.
Table 1
Baseline characteristics of demographic data.
Characteristics
|
|
Age, years
|
66.5 ± 14.5
|
Sex male/female
|
35 (58) / 25 (42)
|
BMI, kg/m2
|
22.7 ± 4.8
|
Location
|
Cervical 4 (7) / Thoracic 21 (35) / Lumbar 35 (58)
|
Duration from the date of onset to the first visit to our hospital, day
|
41 ± 53
|
Duration from the first visit to operation, day
|
25 ± 43
|
Hospitalization, day
|
69 ± 62
|
Creatinine, mg/dL
|
1.7 ± 2.3
|
eGFR, mL/min/1.73m2
|
80 ± 61
|
BUN mg/dL
|
26 ± 17
|
Albumin, g/dL
|
3.1 ± 0.7
|
Total cholesterol, mg/dL
|
161 ± 42
|
Hemoglobin, g/dL
|
11.3 ± 2.3
|
Platelet, 103/µL
|
29.3 ± 13.4
|
White blood cell, 103/µL
|
9.2 ± 5.1
|
Lymphocyte %
|
18 ± 11
|
Neutrophil %
|
73 ± 12
|
ESR, mm/1h
|
76.6 ± 35.2
|
CRP, mg/dL
|
9.6 ± 10.1
|
CONUT
|
4.9 ± 3.4
|
mGPS
|
1.5 ± 0.6
|
CAR
|
3.4 ± 4.2
|
NLR
|
8.0 ± 11.1
|
PNI
|
38.1 ± 7.9
|
PLR
|
308 ± 291
|
Dialysis
|
6 (10)
|
Diabetes mellitus
|
23 (38)
|
Smoker
|
12 (20)
|
Use of steroids
|
2 (3)
|
Use of synthetic or biological disease-modifying anti-rheumatic drugs
|
0 (0)
|
Use of anticancer drug
|
1 (2)
|
Use of antibiotics at admission
|
14 (23)
|
Kulowski classification acute/subacute/chronic
|
7 (12) / 27 (45) /26 (43)
|
JOA score %
|
44.2 (25.7)
|
Data are presented as mean ± standard deviation or n (%). BMI, body mass index; BUN, Blood urea nitrogen; CAR, CRP-albumin ratio; CONUT, controlling nutritional status; CRP, C-reactive protein; eGFR, estimate glomerular filtration rate; ESR, erythrocyte sedimentation rate; JOA, Japanese Orthopaedic Association; mGPS, modified Glasgow prognostic score; NLR, neutrophil–lymphocyte ratio; PLR, platelet–lymphocyte ratio; PNI, prognostic nutrition index.
Regarding the types of surgery, the posterior approach included 3 cases of decompression, 2 of fixation, and 11 of decompression and fixation. The posterior and anterior approach included 10 cases of posterior decompression and fixation, and anterior curettage and fixation by bone graft, 11 of posterior fixation by percutaneous pedicle screw, and anterior curettage and fixation by bone graft (1 case bone cement), and 1 case of posterior decompression and anterior fixation by bone graft.
Table 2
Baseline radiographic characteristics and operative data.
Characteristics
|
|
Griffith early/destructive/sclerotic
|
27 (45)/32 (53)/1 (2)
|
MRI stage Ⅰ/Ⅱ/Ⅲ/Ⅳ/Ⅴ
|
1 (2)/6 (10)/32 (55)/14 (24)/5 (9)
|
No. of infected vertebral bodies
|
2.1 ± 0.9
|
Epidural abscess
|
19 (32)
|
Psoas abscess
|
15 (25)
|
Paraspinal abscess
|
5 (8)
|
Operative time, min
|
243 ± 221
|
Blood loss, mL
|
392 ± 791
|
Number of operations
|
1.6 ± 0.8
|
Biopsy positive
|
17 (31)
|
Data are presented as the mean ± standard deviation or n (%); MRI, magnetic resonance imaging. |
Table 3
Differences in blood draw results between the time of admission and discharge
Characteristic
|
Admission
|
Discharge
|
p-value
|
Albumin, g/dL
|
3.1(0.7)
|
3.1(0.5)
|
0.055
|
White blood cell, 103/µL
|
9.2(5.1)
|
5.5(1.8)
|
≺0.001*
|
Lymphocyte count, %
|
18(11)
|
26(10)
|
≺0.001*
|
Neutrophil count, %
|
73(12)
|
59(14)
|
≺0.001*
|
CRP, mg/dL
|
9.6(10.1)
|
0.8(1.4)
|
≺0.001*
|
JOA score %
|
44.2(25.7)
|
60.7(25.4)
|
≺0.001*
|
*p < 0.05. CRP, C-reactive protein; JOA, Japanese Orthopaedic Association |
Identification of factors associated with the duration required to control infection
The duration required to control an infection was found to be significantly associated with BMI (p = 0.004), BUN (p = 0.015), CRP (p < 0.001), mGPS (p = 0.015), CAR (p < 0.001), NLR (p = 0.023), Kulowski classification (p = 0.049), MRI stage (p = 0.029), the number of infected vertebral bodies (p = 0.030), and psoas abscess (p = 0.032) in the univariate Cox regression analysis (Table 4).
Table 4
Univariate Cox regression analysis of demographic and radiographic data.
Characteristics
|
HR
|
95% CI
|
p
|
Sex male/female
|
1.134
|
0.667–1.902
|
0.64
|
Age, years
|
1.000
|
0.979–1.017
|
0.95
|
BMI, kg/m2
|
0.917
|
0.867–0.972
|
0.004*
|
disease duration, day
|
1.000
|
0.996–1.004
|
0.70
|
duration from the first visit to operation, day
|
1.006
|
1.000-1.015
|
0.072
|
Creatinine, mg/dL
|
1.079
|
0.965–1.241
|
0.19
|
eGFR, mL/min/1.73m2
|
0.998
|
0.994–1.002
|
0.22
|
BUN mg/dL
|
1.018
|
1.003–1.035
|
0.015*
|
Albumin, g/dL
|
0.754
|
0.531–1.105
|
0.14
|
Total cholesterol, mg/dL
|
0.999
|
0.991–1.007
|
0.73
|
Hemoglobin, g/dL
|
0.967
|
0.875–1.073
|
0.53
|
Platelet, 103/µL
|
0.984
|
0.967–1.002
|
0.073
|
White blood cell, 103/µL
|
1.013
|
0.967–1.068
|
0.61
|
Lymphocyte %
|
0.978
|
0.955–1.004
|
0.097
|
Neutrophil %
|
1.019
|
0.998–1.041
|
0.071
|
ESR, mm/1h
|
1.006
|
0.998–1.014
|
0.17
|
CRP, mg/dL
|
1.055
|
1.026–1.088
|
< 0.001*
|
CONUT
|
1.055
|
0.964–1.156
|
0.24
|
mGPS
|
1.721
|
1.119–2.558
|
0.015*
|
CAR
|
1.136
|
1.054–1.241
|
< 0.001*
|
NLR
|
1.033
|
1.004–1.073
|
0.023*
|
PNI
|
0.969
|
0.938–1.003
|
0.074
|
PLR
|
1.001
|
1.000-1.003
|
0.18
|
dialysis
|
1.138
|
0.513–3.030
|
0.77
|
Diabetes mellitus
|
1.350
|
0.802–2.315
|
0.26
|
Smoker
|
0.910
|
0.492–1.815
|
0.78
|
Use of steroids
|
1.972
|
0.586–12.346
|
0.31
|
Use of anticancer drug
|
2.493
|
0.538–44.334
|
0.30
|
Use of antibiotics at admission
|
0.971
|
0.542–1.851
|
0.93
|
Kulowski classification acute/subacute and chronic
|
2.188
|
1.003–5.747
|
0.049*
|
JOA score %
|
0.987
|
0.974–1.001
|
0.068
|
Griffith early/destructive and sclerotic
|
1.045
|
0.622–1.767
|
0.87
|
MRI stage Ⅲ, Ⅳ, and Ⅴ / Ⅰ and Ⅱ
|
2.762
|
1.125–5.848
|
0.029*
|
No. of infected vertebral bodies
|
1.410
|
1.033–1.965
|
0.030*
|
Epidural abscess
|
1.553
|
0.902–2.770
|
0.11
|
Psoas abscess
|
1.908
|
1.055–3.676
|
0.032*
|
Paraspinal abscess
|
1.980
|
0.862–5.747
|
0.11
|
*p < 0.05. BMI, body mass index; BUN, Blood urea nitrogen; CAR, CRP-albumin ratio; CI, confidence intervals; CONUT, controlling nutritional status; CRP, C-reactive protein; eGFR, estimate glomerular filtration rate; ESR, erythrocyte sedimentation rate; HR, hazard ratio; JOA, Japanese Orthopaedic Association; mGPS, modified Glasgow prognostic score; NLR, neutrophil–lymphocyte ratio; PLR, platelet–lymphocyte ratio; PNI, prognostic nutrition index, MRI, magnetic resonance imaging.
Next, we investigated independent predictors associated with the duration required to control infection using a stepwise multivariate Cox regression analysis. Based on univariate analysis, the dependent variable was defined as the duration required to control infection, and the independent variables were BMI, duration from the first visit to the operation, BUN, platelet count, lymphocyte count, neutrophil count, CRP, mGPS, CAR, NLR, PNI, Kulowski classification acute/subacute and chronic phase, JOA score, MRI stage ≥ Ⅲ, the number of infected vertebral bodies, and psoas abscess. As a result, CRP (odds ratio = 1.068, 95% confidence interval [CI] 1.033–1.109, p < 0.001), the number of infected vertebral bodies (odds ratio = 1.499, 95% CI 1.072–2.166, p = 0.017), and duration from the first visit to operation (odds ratio = 1.006, 95% CI 1.001–1.014, p = 0.022) were identified as the independent risk factors (Table 5). Although not significant, Kulowski classification was also selected in the final model.
Table 5
Multivariate Cox regression analysis of demographic and radiographic data.
Characteristics
|
HR
|
95% CI
|
p
|
CRP, mg/dL
|
1.068
|
1.033–1.109
|
< 0.001*
|
No. of infected vertebral bodies
|
1.499
|
1.072–2.166
|
0.017*
|
Duration from the first visit to operation, day
|
1.006
|
1.001–1.014
|
0.022*
|
Kulowski classification acute/subacute and chronic
|
2.270
|
0.9543-6.380
|
0.065
|
*p < 0.05. CI, confidence intervals; CRP, C-reactive protein; HR, hazard ratio |
Cutoff value for CRP and the number of infected vertebral bodies
To enhance the clinical utility of the results of this study, we investigated cutoff values for CRP and the number of infected vertebral bodies. For this purpose, we performed a logistic regression analysis using the average number of days of hospitalization for pyogenic spondylitis in the DPC data, i.e., 42 days or more for the long-term hospitalization group and less than 42 days for the short-term hospitalization group. The area under the curve (AUC) corresponding to the CRP proved to be the best indicator for long-term hospitalization. The cutoff value for CRP was 7.87 (Fig. 1). The AUC was 0.75. The sensitivity and specificity were 0.67 and 0.84, respectively. The receiver operating characteristic curve for the number of infected vertebral bodies is shown in Fig. 2. The cutoff value for the number of infected vertebral bodies was 3 (Fig. 2). The AUC was 0.67. The sensitivity and specificity were 0.30 and 0.96, respectively.