Demographic and clinical data
Among the 118 patients, 45 were excluded due to surgery-related development (n=25), tuberculosis (n=12), ankylosing spondylitis (n=2), and follow-up loss (n=6). The final analyses were performed on 73 patients (44 men and 29 women) with a mean age of 64.73 ± 11.61 (42-84) years. There were no statistically significant differences in age, gender, underlying diseases, and history of spinal procedures between the two groups. However, there were statistically significant differences in the extent of affected PVO lesion [1.49 ± 0.93 (1-5) vs. 2.09 ± 1.28 (1-5) levels, p = 0.022] and the presence of epidural abscess (43.9% [18/41] vs. 84.4% [27/32], p = 0.001), except the presence of psoas muscle and paraspinal abscess. The recurrence rates were 7.3% (3/42) and 6.3% (2/32), respectively (p = 1.000). Detailed data are mentioned in Table 1.
Microorganisms and antibiotics
The rate of causative bacterial identification was only 43.8% (32/73) with 8 (25%) from blood culture, 15 (46.9%) from tissue culture of CT-guided needle or open surgical biopsy of the PVO lesion, and 9 (28.1%) from both blood and tissue cultures. According to the culture methods, 17 of 73 patients (23.3%) with blood cultures, 7 of 37 patients (18.9%) with CT-guided needle biopsy, and 17 of 36 patients (47.2%) with open surgical biopsy were identified. Microbiologic findings of the 32 patients in the CP group are shown in Table 2. The most common causative microorganism was Staphylococcus aureus (40.6%, 13/32). Five cases of recurrence were noted: one with methicillin-resistant Staphylococcus epidermidis, one with Enterobacter, and three with culture-negative cases. Detailed data are mentioned in Table 2.
The antibiotic regimens for the treatment of PVO are summarized in Table 3. β-Lactam and glycopeptide were the mainly used as the effective antibiotics in both groups. The use of glycopeptide did not significantly differ between the two groups. However, quinolones were used statistically significantly more in the CN group (22.0% [9/41] vs. 3.1% [1/41], p = 0.036). Antibiotics were administered before tissue culture of the PVO lesion in 71.2% cases (52/73), and there was no significant difference between the two groups (78.0% [32/41] vs. 62.5% [20/32], p = 0.194). There was no statistically significant difference in the duration of total antibiotics between the two groups (101.17 ± 52.84 of CN group vs. 84.19 ± 50.29 days of CP group, p = 0.168). However, there were statistically significant differences in the duration of parenteral antibiotics (45.88 ± 16.14 vs. 57.31 ± 24.39, p = 0.019), oral antibiotics (55.29 ± 47.40 vs. 26.84 ± 41.10, p = 0.009), and the incidence of using oral antibiotics (75.6% [31/41] vs. 43.8% [14/32], p = 0.001) between the two groups, respectively. Detailed data are provided in Table 4.
The clinical variables related to the causative bacterial identification are presented in Table 5. The clinical variables with statistical significance identified in a univariate logistic regression analysis were included in a multivariate logistic regression analysis. Epidural abscess and initial VAS score of back pain were statistically significant predictors.
Changes of ESR/CRP and VAS score for back pain during the early 3-month after starting antibiotics
There were statistically significant improvements in ESR, CRP, and VAS score of back pain in both groups during early 3 months, respectively (p < 0.01). In ESR, there were no statistically significant differences at initial, 1 week, 1 month, and 3 months findings between the two groups. However, there were statistically significant differences in CRP (9.21 ± 7.43 vs. 15.17 ± 10.32 mg/dL, p = 0.005) and VAS score of back pain (7.41 ± 0.92 vs. 8.13 ± 1.13, p = 0.004) at initial assessment between the two groups, respectively. However, there were no statistically significant differences in CRP and ESR at 1-week, 1-month, and 3-month between the two groups, respectively. When parenteral antibiotics discontinued or changed to oral antibiotics, there were no statistically significant differences in ESR, CRP, and VAS score of back pain between the two groups. The values of ESR, CRP, and VAS score of back pain in total patients were 44.85 ± 24.99 mm/h, 0.89 ± 1.02 mg/dL, and 3.98 ± 1.08, respectively. Detailed data are provided in Table 6.