Baseline characteristics of all cases in the recurrence group and Non-recurrence group
A total of 653 CSDH cases were enrolled in this study at last (Figure 1). This study group included 561 male cases (85.9%) and 92 female cases (14.1%). Patient’s age ranged from 21 to 100 years with a median of 72 years (interquartile range 64 to 80 years). (Table 1). 8 patients with hospitalized mortality were excluded and there was no mortality at following up in this study.
The descriptive characteristics between groups with and without recurrence are exhibited in Table 1, including the demographics, laboratory, imaging, medication and comorbidity characteristics. In this study, 96 (14.7%) cases were diagnosed as CSDH recurrence, including 16 patients who needed a second operation. Compared with non-recurrence, the cases in the group of recurrence were more likely to be older and have lower level of serum leukocyte, neutrophil and platelet counts. In the meanwhile, lower serum fibrinogen concentration was examined in recurrence cases (p<0.05). Moreover, there was a statistical difference exhibited for serum BUN concentration of both pre and post-operation between two groups (p=0.001 and p<0.001,respectively). Table 2 revealed that the BUN level of preoperation was significantly higher than the postoperative BUN level in the non-recurrence group (p<0.001) while it showed no obvious difference in the recurrence group.
Baseline characteristics of all cases in BUN quartiles
For further exploration, the cases were divided into 4 groups on the basis of quartiles of the postoperative BUN concentration. The cut-off points for this stratification of the BUN concentration into quartiles were: Q1 ≤ 4.0 mmol/L, 4.0 < Q2 ≤ 4.9 mmol/L, 4.9 < Q2 ≤ 6.4 mmol/L and Q4 > 6.4 mmol/L. Table 3 summarized the characteristics of the CSDH cases by the quartiles of BUN. Cases with different postoperative BUN concentration appeared to be similar in most features except for age, Cr, erythrocyte and hemoglobin. These factors would be adjusted for multivariate-adjusted binary logistic regression for good measure. 22 (14%) of 157 patients in the highest quartile of BUN suffered moderate disability at discharge which was statistically higher than in other quartiles.
Association between the BUN concentrations and recurrence
Significant differences were got between the recurrence and non-recurrence groups in BUN concentration quartiles of cases (P = 0.003). The proportion of cases in the lowest quartile (≤4.0 mmol/L) was dramatically low in the recurrence group (P = 0.027), whilst the proportion of cases in the highest quartile (>6.4 mmol/L) was significantly high in the recurrence group (P = 0.012) (Table 4).
In Table 5, with all cases taken as a whole, the condition that CSDH recurrence was interpreted as a dependent variable and the lowest quartile was interpreted as the reference was used for postoperative BUN level in the binary logistic regression models. The highest quartile of BUN concentration (>6.4 mmol/L) was independently estimated as a risk factor of CSDH recurrence with an unadjusted OR of 3.315 (95%CI:1.711–6.423, p<0.001). After adjusting for the confounders including sex, age, current alcohol drinking, current smoking, comorbidities (hypertension, diabetes mellitus, coronary heart disease), medicine (Atorvastatin and PAMBA),and laboratory investigation (platelet, fibrinogen, leukocyte, erythrocyte, hemoglobin, Cr), the highest quartile of BUN remained significantly and independently associated with CSDH recurrence (model 1: OR=2.892, 95% CI:1.463–5.717, p=0.002; model 2: OR=2.939, 95% CI:1.480–5.836, p=0.002; modal 3: OR=3.069, 95%CI:1.488–6.330, p=0.002). There was no multicollinearity between the independent variables in model 3. Furthermore, restricted cubic spline regressions were used to explore the linear relationship between BUN concentration and the risk of CSDH recurrence (Figure 3). Most importantly, it could be observed visually that the highest quartile had significantly high OR value.