Comparison of Clinical Data Between Patients With Complications and Without Complications After Spinal Tuberculosis Surgery: A Propensity Score Matching Analysis

Purpose This study used a propensity score matching (PSM) analysis to explore the risk factors of post-operative complications and compared the differences in clinical data between them following spinal tuberculosis surgery. Methods The clinical data of patients with spinal tuberculosis were collected in our hospital from June 2012 to June 2021, including general information, laboratory results, surgical information, and hospitalization costs. The data were divided into two groups: complication and without complication groups. The baseline data of the two groups were obtained using the PSM analysis. Univariate and multivariate logistic analyses were used to analyze the differences between the two groups. Results A total of 292 patients were included in the PSM analysis: 146 patients with complications and 146 patients without complications. The operation time, incision length, hospital stay, and albumin quantity in the complications group were 162 ± 74.1, 11.2 ± 4.76, 14.7 ± 9.34, and 1.71 ± 2.82, respectively, and those in the without complication group were 138 ± 60.5, 10.2 ± 3.56, 11.7 ± 7.44, and 0.740 ± 2.44, respectively. The laboratory costs, examination costs, guardianship costs, oxygen costs, and total costs in the complications group were higher than those in the without complication group. A significant difference was observed in the albumin quantity by logistic regression analysis (P < 0.05). Conclusion Several costs in the complication group were higher than in the without complication group. The albumin quantity may be an independent factor to predict post-operative complications of spinal tuberculosis by logistic regression analysis.


INTRODUCTION
Spinal tuberculosis is the most common form of extrapulmonary tuberculosis (1). It has become a major health concern in developing countries due to the emergence of drug-resistant strains (2,3). Surgical treatment is known to be the best solution in case the anti-tuberculosis treatment fails (4). However, surgeries are sometimes associated with certain unexpected complications. The results of Zhou et al.'s study revealed that 102 patients with L1-2 vertebral tuberculosis had 14 patients with pulmonary complications after the surgery (5). Zhuang et al. applied oblique lateral interbody fusion to lumbar spinal tuberculosis with 10 (16.95%) complications, including peritoneal injury, neurological injury, infection of incisions, segmental artery, iliac vein lacerations, and instrument failure. However, the anterior-only approach was associated with 37.5% complications (4). Similarly, Susanta et al. reported one case (2.3%) of sinus formation and two cases (4.7%) of superficial wound infection after thoracic tuberculosis surgery (6). Surgeons should completely understand the surgical approach and postoperative complications of spinal tuberculosis, which is of great significance in treating the disease (7).
The global tuberculosis situation is particularly tense in developing countries. The incidence of pulmonary or spinal tuberculosis is low in developed countries, however, the incidence is high in developing countries. The incidence of tuberculosis is high in the African region, which accounted for 71% of global tuberculosis and AIDS cases in 2018 (8). The backward economies of developing countries cannot provide good sanitation and treatment facilities, which is also the reason for the high incidence of tuberculosis (9). Insufficient course of antituberculosis therapy after spinal tuberculosis surgery due to financial burden, which may increase the recurrence of spinal surgery.
In the past, our team's research had found that there were various complications of spine surgery. There was a potential risk of cerebrospinal fluid leakage and reduced albumin after surgery for spinal tuberculosis (10,11). In addition, we also found a risk of internal fixation failure following spinal surgery (12). In a recent study of spinal tuberculosis, our team found that spinal tuberculosis was prone to recurrence after surgery and the risk of pulmonary embolism after multiple surgical treatments (13). Therefore, we collected a large number of clinical data on spinal tuberculosis surgery with the aim of predicting the risk of complications of spinal tuberculosis.
Although PSM analysis has been favored in spinal surgery, including cervical spine surgery (14), thoracic surgery (15), and lumbar spine surgery (16), only a few reports are available on spinal tuberculosis with PSM analysis. However, a combination of PSM analysis and logistic regression analysis for spinal tuberculosis analysis has not been reported. This study was the first to apply a combination of PSM analysis and logistic regression analysis to study the complications after spinal tuberculosis. We compared the clinical data of complication and without complication groups and explored the factors related to complications after tuberculosis surgery.

MATERIALS AND METHODS
The clinical data of patients with spinal tuberculosis were collected in our hospital from June 2012 to June 2021, including general information, laboratory results, surgical information, post-operative complications, and hospitalization costs. A total of 495 patients were enrolled in the study, including 298 males and 197 females. All data were divided into two groups: complication group with 146 patients (Figures 1-3) and without complication group with 349 patients. In this study, 292 patients were included in the PSM analysis: 146 patients in the complication group and 146 patients in the without complication group. The criteria for inclusion were as follows: (1) without other diseases that affected post-operative recovery except hypertension and diabetes, (2) diagnosed with spine tuberculosis by pathological examination, (3) without a history of surgery affecting the spine, (4) complete clinical data, and (5) Lesions involving any part of the spine, including cervical, thoracic, lumbar, and sacral. The criteria for exclusion were as follows: (1) other diseases that affected post-operative recovery except for hypertension and diabetes, (2) without diagnosis of spinal tuberculosis by pathological examination, (3) with a history of surgery affecting the spine, and (4) incomplete clinical data. Post-operative complications of spinal tuberculosis were defined as post-operative wound infection or systemic infection, failure of internal fixation, recurrence of tuberculosis, and surgery-related diseases such as pulmonary embolism, cerebral infarction, and myocardial infarction. According to the fixed vertebral body at the upper end and the fixed vertebral body at the lower end, we choose the appropriate length of incision. "Bottle" was defined as the unit of measurement for albumin quantity. Each bottle of albumin injection was 50 ml, of which albumin accounted for 20%. This study was approved by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University.

Data Collection
The general information of patients who underwent spinal tuberculosis surgery in our hospital was collected, including gender, body mass index (BMI), age, hypertension, diabetes, lesion segment, medical history, marriage, systolic blood pressure, diastolic blood pressure, pain, lower limb pain, number of lower limb pain, fatigue, fever, night sweats, appetite, weight loss, Oswestry Disability Index (ODI), Japanese Orthopedic Association (JOA) scores, visual analog scale (VAS), American Spinal Injury Association (ASIA) impairment scale, occupation, race, days before surgery, smoking, and drinking. The laboratory results of patients were collected, including blood glucose, blood type, C-reactive protein (CRP), hepatitis B surface antigen, white blood cells, hemoglobin, platelets, percentage of neutrophils, percentage of lymphocytes, absolute monocytes, percentage of monocytes, total bilirubin, direct bilirubin, indirect bilirubin, total protein, albumin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), AST/ALT, blood urea, blood creatinine, blood uric acid, and erythrocyte sedimentation rate. The surgical information of patients was collected, including titanium cage, internal fixation, abscess, albumin quantity, recurrence, surgical approach, hospital stay, transfusion, local streptomycin, operation time, bleeding volume, red blood cell transfusion, drainage, kyphosis, post-operative complications, and incision length. The hospitalization costs of patients were collected, including antibacterial costs, western medicine costs, surgical costs, anesthesia costs, laboratory costs, examination costs, guardianship costs, transfusion costs, oxygen costs, high consumable costs, care costs, consultation fees, and total costs. Antibacterial costs were defined as the cost of treatment with antibiotics such as penicillin, streptomycin and cefuroxime etc. Western medicine costs were defined as the cost of oral medication such as painkillers, omeprazole and glutathione etc. Surgical costs were defined as the cost of lesion removal, bone graft fusion, reconstruction of spinal stability, etc., but did not include the cost of purchasing internal fixation devices. High consumable costs were defined as the purchase of internal fixation devices such as pedicle screws, titanium cages. Anesthesia costs were defined as the cost of anesthesia during surgery. Laboratory costs were defined as the cost of blood tests before and after surgery. Examine costs were defined as the cost of pre-operative and post-operative imaging examinations, such as X-ray, CT, MRI, etc. Guardianship costs were defined as the cost of monitoring vital signs after surgery. Transfusion costs were defined as the cost of blood transfusions during hospitalization. Oxygen costs were defined as the cost of oxygen therapy during hospitalization. Care costs were defined as the costs of nurses serving patients during hospital stays. Consultation costs were defined as the cost of asking other departments for consultation and treatment during hospitalization. Total costs were defined as the total cost of hospitalization.

Statistical Analysis
The propensity score usually includes four methods, namely covariate adjustment using the propensity score, matching on the propensity score, stratification on the propensity score, and inverse probability of treatment weighting using the propensity score. The most popular of these in medical research is matching on the propensity score (17,18 (21). We referred to previous research strategies and used the PSM method to analyze the clinical data based on the presence or absence of complications after spinal tuberculosis surgery. PSM can reduce the interference factors of the outcome after matching the baseline data of treatment and control groups (22,23).
Although PSM analysis has been favored in spinal surgery, including cervical spine surgery (14), thoracic surgery (15), and lumbar spine surgery (16), only a few reports are available on spinal tuberculosis with PSM analysis. Univariate and multivariate analyses were usually applied to the analysis of related factors in the study of spinal tuberculosis. A study by Surachai et al. used logistic regression analysis and found that signal cord change and notable Cobb angle could be used as predictive factors for the neurological deficit in spinal tuberculosis (24). Moreover, Wang et al. used logistic  regression analysis and found that the independent risk factors of lower extremity motor or sensory deficits were age, worrying of sickness, location, and spinal compression in patients with spinal tuberculosis (25).
The RStudio software (version 1.4.1717) and R software (version 4.1.0) were used for data analysis. To eliminate the confounding factors that affected the results, we selected certain parameters for the PSM analysis, including gender, BMI, age, hypertension, diabetes, abscess, titanium cage, internal fixation, percentage of lymphocytes, albumin, and lesion segment. The balance of the matched data was tested using the classical method; the alpha was set to 5. In addition, the balance of the matched data was tested using the standard difference method; the alpha was set to 15. The data following the PSM analysis were included in univariate analysis. We refer to previous published articles (26,27), and on the basis of fully considering the clinical significance and statistical significance of the parameters. Parameters with P <0.1 in the univariate analysis were included in the multivariate logistic regression analysis. The validity of the logistic regression model was evaluated by the receiver operating characteristic (ROC) curve. The statistical difference was considered as P-value <0.05.

RESULTS
There were differences in baseline data before propensity matching, but none were statistically significant after propensity matching ( Table 1). The histogram showed a difference between categorical variables, including gender (Figure 4A), hypertension ( Figure 4B), diabetes (Figure 4C), abscess (Figure 4D), titanium cage (Figure 4E), and internal fixation (Figure 4F), before and after matching. The frequency distribution graph showed a difference in quantitative variables, including BMI ( Figure 5A), age (Figure 5B), percentage of lymphocytes (Figure 5C), albumin (Figure 5D), and lesion segment ( Figure 5E) before and after matching. The balance of the standardized differences of covariates was evaluated using the standard difference method; the results showed that the standardized difference of the matched data was significantly smaller than that of the unmatched data ( Figure 5F).
The general data between the two groups using the PSM analysis revealed that the P-value of appetite and occupation was <0.1, and that of other parameters was more than 0.1 ( Table 2).
We constructed a logistic regression model to further explore the risk factors that affected the complications. The parameters with P-values <0.1 were included in the logistic regression analysis, including appetite, occupation, AST/ALT, albumin quantity, surgical approach, operation time, and incision length. There was a significant difference in the albumin quantity using the logistic regression model analysis ( Table 6). Furthermore, there was a significant difference in the intercept of the regression The goodness of fit of the logistic regression model was assessed by HL-test and the chi-square value. P-values were 4.534 and 0.806, respectively. The confidence interval of the odds ratio ( Table 6) was further used to evaluate the logistic regression model. The ROC curve was applied to evaluate the validity of the logistic regression model. The area under the curve (AUC) was 0.75 (Figure 8).
In this study, various complications occurred after spinal tuberculosis surgery during post-operative follow-up ( Table 7). Eight patients were transferred to ICU for treatment due to post-operative respiratory failure, hypoxia, and inability to remove the ventilator. Wound infection occurred in 77 patients. After intensive wound management or two-stage debridement, they can recover well. Pedicle screw fracture occurred in 1 patient, requiring a second surgery to re-implant the internal fixation device.

DISCUSSION
Spinal tuberculosis surgery has become the most common treatment for spinal tuberculosis, including anterior (28), posterior (29), and combined anterior and posterior approaches (30). Moreover, the literature reports that spinal tuberculosis surgery includes minimally invasive procedures and combined open and minimally invasive procedures (7). However, postoperative complications are often unexpected and also a challenge for surgeons (31,32). The complications of spinal tuberculosis vary depending on the location. The complications in cervical tuberculosis have been reported to be more than in non-cervical tuberculosis (33). In addition, post-operative complications of spinal tuberculosis vary depending on the surgical approach. The complications of the anterior approach were higher than in the posterior approach (7).
The parameters with a P-value <0.05 in the univariate analysis included albumin quantity, operation time, and incision length. Patients with classic spinal tuberculosis were often accompanied by weight loss (34); these patients who underwent surgical treatment were prone to poor wound healing due to post-operative nutritional insufficiency (35)(36)(37). Under such conditions, clinicians often use albumin supplementation to promote the recovery of patients (38). In addition, a longer surgical incision results in more secretion of fluid from the incision and increased susceptibility to wound infection (39). This could be the reason that the length of the incision in the complication group in this study was longer than that in the without complication group, and the albumin quantity in the former was more than that in the latter. If the operation time exceeded 3 h, the second group of antibiotics often needed to be increased, which potentially increased the risk of post-operative infection and other complications in case of prolonged operation time (40). Therefore, the operation time in the complication group was higher than in the without complication group.
Various costs, including hospital stay, laboratory costs, examination costs, guardianship costs, oxygen costs, and total costs, were higher in the complication group than in the without complication group. Research by Marah et al. showed that postoperative respiratory failure increased the cost of surgery by more than 50% (41). Mark et al. studied the relationship between surgical complications and hospitalization costs and found that hospitalization costs were on an average of $19,626 higher in patients with complications than in those without complications (42). The findings of Hunt et al. showed that post-operative complications not only increased the length of hospital stay but  also increased the cost of hospitalization (43). This was consistent with the results of our study.
In the present study, only the albumin quantity was significantly different by multivariate logistic regression analysis. In addition, the albumin quantity in the complication group was higher than that in the without complication group. Li et al.'s study of spinal tuberculosis found that the albumin in the anterior approach group decreased from 37 g/L preoperatively to 24.2 g/L post-operatively and that in the posterior approach group decreased from 38.6 to 28.2 g/L, with pneumonia and other complications in both groups  of surgical approaches (44). A close relationship existed between albumin and post-operative complications (45), and albumin supplementation may accelerate the recovery of the post-operative condition (46). The logistic regression model was tested through the HL test method, and the P-value was 0.806 (P > 0.05), which implied no difference between the logistic regression model and the actual application. The value of the area under the ROC curve was 0.75, implying that this regression model can be used in clinical applications.
PSM analysis strategies, such as tumor chemotherapy (47), cervical spine surgery (48), robot-assisted surgery (49), and other areas, are widely used in medical research. The PSM analysis can reduce the difference between the two groups of comparable data and decrease the effect of interfering factors on the results by adjusting the balance of baseline data (22). Such a model allows the researchers to measure the probability of an outcome event occurring between treatment and control groups (50). The pre-matched imbalance and post-matched  balance were assessed using the balance standardized differences (23). We found a significant difference in the laboratory results of albumin value between the complication and without complication groups, which inevitably became a disturbing factor for another result that the albumin quantity was an independent risk by the logistic regression analysis. The albumin values of the two groups before the operation were adjusted to be the same by the PSM analysis, and the results of the study were more comparable. However, this study had certain limitations. (1) Some data were lost after the PSM analysis. (2) More sample data should be collected to obtain more pairs. (3) This subject was a single-center study, and the results need to be verified by a multi-center study.

CONCLUSION
In this study, the costs in the complication group were generally higher than in the without complication group. Multivariate logistic regression analysis found that albumin quantity could be an independent factor for predicting post-operative complications of spinal tuberculosis. Therefore, surgeons should pay more attention to the patient's economy and the occurrence of post-operative complications in clinical work.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics Committee of the First Affiliated Hospital of Guangxi Medical University. The patients/participants provided their written informed consent to participate in this study.