Predictors of perioperative blood transfusion and intraoperative blood loss in thoracolumbar tuberculosis

Background. During surgery for thoracic and lumbar tuberculosis infection, patients can lose a signicant amount of blood and thus require a perioperative blood transfusion. However, the risk factors for increased intraoperative blood loss and perioperative blood transfusion have yet to be identied. The aim of this retrospective study was to determine the predictors of perioperative blood transfusion and intraoperative blood loss in thoracolumbar tuberculosis. Methods. From 2008 to 2018, 336 patients who met the inclusion criteria were enrolled in the study. The predictors of allogenic blood transfusion were identied using univariate and multivariate logistic regression analyses. Univariate and multivariate linear regressions were conducted to investigate the risk factors for intraoperative blood loss. The predictors of high levels of intraoperative blood loss were analyzed by multivariate logistic regression analysis. Results. Altogether, 336 adult patients with thoracic and lumbar tuberculosis were included in this study. The mean age of patients was 49.6 ± 15.5 years old (range 14-85). Our data revealed signicant relationships between blood transfusions and female gender, BMI, vertebral collapse/kyphosis and intraoperative blood loss. Multivariable linear regression analysis revealed that BMI, levels of instrumentation, surgical approach and operative time were independent risk factors for intraoperative blood loss. Specically, a lower BMI, decreased preoperative hemoglobin levels, four or more levels of instrumentation, a combined surgical approach and a prolonged operative time were identied as risk factors for high levels of intraoperative blood loss. Conclusions. This study identied some clinical predictors of perioperative blood transfusion and intraoperative blood loss in patients undergoing thoracic and lumbar tuberculosis surgery. These results may contribute to the planning of preoperative blood transfusions and help to minimize intra- or postoperative complications.


Background
Subacute and chronic spinal infections are usually caused by a wide spectrum of pathogens, of which Mycobacterium tuberculosis is considered the most common. [1] The most common type of spinal tuberculosis is spondylodiscitis. In the last two decades, there has been a resurgence of tuberculosis in developed countries. [2] However, there are still challenges in the management of spinal tuberculosis. [3,4] Surgical treatment is indicated for patients with infections that are resistant to antibiotic therapy or patients with bone destruction, kyphosis, neurologic impairment and severe pain [4][5][6][7]. Generally, debridement and bone grafting are the rst choice of surgical procedures. [8,9] Despite advances in surgical techniques, spinal tuberculosis is still associated with substantial blood loss, and many patients require a perioperative blood transfusion. [4,10] Because tuberculosis is a systemic disease, the use of autologous blood transfusion is not recommended for such patients. In addition to issues regarding shortages and cost, allogeneic blood transfusion can be associated with serious complications and a prolonged operation. [11,12] Thus, identifying patients at a high risk of intraoperative blood loss and those who are likely to require a perioperative blood transfusion is an important step to improve postoperative outcomes, reduce complications and minimize health resource utilization. [13][14][15] To date, few studies have elucidated the risk factors for blood loss during spinal tuberculosis surgery. Research has also yet to identify which clinical factors predict the need for perioperative blood transfusion and blood loss in patients undergoing debridement and reconstruction procedures. Therefore, the objective of the present study was to identify predictors of high intraoperative blood loss and the need for perioperative blood transfusion from our series.

Methods
This study was approved by the university review board. A total of 336 patients with thoracic and lumbar spinal tuberculosis who underwent surgery were identi ed from two independent hospitals between January 2010 and November 2018. The diagnosis of spinal tuberculosis was based on the following previously reported criteria [10,16]: clinical manifestations, radiological evidence, response to antimicrobial therapy and results of microbiological examination. Indications for surgery included vertebral collapse and spinal instability; severe kyphotic deformity or progressive worsening of kyphosis; spinal cord compression by abscess or necrosis; formation of a hollow or sequestrum; or no improvement in antimicrobial therapy. The study population consisted of patients > 18 years old. Patients with recurrence at the lesion site and who underwent deformity surgery without debridement were excluded.
The variables extracted for analysis included patient age, gender, preoperative hemoglobin levels, body mass index (BMI), location of the lesion (thoracic, thoracolumbar, lumbar or lumbosacral spine), vertebral collapse or kyphosis, discitis involved, levels of instrumentation, paraspinal/epidural abscess, surgical approach (anterior, posterior and combined procedure), operative time, the amount of intraoperative blood loss and the need for allogenic blood transfusion. The recorded intraoperative blood loss was obtained from the anesthetist's case records and con rmed by surgeons. Blood loss was further assessed using the electronic records in the institutional database. A blood loss of over 30% of the total blood volume during surgery was set as the cutoff for high levels of intraoperative blood loss. The total blood volume was calculated according to a previously reported method. [15] [17] The surgical procedures were classi ed into three categories. The anterior approach was performed by anterior or lateral-anterior debridement, autologous or allograft bone graft and instrumentation. The posterior approach was performed by posterior debridement, autologous or allograft bone graft and instrumentation. The combined approach was performed by posterior instrumentation combined with anterior debridement and autologous or allograft bone graft.

Statistical Analysis
The independent factors used for prognostic modeling analysis were measured using either categorical or continuous variables. The logistic regression model was used in the univariate or multivariate analysis to identify the predictors of perioperative blood transfusion. To estimate the effect of different factors on intraoperative blood loss, both univariate and multivariate linear regression analyses were performed. The level of signi cance was set at 0.05 in all analyses. Statistical analyses were performed using IBM SPSS statistical software (version 23.0, Armonk, NY, USA).

Patient Demographics and Clinical characteristics
A total of 336 patients with thoracic and lumbar tuberculosis who underwent surgery were evaluated for this study. The mean age of the patients was 49.6 ± 15.5 years old (range 14-85). There were 178 female and 158 male patients, and the mean preoperative hemoglobin level was 122.04 ± 17.37 g/L. Perioperative blood transfusion was performed for 66.1% of the patients. A total of 11.3% of patients (38) had high intraoperative blood loss during surgery. For the entire cohort, the mean intraoperative blood loss was 661.37 ± 557.54 ml. The mean blood loss was 832.13 ± 590.17 ml for the patients who received a blood transfusion and 276.75 ± 187.19 ml for the patients who did not. Table I shows the characteristics of the study patients.

Factors In uencing Intraoperative Blood Loss
Univariate linear regression analysis ( Table 2) revealed that BMI, the number of involved discitis, levels of instrumentation, surgical approach and operative time were correlated with intraoperative blood loss. Moreover, multivariate linear regression analysis revealed that a higher BMI, higher levels of instrumentation, a combined approach and a prolonged operative time were linked to higher intraoperative blood loss ( Table 2). The results showed that the amount of blood loss increased as BMI increased (18.42 mL per 1 kg/m 2 increase in BMI) and as operative time increased (2.33 ml per 1 min increase in operative time). In patients undergoing a combined surgical approach, the amount of blood loss increased by 325.78 ml. The amount of intraoperative blood loss was signi cantly higher (514 ml higher) in patients with four or more levels of instrumentation.  Multivariable logistic regression analysis (Table 3) revealed ve signi cant predictors of high intraoperative blood loss in the entire cohort. Next, we developed a tool that allows us to more accurately estimate the risk of high blood loss during surgery at an individual level. Nomograms were established by the multivariate logistic regression model according to all the signi cant variable factors. Figure 1 shows the nomogram predicting the risk of high amounts of intraoperative blood loss. The C-index of this nomogram is 0.784 (bootstrapped 95% CI [0.729-0.834]). Figure 2 shows a calibration plot comparing the predicted probabilities of a high amount of intraoperative blood loss based on 228 bootstrapped samples. In the univariate regression analysis (Table 4), the numbers of involved discitis, levels of instrumentation, the surgical approach, the operative time and the intraoperative blood loss were signi cantly associated with perioperative blood transfusion. Furthermore, multivariate logistic regression analysis was performed to adjust for interactions between covariates and identi ed four predictors of allogeneic blood transfusion in the perioperative period ( Table 4). The risk of allogenic blood transfusion increased with female sex, vertebral collapse/kyphosis, and intraoperative blood loss. A higher BMI decreased the risk of allogeneic blood transfusion in patients. Durand reported that operative duration, surgical invasiveness, hematocrit, weight, and age were the most in uential variables for predicting blood transfusion. [14] In addition, it has been reported that preoperative hemoglobin levels are an important predictor of allogenic blood transfusion in the perioperative period for patients who underwent surgery. However, in our analysis, preoperative hemoglobin levels did not predict the need for a perioperative blood transfusion among patients with spinal tuberculosis. Our data also revealed that the preoperative hemoglobin levels were signi cantly correlated with high intraoperative blood loss. Preoperative low hemoglobin levels may increase postoperative complications for patients with spinal tuberculosis. [18] In addition, several studies have shown that preoperative anemia is associated with poor outcomes after surgery and increased health care use. [19,20] Therefore, the correction of preoperative anemia may improve the outcome of spinal tuberculosis surgery.
Several clinical factors, including the higher number of involved discitis, higher levels of instrumentation and combined surgical approach, were found to be correlated with the increased rate of blood transfusion and intraoperative blood loss. However, after adjusting for interactions between covariates in the multivariate analysis, the in uence of these factors was revealed to be nonsigni cant, and the odds of transfusion increased by 4.5 times when patients had a pathological fracture or kyphosis. The choice of surgical approach has been reported to be linked to intraoperative blood loss. [21,22] Previous studies reported that one-stage anterior operation had advantages when compared to posterior instrumentation, as both instrumentation and grafting are performed as single-stage surgery through the same incision, which may minimize total blood loss. [21,23] In contrast, other studies illustrated that the posterior approach favors less intraoperative blood loss. This difference may be attributed to the preference and experience of surgeons. Usually, the combined approach often leads to prolonged operative time and more blood loss, which may be associated with signi cant kyphosis and multiple-level lesions. [24] Furthermore, our investigation revealed that increased operative time is strongly associated with a higher likelihood of blood transfusion and a higher amount of blood loss. Four or more levels of instrumentation are also a risk factor for high intraoperative blood loss. In fact, operating time and blood loss are associated with an increase in the number of levels fused or instrumentation in spine surgery. [13,25] Kumar reported the results of their review of 243 patients undergoing surgery for metastatic spinal tumors and identi ed primary tumor, type of surgery, and prolonged surgery time as predictors of increased blood loss. [25] Our results showed that female sex and lower BMI were risk factors for blood transfusion. A possible explanation for the relationships between blood transfusion and gender and BMI is that a smaller body size may increase intraoperative blood loss. [26] In our cohort, the mean age was 49.5 ± 15.6 years old. In North America, Europe and China, tuberculous spondylodiscitis is more commonly seen in adult patients with a mean age of 40 years [1,27]. It has been reported that advanced age did not increase the morbidity associated with spinal operation. [28] For most patients, robust compensatory mechanisms render hypotension an insensitive indicator of shock until more than 30% of the patient's blood volume has been lost; therefore, intraoperative blood loss over 30% of the total blood volume was set as the cutoff in this study. Intraoperative cues indicative of intraoperative blood transfusion include low hematocrit and hemoglobin levels (< 70 g/L), a higher heart rate (> 120 beats/min), a weak peripheral pulse, and cool extremities with pale or mottled skin, among others. [29] Several factors probably contribute to reducing the risk of bleeding during spinal tuberculosis surgery.
Proper surgical approaches, the discontinuation of anticoagulant medications, the correction of preoperative anemia and the preoperative estimation of blood loss are essential in minimizing blood loss in surgery [30]. Intraoperative methods to control blood loss include positioning to reduce intraabdominal pressure, the use of tranexamic acid and the use of topical anti brinolytic agents. Moreover, modi ed anesthesia techniques such as core temperature control and hypotensive anesthesia have shown promising results in safely reducing blood loss [30][31][32].
There were several limitations in this study. This was a retrospective study with a small sample size. A multicenter study should be conducted. Almost all the patients in this cohort had spondylodiscitis. Patients with atypical forms of spondylitis without disc involvement were not included in this study.
There was also a lack of clinical information on commodities, medications, nutritional status and other variables. Tranexamic acid was not given to patients in this cohort; therefore, its effect on spinal tuberculosis surgery was not evaluated. It has been reported that tranexamic acid could reduce both intraoperative-perioperative allogeneic transfusion rates and operative time in spinal surgeries. [33] A recent study demonstrated that tranexamic acid contributes to reducing the drainage and blood transfusion perioperative duration of tuberculosis patients, but it cannot decrease intraoperative blood loss.
[34] The cause of transfusion was also not collected, but patients typically get transfused when their hemoglobin level less than 70 g/L, and they present changes in symptomatic or vital signs. Additionally, the risk of bleeding with regards to other causes of spine surgery was not investigated and a such comparation would be further analyzed in future work. However, despite such limitations, we believe that we assessed several important clinical variables that allowed us to determine the relationships between the evaluated surgeries and blood loss and the risk of blood transfusions.

Conclusions
In conclusion, the ndings of this study have substantial implications for perioperative management in patients with spinal tuberculosis. Our study revealed a signi cant relationship between blood transfusions and female gender, BMI, vertebral collapse/kyphosis and intraoperative blood loss.
Furthermore, a lower BMI, decreased preoperative hemoglobin levels, four or more levels of instrumentation, a combined surgical approach and a prolonged operative time were identi ed as risk factors for high levels of intraoperative blood loss. An understanding of these factors may contribute to preoperative planning and help to minimize postoperative complications.

Declarations
Ethics approval and consent to participate This study was approved by the Medical Ethics Committee of the second a liated Hospital of Harbin Medical University, Heilongjiang. A certi cate of approval has been provided. Due to the retrospective nature of the study, formal informed consent from the participants was not required.

Consent to publish
Not applicable.

Availability of data and materials
The datasets analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests. The prediction nomogram of the high amount of intraoperative blood loss. The nomogram is used by adding up the points identi ed on the points scale for each predictor. According to the sum of these points projected on the bottom scales, the nomogram can provide the predicted probability for intraoperative high blood loss.