Introduction
The growing geriatric population has led to an increased prevalence of symptomatic adult spinal deformity (ASD) [1], with a corresponding increase in the volume of ASD surgeries [2]. ASD encompasses a range of complex spinal conditions, including scoliosis, kyphosis, and neural element compression, which can significantly impact patients’ quality of life and function [3]. Although ASD surgery can substantially improve the quality of life, high complication rates persist [4], including considerable estimated blood loss (EBL) [5], prolonged operative times [6,7], and extended length of stay (LOS) [6,8,9]. These complications exacerbate perioperative morbidity and impose a substantial economic burden [10]. Therefore, identifying and addressing modifiable risk factors is crucial to optimizing preoperative planning, perioperative care, and guiding patient expectations in ASD surgery.
Existing literature on ASD surgery highlights several factors contributing to increased EBL, operative time, and LOS. Predictors of higher EBL include a Cobb angle >50° [11], a greater number of levels fused [11,12], lower preoperative hemoglobin levels [12], and osteoporosis [5]. For operative time, combined anterior-posterior approach, higher frailty risk scores, and three-column osteotomies (3CO) are associated with longer operative durations [7]. Major blood transfusions [12], age, infections, neurologic complications, comorbidities [13], intraoperative complications [13], and institutional factors like academic settings [8] have been implicated in prolonged hospital stay. These findings underscore the multifactorial nature of adverse outcomes, encompassing patient characteristics, surgical factors, and institutional variables.
Despite extensive research on ASD surgery, a significant knowledge gap remains regarding the specific preoperative, intraoperative, and postoperative factors that most substantially impact EBL, operative time, and LOS. Given the potentially rapid escalation of complications in ASD operations, identifying high-risk patients can help improve patient safety and potentially reduce the need for multiple reoperations [14] or mitigate life-threatening outcomes [15]. This study aimed to investigate the preoperative, intraoperative, and postoperative factors most strongly associated with the highest quartile of EBL, operative time, and LOS in ASD surgery patients. We also aimed to determine if these high quartiles are linked to increased mechanical complications, reoperation rates, and adverse patient-reported outcomes measures (PROMs).
Materials and Methods
Study design
This was a single-center, retrospective cohort study of patients who underwent ASD surgery between 2009 and 2021, with a minimum 2-year follow-up. Five fellowship-trained neurosurgery and orthopedic spine surgeons contributed to the registry. Postoperative PROMs were collected with the assistance of five dedicated research staff members. The study was approved by the Institutional Review Board (IRB) of Vanderbilt University Medical Center (IRB #220894), and patient consent was waived due to the retrospective design.
Patient selection and independent variables
Only adult patients (age ≥18 years) undergoing ASD surgery were eligible for inclusion. Inclusion criteria were: fusion of ≥5 vertebral levels, Cobb angle ≥30°, sagittal vertical axis (SVA) ≥5 cm, coronal vertical axis (CVA) ≥3 cm, pelvic tilt (PT) ≥25°, or thoracic kyphosis (TK) ≥60°, and a minimum of 2-years of follow-up. Patients were categorized into two groups based on their EBL, operative time, and LOS: those in the fourth quartile (highest values) and those in the combined first to third quartiles (lower values). This quartile-based grouping allowed us to identify patients at the highest risk for each outcome, with the fourth quartile representing the most extreme values for EBL, operative time, and LOS. This approach is consistent with prior studies that have used quartile thresholds to stratify surgical risk [16].
Independent variables
Patient demographic data, such as age, sex, body mass index (BMI), and comorbidities, were compared between the quartile groups. Preoperative radiographic parameters included coronal measurements of CVA and major Cobb angle, as well as sagittal measurements of the L1–L4 angle, lumbar lordosis (LL), pelvic incidence, sacral slope (SS), corrected SVA, PT, and T1 pelvic angle (TPA) [17].
Intraoperative variables analyzed included total instrumented levels (TIL), number of interbody grafts, primary surgeon, 3CO, and perioperative hemoglobin levels. Mechanical complications, such as proximal/distal junctional kyphosis (PJK/DJK), rod fracture, and pseudarthrosis, were compared. PJK was defined as an angle of ≥10° between the inferior endplate of the upper instrumented vertebra (UIV) and the superior endplate of the vertebra two levels above the UIV, accompanied by a concurrent ≥10° change from preoperative measurements [18].
Outcome variables
The three primary outcomes in this study were: (1) EBL, (2) operative time, and (3) LOS. Each variable was divided into quartiles, and the highest quartile (4th quartile) was designated a priori as the outcome of interest. The comparison group for each outcome was the combined first to third quartiles.
For the study’s second objective, we evaluated whether the aforementioned variables were associated with three secondary outcomes: (1) mechanical complications, (2) reoperation, and (3) 2-year PROMs. The PROMs included the Oswestry Disability Index (ODI) [19], EuroQoL Group (EQ-5D) questionnaire [20], and Numeric Rating Scales for back and leg pain (NRS-BP and NRS-LP, respectively). We also compared the proportion of individuals achieving a minimally clinically important difference (MCID), defined as a 30% improvement from baseline [21], across groups for ODI, EQ-5D, and NRS scores.
Statistical analysis
Statistical analyses involved categorizing patients into the fourth quartile or the combined first to third quartiles for EBL, operative time, and LOS. Continuous variables were summarized using means and standard deviations, while categorical data were summarized using frequencies. The Shapiro-Wilk test assessed normality of distribution, and the F-test evaluated equality of variances across groups for continuous measures. Two-sample t-tests were used for normally distributed data with equal variances, while Wilcoxon rank-sum or Mann-Whitney U tests were used for non-normal data. Chi-square or Fisher’s exact tests were applied to categorical variables when expected cell counts were small. Univariate and multivariable logistic regression analyses were conducted on binary outcomes for EBL, operative time, and LOS, with the fourth quartile coded as 1 and the combined first to third quartiles coded as 0. This quartile-based approach enabled meaningful comparisons between patients with the highest resource utilization and the rest of the cohort. Covariates included age, comorbidities, and preoperative radiographic measurements. Statistical significance was determined using a p-value threshold of <0.05. All analyses were performed using the R software ver. 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria).
Results
The study included 238 patients with a mean age of 63.4±17.4 years and a predominantly female cohort (76.1%, n=181). The mean BMI was 28.9±6.9 kg/m2. Comorbidities included diabetes (18.5%, n=44), chronic obstructive pulmonary disease (COPD; 26.9%, n=64), congestive heart failure (CHF; 14.3%, n=34), hypertension (64.7%, n=154), and osteoporosis (23.9%, n=45). Prior smokers comprised 27.3% (n=65) of the cohort, while current smokers made up 9.2% (n=22) (Table 1). Patients were categorized into the 4th (highest) quartile and 1st–3rd quartiles for EBL (4th, n=59; 1st–3rd, n=169), operative time (4th, n=61; 1st–3rd, n=177), and LOS (4th, n=55; 1st–3rd, n=182).
Estimated blood loss
Patients in the highest EBL quartile did not significantly differ from those in the lower three quartiles in terms of age (67.4±12.7 years vs. 62.5±18.6 years, p=0.164), proportion of females (76.3% vs. 75.1%, p=0.863), or BMI (28.5±5.8 kg/m2 vs. 29.0±7.4 kg/m2, p=0.941). However, a higher proportion of individuals in the 1st–3rd quartiles had diabetes than in the 4th quartile (22.5% vs. 10.2%, p=0.039) (Table 1).
There was no significant difference in the proportion of patients undergoing an anterior-posterior approach between both groups (16.9% vs. 13.0%, p=0.454). Patients in the highest EBL quartile had a different distribution of operating surgeons compared to those in the 1st–3rd quartiles (p=0.005). Patients in the highest EBL quartile had significantly longer operative times (516.2±141.4 minutes vs. 388.2±134.3 minutes, p<0.001), higher rates of 3COs (30.5% vs. 14.8%, p=0.008), higher preoperative hemoglobin (13.1±1.7 g/dL vs. 12.2±1.7 g/dL, p<0.001), and longer LOS (7.9±8.5 days vs. 6.7±11.7 days, p=0.008) compared to the lower three quartiles (Table 2). A significantly higher proportion of individuals in the 4th quartile met the MCID EQ5D QALY (18.4% vs. 4.2%, p=0.031) (Table 3).
A univariable logistic regression revealed that 3CO predicted higher EBL (odds ratio [OR], 2.53; 95% confidence interval [CI], 1.25–5.08; p=0.009) (Table 4). Multivariable analysis revealed that a 3CO (OR, 3.60; 95% CI, 1.42–9.28; p=0.007) and older age (OR, 1.03; 95% CI, 1.00–1.06; p=0.039) were predictive of higher EBL (Table 4).
Operative time
There were no significant differences between those in the 4th versus 1st–3rd quartiles of operative time regarding age (68.5±10.6 years vs. 61.8±19.0 years, p=0.072), proportion of females (79.7% vs. 75.1%, p=0.480), or BMI (30.3±7.4 kg/m2 vs. 28.4±6.7 kg/m2, p=0.079) (Table 1). Patients in the highest operative time quartile had a higher L1–L4 kyphosis compared to those in the lower quartiles (2.1°±18.4° vs. −8.0°±18.0°, p<0.001). The highest quartile also had a larger LL (−17.0°±28.6° vs. −24.6°±33.9°, p=0.036) and lower SS (22.5°±12.2° vs. 29.0°±13.3°, p<0.001) compared to those in the lower 1st–3rd operative time quartiles (Table 1).
A higher proportion of patients in the highest quartile underwent an anterior-posterior approach compared to those in the 1st–3rd quartiles (31.3% vs. 8.5%, p<0.001). There were no significant differences between those in the 4th versus 1st–3rd quartile for operative time regarding the proportion of operating surgeons. Patients in the 4th operative time quartile had a higher EBL (2,160.5±1,338.0 mL vs. 1,218.6±1,106.2 mL, p<0.001), more TIL (12.1±3.2 vs. 10.0±3.0, p<0.001), higher rate of 3COs (27.1% vs. 15.3%, p=0.041), and a longer LOS (8.7±9.0 days vs. 6.3±11.2 days, p<0.001) than those in the lower quartiles (Table 2). Patients in the 4th quartile had a higher rate of rod fractures (30.5% vs. 15.8%, p=0.014), but they more frequently met the MCID for EQ5D QALY (18.8% vs. 4.9%, p=0.029) (Table 3).
Univariate logistic regression analysis revealed that a 3CO (OR, 2.07; 95% CI, 1.01–4.16; p=0.044) and a higher TIL (OR, 1.23; 95% CI, 1.12–1.36; p<0.001) were the strongest predictors of increased operative time. On multivariable analysis, higher TIL (OR, 1.26; 95% CI, 1.12–1.43; p<0.001) and older age (OR, 1.05; 95% CI, 1.02–1.08; p=0.003) were significant predictors of increased operative time (Table 5).
Length of stay
Those in the 4th quartile of LOS had similar age (67.9±9.8 years vs. 62.0±19.0 years, p=0.277), proportion of females (74.5% vs. 76.9%, p=0.716), and BMI (30.2±6.7 kg/m2 vs. 28.5±7.0 kg/m2, p=0.090) as those in the 1st–3rd LOS quartiles (Table 1). Those in the highest LOS quartile had more TIL (10.8±2.8 vs. 10.4±3.3, p<0.001), a similar distribution of operating surgeons (p=0.711), longer operative times (492.1±167.9 minutes vs. 398.0±133.3 minutes, p<0.001), higher rate of 3COs (27.3% vs. 15.4%, p=0.045), and a higher EBL (2,115.6±1,690.1 mL vs. 1,252.3±977.2 mL, p<0.001) than those in the lower three LOS quartiles. There were no significant differences in the proportion of patients undergoing an anterior-posterior approach between the two groups (20.0% vs. 12.6%, p=0.167) (Table 2). Patients in the 4th quartile for LOS reported lower EQ5D QALY scores than those in the lower three quartiles (0.6±0.2 vs. 0.7±0.2, p=0.012) (Table 3).
A univariable logistic regression predicting LOS revealed that the presence of two or more comorbidities (OR, 4.18; 95% CI, 1.63–13.0; p=0.006) and 3CO (OR, 2.06; 95% CI, 0.99–4.19; p=0.048) were the strongest predictors of increased LOS. On multivariable analysis, only TIL (OR, 1.23; 95% CI, 1.09–1.40; p=0.001) was a significant predictor of LOS, such that a higher TIL was associated with a longer LOS (Table 6).
Discussion
This study investigated the factors influencing EBL, operative time, and LOS in a cohort of 238 patients undergoing ASD surgery. Three-column osteotomies were associated with all three outcomes on univariable analysis. Multivariable analysis revealed that advanced age predicted higher EBL, while both advanced age and greater TILs predicted longer operative time. Additionally, greater TILs were associated with extended LOS.
3COs were a significant predictor of higher EBL, longer operative time, and extended LOS in ASD surgeries. The application of 3CO has evolved, shifting from broader use to a more selective approach, typically reserved for rigid deformities or previously fused spines due to the associated high risks and complications [22]. Preferred osteotomy levels have transitioned from mid-lumbar to lower lumbar levels, prioritizing spinal lordosis from L4–S1. Lower lumbar osteotomies are challenging and carry the risk of neurologic deficits from nerve root apraxia or injury [23]. Despite reported improvements in techniques and outcomes, 3COs remain complex and high-risk procedures, requiring careful consideration and meticulous planning [24]. Passias et al. [24] reported similar results in a study on 3COs, with a mean EBL of 1,506 mL and a mean operative time of 378 minutes, compared to the current cohort’s 1,453 mL and 420 minutes, respectively.
Several key factors were associated with EBL, including surgical techniques, older age, and the operating surgeon. 3COs were strongly associated with blood loss, likely due to extensive bony resection, cancellous body bleeding, and the time required [25]. Similar to prior literature [26], older age was predictive of increased EBL. The aging population is prone to developing comorbidities, which, along with decreased elasticity of blood vessels, affect hemostasis in older patients [27]. Moreover, EBL quartiles were significantly different among operating surgeons, indicating that surgeon technique plays a major role in EBL. Notably, a higher preoperative hemoglobin level was observed in the highest EBL quartile. This contradicts a previous study in which lower preoperative hemoglobin was associated with a higher risk of intraoperative blood transfusions [28], warranting further investigation.
Total instrumented levels emerged as the strongest predictor of increased operative time, consistent with previous studies [7]. This is logical, as more levels require additional time for instrumentation, decompression, and fusion. Similar to Passias et al. [7], older age was also predictive of increased operative time, possibly due to age-related challenges such as poor bone quality and increased tissue friability. The current study’s findings align with Daniels et al. [6], showing associations between longer operative time and higher EBL and longer LOS. However, unlike Daniels et al. [6], the present study found that longer operative times were associated with lower sacral slope, indicating a higher pelvic tilt, more compensatory pelvic retroversion, and a more severe deformity, suggesting certain sagittal alignments present unique surgical challenges requiring more time for optimal correction. A combined anterior-posterior approach was more common among patients in the highest operative time quartile, consistent with previous reports, likely due to the additional time required for repositioning and greater surgical invasiveness [29]. Patients in the highest operative time quartile also had higher rates of rod fractures, potentially related to the complexity of deformity correction or extended duration of rod contouring and stress during the procedure. Mechanical stress from rod contouring and notching can compromise rod integrity. Recent advancements, such as patient-specific rods and multi-rod constructs, may help decrease operative time and reduce rod fracture rates [30].
The presence of two or more comorbidities was identified as a strong predictor of increased LOS, highlighting the significant impact of patient health status on postoperative recovery. Patients in the highest LOS quartile had higher rates of CHF and hypertension, consistent with previous reports [8,31], emphasizing the role of cardiovascular comorbidities in prolonging hospital stay. Higher TIL was associated with longer LOS, consistent with previous findings that more extensive surgeries typically require longer recovery periods and are associated with a higher risk of complications [31]. Patients with longer LOS reported lower EQ5D QALY scores, suggesting a potential relationship between extended hospitalization and reduced quality of life. This association has been reported in elderly patients hospitalized for acute illness, which warrants further investigation in ASD surgeries. While 3COs were predictive of increased LOS in univariable analysis, this factor did not remain significant in the multivariable model, indicating that the extent of instrumentation may more substantially impact LOS than the type of osteotomy performed.
Some limitations of this study should be acknowledged. As a single-center retrospective study, the findings may not be generalizable to all ASD populations, and the study design inherently introduces the potential for selection bias and limitations in controlling for all confounding variables. The involvement of multiple surgeons, while reflective of real-world practice, adds variability in surgical technique and decision-making. Stratification by surgical team could have provided additional insights, but the uneven distribution of cases across surgeons would have resulted in underpowered comparisons. Future research should investigate the impact of attending surgeons on perioperative and postoperative outcomes. Our quartile-based approach to stratifying EBL, operative time, and LOS was designed to isolate extreme cases but resulted in uneven group sizes. Other strategies, such as median-split or continuous modeling, may provide different insights and should be considered in future studies. Additionally, the study’s outcomes may have been influenced by unmeasured factors such as anesthesia protocols and postoperative care pathways. Importantly, changes in postoperative radiographic measures were not analyzed. Since a greater degree of correction can significantly impact operative time, future studies should aim to evaluate postoperative radiographic changes, particularly focusing on L4–S1 lordosis. The study’s definition of clinically meaningful pelvic retroversion and PJK, while based on established criteria, might not capture all clinically relevant cases due to the complexity of spinal alignment and variability in surgeon interpretation. Finally, frailty, a recognized predictor of outcomes in elderly patients undergoing ASD surgery, was not assessed. Future studies should incorporate standardized measures of frailty to provide a more comprehensive understanding of outcomes in ASD surgery.
Conclusions
The current study highlights key risk factors contributing to perioperative morbidity in ASD surgery. Three-column osteotomy consistently emerged as the most impactful, influencing all three outcomes. Recognizing this central driver of adverse outcomes can help surgical teams develop more tailored approaches to ASD correction, enhancing patient safety and recovery.
Key Points
Three-column osteotomy predicted blood loss, operative time, and length of stay.
Three-column osteotomy and older age were the strongest predictors of excessive blood loss.
Highest operative time was predicted by higher total instrumented levels and older age.
Higher total instrumented levels primarily predicted prolonged length of stay.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
