INTRODUCTION
Since the publication of enhanced recovery after surgery (ERAS) program for posterior lumbar interbody fusion in 2021, the program has been increasingly applied to various spine surgeries [3]. A series of recovery programs have been shown to reduce hospital length of stay (LOS) and facilitate postoperative recovery [4,8]. Based on the principles of the ERAS program, we developed same-day surgery (SDS) and non-SDS protocols for intradural extramedullary (IDEM) spinal cord tumor surgery, and have implemented those protocols since 2022. In certain cases, preoperative admission is essential for adequate preoperative inpatient care, such as blood glucose management, intravenous (IV) steroid administration, and transfusion for blood, along with other necessary interventions [9]. However, preoperative admission is not required for all patients if they do not need specific preoperative inpatient care. Such cases may undergo surgery on the same day as admission and stay in the hospital postoperatively for continued monitoring and care, a process referred to as SDS. The SDS protocol has already been shown to reduce hospital LOS [7]. This finding thereby prompts consideration of whether SDS could represent a feasible and cost-effective management strategy for patients with IDEM spinal cord tumor who do not require preoperative inpatient care. This study aimed to validate the clinical safety and economic feasibility of our newly developed ERAS-based SDS selection protocol by comparing healthcare costs, clinical and surgical outcomes between SDS and non-SDS groups for IDEM spinal cord tumors using propensity score-matched (PSM) analysis.
MATERIALS AND METHODS
The Institutional Review Board of Asan Medical Center (2026-0912) approval and informed consent were waived due to the retrospective nature of the study and minimal risk to participants.
Study participants
A total of 189 patients underwent surgery for IDEM spinal cord tumors at single institution between 1 January, 2022 and October 31, 2024. The inclusion criteria were as follows : 1) age over 19 years; 2) diagnosis of IDEM spinal cord tumor; and 3) a follow-up period of at least 6 months. The exclusion criteria were : 1) interdepartmental surgeries involving inpatients from other departments; 2) emergency or urgent surgeries; 3) IDEM tumors extending into the intramedullary or extradural space; and 4) a diagnosis of significant spine disorders, such as spinal trauma or ankylosing spondylitis, which could influence outcomes (Fig. 1). Emergency or urgent surgeries referred to procedures performed earlier than the planned surgical date due to rapid neurological deterioration. After applying the inclusion and exclusion criteria, 173 patients were selected for analysis, of whom 58 underwent surgery following the SDS protocol, while 115 were managed using the non-SDS protocol. To ensure comparability between the two groups, PSM was performed in a 1 : 1 ratio based on six covariates, resulting in 48 matched pairs (n=96). This study was conducted in accordance with institutional guidelines.
Demographics
The clinical characteristics of the 48 PSM patients in both the SDS and non-SDS groups were assessed, including smoking history, alcohol use, hypertension, diabetes mellitus, renal disease, cardiac disease, cerebrovascular disease, and pulmonary disease. LOS was defined as the days from admission to discharge, counting the admission day as one day, regardless of admission time (Table 1). Smoking was defined as a history of smoking for prior 6 months, and alcohol use was defined as the intake of at least five drinks per week for prior 6 months. All patients were also classified based on the American Society of Anesthesiologists (ASA) physical status classification system [11]. Tumor locations were divided into three levels : cervical, thoracic, and lumbosacral. Tumor orientation was determined based on the dominant direction observed on axial magnetic resonance (MR) images and classified as anterior, posterior, or lateral. Laminoplastic laminotomy level refers to the vertebral level at which the lamina, previously removed for tumor resection, is reattached. Standardized clinical outcome scores were assessed at 1 month postoperatively to evaluate early recovery. However, all patients were monitored for a mean follow-up period of 11.1 to 13.7 months (Table 1).
Surgical procedures
All procedures were performed by a single surgeon using standard microsurgical techniques. We achieved Simpson grade II for meningiomas, preserved functional nerve roots under intraoperative neuromonitoring for schwannomas, and targeted gross total resection for ependymomas. Postoperative care followed a standardized ERAS protocol, including mobilization within 24 hours and multimodal analgesia.
Surgical and clinical outcomes
Surgical outcomes included operative time, estimated blood loss (EBL), residual tumor, and postoperative complications such as hematoma, wound dehiscence, neurological deterioration due to reperfusion injury, and cerebrospinal fluid (CSF) leak, all of which required additional interventions when they occurred. These outcomes were evaluated through a review of medical records and routine postoperative magnetic resonance imaging (MRI) which was performed on postoperative day 1 (POD 1) and at 3 months after surgery. Additionally, unscheduled imaging studies were promptly obtained in cases where new or worsening neurological symptoms developed during the postoperative period. Clinical outcomes were evaluated using the Visual analog scale (VAS), the modified McCormick scale, the American Spinal Injury Association (ASIA) impairment scale, the Karnofsky performance status (KPS) scale, and the Eastern Cooperative Oncology Group (ECOG) performance status, patient-reported surgical satisfaction using a 5-point Likert scale, and readmission within 30 days. The VAS, the modified McCormick scale, the ASIA impairment scale were evaluated preoperatively and at one month postoperatively. For statistical analysis, ASIA impairment scale was converted to a continuous variable by assigning 1 point per grade in reverse order, such that grade E was scored as 1 and grade A as 5. Changes in the scores from preoperative to 1-month postoperative period were analyzed as continuous variables to compare the SDS and non-SDS groups. In the 5-point Likert scale, a score of 1 indicated “very dissatisfied,” 2 indicated “dissatisfied,” 3 indicated “neutral,” 4 indicated “satisfied,” and 5 indicated “very satisfied” (Table 2) [6].
Healthcare costs
Healthcare costs were determined based on detailed medical billing statements (Table 3) and are presented in US dollars (USD), converted from Korean Won (KRW), using the purchasing power parity (PPP) rates published by the World Bank [1]. A portion of the total hospital costs was covered by the National Health Insurance, while the remaining amount was paid by patients as out-of-pocket (OOP) expense. Surgery and procedure costs encompass not only surgical procedure fees but also surgery-related interventions. Anesthesia fees represent the professional charges associated with providing anesthesia care during surgery, including anesthetic induction, maintenance, airway management, and continuous intraoperative monitoring. Surgical and anesthetic material costs cover the disposable supplies and medications used during surgery and anesthesia. Laboratory and imaging costs included the total expenses for laboratory tests and imaging studies performed preoperatively, intraoperatively, and postoperatively to evaluate the lesion. Pharmaceutical costs reflect the expenses for medications administered during hospitalization, encompassing both oral and parenteral drugs. Facility maintenance costs accounted for charges related to the use and maintenance of hospital infrastructure, separate from clinical services provided to patients, including charges for safety management, rapid response systems, and meals.
Statistical analysis
R software version 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria) was used for all statistical analyses. Categorical variables were presented as frequencies and percentages, whereas continuous variables were reported as means and standard deviations (SDs). Cliff’s delta was used to quantify the magnitude of healthcare costs differences between two groups. Categorical variables were compared using Pearson’s chi-square test or Fisher’s exact test based on whether the expected cell count was below the threshold of 5. Continuous variables were evaluated using Welch’s t-test or the Mann-Whitney U-test. The Mann-Whitney U-test was applied when the data exhibited a non-parametric distribution, while Welch’s t-test was used for parametric distributions with unequal variances. Statistical significance was defined at p<0.05. To address potential ceiling effects and baseline imbalances in neurological and functional status, an analysis of covariance (ANCOVA) was performed for all functional outcome measures, including the modified McCormick scale, ASIA impairment scale, KPS scale, and ECOG status. In this model, postoperative functional scores were compared between groups using their respective preoperative values as covariates (Table 4).
PSM analysis
A greedy nearest-neighbor matching algorithm was employed to match 173 patients in a 1 : 1 SDS to non-SDS ratio by six variables : sex, age, tumor level, pathological findings, tumor location relative to the spinal cord, and the level of laminoplastic laminotomy. These covariates were carefully selected based on their established or potential association with surgical complexity, tumor pathology, and surgical and clinical outcomes to minimize selection bias and ensure balanced comparison between SDS and non-SDS groups in the PSM process. Specifically, preoperative functional status (e.g., ASA class and neurological grades) was intentionally excluded from the PSM variables. Because the primary goal of this study was to validate the safety of our SDS selection protocol and to evaluate the cost differences between the SDS and the non-SDS groups. Therefore, matching for these variables would have defeated the purpose of testing the triage system. Instead, we focused on matching surgical complexity and pathological factors to ensure that the surgical procedures themselves were technically comparable between the groups. The adequacy of the PSM analysis was assessed using three measures : 1) reevaluation of six covariates using Fisher’s exact test or Pearson’s chi-square test for categorical variables, and the Wilcoxon signed-rank test for continuous variables to confirm that there were no significant differences between the two groups; 2) visual assessment of the propensity score distribution before and after matching using kernel density plots (Fig. 2); and 3) setting the caliper width at 0.25 times the SD of the propensity scores (caliper value, 0.024) to avoid excessive imbalance during matching (Fig. 1).
RESULTS
Patient demographics
After PSM, 48 patients were included in both the non-SDS and SDS groups. The mean patient age in the non-SDS and SDS groups was 53.0 and 51.1 years, respectively. Each group included 21 male patients. The LOS was significantly shorter in the SDS group (4.1±4.1 days) compared to the non-SDS group (5.6±4.6 days) (p<0.05). There were no significant differences between the groups in terms of smoking history, alcohol use, hypertension, diabetes mellitus, renal disease, cardiac disease, cerebrovascular accident, and pulmonary disease. According to the ASA physical status classification, 45 patients (93.8%) in the SDS group were classified as class II, and no patients (0.0%) were classified as class III, whereas 45 patients (93.8%) in the non-SDS group were class III (Table 1). Histopathological analysis revealed that meningiomas were diagnosed in 19 patients in the non-SDS group and 17 in the SDS group. Schwannomas were observed in 18 patients in the non-SDS group and 20 patients in the SDS group, while myxopapillary ependymoma was identified in 11 patients in both groups. Tumor orientation relative to the spinal cord was categorized as anterior, posterior, or lateral. Anteriorly located tumors were found in 11 patients in the non-SDS group and 12 patients in the SDS group, posterior in 17 and 16 patients, respectively, and lateral orientation in 20 patients in each group. Laminoplastic laminotomy at onelevel was performed in 16 patients in both groups. Two-level laminotomies were performed in 15 non-SDS and 16 SDS patients, while three-level procedures were carried out in 17 and 16 patients, respectively. The mean follow-up period was 13.7±1.4 months for the non-SDS group and 11.1±1.1 months for the SDS group (Table 1).
Clinical and surgical outcome
The mean operation time was 226.1±85.5 minutes for the non-SDS group and 210.0±59.9 minutes for the SDS group. The EBL was 378±112 mL for the non-SDS group and 358±91 mL for the SDS group, with no statistically significant difference between the groups. Postoperative hematoma was observed in one patient in the non-SDS group and two patients in the SDS group. Wound dehiscence occurred in one patient in the SDS group, with no cases in the non-SDS group. Neurological deterioration due to reperfusion injury and CSF leak occurred in one patient each. Follow-up MRI performed at 3 months postoperatively confirmed no delayed surgical complications, such as late-onset CSF leaks or pseudomeningoceles, in either group.
Preoperatively, the mean VAS score was 7.0±1.5 in the non-SDS group and 7.4±0.8 in the SDS group (p=0.163). At 1 month postoperatively, the mean VAS score decreased to 1.3±1.1 and 1.0±1.0, respectively (p=0.214). The degree of improvement in VAS scores from the preoperative to the 1-month postoperative period was 5.8±1.7 in the non-SDS group and 6.4±1.3 in the SDS group, with no statistically significant difference between the two groups (p=0.129). Due to the criteria used to allocate patients into the SDS and non-SDS groups, significant differences in modified McCormick scale scores and the ASIA Impairment scale scores were observed between the two groups. Preoperatively, the modified McCormick scale scores were 3.2±0.5 in the non-SDS group and 4.8±0.4 in the SDS group (p=0.002), while the ASIA impairment scale scores were 2.3±0.5 and 3.2±0.3, respectively (p=0.001). Patients in the SDS and non-SDS groups were stratified by ASIA, ASA, and the modified McCormick scale, which revealed significant differences in their preoperative status. Clinically, the non-SDS group often presented with advanced myelopathy and motor weakness, whereas the SDS group was characterized by radicular pain or mild sensory symptoms with preserved mobility.
At 1 month postoperatively, the modified McCormick scale scores were 2.9±0.7 in the non-SDS group and 4.3±0.5 in the SDS group (p=0.003), and the ASIA impairment scale scores were 1.7±0.5 and 2.5±0.5, respectively (p=0.002). Nonetheless, the improvements from the preoperative to the 1-month postoperative period were similar in both groups. The modified McCormick scale improved by 0.4±0.6 in the non-SDS group and 0.5±0.5 in the SDS group (p=0.410), while the ASIA impairment scale score increased by 0.6±0.7 and 0.7±0.5, respectively (p=0.394). Surgical satisfaction was high in both groups, with mean Likert scores of 4.9 and 4.8 in the non-SDS and SDS groups, respectively, showing no statistically significant difference. There were no readmissions within 30 days of discharge in both groups (Table 2). To further validate these findings independent of baseline status, an adjusted analysis using ANCOVA was performed. The results revealed no significant differences between the SDS and non-SDS groups in adjusted postoperative scores for the modified McCormick scale (p=0.378), ASIA impairment scale (p=0.163), KPS scale (p=0.860), and ECOG status (p=0.512), confirming comparable functional recovery across both protocols (Table 4).
Healthcare costs
The average total hospital cost was significantly higher in the non-SDS group (10940±770 USD) compared to the SDS group (9740±2370 USD) (p<0.001; Cliff’s delta, 0.565). There were no significant differences between the groups in terms of OOP expense, surgery and procedural costs, anesthesia fees, surgical and anesthetic material costs, laboratory and imaging costs, pharmaceutical costs. However, a significant difference was found in the facility maintenance costs, which were higher in the non-SDS group (1470±1060 USD) compared to the SDS group (870±940 USD) (p<0.001; Cliff’s delta, 0.571).
DISCUSSION
Preoperative assessment is undertaken to evaluate the feasibility of performing IDEM spinal cord tumor surgery under general anesthesia. In certain cases, preoperative inpatient care is necessary prior to surgery, warranting hospital admission before the scheduled surgical date [10]. Examples of preoperative inpatient care before surgery include blood transfusions for hematologic imbalances, IV fluid therapy for nutritional deficits, as well as other medical conditions. However, patients who do not require such preoperative inpatient care do not need to be admitted before surgery, and the surgery can be performed on the day of admission, a protocol referred to as SDS. Since the establishment of the strong recommendations for preoperative, intraoperative, and postoperative care for lumbar fusion by the ERAS society in 2021, these protocols have been increasingly recognized as applicable not only to lumbar fusion but also to spinal tumor surgeries [2]. In 2022, based on the strong recommendations of ERAS program, we developed and implemented SDS and non-SDS protocols in the management of IDEM spinal cord tumors (Fig. 3].
Although baseline neurological status differed between the SDS and non-SDS groups before surgery, the extent of neurological improvement after surgery did not show a statistically difference between the two groups. These findings reflect that ERAS-based SDS and non-SDS protocols yielded not only comparable surgical outcomes but also clinical outcomes, including pain relief, functional recovery, complication rates, patient satisfaction, and short-term readmission (Table 2).
Cost components, including OOP expense, surgery and procedure costs, anesthesia fees, surgical and anesthetic material costs, laboratory and imaging costs, pharmaceutical costs, were comparable between the two groups (Table 3). This result is further supported by negligible to small effect sizes in Cliff’s delta analyses, with 95% confidence intervals including zero across all components, indicating a lack of statistically significant differences. The absence of cost differences in surgery, procedures, and anesthesia indicates that surgical and anesthetic management was applied uniformly in both groups. Although the non-SDS group incurred higher pharmaceutical costs due to preoperative medical preparation, this increase was subsequently offset by the same application of the standardized ERAS-based analgesia pathway after surgery, resulting in no overall difference between the groups. This pathway includes early transition to oral agents, opioid-sparing strategies, and protocolized multimodal analgesia. Total hospital costs were 12.3% lower in the SDS group, with the most substantial reduction observed in facility maintenance costs, which were 69.1% lower than those in the non-SDS group (Table 3). These costs, typically charged on a per diem or per night basis, include the standard facility fee as well as expenditures related to safety management, rapid response systems, meals, and admission services. Higher facility costs in the non-SDS group reflect the intensive nursing care and monitoring required for patients with severe neurological deficits, unlike the functionally independent SDS cohort. As they are directly influenced by the duration of hospitalization, shortening the LOS can result in significant cost savings. In fact, the SDS group had an average LOS that was 1.5 days shorter than that of the non-SDS group, which likely contributed to the observed difference in total hospital costs (Table 1).
Patients were selected for SDS based on their preoperative health status and medical conditions (Fig. 3). Patients with poor ASA physical status, ASIA impairment scale, or modified Mc-Cormick scale scores were considered to have a higher risk of perioperative complications. They required closer observation before and after surgery and were therefore not suitable for same day admission surgery. It is important to emphasize that SDS represents a highly selective, protocol-driven strategy applied to carefully chosen patients with stable general and neurological status. Therefore, the observed reduction in LOS and hospital costs should be interpreted as the clinical outcome of this specific low-risk cohort rather than a universally applicable result for all patients with IDEM spinal tumors.
The fundamental design of this study was aimed at validating the safety and cost-effectiveness of our newly developed SDS selection protocol in real-world practice. By intentionally selecting candidates based on their preoperative status, we sought to determine if this protocolized triage could successfully identify patients suitable for same-day admission without increasing perioperative risks. Consequently, while this represents a selective, low-risk analysis, it provides a valuable model for safe and efficient resource optimization in spinal tumor surgery.
We present the ERAS-based SDS and non-SDS protocols implemented at our institution (Fig. 3). Patients who presented with an ASA physical status classification of ≥3, or a modified McCormick grade of ≥4, or an ASIA impairment scale grade of ≥3 were deemed ineligible for SDS. In addition, those who required preoperative IV injection were also considered suitable candidates for the non-SDS group. In the SDS protocol, surgery is performed on hospital day (HD) 1, while in the non-SDS protocol, it is performed on HD 2. All patients receive preoperative prophylactic antibiotics within 30 minutes prior to skin incision, with intraoperative redosing every 4 hours. Neither group requires the placement of a surgical drain following surgery, and patients in both groups begin sips of water 6 hours after extubation. Postoperative management is conducted from POD 1 to POD 3. Ambulation begins within 24 hours postoperatively. The Foley catheter is removed once toilet ambulation is feasible, or within 12 hours regardless of mobility to assess bladder function through post-void residual urine. Postoperative nausea and vomiting prophylaxis is determined based on Apfel score assessment, with rescue antiemetic therapy administered as needed [5]. Consequently, patients in the SDS group are typically discharged on HD 4, while those in the non-SDS group are discharged on HD 5.
Limitation
Although propensity score matching was applied, an inherent selection bias remains as the SDS protocol was intentionally applied only to patients deemed clinically suitable (ASA <3 and lower neurological severity). This selection process itself may have contributed to more favorable outcomes in the SDS group. Clinical outcome scores were primarily evaluated at one month postoperatively to focus on the immediate effects of the SDS protocol. While our average safety monitoring exceeded 1 year without significant delayed complications, this short-term assessment of functional scores may not fully capture the complete trajectory of long-term neurological recovery. The singlecenter design may limit generalizability, particularly to lowervolume institutions or healthcare systems with different reimbursement structures. Cost analyses were based on PPPadjusted USD derived from Korean billing data, which may not fully reflect real-world economic conditions in other countries.
CONCLUSION
Our validated SDS selection protocol represents a safe and cost-effective alternative for carefully screened patients who do not require preoperative preparation and demonstrate favorable systemic and neurological status. As healthcare systems increasingly seek to optimize resource utilization and enhance patient care, the SDS protocol for IDEM spinal cord tumors may serve as a valuable model for standardized and efficient perioperative management.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Print 
