INTRODUCTION
Degenerative lumbar spine pathologies, including lumbar disc herniation and spinal stenosis, are among the leading causes of chronic low back pain and disability worldwide, particularly in aging populations [1]. In Singapore, as in many developed countries, the burden of lumbar degenerative disease is expected to rise with increasing life expectancy [1]. Surgical intervention becomes necessary when conservative treatments fail, especially in cases of refractory radiculopathy, neurogenic claudication, or functional impairment.
In recent decades, the field of spine surgery has witnessed a paradigm shift toward minimally invasive techniques, driven by the goal of achieving effective neural decompression while minimizing collateral tissue damage [2]. Endoscopic spine surgery (ESS) has emerged as a significant advancement in this context. Unlike traditional open or microscopic techniques, ESS leverages high-definition endoscopic visualization and percutaneous access to perform decompression with reduced soft tissue trauma, shorter hospital stays, and faster postoperative recovery [3,4].
ESS encompasses various techniques, including full-endoscopic interlaminar and transforaminal approaches as well as the increasingly adopted unilateral biportal endoscopic (UBE) technique. UBE decompression, in particular, offers bimanual instrumentation with endoscopic magnification, providing enhanced control and visualization comparable to open surgery, while retaining the benefits of minimally invasive access [5]. Originating in South Korea, the UBE technique has been widely disseminated across Asia and adopted in countries like Vietnam, China and increasingly, Western institutions [6].
Despite its promising benefits, ESS is associated with a well-recognized learning curve. Surgeons must acquire new psychomotor skills and adapt to a distinct operative environment that includes endoscopic navigation, continuous saline irrigation, and limited tactile feedback. Literature suggests that proficiency in UBE decompression may require around 35 cases, depending on the surgeon’s prior experience with arthroscopy or minimally invasive techniques [7,8]. Moreover, early in the learning curve, longer operative times and increased risk of complications such as dural tears or incomplete decompression have been reported [7,8]. In addition, within the Singapore public hospital setting, ESS incurs a price premium due to the additional imaging and equipment required.
While there is robust data from Vietnam, China, and South Korea demonstrating the efficacy and safety of ESS in treating lumbar pathologies [9-11], clinical outcomes in the Southeast Asian context—particularly Singapore—remain underreported. Singapore’s healthcare system is uniquely situated at the intersection of high-resource medical infrastructure and a diverse, aging population with increasing demands for fast recovery and cost-efficient care. To date, there is a paucity of data examining the outcomes of ESS performed across multiple surgeons within a single institution in Singapore. With ongoing advancements in healthcare funding and technological adoption across the region and globally, it is expected that more institutions will confront the decision of introducing UBE into their surgical programmes. This transition inevitably raises key questions: whether the technique is clinically efficacious in their patient population, whether it can be implemented safely, and how best to structure training and support for surgical teams adopting a new endoscopic platform.
This study aims to fill that gap by analyzing the early clinical outcomes of patients who underwent ESS—specifically the UBE approach for lumbar decompression/discectomy.
The objectives of this study are threefold:
(1) To describe the demographic and clinical characteristics of patients undergoing UBE surgery for lumbar decompression/discectomy at a tertiary hospital in Singapore.
(2) To evaluate early clinical outcomes of patients undergoing ESS in a single-center setting involving multiple lead surgeons, and benchmark these outcomes against international standards.
(3) To investigate the clinical learning curve associated with the adoption of UBE surgery among surgeons at our institution.
We hypothesize that ESS in our setting will demonstrate outcomes comparable to global benchmarks and that modifiable patient factors and surgical experience will meaningfully influence recovery trajectories.
MATERIALS AND METHODS
1. Study Design and Setting
This is a retrospective, single-center cohort study conducted at Sengkang General Hospital (SKH), a public tertiary hospital in Singapore under the SingHealth cluster. SKH serves the northeastern region of Singapore and functions in collaboration with national tertiary centers such as Singapore General Hospital and Changi General Hospital. ESS was formally introduced into routine surgical practice at SKH in late 2022.
2. Inclusion Criteria
• Symptoms ≥ 6 weeks
• Diagnosis based on clinical history, physical exam, and imaging
• Lumbar spine pathologies only
• Absence of clinical signs suggestive of lumbar instability
• No evidence of significant infection or spinal fracture
• Minimum of 1-month postoperative follow-up
3. Exclusion Criteria
• History of spinal trauma (n=2)
• Spinal surgery cases that included the use of instrumentation (e.g., screws, rods and cages) (n=4)
• Admissions for non-spinal medical conditions with subsequent referral for spinal surgery (n=1)
• Prolonged hospitalization due to unrelated medical issues (e.g., respiratory or genitourinary infections) (n=2)
• Incomplete records or missing follow-up data (n=59)
• The methodological workflow is illustrated in Figure 1.
4. Data Collection
This study was approved by the SingHealth institutional review board (ECOS Ref No. 2020-2381) prior to study commencement. Patients were informed preoperatively regarding the potential use of anonymized clinical data for research purposes. Data were analyzed retrospectively and deidentified prior to publication, in compliance with institutional ethical standards.
A comprehensive review of electronic medical records was performed to extract and categorize data across 4 domains: patient demographics and comorbidities, preoperative clinical and imaging findings, intraoperative details, and postoperative outcomes.
1) Patient demographics and comorbidities
• Age, sex, ethnicity, body mass index (BMI), smoking status
• Medical history including diabetes mellitus, hypertension, hyperlipidemia, coagulopathy, autoimmune diseases, prior stroke, anticoagulant use, and history of primary or metastatic spinal cancer
• History of previous surgeries, including prior spine surgeries (number, pathology, spinal levels involved, surgical approach)
2) Presenting symptoms and functional status
• Symptom duration (months) and distribution (back pain, radiculopathy, weakness) categorized by spinal regions (cervical, thoracic, lumbar, sacral) with corresponding myotome or dermatome levels
• Laterality (unilateral right/left or bilateral)
• Functional status including independence in activities of daily living and bedbound status
3) Preoperative imaging and clinical evaluation
• Imaging modalities utilized (magnetic resonance imaging [MRI], computed tomography), regions scanned (cervical, thoracic, lumbosacral, whole spine)
• Detailed imaging findings, including affected spinal levels, presence of spondylolisthesis, spondylosis, radiculopathy, spinal fractures, and neoplastic involvement
4) Operative data
• Date and duration of operation, operating surgeon, assistants, and total surgical team size
• Surgical technique details: type of procedure (e.g., discectomy, laminectomy, facetectomy), spinal levels operated on
• Use of navigational tools (electromagnetic navigation, fluoroscopy)
• Intraoperative parameters including anesthesia type, intraoperative complications (e.g., dural tear, nerve injury), blood loss, nerve monitoring usage, and conversion to open surgery
5) Postoperative outcomes
• Immediate postoperative course including intensive care unit admission, time to ambulation, hospital length of stay, pain scores
• Neurological status: new or improved sensory and motor deficits categorized by spinal region and myotome/dermatome level
• Functional recovery metrics such as spinal range of motion and physiotherapy assessment scores
• Longitudinal follow-up data including pain scores (VAS) and quality of life assessments (SF-36) at baseline (preoperation), 1 month, 3 months, 6 months, and 12 months postoperatively
Given the short postoperative hospital stay (mean=1.38 days), a proportion of patients did not attend subsequent outpatient follow-up, likely reflecting satisfactory early recovery, absence of recurrent symptoms, and minimal need for further rehabilitation. To address this, supplementary follow-up was conducted via telephone consultations. These calls were performed with prior informed consent and utilized a standardized questionnaire that included confirmation of operative details to verify patient identity and surgical procedure, as well as assessment of postoperative outcomes such as the SF-36 health survey and VAS pain scores. Multiple attempts were made to contact patients or their caregivers at different times and days to maximize response rates and data completeness.
For patients who remained unreachable despite repeated efforts, their data were classified as lost to follow-up. Considering variability in follow-up adherence—where some patients completed assessments beyond 1 month but not up to 12 months—outcome measures were analyzed separately at predefined postoperative intervals (1 month, 3 months, 6 months, and 12 months). This method enabled inclusion of the maximum number of patients with available data at each time point, thereby minimizing data loss and improving the robustness and representativeness of longitudinal outcome analyses. Patients lost to follow-up at later intervals were excluded from those specific analyses but included in earlier time-point analyses where data were available.
5. Statistical Analysis
Data were analyzed using R. Continuous variables are reported as mean±standard deviation. Pre- and postoperative comparisons were made using paired t-tests. Pearson correlation assessed relationships between patient factors and outcome measures. A p-value <0.05 was considered statistically significant.
6. Surgical Technique and Perioperative Management
All surgical procedures were conducted at a single tertiary institution by 4 consultant-level lead spine surgeons. The operating surgeons had completed formal orthopedic training and obtained Fellowship of the Royal College of Surgeons certification, with a minimum of 15 years of overall clinical practice. Their prior experience included several years of independent spine surgical practice, with established familiarity in minimally invasive spinal techniques such as microscopic decompression and tubular approaches. This surgeon profile reflects real-world adoption of UBE by conventionally trained spine surgeons transitioning from established minimally invasive techniques, rather than outcomes derived from a single high-volume endoscopic specialist. While minor inter-surgeon technique variation may be present, all operations adhered to a standardized protocol for UBE decompression. Patients were positioned prone under general anesthesia. Intraoperative fluoroscopic guidance was employed to confirm level and trajectory, supplemented by continuous neuromonitoring to enhance neural safety throughout the procedure.
Portal placement strategy was standardized according to the side of pathology. For unilateral decompressions, left-sided and right-sided pathologies followed internally consistent but mirrored portal configurations.
For left-sided decompressions, both the viewing and working portals were placed to the left of the midline. In the medial-lateral plane, the portals were positioned between the spinous process and the left medial pedicular line. Cranio-caudally, the viewing portal was placed cranial to the intervertebral disc, while the working portal was placed caudal to the disc. The 2 portals were typically spaced approximately 3–4 cm apart.
For right-sided decompressions, both portals were placed to the right of the midline, between the spinous process and the right medial pedicular line. In contrast to left-sided cases, the cranio-caudal orientation was reversed: the viewing portal was positioned caudal to the disc space, and the working portal was placed cranial to the disc space.
For bilateral decompressions, a unilateral left-sided approach was used, with portal placement identical to that described for left-sided decompressions.
The working portal was established approximately 2 cm caudal to the targeted intervertebral disc level, and the viewing portal approximately 1 cm cranial to the disc level. Skin incisions were made at the intersection of the spinolaminar junction and medial pedicular line under fluoroscopic guidance. A 0° arthroscope was introduced through the cranial portal. Saline irrigation was facilitated via a gravity-fed system, with the height of the saline reservoir dynamically adjusted intraoperatively to optimize visualization and reduce epidural pressure. Pressure pumps were deliberately avoided due to concerns regarding increased risk of neural compromise, particularly neuropraxia.
Initial access to the operative field was established using serial dilators, following which a radiofrequency (RF) wand was used to perform soft tissue dissection and to dock the endoscopic cannula against the lamina at the Spino-laminar junction. Hemostasis during this phase was meticulously achieved with bipolar RF coagulation to maintain a clear visual field and reduce intraoperative bleeding.
Laminotomy and decompression followed a structured sequence. For example, in an L4–5 decompression, the procedure began with an inferior laminotomy of the cranial vertebra (caudal edge of the L4 lamina), followed by a medial facetectomy. Subsequently, a superior laminotomy of the caudal vertebra (cranial edge of the L5 lamina) was performed. To minimize the risk of iatrogenic instability, the lateral extent of medial facetectomy was limited to the level of the medial pedicular wall, allowing adequate pedicle-to-pedicle decompression of the traversing nerve root while preserving the lateral facet complex and pars interarticularis. This extent of facet resection has been shown to maintain segmental stability [12]. Patients with markedly sagittal oriented facet joints, which may predispose to postoperative instability, were identified preoperatively on imaging and managed with instrumented fusion rather than decompression alone.
Bony decompression was achieved using a high-speed burr. Throughout this phase, the ligamentum flavum was preserved to serve as a protective barrier over the dura until decompression was complete. Once bone work concluded, the ligamentum flavum was excised en bloc or in 2 layers if hypertrophic. The flavectomy sequence involved midline splitting of the ligament, followed by sequential detachment along its caudal, lateral, and cranial margins. Care was taken to delineate a clear plane between the dura and the flavum using blunt instruments such as nerve hooks and Penfield dissectors to prevent dural injury.
Following bony decompression and flavectomy, discectomy was performed only when indicated, and always after neural decompression was confirmed. The dural sac was gently retracted medially with a nerve root retractor. Disc removal was performed using pituitary forceps. No internal decompression or annuloplasty was routinely performed to minimize iatrogenic instability. Completion of decompression was confirmed intraoperatively by observing free mobilization of both the traversing nerve root and dural sac using a Penfield dissector.
For multilevel decompression, previously established working portals were repurposed as viewing portals for subsequent caudal levels, minimizing the need for additional incisions. Final hemostasis was achieved using bipolar RF and application of a hemostatic matrix (Floseal, Baxter) left in situ for approximately 3 minutes. Placement of drainage catheters was not routinely required. Wound closure was performed with non-absorbable Prolene® sutures in a simple interrupted fashion at the dermal level.
7. Postoperative Care and Follow-up Protocol
Patients were advised on strict activity modification postoperatively. Formal physiotherapy was deferred for the first 6 weeks. Patients were instructed to avoid forward flexion, axial rotation, and heavy lifting for a period of 3 months. Routine postoperative imaging was not performed in the absence of clinical indications suggestive of complications or recurrence, in accordance with institutional guidelines.
8. Complication Management and Intraoperative Considerations
Neural protection was prioritized through the use of intraoperative neuromonitoring and adherence to strict surgical visualization protocols. The risk of dural or nerve injury was mitigated by preserving the ligamentum flavum during bony work, establishing a clear tissue plane prior to flavectomy, and avoiding excessive mechanical manipulation. Intraoperative bleeding was minimized through early and methodical use of RF coagulation, while impaired visualization due to bleeding or debris was addressed via saline irrigation through bidirectional portal flow.
In the event of an incidental durotomy, TachoSil fibrin sealant patches were applied in an inlay fashion to ensure watertight closure. Patients with dural tears were placed on 24-hour bed rest postoperatively and monitored closely during early follow-up for signs of cerebrospinal fluid leakage or pseudomeningocele formation.
RESULTS
Patient demographics and clinical characteristics are detailed in Table 1. The mean age of the cohort was 56.8±14.9 (range, 20–85) years, with a slightly higher proportion of females (54.1%) compared to males (45.9%). The majority of patients were of Chinese ethnicity (75.7%), followed by Malay (15.3%), Indian (1.8%), and other ethnic groups (7.2%). The mean BMI was 26.9±4.9 kg/m2, classifying the cohort within the overweight category based on Asian BMI standards. Most patients were nonsmokers (85.6%). Common comorbidities included hypertension (50.4%), hyperlipidemia (48.6%), and diabetes mellitus (17.1%).
Preoperative presenting symptoms and operative characteristics are summarized in Table 2. Radiculopathy was the most common presenting symptom, reported in 71 patients (64.0%), followed closely by back pain in 70 patients (63.1%). Other presenting symptoms included leg pain in 39 patients (35.1%), neurogenic claudication in 22 patients (19.8%), leg numbness in 21 patients (18.9%), and leg weakness in 10 patients (9.0%). The most frequently operated levels were L4–5 (63.1%) and L5–S1 (29.7%), followed by L3–4 (9.9%) and L2–3 (3.6%). Most procedures were single-level decompressions (94.6%), and were performed on the left side in 39.6% of cases, the right side in 34.2%, and bilaterally in 26.1%.
Patients demonstrated significant improvement in both back and leg pain following ESS, as reflected in patient-reported outcome measures (PROMs) in Table 3. Mean VAS reductions for leg pain peaked at 3 months (4.18 points, p<0.001) and remained statistically significant through 12 months. Similarly, back pain showed significant improvement at all postoperative time points, with the greatest reduction observed at 3 months (3.07 points, p<0.001).
Corresponding improvements were also noted across multiple SF-36 domains (Table 4), particularly in physical functioning, role physical, and bodily pain, all of which exhibited statistically significant gains at 3 and 6 months postoperatively (p<0.001). These findings suggest that ESS not only delivers sustained pain relief but also facilitates early functional recovery and enhances overall quality of life as captured by PROMs.
Four intraoperative dural tears occurred (3.6%), all managed nonsurgically with no long-term sequelae. No conversion to open surgery was noted. There was no requirement for subsequent reoperation in these patients. Mean length of hospital stay was 1.38 days; 93.7% of patients were discharged within 48 hours.
A quantitative assessment of operative time relative to the cumulative case number was performed to characterize the learning curve of a single surgeon performing UBE ESS (Figure 2). Operative time analysis was restricted to a single surgeon’s first 80 single-level UBE ESS cases to minimize variability from differing start times and prior endoscopic experience among surgeons. This approach ensured a more consistent evaluation of the learning curve.
Linear regression analysis demonstrated a statistically significant negative correlation between operative time and case sequence, with the model defined as: y=-0.61 x + 178.826 where, y is operative time in minutes and x is the cumulative case number. The correlation coefficient was weakly negative (r=-0.262), and the coefficient of determination was low (R²=0.069), indicating that approximately 6.9% of the variability in operative time is explained by surgical experience alone (p=0.0187). This suggests additional factors influencing operative duration.
To evaluate external variables potentially influencing operative time, regression analyses were performed for patient age and BMI against operative time. Age demonstrated no significant association with operative time (p=0.155), whereas BMI showed a statistically significant correlation (p=0.023), indicating that patient BMI may be a confounding factor affecting operative duration independently of surgical experience.
Further subgroup analysis comparing mean operative times between early case cohorts and their respective remainders revealed statistically significant reductions in operative time after the initial 20 cases (Table 5). The mean operative time for the first 20 cases was 173 minutes (n=20), significantly longer than the 148 minutes observed in the remaining 60 cases (p=0.048). This trend persisted with increased sample size: the first 30 and 40 cases had mean operative times of 169 and 171 minutes, respectively, compared to 145 and 137 minutes in the remainder, with p-values of 0.037 and 0.005, respectively. The first 10 cases showed a trend toward longer operative times (mean 170 minutes) compared to the remainder (152 minutes), though this did not reach statistical significance (p=0.192).
DISCUSSION
Our results indicate that ESS via the UBE technique produces excellent clinical outcomes when used in the management of lumbar degenerative disease. Significant and sustained postoperative improvements in pain and function were documented, paralleling outcomes reported in regional studies from Vietnam and South Korea [9,11]. Patients reported substantial decreases in VAS scores from baseline for both back and leg pain at all follow-up intervals up to 12 months. Likewise, SF-36 assessments conducted at multiple time points revealed statistically significant improvements across multiple domains. These include physical function, role limitations due to physical health, bodily pain, vitality and social function.
The complication profile was limited and consistent with previous studies that primarily highlight a risk of intraoperative dural tears [13,14]. Early mobilization and discharge were achieved in the majority of cases, favoring the viability of UBE in short-stay and ambulatory surgical settings when compared to open decompression—an increasingly relevant consideration in Singapore’s value-driven care model.
As discussed in our results, the operative time trends demonstrated a clear learning curve with progressive reductions as case volume increased—particularly after the first 20 to 30 cases. However, external patient factors such as BMI contributed to variability in operative time, emphasizing the importance of these variables in surgical performance assessment and operative planning.
These findings align with prior studies documenting the UBE ESS learning curve. Liu et al. [7] reported that the learning curve plateaued after 37.5 cases, showing progressive reductions in operative time and complication rates with increasing surgeon experience. Similarly, Peng et al. [8] observed a significant decline in operative times after approximately 32 cases, emphasizing the importance of case volume in achieving proficiency. BMI has also been documented to influence operative time variability beyond surgical experience [15,16].
This study has several limitations inherent to its retrospective, non-randomized design. First, there was no control or comparison group undergoing microscopic or tubular decompression, and therefore no conclusions regarding comparative efficacy can be drawn. This limitation will be addressed in future series from our institution incorporating comparative cohorts.
Second, a formalized selection algorithm defining eligibility for UBE versus alternative decompression techniques was not employed, particularly during the early adoption phase. It is plausible that patients with more favorable anatomy or fewer comorbidities were preferentially selected initially, as is common in the introduction of novel surgical techniques. However, this pragmatic selection approach reflects real-world clinical practice and is consistent with prior learning-curve studies in minimally invasive spine surgery, where early case selection prioritizes patient safety and procedural familiarity. Importantly, the primary aim of this study was not to establish superiority over existing techniques, but to evaluate the feasibility, safety, patient-reported outcomes, and learning curve associated with UBE implementation in a regional hospital setting.
Additionally, the single-center nature of the study and the absence of routine postoperative MRI or other objective radiological endpoints limited detailed anatomical outcome assessment and necessitated reliance on patient-reported measures. Nonetheless, this study provides meaningful insight into intermediate-term (≥1 year) patient-reported outcomes following the adoption of UBE in a multisurgeon tertiary care environment.
The integration of comprehensive PROMs with detailed perioperative and operative metrics provides clinically relevant insight into procedural feasibility, safety, and learning-curve characteristics within a real-world practice environment. The inclusion of multiple surgeons strengthens the external validity of the findings by reflecting outcomes across varying levels of surgical experience, rather than those of a single expert operator. In addition, the use of a structured follow-up protocol, supplemented by telephone consultations to minimize missing data, enhances the completeness and reliability of outcome assessment.
Together, these methodological strengths support the role of ESS as a viable decompression strategy associated with meaningful improvements in patient-reported quality of life and functional outcomes during early institutional adoption. Furthermore, this study establishes a foundation for future prospective, multicenter controlled studies in Singapore, which are warranted to further evaluate comparative effectiveness and cost-efficiency as ESS techniques continue to evolve.
While ESS offers clinical advantages and favorable patient outcomes, it carries a notable cost premium and requires a significant learning curve for surgeons. Adoption of UBE ESS necessitates investment in specialized equipment and often involves longer operative times during the early phase of proficiency development, which can affect resource utilization and clinical workflow efficiency.
CONCLUSION
ESS, particularly the UBE approach for lumbar decompression/discectomy, is a minimally invasive and effective intervention for degenerative lumbar spine conditions in Singapore’s tertiary hospitals. Our study demonstrates favorable early clinical outcomes, low complication rates, and a manageable learning curve. These findings support broader integration of ESS into mainstream spinal care pathways and warrant further multicenter prospective research.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Print 
