INTRODUCTION
Lumbar disc herniation (LDH) is one of the most common causes of sciatica and low back pain, significantly impacting patients’ quality of life [1]. For patients whose symptoms persist or worsen despite at least 6 weeks of conservative treatment, surgical intervention is considered [2]. For decades, the undisputed “gold standard” in the surgical treatment of LDH has been the lumbar microdiscectomy (MD) [3]. Since its independent introduction by Yasargil and Caspar in 1977 [4], MD has proven its superiority over traditional open discectomy through smaller incisions, reduced tissue damage, and excellent visualization of the surgical field. Nevertheless, MD still entails measurable spinal tissue trauma, as achieving sufficient nerve root decompression and an adequate surgical corridor often necessitates some degree of soft-tissue manipulation, with limitations in visualization necessitating dural retraction, and challenges in managing epidural bleeding [5].
Endoscopic spine surgery, including endoscopic discectomy (ED), has evolved over the past 30 years and continues to undergo rapid advancements aimed at improving efficacy, minimizing complications, and expanding its indications [6,7]. Compared to microscopy, proponents argue that various types of endoscopy reduce postoperative pain, hospital stay, and complications, potentially making it a replacement for MD [5,8,9]. However, the evidence remains mixed. While some studies report notable short-term advantages of ED in terms of reduced postoperative pain and quicker functional recovery, few studies demonstrate that long-term clinical outcomes are largely comparable to those achieved with MD. Nevertheless, questions remain regarding the extent to which these short-term benefits justify the associated technical complexity and learning curve [10-12].
This systematic review and meta-analysis aim to compare the efficacy, safety profile, and recurrence risk of ED versus MD for LDH. Specifically, we evaluate patient-reported outcomes (PROMs), perioperative parameters, and complications, with subgroup analyses based on different endoscopic approaches. Through this review, we seek to determine whether ED offers benefits equivalent to or surpassing MD or if its advantages remain technique-specific and context-dependent.
MATERIALS AND METHODS
1. Search Strategy and Eligibility Criteria
This systematic review and meta-analysis was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines for meta-analysis [13] and was preregistered with PROSPERO (Registration No. CRD420 251086356). A comprehensive literature search was performed in PubMed, Embase, Scopus, and Web of Science to identify relevant studies on surgical interventions for LDH. The search strategy included terms related to “microscopic discectomy,” “full endoscopic discectomy,” “percutaneous endoscopic lumbar discectomy,” “transforaminal endoscopic discectomy,” “percutaneous transforaminal endoscopic discectomy,” “interlaminar endoscopic discectomy,” “percutaneous endoscopic interlaminar discectomy,” “biportal endoscopic spine surgery,” and “unilateral biportal endoscopy.” Manual screening of reference lists from eligible studies was also conducted to ensure completeness. Eligible studies were randomized controlled trials (RCTs), and prospective cohort studies in English from 1 January 2000 to 30 June 2025, enrolling adult patients (≥18 years) who underwent one of the following procedures: microscopic discectomy (MD), transforaminal endoscopic lumbar discectomy (TELD), interlaminar endoscopic lumbar discectomy (IELD), or unilateral biportal endoscopy discectomy (UBED), and compare outcomes between microscopic surgery with any type of endoscopy. Only studies that reported relevant clinical outcomes with sufficient quantitative data were included.
2. Study Selection and Outcome Measurement
Two authors independently screened the titles and abstracts retrieved from the database queries. Any disagreements on whether to include or exclude were discussed and reviewed with the third senior investigator. Baseline characteristics of each study were recorded, including age, sex, follow-up period, type of surgery, and outcome measurement. Primary outcome was mean change of PROM measurement between baseline and each time-point, including the Oswestry Disability Index (ODI) and visual analogue score (VAS) for back and leg at ≤3 months (short-term), at 12 months, and at ≥2 years (long-term). Change scores (baseline minus follow-up) were analyzed for ODI and VAS so that positive values indicated greater improvement. Operative time (OT) and length of stay (LOS) were analyzed as raw follow-up means (smaller values better). Binary outcomes comprised any peri-/postoperative complication and recurrence rate.
3. Data Extraction and Risk of Bias Assessment
Data extraction was conducted independently by 2 reviewers using a standardized form, collecting relevant information on study design, participant characteristics, surgical intervention, comparator, and outcome measures. Risk of bias for included RCTs was assessed using the Cochrane Risk of Bias 2.0 tool [14], while the ROBINS-I (Risk Of Bias In Non-randomized Studies – of Interventions) tool was used for nonrandomized studies [15]. Two authors independently performed risk of bias assessment and resolved disagreements by discussion or, if necessary, consultation with a third reviewer. The robvis (Risk-of-bias VISualization) tool was employed to create the summary and trafficlight plots of the risk of bias results [16].
4. Certainty of Evidence (GRADE Assessment)
We evaluated the certainty of evidence for all primary outcomes using the GRADE framework, which considers 5 domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. RCT evidence begins at high certainty and is downgraded based on methodological limitations. For outcomes combining RCTs and observational studies, the body of evidence was downgraded if observational studies had substantial risk of bias or contributed heavily to pooled estimates. Certainty of evidence was rated as high, moderate, low, or very low. A summary of findings table was constructed to present pooled effect sizes alongside GRADE certainty ratings.
5. Data Synthesis and Statistical Analysis
All statistical analyses were conducted using R ver. 4.4.2 (R Foundation for Statistical Computing, Austria) with the meta package (ver. 6.5-0). For continuous outcomes (e.g., OT, LOS, VAS, ODI), pooled mean differences and 95% confidence intervals (CIs) were calculated using the inverse variance method. For binary outcomes (e.g., recurrence, overall complications, dural tears, wound infections), odds ratios (ORs) with 95% CIs were calculated using either the Mantel-Haenszel or inverse variance method, depending on event rate and model fit. Subgroup analyses were performed by endoscopic subtype (TELD, IELD, unilateral biportal endoscopy [UBE]) where data permitted. A random-effects model using the Paule-Mandel estimator was employed to account for between-study heterogeneity. The Hartung-Knapp method was applied to produce robust CIs. Heterogeneity was assessed using the I² statistic, τ², and Cochran Q test, with I² >50% considered substantial. A p-value of <0.05 was considered statistically significant.
Funnel plots were used for visual inspection of publication bias, and Egger regression test was applied for statistical evaluation. Although some outcomes included fewer than 10 studies, Egger test was performed and interpreted with caution due to limited statistical power.
RESULTS
1. Characteristics and Quality of Included Studies
A total of 11,326 records were identified through comprehensive database searches of PubMed, Embase, Scopus, and Web of Science (Fig. 1). After removing duplicates and conducting title, abstract, and full-text screening, 17 studies met the eligibility criteria and were included in the quantitative synthesis (Fig. 1). Of these, 9 were RCTs and 8 were prospective cohort studies, representing 10 countries: Korea, India, Germany, Brazil, the United Kingdom, the Netherlands, Portugal, Russia, Ukraine, and Egypt. Together, these studies enrolled 3,115 patients; 1,793 in the MD group and 1,322 in the ED group. The mean followup duration across studies was 23.15 months (range, 1–60 months), and the overall mean patient age was 43.02 years.
All studies compared MD with one or more types of ED, including IELD, TELD, and UBE. The primary outcomes assessed included the ODI, VAS scores for back and leg pain, OT, LOS, complication rate, and recurrence rate. Detailed study characteristics are summarized in Table 1.
Among the 9 RCTs assessed using the RoB 2.0 tool, 33% were rated as low risk, 44% had some concerns, and 22% were judged high risk, most commonly due to deviations from intended interventions (D2) and missing outcome data (D3) (Fig. 2A). Of the 8 nonrandomized studies evaluated with ROBINSI, one (12.5%) demonstrated a serious risk of bias, primarily due to incomplete follow-up, while the remaining studies were judged to have moderate risk, mainly relating to confounding and outcome measurement (Fig. 2B).
Short-term and 1-year ODI outcomes demonstrated moderate-certainty evidence favoring ED, whereas long-term ODI was based on very low-certainty evidence due to imprecision and high heterogeneity. Pain outcomes (VAS leg/back) ranged from low to very low certainty for short- and long-term time points because of substantial variability among studies, while 1-year back pain showed moderate certainty. Recurrence demonstrated moderate-certainty evidence indicating higher risk with ED overall, driven predominantly by the TELD subgroup. IELD and UBE recurrence estimates were supported by low-certainty evidence, reflecting imprecision from wide CIs (Table 2).
2. Perioperative Outcome
1) Operative time
There were 13 studies [17-29] that reported OT, including 2,291 patients with 1,274 and 1,017 patients in microscopy and endoscopy groups, respectively. The pooled analysis demonstrated no statistically significant difference in OT between endoscopic and MD. The overall mean difference was -5.89 minutes (95% CI, -16.43 to 4.66; p=0.25), favoring the endoscopic approach, but the CI crossed zero. There was considerable heterogeneity across studies (I²=96.3%, τ²=258.64, p<0.0001), suggesting substantial variability in OT estimates between trials. The 95% prediction interval ranged from -42.35 to 30.58 minutes, indicating that future studies may observe a wide range of effects (Fig. 3A).
2) Hospital stays
Thirteen studies [17,19-30] comprising 2,258 patients (976 in the endoscopic group and 1,282 in the microscopic group) were included in the meta-analysis of length of hospital stay. The pooled analysis demonstrated a statistically significant reduction in hospital stay for patients undergoing ED compared to those receiving MD. The overall mean difference was -2.43 days (95% CI, -3.62 to -1.23; p<0.05), favoring the endoscopic approach. Heterogeneity among the studies was substantial (I²=95.9%, τ²=2.89, p<0.0001), indicating considerable variability in effect sizes across studies. The 95% prediction interval ranged from -6.40 to 1.55 days, suggesting that while most future studies are likely to observe a benefit for endoscopy, some may report no difference (Fig. 3B).
3. Patient-Reported Outcomes
1) Short-term ODI (≤3 months)
Seven studies (481 patients in the endoscopic group and 520 in the microscopic group) [17,19,23-26,28,29] reported the mean change in ODI from baseline during the early postoperative period. Using a fixed-effects model (I²=48.7%), ED was associated with significantly greater short-term improvement in disability scores compared to MD. The pooled mean difference was 2.13 points (95% CI, 0.58–3.67; p=0.057). The 95% prediction interval ranged from -3.97 to 7.53, suggesting moderate between-study variability (Fig. 4A). Subgroup analysis revealed that the short-term advantage was most pronounced in patients undergoing TELD (mean difference, 3.72; 95% CI, 2.36–5.09), while no significant difference was observed between UBE and microscopy (mean difference, 0.08; 95% CI, -9.05 to 9.21) (Supplementary Figs. 1C and 2C).
2) 1-Year ODI
Six studies (364 endoscopic, 440 microscopic) [17,23-25,28,29] assessed ODI change at approximately 12 months postoperatively. The fixed-effects analysis showed a statistically significant benefit favoring the endoscopic approach, with a pooled MD of 2.29 points (95% CI, 0.44–4.15; I²=13.4%; p<0.3288). Heterogeneity was low (I²=13.4%), and the prediction interval ranged from -0.45 to 5.04 (Fig. 4B). Subgroup analysis between MD and TELD shows no significant difference in ODI at 1-year follow-up (Supplementary Fig. 3C).
3) Long-term ODI (≥2 years)
Three studies (436 endoscopic, 442 microscopic) [17,20,27] reported long-term change in ODI. A random-effects model was used due to high heterogeneity (I²=71.7%). The pooled MD was -0.06 (95% CI, -11.72 to 11.61), indicating no significant difference between the techniques. The wide prediction interval (-20.32 to 20.21) reflects substantial variability in long-term outcomes (Fig. 4C). Subgroup analysis showed no significant difference between MD and TELD (mean difference, -0.06; 95% CI, -11.72 to 11.61) (Supplementary Fig. 4C).
4) Short-term VAS Leg Pain (≤3 months)
Nine studies comprising 1,047 patients (504 in the endoscopic group and 543 in the microscopic group) [17,19,23-26,28,29,31] reported the mean change in VAS leg pain from baseline in the early postoperative period. The random-effects model showed no statistically significant difference between the 2 techniques, with a pooled mean difference of -0.02 points (95% CI, -0.63 to 0.60). The prediction interval ranged from -1.63 to 1.60, indicating that future studies may also find no clinically meaningful difference. Heterogeneity was moderate to high (I²=71.7%, τ²=0.24, p=0.0004), suggesting variability in effect sizes across studies (Fig. 5A).
5) 1-Year VAS leg pain
Seven studies (387 endoscopic, 463 microscopic) assessed the mean change in leg pain at 1-year follow-up [17,23-25,28,29,31]. The pooled analysis using a random-effects model revealed no significant difference between groups (mean difference, 0.52 points; 95% CI, -0.12 to 1.17). The prediction interval ranged from -1.19 to 2.23, indicating that future studies could favor either approach. Heterogeneity was substantial (I²=81.4%, τ²=0.41, p<0.0001) (Fig. 5B). Subgroup analysis between MD and TELD shows no significant difference in leg pain at 1-year follow-up (Supplementary Fig. 3B).
6) Long-term VAS leg pain (≥2 years)
Three studies with 878 patients (436 endoscopic, 442 microscopic) reported long-term changes in VAS leg pain [17,20,27]. The pooled analysis using a random-effects model showed no statistically significant difference between ED and MD (mean difference, 0.90 points; 95% CI, -1.78 to 3.58). The prediction interval was wide (-4.27 to 6.07), and heterogeneity was very high (I²=89.7%, τ²=1.04, p<0.0001), limiting the reliability of the estimate (Fig. 5C).
7) Short-term VAS back pain (≤3 months)
Nine studies with a total of 1,047 patients (504 endoscopic, 543 microscopic) [17,19,23-26,28,29,31] reported the mean change in VAS back pain in the early postoperative period. The random-effects model showed no statistically significant difference between the groups (mean difference, 0.39; 95% CI, -0.24 to 1.02; p>0.05), suggesting similar early pain relief. The 95% prediction interval (-1.33 to 2.11) indicates that future studies could favor either group. Heterogeneity was substantial (I²=76.0%, τ²=0.48, p<0.0001) (Fig. 6A).
8) 1-Year VAS back pain
Seven studies (387 endoscopic, 463 microscopic) evaluated VAS back pain change at 1 year [17,23-25,28,29,31]. A fixed-effects model was used due to moderate heterogeneity (I²=48.8%). The analysis showed a statistically significant difference favoring the endoscopic group, with a pooled mean difference of 0.49 points (95% CI, 0.21–0.77; p<0.01). The prediction interval (-0.98 to 1.75) suggests variability across future settings, although the effect size remains small (Fig. 6B). Subgroup analysis between MD and TELD shows no significant difference in back pain at 1-year follow-up (Supplementary Fig. 3A).
9) Long-term VAS back pain (≥ 2 years)
Three studies with 878 patients (436 endoscopic, 442 microscopic) reported long-term outcomes [17,20,27]. The pooled analysis showed no statistically significant difference (mean difference, 1.46; 95% CI, -2.40 to 5.33; p>0.05), and the wide prediction interval (-6.13 to 9.05) highlights considerable uncertainty. Heterogeneity was high (I²=95.2%, τ²=2.30, p<0.0001) (Fig. 6C).
10) Subgroup analysis
Subgroup analyses comparing TELD and UBE with MD showed no statistically significant differences in VAS back, VAS leg, or ODI at short-term, 1-year, or long-term follow-up. These findings indicate that both TELD and UBE achieve outcomes equivalent to MD, with no evidence of superiority in functional recovery or pain relief (Supplementary Figs. 1-4).
4. Complications and Recurrence
All studies reported recurrence rate and overall complication rate [17-33]. Recurrence occurred in 61 patients (3.4%) in the MD group and 73 patients (5.5%) in the ED group. The fixed-effect model demonstrated a statistically significant higher risk of recurrence in the ED group compared to the MD group (OR, 1.90; 95% CI, 1.33–2.72; p<0.001). The prediction interval ranged from 1.28 to 2.82. Importantly, heterogeneity among studies was negligible (I²=0%, τ²=0, p=0.991), indicating consistent findings across studies (Fig. 7).
Subgroup analyses comparing recurrence rates between MD and various endoscopic techniques were conducted. In the TELD subgroup, 7 studies were included with a total of 836 patients in the TELD group and 1,208 in the microscopic group. The recurrence rate was significantly higher in the TELD group (OR, 2.01; 95% CI, 1.11–3.64), with no significant heterogeneity observed (I²=0.0%, p=0.9225). In contrast, the IELD subgroup, which included 5 studies comprising 144 IELD and 457 microscopic cases, did not show a significant difference in recurrence (OR, 1.30; 95% CI, 0.45–3.74; I²=0.0%; p=0.9891). Similarly, in the UBE subgroup (n=375), the pooled analysis revealed no significant difference in recurrence compared to MD (OR, 2.03; 95% CI, 0.23–17.55; I²=0.0%; p=0.8474) (Fig. 7).
Total complication rates were comparable between groups, with 110 events (6.1%) in the MD group and 60 events (4.5%) in the ED group (Fig. 8A). Using a fixed-effect model, the pooled OR for overall complications favored the ED group but did not reach statistical significance (OR, 0.80; 95% CI, 0.55–1.15; p=0.229). The prediction interval ranged from 0.26 to 2.24. Between-study heterogeneity was low (I²=35.7%, τ²=0.1901, p=0.0903), supporting the appropriateness of the fixed-effect model (Fig. 8).
Dural tears occurred in 33 patients (1.8%) in the MD group and 12 patients (0.9%) in the ED group. Using the fixed-effect model, the pooled OR was 0.89 (95% CI, 0.45–1.73), indicating no significant difference between groups, with a wide prediction interval (0.40–1.95). There was no evidence of heterogeneity across studies (I²=0.0%, τ²=0, p=0.5829), suggesting consistency in the reported outcomes (Fig. 8B).
Wound complications were reported in 46 patients (2.6%) in the MD group and only 1 patient (0.1%) in the ED group. Across 17 studies, the pooled OR using a fixed-effect model was 0.17 (95% CI, 0.07–0.42), indicating a significantly lower risk of wound complications in the ED group. There was no evidence of heterogeneity (I²=0.0%, τ²=0, p=0.9986), and the prediction interval ranged from 0.06 to 0.48, suggesting a consistent benefit of the endoscopic approach across studies (Fig. 8C).
5. Publication Bias
Funnel plot inspection revealed no apparent asymmetry for most outcomes, suggesting a low risk of publication bias (Supplementary Figs. 5-10). Egger test showed no statistically significant bias for any outcome, including OT (p=0.322), LOS (p=0.926), short-term ODI (p=0.326), 1-year ODI (p=0.916), long-term ODI (p=0.296), short-term VAS leg (p=0.287), 1-year VAS leg (p=0.916), long-term VAS leg (p=0.232), short-term VAS back (p=0.066), 1-year VAS back (p=0.422), long-term VAS back (p=0.172), recurrence (p=0.059), and overall complication (p=0.477). While Egger test was performed despite fewer than 10 studies per outcome, limiting statistical power, the results collectively indicate minimal evidence of publication bias.
DISCUSSION
This systematic review and meta-analysis provides a comprehensive evaluation of the current evidence comparing various modalities of ED with the established gold-standard MD for the treatment of LDH. The cumulative evidence, drawn from 17 studies encompassing 3,115 patients, presents a bifurcated and nuanced conclusion. While ED solidifies its role as a truly minimally invasive procedure with demonstrable and statistically significant perioperative advantages, it did not clearly demonstrate clinical superiority in long-term PROMs. The key insight from these findings is that the umbrella term “endoscopic discectomy” is not uniform, and its associated risks and benefits are highly dependent on the specific surgical approach.
1. Perioperative Gain and PROMs
Across included studies, both ED and MD were associated with substantial improvements in disability and pain scores, confirming their efficacy in alleviating symptoms of LDH. ED demonstrated greater short-term improvements in ODI within the first 3 postoperative months, with a modest advantage persisting at one year. However, long-term follow-up revealed no significant differences between techniques, a finding consistent with previous meta-analyses reporting comparable functional [34-36].
The superior short-term clinical outcomes and back pain relief with ED may be attributable to its less invasive nature, characterized by reduced soft tissue and bony disruption [32]. IELD minimizes muscle dissection and bone removal to access the disc, whereas the TELD utilizes the natural safe corridor of Kambin triangle, thereby avoiding resection of muscle or bone structures [19]. In subgroup analyses, only TELD demonstrated significantly greater ODI improvement compared with MD, whereas no significant differences were observed for UBE or IELD. Previous meta-analyses similarly reported slightly better ODI improvements favoring TELD over IELD, though these differences were small and did not reach the minimal clinically important difference [37]. Given that TELD is the least invasive approach, preserving natural anatomical structures and avoiding nerve root retraction, this may partly explain its relative advantage compared with UBE and IELD [19,38]. VAS scores for leg and back pain improved comparably between ED and MD across both short- and long-term follow-up, similar to the result of previous meta-analyses [34-36], the only exception was a slight superiority of ED in VAS back pain, with a pooled mean difference of 0.49 points; however, this small magnitude of difference is unlikely to be clinically meaningful. These findings support the concept that, regardless of surgical approach, adequate neural decompression and disc removal can provide sufficient pain relief in patients with LDH. Our results are consistent with prior meta-analyses reporting comparable pain improvement between ED and MD [35]. Another factor to be considered is that we may need a more sensitive measure of daily function and back pain assessment than ODI to tease out the difference between these subtle advantage.
OT was similar between ED and MD in the pooled analysis, in line with earlier reports [36,39]. Considerable heterogeneity across studies was observed, likely reflecting variations in surgical techniques and surgeon experience. Although our study did not specifically assess this factor, prior evidence suggests that ED may prolong OT for less experienced surgeons [40]. In contrast, when performed by experienced surgeons, OT does not differ significantly between the 2 techniques.
Hospital stay was significantly shorter in the ED group compared with MD, with a pooled mean difference of -2.43 days. This finding is concordant with previous studies reporting reduced LOS favoring ED, reinforcing the robustness of this result over time. Notably, 2 studies included in our analysis [24,29], both of which evaluated UBE compared with MD, did not show this benefit. This aligns with prior reports suggesting that UBE may prolong hospital stay compared with full-endoscopic techniques and might result in a similar LOS to MD [41]. Collectively, these findings highlight a potential advantage of ED in reducing hospitalization, particularly when performed using full-endoscopic approaches. However, the same hospital stay metrics may not readily apply across all settings. For example, in regions like the United States where discectomies are frequently performed on an outpatient basis, patients are often discharged within hours after surgery [42]. These variations are driven by differences in health system structure, reimbursement policies, and clinical pathways. As such, while ED offers a reproducible benefit in minimizing hospitalization, its advantage may appear less pronounced in systems already optimized for rapid discharge.
2. Safety Profile and Complications
The overall complication rate in our analysis showed no significant difference between ED and MD, consistent with previous meta-analyses [5,35,39]. However, when focusing on specific complications, wound-related events were significantly less frequent in the ED group, with a pooled OR of 0.17 (95% CI, 0.07–0.42), accompanied by low heterogeneity and a narrow prediction interval. This reduction may be explained by the inherent features of ED techniques, whether IELD, TELD, or UBE, all utilize smaller incisions and less soft-tissue dissection, resulting in reduced dead space and improved local perfusion. Moreover, the use of continuous saline irrigation further lowers the risk of surgical site infection [43,44]. Although the baseline rate of wound complications in MD is relatively low, our pooled analysis demonstrated a statistically significant additional reduction with ED.
Our meta-analysis found no statistically significant difference in the incidence of dural tears between ED (0.9%) and MD (1.8%). Although not significant, the numerical trend favored ED, suggesting that reduced exposure and limited soft-tissue retraction in endoscopic surgery may lower the risk. Prior reviews have similarly reported dural tear rates of 0.5%–2.5% for full-endoscopic discectomy, compared to 1%–7% for MD [5,10]. Advantages of ED include minimal neural retraction, continuous irrigation for clear visualization, and smaller bony removal, all of which may reduce durotomy risk. However, when a dural tear occurs, repair is technically more challenging in ED. Surgeons often rely on conservative measures such as fibrin sealant or fat grafts for small tears, and large durotomies may necessitate conversion to open surgery [45]. Thus, while ED may lower the risk of dural injury, its management requires readiness with endoscopic patching techniques or bailout strategies. As newer endoscopic suturing devices are developed, management options may improve. Clinically, ED is at least as safe as MD regarding durotomy risk, but surgeons must remain vigilant and trained to manage this complication.
Regarding the recurrence rate, like the previous meta-analysis [8], our study also found ED is associated with a significantly elevated risk of recurrence when compared to MD. However, our subgroup analysis revealed that the increased risk of recurrence is not uniform across all endoscopic techniques but appears to be primarily associated with TELD. Compared with MD, TELD was associated with a statistically significant increase in recurrence risk, while other endoscopic methods, such as IELD and UBE discectomy, did not demonstrate such an association. The anatomical and technical limitations of TELD likely contribute to this finding. During TELD, working through a very narrow foramen, the surgeon often removes only the herniated disc fragment without significant decompression of the spinal canal and nerve root. Additionally, if the herniation is large or sequestered, some fragments can be missed. By contrast, in MD, the surgeon typically decompresses the spinal canal, then opens the annulus and removes loose nuclear material under direct vision, potentially reducing the risk of residual fragments. The interlaminar endoscopic approach, in effect, mimics the open technique by allowing direct access to the thecal sac and disc through an enlarged window, which may explain why IELD did not suffer a higher symptomatic recurrence rate in our data. Similarly, biportal endoscopy permits broader decompression, which might prevent missed fragments and may buffer the clinical impact of subsequent herniations. When a disc reherniation occurs after TELD, the absence of additional decompressive space may predispose patients to earlier and more symptomatic recurrence [46].
3. Surgical Invasiveness
ED techniques consistently demonstrate reduced surgical invasiveness compared to MD. Choi et al. [19] compared MD with TELD, percutaneous endoscopic interlaminar discectomy (PEID), and unilateral biportal endoscopic discectomy (UBED), showing that TELD caused the least muscle injury, as evidenced by the lowest creatine phosphokinase (CPK) and C-reactive protein levels, minimal muscle edema on magnetic resonance imaging (MRI), and the shortest hospital stay and OT. UBED showed intermediate invasiveness, while MD resulted in the greatest tissue trauma. However, Wang et al. [47] (2023) found no significant difference in multifidus muscle preservation between UBED and PEID on 12-month MRI, suggesting both approaches are similarly muscle-sparing over the long term. Akçakaya et al. [48] (2016) further supported lower muscle trauma in fully ED, with significantly reduced CPK levels at 6, 12, and 24 hours postoperatively compared to MD. Two RCTs by Park et al. [24,29] further supported the minimally invasive nature of biportal ED, reporting significantly lower postoperative CPK levels, reduced early surgical site pain, better scar outcomes, and fewer wound complications compared to MD, with no difference in long-term outcomes. These findings confirm that ED offers meaningful reductions in surgical invasiveness while maintaining clinical efficacy.
4. Cost-Effectiveness
Recent studies have consistently highlighted the potential economic advantages of ED techniques over traditional MD for treating LDH, though findings vary by perspective and specific approach. In a 2019 retrospective analysis, Choi et al. [46] found that ED, particularly TELD, incurred lower primary hospital costs due to shorter stays and reduced anesthesia needs compared to MD, ED saved an additional net of $8,064 per quality-adjusted life years (QALY). Furthermore, there is no difference in cost-effectiveness among TELD, IELD, and UBED. Building on this, a 2022 randomized noninferiority trial by Gadjradj et al. [49] demonstrated that percutaneous transforaminal ED was noninferior clinically to MD, yielding similar QALYs but lower societal costs driven by reduced absenteeism and healthcare utilization, reinforcing percutaneous transforaminal endoscopic discectomy’s cost-effectiveness from a societal viewpoint despite insurance limitations. Emerging evidence in 2025 further supports these trends while introducing nuances. A 2025 retrospective study in Singapore compared lumbar ED variants to minimally invasive MD, finding UBED more cost-effective due to lower equipment costs (using versatile hospital-owned tools) and similar complication rates, despite longer OTs, positioning it as a viable alternative with a manageable learning curve [50]. However, a 2025 U.S. Medicare analysis revealed declining reimbursements for ED (-27.51% mean per procedure from 2017–2021) amid limited adoption, contrasting with stable MD reimbursements, suggesting barriers to widespread implementation despite potential cost savings [51].
This review has some limitations. High heterogeneity in outcomes like OT and hospital stay, driven by differences in surgical subtypes, surgeon expertise, and patient demographics, reduces the precision of pooled estimates. Possible selection bias in the indication of endoscopic surgeries also may suggest reason for high heterogeneity in outcomes. The mean follow-up of 24 months restricts long-term conclusions on efficacy and recurrence. Subgroup analyses and sensitivity analysis of endoscopic subtypes (TELD, IELD, UBE), type of studies, surgeon experiences, and study region were limited by inconsistent data, and factors like cost-effectiveness and learning curves were underreported. Moreover, the number of IELD and UBE studies was small, which limits the strength of subgroup comparisons and necessitates cautious interpretation of their relative effects. Furthermore, our assessment of publication bias is restricted by the small number of studies contributing to several outcomes. Funnel plots and Egger tests are underpowered when fewer than 10 studies are available; therefore, any observed asymmetry should be interpreted with caution. The certainty of evidence for several outcomes was reduced after GRADE assessment. Many RCTs demonstrated some concerns or, particularly regarding deviations from intended interventions, missing data, and lack of blinding. High heterogeneity affected several pain-related outcomes, leading to downgrades for inconsistency. Imprecision further reduced certainty for long-term outcomes due to wide CIs that spanned clinically important benefit and harm. As a result, the strength of evidence varies substantially across outcomes, limiting the ability to make definitive long-term comparative conclusions. The outcomes of ED, such as OT and complication rates, are heavily influenced by surgeon experience. Prior analyses demonstrate that 60–75 cases are typically required to overcome the learning curve, which may explain some heterogeneity across studies [52]. Future studies need standardized outcomes, longer follow-up, and direct subtype comparisons to address these gaps. Ongoing advancements in endoscopic techniques, together with results from currently registered RCTs [53], may eventually shift the balance. If these trials demonstrate durable long-term superiority, endoscopic surgery could redefine the standard of care, until then, a tailored, hybrid approach remains the most pragmatic strategy for optimizing patient outcomes.
Clinically, our findings imply that ED, particularly IELD and UBE, represents a viable alternative to MD for LDH, offering faster recovery and reduced hospital resource utilization without compromising long-term efficacy. However, the higher recurrence risk with TELD warrants careful patient selection, such as avoiding it in cases with high-grade herniations or calcified discs, and emphasizes the importance of surgeon training to mitigate learning curve effects. Future research should prioritize large-scale RCTs with longer follow-up (>5 years), standardized PROM assessments, and cost-effectiveness analyses to better delineate ED’s efficacy. Head-to-head comparisons of ED subtypes via network meta-analyses would further clarify optimal indications.
CONCLUSION
Addressing whether ED will replace MD as the gold standard for LDH, our analysis concludes “not yet”; better designed studies are needed that control for uniform surgical approach (e.g., IELD or UBE) and pathology (e.g., extruded paracentral herniated nucleus pulposus) in the future. The long history of excellent outcomes with MD sets a high bar that endoscopic techniques have now met, but not yet exceeded. Endoscopic techniques offer reduced invasiveness, shorter hospital stays, and fewer wound complications, yet they fall short of supplanting MD due to comparable long-term functional outcomes and a doubled reherniation risk with transforaminal approaches like TELD. Subtype-specific applications (e.g., IELD, UBE) position endoscopy as a viable alternative in select cases; however, MD remains the benchmark for safety and efficacy. As techniques evolve and high-quality trial data mature, the balance may gradually tilt toward endoscopic surgery, though for now the evidence supports cautious, selective adoption rather than full replacement.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement
Print
