Awake Anesthesia in Endoscopic and Minimally Invasive Transforaminal Lumbar Interbody Fusion (TLIF): A Systematic Review With Meta-analysis of Minimally Invasive TLIF (Spinal Versus General Anesthesia)
Article information
Abstract
Objective
Transforaminal lumbar interbody fusion (TLIF) is commonly performed in patients with lumbar stenosis and segmental instability. Endoscopic and minimally invasive (MIS) TLIF have recently attracted increasing interest as less invasive surgical approaches. Although general anesthesia (GA) remains the standard anesthetic strategy, awake protocols, including spinal anesthesia (SA) and monitored anesthesia care (MAC), may offer potential advantages by reducing anesthetic risk. This review assessed the current literature on awake protocols, including SA and MAC, in patients undergoing endoscopic and MIS TLIF, with a focus on perioperative and clinical outcomes.
Methods
A systematic database search of PubMed, Embase, Web of Science, and CINAHL was performed from inception to 25 August 2025 in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines. A proportional meta-analysis using a random-effects model was performed for perioperative outcomes, and funnel plots were used to assess potential publication bias. Study quality was evaluated using the MINORS (Methodological Index for Non-Randomized Studies) criteria.
Results
Eleven studies involving 351 patients were included, with a mean follow-up of 19 months. In the meta-analysis, MIS TLIF performed under SA showed a trend towards improved perioperative outcomes, including length of stay, estimated blood loss, and operative time, compared with GA. However, the results were inconsistent and were not consistently statistically significant, and substantial heterogeneity limited interpretation. The analysis of endoscopic TLIF performed under MAC was limited to descriptive synthesis because of the lack of comparative datasets and heterogeneity in outcome reporting, precluding formal comparative inference between MAC and GA. No major complications were reported.
Conclusion
MIS TLIF performed under SA or GA demonstrated comparable perioperative and clinical outcomes, with observed differences representing nonsignificant trends in predominantly low-level retrospective evidence. Endoscopic TLIF performed under MAC remains supported only by limited descriptive evidence, precluding quantitative synthesis. Overall, the current evidence is exploratory and hypothesis-generating. Higher-quality prospective comparative studies with standardized outcome reporting are needed to clarify the optimal anesthetic strategy for each distinct surgical approach.
INTRODUCTION
Lumbar degenerative disease is a major cause of disability worldwide and frequently necessitates surgical intervention when conservative management fails [1,2]. Transforaminal lumbar interbody fusion (TLIF) is a widely utilized technique that restores disc height, achieves neural decompression, and provides segmental stability [3]. Minimally invasive (MIS) and endoscopic TLIF approaches have gained popularity in recent years as a less invasive option, with reduced iatrogenic soft tissue injury, blood loss, and postoperative morbidity compared with conventional open techniques [3]. With an aging population worldwide, this is particularly advantageous in treating older patients with more comorbidities [4].
Traditionally, TLIF is performed under general anesthesia (GA), which provides optimal airway control and ability to perform long surgeries in the prone position [5]. However, growing evidence has demonstrated advantages of awake approaches such as spinal anesthesia (SA) and monitored anesthesia care (MAC) [6]. A recent systematic review in 2023 compared SA versus GA in spine surgery. Rajjoub et al. [7] reported shorter surgery times, hospital length of stay (LOS) and lower complication rates including urinary retention and postoperative nausea for the SA group. Parallel to these surgical advancements, there has been increasing interest in performing spinal procedures under MAC or SA, thereby avoiding the risks associated with GA. This is particularly relevant in elderly or comorbid populations, where anesthetic-related complications may significantly impact outcomes [7].
However, most of the existing literature evaluates awake techniques across heterogeneous spinal procedures. While the use of SA for decompression surgery alone is well established, the evidence specific to more complex surgeries such as endoscopic or MIS TLIF remains limited. Given the rising interest in awake minimally invasive fusion techniques, a focused evaluation of outcomes in endoscopic and MIS TLIF performed without GA is warranted. Therefore, the aim of this systematic review was to assess the current literature evaluating awake protocols (SA and MAC) in patients undergoing endoscopic and MIS TLIF, with a focus on perioperative and clinical outcomes.
MATERIALS AND METHODS
This systematic review was planned, conducted, and reported in accordance with the Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA) guidelines [8]. The review was registered on International Prospective Register of Systematic Reviews (PROSPERO) database (registration No. CRD420261288135). Registration occurred after completion of the literature search but before data extraction and analysis. The study was conducted in accordance with the registered protocol.
1. Search Strategy
We conducted a comprehensive database search of Embase, PubMed, CINAHL and Web of Science from inception until 25 August 2025. These databases were chosen as they provided a wide range of biomedical articles from international journals [9].
The search strategy used consisted of the following terms: ((Fusion AND transforaminal AND lumbar) OR TLIF OR Transforaminal Lumbar Interbody Fusion) AND (endoscopic OR endoscopy OR minimally invasive) AND (awake OR without general anaesthesia OR local anaesthesia OR percutaneous OR epidural OR regional OR spinal anaesthesia OR monitored anaesthesia care). Full search terms are provided in the supplementary material.
In addition, a snowball strategy was applied, looking at references and citations of relevant articles with similar topics.
2. Eligibility Criteria
1) Inclusion criteria
We included studies that involved patients above 18 years old, of any gender or race, who underwent either endoscopic or MIS TLIF without GA. Awake protocols included either MAC or SA. Studies were accepted regardless of the number of levels operated on. We included follow-up studies (including retrospective or prospective cohort studies/nonrandomized controlled trials) and case series with >8 people. Only studies written in English were included.
2) Exclusion criteria
We excluded studies which analyzed people who went for procedures under GA, translaminar procedures or with duplicate study populations. Systematic reviews, meta-analysis, case reports and articles where full-text was unavailable were also excluded.
3. Study Selection and Data Extraction
The review process was completed using Covidence [9].
Two reviewers (AT and ST) independently screened through the titles and abstracts based on the inclusion and exclusion criteria. Following that, the full texts of shortlisted studies were retrieved. For studies originating from the same institution or research group, recruitment periods, study centers and eligibility criteria were reviewed to assess potential population overlap. Where available, these were cross-checked to minimize the risk of duplicate patient inclusion. The 2 reviewers (AT and ST) then independently extracted data from these full texts. From each study, we extracted study reference (author, year of publication, country of study, study design/duration/objectives), demographics of subjects (number of participants, gender, age range, comorbidities) and results of the study (perioperative parameters, postoperative functional and neurologic status changes and other relevant scores/measurements). Perioperative parameters consisted of operative time, estimated blood loss (EBL) and LOS. Functional and neurologic status were measured by changes in Oswestry Disability Index (ODI) and visual analogue scale (VAS) for back/leg pain. Complication rates were also extracted.
GA, MAC, and SA were defined at first use and consistently applied throughout the manuscript, with MAC and SA reflecting reporting patterns in endoscopic and MIS TLIF studies respectively.
Disparities were resolved by discussion and consultation with the review team. The final included studies were decided in consultation with the senior author (JL). Each step of the selection process was outlined in PRISMA.
4. Statistical Analysis
Data analysis was performed using R statistical computing software (Version 2024.12.1+563, Tidyverse and stats packages) [10]. Studies comparing MIS TLIF under GA and SA were analyzed separately from studies comparing endoscopic TLIF under GA and MAC due to fundamental procedural differences.
A random-effects meta-analysis was performed to compare perioperative outcomes between GA and SA in MIS TLIF. Funnel plots were generated to assess publication bias where appropriate, although interpretation was limited by the small number of included studies [11,12]. I2 of 50% to 75% was considered moderate heterogeneity and greater than 75% considered high heterogeneity. A p-value less than 0.05 was considered statistically significant. No meta-regression or sensitivity analysis was performed due to limited study numbers and heterogeneity in reported outcomes. Functional and neurological outcomes in MIS TLIF studies were not pooled due to substantial heterogeneity in outcome reporting.
All outcomes in endoscopic TLIF studies comparing GA and MAC were analyzed descriptively due to substantial heterogeneity and limited study numbers [13,14]. Standardized mean differences (SMDs) were calculated to facilitate cross-study comparison. SMDs were reported descriptively to illustrate the direction and relative magnitude of treatment effects rather than to derive a pooled summary estimate. A narrative synthesis was additionally undertaken to summaries study characteristics and findings.
5. Quality Assessment
The quality of all included studies was independently evaluated by 2 authors (AT and ST). The Methodological Index for Non-Randomized Studies (MINORS) was applied, using an 8-item index (with a global ideal score of 16) for noncomparative studies and 12-item version (maximum score 24) for comparative studies [15]. Any discrepancies in scoring were resolved through discussion with the senior author (JL).
RESULTS
1. Study Selection
Using the above search strategy, a total of 1,132 articles were identified. 798 studies were sent for title and abstract screening after removing duplicates. After reviewing 25 full-text articles, 11 studies were eventually included in the review [16-26]. Details of the search strategy and selection of the included studies are summarized in Figure 1.
PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) diagram.
2. Study Characteristics and Quality
The 11 studies included 3 cohort studies, 2 case control studies and 6 case series (Table 1). Studies were mostly from the United States (US) (n=10), while the remaining one study was from South Korea. Four studies reported the use of endoscopic TLIF and 7 studies reported the outcomes of MIS TLIF. The authors observe that the awake protocol choice of anesthesia for all the included endoscopic TLIF studies was MAC [6], while that of the MIS TLIF studies was SA. A total of 351 patients were included. Six studies (all MIS TLIF) only reported perioperative outcomes while the remaining studies had a mean follow-up period of 19 months. The mean age across all studies ranged from 44–67 years. Patient comorbidities were inconsistently and heterogeneously reported across studies, precluding meaningful summarization or stratified analysis.
Summary of included studies
Table 2 presents the MINORS assessment results. Noncomparative studies achieved scores ranging from 10 to 12 (median, 11; IQR, 10–12), while comparative studies scored between 16 and 19 (median, 18; IQR, 17–19), reflecting overall moderate to high methodological quality. Items such as a clearly defined study aim, inclusion of consecutive patients, and use of appropriate endpoints were consistently well addressed. Among comparative studies, the adequacy of control groups, baseline equivalence, and statistical methodology were also generally robust.
Methodological index for nonrandomized studies assessment
Nonetheless, several methodological shortcomings were identified. Prospective data collection and formal sample size calculations were infrequently performed, and reporting on loss to follow-up and blinded outcome assessment was variable. These limitations may introduce bias and reduce precision. Therefore, the reported treatment effects should be interpreted cautiously, as studies with higher risk of bias may overestimate clinical outcomes.
3. MIS TLIF (SA vs. GA)
The perioperative parameters reported included EBL, operative time, LOS and postoperative opioid use (Table 3). Forest and funnel plots were performed for LOS, EBL and operative time (Figures 2–7). Three studies were not included in these plots (2 did not report mean scores, 1 single-arm study).
Summary of perioperative parameters for included studies
Forest plot of studies comparing length of stay (hours) of patients who underwent minimally invasive transforaminal lumbar interbody fusion under spinal versus general anesthesia. GA, general anesthesia; SA, spinal anesthesia; SMD, standardized mean difference; CI, confidence interval.
Funnel plot of studies comparing length of stay (hours) of patients who underwent minimally invasive transforaminal lumbar interbody fusion under spinal versus general anesthesia.
Forest plot of studies comparing estimated blood loss (EBL) (mL) of patients who underwent minimally invasive transforaminal lumbar interbody fusion under spinal versus general anesthesia. GA, general anesthesia; SA, spinal anesthesia; SMD, standardized mean difference; CI, confidence interval.
Funnel plot of studies comparing estimated blood loss of patients who underwent minimally invasive transforaminal lumbar interbody fusion under spinal versus general anesthesia.
Forest plot of studies comparing operative time (minutes) of patients who underwent minimally invasive transforaminal lumbar interbody fusion under spinal versus general anesthesia. GA, general anesthesia; SA, spinal anesthesia; SMD, standardized mean difference; CI, confidence interval.
Funnel plot of studies comparing operative time (minutes) of patients who underwent minimally invasive transforaminal lumbar interbody fusion under spinal versus general anesthesia.
Four studies (n=313; GA: 154; SA: 159) reported LOS, demonstrating a nonsignificant increase with GA compared with SA (SMD=1.77; 95% confidence interval [CI], -0.60 to 4.15; I²=96.6%) (Figure 2). Three studies (n=238; GA: 108; SA: 130) reported EBL, showing a small nonsignificant increase with GA (SMD=0.30; 95% CI, -0.05 to 0.66; I²=0%) (Figure 4). Four studies reported operative time (n=313; GA: 154; SA: 159), with GA associated with a nonsignificant longer operative time (SMD=2.98; 95% CI, -1.00 to 6.96; I²=97.7%) (Figure 6). Funnel plot analysis demonstrated asymmetry in LOS (Figure 3) and operative time (Figure 7), likely reflecting substantial heterogeneity.
Postoperative opioid consumption was reported in 4 studies but not pooled due to inconsistent reporting units (Table 3). There was lower dosage of opioids used for the SA group in all studies, with 2 demonstrating statistical significance (p<0.05).
The functional and neurologic outcomes assessed included the ODI, VAS back and VAS leg scores (Table 4). Minimal clinically important difference (MCID) thresholds for these scores were extracted from published literature specific to each surgical technique (endoscopic TLIF and MIS TLIF) and applied accordingly where reported [27,28]. Two studies each reported ODI, VAS back, and VAS leg scores. All studies except for Ononogbu-Uche et al. (VAS leg 1.9<MCID of 2.1) reported mean change above MCID across scores [28,29]. Only one study reported mean change of ODI and VAS scores for SA versus GA [25]. Hence, there were not enough cohorts to compare GA and non-GA outcomes directly.
Summary of functional and neurologic outcomes scores for the included studies
4. Endoscopic TLIF (MAC)
For endoscopic TLIF under MAC, we were unable to pool perioperative parameters for meta-analysis due to heterogeneity (Table 3). Only 2 of the studies reported both mean and standard deviation, while none of them were comparative.
Four studies reported ODI, 2 studies reported VAS back and 1 study reported VAS leg scores (Table 4). All studies but Kolcun et al. [14] (ODI 12.4<MCID of 15) reported mean change above MCID across scores [27]. However, findings were descriptive in nature due to lack of comparative datasets and heterogeneity in reporting, precluding any formal comparative inference between MAC and GA.
5. Complications
Five studies reported complications attributable to anesthetic technique (Table 5). The most common complication was nausea and vomiting, others include postoperative ileus, transitory dysesthesia and anxiety. Sekerak et al. [23] reported no significant difference in complication rate (nausea and vomiting) between the SA and GA groups.
Complications
DISCUSSION
In this study, we observed comparable perioperative and clinical outcomes for MIS TLIF performed under SA or GA, based on predominantly low-level retrospective evidence. Meanwhile, endoscopic TLIF performed under MAC remains supported only by limited descriptive evidence, precluding quantitative synthesis. Overall, current evidence is exploratory and hypothesis-generating.
The present review intentionally analyzed endoscopic TLIF under MAC separately from MIS TLIF under SA, given the fundamental procedural and perioperative differences between these approaches. Endoscopic TLIF is typically performed through smaller working channels with reduced soft tissue disruption and is therefore more amenable to conscious sedation with local anesthesia. In contrast, MIS TLIF often requires sequential dilation, tubular retraction and greater manipulation of paraspinal musculature, making SA a more commonly utilized awake anesthetic strategy. These procedural differences introduce substantial clinical heterogeneity and limit the validity of directly pooling endoscopic MAC and MIS SA cohorts within comparative meta-analysis.
Current literature has reported several potential advantages of awake procedures over GA. Firstly, the use of awake procedures may reduce the GA risk for patients with major medical comorbidities. GA is associated with intraoperative hemodynamic fluctuations and ventilation, which increases risk of myocardial infarction, respiratory depression and renal failure. Avoiding GA may be particularly advantageous in high-risk individuals with multiple comorbidities, who may otherwise be deemed too high risk to proceed with surgery. In today’s context of an aging population with increasing prevalence of symptomatic degenerative spine disease [30], this gives more people a chance to undergo spinal surgery and regain their quality of life.
In addition, common GA side effects like giddiness, nausea, vomiting and sometimes delirium in older patients are also avoided. Postoperative pain is also reduced with the need for less narcotics [23,26]. This may facilitate earlier ambulation and postoperative recovery, thus reducing the LOS and risk of hospital related infection [20,31]. Hence, by improving patient comfort and allowing them to return home earlier, this would potentially improve patient experience.
Finally, awake procedures can potentially improve surgical conditions. In the context of TLIF, patients can report radiating symptoms in real time, which provides surgeons with instant feedback on whether any nerve is being damaged during certain portions of the surgery such as decompression or insertion of screws. Although no studies in the current review reported endoscopic TLIF performed under SA, the potential advantages of this anesthetic technique warrant consideration. The use of SA lowers down the patient’s blood pressure as compared to GA, hence in the context of endoscopic TLIF, the authors postulate less bleeding and possibly better scope visualization. SA also aids in muscle relaxation, thus reducing muscle twitching intraoperatively. These factors may optimize the surgical environment and potentially contribute to reducing operation timing and operating theater (OT) turnover time.
While awake anesthesia techniques offer potential advantages in TLIF, several limitations and challenges must be acknowledged. Firstly, there is a nonnegligible risk of conversion to GA, particularly when major intraoperative complications arise (e.g., massive bleeding or complex dural tears). Hence, both the surgical and anesthetic teams must have robust contingency plans, as well as protocols for rapid conversion if needed. Moreover, airway management in prone patients under light sedation also remains a particular concern, especially in the context of intraoperative desaturation [32]. This makes vigilant hemodynamic monitoring and titration of sedatives essential. Kolcun et al. [14] reported 4 cases of conversion to GA, due to emesis (n=2), epistaxis (n=1), and extreme anxiety (n=1). The authors adjusted their protocol to prevent further cases of conversion. To prevent epistaxis, oxymetazoline spray was given preoperatively and ondansetron with glycopyrrolate was administered as part of the preoperative medication regime to avoid intraoperative vomiting. Kolcun et al. [14] reflect that intraoperative emesis may be predisposed by a history of reflux. Future studies may screen patients for risk factors that predispose to GA conversion and adopt modifications in preoperative regime to reduce conversion rate. For high-risk patients, GA may still be advantageous in maintaining consistent airway control and predictable anesthetic depth from the start.
With regards to SA, the duration and extent of surgery are inherently constrained by the effective sensory and motor block time of the spinal anesthetic. This limits SA primarily to one to 3 level fusion [33], which is also observed in our review. We observed that a majority of included studies reported 1 to 2 level TLIF, with only Navarro-Garcia de Llano et al. [22] reporting for 3 levels (n=2). SA may also compromise intraoperative neuromonitoring by suppressing motor evoked potentials and somatosensory evoked potentials [34]. Hence if SA is used for more complex fusion cases, surgeons may have to utilize real-time navigation systems (e.g., O-arm imaging). Hence, GA would likely be advantageous in complex multilevel cases where a longer duration is needed for operation or when navigation systems are not readily available.
SA would be more complicated in patients with altered anatomy such as severe scoliosis and degenerative spinal disease [35]. Adjustment of medications may also be complex for patients with extreme body heights, especially short-statured patients, such as those with achondroplasia. Anxiety, inability to lie prone for a duration of time when awake, or complex spinal deformity will also limit the suitability of use of SA. Finally, successful use of SA or MAC in spinal surgery also ultimately depends on a good relationship between surgeon and anesthetist. Trust and good communication are required between both teams to ensure appropriate decision making in critical situations.
1. Limitation
The authors recognize several limitations. Firstly, there was a small number of studies available (n=11), with particularly limited representation of individual minimally invasive techniques, especially endoscopic TLIF (n=4). This restricts the granularity of comparative analyses and highlights the need for future studies directly comparing MAC versus GA and SA versus GA to strengthen evidence synthesis.
Next, clinical outcomes, anesthesia regime and exact surgical techniques that were reported in the included studies were highly heterogeneous. Perioperative outcomes reported were limited and few studies reported important data including time to ambulation and OT turnover time. This limited our ability to pool outcomes together for greater impact conclusions. For outcomes that were pooled, the very high between-study heterogeneity reduced confidence in the pooled estimates and limited definitive conclusions regarding comparative effectiveness.
There was also variability in study sizes among the included studies, ranging from small case series (n=10) to larger cohorts (n=101). This may limit the robustness and generalizability of the pooled estimates, particularly given the potential for small-study effects. A further limitation is the potential for partial population overlap among studies originating from the same institution and research group, as overlapping recruitment periods and centers could not be fully excluded. However, any such overlap is unlikely to substantially affect the overall findings given the study-level nature of the data and methodological heterogeneity. Ten of the 11 included studies originated from the US, with only one from South Korea, resulting in a strong geographical imbalance. This limits external validity as differences in healthcare systems, anesthetic practices and perioperative pathways may affect outcomes. Consequently, the findings may not be fully generalizable to other international settings and should be interpreted with caution.
The inconsistent and heterogeneous reporting of comorbidities precluded meaningful summarization or subgroup analysis, introducing potential bias and limiting the ability to identify patient populations most appropriate for SA or MAC. Furthermore, all included studies were retrospective observational in design, and no randomized controlled trials were available, which inherently increases susceptibility to selection bias and confounding.
There were few comparative studies, making it hard to come to conclusions if awake procedures are superior to GA. In addition, MCID thresholds were derived from technique-specific literature rather than a single standardized source. While this approach preserves methodological consistency within each surgical cohort, variability across studies limited direct comparability of MCID attainment between groups.
Publication bias was assessed qualitatively using funnel plot asymmetry where sufficient studies were available. However, the interpretation of funnel plots in our study has several limitations. Firstly, the relatively small number of included studies reduces the reliability of funnel plot asymmetry as statistical power to detect true publication bias is limited when such few studies are available. Secondly, substantial variation on surgical techniques, anesthetic protocols, patient selection and study design among the included studies may further confound funnel plot interpretation. Most of the included studies were retrospective observational cohorts with variable sample size and hence, the possibility of small-study effects and residual reporting bias cannot be excluded.
Finally, PROSPERO registration was completed after the literature search had been conducted. Although the review protocol and analytical framework had been predetermined, retrospective registration may increase the perceived risk of selective reporting.
2. Clinical Implications
In the clinical context, surgeons and anesthetists would need to work closely with each other to choose an anesthetic plan customized to each patient’s surgical needs and medical comorbidities. MAC and SA may potentially be advantageous in suitable patient groups. These techniques may complement enhanced recovery after surgery pathways in appropriately selected patients, with the potential to improve patient turnover, cost savings and outreach to individuals who require TLIF. These advantages have been observed in other more common orthopedic surgeries such as total hip replacement [36], where MAC and SA have been established as the standard for years. Thus, we postulate that the benefits of these anesthesia techniques in the context of TLIF may only be observed after longer follow-up on larger sample sizes.
3. Recommendations for Future Research
This calls for more prospective cohort studies and randomized controlled trials with larger population sizes. Outcomes measured need to be more standardized and holistic, to properly assess perioperative outcomes as well as pain, function and neurology. Patient satisfaction of the surgeries should also be included. More comparative studies against GA are needed, and future studies could also compare between the awake methods (SA and MAC). Future papers could investigate cost-effective measures and the effectiveness of collaborative training between surgeons and anesthetists to specialize in awake TLIF. Further studies can also have more comprehensive analysis of adverse outcomes and complications.
CONCLUSION
MIS TLIF performed under SA or GA demonstrates comparable perioperative and clinical outcomes, with observed differences representing nonsignificant trends within predominantly low-level retrospective evidence. Endoscopic TLIF performed under MAC remains supported only by limited descriptive evidence, precluding quantitative synthesis. Overall, current evidence is exploratory and hypothesis-generating. Higher-quality prospective comparative studies with standardized outcome reporting are required to clarify the optimal anesthetic strategy within each distinct surgical approach.
Notes
Conflicts of interest
S Lee and JW Hur, members of the Editorial Board of Journal of Minimally Invasive Spine Surgery & Technique, are the authors of this article. However, they played no role whatsoever in the editorial evaluation of this article or the decision to publish it. The other authors have nothing to disclose.
Funding/Support
This study received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.