Is Intraoperative Neuromonitoring with SSEPs and EMG Predictable for Postoperative Neurologic Deficit in Posterior Lumbar Fusion Surgery? A Retrospective Cohort Analysis

Abstract

Study Design

Retrospective, single-center cohort study.

Objective

To evaluate intraoperative neuromonitoring (IONM) with free-run electromyography (EMG) and somatosensory evoked potentials (SSEPs) during primary posterior lumbar interbody fusion (PLIF) for degenerative conditions and associations with postoperative motor deficits (PMD).

Methods

Patients undergoing PLIF from 2015 to 2020 were reviewed. Revision fusions, deformity corrections, and procedures in proximity to the conus were excluded. Patient characteristics, comorbidities, surgical details and intraoperative EMG and SSEP recordings were reevaluated. PMDs were defined as any decline of ≥1/5 strength grade compared to preoperative. Test accuracy and predictive value of SSEP and EMG events for PMD were calculated.

Results

401 patients (48.9% females, mean age 61 years, mean BMI 28.6) were included. One- and two-level fusions accounted for 67.8% and 27.7% of cases, respectively, most commonly involving L4/5 (67.8%) and L5/S1 (51.4%). EMG events occurred in 29.4% (n = 118) and SSEP events in 4.5% (n = 18). SSEP events were significantly associated with PMD (P = 0.043), whereas EMG events were not (P = 0.463). In multivariable regression, SSEP events predicted PMD with odds ratios of 3.85 for any SSEP event and OR 10.41 for persistent SSEP signal loss (both P = 0.002). Test performance of SSEP was limited (sensitivity: 13.6%; positive predictive value 16.7%).

Conclusion

In posterior lumbar interbody fusion, SSEP events are associated with postoperative motor deficits, whereas EMG events are not. However, the overall test accuracy of IONM in predicting neurologic deficits remains limited. Instead of routine utilization, IONM should be tailored to the individual case.

Introduction

Iatrogenic neurologic injury is a serious complication in spine surgery, with potentially lasting medical, socio-economic, and legal consequences.^1,2 Intraoperative neurophysiologic monitoring (IONM) has become a widely adopted adjunct in spine surgery, with the goal of identifying potentially harmful maneuvers before permanent neurologic damage occurs.^3,4 IONM employs multimodal techniques to assess neural integrity through including somatosensory evoked potentials (SSEPs), free-run and triggered electromyography (EMG), transcranial motor evoked potentials (MEPs).

IONM is considered standard in certain high-risk spine surgeries, such as deformity correction or intramedullary tumor resections, where its diagnostic value is well supported.^5,6 In some geographic regions, IONM has also become routine in spine surgeries for degenerative lumbar conditions. ³ However, its role in these cases is being questioned, ⁷ as the broad adoption of IONM in these cases may increase health care costs, prolong presurgical preparation and positioning, extend operative times and disrupt workflow due to stimulations protocols.^7-9 It may also require more restrictive anesthesia management, and may have adverse reactions.^10-12

At our institution, IONM is routinely employed in all spine surgeries. For lumbar fusion cases without elevated risk for neurologic injury, monitoring is typically limited to free-run EMG and SSEPs. These modalities provide a degree of neural surveillance while avoiding some of the aforementioned drawbacks associated with more extensive neuromonitoring. However, their diagnostic accuracy for detecting isolated nerve root injury is limited due to overlapping radicular innervation and variable sensitivity to focal deficits^13,14 Additionally, IONM signals may be affected by patient positioning, anesthesia, and other non-pathologic variables, making interpretation difficult.^9-11

Given these limitations, the true utility of routine IONM in degenerative lumbar fusion procedures remains unclear. We therefore conducted a single-institution, retrospective analysis of a homogenous cohort undergoing posterior lumbar interbody fusion (PLIF) for degenerative conditions. Our aim was to provide a more granular evaluation of IONM events in this setting and assess whether such events were predictive of postoperative motor deficits (PMDs).

Methods

After obtaining institutional review board approval (#2022-0108), we retrospectively reviewed patients over the age of 18 years who underwent open posterior decompression laminectomy and instrumented fusion with pedicle screw fixation and posterior lumbar interbody fusion (PLIF) for degenerative disc disease with central and/or foraminal stenosis at a single tertiary spine center between 01/01/2015 and 12/31/2020. IONM is routinely performed in all lumbar fusion cases in our hospital and was utilized in all included cases. Interbody fusion was performed using unilateral PLIF cages at all instrumented levels. Exclusion criteria included: anterior or lateral interbody fusion procedures, revision fusions, deformity correction procedures involving three-column osteotomies, and surgeries extending to L1 or higher due to potential conus medullaris proximity.

Patient demographics, comorbidities, estimated blood loss (EBL), operative time, and number and location of levels fused were extracted from the electronic medical record. IONM reports and pre- and postoperative neurological assessments were reviewed as detailed below. Operative notes were examined to determine whether reported IONM changes resulted in alterations to the surgical workflow. The study reporting is in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. ¹⁵

Neurologic Examination and New Postoperative Motor Deficits (PMD)

At our institution, a standardized neurological exam is performed preoperatively in the holding area by a physician assistant specializing in spine surgery. The exam includes assessment of lower extremity motor strength, sensation, and reflexes. Any discrepancies between this and the outpatient clinic documentation are reviewed and verified by the attending surgeon. Motor strength is graded on a 0-5 scale: 0 no muscle activation; 1 muscle contraction without movement; 2 movement with gravity eliminated; 3 movement against gravity; 4 movement against mild resistance; 5 full strength. The following muscle groups are routinely examined: hip flexion (L1-L3), knee extension (L2-L4), knee flexion (L5-S1), ankle dorsiflexion (L4), great toe extension (L5), and ankle plantarflexion (S1). Postoperative motor exams were performed on postoperative day 1 by the same clinical team. Any decline of ≥1 motor grade in any muscle group compared to baseline was considered a postoperative motor deficit (PMD). In patients with preexisting weakness, a PMD was defined as further decline in the same muscle or new weakness in another group. Sensory or reflex changes were not included in this analysis. PMDs were further categorized as loss of 1 grade or ≥2 grades.

IONM Protocol

SSEPs were recorded using needle electrodes placed posterior to the medial malleolus (tibial nerve) and the distal ulnar forearm (ulnar nerve) to account for anesthetic and positioning effects. At the discretion of the neurophysiologist or surgeon, additional electrodes were placed to stimulate the peroneal and/or saphenous nerves. SSEPs were recorded from CPz-Fpz, CPz-CPc, and CPi-CPc for the lower extremities, and CPc-CPi, CPc-CPz, and CPc-Fpz for the upper extremities. Stimulation parameters included intensities of 15-40 mA for tibial nerve and 15-100 mA for saphenous nerve, with pulse widths of 200-450 µsec. Biphasic stimulation was routinely used to minimize electrode-related tissue injury. The stimulation repetition rate (rep-rate) were set between 4.75-4.78 Hz and adjusted in real time by the neurophysiology technician. Free-run EMG was recorded bilaterally from the vastus lateralis, rectus femoris, tibialis anterior, medial gastrocnemius, and abductor hallucis. Pre-incision baseline signals were recorded after final patient positioning and served as the reference for identifying intraoperative changes.

Anesthesia was managed at the anesthesiologist’s discretion, and no neuromonitoring-specific protocols were mandated. Vecuronium (0.1 mg/kg) was provided to facilitate endotracheal intubation and then titrated to 2/4 twitches, guided by train of 4 (ToF) monitoring. A combination of propofol and inhalational agents, such as isoflurane or sevoflurane, was used, with volatile agents typically maintained below 0.5 minimum alveolar concentration (MAC) to preserve neuromonitoring signal integrity. No adverse events related to stimulation sites were reported during the study period. ¹⁶

IONM Events

IONM reports were reviewed for any adverse events in SSEPs or free-run EMG. Triggered EMG and MEP were not considered for analysis because of their infrequent usage in the investigated cohort. Both the IONM reports and original neurophysiological tracings for all patients were re-evaluated by a board-certified neurologist, who was blinded to postoperative neurologic outcomes. SSEP events were defined as a >50% amplitude reduction, whereas a >10% latency increase from baseline were not considered, as previous studies have shown to be not associated with neurologic injury in spine surgery.^17,18 Events were classified into transient and persistent signal changes by the neurologist. EMG events were defined as acute discharges or spikes. Interpretation was qualitative and based on waveform morphology, as no universally accepted amplitude threshold exists in free-run EMG monitoring.

Test Accuracy

Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated for SSEP and EMG events. A true positive was defined as any reported SSEP or EMG event with a postoperative ipsilateral PMD. A false negative event was defined as a PMD without a recorded IONM event.

Statistical Analysis

Statistical analysis was performed using R-Studio version 2023.09.1 Shapiro-Wilk tests were used to determine normality. Median and interquartile ranges (IQR) were calculated as most of the parameters were non-normally distributed. Kruskal-Wallis test was used to examine the difference of continuous variables between 2 groups. Chi-square testing was utilized for comparing the categorical variables between 2 groups. Multivariable logistic regression that adjusted for age was used to assess the relationship between EMG and PMD, and between SSEP and PMD. Statistical significance was defined as P < 0.05.

Results

The initial hospital record query identified 589 patients undergoing instrumented PLIF with IONM. After applying inclusion and exclusion criteria, 401 patients were eligible for analysis (Figure 1). Mean age was 61 ± 12 years, 48.9% were female, and the mean BMI was 28.6 ± 5.2 kg/m². Of the procedures, 67.8% were one-level and 27.7% were two-level fusions, most commonly involving L4/5 (67.8%) and L5/S1 (51.4%). Preoperative strength deficits were documented in 82 patients (20.5%), most commonly affecting the EHL (n = 50), IP (n = 31), Quad (n = 27), TA (n = 24), HS (n = 22), and GS (n = 10).

Figure 1.

Study flowchart.

EMG and SSEP Events

In 131 patients (32.7%), an IONM event was recorded in the final report; while 270 patients (67.3%) had no reported IONM event. The majority of these IONM events were EMG-related (n = 118), while SSEP event were less common (n = 18). Concomitant EMG and SSEP events were observed in 7 cases.

Comparing the 118 patients with EMG events (29.4%) to the 283 without (70.6 %), there were no significant difference in patient demographics, numbers of levels fused, or localization of levels instrumentation level (Table 1). Similarly, the prevalence of preoperative motor deficits was not associated with EMG or SSEP events. However, patients with EMG events had significantly longer operative times of 240 min (IQR 210-255 mi) vs 210 min (IQR 195-240 min) in patients without EMG events (P = 0.004).

Table 1.

Patient Characteristics Stratified by EMG Event and SSEP Event.

	EMG event (n = 118)	No EMG event (n = 283)	P-value	SSEP event (n = 18)	No SSEP event (n = 383)	P-value
Age [years]	63 (55-70)	64 (53-70)	0.545	65 (60-71)	63 (53-70)	0.221
Sex (female %)	48.3	49.1	0.882	38.9	49.4	0.386
BMI [kg/m2]	27.5 (25.1-31.5)	28.3 (25.1-32.0)	0.334	29.5 (25.9-33.5)	28.1 (25.1-31.8)	0.218
ASA			0.304			1.000
I	2 (1.7%)	10 (3.5%)		0 (0%)	12 (3.1%)
II	106 (89.8%)	238 (84.1%)		16 (88.9%)	328 (85.6%)
III	10 (8.5%)	35 (12.4%)		2 (10.5%)	43 (11.3%)
Diabetes (%)	11 (9.3 %)	33 (11.7%)	0.495	4 (22.2%)	40 (10.3%)	0.123
Preop motor deficit	21 (17.8%)	61 (21.6%)	0.395	2 (11.1%)	79 (20.9%)	0.548
OP time [min]	240 (210-255)	210 (195-240)	0.004	225 (180-240)	210 (195-245)	0.892
EBL [ml]	350 (250-600)	300 (200-500)	0.159	675 (300-1000)	300 (200-600)	0.007
No. levels fused	1 (1-2)	1 (1-2)	0.670	1 (1-2)	1 (1-2)	0.899
Instrumentation
L2/3	3 (2.5%)	10 (3.5%)	0.610	0 (0%)	13 (3.4%)	0.545
L3/4	23 (19.5%)	39 (13.8%)	0.149	2 (11.1%)	60 (15.7%)	0.454
L4/5	79 (66.9%)	193 (68.2%)	0.807	13 (72.2%)	259 (67.6%)	0.683
L5/S1	59 (50.0)	147 (51.9%)	0.723	10 (55.6%)	196 (51.2%)	0.716
PMD (≥1 grade)	8 (6.8%)	14 (4.9%)	0.463	3 (16.7%)	19 (5.0%)	0.068
PMD (≥2 grades)	1 (0.8%)	4 (1.4%)	0642	5 (1.3%)	0 (0%)	0.794

EMG = electromyography; SSEP = somatosensory evoked potentials; BMI = body mass index; ASA = American Society of Anesthesiologists Classification; OP time = time in operating room; EBL = estimated blood loss. No. levels fused = number of levels/segments fused; PMD = postoperative new motor deficit of = 1 or ≥2 motor grades. Bold type indicates statistical significance, defined as p < 0.05.

18 patients had SSEP events (4.5%), of which 9 were transient signal loss with full recovery until the end of the procedure and 9 were persistent. When comparing patients with and without SSEP events (n = 18 vs n = 383), the SSEP group had significantly higher EBL, with a median of 675 mL vs 300 mL (P = 0.007) (Table 1). All other demographic and surgical characteristics, including operative time, were similar between the 2 groups. Only 1 SSEP event resulted in a change to intraoperative management, based on our retrospective chart review (see Illustrative Case 1).

PMD vs No PMD

PMDs were identified in 22 of 401 patients (5.5%). Seventeen patients experienced a loss of 1 strength grade, while 5 had a loss of 2 strength grades. The extensor hallucis longus (EHL) was the most commonly affected muscle group (18 of 22 cases), including 14 isolated EHL deficits and 5 with concomitant involvement of other muscle groups. The second most commonly affected muscle was the tibialis anterior (TA) (n = 4), followed by the hamstrings (n = 3). Patient age, sex, and BMI were similar between those with and without a PMD, but patients with a PMD were more likely to be classified as American Society of Anesthesiologists (ASA) class III (22.7% vs 10.6%) and less frequently as ASA class II (68.2% vs 86.8%) (P = 0.039) (Table 2). There were no significant differences in levels instrumented, number of fused segments, or operative time between groups. PMD was also not associated with the presence of a preoperative neurologic deficit. However, PMD was associated with higher EBL, with a median of 475 mL in the PMD group vs 300 mL in those without PMD (P = 0.023).

Table 2.

Patient Characteristics Stratified by PMD Versus No PMD.

	PMD (n = 22)	No PMD (n = 379)	P-value
Age [years]	64 (53-70)	62 (54-70)	0.972
Sex (female %)	45.5	49.1	0.741
BMI [kg/m2]	28.3 (24.9 – 30.7)	28.1 (25.1-31.9)	0.998
ASA			0.039
I	2 (9.1%)	10 (2.6%)
II	15 (68.2%)	329 (86.8%)
III	5 (22.7%)	40 (10.6%)
Diabetes	5 (22.7 %)	39 (10.3%)	0.070
Preop motor deficit	2 (9.1%)	80 (21.1%)	0.275
OP time [min]	210 (184-240)	210 (195-250)	0.345
EBL [ml]	475 (300-650)	300 (200-600)	0.023
Number of levels fused	1 (1-2)	1 (1-2)	0.859
Instrumentation
L2/3	0	13 (3.4%)	0.377
L3/4	1 (4.5%)	61 (16.1%)	0.145
L4/5	14 (63.6%)	258 (68.1%)	0.665
L5/S1	14 (63.6%)	192 (50.7%)	0.236
IONM event	10 (45.5%)	121 (31.9%)	0.188
EMG event	8 (6.8%)	14 (4.9%)	0.463
SSEP event	3 (15.8%)	19 (5.0%)	0.043

PMD = (new) postoperative new motor deficit of ≥1 motor grade; BMI = body mass index; ASA = American Society of Anesthesiologists Classification; OP time = time in operating room; EBL = estimated blood loss. No. levels fused = number of levels/segments fused; EMG = electromyography; SSEP = somatosensory evoked potentials. Bold type indicates statistical significance, defined as p < 0.05.

Of the 22 PMDs, 12 had resolved during the patient’s hospital stay, and 4 were improving, with 6 persisting until discharge. All 4 patients who were improving during the hospital stay ultimately achieved full recovery: 3 by the first postoperative visit at 6 weeks, and 1 by the visit at 3 months. Among the 6 patients whose PMD persisted through discharge, 4 eventually demonstrated full motor strength on outpatient follow-up (1 at 6 weeks, 1 at 3 months, 1 at 6 months, and 1 at one year). Two patients had persistent deficits at their last follow-up at 2 years. One had a 4/5 unilateral EHL weakness with normal sensation, while the other had unilateral 4/5 weakness in both EHL and TA, along with decreased sensation in the L5 distribution. Despite the persistent weakness, the patient was very satisfied with the surgical outcome, as both radicular and back pain had completely resolved, and maintained an active lifestyle.

EMG and SSEP Test Accuracy and Predictive Value

No significant association was found between EMG events and PMD (P = 0.463) (Table 3). Sensitivity and specificity of an EMG event to detect a PMD was 36.4% and 71.0%, respectively. PPV and NPV of an EMG event were 6.8% and 95.1%, respectively. Of the 18 patients with SSEP events, 3 developed a PMD (15.8%). All of these 3 had persistent signal loss until the end of the procedure. The other 6 cases with persistent loss in SSEP did not show any postoperative sensorimotor deficit.

Table 3.

Crosstabulation of EMG Events and PMD.

	PMD	No PMD	P-value
No EMG event	14 (63.6%)	269 (71.0%)	0.463
EMG event	8 (36.4%)	110 (29.0%)

PMD = (new) postoperative motor deficit of ≥1 motor grade; EMG = electromyography.

The presence of a PMD was significantly associated with SSEP events (P = 0.043) (Table 4). The sensitivity and specificity of SSEP events to detect a PMD was 13.6% and 96.0%, respectively. PPV and NPV were 16.7% and 95.0%, respectively. For persistent SSEP signal loss, sensitivity was 13.6%, specificity 98.4%, PPV 33.3%, and NPV 95.2%. In multivariable logistic regression analysis, EMG events were not significant predictors of PMD (Table 5). SSEP events demonstrated a significant association with PMD in models adjusted for age and number of levels fused, with an odds ratio of 3.84 (95% CI: 1.01-14.53, P = 0.048) and 3.85 (95% CI: 1.02-14.45, P = 0.046), respectively. (Table 6). Persistent SSEP signal loss was a strong predictor of PMD, with odds ratios of 9.87 (95% CI: 2.27-43.00, P = 0.002) when adjusted for age, and 10.41 (95% CI: 2.38-45.59, P = 0.002) when adjusted for number of levels fused (Table 7).

Table 4.

Crosstabulation of SSEP Events and PMD.

	PMD	No PMD	P-value
No SSEP event	19 (86.4%)	364 (96.0%)	0.043
SSEP event	3 (13.6%)	15 (4.0%)

PMD = (new) postoperative motor deficit of ≥1 motor grade; SSEP = somatosensory evoked potentials. Bold type indicates statistical significance, defined as p < 0.05.

Table 5.

Association Between EMG Events and Postoperative Motor Deficit: Multivariable Logistic Regression Adjusted for Age and Number of Levels Fused.

	Odds Ratio	Lower 95% CI	Higher 95% CI	P-value
EMG	1.39	0.57	3.42	0.468
Age	1.01	0.97	1.04	0.918
EMG	1.39	0.57	3.42	0.468
No. Levels	0.83	0.38	1.78	0.625

CI = Confidence intervals. EMG = electromyography; SSEP = somatosensory evoked potentials.

Table 6.

Association Between SSEP Events and Postoperative Motor Deficit: Multivariable Logistic Regression Adjusted for Age and Number of Levels Fused.

	Odds Ratio	Lower 95% CI	Higher 95% CI	P-value
SSEP	3.84	1.01	14.53	0.048
Age	0.99	0.96	1.04	0.990
SSEP	3.85	1.02	14.45	0.046
No. Levels	0.82	0.38	1.79	0.624

CI = Confidence intervals. EMG = electromyography; SSEP = somatosensory evoked potentials. Bold type indicates statistical significance, defined as p < 0.05.

Table 7.

Association Between Persistent SSEP Signal Loss and Postoperative Motor Deficit: Multivariable Logistic Regression Adjusted for Age and Number of Levels Fused.

	Odds Ratio	Lower 95% CI	Higher 95% CI	P-value
SSEP	9.87	2.27	43.00	0.002
Age	1.00	0.96	1.04	0.942
SSEP	10.41	2.38	45.59	0.002
No Levels	0.76	0.33	1.72	0.509

CI = Confidence intervals. EMG = electromyography; SSEP = somatosensory evoked potentials. Bold type indicates statistical significance, defined as p < 0.05.

Illustrative Case Examples

To complement the statistical findings, we present 2 illustrative cases from our cohort that highlight the clinical relevance and interpretive challenges of intraoperative neuromonitoring events in posterior lumbar fusion surgery.

Case 1

A 66-year-old female presented with neurogenic claudication and bilateral L4 radicular pain without preoperative sensorimotor deficits. After failure of conservative treatment, preoperative imaging revealed a dynamic grade 1 degenerative spondylolisthesis at L3/4 with associated severe central canal stenosis. The patient underwent decompression laminectomy and posterior instrumented fusion with posterior lumbar interbody fusion (PLIF) at L3/4 (Figure 2). Intraoperatively, free-run EMG activity was noted in the bilateral lower extremities, and 2 episodes of intermittent SSEP attenuation were recorded (Figure 3). The first event occurred during the surgical approach and was suspected to be related to patient positioning, although baseline SSEP signals were initially intact. In response, the surgical plan was adjusted: decompression was performed immediately and before instrumentation, which reversed the usual operative sequence. The SSEPs returned back to baseline on the left and were partially recovering on the right within 30 min, as a second attenuation occurred during instrumentation and PLIF cage insertion. SSEP signals began to improve and returned to baseline without further intervention, and no changes to instrumentation were made. The patient awoke without new motor or sensory deficits, and lower extremity strength remained 5/5 postoperatively. Intraoperative and postoperative radiographs confirmed appropriate hardware placement.

Figure 2.

MRI and Radiographic Imaging of Illustrative Case 1. (A) Preoperative axial and sagittal T2-weighted MRI imaging showing a degenerative spondylolisthesis at the level L3/4 with severe central stenosis (top left and right) and facet joint diastasis (bottom left). (B) Anteroposterior and lateral radiographs confirming appropriate implant positioning of the pedicle screw construct and posterior lumbar interbody fusion cage.

Figure 3.

Intraoperative Neuromonitoring of Illustrative Case 1. (A) IONM report and (B) IONM traces demonstrating the transient loss of bilateral lower extremity SSEPs. Tibial Nerve SSEP Recordings Are Shown.

Case 2

A 69-year-old female presented with back pain and right-sided L4 and L5 radicular symptoms unresponsive to conservative management. Neurologic examination revealed normal sensation and 5/5 strength in all lower extremity muscle groups. Imaging demonstrated degenerative lumbar scoliosis with asymmetric disc collapse at L4/5 and L5/S1, and degenerative spondylolisthesis at L5/S1 causing right-sided foraminal stenosis. She underwent selective decompression and posterior instrumented fusion from L4 to S1 with PLIF (Figure 4). Throughout the procedure, free-run EMG remained stable without abnormal discharges, and SSEPs of both upper and lower extremities were pristine and did not show any signal alterations (Figure 5). Despite uneventful IONM, the patient developed a new left-sided partial foot drop on postoperative day 1, with 3/5 in TA and 2/5 in EHL. Postoperative CT and MRI confirmed correct hardware placement, adequate decompression, and no evidence of compressive hematoma or nerve impingement. The deficit was attributed to intraoperative transient nerve root inflammation or neuropraxia not detected by IONM. She was treated with intravenous dexamethasone and oral gabapentin, with gradual improvement in motor function. By postoperative day 7, strength improved to 4/5 in TA and 3/5 in EHL. At 6 and 12 weeks postoperatively, both muscles were 4/5. At 6 months TA strength returned to 5/5, with persistent 4/5 EHL weakness. Full strength was regained at her 1-year follow-up and maintained on all subsequent visits.

Figure 4.

MRI and Radiographic Imaging of Illustrative Case 2. (A) Preoperative MRI. Top left (T2) and bottom left (T1) parasagittal view showing the foraminal stenosis at L4/5 and L5/S1 on the right in place of multilevel degenerative disc disease (T2 midline sagittal on right) (B) Anteroposterior and lateral radiographs confirming appropriate placement of pedicle screws at L4-S1 and PLIF cages.

Figure 5.

Intraoperative Neuromonitoring of Illustrative Case 2. (A) IONM report and (B) IONM traces demonstrating stable SSEPs and no EMG abnormalities throughout the case. Tibial Nerve SSEP Recordings Are Shown.

Discussion

This study analyzed IONM recordings, including EMG and SSEP, in a cohort of 401 patients undergoing instrumented PLIF for degenerative conditions. The key findings include:

• • Free-run EMG events were common (29.4%), but SSEP events were rare (4.0%), with only 3 patients (0.7%) in the entire cohort having attenuated signals until the end of the procedure.

• • SSEP events were associated with new postoperative motor deficits, but their sensitivity and positive predictive value were low.

• • EMG events were not associated with postoperative motor deficits.

• • These findings suggest that routine use of intraoperative SSEP and EMG monitoring in degenerative spine surgery may not be justified, and that IONM application should instead be tailored to individual patient and surgical factors.

This single-institution study adds to the existing literature on the diagnostic value of IONM in spine surgery, with a particular focus on open posterior fusion procedures for degenerative conditions. A recent analysis of a New York statewide database reported a marked increase in the use of IONM, rising from 79 cases in 2007 to 6201 in 2018. ³ However, in their multivariable analysis, the use of IONM did not reduce the incidence of neurological deficits, suggesting that routine application may not be warranted. Our single-center study offers a more granular evaluation of the utility of IONM in degenerative spine surgery. At our institution, SSEP and free-run EMG are consistently monitored in all lumbar fusion cases. Additional modalities, such as triggered EMG or transcranial MEP, may be employed based on interdisciplinary discussions. In our findings, EMG events occured in 29.4% of cases but were not associated with the development of neurologic deficits. Conversely, SSEP events were rare, observed in only 4.0% of patients, yet showed a positive association with PMD. Systematic reviews and meta-analyses by Chang et al. and Reddy et al. suggest that SSEP, transcranial MEP, and triggered EMG possess predictive value for postoperative neurologic deficits; however, each modality has shown only moderate to poor sensitivity.^19-21 In contrast, a multi institutional study by Wilent et al. ²² reported 100% sensitivity and 98% specificity of transcranial MEPs for detecting postoperative motor deficits. However, the incidence of PMD in their study was 0.%, with 14 out of 4386 cases affected, which is lower than what has been reported in other series. Notably, all deficits were functionally significant, such as foot drop or heel drop in their study. This contrasts with the findings from the meta-analysis by Reddy et al., ²⁰ which reported MEP sensitivities of 29% for transient changes and 47% for persistent changes.

In our instrumented PLIF cohort, the rate of patients with PMD was 5.5%, similar to previous studies ranging from 6.6% to 7.4%,^23,24 but most of the PMDs in our study were loss of only 1 motor strength grade with the EHL being the most commonly affected muscle. The incidence of EMG changes during surgery in almost one-third of patients in our study is consistent with prior studies demonstrating EMG activity in 40% of patients undergoing lumbar fusion surgery. ²⁵ Further, the weak correlation between intraoperative EMG abnormalities and new PMD observed in this study (sensitivity of 36.4%) aligns with existing literature. Although brief spikes in EMG activity is not expected to result in neurologic injury, it may reflect the extent of spinal nerve manipulation and retraction during surgery, which could theoretically lead to neuropraxia and postoperative motor deficits. A meta-analysis by Alvi et al. found a pooled sensitivity of 49.6% for EMG events during lumbosacral surgery, along with a high false-positive rate of 39.6% for motor deficits. ²⁶ The high number of false positives and low positive predictive value in our study suggest that intraoperative EMG, while sensitive to nerve root irritation, is less effective in identifying clinically meaningful nerve injury.

SSEP events, in contrast, showed a statistically significant association with PMD. The odds of developing a postoperative motor deficit were approximately ten times higher in patients with persistent SSEP signal loss. However, the sensitivity was only 13.6%, also limiting its utility as a screening tool. Previous studies have reported conflicting evidence to support SSEP as monitoring during spinal fusion surgeries.^19,27,28 Previous meta-analyses have reported higher pooled sensitivities, such as 67.5%, and high specificities, such as 96.8%, with postoperative neurologic changes being 22 times more likely to occur in patients with intraoperative SSEP alerts.^26,29 Thus, while intraoperative SSEP events may be rare, they are more likely to correlate with true postoperative neurologic deficits.^19,26

The role of IONM during lumbar spinal fusion remains a subject of debate. While several societies acknowledge its usefulness, there is no consensus that IONM constitutes a standard of care in low-risk procedures.^30-32 Some large database studies have shown reduced neurologic complication rates with IONM use, though these findings are not consistently supported by meta-analyses.^33,34 Our findings suggest that routine IONM using SSEPs and EMG is not justified in low-risk degenerative spine surgery. Instead, its application should be individualized. In our cohort, 18 patients experienced an SSEP event, but only 3 developed a PMD, yielding a positive predictive value of 16.7%. Intraoperative management was altered based on IONM findings in only 1 patient (Case 1), in whom decompression was performed before screw insertion due to persistent SSEP changes. Whether this patient would have developed a PMD without IONM remains uncertain. Given the low sensitivity and otherwise limited actionability of SSEP changes, the likelihood that intraoperative SSEP events meaningfully alters management in low-risk lumbar fusion procedures appears limited.

Although the test accuracy of EMG and SSEP was limited in our cohort, their use may still help identify potential nerve injuries in patients at risk during more complex procedures, such as long-segment fusions spanning the thoracolumbar junction. ²⁴ Prior studies have shown that EMG offers higher sensitivity but lower specificity, whereas SSEP demonstrates low sensitivity but high specificity. ²⁴ In cases involving high-risk maneuvers or patients with elevated risk of nerve injury, we recommend the use of multimodal monitoring, including MEPs, which have been shown to offer higher diagnostic accuracy. ³⁵ Furthermore, because the tibial nerve used in standard SSEPs contains fibers from L4 to S1, focal injury to a single nerve root may go undetected. This may explain why, in Case 2, no signal alterations were observed intraoperatively. The patient likely experienced irritation of a single nerve root, presumably the left L5, which standard SSEP monitoring may have missed. Incorporating saphenous nerve SSEPs, with stimulation below the knee, provides the only lower extremity SSEP capable of selectively detecting L4 nerve root injury and has been shown to effectively identify femoral nerve damage. ³⁶ However, no equivalent selective SSEP exists for the L5 or S1 nerve roots.

IONM also has drawbacks. In our study, patients with EMG events had longer operative times, averaging 30 minutes more than those without. SSEP events were associated with higher EBL, possibly due to ischemic changes from reduced perfusion of nerve roots. ³⁷ Employing meticulous hemostatic techniques during lumbar spinal fusion procedures may help lower the risk of SSEP events and reduce the likelihood of PMD by maintaining adequate neural perfusion. Alternatively, longer operative times and increased blood loss may simply reflect more complex pathology, such as advanced central or foraminal stenosis, which may inherently carry a higher risk of IONM changes. Despite their improved diagnostic performance, MEPs also have notable limitations. These include the risk of tongue bite injury, potential disruption of surgical workflow due to stimulation protocols, prolonged operative times, and the requirement for more restrictive anesthesia management. ¹²

Another important consideration in the use of IONM is its medicolegal role. A survey of spine surgeons reported that medicolegal concerns were the most common reason for utilizing IONM. ⁸ Given the significant legal implications of neurologic injury, the documentation provided by IONM can be crucial, offering a timeline of intraoperative events. ⁸ However, it remains unclear whether such documentation provides any legal protection in the event of litigation related to a postoperative neurologic deficit. Over the years, IONM has featured in several malpractice cases involving lumbar fusion surgery, with some arguing it represents the standard of care depending on geographic region. While the details of IONM use in these cases are often unclear, the variability reinforces the need for standardized protocols regarding its appropriate application and interpretation.

More broadly, it is important to recognize that although intraoperative alerts are intended to prevent neurologic injury, there are currently no validated protocols on how to respond to them. ³⁸ Our study demonstrates that these alerts are not always predictive of postoperative deficits, yet they frequently require immediate intraoperative decision-making. Responses may include checking patient positioning and electrode placement, adjusting anesthesia, managing blood pressure and perfusion, revising implants, extending decompression, or performing a wake-up test. Each of these actions carries potential risk, and the decision is further complicated by the possibility of false-positive alerts, as demonstrated in this study. The lack of standardized, evidence-based guidance underscores the need for further research. In our view, establishing clear protocols that define when IONM is indicated and how to respond to alerts would enhance patient safety and improve consistency in spine surgery practice. Universal application of IONM without clear criteria may not be effective and could lead to unnecessary interventions.

Future studies should prioritize the development and validation of response strategies for IONM alerts, evaluating both their immediate impact and long-term patient outcomes. In addition, integrating IONM findings with intraoperative observations and clinical judgment may offer a more comprehensive and individualized approach to surgical decision-making.

Limitations

The limitations of this study primarily arise from its retrospective design. Despite standardized protocols for neurologic examinations at our institution, there is a possibility of variability or inaccuracies in examination or documentation for some patients. Likewise, while IONM is documented following standardized procedures, the assessment may still be subject to the neurophysiologist’s subjective judgment. Sensory and reflex examinations are important components of perioperative neurologic assessments, but due to inconsistencies in documentation, this study focused solely on motor deficits to enhance reliability. Another limitation is that SSEPs do not directly evaluate motor pathways and, in theory, have limited association with postoperative motor deficits. However, SSEP changes may still reflect iatrogenic effects on the exiting nerve roots, which can contribute to motor dysfunction. Additionally, while various IONM modalities exist, this study included only SSEP and free-run EMG, as these represent the standard of care for degenerative lumbar procedures at our institution. Therefore, our findings may not be generalizable to other modalities such as transcranial or triggered MEPs, which are not routinely used for surgeries below the conus. Furthermore, the occurrence of IONM events may vary between surgeons based on intraoperative preferences, particularly regarding the degree of muscle relaxation used. Greater muscle relaxation may reduce EMG responsiveness, resulting in fewer detected events. Similarly, the incidence of postoperative motor deficits may be influenced not by IONM itself, but by how the surgical team responds to alerts. Thus, both the detection of IONM events and the development of PMDs can vary across surgeons and surgical practices, which may introduce variability in outcomes that this retrospective study was not able to capture.

Conclusions

In posterior lumbar interbody fusion, SSEP events are associated with postoperative motor deficits, whereas EMG events are not. However, the overall test accuracy of IONM in predicting neurologic deficits remains limited. In the absence of elevated neurologic risk, routine SSEP and EMG monitoring may have limited value during posterior lumbar interbody fusion and should be selectively applied in cases where the risk of neurologic injury is elevated and where signal changes are more likely to be actionable. In such scenarios, consideration should be given to incorporating additional IONM modalities that offer greater sensitivity to neurologic compromise, such as MEPs.

Ethical Statement

Ethical Approval

The hospital’s institutional review board (IRB) approved the conduct of the study (#2022-0108).

Informed Consent

Written and informed consent to participate in this study and to publish their data was obtained from all individual participants included in the study.

ORCID iDs

Marco D. Burkhard https://orcid.org/0000-0003-1501-1952

Gisberto Evangelisti https://orcid.org/0000-0002-5185-4482

Franziska C. S. Altorfer https://orcid.org/0000-0001-5497-6808

George Gorgy https://orcid.org/0009-0006-0961-4975

Alexander P. Hughes https://orcid.org/0000-0001-7293-9672

Andrew A. Sama https://orcid.org/0000-0002-3014-7287

Data Availability Statement

The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.*