Postoperative complications after anterior cervical discectomy and fusion in patients with versus without carotid artery stenosis

Abstract

Background

Carotid artery stenosis (CS) may elevate risk of stroke related to anterior cervical discectomy and fusion (ACDF) due to manipulation of the carotid artery and alteration in flow dynamics. Prior studies assessing this potential association are limited by small sample size and assessment of in-hospital stroke only.

Methods

A retrospective cohort study was conducted to compare 90-day adverse events following ACDF in patients with versus without CS. Adult patients (age >17 years) without a history of stroke who underwent isolated 1-3 level ACDF were identified in the 2010-2023 PearlDiver M170 database. Patients with CS were matched 1:4 to patients without CS based on age, sex, and Elixhauser comorbidity index (ECI). Adverse outcomes, emergency department (ED) visits, and readmissions in the 90 postoperative days were identified and compared between groups with univariable and multivariable analyses, the latter controlling for age, sex, and ECI. The timing of stroke following surgery was also plotted on a Kaplan-Meier survival curve and compared by log-rank test. A secondary analysis was conducted comparing CS patients with and without a history of revascularization procedure.

Results

Among patients in the database undergoing ACDF, 17,772 with and 423,401 without CS were identified. Following matching, final cohorts consisted of 16,888 CS patients and 67,355 matched controls. CS was associated with greater odds of cardiac events (OR=1.74, p<.001) and stroke (OR=1.49, p<.001), but not other assessed adverse outcomes. The incidence of stroke continued to diverge for the 2 groups out to 90-days postoperatively. Of the CS patients, 2.8% had prior carotid revascularization. These patients had no difference in odds of stroke when compared to patients who had not undergone revascularization (p=.522)

Conclusions

ACDF patients with CS were found to be at significantly elevated odds of cardiac events and stroke (adverse outcomes associated with atherosclerotic burden), but not other assessed adverse outcomes. Surveillance of related symptoms through the perioperative period is importantly considered.

Introduction

The incidence of perioperative stroke in patients undergoing cervical discectomy is estimated to be 0.13% in-hospital [1]. During anterior cervical discectomy and fusion (ACDF), the surgical approach necessitates retraction of the carotid sheath [2,3], which may alter blood flow [3] and potentially affect cerebral perfusion [4,5]. These effects may contribute to stroke with ACDF, particularly if there is pre-existing carotid disease. Globally, 1%–2% of adults have underlying asymptomatic carotid stenosis (CS), with prevalence increasing in older age groups [6]. Given that CS is the underlying cause of 8%–15% of ischemic strokes, [7] the potential association of CS with stroke and ACDF procedures is of interest.

Case reports have described perioperative stroke in patients undergoing anterior cervical surgeries [5,[8], [9], [10], [11], [12], [13], [14], [15], [16], [17]]. Most patients identified had either known or undiagnosed CS, [5,9,10,14,15] with others having varied vascular risk factors, [8,12,16] or vascular anomalies, [17] and only one identified with no apparent risk factors [13]. All reported strokes occurred ipsilateral to the surgical approach, [5,[8], [9], [10], [11], [12], [13], [14], [15],17] with the exception of bilateral thalamic strokes attributed to vertebral artery thrombosis [16]. Many were directly or indirectly attributed to retraction or manipulation of the carotid artery [5,[12], [13], [14], [15],17].

In a retrospective multicenter case series of 17,625 patients undergoing anterior cervical surgery, Härtl et al. reported zero patients who experienced carotid artery injury or stroke, though noting case reports such as those above [18]. Chugtai et al. [19] utilized a 2009-2013 New York statewide database to assess patients with and without carotid stenosis and their relative risk of stroke following ACDF. Only 61 patients with carotid stenosis were identified, but they found a significantly higher incidence of postoperative stroke during hospitalization (6.6% vs. 0%), though none of these strokes were associated with in-hospital mortality. Notably, this study included only a small sample of patients with carotid stenosis and investigated in-hospital stroke only, potentially failing to characterize stroke outside of the acute perioperative period. Further, this study did not stratify for prior carotid revascularization procedures.

Chugtai et al. [19] also considered the risk of nonstroke adverse events for ACDF patients with versus without CS. Of nonstroke adverse events, the only significant differences identified were elevated incidences of acute renal failure (27.9% vs. 4.9%) and sepsis (18.0% vs. 4.9%) in patients with relative to without CS. While other studies of ACDF patients with versus without CS were not identified, hyperlipidemia, a known risk factor for atherosclerosis and development of arterial stenosis, has been associated with myocardial infarction, renal failure, and urinary tract infection/incontinence following ACDF, [20] suggesting the risk of these complications may extend broadly to patients with atherosclerosis.

The present study sought to leverage a large national insurance database to investigate postoperative outcomes of ACDF in patients with relative to without CS. Given the manipulation of the carotid artery with ACDF approach, perioperative stroke was of interest.

Methods

Study population

This retrospective cohort study was conducted using the 2010 to 2023 Q1 M170PearlDiver administrative database, which is widely established in the spine literature [[21], [22], [23], [24], [25], [26], [27]]. As all data accessed in PearlDiver is aggregated and deidentified, our Institutional Review Board (IRB) has determined studies using this database to be exempt from review.

Adult patients undergoing ACDF were identified using Current Procedural Terminology (CPT) codes 22551 and 22554, and codes 22552 and 22585 were used to identify additional levels. Patients undergoing 1-3 level ACDF were included. Patients undergoing any lumbar, thoracic, posterior cervical, or anterior cervical corpectomy or disc arthroplasty procedures on the same day were excluded. Additional exclusion criteria included age <18 years old, having a new diagnosis of spinal infection, fracture, or neoplasm in the 90 days prior to surgery, having a prior International Classification of Diseases (ICD)-9/10 code for stroke or post stroke deficit, and patients not active in PearlDiver for 90 days following index surgery.

ICD-9/10 codes were used to identify patients with carotid stenosis (ICD-10-D-I6521, ICD-10-D-I6522, ICD-10-D-I6523, ICD-10-D-I6529, ICD-9-D-43310). Patient characteristics were abstracted including: age, sex, and Elixhauser Comorbidity Index (ECI, a measure of comorbidity burden) [28]. Patients with and without CS were matched 1:4 based on age, sex, and ECI, using PearlDiver’s MATCH command for exact matching to minimize confounding variables. Patients having undergone prior carotid revascularization (endarterectomy, stenting, or bypass) were identified by CPT codes: CPT-35301, CPT-0075T, CPT-35501, CPT-35506, CPT-35507, CPT-35508, CPT-35509, CPT-35510, CPT-35526, CPT-35601, CPT-35606, CPT-35626, CPT-35642, CPT-37215, CPT-37216, CPT-37217, CPT-37218, CPT-S2211.

Postoperative outcomes

Adverse events in the 90 days postoperatively were identified using ICD codes, using methodology previously described [21,22,26,29,30]. Serious adverse events were noted if one of the following occurred: cardiac event (myocardial infarction or cardiac arrest), sepsis, deep vein thrombosis (DVT), or pulmonary embolism (PE). Minor adverse events were noted if there was one of the following: acute kidney injury (AKI), transfusion, pneumonia, or urinary tract infection (UTI). A procedural adverse event was noted for any of the following: stroke, esophageal perforation, surgical site infection (SSI), dysphonia, hematoma, dysphagia, or wound dehiscence. ICD-9 and ICD-10 codes used to identify stroke are included in Supplemental Material 1. Any adverse event was noted if there was a serious or minor adverse event.

Ninety-day postoperative emergency department (ED) visits were identified using CPT codes (CPT-99281 to CPT-99285) denoting ED level of care. Readmissions were noted based on the presence of any new admission code within ninety days of surgery.

Data analyses

Patient demographics in unmatched cohorts were compared using Student's t-tests or chi-squared tests, as appropriate. These analyses were then repeated after matching.

The incidences of adverse events were compared with chi square tests. Multivariable logistic regression was used to compare the odds of adverse events individually and in aggregate between the matched cohorts of patients with and without CS, controlling for age, sex, and ECI. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were reported.

A Kaplan-Meier survival curve was generated to compare 90-day stroke-free survival by log rank test. Stroke incidence was also compared between patients with CS who had prior carotid revascularization relative to those who had not using chi square tests and multivariable logistic regression.

Statistical analyses were performed using PearlDiver Bellwether software and GraphPad Prism 8. Statistical significance was defined as a p-value <.05, with Bonferroni correction applied to each analysis to account for multiple comparisons.

Results

Study population

Carotid stenosis was noted for 17,772 (4.0%) of 441,173 ACDF patients identified (Table 1). On average, patients with CS were older (64.6 ± 9.5 years vs. 54.9 ± 11.2 years), more likely to be male (51.5% vs. 44.5%) and had a higher comorbidity burden (ECI 6.7 ± 3.8 vs. 3.4 ± 3.1) (p<.001 for all) than patients without CS.

Table 1.

Patient characteristics of adult patients with and without carotid stenosis who underwent anterior cervical discectomy and fusion

	Unmatched sample			Matched sample (4:1)*
	Patients without CS	Patients with CS	p-value	Patients without CS	Patients with CS	p-value
Total	423,401 (96.0%)	17,772 (4.0%)		67,355	16,888
Age (mean ± SD)	54.9±11.2	64.6 ±9.5	<.001	64.2±9.4	64.2±9.4	.801
Sex			<.001			.981
Female	234,607 (55.4%)	8,615 (48.5%)		33,009 (49.0%)	8,274 (49.0%)
Male	188,794 (44.5%)	9,157 (51.5%)		34,346 (51.0%)	8,614 (51.0%)
ECI (mean ± SD)	3.4±3.1	6.7±3.8	<.001	6.4±3.5	6.4±3.6	.498

^⁎Matching criteria included age, sex, and Elixhauser Comorbidity Index (ECI). SD, standard deviation.

Following matching for patient age, sex, and ECI, there were 16,888 patients with CS and 67,355 without. The previously noted differences in patient demographics were no longer significant.

Postoperative outcomes

Between the matched cohorts of patients with and without CS, there were no differences in the aggregated incidences of any adverse events, serious adverse events, minor adverse events, or procedural adverse events in the 90 days following surgery. In terms of individual adverse events, patients with CS had a significantly increased incidence of cardiac event (1.1% vs. 0.6%) and stroke (0.9% vs. 0.6%) (p<.001 for both). There were no significant differences in the incidences of any other assessed individual adverse event between the cohorts. Rates of ED visits and readmissions were also not significantly different (p=.040, p=.980). Univariable comparisons are presented in Table 2.

Table 2.

Univariable analysis of 90-day adverse events among matched cohort of adult patients with and without carotid stenosis following ACDF

	Patients without CS	Patients with CS	p-value
Total (N)	67,355	16,888
Any adverse event	6,240 (9.3%)	1,655 (9.8%)	.034
Serious adverse events	2,053 (3.0%)	575 (3.4%)	.018
Cardiac event	417 (0.6%)	182 (1.1%)	<.001
Sepsis	732 (1.1%)	196 (1.2%)	.435
Deep vein thrombosis	765 (1.1%)	180 (1.1%)	.465
Pulmonary embolism	508 (0.8%)	101 (0.6%)	.037
Minor adverse events	4,552 (6.8%)	1,179 (7.0%)	.311
Acute kidney injury	1,347 (2.0%)	369 (2.2%)	.136
Transfusion	164 (0.2%)	44 (0.3%)	.755
Pneumonia	1,475 (2.2%)	388 (2.3%)	.412
Urinary tract infection	2,407 (3.6%)	617 (3.7%)	.634
Procedural adverse events	5,426 (8.1%)	1,452 (8.6%)	.022
Stroke	421 (0.6%)	157 (0.9%)	<.001
Esophageal perforation	31 (<0.1%)	11 (0.1%)	.423
Surgical site infection	440 (0.7%)	129 (0.8%)	.129
Dysphonia	532 (0.8%)	150 (0.9%)	.220
Hematoma	463 (0.7%)	125 (0.7%)	.494
Dysphagia	4,277 (6.3%)	1,112 (6.6%)	.273
Wound dehiscence	181 (0.3%)	38 (0.2%)	.361
ED Visit	9,124 (13.5%)	2,391 (14.2%)	.040
Readmission	3,807 (5.7%)	956 (5.7%)	.980

Bonferroni correction applied, p<.002 considered significant. Statistically significant values are bolded.

On multivariable logistic regression analysis, controlling for age, sex, and ECI, patients with CS did not have significantly different odds of experiencing any adverse events, serious adverse events, minor adverse events, or procedural adverse events in aggregate (p=.045, .023, .381, .027, respectively). Patients with CS had higher odds of cardiac event (OR=1.74) and stroke (OR=1.49, p<.001 for both) (Table 3 and Fig. 1). There were no significant differences in the odds of any other individual adverse event, ED visits, or readmissions.

Table 3.

Odds of 90-day adverse events among matched cohort of adult patients with relative to without carotid stenosis following ACDF^⁎

	Patients with carotid stenosis [OR (95% CI)]	p-value
Any adverse event	1.06 (1.00, 1.12)	.045
Serious adverse events	1.12 (1.01, 1.23)	.023
Cardiac event	1.74 (1.46, 2.08)	<.001
Sepsis	1.06 (0.90, 1.24)	.470
Deep vein thrombosis	0.93 (0.79, 1.10)	.407
Pulmonary embolism	0.79 (0.63, 0.97)	.029
Minor adverse events	1.03 (0.96, 1.10)	.381
Acute kidney injury	1.09 (0.96, 1.22)	.174
Transfusion	1.06 (0.75, 1.47)	.729
Pneumonia	1.04 (0.93, 1.17)	.455
Urinary tract infection	1.02 (0.93, 1.11)	.686
Procedural adverse events	1.07 (1.01, 1.14)	.027
Stroke	1.49 (1.23, 1.78)	<.001
Esophageal perforation	1.41 (0.68, 2.73)	.323
Surgical site infection	1.17 (0.96, 1.42)	.122
Dysphonia	1.12 (0.93, 1.34)	.213
Hematoma	1.07 (0.88, 1.30)	.483
Dysphagia	1.04 (0.97, 1.11)	.309
Wound dehiscence	0.83 (0.58, 1.17)	.301
ED Visits	1.05 (1.00, 1.10)	.048
Readmission	1.00 (0.93, 1.07)	.941

Bonferroni correction applied, p<.002 considered significant. Statistically significant values are in bold. CI, confidence interval; OR, odds ratio.

^⁎Controlling for age, sex, ECI.

Fig. 1.

Odds of 90-day adverse events for patients with carotid stenosis relative to patients without carotid stenosis following ACDF forest plot depicting the odds of 90-day adverse events for patients with carotid stenosis relative to those without carotid stenosis following anterior cervical discectomy and fusion. Black bars indicate statistical significance (p<.05). ED, emergency department.

Sub-analyses

A Kaplan Meier survival curve of time to stroke in the 90 days postoperatively was generated and compared by log rank test (Fig. 2). A time curve of percentage of patients remaining stroke-free was significantly different in patients with CS relative to patients without CS (p<.001), with the most notable divergence of the curves occurring after day 14 postoperatively.

Fig. 2.

Survival to Stroke Following ACDF Kaplan-Meier analysis of patient stroke-free survival following anterior cervical discectomy and fusion for patients with carotid stenosis relative to those without carotid stenosis. Survival was significantly different between cohorts (p<.001). ACDF, anterior cervical discectomy and fusion; CS, carotid stenosis.

Of 17,772 patients with CS identified undergoing ACDF, prior carotid revascularization had been performed for 498 (2.8%). When comparing ACDF patients with versus without history of prior revascularization procedures, there were no differences in stroke incidence by univariable analysis (p=.593) or in odds of stroke by multivariable analysis (OR=1.31) (p=.522).

Discussion

The current study, to our knowledge, represents the largest cohort of patients with carotid stenosis undergoing ACDF. Prior to matching, the cohort with CS was more likely to be older, male, and have higher comorbidity burden. This is consistent with prior meta-analyses of demographics of patients with CS [25]. After 1:4 matching, there were 16,888 patients with CS and 67,355 without CS identified, each with similar patient demographics. Of note, on univariable and multivariable analysis, patients with CS who underwent ACDF had a statistically significant increase in cardiac (1.1% vs. 0.6%) and stroke events (0.9% vs. 0.6%) postoperatively (p<0.001 for both). Although this is a relatively small increase in risk, it is not negligible and is worthy of consideration given the debilitating outcomes following such events. Further, given contemporary literature reports an annual stroke incidence of approximately 1% in patients with carotid stenosis, [31,32] a 90-day incidence of 0.9% in these patients represents a dramatic elevation from baseline risk.

CS is a well-established risk factor for ischemic stroke [[33], [34], [35], [36]]. Manipulation of the carotid artery during ACDF is known to disrupt blood flow, [3] and could disrupt atherosclerotic plaques which predispose patients to stroke. The timing of stroke risk after ACDF has not been well studied. Prior cohort studies have focused on strokes occurring during initial hospitalization, [1,19] theorizing that most strokes would occur in the acute postoperative period, but potentially failing to characterize strokes occurring later in the postoperative period. While most case reports describe strokes occurring intraoperatively or hours to days postoperatively, one case report describes a stroke 18 days after surgery that was attributed to carotid manipulation, suggesting the “at risk” window may be larger than the initial hospital stay [12]. The currently presented study observed continued divergence in stroke rate extending late into the postoperative period, even after discharge. This finding suggests a persistently elevated risk into the late postoperative period in such high-risk patients which has not been previously characterized, and that prior work focused on only the initial hospital stay may fail to characterize this continued risk. Further research is needed to characterize whether this divergence in odds of stroke is directly related to carotid manipulation, or instead due to the underlying elevation in stroke risk in patients with CS and not strictly procedural.

The presence of CS is strongly associated with atherosclerosis elsewhere, particularly coronary artery disease, which likely explains the study findings of increased cardiac events after ACDF in patients with CS [6,37]. However, there were no significant differences in odds of any other individual or aggregate adverse events, including procedural adverse events. This finding suggests the risk of adverse outcomes in patients with CS is primarily related to atherosclerotic burden, and is not diffuse, in contrary to what was found in prior studies of patients with dyslipidemia and CS which noted elevated risk of general medical complications including acute renal failure, UTI, and sepsis [19,20].

In a sub analysis of CS patients with versus without prior revascularization, no significant difference in stroke rate was noted. One interpretation of this finding would be that revascularization is not protective against stroke. However, according to current guidelines, typically only those with high grade (70%–99%) or symptomatic moderate grade (50%–69%) CS and low surgical risk would be considered for surgery [[38], [39], [40], [41]]. Given the expected overrepresentation of severe symptomatic CS in the cohort having undergone revascularization, this may indeed represent a protective effect, with intervention potentially leading to a return to baseline risk for a patient with lower-grade CS.

A prior case report described ACDF in a patient with no significant collateral blood vessels to the carotid artery territory, in which case somatosensory evoked potential monitoring detected changes suggestive of cerebral ischemia due to carotid compression [42]. These changes lead the operative team to intermittently remove the retractor over the course of surgery so that carotid compression would not be prolonged. The patient had no postoperative neurologic deficits, suggesting that timely recognition of somatosensory evoked potential differences can potentially avert neurovascular complication, and emphasizing the importance of multimodal neuromonitoring in anatomically or vascularly high-risk patients, such as patients with carotid stenosis. Further, to date, there are no studies that quantitatively measure intraoperative carotid compression pressure during anterior cervical approaches. Future work incorporating pressure-sensing retractors, ultrasound-based compression metrics, or correlating perfusion surrogates such as cerebral oximetry with retraction force is necessary to better characterize safe operative techniques in high-risk patients.

As with all database studies, the present study is limited by coding fidelity. An attempt was made to evaluate concordance of CS laterality with stroke laterality. However, the vast majority of ICD-10 codes utilized for stroke did not specify laterality or vascular territory, making this analysis unfeasible. Further, carotid stenosis ICD-10 codes do not specify percent occlusion to allow for evaluation of the impact of degree of stenosis on outcomes. Due the nature of the PearlDiver database, the impact of intraoperative procedural details, postoperative nursing protocols, and patient lifestyle factors could not be investigated, all of which may play a role in the development of adverse events. All patients with prior stroke were excluded from this study to prevent any lingering codes from being detected as new stroke after surgery. As such, we likely then underestimate stroke risk in the CS population presenting to care by removing these high-risk patients, and findings may not be generalizable to these patients.

Conclusion

In summary, ACDF patients with CS were found to be at significantly elevated odds of cardiac events and stroke (adverse outcomes associated with atherosclerotic burden), but not other assessed adverse outcomes. Preoperative screening with consideration for carotid artery duplex ultrasound in high-risk patients and revascularization procedures when indicated by current vascular surgery guidelines may be warranted. Further study investigating post-ACDF stroke risk based on specific CS grade and history of revascularization procedure is needed. Surveillance for stroke-related symptoms throughout the postoperative period is important to consider.

Declarations of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.