Outcomes after cervical transforaminal epidural steroid injections vs cervical interlaminar epidural steroid injections: A propensity matched comparison study utilizing a large national database Authors: Andrew R Stephens, Gabriella H Wozniak, Ramzi El-Hassan, Hemant Kalia Published: 2026 ABSTRACT AbstractBackgroundCervical transforaminal and interlaminar epidural steroid injections (CTFESI and CILESI) are commonly used to treat cervical radicular symptoms. There is limited literature comparing the two procedures with some concerns of increased risk of complications after CTFESI.ObjectiveCompare the rates of cervical spine surgery, post-procedure opioid use, and complications after CTFESI and CILESI.MethodsRetrospective propensity matched cohort comparison utilizing a large global database was performed of adult patients with diagnostic codes for cervical radiculopathy undergoing CTFESI or CILESI. Each cohort was matched for age, sex, race/ethnicity, BMI and obesity codes, nicotine use, mental health diagnoses, chronic cardiovascular, pulmonary and renal diseases, pre-procedural opioid use, and prior spine surgery. The primary outcome was the rate of eventual anterior cervical decompression and fusion, total disc replacement, and/or neuroforaminotomy. Secondary outcomes included post-procedure opioid use and major (stroke, myocardial infarction, paralysis, and seizure) and minor (headache, pallor, syncope, flushing) complications.ResultsThe rate of cervical spine surgery was higher for CTFESI compared to CILESI. Post-procedure opioid use was also higher for the CTFESI. Compared to CILESI, CTFESI was associated with greater risk of stroke, seizures, myocardial infarction and minor complications. There was no difference in risk of paralysis.ConclusionThis study demonstrates that patients undergoing CILESI were less likely to use opioid medications post-procedurally, have less major and minor complications and lower likelihood of having surgery as compared to CTFESI. These differences may be due to sampling biases between the two groups that were part of the clinical decision process in selecting the approach.Keywords: Cervical radiculopathy, Cervical transforaminal epidural steroid injection, Cervical interlaminar epidural steroid injection, Surgical rates, Complications Introduction Cervical radiculopathy is a common cause of pain treated by musculoskeletal, spine, and pain providers [1]. For patients with pain refractory to conservative care, epidural steroid injections can improve symptoms and function allowing greater participation in rehabilitation efforts [2]. Both cervical transforaminal and interlaminar epidural steroid injections (CTFESI and CLIESI) are commonly performed by interventional spine and pain physicians to treat cervical pathology. Though superiority of each has been postulated, no work to date has led to a conclusive preference of either approach. Literature is limited in quantity, with few high-quality studies [3]. Overall, the findings are mixed [3,4]. Many providers employ each with some preferring CTFESI to target single-level pathology and CILESI if multilevel pain generators are identified. There have also been reports of major adverse events after cervical spine injections in case studies [[5], [6], [7], [8], [9], [10], [11], [12], [13]], but comparison trials and larger retrospective cohort studies have not demonstrated this increased risk [[14], [15], [16]]. More work is needed to determine what advantages, if any, either approach offers to patients. Study purpose The purpose of this study was to describe outcomes after CTFESI and CILESI using a large national database. The primary aim was to describe rates of cervical spine surgery. The secondary aims were to compare risk of major and minor complications after the procedure. An additional secondary aim was to compare the rates of post-procedure opioid use. This study's primary null hypothesis was that there would be no differences in surgical rates between the two procedures. The secondary null hypothesis was that rates of post-procedure complications and opioid use would not differ. Methods Database To accomplish the aims, TriNetX was utilized. TriNetX was founded in 2013 and is a large global database and includes over 200 participating health care centers across 30 countries. TriNetX contains de‐identified data, with analytic capabilities and is General Data Protection Regulation, Health Insurance Portability and Accountability Act, and General Personal Data Protection Law compliant [17]. The TriNetX platform allows researchers with real‐time access to regularly updated electronic health record data across all participating health care centers. TriNetX houses patient encounter data from approximately 2007 [18]. All ICD-10 and CPT codes are linked to over 150 million de-identified patient data. TriNetX has been used to evaluate outcomes in the spine patient population [[19], [20], [21], [22]]. Patient populations Given that a database with de-identified patient data was used, IRB approval was not necessary though the authors have previously obtained IRB approval to evaluate outcomes for CTFESI and CILESI at their home institution. Patient cohorts were obtained by querying the database for patients with ICD-10 codes for cervical radiculopathy and either CTFESI (CPT 64479) or CILESI (CPT 62321). The first unique injection for each individual patient was included in the analysis. Utilizing TriNetX internal software, the CTFESI and CILESI cohorts were matched for patient age, sex, race/ethnicity, BMI and obesity codes, nicotine use, mental health diagnoses, chronic cardiovascular (coronary artery disease, diabetes, hypertension etc), pulmonary (chronic obstructive pulmonary disease, asthma) and renal diseases (chronic renal failure), pre-procedural opioid use, and prior spine surgery. The database is limited in the ability to review imaging findings (level(s) of pathology), non-coded clinical information such as pain severity, reported functional limitations, and neurologic deficits. These factors, which likely are involved in deciding between approaches and the ultimate need of surgery, were unable to be controlled for within the propensity score matching. Within the TriNetX platform researchers can perform propensity score matching directly on the TriNetX website. The embedded statistical platform runs logistic regression based on user-specified variables to obtain a list of propensity scores and uses 1:1 greedy nearest-neighbor PSM to obtain the matched cohort. Table 1 demonstrates demographic data representing the propensity matched cohorts. There were no significant differences between matched variables. Table 1Demographics of matched CTFESIa and CILESIb cohorts.Table 1NCTFESI = 26,388CILESI = 26,388VariableMean/NumberSD/%Mean/NumberSD/%Standardized Mean DifferencesAge5512.754.912.90.007Race White20,09581.3%20,24681.9%0.022 Black19257.8%19968.1%0.016 Asian4031.6%3871.6%0.000 Other8523.4%7202.9%0.040 Unknown12985.3%12385.0%0.019 American Indian or Native Alaskan1030.4%910.4%0.000 Native Hawaiian or Pacific Islander340.1%320.1%0.000Sex Male11,75747.6%11,54946.7%0.025 Female12,83051.9%13,03752.2%0.008Pre-procedure Spine Surgery16686.8%14545.9%0.052Mental Health Disorders12,93052.3%12,70851.4%0.025Pre-procedure Opioid Use17,08069.1%16,93868.5%0.018Nicotine Dependence468219.0%460518.6%0.014Hypertension987340.0%948738.4%0.046Pulmonary Heart Disease7733.2%6812.8%0.033Atrial Fibrillation9603.9%8913.6%0.022Heart Failure7543.1%6882.8%0.025Cardiomyopathy3891.6%3701.5%0.011Aortic Aneurysm100.0%100.0%0.000Peripheral Vascular Disease7683.1%6932.8%0.025Chronic Obstructive Pulmonary Disorder16586.7%15346.2%0.029Emphysema6222.5%5972.4%0.009Bronchiectasis1350.5%1350.5%0.000Chronic Kidney Disease13575.5%12775.2%0.019Diabetes354514.3%336713.6%0.029Overweight or Obesity569023.0%545322.1%0.030BMI < 202441.0%2310.9%0.015 20-29.9263710.7%250910.2%0.023 30-30.98853.6%7623.1%0.039 31-31.98453.4%7493.0%0.032 32-32.97843.2%7142.9%0.025 33-33.97643.1%6802.8%0.025 34-34.97182.9%6522.6%0.026 35-35.97242.9%6662.7%0.017 36-36.96642.7%6102.5%0.018 37-37.95852.4%4922.0%0.039 38-38.95272.1%4751.9%0.020 39-39.94832.0%4151.7%0.031 40-44.99713.9%9103.7%0.015 45-49.94551.8%4281.7%0.011 50-59.92401.0%2300.9%0.015 60-69.9410.2%350.1%0.037 70+190.1%180.1%0.000aCTFESI: Cervical transforaminal epidural steroid injection.bCILESI: Cervical interlaminar epidural steroid injection. Demographics of matched CTFESIa and CILESIb cohorts. CTFESI: Cervical transforaminal epidural steroid injection. CILESI: Cervical interlaminar epidural steroid injection. Outcome measures The primary outcomes assessed were cervical spine surgery within 1 year post-procedurally and at final follow-up. Twenty years, was selected to reflect final follow-up as data entered into TriNetX begins at about 2007. Cervical spine surgeries evaluated included patients undergoing total disc replacement (CPT 22856), anterior cervical decompression and fusion (ACDF, CPT 22551) and/or neuroforaminotomy (CPT 63045). Surgeries were selected in TriNetX using applicable CPT codes. For the secondary outcomes, opioid analgesic medications were selected utilizing RX medical codes. TriNetX RX medical codes include categories of medications prescribed (antibiotics, antiepileptics, opioids etc) and specific medications. In this study, the categorical ‘opioid’ RX code was selected rather than selecting specific opioid medications. Opioid prescriptions within 6 months post-procedure were analyzed. RX codes are based on prescription sent to the pharmacy. The database is unable to determine the dose, number, and duration of use. Rate of major (stroke, seizures, paralysis, and myocardial infarction) and minor complications (headache, pallor, syncope, and flushing) diagnoses on the day of or day after the procedure were compared between the two matched cohorts. Statistical analysis A propensity matched comparison of the primary (cervical spine surgery) and secondary outcomes (major/minor complications and opioid use) were selected as outcome endpoints. Standardized mean differences were calculated between matched items. Outcomes were compared with odds ratio using the Measures of Association and Number of Instances. Statistical significance was assessed at P < 0.05 and 95% confidence interval (CI) were calculated. All statistical analyses were performed using the TriNetX platform. Results Prior to matching there were 29,216 CTFESI and 95,614 CILESI for a total of 124,830 cervical epidural injections performed during the study time period. For CTFESI, dexamethasone was used for 64% of cases, betamethasone was used for 12%, methylprednisolone was used for 9%, and triamcinolone was used for 9% of cases. For CILESI, dexamethasone was used for 49% of cases, betamethasone was used for 9%, methylprednisolone was used for 22%, and triamcinolone was used for 14% of cases. After successful propensity matching (Fig. 1), 26,388 patients in each cohort were included in the analysis. Our study demonstrated that patients were more likely to undergo any cervical spine surgery within 20 years after CTFESI than after CILESI (N = 3849 vs 2130; 16.1% vs 8.9%, p < 0.001). Patients undergoing CTFESI were more likely than CILESI to undergo ACDF (N = 3035 vs N = 1847; 11.5% vs 7.0%, p < 0.001), neuroforaminotomy (N = 844 vs N = 343; 3.2% vs 1.3%, p < 0.001), and TDR (N = 660 vs N = 317; 2.5% vs 1.2%, p < 0.001). A larger percentage of patients who underwent CTFESI compared to CILESI underwent surgery within 1 year: any (N = 3378 vs N = 1794; 12.8% vs 6.8%, p < 0.001), ACDF (N = 2507 vs N = 1267; 9.5% vs 4.8%, p < 0.001), neuroforaminotomy (N 528 vs 317; 2.0% vs 1.2%, p < 0.001) and TDR (N = 475 vs N = 290; 1.8% vs 1.1%, p < 0.001). The odds ratios are presented in Table 2. Distribution of propensity scores before and after matching. Cohort 1 – cervical transforaminal epidural steroid injection. Cohort 2 – cervical interlaminar epidural steroid injections. Table 2Odds of a Patient Undergoing Cervical Spine Surgery after CTFESIa compared to CILESIb.Table 2SurgeryOdds Ratio95% Confidence Interval20 Years Post-ProcedureAnyCTFESI vs CILESI2.01.9, 2.1Anterior Cervical Decompression and FusionCTFESI vs CILESI1.71.6, 1.8NeuroforaminotomyCTFESI vs CILESI2.52.2, 2.8Total disc replacementCTFESI vs CILESI2.11.8, 2.41 Year Post-procedureAnyCTFESI vs CILESI2.01.9, 2.2Anterior Cervical Decompression and FusionCTFESI vs CILESI2.11.9, 2.3NeuroforaminotomyCTFESI vs CILESI1.71.4, 2.0Total disc replacementCTFESI vs CILESI1.71.4, 2.0aCTFESI: Cervical transforaminal epidural steroid injection.bCILESI: Cervical interlaminar epidural steroid injection. Odds of a Patient Undergoing Cervical Spine Surgery after CTFESIa compared to CILESIb. CTFESI: Cervical transforaminal epidural steroid injection. CILESI: Cervical interlaminar epidural steroid injection. Post-procedure opioid use was significantly less for patients who had a CILESI compared to CTFESI (N = 9077 vs N = 10,159; 34.4% vs 38.50%, p < 0.001). The risk of stroke codes was higher for CTFESI compared to CILESI (N = 106 vs 53; 0.4% vs 0.2%) with an odds ratio of 2.3 (95% CI 1.6, 3.2; p < 0.001). Myocardial infarction codes were higher in CTFESI than CILESI (N = 34 vs N = 11; 0.13% vs 0.04%) with odds ratio of 3.4 (95% CI 1.7, 6.9 p = 0.0003). Seizures were significantly higher in the CTFESI cohort (N = 53 vs N = 26; 0.2% vs 0.1%) with an odds ratio of 1.7 (95% CI 1.1, 2.7; p = 0.02). Minor complications occurred more frequently in the CTFESI group compared to the CILESI (N = 475 vs N = 158; 1.8% vs 0.6%) with an odds ratio of 3.0 (95% CI 2.5, 3.6; p < 0.001). There was no difference in the risk of paralysis (N = 15 vs N = 13; 0.06% vs 0.05%; p = 0.11). There were no deaths in either cohort. Discussion Our primary finding was that post-procedure cervical spine surgery was less commonly utilized after CILESI compared to CTFESI. Within the first year, the overall rate of cervical spine surgery was 6.8% for patients undergoing CILESI and 12.8% for patients undergoing CTFESI. In the 20-year post-procedural interval the overall rate of cervical spine surgery was 8.9% in the CILESI cohort, and 16.1% in the CTFESI cohort. This is consistent with prior literature [14,23,24]. The eventual need for surgery has been reported to be as high as 18% for CTFESI [25]. This may be explained due to the fact that CILESI were more likely to be performed with particulate steroids. These differences can be explained by the fact that CTFESI is commonly used as a tool for surgical planning [26,27]. Our secondary finding demonstrated CILESI was associated with less post-procedure opioid prescriptions. There is little literature evaluating narcotic use after cervical spine injections. In this study, chart review could not be conducted to confirm reasons for opioid continuation after injection. This makes it difficult to extrapolate opioid-use as post-procedure outcome measure. In the setting of a large study population increased opioid use does represent a trend towards higher levels of pain. The correlation may be a result of the interventions themselves, or factors that lead to the decision to perform a CTFESI or CILESI. For example, a single dominant pathology, such as a large extrusion with severe nerve compression at a single-level, may more often lead to a decision of a CTFESI rather than a CILESI. This type of pathology may also be more likely to be offered a surgery given that the offending pathology is clear. While this is intuitively apparent, there are no studies evaluating trends of choice between CTFESI or CILESI relative to specific pathology, nor are there studies evaluating the outcomes of these interventions based on specific pathology. Provider training and procedure practices may also have a strong influence on both the procedure being offered, and the utilization of opioid prescriptions in managing cervical radicular pain. The difference in opioid-use between the groups, could be a result of post-surgical opioid use. The rate of opioid use after orthopedic surgeries is as high as 98% in the immediate post-operative period [28], and new chronic opioid use in previously opioid-naive patients after cervical fusion has been shown to be around 12% [29]. If the higher surgical rate in the CTFESI cohort is factored into opioid-use, it may be that the two cohorts are comparable. Future studies will have to be conducted to control for this. There are several well designed studies assessing outcomes of CTFESI and CILESI. Levin et al. demonstrated that more patients undergoing CILESI experienced greater pain relief than CTFESI [30]. McCormick et al. demonstrated that more patients undergoing CILESI experience improvement in reported pain and function compared to CTFESI [14]. Other studies have shown equivocal outcomes between the two cohorts. Lee and Lee demonstrated similar outcomes between CTFESI and CILESI [31]. In a systematic review by Lee et al. they found no significant difference in pain reduction between CTFESI vs CILESI [3]. Again, none of these studies evaluated the degree of pathology on advanced imaging to the outcomes themselves, especially in the setting of a comparison between CTFESI and CILESI. This is likely an important factor in outcomes of either CTFESI or CILESI, as it has been demonstrated in the lumbar spine that the degree of nerve root compression is directly correlated with the expected efficacy of lumbar transforaminal epidural steroid injection [32] as well as the propensity to progress to surgical intervention. In our matched cohorts, this study found that CILESI had lower risks of major complications. It should be noted that the overall risk of complications for both procedures was very low and the absolute difference between the two cohorts was small. Prior single cohort and comparison studies demonstrated minimal to no complications after CTFESI or CILESI [[14], [15], [16],21,30,31,33]. Our comparative study is insightful given the large number of patients included. It is limited, however, as there is no means to verify if the diagnostic codes were related to the procedure or were incidental. Although potential demographic and comorbidity confounders were controlled for in the propensity matching, the database is limited in the ability to determine the nuance of these diagnoses. For example, although two patients may both have a diagnosis of hypertension, there is no way of knowing if the hypertension is well controlled. The matching may balance the cohorts by diagnostic codes but the patient with uncontrolled hypertension would be more likely to experience complications than the patient with controlled hypertension. The inclusion of myocardial infarction as a complication highlights this point. There is no known mechanism that supports CTFESI or CILESI directly being a cause of myocardial infarction. Given this, it is likely that the CTFESI group had higher risk factors for cardiovascular events than the CILESI group. This may account for at least a portion of the higher rates of stroke seen within the CTFESI group. There is, however, a greater risk for vascular uptake with CTFESI and if performed with particulate steroid lead to an increase risk of stroke. Recent large cohort studies have demonstrated no major complications including stroke with use of dexamethasone [15,16,33]. This study was not designed to guide interventionalists on approach, given that this is a highly patient-specific decision that should be made on a case-by-case basis. However, it does serve to further elucidate the relative risks between these two procedural approaches and provide further context for a more informed decision. There are inherent limitations when utilizing a large database retrospectively, largely stemming from the inability to confirm and provide clinical context. Though the database was queried for diagnostic codes for cervical radiculopathy, the exact pathology (disc herniation, neuroforaminal stenosis etc.) was unable to be evaluated. Furthermore, there was no means to verify if the diagnostic codes, such as those identifying complications, were related to the procedure or were incidental. Imaging findings, severity of pain, functional limitations, and neurologic deficits were unable to be evaluated and controlled for in the propensity score matching. These factors may confounders for which a physician is more likely to select one procedure over another and be more likely to lead to surgical outcomes. Additionally, we were unable to assess patient-specific outcomes of pain or function. We were also unable to be account for the impact of repeat injections on the outcomes. Epidural and surgical levels and indications were unable to be ascertained to determine the direct correlation with the pathology for which the cervical epidural steroid injections were performed. Conclusion This study demonstrates that cervical spine surgery is significantly less common after CILESI than CTFESI and that patients were prescribed fewer opioid narcotics after CILESI. Complications rates were also lower after CILESI as compared to CTFESI. These findings may be due to sampling biases inherent in the clinical decision process of the procedural approach, which were unable to be controlled for in the propensity matched comparison. Funding statement This study was not supported by funding. References