The decision to proceed with any intervention not only requires understanding of the benefits but also the risks. Complications reported with interlaminar cervical epidural steroid injections include dural puncture, bloating, nausea and vomiting, vasovagal reaction, facial flushing, fever, nerve root injury, pneumocephalus, epidural hematoma, subdural hematoma, stiff neck, Cushing’s syndrome, transient paresthesias, hypotension, respiratory insufficiency, transient blindness, epidural abscess, paralysis, cord injury, and death [39–51].
Complications reported with transforaminal cervical epidural steroid injections include neck pain, transient increased radicular pain, nausea, vasovagal reaction, dural puncture, non-specific headache, transient lightheadedness, dyspepsia, fluid retention, transient global amnesia, vertebral artery injury, paralysis, cord infarction and cerebellar infarction, and death [52–57].
Blinded and fluoroscopically guided interlaminar CESI
Waldman  prospectively reported upon complications with blind interlaminar C5-6 or C6-7 epidural steroid injections utilizing loss of resistance technique on 215 subjects receiving 790 injections. Complications were recorded immediately at 6 weeks by the pain management physician or nurse. Two individuals suffered dural puncture with headache requiring blood patching. There were 3 vasovagal reactions and one superficial infection.
Botwin et al.  retrospectively reported complications with fluoroscopically guided C6-7 or C7-T1 interlaminar epidural steroid injections on 157 subjects receiving 345 injections. Data were obtained at 24 h from the ambulatory surgical center questionnaire and 3 week physician follow-up. There was 6.7% neck pain, 4.6% non-positional headache, 1.7% insomnia, 1.7% vasovagal reaction, 1.5% facial flushing, 0.3% fever, and 0.3% dural puncture incidence. The overall rate of complications was 16.8%.
Fluoroscopically guided transforaminal CESI
Ma et al.  reviewed records of 1,036 cervical transforaminal epidural steroid injections in 844 subjects. Immediate complications were recorded by the radiologist performing the procedure. If any complications occurred after discharge, the subjects were told to contact their referring physician. Hence, complications that occurred after discharge are at risk of being underreported. The authors’ report complications occurred in 14 subjects (1.66%). These included headache/dizziness (0.59%), transient pain or weakness (0.71%), hypersensitivity reaction (0.12%), transient global amnesia (0.12%), vasovagal reaction (0.12%), and wrong site injection (0.36%).
Huston et al.  performed a prospective, controlled study with independent interviewer of lumbar and cervical selective nerve root injections on 151 subjects who received 306 injections. The control group was 60 subjects with similar demographics and spinal complaints that did not undergo an intervention at time of interview. Procedural complications were recorded by the interventionalist. Immediate, 1 week and 3 month complications were recorded by an independent interviewer. Of the cervical group, there were 89 cervical selective nerve root injections performed on 37 subjects. There were no major complications. There was one dural puncture but the subject did not develop a headache and no treatment was needed. In the cervical group immediate complications were increased pain at injection site 22.7%, increased radicular pain 18.2%, lightheadedness 13.6%, increased spine pain 9.1%, non-specific headache 4.5%, and nausea 3.4%. One week follow-up compared to the control group had significance for increased pain at injection site (P = .001), non-specific headache (P = .019), and non-spinal headache (P = .002). At 3 months follow-up 2 subjects complained of increased neck pain. One would repeat the injections again. The other subject would not repeat the injection. This subject did have complete resolution of radicular pain from a herniated disk but was not satisfied because of persistent neck pain.
Interlaminar versus transforaminal CESI
Incidence of dural puncture with interlaminar CESI ranges from 0.25% to 2.00% [21, 58, 60] and transforaminal CESI 1.12% . With dural puncture the procedure is discontinued to avoid subarachnoid instillation of local anesthetic or corticosteroid. Instillation of local anesthesia could result in spinal anesthesia with respiratory depression, hypotension, and syncope . Additionally, subarachnoid corticosteroid injection has been hypothesized to cause arachnoiditis [62, 63]. Celestone Chronodose has been evaluated in sheep and found to result in arachnoiditis at suprapharmacologic levels . Epidural injections of triamcinolone and methyprednisolone in animal studies did not result in arachnoiditis or nerve root injury [65, 66].
After dural puncture, there is still the concern of spinal headache. However, dural puncture does not always result in a spinal headache [53, 58, 60, 61]. If a headache does occur, treatment may consist of strict bed rest, hydration, analgesics, and caffeine . For severe or persistent headache interlaminar or transforaminal blood patching may be performed [58, 68].
The risk of dural puncture may be higher with an interlaminar than a transforaminal approach. Cryomicrotome studies of the cervical epidural space report absence of the posterior epidural space above C7-T1. In another study, the posterior epidural space at C5 has been reported at 1–1.5 mm . Furthermore, there is an absence of the interspinous ligament in the cervical spine  along with half of specimens being deficient in the midline of the ligamentum flavum. Hence, blind injections utilizing loss of resistance technique may inadvertently puncture the dura. With transforaminal CESI, utilizing multi-planar imaging should avoid inadvertent dural puncture. On the AP view, the needle should not be advanced beyond the 6 o’clock (midline) position of the lateral mass. Dural puncture typically occurs by advancing the needle too far in the oblique plane. The technique section will discuss the need for frequent multi-planar imaging while advancing the spinal needle.
Non-positional headaches occurred with cervical selective nerve root injections in 4.5% and was found to be statistical significant when compared to a control population . For fluoroscopically guided interlaminar CESI, non-positional headache occurred in 4.6%. The rate is comparable between the two techniques. These headaches have been attributed to alterations in CSF pressure which is related to rapid injection and/or higher volumes [70, 71]. These headaches are typically transient and respond to oral analgesics. These headaches can be minimized by injecting slowly and the use of lower volumes.
Transient blindness after epidural steroid injection has been reported primarily with lumbar interlaminar and caudal injections. However, Kao  reported a case of impaired visual acuity following a C6-7 interlaminar ESI with 4 ml lidocaine and 2 ml triamcinolone. Five days after the second CESI, the patient developed headache, vomiting, neck stiffness, and decreased left visual acuity. A serous retinal detachment was found. The detachment resolved after 4 weeks but still with impaired vision. By 4 months, the patient reported normal vision. The cause of the retinal detachment is unknown. The author hypothesizes the following possibilities: (1) stress during the injection resulted in neuroendocrine alteration, (2) corticosteroid may have damaged the choroicapillaris with altered permeability, and (3) the corticosteroids may have elevated cathecholamine levels resulting in constriction of choroids vessels. If due to the steroids, then any utilization of corticosteroid whether oral, intra-muscular, or epidural with systemic absorption could potentially result in retinal detachment. Another possibility is increased CSF pressures which could explain the headache and vomiting the patient experienced. While the volume was 6 ml, rapid instillation could increase CSF pressures.
Bilateral retinal hemorrhages with transient blindness after caudal and lumbar ESI has been reported [72–75]. The retinal hemorrhages were felt to be related to increased CSF pressure created by rapid instillation and large volumes of medication injected with ESI [72, 73]. Volumes of 20 ml can increase CSF pressures . Volumes of lidocaine and bupivicaine have ranged from 2 to 20 ml with cervical interlaminar ESI [21, 58, 76]. Larger volume CESI’s with rapid instillation should be avoided.
Cushing’s syndrome has been reported after interlaminar CESI with 60 mg of methylprednisolone . The syndrome resolved after 12 months. Cushing’s syndrome has also been reported after lumbar ESI. While Cushing’s syndrome has not been reported with transforaminal ESI, the occurrence is probably related to the systemic absorption of the corticosteroid as opposed to the technique of delivery. Additionally, systemic affects of corticosteroids can result in hyperglycemia in diabetics and needs to be closely monitored.
Infection is always a risk whenever the skin is punctured. Huang et al.  present a case of epidural abscess following CESI. The patient initially presented with increased neck pain and chills. The patient subsequently developed left arm pain, paresthesia, and weakness but was neurologically intact in the lower extremities. An epidural abscess from C4 to C6 was seen on gadolinium-enhanced MRI. The patient underwent surgical decompression, irrigation, and debridement. The patient was placed on intravenous antibiotics. By 7 months post-operative, the patient regained baseline neurologic function . Whether one technique is more predisposed is unknown. The procedure should be done with sterile technique. Additionally, the interventionalist should not touch the spinal needle tip. Both techniques are at risk for infection.
Nerve root injury and transient paresthesias have been reported after interlaminar CESI [46, 60]. After a blind interlaminar C5-6 ESI, nerve root injury was postulated as the initial injury resulting in complex regional pain syndrome . For cervical selective nerve root injections transient increased radicular pain occurred in 18.2% but with no episodes of nerve root injury .
Paralysis: interlaminar and transforaminal CESI
Epidural hematoma after fluoroscopically guided C5-6 interlaminar CESI requiring surgical evacuation has been reported . The patient had near full recovery of paralysis. Another case occurred within half and hour after painful paresthesia with the introduction to a Touhy needle at the C6-7 interspace . The patient had incomplete recovery prior to surgery with high-dose intravenous methylprednisolone. The patient also required immediate surgical evacuation of the hematoma. Puncturing of the epidural venous plexus is the probable etiology. Whether the risk of epidural hematoma and subsequent quadriplegia is greater with interlaminar versus transforaminal CESI is unknown. However, with tranforaminal injections the needle is not placed directly into the spinal canal. Puncturing a radicular vein or artery within the foramen may be less likely to result in thrombosis formation with cord compression as the needle is not within the spinal canal. However, there may be other serious consequences of vascular puncture with transforaminal injections.
Brouwers  reported a case of tetraplegia following a right C6 transforaminal ESI with bupivicaine/triamcinolone mixture. The paralysis was consistent with anterior spinal artery syndrome. Another case was reported of cord infarction after left C6 transforaminal ESI with the patient suffering incomplete tetraplegia . In another case, digital subtraction revealed puncture of a radicular artery terminating with several branches in the region of the spinal cord despite correct technique with a transforaminal cervical epidural injection . The procedure was done with live fluoroscopy when injecting contrast. Fortunately, the procedure was abandoned without sequelae after noting the vascular flow. Tiso et al.  reported a case of cerebellar infarction after a C6 transforaminal CESI with bupivicaine/triamcinolone mixture. Intra-vascular injection of particulate steroid resulting in embolic occlusion through the vertebral artery with subsequent infarction was postulated as the cause. Pathology revealed bilateral cerebellar and occipital cortex infarction, thromboembolism of the leptomeningeal artery.
Light microscopy of steroid particulate size found up to 50 µm particle size for methylprednisolone, triamcinolone, betamethasone sodium phosphate and acetate, dexamethasone, and betamethasone sodium phosphate . Only betamethasone sodium phosphate had no particles greater than 50 µm. Less than 5% of particles were greater than 50 µm for methylprednisolone, betamethasone sodium phosphate and acetate, and dexamethasone. Methylprednisolone and triamcinolone had a tendency to coalesce into large aggregates of greater than 100 µm, which could result in sludging . Particle and aggregate size is relevant when one considers the diameter of the artery system: artery >50 µm, metarteriole 20–50 µm, arteriole 10–15 µm, and capillary 5–8 µm. Based upon dimensions, methylprednisolone and triamcinolone sludge could block smaller arteries and arterioles and result in ischemia . Dexamethasone and betamethasone sodium phosphate would be better choices to avoid vascular occlusion. However, betamethasone sodium phosphate is not available commercially. While betamethasone sodium phosphate can be obtained through a compounding pharmacy, it is not recommended. Meningitis with subsequent death has been linked to compounding pharmacies making betamethasone sodium phosphate . Hence, at this time, the corticosteroid recommended for transforaminal CESI is commercially available dexamethasone . Furthermore, there was no statistical significance difference between outcomes with dexamethasone versus triamcinolone for cervical radicular pain .
Temporary paralysis was reported after a diagnostic C7 transforaminal injection with 0.8 ml 2% lidocaine . Sixty-seconds after injection the patient felt unwell. In the next 2–3 min the patient developed quadraparesis that resolved after 20 min. The neurologic deficits were consistent with anterior spinal artery syndrome. The authors recommend injecting local anesthetic first and separate from the steroid .
The current hypothesis of tetraplegia following transforaminal epidural steroid injection relates to arterial injection of corticosteroid into a radiculomedullary artery with subsequent occlusion. The radiculomedullary arteries are major feeders to the anterior spinal artery. These arteries can arise anywhere from C3 to C8. Occlusion of a radiculomedullary artery can result in spinal cord infarction—anterior spinal artery syndrome. While vertebral artery puncture should be avoided by adhering to correct technique, Baker et al.  demonstrated a correct technique which can still result in injection into a radicular artery. When performing a transforaminal CESI, the following are modified recommendations to minimize the chance of radicular artery injection of corticosteroid : (1) once the needle is in place, tubing should be connected to the spinal needle hub and syringe. This is to prevent inadvertent needle movement when attaching different syringes of injectate. (2) Injection of contrast under live fluoroscopy—evaluating for any vascular flow. Digital subtraction imaging may be of benefit.  If no vascular flow with contrast, then proceed to injection of local anesthetic under live fluoroscopy. The patient is then monitored for at least 90 s. This is based upon the case report of Karasek and Bogduk  in which initial symptoms of radicular artery injection of local anesthetic occurred at 1 min and paralysis in 2 min. After the time interval expires, the patient should be asked about a metallic taste, peri-oral numbness, auditory changes, agitation suggesting local anesthetic toxicity. Additionally pin-prick is tested in the hands and lower extremities along with movement of the hands and feet. If no signs are present to suggest anterior spinal artery syndrome or intra-arterial injection, then proceed to step 4. (4) Before injecting dexamethasone, confirm needle placement has not changed utilizing fluoroscopy. Then slowly inject dexamethasone.
The occurrence of tetraplegia has opened the debate of whether transforaminal or interlaminar CESI should be performed. However, tetraplegia has occurred after interlaminar CESI. Cord injury following two cases of fluoroscopically guided interlaminar C5-6 ESI occurred in sedated patients . The authors suggested the sedation did not allow the patients to respond to cord penetration by the needle. However, needle penetration of the cord in alert patients can be without pain or paresthesias . While cord puncture may not be painful, injection of contrast agent into the cord produced pain . Tetraplegia followed a fluoroscopically guided C6-7 interlaminar ESI . While the cause is unknown, the authors hypothesized ischemic injury to the cord . Bromage and Benumof reported a case of paralysis consistent with anterior artery syndrome in an individual undergoing spinal epidural anesthesia at T12 . While the cause was unknown, the case report raised caution in epidural injections performed above the termination of the cord .
The cord is at risk of puncture with interlaminar CESI. The ligamentum flavum in the cervical region was found to be deficient in the midline in half of specimens . Furthermore, the interspinous ligament is absent in the cervical spine . When utilizing a loss of resistance technique, lack of resistance from absence of the interspinous ligament and unfused ligamentum flavum could lead to inadvertent dural and cord puncture. Performing the procedure under fluoroscopic guidance should help avoid inadvertent cord penetration.
Both interlaminar and transforaminal CESI’s have case reports of catastrophic neurologic complications and death. The incidence of these complications while felt to be rare is unknown. Derby et al.  surveyed instructors of the International Spine Intervention Society. While not scientific, the survey revealed no major complication, no paralysis or death, in 4,389 interlaminar and 1,579 transforaminal injections. A national databank of complications would be helpful in determining the incidence of rare complications. This would be helpful in developing safer techniques and in appropriate consenting of patients undergoing CESI’s. This system would be beneficial for the continuous improvement of patient care. Unfortunately, the current medical malpractice legal climate in the United States creates challenges in developing such a system.