THERE ARE A WIDE VARIETY OF POTENTIAL PAIN generators in the spine, including the zygapophysial joints, intervertebral discs, nerve roots, sacroiliac joint, muscles, tendons, and ligaments. Interventional pain management techniques are performed for either diagnostic or therapeutic purposes. Ideally, prior to the procedure a thorough history and physical examination have been performed and a specific diagnosis has been made. The procedures should be target specific and performed in a technically sound manner with proper technique. Current literature suggests that in some cases procedures performed without guidance (i.e., “blind”) that are not target specific may not be better than sham treatments; although there may be indications for such procedures in certain conditions, they will be sparingly covered in this chapter. Other injections that are target specific but do not have credible efficacy data will also not be discussed.

Additionally several key principles must be considered when reviewing the literature for interventional procedures, including the importance of categorical outcomes, appropriate time frame for follow-up, and the effects of heterogeneity and prevalence data on outcomes. These will all be discussed prior to reviewing the general indications for and evidence of efficacy for the various techniques that target a variety of pain generators. Therefore, each section will cover the basic principles and evidence-based outcomes for the respective interventions, including procedures targeting the zygapophysial joints, intervertebral discs, the epidural space, and the sacroiliac joint.

Prior to reviewing the types and efficacy of pain interventions, it is important to understand the method of analysis. Studies have found that as little as 30% of pain relief may be clinically meaningful in chronic pain conditions.¹ Fifty percent of pain relief has been established as what patients consider “much improved” for pain in general and improved quality of life, and is also the minimal clinically important change for radicular pain.^1–3 Subsequently, in the pain literature 50% of pain relief is the most commonly used dichotomous outcome measure.⁴ Studies failing to reach these thresholds do not demonstrate a clinically meaningful outcome, despite at times being statistically significant.

An appreciation of the distribution of the outcome variables is critical in correctly interpreting the data. Unfortunately, a common method to report outcomes from pain studies which utilize the visual analog scale (VAS) pain scores is by comparing the mean pain scores of the groups without regards to the distribution of the data. However, for this to be a statistically valid way to analyze ordinal VAS data, the assumption is that VAS pain scores are normally distributed. Unfortunately, pain scores are rarely, if ever, normally distributed.

Consider that if an intervention is successful in achieving significant relief in a portion of subjects, there will be two distinct groups when analyzing outcome data. One group of responders will be distributed around a low VAS value, and the nonresponders will be distributed around the baseline VAS value, resulting in an overall bimodal distribution. Interpreting the mean data for post intervention patients without regard to the distribution of the data leads to an interpretation error (Fig. 37–1).

Adequate definition of the study patient populations and intervention is also critical. For example, patients can have lumbar radicular pain due to intervertebral disc herniation, spinal stenosis, or even a facet synovial cyst. These heterogeneous conditions have very different natural histories and likely different responses to treatment, yet many studies may combine the outcomes of all three groups of patients in a single study. Similarly, there are significant differences between the types of injections. For instance, an interlaminar epidural injection may disperse injectate dorsally along a path of least resistance to multiple levels, while a transforaminal epidural injection places an injectate directly over an affected nerve both ventrally and dorsally.

Results are unclear when studies combine the outcomes of these different procedures together into a single study. When combining heterogeneous groups of patients together and exposing them to a variety of different interventions, it is unavoidable that any potential significant differences may be diluted out. This type of inappropriate lumping of heterogeneous conditions may lead to misunderstanding and confusion. Ideal studies will employ regression analysis to determine if the relationship between the independent and dependent variables exists despite the effect of confounding variables.

The last consideration in pain intervention literature is the appropriateness of the follow-up interval. Again, consider radicular pain due to disc herniation, which is known to have a favorable natural history and improve over a period of weeks to months. When studying such a condition, a short-term follow-up is appropriate, as extending beyond the period in which natural recovery is likely would clearly fail to show a difference. However, although a short follow-up is appropriate for conditions with a favorable natural history (e.g., trochanteric bursitis), it would be inappropriate to have only a short follow-up for known chronic conditions (e.g., spinal stenosis). Even more challenging is the progressive nature of spine pathology, which makes determining treatment effects challenging.

As outlined earlier, interpreting pain intervention literature is complex and limited by the quality of the research.

The zygapophysial joint (aka Z-joint or facet joint) is a diarthrodial synovial joint that contains a joint space within a fibrous capsule and occurs in the cervical spine from C2–C7 and at all levels of the thoracic and lumbar spine. Z-joints are generally innervated by the medial branches of the dorsal rami above and below the joint space. They are accepted as potential pain generators based on discoveries of facet joint nociceptors: capsular distention of these joints induces pain, and anesthetizing the innervation of the joints relieves this pain.^5–11 Up to 60% of cervical neck pain and 45% of low back pain may be due to the Z-joint.^12–16 The most common cause of Z-joint pain is osteoarthritis.¹⁷ The most common levels affected are C2–C3 and C5–C6 in the cervical spine and L4–L5 and L5–S1 in the lumbar spine.^18–20 Despite the relatively common presence of facet-mediated pain, there are no specific physical exam maneuvers or radiographic findings that can always correctly identify patients suffering from it. Therefore, interventions aimed at not only relieving pain but also correctly diagnosing patients with Z-joint–mediated pain are of great value (Figs. 37–2 and 37–3).

Medial Branch Blocks

Z-joints are generally innervated by the medial branches of the dorsal rami above and below the joint space, though the numbering nomenclature differs in the cervical and lumbar spine, so that in the cervical spine the C5–C6 joint is innervated by the C5 and C6 medial branches of the dorsal rami,²¹ whereas in the lumbar spine the L4–L5 Z-joint is innervated by medial branches of the L3 and L4 dorsal rami.²² Therefore, each facet joint is innervated by two or more adjacent spinal nerves (Fig. 37–4).

This is the basis of medial branch blocks (MBBs), which are diagnostic procedures that involve anesthetizing the medial branches of the dorsal rami (Figs. 37–5, 37–6, and 37–7)

The test is designed to evaluate if a patient’s pain is mediated by the anesthetized nerve. In other words, pain relief from medial branch blocks implies that the anesthetized nerve mediates the patient’s pain and similarly that the nerve innervates the anatomic pain generator, namely (or most often) the zygapophyseal joint.

The purpose of an MBB is to precisely anesthetize the targeted nerve in order to evaluate the patient’s response. Therefore, careful documentation of the patient’s pre- and post-procedure pain level is of utmost importance. In general, the patient must be experiencing his or her typical pain prior to the procedure. After the injection, pain level should be recorded in a pain diary in a systematic fashion at intervals of at least 30 to 60 minutes for the remainder of the day. Patients may also track pain medication requirements and ability or inability to perform tasks that are typically pain limited.

Full technical description of the procedures is available from the Spine Intervention Society.²³ When MBBs are performed according to guidelines, no significant complications have been reported.²³ Technical complications such as thecal puncture only occur if the procedural guidelines are not followed and the needle is grossly misplaced.

Validity of Medial Branch Blocks

When performed with proper technique, medial branch blocks have been shown to be target specific.^24,25 Kaplan et al showed that the lumbar Z-joint could have pain provoked by capsular distention, and the pain could be attenuated by anesthetizing the medial branches prior to repeating the provocative capsular distention in eight of nine subjects.⁷ Medial branch blocks have also been shown to accurately diagnose true zygapophyseal joint pain with minimal false positives if stringent criteria are used. Dual-controlled anesthetic medial branch blocks (discussed later), which provide 100% relief for the expected duration of anesthetic used, are required for a test to be considered positive. The specificity of cervical MBBs is 88% though only 54% sensitive.²⁶ The specificity of dual-controlled blocks in the lumbar spine is not as clearly defined.

Dual Comparative Blocks

Depending on how a positive response is defined, rates of false-positive results from single MBBs have been reported to be as high as 27% in the cervical spine and between 17% and 41% in the lumbar spine.^15,27–29 Many argue that such high false-positive rates when only a single MBB is used renders them invalid.^29–31 To mitigate false-positive responses both dual and triple blocks have been proposed. Dual anesthetic blocks include performing the procedure twice with two different anesthetics of differing duration. A triple block includes a dual block in addition to performing the procedure with saline as a placebo. Triple blocks are not commonly performed for a variety of reasons such as cost, efficiency, and the ethics of performing invasive procedures with placebo medications. Dual comparative blocks are a viable alternative to placebo-controlled blocks.²⁶ A true positive finding with comparative local anesthetic blocks is when a patient reports duration of pain relief that corresponds to the expected duration of action of the anesthetic used; in essence, that the patient experiences pain relief with both blocks but longer relief when the longer-acting anesthetic is used.³² One study has shown that including both concordant (length of pain relief equals expected duration of anesthetic) and nonconcordant (length of pain relief does not match expected duration) methods does not significantly decrease the positive predictive value of the test.²⁶ As such, it is not uncommon clinically to accept pain relief after two medial branch blocks, regardless of duration, as a positive response.

False Positives and False Negatives

Understanding false positives and false negatives is important in the context of MBBs, given that they are a diagnostic test. Patients may have an innate desire for the test to work, predisposing them to reporting a false-positive outcome. Excessive use of anesthetic may anesthetize additional structures and may also lead to a false-positive test, though this can be prevented by using only 0.2 to 0.3 mL of anesthetic. Alternatively, if the patient is not in their usual state of pain prior to the procedure, they have new procedural-related pain after the procedure, or if they have multiple concurrent pain generators, they may inaccurately report a lack of pain relief even if the correct nerve was anesthetized, resulting in a false negative. Technically, the target nerve not being anesthetized whether due to anomalous innervations or nerve location, improper technique, or vascular uptake of anesthetic itself may also result in a false negative.^33,34

MBB as a Predictor of Response to Treatment

The practical use of MBBs is directly tied to the ability of MBBs to predict the likelihood of response to treatment, in this case radiofrequency neurotomy or radiofrequency ablation (RFA). As more stringent criteria are used to define a “positive block,” the likelihood of success with RFA increases accordingly. Few patients will have 100% pain relief to both dual-controlled MBBs; however, of those who are considered positive responders, it is highly likely that their pain is originating from the anesthetized nerve, and as such RFA of that nerve is likely to alleviate their pain. Alternatively, if a patient is only required to have 50% relief to dual-controlled MBBs, many more patients will be considered positive responders. However, some of these patients will likely represent false-positive responses and accordingly may not respond to the ensuing RFA. Both scenarios must be balanced, as neither too many patients undergoing an unneeded radiofrequency neurotomy nor patients being withheld from a potentially beneficial treatment is ideal. In other words, as the criteria for what constitutes a positive MBB becomes more stringent (50% vs. 80% vs. 100%), the likelihood of successful RFA also increases but the number of people proceeding to RFA decreases. Indeed, this has consistently been shown in the literature.^35–39

Conclusion

Medial branch blocks are safe and valid tests that have both diagnostic and therapeutic utility, as they predict response to RFA in the management of zygapophyseal joint pain in both the cervical and lumbar spine. Medial branch blocks are indicated in a patient who has chronic or subacute pain thought to be mediated by structures innervated by the medial branches—in most cases the Z-joints, and confirmation of this will alter management such as proceeding to RFA.

It stands to reason that if Z-joints are innervated by the medial branches of the dorsal rami and that anesthetizing the medial branches results in pain relief arising from the Z-joints, then longer-lasting disruption of the medial branches will result in longer-lasting pain relief. This is the theory behind thermal RFA, also known as radiofrequency neurotomy. Technically, RFA uses a radiofrequency generator to transmit energy, using the body as a resistor to generate heat at the tip of an electrode. The heat causes ionic agitation and friction, protein denaturing, cellular membrane disruption, increased tissue permeability, and finally tissue neurolysis of the targeted nerve. There are three types of RFA: low-intensity RFA, cooled RFA, and pulsed RFA.

The most common and well-studied neurolysis is thermal low-intensity RFA. It is performed by applying a specific temperature to the target constantly for 60 to 90 seconds. With this technique, the probe must be parallel to the targeted nerve, as opposed to perpendicular, to achieve a sufficient lesion size⁴⁰ (Fig. 37–8).

RFA is not curative or permanent, as the cell bodies remain intact and are capable of axonal regeneration. Nonetheless, average relief has been shown to be around 400 days.⁴¹ When symptoms recur after a successful RFA, repeat RFA has been shown to be safe and capable of reinstating pain relief.^41–44

RFA is generally considered safe. In the cervical spine, dysesthesia, dizziness, and hypersensitivity in the distribution of the third occipital nerve occasionally occur.⁴⁵ In the lumbar spine, RFA results in lumbar multifidus denervation.^35,46 The full effect that this has on multifidus function, morphology, and segmental anatomy has not been fully described.⁴⁷

Descriptive studies of Z-joint denervation in the neck appeared as early as the 1970s.⁴⁸ In 1995, Lord et al developed a standardized approach to both the procedure itself and the methodology used to assess it.⁴⁹ Since that time, multiple studies using present technique guidelines have demonstrated effectiveness of RFA in treating facet-mediated pain without any negative studies refuting this evidence.^{35,41–44,36,38,49–53} Some of these studies are further discussed later.

The landmark study by Lord et al in 1995 was a double-blind, randomized, placebo-controlled study that proved the efficacy of cervical RFA by showing that it results in real pain relief and not relief only attributed to placebo effects.⁴⁹ In this study 24 patients were diagnosed with chronic C3–C4 to C6–C7 cervical zygapophyseal joint pain and selected based on response to double-blind, placebo-controlled diagnostic blocks. At 27 weeks, seven patients (58%) in the active treatment group and only one patient (8.3%) in the control group were free of pain. In the treatment group, the median time to return of at least 50% of pain was 263 days compared to only 8 days in the placebo-controlled group. Indeed, using proper technique and proper patient selection, cervical RFA was shown to provide lasting pain relief that is not attributable to placebo effects.⁴⁹ The efficacy of RFA of the C2–C3 zygapophysial joint was less convincing in this early study due to more variable anatomy of the third occipital nerve.⁴⁹ However, Govind et al⁴² later evaluated a revised technique of RFA for the treatment of third occipital headache, using electrodes closely placed at three separate targets, and subsequently achieved successful outcome in 43 of 49 (88%) patients who were selected after a positive response to diagnostic blocks. A study by Dreyfuss et al in 2000 was the first to show that under optimal conditions, properly selected patients experienced significant pain relief with lumbar RFA.³⁵ In this prospective study, patients with zygapophysial joint pain that was relieved by at least 80% by two controlled diagnostic medial branch blocks proceeded to lumbar RFA. Sixty percent of patients had at least a 90% reduction in pain and 87% had at least a 60% reduction in pain lasting 12 months.³⁵

Observational studies have replicated these findings in the clinical setting.⁵⁰ One study evaluated 104 patients from two different practices who were selected for treatment with cervical RFA based on complete pain relief following dual-controlled comparative diagnostic MBBs.⁵⁰ Successful outcome was defined as at least 80% pain relief for 6 months, restoration of all their desired activities of daily living (ADLs), no other health care required for their neck pain, and return to work if they had not previously been working. Even with this stringent definition of success, the average success rate between the two practices was 66%.⁵⁰ Moreover, the median duration of pain relief in Practice A was 17 months and in Practice B 20 months. MacVicar reported on a similarly designed study but this time looking at lumbar RFA in 106 patients. Success was defined as 80% pain relief at 6 months as well as restoration of at least four ADLs, no other health care for their back pain, and return to work if they had not been previously working.⁴⁴ Again, even with these strict criteria for success, Practice A demonstrated a 58% success rate and Practice B demonstrated a 53% success rate, with median duration of complete pain relief after the first successful RF neurotomy of 15 months in both groups. These larger clinical observational studies suggest external validity to the initial studies.

Conclusion

Using present technique guidelines in patients selected via dual-controlled MBBs, radiofrequency ablation is a validated procedure with multiple studies demonstrating effectiveness and without any negative studies refuting this evidence.^{35,41–44,36,38,49–53} RFA provides significant relief in a high percentage of patients, with increasingly more dramatic responses in studies with more stringent criteria applied to the preceding MBBs. While rates of success vary depending on the study, a common generality useful for patient education stems from the initial Dreyfuss et al study, which showed that roughly 60% of patients can expect 80% relief and 80% of patients can expect 60% relief for up to 12 months after lumbar RFA.³⁵ Moreover, symptoms do recur, but RFA is repeatable and often reinstates the relief experienced with the initial procedure.^41–43,50

Diagnostic Injections

Historically, intra-articular injection of the facet joints with local anesthetic has been performed as a diagnostic test for zygapophyseal joint pain. Unfortunately, intra-articular diagnostic blocks have never been validated. Even more, the validity of medial branch blocks (discussed earlier) to diagnose facet-mediated pain has been established. As such, there is little to no role for diagnostic intra-articular injections of the facet joints with current data. In addition to improved diagnostic utility and predictive value, medial branch blocks are used in place of intra-articular anesthetic blocks because they are easier and safer to perform.²³

Therapeutic Injections

There are limited data on the therapeutic utility of intra-articular Z-joint steroid injections (Fig. 37–9).

The first report of using intra-articular steroid injections for relief of cervical zygapophyseal pain was by Dory in 1983, who reported that 9 out of 14 patients had at least 50% pain relief after intra-articular injection of triamcinolone. In only four of nine patients was 50% relief sustained for at least 1 month or longer.⁵⁴ A number of observational cervical spine studies were published over the following decade with mostly similar results. In the lumbar spine, uncontrolled observational studies have shown more variable rates of success rate, ranging between 18% and 63%.⁵⁵ The only randomized study that investigated the Z-joint injections in the cervical spine was published in 1992 by Barnsley et al.⁵⁶ The study showed no difference between the group receiving injections with bupivacaine and the group receiving injections with betamethasone, with both groups demonstrating that less than half of patients in either group achieved 50% relief for at least 1 week. Two additional randomized controlled trials (RCTs) evaluated the efficacy of lumbar Z-joint intra-articular steroid injections. One study evaluated patients who had immediate pain relief from intra-articular anesthetic injection and then randomized them to receive either intra-articular corticosteroid or intra-articular saline.⁵⁷ There were no differences between the groups at 1 and 3 months; however, at 6 months the intra-articular steroid group was statistically more likely to have improved.⁵⁷ Improvement only being seen at 6 months and not at 1 or 3 months is not readily explained by the biochemical properties of steroids, casting question on the findings. The other RCT included a group receiving high-volume (8 mL) injections, and given that the Z-joint can only hold 1 to 2 mL, the validity of this study is questionable at best.⁵⁸ Another randomized study compared the effect of intra-articular Z-joint steroid injection with intramuscular steroid injection and found slightly superior outcomes with respect to physical function and reduction in nonsteroidal anti-inflammatory (NSAID) use in the intra-articular group.⁵⁹ However, the effect size of these results was small, there was no control arm to the study, and all subjects in the intra-articular group received bilateral injections at L3–L4, L4–L5, and L5–S1 as opposed to selecting the joints thought to be most likely involved.

Studies that have used facet uptake on single photon emission computerized tomography (SPECT) scans to select what joints to inject have shown more favorable results. Two such studies have shown statistically significantly better outcomes when compared to injections done on joints that were either negative for uptake on SPECT^60,61 or joints that were selected based on clinical judgment alone.⁶¹ Similarly, another study of patients with low back pain and isolated Z-joint inflammation seen on SPECT imaging found that 61% of patients who had intra-articular anesthetic plus steroid injections into these joints reported 50% relief at 12 weeks compared to only 26% of those who received medial branch perineural steroid and anesthetic injections.⁶²

Clearly, the evidence on the therapeutic utility of intra-articular facet injections is mixed and limited. Some have gone as far as to suggest that intra-articular Z-joint steroid injections are no better than sham injections.⁶³ However, it is likely that there is a subset of the population, currently identified as those with joint inflammation seen on SPECT, that may significantly respond to intra-articular steroids for relief of facetogenic pain. But given the robust evidence in support of RFA for facetogenic pain, the clinical utility of intra-articular steroid Z-joint injections is currently rather limited.

Discography

Intervertebral disc pathology can play role in back pain.⁶⁴ Discs are innervated with nociceptive pain receptors by branches of the sinuvertebral nerves, gray rami communicans, and lumbar ventral rami.^65–68 Discography is the injection of contrast into the nucleus pulposis to evaluate disc morphology^69–71 (Fig. 37–10).

Provocation discography is injection of contrast medium into the nucleus pulposis of a disc in an attempt to re-create and thus diagnose a patient’s pain. If further anatomic confirmation of a positive provocative test is desired, subsequent magnetic resonance imaging (MRI) or computerized tomography (CT) scans of the injected disc can then be pursued. However, often many of the morphologically abnormal discs are not painful with provocation.⁷²

The main purpose of discography is to accurately diagnose the pain generator for potential treatment purposes. However, there are currently no studies available that demonstrate a better response to a given therapy in subjects with discography-proven discs as opposed to those selected by clinicians. Moreover, there are currently limited options for treating discogenic pain, regardless of how it is diagnosed. Some have argued there is a role for discography in surgical planning, though the studies that have evaluated this are equivocal, at best, when viewed collectively.^73–76 Currently, discography does not have any well-defined therapeutic utility associated with it.

Also of concern is the many potential downsides to discography. Discography carries with it an inherent risk of infectious discitis.^23,77 Even more concerning, a 10-year matched cohort study that evaluated asymptomatic patients who underwent three-level provocative discography and MRI evaluation and were compared with matched cohorts who underwent MRI evaluation but no discography found that discs exposed to discography had significantly higher progression toward disc pathology.⁷⁸ Discography also has an inherently high false-positive rate due to its provocation nature, with rates in systematic review being reported as 6% to 10%,⁷⁹ though other studies report values as high as 40% in chronic pain patients.⁸⁰

Therefore, there is no high-level evidence that validates the utilization of discography in the management of discogenic pain; additionally, the long-term detrimental sequelae of discography make its utility questionable at best.

In contrast to the earlier procedures targeted at the Z-joints or intervertebral discs in order to diagnose or treat predominantly axial pain, epidural steroid injections (ESIs) target the spinal nerves and are indicated for the treatment of radicular pain (radiculitis). The most common causes of radicular pain are intervertebral disc herniation and spinal stenosis. Pure mechanical compression of spinal nerves produces paresthesia and motor weakness but not pain,⁸¹ and the degree of nerve root compression does not correlate to pain severity.^82–84 Not surprisingly then, radicular pain itself may be caused primarily, if not exclusively, by inflammation.

Inflammation is an essential component to painful radiculitis.^22,85–87 Inflammatory mediators such as phospholipase A2, prostaglandin E2, leukotrienes, nitric oxide, immunoglobulins, and cytokines such as interleukin-6 and tumor necrosis factor alpha are all involved in the inflammatory component of radiculopathy.^88–90 Various inflammatory markers or cells are required for the dorsal root ganglion to generate the painful discharges in radiculitis,⁹¹ are found within the nucleus pulposus itself, are found in high concentrations at sites of disc herniation,^87,92,93 and are seen on histopathology of nerve root specimens taken from decompression surgery.^87,94 Corticosteroids inhibit phospholipase-2 and leukocyte aggregation at the site of inflammation; prevent degranulation of granulocytes, mast cells, and macrophages; prevent transmission of nociceptive C-fibers, and stabilize ectopic discharge of neuronal membranes.^89,95,96 As such, local administration of corticosteroid can theoretically result in symptom relief.⁹⁷ With this in mind, it is not surprising that the evidence for ESIs is strongest for radicular pain due to disc herniation and less convincing for that due to stenosis. It also reasons, then, that as it addresses a painful inflammatory process affecting a spinal nerve, ESIs are not indicated for axial pain due to other causes.

In fact, a recent study evaluated patients with back, buttocks, or leg pain due to lumbar stenosis seen on imaging and randomized them to receive ESI (interlaminar or transforaminal) of either anesthetic alone or anesthetic plus steroid and concluded that epidural injections of steroids plus lidocaine offered minimal or no benefit compared to lidocaine alone at 6 weeks.¹⁷⁷ This was a well-done study that must be considered when discussing ESIs for the treatment of lumbar stenosis causing low back, buttocks, or leg pain. However, many studies that evaluate ESI for the treatment of radicular pain do not differentiate whether it is due to stenosis or disc herniation, making it difficult to draw clear distinctions between the two in the literature at times. Nonetheless, later in the chapter such distinctions will be noted when available. When such distinctions are available, though, it is evident that the best evidence for ESI is for radicular pain due to disc herniation alone.

ESIs can be performed via a variety of approaches. The interlaminar approach targets the posterior epidural space between the dura anteriorly and the ligamentum flavum posteriorly (Figs. 37–11, 37–12, and 37–13)

The transforaminal approach targets the suspected spinal nerve directly in the neuroforaminal space (Figs. 37–14, 37–15, and 37–16) but also has anterior and subarticular spread. In the lumbar spine, a caudal approach through the sacral hiatus is also possible.

Interlaminar or caudal injections were historically done until the transforaminal approach was later developed for the lumbar spine in 1952 by Robecchi and Capra⁹⁸ and later in 1988 by Bard and Laredo for the cervical spine.⁹⁹ The interlaminar approach results in diffuse spread of injectate that flows along the path of least resistance, often leading to unilateral spread to the side with less stenosis, and rarely achieves ventral spread.¹⁰⁰ The transforaminal approach is now favored as it provides more precise placement of steroid into the specific spinal nerve and its dorsal root ganglion that is suspected to be the cause of pain. Derby et al first postulated the additional superior effect of the transforaminal approach due to its ability to provide a high concentration of injectate directly to the ventral epidural space and to the posterior annulus.¹⁰¹ Ackerman et al later proved that a transforaminal approach was most likely to achieve ventral spread compared to other approaches.¹⁰²

Risks of Epidural Steroid Injections

Complications and Side Effects

Overall incidence of immediate or delayed complications for cervical and lumbar ESIs ranges between 1.6% and 20.0% depending on how they are defined.^103–108 The most common side effects include injection site pain, vasovagal reaction, increased radicular pain, lightheadedness, increased pain caused by direct trauma to the spinal nerve, nausea, nonpositional headache, vomiting, facial flushing, and elevated blood pressure.^97,106–110 Anesthetic medications, contrast, and steroids can all cause allergic reaction. Steroids can also cause myopathy, fluid retention, hypertension, mood abnormalities, menstrual irregularities, hyperglycemia, and iatrogenic Cushing’s syndrome.^111–114

There are also complications that are potentially more serious as a result of the procedure’s proximity to the spinal canal and related structures. Bleeding can result in epidural hematoma and compression of the spinal cord or spinal nerves, though this is quite rare.^115,116 Dural puncture can result in positional headache. If not identified prior to intrathecal administration of anesthetic, it can also result in cauda equina, arachnoiditis, or meningitis.¹⁰³ Infection is another known risk, which can lead to epidural abscess, discitis, osteomyelitis, and meningitis.^117–121

The most serious complications associated with ESIs are spinal cord infarction with subsequent paralysis,^122–125 cerebellar and brainstem infarction with and without herniation,^126,127 and death.^128,129 There is a large volume of evidence that these neurologic complications arise after the injection of a particulate corticosteroid into an artery that supplies the central nervous system^{123,130–134} (Fig. 37–17).

Such arteries (including the vertebral arteries and radiculomedullary arteries) are most often located in or near the transvertebral foramen. While interlaminar and even caudal injections have case reports of paralysis in the literature,¹³⁵ the risk is greatest with transforaminal injections given the known anatomy. Pain interventionalists must have intimate knowledge of the arterial supply of the spinal cord (Fig. 37–18).

Multiple safeguards can be used to detect intravascular needle placement and avoid intra-arterial injection, and include use of proper technique according to the Spine Intervention Society guidelines, injection of contrast under real-time fluoroscopy, anesthetic test dose, use of low-volume extension tubing, an infraneural approach, and potential use of digital subtraction angiography.^{123,136–140}

The vast majority of literature that supports the efficacy of transforaminal ESI (TFESI) utilizes particulate steroids.^141,142 There is also a theoretical argument that use of particulate steroids may result in better outcomes. However, the vast majority of studies that have compared outcomes between particulate and nonparticulate steroids for ESI have found no significant difference in outcomes.^{130,143–148}

Most importantly, when selecting what steroid to use for TFESI, safety is paramount. Currently, all of the literature on permanent neurologic complications due to these procedures is exclusively associated with the use of particulate corticosteroids.^{122,123,127,133,149–156} Of the commercially available steroids, only dexamethasone phosphate (Decadron) is considered nonparticulate. In addition to the available reports of catastrophic neurologic complications with particulate steroids, dexamethasone does not carry the same theoretical risk of an embolic effect that particulate steroids do, given that dexamethasone has been shown to have particles five to ten times smaller than red blood cells, contain few particles, and do not aggregate.¹³² Considering this, dexamethasone should be considered the only appropriate steroid for cervical TFESI. For lumbar TFESI, dexamethasone should also be considered first-line treatment, with other steroids only considered in exceptional circumstances.

While interlaminar and even caudal injections have case reports of paralysis in the literature,¹³⁵ these procedures have a lower risk of complications due to the vascular anatomy of the spinal column. There are currently insufficient data to give a clear recommendation on which corticosteroid should be utilized for these procedures.

The most appropriate dose of all corticosteroids for ESIs, including dexamethasone, has yet to be established. Most studies that investigated lumbar TFESI used either low-dose methylprednisolone (40 mg) or high-dose methylprednisolone (80 mg).¹⁴¹ Dexamethasone is roughly fivefold more potent per milligram of methylprednisolone.

Number of Injections

There is no indication or literature that supports planning on or necessitating more than one injection, such as the common misconception that a “series of three” is needed. Pooled data analysis has revealed that 94 +/− 2% of patients with successful outcomes from TFESI did so with only one injection.¹⁴¹ After an initial injection, depending on response, however, there are times where additional repeat injections may be warranted. Repeat lumbar TFESI performed for recurrence of radicular pain has been shown to recover most or all of the benefit achieved after the initial injection. Repeat lumbar TFESI within a 3-month window can provide cumulative benefit in certain cases as well.¹⁵⁷

Cervical Epidural Steroid Injections

Cervical interlaminar epidural steroid injections (CILESI) require use of fluoroscopy, as without image guidance, the epidural space is missed 53% of the time.¹⁰⁰ Literature on the efficacy of such injections is very limited, however. Three observational studies that utilized non–image-guided CILESI injection reported success rates between 56% and 79% for up to 12 months.^158–160 Literature on fluoroscopically guided CILESI is even more limited, with a single study showing 50% relief in 77% of patients at 1 year, though the same rate of success was found in the group that received cervical interlaminar epidural anesthetic without steroid.¹⁶¹

The evidence in support of cervical transforaminal epidural steroid injections (CTFESI) is more robust. A recent comprehensive review of the literature on CTFESI, which included 16 different studies, most of which were observational in nature, concluded that cervical TFESI for cervical radicular pain results in approximately 40% of patients having at least 50% improvement in pain at 4 weeks.¹⁴² Select studies are briefly summarized as follows. Three cohort studies only reported group mean data following CTFESI and showed mean improvement on VAS of 5.4 at 1 year, 5.5 at 6 months which decreased to 3.2 at 12 months, and 4.3 at 6 weeks, respectively.^144,162,163 Four additional studies reported outcomes in patient cohorts after CTFESI using categorical outcomes that defined success as at least 50% improvement on VAS and reported successful outcomes as follows: 56% at 3 months and 50% at 12 months,¹⁶⁴ 48% at 12 to 14 weeks,¹⁶⁵ and 63% at 4 weeks.¹⁴³ A larger study that was initially designed to compare outcomes of different steroids used for TFESI found that regardless of what steroid was used, 260 of 441 (59%) patients had at least a 2-point improvement on VAS at 4-week follow-up.¹⁴⁴ Lastly, in a study that compared the use of CT to fluoroscopy for performing CTFESI, 36 of 65 (55%) patients in the fluoroscopy group had successful outcomes at 8 weeks, while 37 of the 51 (73%) in the CT groups also had a successful outcome.¹⁶⁶

It is worth noting that in 2007 a randomized study compared cervical transforaminal injection of anesthetic plus steroid to cervical transforaminal injection of anesthetic alone and reported no difference between groups. However, the study was flawed, in that the criteria for enrollment was at least 50% pain relief after diagnostic transforaminal injection of anesthetic and then this same intervention (transforaminal injection of anesthetic) was used as the control arm, resulting in significant selection bias.¹⁶⁷

Two studies also reported on the surgical-sparing effects of CTFESI. One study was a prospective study of 21 patients who were on the waiting list for surgery but were offered CTFESI first and reported 5 of the 21 had significant enough relief postinjection to avoid surgery.¹⁶⁸ The other was a retrospective study of 70 patients with cervical disc herniation resulting in radicular pain who had opted for surgical management but were offered CTFESI in an attempt to avoid or delay surgery and reported that 44 of the 70 had substantial enough relief to avoid surgery.¹⁶⁹

The data on caudal epidural steroid injections (CESI) and interlaminar epidural steroid injections (ILESI) for lumbosacral radicular pain are very limited, as the majority of available studies did not utilize image guidance for the injections. This is very problematic, as without image guidance there is a 30% to 40% rate of missing the epidural space.^2,170 The current standard of care dictates fluoroscopy be used for such injections. The only study that evaluated fluoroscopically guided CESI for radicular pain compared it to surgically targeted placement of steroid around the affected nerve and showed that both groups demonstrated equivalent improvement for up to 6 months.¹⁷¹ Results of RCTs in the management of radicular pain due to herniated nucleus pulposis have been mixed, with some suggesting good efficacy while others have been inconclusive.^172,173 There are no studies that specifically analyzed the efficacy of image-guided ILESI for lumbar radicular pain due to disc herniation. This may be in part due to a shift in clinical practice towards TESIs. This shift is supported by multiple studies that have compared the various approaches and demonstrate the superiority of the transforaminal approach to ILESI.^174–177

Careful review of the available literature that specifically investigates transforaminal injections for radicular pain reveals significant and positive findings, most dramatically for herniated disc pathology. The best-designed study to date that investigated the efficacy of TFESI was by Ghahreman et al in 2010. The study was a prospective, randomized study with five arms that compared TFESI to transforaminal local anesthetic, to transforaminal saline, to intramuscular steroid, to intramuscular saline. They found that a significantly greater proportion of patients in the TFESI group (54%) achieved at least 50% pain relief at 1 month compared to the other four arms (ranging between 7% and 21%).⁵ Pain relief was also “corroborated by significant improvements in function and disability, and reduction in the use of other health care” in the TFESI group. The number needed to treat to obtain at least 50% pain relief at 4 weeks was only three patients.⁵ Lastly, the design of the study also makes evident that both the medication (steroid) and route of delivery (transforaminal) combine to form a compound intervention that is unique and different from systemic steroids and from transforaminal administration of other compounds.⁵

An RCT that evaluated the efficacy of TFESI steroid versus transforaminal saline for treatment of radicular pain due to disc herniation found benefit of TFESI in those with L3–L4 herniation and L4–L5 herniation (but not L5–S1 herniation) at 4 weeks.¹⁷⁸ The same study found that at 1 year, TFESI was found to prevent progression to surgery within the same subgroup of contained disc herniations, and when compared to the control group found a savings of $12,666 per responder on average.¹⁷⁸ A recent study compared ILESI or TFESI for lumbar radicular pain due to stenosis of disc herniation to treatment with gabapentin-favored ESI, with 66% of those in the ESI groups vs. 46% in the gabapentin group being characterized as “responders” (p < 0.02).¹⁷⁹ Another study compared TFESIs to trigger-point injections for lumbosacral radiculopathy secondary to herniated nucleus pulposis and again favored TFESI, with 84% success in the TFESI group compared to 48% in other.¹⁸⁰

There are also multiple observational studies that evaluated patient cohorts of various sizes that demonstrated significant successful relief of lumbosacral radicular pain after TFESI in varying percentages of patients ranging from 41% to 75%.^181–184 The largest of these studies included 2,024 patients who received TFESI for radicular pain due to disc herniation or foraminal stenosis and reported that 45.6% had at least 50% pain relief at 2 months. Even more, in patients with less than 3 months of pain, the success rate increased to 68.3% at 2 months.¹⁸² Other studies that have investigated specifics of TFESI such as what type of steroid to use or what transforaminal approach is best can also be looked at as a whole for further evidence. Three such studies that specifically reported on patients with radicular pain due to disc herniation and defined success as at least 50% improvement reported success rates of 70% at 4 weeks,¹⁸⁵ 62% at 6 months,¹⁸⁶ and 73% at 3 and 6 months.¹⁴⁶

Lumbar TFESI has also been shown to have a surgical-sparing effect. Of 30 patients with severe lumbar radiculopathy who were considered surgical candidates but underwent TFESI, only 3 proceeded to surgery within the next 6 months and at mean follow-up of 3.4 years 79% of patients not lost to follow-up still had not required further intervention.¹⁸⁷ A retrospective review of 69 patients with symptomatic herniated lumbar discs reported surgical-sparing effect in 77% of patients at 1.5 years.¹⁸⁸ Another study that looked at 55 patients scheduled for surgery for lumbar radicular pain due to degenerative stenosis or disc herniation but first received TFESI deferred surgical intervention 71% of the time, which was significantly greater than the control group.¹⁸⁹ This effect was maintained for 5 years.¹⁹⁰ All three studies provide evidence that TFESI is an effective means of preventing surgical intervention in a significant amount of patients with radicular pain, with the strongest evidence again for those whose symptoms are due to herniated nucleus pulposis.

In 2013 MacVicar et al published a thorough and comprehensive review of the literature with accompanying systematic analysis of all published data regarding lumbar TFESI.¹⁴¹ After plotting the success rates of outcome studies, pragmatic studies, and explanatory studies, MacVicar et al stated that with regard to TFESI for radicular pain due to disc herniation, the literature is “abundant” and of “higher quality” and reveals that “about 60% of patients [seem] to achieve at least 50% relief of pain at between 1 and 2 months but that only 40% maintain this outcome for 12 months.”¹⁴¹ They succinctly concluded that TFESI is effective (more so in patients with contained disc herniations, low-grade compression, and acute symptom duration),^{102,186,191,192} statistically more than placebo effects,^5,189 and reduces the burden of disease by improving function^5,176,180 and reducing the need for surgery,^187–190 and is ultimately cost-effective.^141,178

In conclusion, there is very little evidence that analyzes the efficacy of fluoroscopically guided CILESI even though it is still common in clinical practice. In the lumbar spine, evidence for ILESI and CESI is also sparse. With respect to cervical TFESI a recent review conservatively concluded that 40% of patients have at least 50% improvement in cervical radicular pain at 4 weeks after CTFESI.¹⁴² In the lumbar spine, another review of TFESI concluded that about 60% of patients seem to achieve at least 50% relief of pain at between 1 and 2 months.¹⁴¹ The evidence for ESI is indeed most robust for the use of TFESI to treat lumbar radicular pain due to herniated intervertebral discs. Both cervical and lumbar TFESI have been shown to have a significant surgical-sparing effect as well.

Radiculitis is a process in which inflammation plays a primary role, especially in the case of disc herniation. Not surprisingly, direct deposition of steroid medication to the affected nerve roots can provide significant relief in many patients. It is important to also realize that ESI also has associated risks, at times catastrophic. Such procedures should not be taken lightly. ESI must be performed in the safest way possible, including use of appropriate technique, appropriate safety measures, and appropriate steroid selection. The best available evidence is clear, though, that ESI provides relief for radicular pain that is significant, clinically meaningful, and not due to placebo effects.

Introduction

The prevalence of sacroiliac joint complex (SIJC) pain varies depending on how it is defined and the population examined. The clinical presentation of SIJC pain is often nonspecific, with pain reported in the buttocks (94%), low back (72%), and thigh (48%).¹⁹³ In the absence of trauma or confirmed rheumatologic sacroiliitis, there is no correlation between pain and radiologic imaging. Traditionally, intra-articular anesthetic injections have been used as a diagnostic tool, though only 15% to 30% of patients suspected to have SIJ pain on history and physical are found to respond to intra-articular anesthetic block.^194,195 This likely reflects deficiencies not only in clinical diagnosis but also with intra-articular diagnostic injections themselves. SIJC pain can arise from intra- and extra-articular etiologies. The encapsulated diarthrodial synovial sacroiliac joint itself is certainly one potential cause as evidenced by pain referral maps generated by Fortin and colleagues immediately after forced distension of the joint capsule itself.¹⁹⁶ Additionally, studies have demonstrated that the posterior ligaments are also potential pain generators.¹⁹⁷ Unfortunately, though, little differentiation is made between intra-articular pain versus posterior ligamentous pain in research and practice. Understanding of the associated anatomy is further confounded by debate over the innervation of the SIJC, with suggested innervations including posterior elements from the L4 and L5 dorsal rami and lateral branches of S1–S3, as well as anterior elements from the trunk of the lumbosacral plexus, obturator nerve, and superior gluteal nerves.¹⁹⁸ The lack of clear definition, diagnosis, and understanding of the anatomy confounds the interpretation of available literature on SIJC interventions. That being said, when discussing sacroiliac pain, a useful framework to consider is two distinct and overlapping entities: both intra-articular and capsular pain generators, which have both anterior and posterior innervation, as well as posterior ligament pain, which is likely exclusively posteriorly innervated.