Sedation

An important, potentially confounding factor when performing diagnostic blocks concerns the use of sedation. Logically, one would assume that administration of opiates and sedatives before a diagnostic block would increase the false-positive rate; however, Manchikanti and colleagues³³ found that this proportion of patients was relatively small, and there was no difference with use of saline, opiate, or sedative. In a randomized study of 60 patients, Manchikanti and colleagues³³ titrated medication to relaxation using saline, midazolam, or fentanyl. They found that only 50% of patients receiving sodium were relaxed, whereas 100% of patients receiving either fentanyl or midazolam were relaxed. In all groups, 10% of the patients reported greater than 80% pain relief with active motion testing. Even so, typically, one limits or omits sedation before a diagnostic injection, however.

Another method to mitigate false-positive results is to administer a short-acting sedative such as propofol and delay testing to 30 minutes after the procedure to minimize the sedative effects. If the patient is anxious or especially if the patient routinely takes significant doses of opiates and may be in early opiate withdrawal owing to NPO status, the patient could be sedated and tested after the administration of a judicious dose of opiate.

Biopsychosocial Factors

More recently, authors have reiterated the importance of shifting the concept of “backache.” Kikuchi³⁴ recommended changing the term spinal disorder to biopsychosocial pain syndrome and the term morphologic abnormality to mechanical, functional disorder. According to Kikuchi,³⁴ morphologic and structural abnormalities do not always explain all of a patient’s pain, and chronic backache should not be seen as an isolated spinal disease. A significant amount of scholarship has been devoted to enumerating the psychosocial factors associated with spinal pain. In a classic study comparing workers with symptomatic disc herniation (requiring surgery) versus asymptomatic workers, significant differences were found in three areas: presence of nerve root compromise, psychosocial factors (depression, anxiety, marital status, self-control), and work perception (satisfaction, job loss, occupational stress, intensity of concentration).³⁵ Of the risk factors, two of three were functional, not morphologic.

The authors agree with this model and acknowledge that psychosocial factors have a significant impact on spinal pain; the authors also acknowledge the importance of treating the “bio” component of the biopsychosocial syndrome. Today’s spine specialists readily acknowledge that patients often benefit from a multidisciplinary approach to treatment. Research continues to investigate the best way to treat patients with chronic spinal pain and significant psychosocial comorbidities.

Although the long-term results of treatment for chronic axial back pain may be influenced by various psychosocial factors,^36,37 the possible effect of psychosocial factors in determining the patient’s tested perception of pain and functional improvement with treatment does not indicate that the diagnosis was incorrect. In many cases, when a “biologic” pain generator can be correctly identified and treated, the psychosocial distress resolves. If it is true that psychological variables determine whether a patient admits relief on various testing instruments, does the evidence of physiologic distress noted on test scores reverse when the patient’s chronic pain is relieved?

Wallis and colleagues³⁸ studied 17 patients after whiplash injury with a single symptomatic cervical zygapophyseal joint who were enrolled in a randomized controlled trial of percutaneous radiofrequency neurotomy. At 3 months after the procedure, all patients whose pain was relieved had complete resolution of preoperative psychological distress; in contrast, all but one of the patients who did not experience pain relief continued to experience psychological distress. Manchikanti and colleagues³⁹ found no correlation between somatization disorder and inappropriate Waddell signs and symptoms to response in pain relief after a comparative double block protocol for diagnosing facet pain. Derby and colleagues⁴⁰ found no difference in response to pressure-controlled disc stimulation between patients with abnormal psychometric Distress Risk Assessment Method scores and asymptomatic volunteers.

Pathophysiology of Facet Pain

Traumatic injury to lumbar and cervical zygapophyseal joints is common and probably occurs to a lesser extent to the thoracic joints. In the setting of trauma, there is a clear pathophysiologic difference between facet joints and nontraumatic controls. Zygapophyseal joint sections from autopsy specimens of individuals with a past history of trauma but dying of natural causes show significant age-related, gender-related, and trauma-related changes in the bone, cartilage, and soft tissues, including subchondral sclerosis, fibrillation and splitting of cartilage, and cartilage length differences, versus subjects with no history of trauma.⁴⁶ Histologic sections of the lumbar zygapophyseal joints of mostly motor vehicle accident victims revealed fractures of the superior articular process, central infarctions of the subchondral bone plate, and capsule tears including the ligamentum flavum.^47,48 Most tissue sections (77%) show soft tissue injuries, and approximately 30% (11 of 33) show fractures and infarctions. Sections from the cervical spine in trauma victims show similar injuries to the zygapophyseal joint articular cartilage and annular lesions in the intervertebral discs and cartilaginous endplates (Fig. 15–1).⁴⁹ In the lumbar and cervical spines, lesions were found exclusively in the trauma patients and in none of the patients in the control group.^46,49,50

FIGURE 15–1 Postmortem studies of motor vehicle victims reveal the following sites of injury; many of these injuries may not be visible on routine MRI.

(From Bogduk N, McGuirk B: Management of Acute and Chronic Neck Pain: An Evidence-based Approach. Philadelphia, Elsevier, 2006.)

Many of these pathoanatomic findings are occult on routine x-ray, CT scan, and magnetic resonance imaging (MRI) but may be the cause of ongoing neck pain in survivors of motor vehicle accidents or other significant trauma. In a small study with short-term follow-up (approximately 3 months), Eisenstein and Parry⁵¹ examined zygapophyseal joints in 12 patients who underwent successful fusion for zygapophyseal joint mediated pain (diagnosed by provocation arthrography, intra-articular blocks, and negative discography) versus controls and found histologic changes similar to changes of chondromalacia patellae and osteoarthritis of large joints. The most frequent finding was focal full-thickness cartilage necrosis or loss of cartilage with exposure of subchondral bone; osteophyte formation was absent in all specimens.⁵¹ Degenerative histologic findings alone do not make a definitive diagnosis of facet syndrome, however. Ziv and colleagues⁵² reported a high proportion of coarsely fibrillated or ulcerated (or both) facets in fresh cadaveric spines from young adults (30 to 50 years old); such degeneration remains constant throughout adulthood.

Traumatic and repetitive injury leading to painful “facet arthritis”⁴⁸ may cause pain because zygapophyseal joints and their capsules are heavily innervated structures subject to high stress and strain during spinal loading.⁵³ Joints comprise free and encapsulated nerve endings containing substance P and calcitonin gene-related peptide.^54–56 Substance P, calcitonin gene-related peptide, and immunoreactive sensory and autonomic nerves are found in zygapophyseal joint synovial membranes.⁵⁷ Facet capsules contain low-threshold mechanoreceptors, mechanically sensitive nociceptors, and silent nociceptors.⁵⁸ These low-threshold and high-threshold mechanoreceptors fire when the joint capsule is stretched or compressed, and their firing can be suppressed by injected lidocaine and hydrocortisone.⁵⁹ In animal models, induced inflammation decreases the threshold of nerves within the joint capsules and causes elevated baseline discharge rates.⁵⁹ In animal models of knee arthritis, acute inflammation sensitizes fine articular afferents, which become active at rest and respond more vigorously to routine painless joint range of motion.⁶⁰

Excessive stretching damages the zygapophyseal joint capsules and causes axonal swelling, retraction balls, and inflammation.⁵⁸ The result is hyperexcitability and spontaneous firing, which are synonymous with neuropathic pain. Capsular injury during a whiplash injury may cause persistent neck pain secondary to chronic capsular overstretching. Animal studies suggest that facet capsule strains comparable to strains previously reported for whiplash kinematics and subcatastrophic failures of this ligament activate nociceptors within the capsule.^61–65 In addition, chronic capsular loading in animals may cause central inflammation resulting in mechanical hyperalgesia and in some cases centrally maintained pain.^56,63,66,67

Animal studies showing central sensitization are consistent with the widespread hypersensitivity documented in whiplash patients. Although focal sensitization to mechanical stimuli may be found 3 months after whiplash injury, which mostly resolves by 6 months, some patients develop persistent pain with symptoms that are consistent with chronic neuropathic pain.⁶⁸ Patients with persistent pain at approximately 6 months show signs of more widespread hyperalgesia⁶⁹ and hypersensitivity to cutaneous and muscular stimulation in neck and lower limb consistent with central hypersensitivity.^10,70

If chronic pain originating from injury to the zygapophyseal joints is due to capsular stretch and maintained by central hypersensitivity, local anesthetic with or without corticosteroid should suppress nociceptive input for the duration of the local anesthetic effect and in some cases (e.g., similar to a sympathetically maintained pain state) for days to weeks.⁷¹ If most of a patient’s pain is due to mechanical and central causes, there would be no reason why local anesthetic and corticosteroid would be more effective than local anesthetic alone.^72–74 Former and latter logical outcomes are supported in prospective and randomized controlled trials,⁷⁴ although alleviating pain with medial branch blocks was shown in one study to relieve pain for an average of several months.^75,76

Decreasing peripheral input 6 months or longer by heat ablation of medial branches relieves pain⁷⁷; the pain typically returns within the expected time it takes for the medial branches to regenerate. Such prolonged relief of pain, if accompanied by resolution of widespread hypersensitivity, would imply that central hypersensitivity is reversible when the peripheral source of input is interrupted. That is, if there is a concern that persistent central hypersensitivity would lead to failure of a proposed localized or segmental stabilization procedure, resolution of widespread and local hypersensitivity after medial branch neurotomy might predict that decrease in nociceptive input by surgical stabilization would be successful. Zygapophyseal joint pain often occurs at more than one segment, however, and one must identify adjacent or skipped level sources of zygapophyseal joint pain, especially if one is considering surgical fusion or arthroplasty.

When cadaveric lumbar spines are anteriorly fixated at one level, motion is transferred to adjacent segments causing increased capsular stretch in the adjacent facet joints.⁷⁸ In extension, cervical arthroplasty models exhibit significant increases of facet force at the treated level. In the fusion model, the facet forces decrease at the treated segment and increase at the adjacent segment.⁷⁹ Failure to recognize symptomatic pathology at an adjacent level or the same level may lead to early or late return of pain. This is not a failure of the diagnostic blocks; it is a failure to obtain a thorough diagnosis.

Rationale for Control Blocks in Diagnostic Facet Intra-articular and Medial Branch Blocks

Can a diagnosis of facet syndrome be made without injections? The diagnosis of zygapophyseal joint pain is typically hypothesized based on clinical findings and imaging studies. Most clinicians rely on a variety of favorite criteria, such as localized unilateral pain that is worse in extension, pain worse in the morning and better with gentle movement, concordant pain provoked with palpation approximately 1 cm lateral to the midline over the zygapophyseal joints, and imaging studies showing signs of facet degeneration. A “facet syndrome”⁸⁰ diagnosed by clinical findings has not been substantiated, however, if one uses as a reference standard the relief of pain after placebo-controlled anesthetic blocks.^23,27,81 The current best evidence has not found any individual clinical finding or cluster of findings that can predict response to the reference standard of pain relief after local anesthetic block of the medial branches or intra-articular zygapophyseal joint block.

The purpose of diagnostic facet blocks is to establish the diagnosis or rule it out, similar to a liver biopsy of a suspicious lesion. Diagnostic facet blocks are a tertiary intervention in patients with chronic pain that has not resolved with time and conservative care. The current standard for the diagnosis of facet mediated pain is the use of controlled differential (double) blocks to confirm or refute one’s hypothesis that the facet joint is a pain generator. Because of the high false-positive rates of single diagnostic blocks, a single block does not constitute a diagnosis, and control blocks are essential to decrease the incidence of false-positive responses.⁸² The reported false-positive rate of a single diagnostic block ranges from 17% to 63%.⁸³ In a retrospective review of 438 patients using a double block paradigm requiring 80% relief, the false-positive rates for a single diagnostic block were 45%, 42%, and 45% for the cervical, thoracic, and lumbar regions.⁸³

Although Cohen and colleagues⁸⁴ showed that treatment results after medial branch neurotomies were not changed by requiring a placebo control, the current published standard of interventional societies requires a confirmatory block before making a diagnosis of zygapophyseal joint mediated pain. The best-studied double block protocol requires a difference in pain relief duration based on a shorter or longer acting medication—typically, greater than 1 hour for lidocaine and greater than 2 hours for bupivacaine.^85,86 Because the goal is to have a placebo control and because consent and patient compliance issues hinder using a saline block control, one may argue that any prior diagnostic injection in which the patient reports no relief is a valid control block, especially because the patient and the physician were anticipating relief. Because many insurance companies in the United States no longer authorize or consider double blocks medically necessary, the occurrence of a previous “negative” block evaluation could be the negative control.

When the treatment is relatively benign, convincing relief with physician and staff testing after facet or medial branch injections in an older patient with clinical symptoms consistent with zygapophyseal joint pain may not justify confirmatory injections.⁸⁴ A young patient who has little or no facet abnormalities, who is on a significant dose of narcotics, and who reports less than convincing approximately 80% relief should undergo a second confirmatory injection before considering interventional or surgical treatment based on the block results.

The rationale for how many and which levels to test is debated. Levels are typically chosen based on known pain referral patterns, prevalence studies, and localized manual palpation. Using a comparative double block control, Manchukonda and colleagues⁸³ found that most often two joints were symptomatic in the lumbar spine and three adjacent joints were symptomatic in the thoracic and cervical spine. The most logistically efficient and least costly method is to rule out facet pain globally on the side or sides of the patient’s pain and at the proximity of approximately two to three adjacent levels. One would inject (in the case of lumbar spine) the L2-5 medial branches to block the L3-4, L4-5, and L5-S1 zygapophyseal joints (Fig. 15–2 shows medial branch anatomy of the lumbar spine; Fig. 15–3 shows a lumbar medial branch block). If no relief occurred and there was no evidence that the higher joints were involved, and if one is confident that the facets were denervated, one can eliminate facet pain from the diagnosis. If the result is positive, one can perform more selected denervation on a confirmatory injection. Although this could be argued to be the most efficient method, many third-party payers in the United States limit injections to two levels per session.

FIGURE 15–2 Anteroposterior radiograph of lumbar spine showing course of medial branches of L1-4 dorsal rami, L5 dorsal ramus, and their articular branches to lumbar zygapophyseal joints. Medial branches are blocked at junction of superior articular process and transverse process.

(From Bogduk N [ed]: Practice Guidelines for Spinal Diagnostic and Treatment Procedures. San Francisco, International Spine Intervention Society, 2004.)

FIGURE 15–3 Oblique view of right L4 medial branch block. A 3.5-inch 25-gauge needle is placed at junction of L5 superior articular process and transverse process. A total of 0.5 mL of contrast dye is injected along medial branch. SAP, superior articular process.

(Courtesy Richard Derby, MD.)

Diagnostic Accuracy

The features of a test that indicate its diagnostic accuracy are sensitivity and specificity. From these measures, the false-positive and false-negative rates and predictive value can be derived. Sensitivity refers to the ability of the test to identify correctly patients with the disease. Several factors that affect sensitivity should be discussed. Vessels accompany the medial branch as they course around the waist of the superior articular pillar, and injecting local anesthetic into a vessel rather than around the nerve may cause a false-negative response.

Kaplan and colleagues⁸⁷ reported that medial branch blocks may fail because of venous uptake. Venous uptake occurred in 7 of 20 (35%) medial branch blocks. If venous uptake was encountered, repositioning of the needle resulted in joint anesthesia only 50% of the time. When venous uptake was encountered, the subjects were brought back for a later injection. These findings stress the importance of using contrast medium for medial branch blocks and carefully observing the flow pattern. Kaplan and colleagues⁸⁷ also found that in 11% of cases they were unable to anesthetize the joint, even in the absence of venous uptake. Medial branch blocks in the lumbar spine would have an 11% false-negative rate; this may have been due to anomalous or collateral facet innervation or insufficient volume of local anesthetic reaching the target nerve.

Rarely discussed is the consistency with which one may expect a longer duration of action of bupivacaine versus lidocaine. Although bupivacaine has a longer duration of action than lidocaine, the mass of drug reaching the nerve is the most important variable, and one cannot guarantee that the same amount will be available on consecutive sessions. In addition, most patients are not kept in the recovery area for the duration of local anesthetic to be evaluated, and the duration depends on a patient’s self-reporting, which may or may not be consistent between injections. Lord and colleagues⁸⁶ showed that using a double block comparative standard, 65% of patients failed to recognize the difference in duration of pain relief but did accurately distinguish a separate placebo-controlled block with saline.

The crucial factors for specificity of the diagnosis of zygapophyseal joint pain are accurate targeting of the intended structure under fluoroscopy, confirmed by contrast medium, and the delivery of the appropriate volume of local anesthetic. Intra-articular joint blocks are specific, unless there is a medial capsular tear or injection of excessive volumes (>1 mL injected into a lumbar zygapophyseal joint or 0.3 mL injected into a cervical zygapophyseal joint), which may rupture the capsule and spread medially into the epidural space.^88,89 Many early studies of facet intra-articular blocks used 2 to 8 mL per injection. When reviewing negative studies regarding facet injections and the systematic reviews that still quote these studies, the discerning reader should check the total volume of injectant used.^89,90 Destouet and colleagues⁸⁹ found that volumes of injectant of 0.5 to 1.5 mL commonly ruptured the superior recess of the capsule and extravasated; in later studies, Destouet and Murphy⁹¹ aspirated the 0.5 to 1.5 mL of contrast dye before adding local anesthetic and steroid. Cadaveric studies performed with variable volumes of methylene blue injected into facet joint (1 to 4 mL) showed that the dye extended not as expected into the paraspinal tissues but rather into the epidural space and around the spinal nerves.⁹² Moran and colleagues⁹² described the facet capsule as thick dorsally, whereas anteriorly the facet synovial membrane is contiguous with the ligamentum flavum, and the adipose tissue in the superior recess is in direct contact with the adipose tissue around the spinal nerve. Randomized controlled studies validate the specificity of medial branch blocks for relief of zygapophyseal joint mediated pain.⁹³

To maintain the specificity of medial branch block, a low volume of local anesthetic is used to anesthetize a specific cervical⁹³ or lumbar medial branch.^87,88 If volumes greater than 0.5 mL are used, the close proximity of the lumbar lateral and intermediate branch to the medial branch potentially might increase false-positive rates by blocking paraspinal soft tissues (ligaments or muscles or both).^20,94 In addition, volumes greater than 0.5 mL on the superior edge of the lumbar transverse process may spread onto the dorsal and ventral spinal nerves.⁸⁸ Decreasing input into the central nervous system by anesthetizing any structure or nerve could relieve pain by decreasing or modulating central input. Although this is a probable confounding factor, the reference typically cited is the study by North and colleagues,⁹⁵ which reported an unacceptably high false-positive rate for diagnostic blocks in patients with low back pain. The conclusion reached by North and colleagues⁹⁵ was that diagnostic blocks had limited specificity. Their protocol clearly lacked diagnostic specificity, however. They used an excessive 3 mL volume of bupivacaine for the medial branch blocks. Because 0.5 mL of local anesthetic placed too close to the superior edge flows onto the exiting root, the reported decrease in sciatic pain was more likely due to anesthetizing the spinal nerve and not a false-positive effect of neuromodulation.⁸⁸

Because patients in the study by North and colleagues⁹⁵ reported an average relief of 75% to 80% after anesthetizing the sciatic nerve, operators must be aware that blocking this nerve anywhere along its course may result in the report of pain relief. As previously mentioned, relief of pain for the duration of the local anesthetic effect cannot distinguish between reversible inflammatory or compressive causes of pain and nonreversible neuropathic causes of pain.

Although North and colleagues⁹⁵ found that 3 mL of local anesthetic placed into muscles at several levels has relatively minimal effect on sciatic pain, some authors question whether anesthetizing the needle track increases false-positive responses.²⁰ Patients probably report more pain relief when the needle track is anesthetized. Rather than a false-positive response, one would expect better pain relief when patients are not experiencing lingering needle-related pain at the same time as pain relief resulting from anesthetizing the zygapophyseal joints.

Lumbar Spine: Facet Syndrome

Lumbar Zygapophyseal Joint Pain

Lumbar zygapophyseal joint pain is common and seems to become more prevalent as patients age and as the duration of chronic back pain increases. The prevalence is directly inversely correlated, however, to the stringency of the reference standard. Prevalence studies consistently find the zygapophyseal joint as a common source of pain, but diagnosis using a single anesthetic block has a potential high placebo rate versus a double comparative block control requiring a longer duration of relief after lidocaine versus bupivacaine.

Studies reporting the prevalence of zygapophyseal joint pain report figures ranging from 15% to 52%. In 1994, Schwarzer and colleagues⁹⁸ established the prevalence of zygapophyseal joint pain using a double block comparative protocol in 96 patients with a mean age of 38 years and mean duration of low back pain of 16 months mostly secondary to work-related injuries and trauma presenting to two tertiary U.S. clinics. (Fig. 15–4 shows a lumbar intra-articular zygapophyseal joint block.) Schwarzer and colleagues⁹⁸ found that the combination of discogenic pain and zygapophyseal joint pain is uncommon. In a group of 176 patients, 47% had initial relief with a screening lidocaine block, but only 15% had 50% or greater relief with a confirmatory block.

FIGURE 15–4 Right L4-5 zygapophyseal joint intra-articular injection. Note filling of superior and inferior subcapsular capsular recesses with contrast dye (arrow). Needle hub partially overlies inferior capsular recess.

(Courtesy of Richard Derby, MD.)

Manchikanti and colleagues,⁹⁹ also using a double block protocol but requiring 75% pain relief and a differential response, reported an even higher initial response of 81 of 120 (67.5%) to lidocaine medial branch blocks and a much greater percentage (45%) of the total reporting longer 75% relief after confirmatory bupivacaine medial branch blocks. This patient group was older, however, than Schwarzer’s group with a mean age of 47 years and with a longer mean duration of low back pain of 47 months. The false-positive rate for one block was 41%. In a later study, Manchikanti and colleagues¹⁰⁰ revised the prevalence of zygapophyseal joint pain downward from 45% to 27% (95% confidence interval 22% to 32%). Schwarzer and colleagues¹⁰¹ studied an older group of patients with a mean age of 57 years referred to an Australian rheumatology clinic with low back pain for an average of 7 years. A diagnosis of zygapophyseal joint pain was made in 40% (95% confidence interval 27% to 53%). Requiring 90% relief of original pain, the prevalence was 32%, and requiring 100% relief, the prevalence was 11%. Manchikanti and colleagues¹⁰⁰ reported an even higher prevalence of 52% zygapophyseal joint pain in a group of patients 65 years old or older.

No consistent history, physical examination, or imaging findings correlated to positive block responses have been found. In the early 1980s, uncontrolled single, variable injectant volume, intra-articular zygapophyseal joint injections were used as the reference standard for identifying lumbar zygapophyseal joint pain, and these authors reported correlations with various history or physical examination findings.^51,102 Some studies^103,104 reported that a cluster of five of seven features (Revel criteria) could predict a 75% decrease in pain after a single intra-articular block. The seven items in the cluster are age older than 65 years, pain well relieved by recumbency, no exacerbation of pain with coughing and sneezing, no exacerbation of pain with forward flexion, no exacerbation of pain with extension, no exacerbation of pain with rising from flexion, and no exacerbation of pain with the extension-rotation test. Subsequent well-conducted studies did not replicate these studies.^28,105 As mentioned, most of these earlier studies used single medial branch blocks, which have been reported to have 25% and 38% false-positive rates for the diagnosis of zygapophyseal joint pain.^106,107

Newer studies of clinical correlations refined the technique with an appropriate injectant volume and a confirmatory double block paradigm with either a second intra-articular injection or a medial branch confirmatory injection with bupivacaine lasting longer than the pain relief after a prior lidocaine block.^{28,92,101,106,108,109} These studies did not find any clinical correlates with history or physical examination. In particular, extension and rotation were not predictive of response. A systematic review of all published studies comparing clinical outcome after local anesthetic blocks and clinical signs and symptoms found no consistent clinical features with a high specificity.¹¹ The review found several clinical features with a high sensitivity, however, and these features may be cautiously used to exclude the diagnosis of facet mediated pain. These features include pain not increased with cough, pain not relieved with recumbency, and pain that can be centralized.¹¹ There are no consistent reproducible history or physical examination criteria that predict a positive response to a facet block. History and physical examination are better at ruling out facet mediated pain than diagnosing facet pain.

The current best evidence also shows that radiologic imaging, with a few more recent exceptions, does not correlate with response to zygapophyseal joint blocks. The conflicting evidence that radiologic imaging may predict outcome from uncontrolled lumbar zygapophyseal joint blocks may be partially due to lack of rigor in the reference standard used to define a positive response in earlier studies.²⁰ In 1979, Carrera¹¹⁰ reported that 73% (n = 63) of patients describing pain relief after uncontrolled intra-articular injection of 2 to 4 mL of local anesthetic had CT evidence of lumbar facet disease versus 13% who had no evidence of disease. It is well accepted, however, that injectant volume should not exceed 1 mL; otherwise, the injection loses specificity, with a leak of local anesthetic around the nerve root and along vertebral levels within the epidural space.⁹²

A large study by Jackson and colleagues¹¹¹ of 390 patients found no relationship between imaging and pain relief after uncontrolled intra-articular lumbar zygapophyseal joint injections. Supporting the findings by Jackson and colleagues,¹¹¹ Schwarzer and colleagues,¹¹² in the only study using placebo-controlled injection, found no correlation between CT findings and a positive response comparing local anesthetic with saline blocks in 63 patients when more stringent criteria of controlled injections were used as the reference standard. Similarly, Cohen and colleagues¹¹³ found no relationship in 192 patients between MRI findings of zygapophyseal joint hypertrophy or degeneration and response to medial branch neurotomies based on positive response to a single medial branch block. Kawaguchi and colleagues¹¹⁴ likewise found no significant relationship between low back pain symptoms and radiographic abnormalities in a group of 106 patients with rheumatoid arthritis.

The intriguing bright spot on the horizon is the finding that where MRI or single photon emission computed tomography (SPECT) shows imaging findings consistent with either “inflammation” or “edema,” a stronger correlation emerges (Fig. 15–5). Although not confirming the diagnosis of zygapophyseal joint pain with a reference standard, Friedrich and colleagues¹¹⁵ more recently found that an estimated 14% (21 of 145) of patients with low back pain had MRI evidence of facet joint edema, and follow-up MRI scan showed “almost perfect” agreement between change in pain and a reduction in intensity of edema on sagittal short tau inversion recovery (STIR) images. Radionuclide bone scintigraphy detects bone areas with synovial changes (inflammation or hyperemia) or increased osteoblast activity and degenerative regions with a high degree of remodeling. Osteophytes in process of growing show a high degree of bone scan activity. As mentioned earlier, a positive lumbar SPECT scan predicts a statistically significant reduction in pain after facet blocks.³¹

FIGURE 15–5 L5-S1 axial T2-weighted MRI showing facet edema (right greater than left), suggestive of instability. In upright weight-bearing position, anterolisthesis was noted; in unloaded supine position, anterolisthesis reduced, and zygapophyseal joints gapped and filled with fluid.

(Courtesy of Richard Derby, MD.)

Zygapophyseal Joint Pain Referral Maps

Pain referral patterns have been studied using stimulation of patients during provocative diagnostic injections¹¹⁶ by injection of hypertonic solutions into normal and abnormal subjects⁸⁰ or by electrical stimulation of medial branches.^80,117 Most studies showed distinct but overlapping referral areas; it is likely that the pain referral patterns obtained in normal volunteers are smaller because of less sensitization. There are also limits to the referral maps; Mooney and Robertson⁸⁰ reported on lumbar facet referral maps in normal volunteers and subjects with a positive diagnostic facet block (Fig. 15–6). Under fluoroscopic guidance, they injected contrast dye (unspecified volume) followed by 3 to 5 mL of hypertonic saline. Some of the distal extremity pain seen in the diagrams may be due to excessive volume of saline with irritation of the sciatic nerve roots. Given the lack of sensitivity and specificity of history, physical examination, and imaging and until more research is performed with finite injectant volumes (in patients with confirmed dual positive blocks), these referral maps can be used as a starting point to guide selection of levels to be injected.

FIGURE 15–6 Pain referral patterns for asymptomatic (normal) and symptomatic (abnormal) subjects obtained by intra-articular zygapophyseal joint injection of contrast dye followed by 3 to 5 mL of hypertonic saline.

(From Mooney V, Robertson J: The facet syndrome. Clin Orthop Relat Res [115]:149-156, 1976.)

Cervical Spine Facet Syndrome

Cervical Zygapophyseal Joint Pain

Based on the confirmatory block paradigm, the cervical facet joints are a common source of chronic neck pain; the prevalence of cervical facet syndrome is greater than the prevalence of lumbar facet syndrome. Cervical discogenic pain shares referral patterns with facet pain, but it is far less common.¹³² Based on comparative blocks of cervical facet joints causing chronic neck pain with either associated headache or shoulder pain, the C2-3 (36%) and C5-6 facet joints (35%) were the most common pain generators.¹³³ After whiplash injury, Level I prospective clinical studies provide evidence that facet joints are the most common source of chronic pain.^134,135 Cervicogenic headache stemming from the C2-3 facet after whiplash has a 53% prevalence.¹³⁴

Often neglected are C0-1 and C1-2 joints in evaluation of upper neck pain and headache. Dreyfuss and colleagues¹²⁹ studied the referral patterns for the atlantoaxial and lateral atlantoaxial joints. In 2002, Aprill and colleagues¹³⁶ failed to confirm the null hypothesis that lateral atlantoaxial joints are not a common source of occipital headache. These investigators found that of 34 patients presenting with symptoms and signs of atlantoaxial joint pain, 21 obtained complete relief of headache after diagnostic injection of local anesthetic. Pain referral patterns have been defined in C2-3 through C7-T1 facets (Fig. 15–7).¹⁵ Innervation of the cervical facet joints is well described (Fig. 15–8).¹⁵ The cervical zygapophyseal joints can be blocked either by medial branch blocks or with intra-articular injections (Fig. 15–9).

FIGURE 15–7 Patterns of referred pain from cervical zygapophyseal joints in normal volunteers from Dwyer A, Aprill C, Bogduk N: Cervical zygapophyseal joint pain patterns. 1:A study in normal volunteers. Spine 15:453-457, 1990.

(From Bogduk N [ed]: Practice Guidelines for Spinal Diagnostic and Treatment Procedures. San Francisco, International Spine Intervention Society, 2004.)

FIGURE 15–8 A, Lateral view of cervical spine showing variable locations of medial branches. At C3, location of C3 deep medial branch is shown. Inset shows location of third occipital nerve (TON). Shaded area shows where C3 deep branches and TON overlap. C5 medial branch is located in the middle of the articular pillar; at C6 and C7, medial branches are located progressively higher. B, Anteroposterior view of cervical medial branches.

(From Bogduk N [ed]: Practice Guidelines for Spinal Diagnostic and Treatment Procedures. San Francisco, International Spine Intervention Society, 2004.)

FIGURE 15–9 Lateral fluoroscopic view of C3-4 zygapophyseal joint injection using 3.5-inch 25-gauge needle. Note contrast dye in posterior and anterior capsular folds (arrows).

(Courtesy Richard Derby, MD.)

Prevalence rates for pain originating from cervical facets range from 36% to 60%. The false-positive rate for a single, uncontrolled block is 27% (95% confidence interval 15% to 38%).¹³⁷ The following prevalence rates (mean [95% confidence interval]) are reported from studies using either a double block or a triple block paradigm (normal saline as a placebo): 54% (40% to 68%),¹³⁴ 36% (27% to 45%),¹³⁸ 60% (33% to 64%),¹³⁵ and 60% (50% to 70%).¹³⁹ Manchikanti and colleagues¹⁰⁰ restudied the prevalence of cervical zygapophyseal joint pain in a larger group of patients and found a similar 55% (95% confidence interval 49% to 61%) prevalence. The most recent study by Manchikanti’s group in 2007,⁸³ of 438 patients requiring 80% relief of pain for 2 hours’ duration with lidocaine and 3 hours’ duration with bupivacaine, reported a prevalence of 39%. Corroborating the high prevalence of cervical zygapophyseal joint pain, Yin and Bogduk^139a in a private practice clinic audit found a 55% prevalence of cervical zygapophyseal joint pain using a strict double comparative block protocol.

Similar to lumbar zygapophyseal joint pain, there are no high-quality studies showing a particular set of clinical features that can predict results of diagnostic cervical facet or medial branch blocks.¹⁴⁰ With diagnosis by medial branch blocks, one exceptionally skilled manipulative therapist was able to identify all 15 subjects with diagnostic block–proven symptomatic zygapophyseal joints and specify the correct symptomatic segment. None of the five patients with asymptomatic joints was misdiagnosed as having symptomatic zygapophyseal joints.¹⁴¹ A later follow-up study by the same group failed to confirm the apparent high specificity and sensitivity, however, and reported a high sensitivity but low specificity and concluded that manual examination of the cervical spine lacks validity for the diagnosis of cervical zygapophyseal joint pain. In the study by Aprill and colleagues¹³⁶ of C1-2 facet pain as a source of occipital headache, only 60% of the patients shared clinical criteria that predicted a positive response to the block.

Advanced imaging has not been correlated with positive responses to diagnostic blocks. Hechelhammer and colleagues¹⁴² found no relationship between short-term pain relief after cervical intra-articular and pericapsular injection of local anesthetic and corticosteroid and the degree of osteoarthritis graded on a CT scan.

Thoracic Spine

The prevalence of patients who complain of chronic upper back or mid-back pain ranges from 3% to 22%.^30,143,144 One survey study of 35- to 45-year-olds estimated the prevalence of thoracic pain to be 15%.¹⁴⁵ Thoracic zygapophyseal joint pain referral patterns have been reported (Fig. 15–10).^146,147 Thoracic medial branch anatomy has also been described (Fig. 15–11).¹⁴⁸

FIGURE 15–10 Maps of referred pain patterns in segments indicated. A, Based on Dreyfuss et al¹⁴⁶ in normal volunteers. B, Based on Fukui et al¹⁴⁷ in patients with single positive facet block.

(From Bogduk N [ed]: Practice Guidelines for Spinal Diagnostic and Treatment Procedures. San Francisco, International Spine Intervention Society, 2004.)

FIGURE 15–11 A and B, Composite sketch of work by Chua and Bogduk¹⁴⁸ with radiographs of cadaveric thoracic spines. Medial branches of thoracic dorsal rami marked with wires to depict location with respect to transverse processes. Note middle thoracic levels, where medial branches are within intertransverse space versus crossing transverse process.

(From Bogduk N [ed]: Practice Guidelines for Spinal Diagnostic and Treatment Procedures. San Francisco, International Spine Intervention Society, 2004.)

There are no pathognomonic clinical or radiographic findings by which thoracic zygapophyseal joint pain may be diagnosed.¹⁴⁹ As with the cervical and thoracic spine, diagnosis is by suspicion and, at a minimum, the pain pattern should correlate with established pain referral maps.¹⁵ The methods physicians apply clinically to the diagnosis and treatment of thoracic facet joint pain rest largely on research done in the lumbar and cervical spine. This is not an entirely unreasonable approach based on what clinicians know in general regarding facet anatomy and innervation; however, more research is needed.

Investigators have mapped out the referral patterns for the thoracic joints. These findings are often used as a starting point to select which thoracic facets to block.^146,147 Dreyfuss and colleagues¹⁴⁶ mapped out thoracic facet joint referral patterns in normal volunteers and found that capsular distention did not provoke pain in 27.5% of volunteers. Fukui and colleagues¹⁴⁷ mapped out referral patterns in patients with suspected thoracic zygapophyseal joint pain who had a positive response to local anesthetic in C7-T1 to T2-3 and T11-12 facet joints. There was considerable overlap between the C7-T1 and T2-3 thoracic joints, and pain maps from these joints are not considered reliable enough to identify the symptomatic segmental level. Dreyfuss and colleagues¹⁴⁶ studied nine asymptomatic volunteers who underwent 40 provocative thoracic facet injections from T3-4 to T10-11. Referral patterns were consistently unilateral. The area of the most intense pain for segments from T2-3 to T11-12 was one level inferior and lateral. Significant overlap occurred over three to five levels. The researchers found that needle position can be confirmed with 0.1 to 0.3 mL of contrast dye, and adequate blocks can be achieved with a volume of 0.5 to 0.6 mL. Normally, thoracic zygapophyseal joints cannot hold more than 0.75 mL (Fig. 15–12 shows a typical thoracic zygapophyseal joint block).¹⁵

FIGURE 15–12 Left T3-4 zygapophyseal joint intra-articular injection. Note circular zygapophyseal joint arthrogram (arrow).

(Courtesy of Richard Derby, MD.)

One research group has performed the three studies in the literature using a controlled, double block paradigm, requiring 75% to 80% relief based on the duration of the local anesthetic used.^150–152 Combining all three studies with patients presenting with chronic middle or upper spinal pain (n = 183), using dual blocks obtains a 40% prevalence of thoracic facet syndrome, with a false-positive rate of 42% if using a single block paradigm.²⁶

What is the predictive value of a positive dual block? In other words, how well do patients fare who have positive dual blocks and undergo therapeutic intervention? Research is limited in this regard. One systematic review²⁶ reported that only therapeutic thoracic medial branch blocks received a 1A or 1B/strong recommendation. Manchikanti and colleagues^153,154 performed two studies. In the first study, 55 consecutive patients were studied; greater than 70% of patients had statistically significant relief (defined as >50% relief) at 3, 6, and 12 months. Most patients received four injections of bupivacaine with or without 1 mL of sarapin and 1 mg of methylprednisolone per milliliter of solution with 1 to 1.5 mL of solution injected per nerve. In the second study of 48 patients with positive dual blocks, 24 patients received bupivacaine, and 20 patients received bupivacaine plus betamethasone. Statistically significant (>50%) pain relief was reported in both groups at all time points up to 1 year. In the systematic review of radiofrequency neurotomy, only two studies were on thoracic medial branch neurotomy; however, both were of low quality and failed to meet inclusion criteria for the review because of lack of diagnosis by controlled blocks, small patient sample, and other methodologic shortcomings.²⁶ More research is needed in regard to diagnosis and treatment of thoracic pain so that the evidence can be graded and systematically reviewed, the caveat being that a lack of evidence is not equivalent to no evidence.

Summary

Chronic disabling spinal pain in a patient suggestive of facet syndrome that is unresponsive to usual care may be considered for diagnostic comparative facet intra-articular or medial branch blocks. The levels to be investigated are typically chosen by pain referral patterns described by the patient, which are correlated with validated zygapophyseal joint pain referral patterns. Upper neck pain and headache are most commonly caused by the C2-3 zygapophyseal joint, and neck pain with shoulder girdle pain is most commonly caused by the C5-6 zygapophyseal joint. The clinician should not neglect C0-1 and C1-2 as potential pain generators. Evaluation of the exact level of thoracic facet pain can be more challenging because pain may be referred over more than three segments. Lumbar zygapophyseal joint referral patterns are also reported in the literature; zygapophyseal joint pain may be localized or referred to the lower extremity.

Although comparative double blocks are considered the reference standard for diagnosis, routine history, physical examination, x-rays, and advanced imaging should be obtained for completeness. The clinician often finds elements that rule out facet syndrome and are more suggestive of disc pathology, radiculopathy, or “red flag” conditions that require different diagnostic and treatment methods. There are also cases where a history of trauma, particularly whiplash, is highly suggestive of pain of zygapophyseal joint origin, with a known greater than 50% prevalence in the cervical spine. Certain specific imaging findings, if present, also may suggest facet syndrome, such as a positive SPECT scan, approximately 2 mm edema on axial MRI of lumbar zygapophyseal joints, or a single zygapophyseal joint with markedly deforming arthropathy compared with other joints.

In terms of testing protocol, one’s preference to perform medial branch blocks, intra-articular zygapophyseal joint injections, or both varies depending on the situation and preference of the physician. If one is confirming zygapophyseal joint mediated pain in preparation for possible medial branch neurotomy, one could argue that medial branch block should be the method of choice. Because one medial branch innervates adjacent joints, if one is considering one of the current or emerging stabilization or intra-articular spacer facet devices, diagnosis should include or be limited to a specific intra-articular block. In addition, the medial branches innervate the multifidus muscle and supply branches to the supraspinous and interspinous ligaments and fascia. In the case of C0-1 and C1-2, intra-articular injections are the only practical method of diagnosing facet pain.

As noted earlier, preprocedural and postprocedural evaluation should be performed by unbiased personnel and checked by the physician using standardized instruments. Evaluation after the procedure includes VAS of standard provocative maneuvers and positions and a report of subjective percent relief of pain. Ideally, the patient would be tested at approximately 15 minutes after lidocaine block and approximately 30 minutes after bupivacaine block. A subject should have at least 50% relief for a positive response to be considered; at least 70% relief is more convincing. Usually, two to three levels are evaluated per session. Depending on the importance of refuting or confirming whether a particular zygapophyseal joint is symptomatic, one may select fewer joints if needed.

Several technical parameters must be met to obtain useful diagnostic information. Diagnostic volumes must be appropriate. Contrast medium must always be used to confirm accurate target identification. For intra-articular zygapophyseal joints blocks, injectant volumes should be limited to 0.3 mL, 0.75 mL, and 1 mL in the cervical, thoracic, and lumbar spine. For medial branch blocks, needle position must be confirmed with injection of a small volume (0.3 to 0.5 mL) of contrast dye and the same volume of local anesthetic. The interventionalist should observe for venous uptake or undesirable flow patterns. If there is venous uptake, there is only a 50% chance of successfully anesthetizing the joint, so the interventionalist may consider bringing the patient back at a later date or interpreting the results of the block accordingly.

Infection may occur after any interventional procedure. Various infections are reported after zygapophyseal joint injections, including paraspinal abscess,¹⁵⁵ facet abscess,¹⁵⁶ osteomyelitis,¹⁵⁷ and epidural abscess.¹⁵⁸ In addition to infections, subdural injections or injection into the spinal cord may occur. A case of transient tetraplegia¹⁵⁹ was reported during a cervical facet injection performed without fluoroscopy and most likely was an accidental subdural injection of local anesthetic. Even when using fluoroscopy there is a risk of accidental subdural injection or potential spinal cord injection. The danger is especially real when performing cervical intra-articular injection using a lateral technique. Using this technique, the needle is passed laterally using a lateral fluoroscopy view. If the anteroposterior view is not periodically checked, one may not recognize passage of the needle through the facet and dura and then into the cord. In a thin individual, the cord may be reached with a 1-inch needle. Keeping the needle directly over the inferior or superior facet and touching the bone before entering the joint helps the interventionalist avoid accidentally entering the spinal canal.

Pathophysiology

The sacroiliac joint has long been recognized as a synovial, fluid-filled diarthrodial joint between the sacrum and ilia with thick 6-mm sacral cartilage and thinner, approximately 1-mm iliac cartilage (Fig. 15–13). The joint is auricular or C-shaped with the convex side of the “C” facing anteriorly and inferiorly (Fig. 15–14).¹⁶⁸ Although the anterior portion is no more than a thickened capsule, the posterior capsule blends into the extensive, thick posterior ligamentous structures, which bind the sacrum to the spine and bilaterally to the ilia. After puberty, the iliac surface develops a convex ridge, and the sacral surface develops a corresponding concave depression. These articular surfaces allow slight movement between the contiguous bony surfaces.

FIGURE 15–13 Anteroposterior view of left sacroiliac joint injection. Note contrast dye filling capsule, including inferior capsular recess (arrows).

(Courtesy Richard Derby, MD.)

FIGURE 15–14 Lateral view of sacroiliac joint injection. Note contrast dye filling joint space. Also note C shape of joint facing anteriorly.

(Courtesy Richard Derby, MD.)

Although early in life gliding motions in all directions are permitted, by the middle of the 2nd decade of life the joints develop prominent ridges centrally along the entire length of the iliac surface and a corresponding groove along the sacral surface. Bowen and Cassidy¹⁶⁸ believed that this interdigitation of the joint surfaces restricts motion to a sagittal rotation or posterosuperior-anteroinferior “nodding” along the crest of the interdigitations. The motion is complex, however, and usually limited to less than 4 degrees of rotation and less than 1.6 mm of translation. Significant motion occurs only after severing the interosseous ligament.¹⁶⁹ It is unclear whether a type or degree of sacroiliac joint motion causes pain in older individuals. Beyond the 6th decade, cadaveric specimens commonly show a central region of ossification of the interosseous sacroiliac ligament and the presence of ridges and depressions, which likely result in little to no movement of the sacroiliac joint in these older individuals.¹⁷⁰ Although restricted by para-articular osteophyte formation, intra-articular bony ankylosis may be rare.¹⁶⁸

Several investigators have studied the innervation of the sacroiliac joint. Nakagawa¹⁷¹ reported innervation from the ventral rami of L4 and L5; the superior gluteal nerve; and the dorsal rami of L5, S1, and S2. An anatomic dissection of the innervation of the sacroiliac joint was performed by Yin and colleagues¹⁷² for the purpose of defining the exact position of the nerves for “sensory stimulation–guided sacroiliac joint radiofrequency neurotomy.” These authors dissected cadavers and placed small-gauge wires adjacent to the lateral branch nerves entering the joint and over the dorsal sacrum to the dorsal sacral foramen from S1 to S3. In 1998, Willard¹⁷³ reported dissection of 10 cadavers that revealed that the S1 and S2 lateral branches provide the primary innervation of the sacroiliac joint and associated dorsal ligaments. Occasional contribution was found by S3 but not S4. Predominant innervation from lateral branches of S1 was also reported by Grob and colleagues.¹⁷⁴ These authors found dorsal nerves derived from S1-4 exclusively innervated the sacroiliac joint and associated ligaments. Nerves were distributed to superficial and deep dorsal sacroiliac ligaments and to the sacrotuberous and sacrospinous ligaments. Emerging from the sacral foramen, the nerves course laterally, sandwiched between superficial and deep portions of the sacroiliac ligaments. There is a great variability in the location and number of lateral branch nerves side to side and between individuals.¹⁷² Currently, the standard for blocking the sacroiliac joint is to block the L5 dorsal ramus and S1-3 lateral branches.

Berthelot and colleagues^174a used the term sacroiliac joint lato-sensu to describe pain from the sacroiliac joint that may be emanating from adjoining ligaments rather than simply the synovial joint. These ligaments include the iliolumbar ligaments, dorsal and ventral sacroiliac ligaments, and sacrospinous and sacrotuberous ligaments. The prevalence of pain originating from these structures has received little formal study, and there is no validated technique to diagnose ligamentous pain. Nevertheless, sacroiliac joint ligamentous pain is proclaimed as a frequent primary source of low back and buttock pain by orthopaedists.¹⁷⁵ More importantly, a negative response to a sacroiliac joint injection does not mean that pain does not originate from the iliolumbar ligament and sacroiliac joint ligaments. A more recent histologic study found calcitonin gene-related peptide and substance P immunoreactive nerve fibers in the normal sacroiliac joint anterior capsular ligament and interosseous ligament. The authors of the study opined that diagnostic infiltration techniques for sacroiliac joint pain should employ extra-articular and intra-articular approaches.¹⁷⁶

In contrast to the zygapophyseal joints, the sacroiliac joint supporting ligaments are thick, and intra-articular injected local anesthetic may not adequately diffuse into the sacroiliac ligaments. Using a single or comparative block protocol, one can investigate sacroiliac joint ligaments by fluoroscopically guided injections of local anesthetics into the ligaments. Ligamentous injections have not undergone rigorous academic inquiry, however, and because the injections are rarely or poorly reimbursed by third-party payers and treatment of ligamentous laxity typically involves unreimbursed “prolo-therapy,” there is little incentive for expensive investigations. The information is important, however, and differential pain arising from the sacroiliac joint versus sacroiliac joint ligaments is reported.

In a comparative study, Murakami and colleagues¹⁷⁷ performed periarticular injections in 25 patients and intra-articular injections in another 25 patients. Periarticular injections relieved on average 92% pain in 100% of the injected patients compared with only 9 of 25 patients receiving intra-articular injections. All 16 patients not receiving relief by intra-articular injections were improved after periarticular injections. The presence of other structural abnormalities does not rule out the sacroiliac joint as a primary source of pain. Weksler and colleagues¹⁷⁸ studied 55 patients with herniated discs with axial and referred leg pain, without objective neurologic deficits but with positive sacroiliac provocation tests. Using intra-articular injection of local anesthetic as the reference standard, the mean baseline VAS pain score decreased 30 minutes after injection from 7.8 to 1.3. In 46 patients 8 weeks after injection, VAS scores ranged from 0 to 3.

The question of whether fusion surgery leads to increased stress on the sacroiliac joint and may be a cause of failed back surgery syndrome was first raised by Frymoyer and colleagues,¹⁷⁹ although their method of assessing sacroiliac joint pathology yielded a negative result; in 1978, Frymoyer and colleagues¹⁷⁹ evaluated patients with radiographs (no diagnostic blocks) 10 years after posterior fusion versus postdiscectomy and found no significant difference in radiographic abnormalities; they opined that sacroiliac pain was “noncontributory” to persistent low back pain after surgery. In their subject population, they believed that the graft donor site was a more common pain generator. Fusion to the sacrum might be expected to stress the sacroiliac joints and lead to late failures or to early failures owing to undiagnosed sacroiliac joint pain. Ha and colleagues¹⁸⁰ prospectively examined 37 patients undergoing posterolateral lumbar and lumbosacral fusions; 22 patients had a floating fusion, and 10 patients had a lumbosacral fusion. CT scans of the sacroiliac joint were performed before surgery and at 2 weeks, 1 year, and 5 years after surgery and compared with 34 matched controls. The incidence of sacroiliac joint degeneration was 75% in the fusion group versus 38.2% in the control group and greater in patients fused to the sacrum. Both groups reported significant improvements in VAS and Oswestry Disability Index scores, and there was no difference in scores between the two groups.

More recent research has shown that the sacroiliac joint can be a significant source of pain after fusion. Biomechanical models seem to support these conclusions. Ivanov and colleagues¹⁸¹ performed a finite element study with lumbosacral models and fusion constructs and found that fusion to the sacrum increased motion and stresses at the sacroiliac joint. Cadaveric studies show that disruption of the ventral band of the iliolumbar ligament significantly increases sacroiliac joint mobility.¹⁸² Ebraheim and colleagues¹⁸³ evaluated the prevalence of sacroiliac joint disruption by CT scan in 24 patients after fusion with persistent “donor site pain” after posterior superior iliac crest graft harvesting. They found a high prevalence of persistent sacroiliac joint pain in patients with inner table disruption. Patients with violation of the synovial portion of the sacroiliac joint had severe degenerative changes on CT versus mild to moderate degeneration with inner table disruption only. It seems that the original hypothesis by Frymoyer and colleagues¹⁷⁹ that sacroiliac joint dysfunction was the cause of donor site pain may have been correct.

What is the evidence for using diagnostic blocks as the reference standard? Diagnosis of sacroiliac joint pain has been reported by researchers using single and dual blocks; with these methods, prevalence rates of sacroiliac joint pain after lumbar fusion range from 27% to 35%. Maigne and Planchon¹⁸⁴ studied 40 patients after fusion with continued pain using 75% pain relief after a single sacroiliac joint intra-articular injection as the “gold standard.” They reported a 35% rate of positive blocks. The only characteristic that distinguished the positive from the negative responders was a different distribution of postoperative pain compared with preoperative pain. A pain-free interval of 3 months after surgery was significant; however, increased uptake in the sacroiliac joint on bone scintigraphy or posterior iliac bone graft harvesting was not significant.

Katz and colleagues¹⁸⁵ studied 34 patients after lumbosacral fusion with continued pain thought to be due to sacroiliac joint with intra-articular injections of local anesthetic and corticosteroids. Eleven patients (32%) had greater than 75% pain relief with local anesthetic and a minimum of 10 days’ continued pain relief (with steroid) and were considered to have definite sacroiliac joint pain. Another 10 patients (29%) had greater than 75% relief with local anesthetic but no long-term relief. There was no correlation between the donor site and pain side. Irwin and colleagues¹⁶³ used dual comparative sacroiliac joint blocks as the reference standard to define sacroiliac joint pain and found that the 27% positive responders tended to be older. They found no statistical relationship between age, body mass index, and gender.

Diagnostic Accuracy of Clinical History and Physical Examination for Sacroiliac Pain

The diagnostic utility of history and accepted sacroiliac joint physical examination tests was first rigorously examined by Dreyfuss and colleagues in 1996.¹⁸⁶ Their study was designed to determine if any single or combination of 12 history and physical examination findings could predict intra-articular sacroiliac joint pain as judged against a single positive intra-articular sacroiliac joint block with greater than 90% pain relief. In 85 patients, there were 45 positive blocks. None of the 12 physical examination tests or the presence of 5 to 12 positive tests or any combination of these 12 tests correlated with the presence of sacroiliac joint pain. One important historical feature was notable, however: only 2 of 45 patients drew pain above the L5 level, suggesting that pain below L5 is more likely to be of sacroiliac joint origin. Maigne and colleagues¹⁸⁷ reached a similar conclusion using dual comparative blocks: no single provocation test reached statistical significance in the 10 patients (18.5%) who had temporary pain relief on the confirmatory injection.

Although no single provocative maneuver has been shown to be of diagnostic value, using the dual block paradigm, several studies have obtained highly acceptable sensitivity (85% to 91%) and specificity (78% to 79%) rates by combining three or more sacroiliac joint pain provocation tests for diagnosis by physical examination.^{11,164,165,167,188} There is some slight variation in the tests used by various authors, but in summary they include the following provocation tests: thigh thrust, distraction test, Gaenslen test, Patrick sign, compression test, midline sacral thrust test, and heel drop test. Specificity increased to 87% if the patient’s pain did not centralize or could not be made to move toward the spinal midline (which is typical of discogenic pain).¹⁸⁹ When three or more provocation tests (distraction, compression, thigh thrust, Patrick sign, Gaenslen test) are negative, the likelihood of sacroiliac joint pain is very low (6% to 15%); when all provocation tests are negative, the sacroiliac joint was never the source of pain.^{164,165,167,189}

In terms of pain referral maps, Slipman and colleagues¹⁹⁰ and Dreyfuss and colleagues¹⁸⁶ concluded that of all alleged signs of sacroiliac joint pain, maximum pain below L5 coupled with pointing to the posterior superior iliac spine or tenderness just medial to the posterior superior iliac spine (sacral sulcus tenderness) has the highest positive predictive value of 60%; if these do not exist, the likelihood of sacroiliac joint pain is less than 10%. Although the maximal sacroiliac joint pain is below L5, pain can be referred into the entire lower extremity, with 94% of patients reporting buttock pain, 48% reporting thigh pain, and 28% reporting lower leg pain (Fig. 15–15).^187,191,192 Referral to the lower extremity is possible from sacroiliac joint pain and cannot reliably be distinguished from other pain sources (e.g., S1 radiculopathy).^191,193

FIGURE 15–15 Density of referral zones for sacroiliac joint pain. 0.5+ is the least common referral zone; 4+ is the most common referral zone.

(From Dreyfuss P, Dreyer S: Sacroiliac joint pain. J Am Acad Orthop Surg 12:255-265, 2004.)

Lastly, although pain referral patterns between responders and nonresponders are similar, Fortin and colleagues¹⁹⁴ described an area of pain approximately 3 × 10 cm just inferior to the posterior superior iliac spine that was common in all their subjects with sacroiliac joint pain. More recently, Murakami and colleagues¹⁹⁵ studied the specificity and sensitivity of the “Fortin” point with periarticular injections. Labeled the one finger test, 18 of 38 patients pointed to a location of pain at the posterior superior iliac spine or within 2 cm of the posterior superior iliac spine, which had a positive effect with periarticular sacroiliac joint block. The authors recommended that sacroiliac joint pain should be considered in patients who can point to their pain using one finger in the vicinity of the posterior superior iliac spine.

Systematic reviews report various conclusions regarding the specificity of the physical examination and sacroiliac joint block to diagnose sacroiliac joint pain based on the authors’ assessment of the diagnostic accuracy of diagnostic sacroiliac joint blocks. The review by Berthelot and colleagues¹⁹⁶ concluded that sacroiliac joint blocks and sacroiliac joint maneuvers were unreliable for diagnosing sacroiliac joint pain. In contrast, Hansen and colleagues¹⁹⁷ concluded in their review that there was moderate evidence for the specificity and validity of diagnostic sacroiliac joint injection and limited evidence for the accuracy of provocative maneuvers. Using a comparative double block reference standard, the most recent meta-analysis and systematic review concluded that the pooled data of the thigh thrust test, compression test, and three or more positive stress tests showed discriminative power for diagnosing sacroiliac joint pain.¹⁹⁸

Diagnostic Accuracy of Imaging

No imaging studies consistently provide findings that are helpful to diagnose primary sacroiliac joint pain. CT, MRI, and bone scan are done predominantly to exclude other causes of pain rather than to diagnose mechanical sacroiliac joint pain. Among patients referred to a low back pain clinic with a variety of pathologies, Hodge and Bessette¹⁹⁹ found a high percentage (75%) of patients with sacroiliac joint arthritis shown on CT scan. Although these authors did not confirm the diagnosis with sacroiliac joint injections, they opined that sacroiliac joint arthritis should be considered a possible diagnosis.

There is limited diagnostic value of CT scan in mechanical sacroiliac joint disease as defined by pain relief after sacroiliac joint blocks under CT scan guidance. Comparing the CT scans of patients diagnosed with sacroiliac joint pain using image-guided analgesic sacroiliac joint blocks with a matched control group of asymptomatic patients, Elgafy and colleagues²⁰⁰ reported that an abnormal sacroiliac joint CT scan had a sensitivity of 57% and a specificity of 69%. Although sacroiliac joint scintigraphy can detect early sacroiliitis,²⁰¹ stress fractures, infection, and tumors, the sensitivity of bone scans for detecting mechanical sacroiliac joint pain is poor (range 12% to 46%).^202,203 Patients with a positive bone scan are likely to have mechanical or arthritic sacroiliac joint pain with a reported specificity of 90% to 100%.^190,203

Diagnostic Accuracy of Sacroiliac Joint Injections

The current standard for diagnosing sacroiliac joint pain is pain relief after dual controlled sacroiliac joint injections, owing to the high false-positive rate of single blocks.¹⁶² When blocking the sacroiliac joint or lateral branches of the sacroiliac joint, imaging guidance must be used. The success of “blind” intra-articular injection is only 22%.²⁰⁴ A positive response should include approximately 70% relief for 1 to 2 hours of relief after a lidocaine block and 3 to 4 hours of relief after a confirmatory block with bupivacaine. Although the reference standard is reasonable, there are several caveats for the diagnosis of mechanical pain originating within the sacroiliac joint. Patients may exhibit extra-articular or periarticular sacroiliac joint pain or perhaps both. As noted earlier, Murakami and colleagues¹⁷⁷ relieved a significant amount of sacroiliac joint pain with periarticular injections. In a retrospective review of 120 patients, subjects who received intra-articular and periarticular injections had superior pain relief compared with subjects receiving intra-articular injections alone.²⁰⁵

False-positive results may occur secondary to leak of contrast dye through capsular tears, which may be present even in asymptomatic individuals. Extracapsular flow is present in 61% of sacroiliac joint intra-articular injections in patients.²⁰⁶ Of sacroiliac joint intra-articular injections, 27% show extravasation that communicates with nearby neural structures, including dorsal sacral foramina extravasation, superior recess extravasation at the sacral ala level to the fifth lumbar epiradicular sheath, and ventral extravasation to the lumbosacral plexus.²⁰⁶ Patients who have postblock extremity numbness are usually considered to have a leak, and the block is typically repeated at a different session.

More important is the potentially significant false-negative response rate because of a failure to anesthetize extracapsular pain sources mentioned previously. Block of the sacroiliac joint dorsal innervations may offer a solution because the block would potentially denervate intra-articular and extra-articular pain sources. Because the sacroiliac joint and ligaments are innervated, similar to the zygapophyseal joint, the joint and capsules are regarded as the same structure. In contrast to the zygapophyseal joint, the sacroiliac joint is surrounded by thick supporting ligaments, and intra-articular injected local anesthetic may not anesthetize the ligaments.

Dreyfuss and colleagues¹⁹³ used a double-blind randomized controlled trial to assess ability of single-site, single-depth L5 dorsal ramus and S1-3 lateral branch blocks to anesthetize the sacroiliac joint in 19 volunteers, using sacroiliac joint fluid distraction before and after blocks to determine effectiveness. The authors reported that only 40% of the volunteers did not feel distention after the blocks. The poor results prompted a cadaveric study of multisite, multidepth blocks to anesthetize the joint. L5 dorsal ramus block was performed at the standard location of the S1 superior articular process and the sacral ala; S1-2 lateral branches were blocked (right side) at the 2:30 o’clock, 4:00 o’clock, and 5:30 o’clock positions; and S3 lateral branch was blocked at the right 2:30 o’clock and 4:00 o’clock positions. The lateral branch blocks were performed 8 to 10 mm lateral to the posterior sacral foramen. A 0.2 mL volume of green dye was injected on the dorsal sacral plate, and an additional 0.2 mL was injected 2 to 3 mm above the sacral plate.

Dissection revealed S1-3 lateral branch nerves were stained in 91% (31 of 34) of cases. Employing the same protocol on 20 volunteer subjects using intraosseous ligament probing and capsular distention, Dreyfuss and colleagues¹⁹³ found that 86% of the sham local anesthetic injection subjects retained the ability to feel capsular distention, leading the authors to conclude that lateral branch blocks do not reliably block the intra-articular portion of the joint and that intra-articular blocks do not reliably block the extra-articular ligaments. One may conclude that to evaluate fully intra-articular and extra-articular pain sources, dorsal ramus and lateral branch blocks and intra-articular injections should be done. The caveat is the nerve blocks were successful in 70% of cases leaving a potential 30% false-negative cases. Injecting larger volumes or injecting the ligaments directly may potentially reduce the false-negative results with the risk of increasing false-positive results secondary to leak of local anesthetic through the posterior foramen.

Predictive Value

Surgical fusion outcomes for mechanical sacroiliac joint pain are reported for only a few small case series audits of initial outcomes after several “new” techniques for fusing the sacroiliac joint.^207–210 Published case series use pain relief after image-guided analgesic sacroiliac joint injections as the reference standard for diagnosing sacroiliac joint pain. Although Schutz and Grob²⁰⁸ reported an 82% unacceptable outcome after bilateral sacroiliac joint fusion in 17 patients based on results from sacroiliac joint anesthetic block, three other studies using novel techniques reported more favorable results for mostly unilateral fusions. Al-Khayer and colleagues²⁰⁹ reported an approximate 50% decrease in VAS and a 14-point decrease in Oswestry Disability Index in nine patients at 2 years after percutaneous sacroiliac joint arthrodesis using a Hollow Modular Anchorage screw (Aesculap, Sheffield, UK). Using percutaneously inserted fusion cages and bone morphogenetic protein, Wise and Dall²⁰⁷ reported an average back pain VAS improvement of 4.9 and leg pain VAS improvement of 2.4 in 13 patients at 6 months. Finally, Ziran and colleagues,²¹⁰ using CT-guided sacroiliac joint blocks as a reference standard, percutaneously fused 17 patients with recalcitrant sacroiliac joint pain and found a statistically significant correlation (P < .02) between final postoperative pain scores and preinjection and postinjection pain scores.

Evidence has been limited based on observation studies assessing the outcome of various treatments for sacroiliac joint pain. Cohen and colleagues²¹¹ selected patients for various types of radiofrequency neurotomy of the L4 medial branch, L5 dorsal branch, and S1-3 lateral branches using the reference standard of a single sacroiliac joint intra-articular block with greater than or equal to 75% relief of pain for 2 hours after injection of 2 mL of bupivacaine. Of 18 patients, 13 obtained satisfactory relief of pain with average scores reduced by 60%, 50%, and 57% at 1 month, 3 months, and 6 months. Only two patients in the placebo group obtained relief; pain scores of the placebo subjects were unchanged from baseline. Yin and colleagues¹⁷² used dual injection into the sacroiliac joint intraosseous ligament to diagnose sacroiliac joint pain. Of patients, 64% reported a minimum of 60% subjective pain relief for a minimum of 6 months after sensory stimulation–guided sacral lateral branch radiofrequency neurotomy.

Summary

Sacroiliac joint pain is a significant cause of chronic low back pain that is diagnosable and treatable with precision injection techniques. Prevalence of sacroiliac joint pain, based on a dual differential block protocol, ranges from 10% to 38%; for single, uncontrolled blocks, the false-positive rate is 54%.^163–167 The sacroiliac joint as a pain generator is no longer disputed. Current research also suggests that the sacroiliac joint is a significant source of persistent pain after lumbar fusion and may be a cause of graft donor site pain.^183–185 Although motion is limited and complex, the joint is known to rotate less than 4 degrees and to translate less than 1.6 mm. Anatomic studies have elucidated the innervation to the joint, with most practitioners directing diagnostic and therapeutic interventions to the L5 dorsal ramus and S1-3 lateral branches.^171,173 Sacroiliac joint pain is now thought to emanate from the joint itself and extra-articular ligamentous sources. Interventionalists are just beginning to diagnose and treat putative extra-articular pain generators.

In contrast to the history and physical examination for zygapophyseal joint pain, certain diagnostic features for sacroiliac joint pain have been validated by controlled blocks. Maximal pain below L5 coupled with pointing to the posterior superior iliac spine has a predictive value of 60%.^186,190 Although no single physical examination test has been shown to be of diagnostic value, using the dual block paradigm, several studies have shown high sensitivity (85% to 91%) and specificity (78% to 79%) by combining three or more provocative sacroiliac joint maneuvers.^165,167 Specificity increases to 87% if the patient’s pain cannot be centralized.²¹² Diagnostic imaging of sacroiliac joint pain has not been shown to be helpful, other than excluding nonmechanical causes of sacroiliac joint pain. The sensitivity of bone scans for detecting sacroiliac joint pain is poor (range 12% to 46%).^202,203 CT scan of the sacrum in a patient with persistent low back or buttock pain after lumbar fusion may be useful, particularly if the synovial joint has been violated. In these patients, severe degenerative changes were found on CT scan.¹⁸³

The current standard for diagnosis of sacroiliac joint pain is approximately 70% relief of pain for 1 to 2 hours after lidocaine block and 3 to 4 hours after bupivacaine block. Total volume should be limited to 1.5 mL. The interventionalist should carefully study the joint arthrogram for any evidence of extravasation via the dorsal sacral foramina, superior joint recess and fifth lumbar epiradicular sheath, or ventral capsule to the lumbosacral plexus because this can cause false-positive responses. Not all patients obtain relief from intra-articular joint injections, and extra-articular sources of pain must be evaluated as well. Other techniques for diagnosis and treatment of the sacroiliac joint include targeting the L5 dorsal ramus and S1-3 lateral branches.

If pain persists, new techniques have also been described for blocking the interosseous sacral ligaments.²⁰⁵ Regarding the predictive value of diagnostic sacroiliac joint injection for sacroiliac joint arthrodesis, some case studies show poor results for arthrodesis; other studies using novel techniques report better results.^207–210 Neurotomy of sacroiliac joint lateral branches after diagnostic block has shown promising results in an observational study.²¹³ Other researchers have also shown promising results with periarticular blockade.¹⁷⁷

Assessment of Effect

Only the study by Yeom and colleagues²²⁰ determined the optimal cutoff level in the percent relief of pain reported by a patient after a procedure needed to qualify for a positive response. Using receiver operator curves, Yeom and colleagues²²⁰ chose a cutoff of 70% subjective relief of pain after a lumbar transforaminal block as the best value to provide optimal accuracy but stated that this level could be adjusted depending on the importance of avoiding false-negative versus false-positive results.

The authors recommend adjusting the cutoff criteria between 50% and 90% depending on the importance of avoiding false-negative versus false-positive blocks. In the authors’ opinion, it is probably best not to treat results as a dichotomous variable but rather as a data point that is more or less likely to indicate the root is a source of pain. If a discrete cutoff is required, 70% is a good compromise. One might also consider requiring a similar degree in change of VAS improvement or, if inconsistent with the patient’s subjective report of pain relief, asking the patient why the discrepancy exists or performing a confirmatory injection. The patient often reports a global relief of pain, whereas diagnostically one is interested only in the degree of pain relief of the particular extremity distribution being evaluated. Relief or nonrelief of axial pain is important information but not pertinent to the location of the patient’s extremity pain and to surgical outcomes.

Although provocation of concordant pain was frequently used in the past and perhaps is useful information, more recent studies use techniques to avoid creating pain during injection.^220,222 Pain referral patterns obtained by electrical stimulation may be considered as supplementary proof or nonproof.²²²

Sensitivity

The most likely causes of low sensitivity or a high rate of false-negative injections are inadequate blocks owing to poor spread around the root, failure to reach pathologic site, dilutional effects with inadequate mass of anesthetic reaching the root, or poor diffusion because of scarring.²²⁰ van Akkerveeken²²¹ calculated 100% sensitivity in 46 patients using 0.2 to 0.5 mL of 0.5% bupivacaine (with provocation) and reported 100% of pain relief at 1 hour. Yeom and colleagues,²²⁰ using 1 mL of 2% lidocaine without considering provocation, calculated a lower sensitivity of 57% (27 of 47) in all patients, increasing to 71% (25 of 35) when injections with inadequate spread were excluded. The causes of the inadequate blocks were spread of injectant into adjacent tissues in 4 of 10 patients, block by huge herniation in 4 of 10 patients, and intraepiradicular sheath injection in 2 of 10 patients. Although Yeom and colleagues²²⁰ had no explanation in the remaining 10 cases, these false-negative results might be explained by a paracentral herniated disc, which, although affecting primarily the traversing root, may also cause chemical irritation of the exiting root.

In addition, Dooley and colleagues²²³ found that the most common reason for typical pain provocation during lumbar block with incomplete pain relief is multilevel pathology. The most probable cause in obvious cases is an inadequate block performed at a location distal to the structural entrapment. Diagnostic injections are often performed in patients with long-standing chronic pain and patients with prior surgery who may have intraneural and extraneural scarring. In such cases, local anesthetic may not penetrate the nerve effectively, and incomplete relief would be expected. Using a more concentrated anesthetic or an anesthetic that preferentially blocks nociceptors (e.g., bupivacaine) may reduce these false-negative responses.

Specificity

Because surgery is often less effective in patients with equivocal structural pathology, in patients with atypical, long-standing pain or prior surgery, one would ideally want to have minimal or no pain relief after the block of an asymptomatic root. van Akkerveeken²²¹ used a 0.2- to 0.5-mL volume of 0.5% bupivacaine and required 100% pain relief for 1 hour. He reported a specificity of “around” 90%.²²¹ In the lumbar spine, using 1 mL of 2% lidocaine and a cutoff value of equal or greater than 70% pain relief, Yeom and colleagues²²⁰ calculated 86% (50 of 58) specificity, which increased to 91% (43 of 47) specificity after excluding 7 patients with overflow of local anesthetic. Although this overflow was thought to be a probable cause of false-positive blocks in 4 of 11 cases, 7 of 11 cases were true-negatives, indicating that the estimated overflow when using 1 mL is about 20% (10 of 47) and with a potentially clinically observable effect in less than 10%.

Furman and colleagues²²⁴ showed that even after injecting only 0.5 mL, the contrast pattern indicated nonselective flow in 30% of lumbar injections. The mass of drug overflowing at these low volumes may not be significant and is consistent with van Akkerveeken’s higher, approximately 90% specificity. North and colleagues⁹⁵ reported an average 50% relief of sciatic pain when blocking the medial branches at several levels using a 3-mL volume, which would spread into the neuroforamen and epidural space, making the putative medial branch block nonspecific.⁸⁸ Nevertheless, convergence may be an alternative explanation of less than 50% pain relief in some cases, and a nonspecific “placebo” response may explain some or most false-positive responses.

Predictive Value

Many, mostly retrospective, observational studies describe in variable levels of detail the predictive value of lumbar root blocks. One retrospective study included the surgical predictive value of cervical and lumbar injections.²²² Another prospective, diagnostic cervical selective root block study compared the diagnostic value of imaging with the short-term surgical predictive value of the test.²²⁵ No studies to date support the use of diagnostic thoracic selective root injections, although this is primarily because the thoracic spine is not often studied because of the low prevalence of herniated thoracic discs.

In the only prospective outcome study, van Akkerveeken,²²¹ in his doctoral thesis, presented a series of studies correlating the value of selective root blocks to diagnosis of various lumbar entrapment syndromes and later summarized the data in a journal publication in 1993. A positive response was provocation of concordant pain and “disappearance” of leg pain after 0.2 to 0.5 mL of 0.5% bupivacaine. He studied patients with radiologic signs of nerve root entrapment but without localizing neurologic signs who subsequently underwent surgical decompression. Excluding the patients who had positive blocks and refused surgery, van Akkerveeken²²¹ reported a positive predictive value of 95% with a 95% confidence interval of 77% to 100%.