| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *Pain Management Division, Department of Anesthesiology and Critical Care Medicine, Johns Hopkins School of Medicine, Baltimore, Maryland; and Department of Surgery, Walter Reed Army Medical Center, Washington, DC.
Address correspondence and reprint requests to Steven P. Cohen, MD, Johns Hopkins Pain Management Division, 550 North Broadway, Suite 301, Baltimore, MD 21205. Address e-mail to scohen40{at}jhmi.edu.
Abstract
BACKGROUND: Since their first description more than 80 yr ago, the use of diagnostic spinal injections to predict surgical outcomes has been the subject of intense controversy. Because there are no standardized guidelines or substantive reviews on this topic, their use has remained inconsistent.
METHODS: Diagnostic procedures included in this review were lumbar and cervical discography, lumbar facet blocks, lumbar and cervical selective nerve root blocks, and sacroiliac (SI) joint injections. We garnered materials via MEDLINE and OVID search engines, books and book chapters, bibliographic references, and conference proceedings.
RESULTS: The lack of randomized, comparative studies for all blocks limited the conclusions that could be drawn. For the data that do exist, there is limited evidence that lumbar discography improves fusion outcomes, and no evidence that it influences disk replacement results. Although limited in scope, the current literature supports the notion that cervical discography improves surgical outcomes. There is strong evidence that lumbar selective nerve root blocks improve the identification of a symptomatic nerve root(s), and moderate evidence that both lumbar and cervical nerve root blocks improve surgical outcomes. The data supporting surgery for facet arthropathy are weak, and the use of screening blocks does not appear to improve outcomes. The data supporting SI joint fusion for degenerative, nontraumatic injuries are similarly weak. Because the most reliable method to diagnose a painful SI joint is with low volume, diagnostic injections, one might reasonably conclude that screening blocks improve surgical outcomes. However, this conclusion is not supported by indirect evidence.
CONCLUSIONS: The ability to evaluate the effect of diagnostic blocks on surgical outcomes is limited by a lack of randomized studies, methodological flaws, and wide-ranging discrepancies with regard to injection variables, surgical technique, and outcome measures. More research is needed to optimize injection techniques and determine which, if any, diagnostic screening blocks can improve surgical outcomes.
Spinal injections are the most frequently performed interventions in pain clinics across the United States, accounting for upwards of 75% of all procedures (1). For patients and many physicians, the ostensible reason for performing these procedures is pain relief. However, the high failure rate for surgery and nonsurgical interventions (2,3), coupled with the burgeoning economic cost of low back pain (LBP) (4), have led many experts to question the "shot-gun approach" to LBP management.
The use of diagnostic injections to identify the source of LBP dates back to the 1920s when von Gaza (5) used nerve blocks to illuminate obscure pain pathways. Motivated by the futility of treating pain without a proper diagnosis, Steindler and Luck (6) used procaine injections in the 1930s to identify specific pain generators in patients with chronic LBP. In the intervening years, spinal injections have been periodically advocated as both diagnostic and prognostic screening tools before surgery (7–12), but their use in this capacity has been sporadic and inconsistent.
Spurred by a wave of highly publicized articles questioning the wisdom of unrestrained surgery for LBP (2,3,13,14), a new philosophy has emerged emphasizing precision diagnosis together with high-tech interventions (15). Termed "reductionism" by Bogduk and McGuirk (15), one aspect of this strategy stresses the diagnostic and prognostic utility of nerve blocks as a prerequisite for optimizing treatment outcomes. Yet, despite the push towards improved diagnosis and refined selection criteria, the routine use of spinal nerve blocks as a screening tool for surgical intervention remains unproven and unpracticed. The purpose of this review is to systematically explore whether diagnostic spinal injections improve outcomes for a wide range of surgical procedures. Articles selected for inclusion were obtained via MEDLINE and OVID search engines, books and book chapters, and bibliographic references dating to the early 1900s.
PROVOCATIVE DISCOGRAPHY
Provocative discography is one of the most controversial diagnostic tests in pain medicine. First developed in 1948 as a diagnostic tool for herniated nucleus pulposus (16), the use of discography to diagnose radicular pain has since been supplanted by safer and more accurate imaging techniques such as computed tomography (CT) and magnetic resonance imaging (MRI). The rationale for performing discography is that it remains the only tool that purports to correlate symptoms with pathology. This reasoning seems justified considering that the estimated lifetime prevalence of LBP ranges from 60% to more than 80% (17,18), and that 60% of asymptomatic subjects display radiologic evidence of lumbar disk degeneration (19,20). Although the prevalence rates for neck pain are somewhat less than for LBP, the numbers are equally compelling. Whereas 16%–22% of adults suffer from chronic neck pain (21,22), over 70% of asymptomatic individuals have radiological evidence of cervical disk pathology.
The main criticism of discography is that disk stimulation may provoke pain in normal disks. The reported incidence of false-positive discography is contingent on multiple factors, including, but not limited to, investigator perspective (i.e., most studies that report high false-positive rates were done by spine surgeons), the use of manometry, criteria for positive discograms (e.g., the use of facial expressions or combined pain provocation and analgesic response to local anesthetic [LA] injection) and the patient population studied. For lumbar discography, published false-positive rates range from <10% (25,26) to upwards of 50% in susceptible individuals (27–30). The reported false-positive rate for cervical discography ranges from <5% to 27%, being higher in patients with chronic neck pain than in asymptomatic subjects (31,32). Previous studies conducted with lumbar disk provocation have found that patients with somatization disorder or hypochondriasis, depression, chronic pain complaints, and high psychometric scores are more likely to report concordant pain production in normal appearing disks (27–29). The main criticism regarding studies that have attempted to quantify false-positive discography rates is that disk stimulation in asymptomatic volunteers may not be an accurate indicator of pain provocation in nonpainful disks in subjects with LBP. One of the hallmarks of a positive discogram is concordant pain provocation, which is not possible in people without low back symptoms.
In addition to false-pain provocation, markedly degenerative disks may also fail to provoke pain during disk injection (33–36), especially in the elderly (34). In recent reviews by Cohen et al. (37,38), the authors estimated that 15%–25% of degenerative disks fail to elicit concordant pain during disk stimulation (Fig. 1).
|
Lumbar Discography and Fusion Outcomes
Despite the widespread use of lumbar discography as a presurgical screening tool, few studies have evaluated its effect on surgical outcomes. The relative lack of controlled studies is further compounded by the widespread variability in outcomes and the controversy surrounding spinal arthrodesis for discogenic LBP. The surgical outcomes for the treatment of internal disk disruption are widely acknowledged to be inferior than for radiculopathy, with the reported success rates ranging from <50% to >80% (3,39,40). Moreover, the few randomized studies that do compare arthrodesis to conservative treatment demonstrate mixed results (2,3,14). The presence of concomitant pain sources in most patients with discogenic pain, along with inconsistent clinical outcomes even with a technically successful arthrodesis, are factors that must be considered when evaluating the predictive value of discography for surgical outcomes.
In the largest and most methodologically sound study attempting to correlate discographic findings with arthrodesis results, Colhoun et al. (41) found a strong correlation between positive discography and surgical outcomes. The authors prospectively collected outcome data on 162 patients who underwent preoperative discography for axial LBP. In the 137 patients with concordant pain on discography, 89% had a favorable outcome at the mean follow-up period of 3.6 yr. In the 25 patients whose disks showed morphological abnormalities but elicited no provocation of symptoms, only 52% reported significant benefit. The surgical treatments evaluated were mainly spinal fusions.
Some retrospective observations have failed to duplicate these results. Esses et al. (42) examined the role of external spinal fixation in predicting the success of spinal fusion in 35 patients with refractory LBP. Thirty-two patients underwent provocative discography before fixator placement. To summarize these findings, neither concordant pain provocation nor discographic evidence of disk degeneration predicted pain relief with external fixator placement, or subsequent spinal fusion. The main flaw in this study is that it was not designed to evaluate the predictive value of discography for spinal fusion; hence, not all patients underwent preoperative disk stimulation.
Madan et al. (43) conducted a retrospective analysis designed to determine the effect provocative discography had on surgical outcomes in 73 patients with chronic LBP. The first 41 patients in this series underwent circumferential arthrodesis without preoperative discography, whereas the last 32 patients proceeded to surgery only if concordant pain was produced during disk stimulation. In the discography group, 75.6% of patients had satisfactory outcomes at the minimum 2-yr follow-up versus 81.2% in the group who did not undergo preoperative discography.
Finally, Derby et al. (11) concluded preoperative discographic screening was valuable based on a retrospective study investigating the influence discography had on outcomes in surgical and nonsurgical LBP patients. Disks were classified as chemically sensitive, mechanically sensitized, or negative/indeterminate. Although no differences were found between outcomes and the type of fusion for the entire sample taken as a whole, subgroup analysis revealed that those patients with chemically sensitized disks had better results following interbody/combined fusion (89% success rate) than after intertransverse fusion (20%), or nonoperative treatment (12%).
With the exception of Derby et al. (11), no study used manometry as a determining factor in discographic interpretation. In addition to enhancing patient safety by preventing over-pressurization of degenerative disks, the use of manometry objectifies discography by providing a reference standard. In two reviews by Cohen et al. (37,38), the authors found no difference in fusion outcomes between studies that used discography and those that did not. In summary, the lack of strong evidence for the use of fusion to treat degenerative disk disease and methodological flaws in the studies make data interpretation exceptionally difficult. For the data that do exist, there is limited evidence that discography improves fusion outcomes in patients with discogenic LBP (Table 1) (44).
|
Lumbar Discography and Disk Replacement Outcomes
Since its introduction in 1966 by Fernstrom (45), lumbar disk replacement has been used to treat discogenic LBP in Europe since the 1980s, and in the United States since the early 2000s. There are currently more than two dozen published studies evaluating disk replacement outcomes, which contain wide variations in outcome criteria and follow-up periods. The success rates in these studies range from around 50% to upwards of 90% (37,38), with approximately half routinely using discography as a preoperative screening tool (Table 2) (46–70). Although the lack of any direct outcome comparison between patients who were selected based on discography results and those who underwent disk replacement based purely on clinical and radiological findings precludes any meaningful conclusions, an indirect comparison does not reveal any significant differences in outcomes between those studies that used preoperative discography and those that did not.
|
Cervical Discography and Spinal Fusion
The evidence supporting cervical fusion to treat cervical discogenic pain is weak and conflicting (3). In a Cochrane review, Jacobs et al. (71) concluded that discectomy alone provides comparable symptom relief to that of fusion, but that it is associated with shorter recuperation times and hospital stays. Discography has thus been advocated as a means to improve surgical outcomes; however, the ability to discern the effect of discography on cervical fusion outcomes is limited by several factors. Foremost, there are no prospective studies comparing outcomes between cohorts who were screened with preoperative discography and those who were not. Second, in those studies that have used discography as a surgical screening tool, publication bias, serious methodological flaws, and wide variability in outcome measures and follow-up periods undermine the conclusions that can be inferred.
Yet, when all the data are assembled, a pattern emerges whereby higher success rates tend to be reported when discography is used as a screening tool before cervical fusion. Perhaps the most relevant study is by Kikuchi et al. (12) who conducted a retrospective survey evaluating surgical outcomes in patients who underwent anterior discectomy and spinal fusion for axial and radicular neck pain. In the 138 patients who were operated based on presurgical discography screening, 80% were either pain-free or had mild discomfort that did not interfere with work 1-yr postprocedure. In a control cohort who underwent cervical fusion without the benefit of discography, only 60% had favorable outcomes. In a prospective study by Hubach (72), 193 consecutive patients with neck pain and neurological symptoms were treated with discectomy and anterior fusion, with the operative levels in most patients being determined by intraoperative discography. In the first 23 patients who were fused without discography, 35% developed pain at an adjacent spinal segment. In the 156 patients who underwent fusion based on discographic abnormalities, only 12% developed juxtafusional pain. These findings indicate that discography may be beneficial not only in determining whether or not to perform surgery, but also in deciding what levels to operate on.
Similar to lumbar discography (37), cervical discography may be more sensitive in detecting anatomical abnormalities than MRI (73), though the clinical relevance of this remains unclear. In an observational study by Zheng et al. (32) the authors found a 64% correlation between MRI results and cervical discograms by level and 24% by patient. Surgical findings were not used as the "gold standard," but compared with CT-discography, MRI was associated with a 51% false-positive and 27% false-negative rate. Satisfactory results were obtained in 76% of patients. An earlier study by Schellhas et al. (31) also demonstrated that discography may be a more sensitive indicator of internal disk disruption than MRI. In 20 disks read as normal on MRI in 10 asymptomatic subjects, 17 were found to have annular tears discographically, although none was concordantly painful. Among 11 radiologically normal disks found in 10 patients with chronic cervical pain, 10 had annular tears during disk injection, with two being concordantly painful. In a study from the 1960s by Klafta and Collis (74), the authors found cervical discography to be less accurate than myelography in predicting surgical findings (72% vs 55%). The authors did not attempt to correlate pain provocation with either surgical findings or outcomes, but did note "pain on injection to be highly indicative of disk abnormalities." These results are in contrast with those of Simmons et al. (9,75), who found discography superior to myelography in predicting surgical findings.
There are more than a dozen studies, all of which are retrospective or observational, that examined outcomes in patients who have undergone anterior cervical fusion based on positive discograms. Most of them have reported success rates of 75%–80% at intermediate to long-term follow-up (75–82). Reported success rates for fusion studies that were done without discography tend to be more variable than those that have used discography as a screening tool. Positive results have ranged from around 30% to upwards of 90%, with most reporting around 60% success rates (71,83–85). Although limited in scope, the current evidence supports using provocative discography as a screening tool before cervical fusion (Table 3) (86).
|
|
LUMBAR FACET BLOCKS AND SPINAL FUSION
The prevalence of lumbar facet arthropathy varies widely, as reported in the literature, being functionally dependent on the diagnostic method and perspective of the investigator (i.e., higher prevalence rates are reported in pain literature than spine surgery literature). Further confounding estimates of prevalence rates are the high incidence of false-positive blocks, estimated at between 25% and 40% using comparative LA blocks or saline controls (93–96). Using single medial branch or intraarticular blocks, reported prevalence rates range from 8% to upwards of 90% (97,98). When placebo-controlled or comparative double blocks are used, false-positive rates decline significantly to between 9% and 40% (93,94,96,99,100). The frequent incidence of false-positive blocks has led some experts to advocate double-blocks as the only reliable method to diagnose the lumbar zygapophysial (l-z) joints as primary pain generators (94,101).
Multiple guidelines and review articles have promoted medial branch blocks (MBB) and intraarticular injections as having comparable diagnostic utility (101–103). However, a critical review of the evidence cited for these claims revealed only two randomized studies comparing MBB and intraarticular injections (Fig. 2) (104–106). Neither study used a crossover design or prescreened patients for l-z joint pain based on diagnostic injections, and both used excessive injectate volumes, which undermines specificity (106). Thus, although MBB and facet joint injections appear to provide similar relief to axial back pain sufferers, the question of which more accurately diagnoses a painful z-joint(s) remains unanswered. In a randomized comparative study comparing the prognostic value of MBB to pericapsular facet injections in patients undergoing radiofrequency denervation, Birkenmaier et al. (107) found better results lasting up to 6 mo postprocedure in the MBB group. But unlike intraarticular facet injections, pericapsular injections are not usually advocated as an accurate means to diagnose facet arthropathy. In general, higher success rates have been found for facet joint radiofrequency denervation in those studies using controlled or confirmatory blocks (108–110).
|
There are several studies assessing the ability of intraarticular facet injections to predict spinal fusion outcomes and surgical findings, all of which are limited by methodological flaws, high volumes in those studies in which injectate composition was listed, and the absence of controlled or comparative blocks. In general, the results of these studies have been disappointing. Bough et al. (111) conducted a retrospective review to determine whether pain provocation during facet arthrography could predict histological findings in 84 patients who underwent facetectomy. Whereas the specificity for provocative l-z joint injection was 75%, the low sensitivity (59%) and negative predictive value (43%) led the authors to conclude facet arthrography was of little value as a surgical screening tool. A flaw in this study was the use of pain provocation to diagnose facet joint pain. Previous studies have found pain provocation to be a less reliable indicator of painful zygapophysial joint than analgesic response to LA infiltration (112). In a small observational study by Esses et al. (42), the authors found a nonsignificant negative association between the response to diagnostic facet blocks and short-term outcomes from external spinal fixation. In a later retrospective study, Esses and Moro (113) found no correlation between the response to facet blocks and success for either operative or nonoperative treatment in 296 patients with axial LBP. These findings are consistent with those of Jackson (114), who found no difference in arthrodesis outcomes between those patients who experienced significant pain relief after l-z injections and those who did not.
In an observational study, Lovely and Rastogi (7) reported excellent or good results with spinal fusion in 23 of 28 patients who obtained >70% pain relief after three facet blocks. However, the findings in this study are undermined by widespread methodological flaws, including the large volumes used during blocks (3–5 mL) and the lack of an adequate comparison group (results were available on only one "negative" facet block that failed to fuse). In summary, the evidence does not support using diagnostic lumbar facet blocks as a predictive tool before spinal fusion. There are currently no studies evaluating the predictive value of either thoracic or cervical facet blocks as screening tools before surgical intervention (Table 4) (115).
|
SELECTIVE NERVE ROOT BLOCKS
Selective nerve root blocks (SNRB) have been used for almost 40 yr to provide diagnostic and prognostic information in patients with radicular pain. Although the terms "SNRB" and "transforaminal epidural injection" are sometimes used interchangeably, the two are different procedures with distinctly separate indications. Transforaminal epidural steroid injections are specifically used to treat radicular pain. Because the injectate is purposefully administered into the epidural space where it ostensibly spreads to adjacent spinal levels, the diagnostic information that a transforaminal epidural injection provides is limited by the lack of specificity. On the other hand, because SNRBs purportedly anesthetize only one spinal nerve, they are often touted as accurate diagnostic and predictive tools for planned decompression procedures (Table 5). In addition to their diagnostic utility, nerve root injections done with steroids and LA have been shown in some studies to provide intermediate-term pain relief and reduce the need for operative intervention (116–121). However, the potential beneficial effects of these procedures, especially transforaminal injections performed with particulate steroids, have been tempered by multiple reports of spinal cord injury, paraplegia and even death (122–125), prompting some clinicians to propose alternative methods of selectively anesthetizing nerve roots (126).
|
Despite the theoretical appeal of SNRB, their diagnostic utility is limited by several factors including a narrow therapeutic window whereby the analgesic effects can be realized and epidural or plexus spread avoided, dermatomal overlap, lack of uniform effects on sensory function, individual variations in dermatomal maps, inability of patients to discern pain provocation and/or pain relief after anesthetic injection, use of excessive injectate volumes, superficial anesthesia and sedation, and inconsistency between nerve root stimulation patterns and dermatomal pain patterns.
One caveat that should be heeded when performing SNRB and other analgesic diagnostic procedures is the need to limit injectate volumes to optimize specificity. In a landmark prospective study by North et al. (127) conducted in 33 patients with clinical and radiological evidence of lumbosacral radiculopathy, the authors performed a battery of LA nerve blocks that included SNRB, sciatic nerve block, MBB, and subcutaneous control injections. Approximately 90% of patients obtained near-complete pain relief after SNRB, 70% obtained almost complete relief after the sciatic block, and a majority received more than 50% pain relief after the MBB. In contrast, the median degree of pain relief after the subcutaneous injection was around 30%. These findings led the authors to conclude that the specificity of diagnostic nerve blocks for the evaluation of spinal pain is low. However, the volumes used during these blocks (3 mL for each procedure) far exceeded that used in current standard practice.
Low-volume SNRB will produce predictable dermatomal sensory changes for between 75% and 85% of patients in the cervical and lumbar spine, being somewhat less valid and reliable in patients with long-standing or multisegmental neurological deficits. One of the first studies to evaluate the validity of SNRB was conducted by Castro et al. (128), who randomized 94 patients to receive CT-guided L4 nerve root injections using 0.5, 1, or 2 mL of contrast. Epidural spread was noted to occur in 48%, 67%, and 75% of subjects in each of these groups, respectively. More significantly, definitive spread to an adjacent nerve root, which may undermine the specificity of a nerve block by causing anesthesia outside the intended dermatome, was found in 24% of the 0.5 mL group, 27% of the 1 mL group, and 33% of injections done with 2 mL of contrast. Spread into the psoas muscle, where the nerve roots converge to become a plexus, was noted in 12% of patients who received 0.5 mL contrast, 33% of subjects who received 1 mL injections, and 68% of people who were injected with 2 mL contrast. These results raise further questions regarding the validity and specificity of SNRB, even when low volumes are used.
Wolff et al. (129–131) performed a series of studies designed to evaluate the sensory and motor effects of segmental nerve root blocks in patients with chronic radicular LBP. In the first study, the authors performed 42 low-volume (0.5 mL) SNRB in 29 patients (129). Using standard dermatomal maps (132), hypesthesia and paresthesias were noted in 80% of patients. However, significant pain reduction was only achieved in 43% of patients. In the second study, 38 low-volume (0.7 mL) SNRB were performed in 20 patients with lumbosacral radiculopathy (130). In all but two patients, hypesthesia to pinprick was noted in the corresponding dermatome level. Although the median decrease in pain score was 4 points on a 0–10 scale, some patients reported no reduction in symptoms. In the third study, the authors performed 20 L4 SNRB in 10 patients with either ropivacaine or lidocaine in a double-blind, crossover fashion (131). In the seven patients with baseline sensory deficits, SNRB produced variable, but nonsignificant, changes in the extent and distribution of preexisting hypesthesia. Clinically significant (
2 points) postblock pain reductions occurred in only eight of the 20 block sessions. These findings indicate that SNRB may be less valid in patients with long-standing, nondermatomal sensory changes.
Several investigations have demonstrated that properly performed cervical SNRB can successfully identify a painful nerve root(s). Anderberg et al. (133) performed cervical SNRB with 1 mL of LA in 20 patients with cervical radiculopathy correlating with single-level MRI pathology. In the 10 patients whose pain was provoked with active neck motion, 9 (90%) reported significant reduction in arm pain (mean 85%). In the 10 patients whose arm pain was not provoked with neck movement, all experienced significant pain reduction (mean 88%). In a similar study conducted in 30 patients with two-level MRI pathology, the same group of authors found a 60% correlation between SNRB results and the most severely degenerated spinal level based on MRI, a 27% correlation between SNRB and neurological examination, and a 23% correlation between nerve root blocks and classical dermatome maps (134). Among the 11 patients with a positive SNRB at two-levels, 6 (55%) had consonant disk pathology on MRI. In only 20% of patients did neurological deficits, MRI and SNRB results correlate.
Following up on the Castro et al. (128) study, Anderberg et al. (135) used multislice CT to evaluate the contrast dispersal patterns of transforaminal nerve root blocks in nine patients with cervical radiculopathy using volumes of 0.6, 1.1, or 1.7 mL. In the two lowest volume groups, a small amount of contrast spread was noted at the adjacent nerve root in two of the three injections; however, because the contrast encompassed less than one-quarter diameter of the nerve root, all blocks were considered valid. In the three 1.7 mL injections, none was selective enough to be valid. Similar to lumbar SNRB, these findings underscore the need to use low volumes to enhance the specificity of cervical SNRB. Finally, Slipman et al. (136) performed a prospective study comparing the pain referral patterns for 134 cases of cervical nerve root stimulation and documented sensory dermatomal maps. The authors found that although the distribution of symptom provocation resembled classic dermatomal patterns, pain and paresthesias were often provoked outside these maps.
SNRB and Lumbosacral Decompression
Several authors have attempted to correlate the results of SNRB with surgical findings and outcomes. Haueisen et al. (8) reported good operative results in a retrospective study conducted in 55 patients with lumbosacral radiculopathy who underwent surgical exploration based on SNRB. In the 46 patients with a "definitive" or highly suggestive nerve root injection, corresponding surgical pathology was found in 93% of patients. In contrast, myelography and electromyography correctly identified the site of pathology in only 24% and 38% of patients, respectively. At follow-up periods ranging from 1 to 5 yr, 49% of patients had minimal or no pain versus 16% of patients who were treated nonoperatively. In an effort to determine the predictive value of SNRB, Stanley et al. (137) performed a prospective study in 50 patients with lumbosacral radicular pain. In the 19 surgically treated patients in whom typical pain was reproduced during injection and relieved after LA infiltration, SNRB correctly identified the level of surgical pathology in 18 patients, which favorably compared with the predictive value of CT (14 patients) and radiculography (12 patients).
In a retrospective evaluation performed in 62 patients with sciatica, Dooley et al. (138) found the combination of concordant pain provocation and analgesic response to LA infiltration during SNRB to have both strong positive and negative predictive value for determining nerve root pathology and predicting surgical outcomes. In the 46 patients whose typical pain was reproduced during needle placement and relieved by low-volume SNRB (Group I), all but one had surgical confirmation of nerve root pathology. In the 32 patients found to have a herniated disk (n = 8), bony entrapment (n = 17) or intraneural adhesions (n = 7) as the primary cause of radicular symptoms, 100%, 82%, and 71% experienced significant relief of radicular pain after decompression, respectively. In the four patients in whom pain was produced by needle placement but not relieved by LA injection (Group II), surgical exploration tended to show multiple root involvement.
Sasso et al. (10) demonstrated that preoperative SNRB can improve both lumbar and cervical surgical outcomes. In a retrospective analysis performed in 101 patients who underwent lumbar (n = 83) or cervical (n = 18) decompression, 90% of the 91 patients with a positive block experienced a good outcome at the mean 16-mo follow-up versus 60% in the 10 patients with negative blocks. The predictive value of these blocks compared favorably with MRI. In summary, there is strong evidence that SNRB can improve accurate identification of a symptomatic lumbar spinal level, and moderate evidence that SNRB improves operative outcomes. Nerve root blocks are most advantageous when low volumes are used, both pain provocation and analgesia are used as criteria for success, and possibly when adjacent/control levels are blocked in addition to the suspected spinal level.
SNRB and Cervical Decompression
Although still promising, the direct evidence supporting SNRB as a predictive tool for cervical decompression outcomes is less robust than for lumbar decompression, owing to a smaller body of literature on the subject. Anderberg et al. (133) conducted a prospective study comparing SNRB results to clinical symptoms and MRI findings in 20 patients with single-level cervical disk disease. In the 18 patients who underwent cervical decompression based on postblock pain relief and radiological findings, all experienced eradication of their radicular symptoms after operative intervention, although five continued to have shoulder pain. The follow-up period in this study was not noted. In a later study performed on 30 patients with multilevel cervical disk pathology, the same group reported good or excellent surgical outcomes in 9 of the 11 patients after low volume, dual-level cervical nerve root block (mean follow-up 31 mo) (134). This compared favorably to patients treated with transforaminal epidural steroid injections (69% success rate in 13 patients) or conservative therapy (50% success rate in eight patients). In the Sasso et al. (10) study reporting better surgical outcomes in patients with positive SNRB (91% good outcomes) than in those with negative SNRB (60% good outcomes), results were not tabulated based on site of pathology. However, 18% of patients presented with cervical radiculopathy. The positive predictive values of SNRB and MRI were similar (91% vs 88%), whereas the negative predictive value for SNRB was significantly higher than that for MRI (40% vs 15%). In conjunction with indirect evidence from data extrapolated from lumbar SNRB studies, there is moderate evidence supporting improved surgical outcomes when cervical SNRB are used before decompression procedures (Fig. 3; Table 6) (139–143).
|
|
|
Sacroiliac (SI) joint pain is a challenging condition affecting between 15% and 25% of patients with axial low back and buttock pain (144,145). Dozens of provocative maneuvers and alignment/mobility tests have been advocated as diagnostic aids in patients with suspected SI joint pain. However, numerous studies and reviews have found no single historical or physical examination feature to be a reliable indicator of a painful SI joint (144,146,147). Similar diagnostic shortcomings have been found for radiological studies, including radionuclide bone scanning, CT, and radiographic stereophotogrammetry (148–151). The lack of reliable alternative diagnostic modalities has led numerous experts to recommend diagnostic SI joint blocks as the most reliable means for diagnosing an SI joint as the primary pain generator (Fig. 4) (144,145,152). To maximize accuracy, these injections should always be performed using radiological guidance. In a prospective, double-blind study by Rosenberg et al. (153), the authors found intraarticular placement in only 22% of SI joint injections guided by anatomical landmarks. But even accurate needle placement using a low injectate volume does not guarantee validity or specificity. A prevalence study by Maigne et al. (154) conducted in 54 patients found that although 19 had a positive response to a screening SI joint block done with lidocaine, only 10 experienced >2 h of pain relief after a confirmatory bupivacaine block. Other studies have found similar false-positive rates for uncontrolled SI joint injections (155).
|
Whereas controlled studies have shown that peri-articular SI joint injections may provide comparable relief to intra-articular injections, the latter are generally regarded as the gold standard for identifying an SI joint(s) as the likely pain generator. Although speculative, the impetus to use intra- rather than peri-articular injections as a prognostic tool before SI joint fusion may be even more compelling, since the ostensible goal of arthrodesis in many cases is to provide stabilization for a painful, mal-aligned or overly mobile joint.
There are several indications for SI joint arthrodesis, including fractures, dislocation/instability, and arthritis. Of these, SI joint pain secondary to degenerative joint disease is the most controversial. Many factors undermine the conclusions that can be drawn regarding the ability of SI joint blocks to improve surgical outcomes, including differences in operative techniques, outcome measures and indications; disparities in the way diagnostic injections were performed; publication bias in the mostly retrospective published reports (i.e., case series and retrospective reports are more likely to be published when positive than negative); failure to properly report outcomes and injection variables; widespread methodological flaws; and lack of any direct comparison between patients who underwent preoperative screening blocks and those who did not. Particularly noteworthy is that all of the studies that used presurgical SI joint screening blocks either used excessive volumes or failed to mention the amount of LA injected. In two retrospective studies from the 1920s and 1930s whereby SI fusion was performed without diagnostic blocks, the earlier study reported excellent results in 85% of patients (22 of 26 patients returned to baseline), whereas the other reported excellent results in 52% of subjects (158,159). In the earlier study, both trauma and nontraumatic patients were included, although results were not separated based on etiology (158). In the later study (159), the primary indication in more than three-quarters of the patients was infection. Neither study used objective outcome measures or specified the follow-up period. In a more recent study by Dabezies et al. (160), compression rods were used to treat 11 cases of mostly trauma-related SI dislocation without the benefit of screening blocks. Although technical success was quantified, no mention was made of patient outcomes except for a brief notation that two patients continued to have mild residual back pain and one experienced worsening of a nerve injury. The average follow-up in this study was 26 mo.
Among the three largest studies evaluating SI fusion contingent on a positive response to fluoroscopically-guided intraarticular SI joint injections, two reported benefit. In a retrospective study assessing the results of 22 operations performed in 21 patients with SI joint osteoarthritis, Waisbrod et al. (161) found that half their patients obtained >50% pain relief at a mean follow-up of 2.5 yr. In the Discussion, the authors estimated that if patients with "psychosomatic disease" are excluded, about a 70% success rate can be expected for SI joint arthrodesis. In another retrospective study, conducted in 20 nontrauma patients with SI joint pain, Buchowski et al. (162) reported significant improvement in all SF-36 categories except for general health in the 15 patients who completed pre- and postoperative outcome assessments, with 60% of patients claiming they would repeat the procedure. Schutz and Grob (163) reported less auspicious results in a recent study evaluating bilateral fusion for SI joint arthropathy. Among their 17 study patients, three reported moderate or complete pain relief at the mean 39-mo follow-up period, with only 18% being satisfied with their results. Flaws in this study include the high volumes (5 mL) used in the 14 patients who underwent preoperative SI joint blocks and the fact that all fusions were bilateral. In most patients with injection-confirmed SI joint pain, the predominant symptoms tend to be unilateral (Table 7) (164–167).
|
In summary, the evidence does not appear to show any advantage in performing preoperative SI joint blocks when the primary surgical indication is trauma-related fracture(s) and/or dislocation. The conclusions regarding the use of SI joint blocks before surgery for arthritis are more complex, partly because the evidence supporting SI joint arthrodesis for degenerative joint disease is weak and inconsistent. Because the evidence is strong that low volume diagnostic blocks are the gold standard for classifying a degenerative joint as painful, one might conclude, a priori, that screening blocks should be done before surgical intervention. However, a review of the disparate outcomes of the literature does not support the conclusion that patient selection based on SI joint blocks will improve the results of SI joint fusion.
DISCUSSION
In an ideal setting, there would be no diagnostic nerve blocks. Noninvasive diagnostic modalities would be 100% accurate, and the evidence that they improved treatment outcomes would be incontrovertible. The only reason pain doctors would insert needles into patients would be to treat their underlying condition. But the world is not perfect, noninvasive modalities are not infallible, and diagnostic spinal injections will probably continue to be done as long as patients have back pain and surgeons are willing to operate on them (Table 8).
|
There are myriad causes for inaccurate diagnostic spinal injections which run the gamut from technical errors, lack of communication between provider(s) and patient, anatomical anomalies, poor patient understanding, expectation bias, placebo effect, overzealous use of superficial anesthesia, inappropriate administration of sedation and/or analgesics, excessive volumes of LA administered during blocks, systemic absorption of LA, and failure to adequately interpret the results of the procedure (106,144,167–170). Because similar shortcomings abound for imaging studies, there will never be a perfect correlation between nerve blocks and less invasive diagnostic tests. The same caveat holds true for nerve blocks and surgery. For these reasons, diagnostic nerve blocks rarely provide a definitive answer to the question of pain causation. Nerve blocks should therefore always be interpreted in context, as one piece of a complex puzzle, in conjunction with historical findings, physical examination, other diagnostic modalities and psychosocial issues. In light of the high failure rates for many surgical procedures and the high false-positive rates associated with most diagnostic spinal injections, negative blocks often provide more useful information than positive blocks (27–30,93–96,154,155).
Perhaps the most striking finding of this analysis is the absence of any methodologically sound, prospective, randomized study examining the impact of diagnostic blocks on surgical outcomes. Although this may render any conclusions we draw as purely speculative, it also underscores the need for better research studies.
For discography, there is weak evidence that preoperative disk provocation improves arthrodesis outcomes for lumbar and cervical fusion. The largest prospective study with the longest follow-up period found a substantial increase in success rates when concordant pain provocation, rather than morphological abnormalities, was used as the basis for surgical intervention (41), whereas smaller, retrospective studies failed to replicate these results (42,43). The evidence supporting cervical discography as a presurgical screening tool is one-sided in favor of outcome improvement, but mitigated by the anecdotal nature of the literature. The preponderance of data neither support surgery as a reliable treatment for facet arthropathy, nor the use of diagnostic facet blocks when surgery is attempted for this indication.
There is more outcome literature available for SNRB than any other diagnostic modality. Yet the evidence remains equivocal because there are no randomized studies comparing surgical outcomes between patients who underwent decompression surgery based on SNRB and those who received other screening procedures. The results of properly performed SNRB appear to correlate better with surgical pathology than imaging studies for lumbar disk surgery. Based on inductive reasoning, one might therefore assume that SNRB should improve surgical results. Although anecdotal, positive SNRB do seem to be associated with better intermediate-term surgical outcomes in both the lumbar and cervical regions. Thus, our conclusion is that there is moderate evidence to support using SNRB to improve surgical decompression outcomes.
Whether diagnostic blocks should be done before SI joint fusion is a complex question that must be asked in the framework of the more overriding question of whether SI joint fusion should even be done for degenerative joint disease. Taken as a whole, the retrospective studies that have examined SI joint arthrodesis outcomes do not show improved outcomes when SI joint blocks are routinely used as screening procedures. However, the wide variations in surgical technique and outcomes, and lack of standardization for the diagnostic injections do not permit definitive conclusions to be drawn as to the utility of SI joint screening blocks. If one operates under the unproven premise that SI joint fusion can be beneficial in a small subset of patients suffering with arthritis, then a priori one must conclude that diagnostic blocks can improve outcomes, because it is generally accepted that diagnostic blocks are the best way to diagnose a painful SI joint (144,147,171).
Driven by better-informed patients, ever-expanding treatment options, and increasing demand by third party payers for stronger evidence of good treatment outcomes, the selection criteria for spinal surgery will continue to evolve. As the frequency with which diagnostic injections are performed continues to increase, so will the need to optimize their validity. Presently, there are little data to support or refute the use of diagnostic injections as screening tools before spine surgery. More research is needed to determine what role, if any, diagnostic injections will play in the evaluation of spine surgical candidates.
Footnotes
Accepted for publication August 13, 2007.
Funded in part by the John P. Murtha Neuroscience and Pain Institute, Johnstown, PA, through the Army Regional Anesthesia and Pain Management Initiative.
The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense.
REFERENCES
