Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
The Spine Journal 8 (2008) 498–504 Clinical Studies Lumbar zygapophysial (facet) joint radiofrequency denervation success as a function of pain relief during diagnostic medial branch blocks: a multicenter analysis Steven P. Cohen, MDa,b,*, Milan P. Stojanovic, MDc, Matthew Crooks, MDa, Peter Kim, MDd, CPT Rolf K. Schmidt, MDb, LTC Cynthia H. Shields, MDb, LTC Scott Croll, MDb, Robert W. Hurley, MD, PhDa a Pain Management Division, Department of Anesthesiology & Critical Care Medicine, Johns Hopkins School of Medicine, 550 North Broadway, Suite 301, Baltimore, MD 21029, USA b Department of Surgery, Walter Reed Army Medical Center, 6900 Georgia Avenue NW, Washington, DC, 20307 USA c Pain Management Division, Department of Anesthesiology & Critical Care, Massachusetts General Hospital, Harvard Medical School, 15 Parkman Street, WACC 324, Boston, MA, 02144 USA d Department of Anesthesiology, University of Southern California School of Medicine, Los Angeles, CA, 90033 USA Received 8 February 2007; accepted 24 April 2007 Abstract BACKGROUND CONTEXT: The publication of several recent studies showing minimal benefit for radiofrequency (RF) lumbar zygapophysial (l-z) joint denervation have led many investigators to reevaluate selection criteria. One controversial explanation for these findings is that the most commonly used cutoff value for selecting patients for l-z (facet) joint RF denervation, greater than 50% pain relief after diagnostic blocks, is too low and hence responsible for the high failure rate. PURPOSE: To compare l-z joint RF denervation success rates between the conventional greater than or equal to 50% pain relief threshold and the more stringently proposed greater than or equal to 80% cutoff for diagnostic medial branch blocks (MBB). STUDY DESIGN/SETTING: Multicenter, retrospective clinical data analysis. PATIENT SAMPLE: Two hundred and sixty-two patients with chronic low back pain who underwent l-z RF denervation at three pain clinics. OUTCOME MEASURES: Outcome measures were greater than 50% pain relief based on visual analog scale or numerical pain rating score after RF denervation persisting at least 6 months postprocedure, and global perceived effect (GPE), which considered pain relief, satisfaction and functional improvement. METHODS: Data were garnered at three centers on 262 patients who underwent l-z RF denervation after obtaining greater than or equal to 50% pain relief after diagnostic MBB. Subjects were separated into those who received partial (greater than or equal to 50% but less than 80%) and near-complete (greater than or equal to 80%) pain relief from the MBB. Outcomes between groups were compared with multivariate analysis after controlling for 14 demographic and clinical variables. RESULTS: One hundred and forty-five patients obtained greater than or equal to 50% but less than 80% pain relief after diagnostic MBB, and 117 patients obtained greater than or equal to 80% relief. In the greater than or equal to 50% group, success rates were 52% and 67% based on pain relief and GPE, respectively. Among patients who experienced greater than 80% relief from diagnostic 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 Department of Defense. Funded in part by the John P. Murtha Neuroscience and Pain Institute, Johnstown, PA and the Army Regional Anesthesia & Pain Medicine Initiative, Washington, DC. 1529-9430/08/$ – see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.spinee.2007.04.022 * Corresponding author. Steven P. Cohen, MD, Pain Management Division, Department of Anesthesiology & Critical Care Medicine, Johns Hopkins School of Medicine, 550 North Broadway, Suite 301, Baltimore, MD 21029, USA. Tel.: (410) 955-1818; fax: (410) 614-7597. E-mail address: [email protected] (S.P. Cohen) S.P. Cohen et al. / The Spine Journal 8 (2008) 498–504 499 blocks, 56% obtained greater than or equal to 50% relief from RF denervation and 66% had a positive GPE. CONCLUSIONS: Using more stringent pain relief criteria when selecting patients for l-z joint RF denervation is unlikely to improve success rates, and may lead to misdiagnosis and withholding a potentially valuable treatment from good candidates. Ó 2008 Elsevier Inc. All rights reserved. Keywords: Denervation; Facet joint; Low back pain; Medical branch block; Predictive value; Radiofrequency; Zygapophysial joint Introduction Lumbar zygapophysial (l-z) joint pain is a challenging condition affecting approximately 15% of patients with chronic, axial low back pain (LBP) [1]. One of the biggest challenges in managing l-z (facet) joint pain is the lack of any standard, reliable treatment. Until recently, radiofrequency (RF) denervation of the nerves innervating the l-z joints was widely considered to be the gold standard for treating lumbar facet arthropathy, having been shown to provide intermediate to long-term relief in patients who positively respond to diagnostic blocks [2–4]. However, the publication of three consecutive negative trials evaluating RF denervation for lumbar and cervical z-joint pain has cast doubts as to the efficacy of RF lesioning [5– 7]. Whereas this has led some clinicians to question the concept of RF lesioning for z-joint pain, others have implicated flaws in technique and selection criteria as the likely causes for treatment failure [8–11]. One variable yet to be examined for its relationship to RF treatment outcome is the pain relief cutoff for designating a diagnostic block as ‘‘positive.’’ Several authors have claimed that the identification of z-joints as pain generators is best accomplished only after complete or near-total relief of back pain after low-volume diagnostic injections [12–15]. In addition to the logic behind this assertion, this claim seems to be borne out by an examination of prospective studies evaluating medial branch RF denervation. Among six prospective studies evaluating lumbar or cervical medial branch RF denervation that used near-total pain relief after diagnostic blocks as an inclusion criterion, all reported excellent outcomes [14–19]. In similar studies using 50% pain relief as the threshold for lesioning, about half reported positive outcomes [3–7,20]. Yet despite the surface validity of this comparison, any conclusions that can be drawn regarding the superiority of choosing a higher analgesic threshold for RF lesioning is limited by a plethora of confounding variables including a lack of direct comparison between outcomes based on the degree of pain relief experienced after diagnostic blocks, publication bias, variations in RF technique and other selection criteria, multiple publications by the same groups of authors, and the fact that most studies that used complete or near-complete pain relief as an inclusion criterion evaluated cervical (rather than lumbar) z-joint denervation. As a source of spinal pain, cervical z-joints have consistently been found to account for a higher percentage of chronic neck pain than lumbar z-joints do in LBP sufferers [1,21,22]. Perhaps most significant is that in almost all of the prospective studies using greater than or equal to 80% pain relief as a cutoff value, either placebo-controlled or comparative local anesthetic blocks were used to minimize the high false-positive rate of uncontrolled z-blocks, estimated at between 25% and 40% [23–26]. In a busy clinical practice, the recommendation to use double blocks is often disregarded on the bases that the definitive RF neurotomy procedure carries a serious complication (eg, epidural abscess, meningitis) rate comparable to that of each diagnostic injection [1]; each diagnostic injection is associated with a small but quantifiable false-negative rate [27]; and the use of controlled diagnostic blocks is not cost effective [28]. This is the case in many academic pain clinics and military treatment facilities, including our own. Even in the positive studies whereby 50% pain relief was designated as the RF threshold, because no subgroup analyses were performed, it is possible that the beneficial effects were realized mostly in patients who obtained nearly complete pain relief from diagnostic blocks. The question as to what the optimal cutoff should be before proceeding to RF lesioning is both clinically and academically relevant. To illustrate, withholding a definitive treatment from someone who experiences only partial pain relief from z-blocks but is nevertheless likely to benefit may potentially lead to misdiagnosis, increased disability, unnecessary interventions, and amplified costs. On the other hand, performing RF lesioning on patients who experience partial pain relief from diagnostic blocks and are consequently predisposed to treatment failure exposes them to unnecessary risks, wastes valuable resources, and threatens the viability of the RF denervation because it undermines the very concept of the procedure. To help determine the optimal pain threshold to guide treatment in patients being considered for RF denervation, we conducted a retrospective analysis of large, multicenter databases comparing outcomes between patients experiencing partial but significant pain relief (greater than or equal to 50% but less than 80%) after diagnostic medial branch blocks (MBB) and those who experience near-total relief (greater than or equal to 80%) after diagnostic injections. 500 S.P. Cohen et al. / The Spine Journal 8 (2008) 498–504 Patients and methods Permission to conduct this study was granted by the Internal Review Boards at three different institutions and all patients who provided informed consent for the procedures. The institutions participating in this study were Johns Hopkins Medical Institutions, Walter Reed Army Medical Center (WRAMC), and Massachusetts General Hospital. The data collected for this study were garnered from the medical records of 298 consecutive patients who underwent lumbar medial branch and L5 dorsal rami RF denervation between July 2003 and July 2006 for suspected l-z joint pain. Thirty-six subjects with ambiguous records or inadequate follow-up were excluded, leaving 262 patients eligible for inclusion. Patients were selected for diagnostic MBB based on presenting symptoms and physical examination. Inclusion criteria for MBB were age greater than 18 years, chronic LBP greater than 3 months duration, and absence of focal neurological signs or symptoms. Excluded from consideration were patients with Grade II or higher spondylolisthesis, symptomatic spinal stenosis, vertebral fractures, untreated coagulopathy, or concomitant medical or psychiatric illness likely to compromise evaluation or treatment. Lumbar medial branch and L5 dorsal ramus blocks Diagnostic MBB were performed using our previously described single [29,30] and conventional multiple needle techniques [4,8,9,17]. Before needle placement, the skin at each entry point was anesthetized using less than 1 mL of 1% lidocaine. Patient with unilateral pain underwent unilateral blocks; those with bilateral or central pain received bilateral blocks. The number of levels blocked varied according to the patient’s symptoms. Correct needle placement was confirmed in both anteroposterior and oblique fluoroscopic views after negative aspiration and contrast injection. At each level, 0.5 mL of bupivacaine or ropivacaine was administered. No patient received intravenous sedation. Consistent with most outcome studies [4,6,8,9,20,31,32], greater than or equal to 80% pain relief during normal activities was used as the criterion for a positive response to account for concurrent sources of back pain. All patients who obtained significant pain relief after MBB proceeded to RF denervation at their next visit. the diagnostic block was determined retrospectively by an attending physician. Radiofrequency denervation RF denervation was performed as an ambulatory procedure using superficial local anesthesia and if necessary, intravenous sedation. With the C-arm intensifier positioned to confer a slightly oblique or anteroposterior view, 22-gauge SMK-C10 cannulas with 5-mm active tips (Radionics, Burlington, MA) were inserted parallel to the course of the nerve until bone was contacted at the junction between the superior border of the transverse process and superior articular process for all medial branch lesions, and at the junction of the ala and articular process of the sacrum for all L5 dorsal ramus lesions. At each site, correct placement was confirmed using electrostimulation at 50 Hz, with concordant sensation achieved at less than 0.5 V. Before lesioning, multifidus stimulation and the absence of leg contractions were verified with electrostimulation at 2 Hz. After satisfactory electrode placement, 0.5 mL of 1% lidocaine was injected through each cannula to reduce thermal pain. The RF probe was then reinserted and a 90 second, 80 C lesion was made using an RF generator (Electrothermal 20S Spine System; Smith and Nephew, Andover, MA or Radionics RF Lesion Generator System, Model RFG3C; Radionics, Valleylab, Boulder, CO). Outcome measures All pain scores were measured using a 0 to 10 visual analog scale or numerical pain rating score. A successful treatment was defined as a greater than 50% or equal to average reduction in preprocedure pain score that persisted at least 6 months after the procedure. For the WRAMC patient subset, a secondary outcome measure, global perceived effect (GPE), was recorded based on the response to three questions routinely administered to all patients in the pain clinic. These questions were 1) My pain is better now than before treatment; 2) The treatment I received improved my ability to perform daily activities; and 3) I am satisfied with the treatment I received and would recommend it to others. An affirmative response to all three of these questions at the 6-month follow-up visit was considered a positive GPE. A negative response to any of these questions constituted a negative GPE outcome. Grading of pain relief after medial branch blocks Statistical analysis Before the procedure and again in the recovery area, patients were instructed to engage in their normal daily activities and to maintain a written 0 to 10 numerical pain diary every 30 minutes for 6 to 8 hours. In addition to pain scores, diaries were used to monitor postblock activities. The classification as to whether patients experienced partial (greater than or equal to 50%, but less than 80%) or nearcomplete (greater than or equal to 80%) pain relief from In addition to the aforementioned outcome measures, other demographic and clinical variables recorded for analysis were age, sex, duration of pain, opioid usage, pain referral pattern(s), presence of paraspinal tenderness, presence of pain worsened by extension and/or rotation (ie, ‘‘facet loading’’), magnetic resonance imaging evidence of l-z joint hypertrophy or degeneration, smoking history, trauma history, presence of diabetes, obesity S.P. Cohen et al. / The Spine Journal 8 (2008) 498–504 (defined as body mass index greater than 30), previous neck surgery, laterality, and number of levels treated. Statistical analyses were performed using STATA version 9.2 (Statacorp, College Station, TX) as in a previous study [9]. The distribution of categorical variables in each group was compared using Pearson’s chi-square. Continuous variables were compared with analysis of variance. Categorical data are reported both by number of patients and percentage. Continuous data are reported as mean and standard deviation unless otherwise indicated. A p value less than .05 was considered statistically significant. Bonferroni correction was used for post hoc analysis. Data were combined across institutions. Regression analysis was used to quantify the association between the many possible predictive variables and clinical outcome. Because the outcome variable was binary (either positive or negative), a logistic statistical model was chosen. Unadjusted univariate analyses were performed, followed by multivariate logistic regression. Variables found to be significant in multivariate logistic regression in a previous publication [9] were included in the multivariate model. Results Data were analyzed on 262 patients. Morphometric, demographic, and clinical characteristics were somewhat heterogeneous among the three treatment centers. WRAMC patients were younger, more often male, and less likely to smoke cigarettes, use opioids, or have a BMI greater than 30. Massachusetts General Hospital patients had a higher incidence of paraspinal tenderness, ‘‘facet loading,’’ and pain radiation below the knee. Johns Hopkins Medical Institutions patients had a higher prevalence of diabetes (Table 1). Because no statistically significant difference with regard to treatment outcome between study centers was found in either univariate or multivariate logistic analysis, all data were combined. The mean age of the subjects was 54.2 years (SD515.4, range 17–89). Their average duration of LBP was 5.7 years (SD55.9, range 0.5–40), and a slight majority (53%) were female. Fifty-five percent of all subjects received greater than or equal to 50% but less than 80% pain relief from the diagnostic block. On average, subjects had three levels treated with denervation (SD50.5, range 2–7). Overall, 54% of the subjects reported greater than or equal to 50% pain relief persisting at least 6 months postprocedure, and 66% of the subjects reported a positive GPE. No single demographic, physical symptom or sign was associated with patient-reported differential pain relief after diagnostic MBB in either univariate analysis, or when all covariates were controlled for using multivariate logistic regression (Table 2). Greater pain relief from diagnostic MBB (80% vs. 50%) was not associated with greater success as defined by greater than or equal to 50% relief of pain persisting for 6 months after RF denervation 501 Table 1 Patient characteristics by study center WRAMC (n5155) JHMI (n567) MGH (n540) p Value* Age in years, mean (SD) 51.2 (15.6) 57.1 (13.6) 61.2 (14.6) 0.002 Sex 0.09 Male 82 (53%) 27 (40%) 15 (38%) Female 73 (47%) 40 (60%) 25 (62%) Success 89 (57%) 36 (54%) 17 (43%) 0.24 Smoking 28 (19%) 21 (32%) 15 (38%) 0.02 Diabetes 14 (10%) 18 (27%) 3 (8%) 0.001 Scoliosis 16 (11%) 12 (20%) 11 (28%) 0.03 Location of symptoms 0.22 Axial 96 (62%) 31 (46%) 18 (45%) Above knee/groin 37 (24%) 22 (33%) 8 (20%) Below knee 22 (14%) 14 (21%) 14 (35%) % Relief from 0.22 diagnostic block 50% 92 (59%) 35 (52%) 18 (45%) 80% 63 (41%) 32 (48%) 22 (55%) Obesity 54 (36%) 36 (54%) 17 (50%) 0.03 Opioid use 35 (23%) 33 (49%) 15 (38%) 0.001 Failed back surgery 30 (19%) 11 (16%) 13 (33%) 0.11 syndrome 6 (6) 5.6 (6.5) 4.8 (4.4) 0.44 Duration of symptoms in years, mean (SD) Facet pathology on MRI 90 (68%) 40 (65%) 27 (68%) 0.86 Pain exacerbation by 88 (58%) 42 (68%) 31 (79%) 0.04 extension/rotation (facet loading) Paraspinal tenderness 78 (54%) 41 (66%) 35 (88%) 0.001 Number of levels 3.1 (0.5) 3.2 (0.6) 3 (0.3) 0.12 treated, mean (SD) WRAMC5Walter Reed Army Medical Center; JHMI5Johns Hopkins Medical Institutions; MGH5Massachusetts General Hospital; SD5standard deviation; MRI5magnetic resonance imaging; RF5radiofrequency; ANOVA5analysis of variance. Success defined as $50% pain relief persisting at least 6 months after RF denervation. Data are presented as number (%) unless otherwise specified. * Age, duration of symptoms, and number of levels treated were compared with ANOVA; categorical data were compared with Pearson’s chi-square. (p5.52). Additionally, greater pain relief from diagnostic blocks was not associated with improved patient satisfaction as measured by GPE (p5.89). The lack of association with either pain reduction by numerical pain rating score or GPE persisted in multivariate analysis (p5.84 and p5.54, respectively; Table 3). Discussion The principal finding in this study is that the degree of pain relief obtained after diagnostic screening blocks does not correlate with l-z joint denervation outcomes. It suggests that using 50% pain relief as the benchmark for a positive diagnostic procedure, which based on the current literature is the most commonly used cutoff, leads to comparable success rates and superior overall outcomes (because it is more inclusive) than using more stringent 502 S.P. Cohen et al. / The Spine Journal 8 (2008) 498–504 Table 2 Patient characteristics by response to diagnostic block Age in years, mean (SD) Sex Male (n5124) Female (n5138) Institution WRAMC (n5155) JHMI (n567) MGH (n540) Success (n5142) GPE improvement (n5100) Smoking (n564) Diabetes (n535) Scoliosis (n539) Location of symptoms Axial back pain only (n5145) Above knee/groin (n567) Below knee (n550) Obesity (n5107) Opioid use (n583) Failed back surgery syndrome (n554) Duration of symptoms, mean (SD) Facet pathology seen on MRI (n5157) Pain exacerbated by extension/rotation (facet loading) (n5161) Paraspinal tenderness (n5154) Table 3 Factors associated with successful outcome* Greater than 50% relief (n5145) Greater than 80% relief (n5117) 53.1 (16) 55.5 (14.7) 63 (43.4%) 82 (56.6%) 61 (52.1%) 56 (47.9%) 92 35 18 76 60 31 21 25 63 32 22 66 40 33 14 14 p Value 0.21 0.16 0.22 (63.4%) (24.1%) (12.4%) (52.4%) (66.7%) (22.8%) (15.2%) (19.2%) (53.8%) (27.3%) (18.8%) (56.4%) (65.6%) (29.2%) (12.3%) (12.6%) 78 (53.8%) 67 (57.3%) 40 27 53 50 31 27 23 54 33 23 (27.6%) (18.6%) (38.4%) (34.5%) (21.4%) (23.1%) (19.7%) (47.8%) (28.2%) (19.7%) 0.52 0.89 0.25 0.50 0.16 0.71 0.14 0.28 0.73 5.99 (6.6) 5.44 (5.0) 0.46 87 (67.4%) 70 (66.7%) 0.90 92 (67.6%) 69 (59.5%) 0.18 80 (58.4%) 74 (68.5%) 0.10 WRAMC5Walter Reed Army Medical Center; JHMI5Johns Hopkins Medical Institutions; MGH5Massachusetts General Hospital; SD5standard deviation; MRI5magnetic resonance imaging; GPE5global perceived effect; RF5radiofrequency; ANOVA5analysis of variance. Success defined as $50% pain relief persisting at least 6 months after RF denervation. GPE percentages calculated from a denominator of 151 patients ($50% n590, $80% n561). Data are presented as number (%) unless otherwise specified. Age, duration of symptoms, and number of levels treated were analyzed with ANOVA. Categorical data were compared with Pearson’s chi-square. criteria for identifying RF candidates. Whether lower thresholds might result in similar outcomes remains unknown because most previous studies used 50% pain relief as the threshold for proceeding to denervation. The rationale for using 50% pain relief as the criteria for RF treatment has been elegantly outlined by Schwarzer et al. [23], who cite the high incidence of concurrent spinal pathology occurring with l-z joint degeneration as the primary reason. This is perhaps best illustrated by the work of Fujiwara et al. [33], who found that whereas lumbar degenerative disc disease frequently occurs in the absence of z-joint degeneration, the reverse is almost never true (ie, patients with severe l-z joint osteoarthritis virtually always have radiologic evidence of degenerative disc disease and/or other spinal pathology). % Relief from diagnostic block Failed back surgery syndrome Duration of symptoms Pain exacerbated by extension/rotation (facet loading) Paraspinal tenderness Successful outcome p Valuey GPE p Valuey 0.95 0.59 0.98 0.48 0.84 0.11 0.33 0.01 0.78 0.40 0.93 0.5 0.54 0.04 0.04 0.09 2.25 0.005 2.25 0.04 GPE5global perceived effect; RF5radiofrequency. Success defined as $50% pain relief persisting at least 6 months after RF denervation. Failed back surgery syndrome, ‘‘facet loading’’ and duration of symptoms were significantly associated with negative outcome. Paraspinal tenderness was significantly associated with positive outcome. * Data are presented as Odds Ratio (95% Confidence Limits). y p Values determined from adjusted multivariate logistic regression analysis. The impetus behind choosing more stringent denervation criteria seems equally compelling. MBB are inherently nonspecific, even when low volumes are injected under fluoroscopic guidance [1,34]. In a cadaveric study conducted by Kellegren [35] in the 1930s, the injection of as little as 0.5 cc of contrast was found to encompass an area of 6 cm2 of tissue. A later study by Dreyfuss et al. [36] found that performing low-volume (0.5 mL) MBB using conventional landmarks resulted in contrast spread into the epidural space or intervertebral foramen in 16% of cases, and between the cleavage plane of the multifidus and longissimus muscles in all injections. In addition, the excessive use of superficial anesthesia, intravenous or oral sedation, and a host of other factors may contribute to analgesia unrelated to nerve blocks [1,37,38]. Thus, even when RF denervation is flawlessly performed, the degree of long-term pain relief achieved is unlikely to match that which follows diagnostic MBB. There are several explanations for our findings. First, the group of patients who experienced near-total pain relief may have contained a higher percentage of placebo responders than their counterparts who obtained only partial pain relief. Whereas placebo responders might also be expected to positively respond to the definitive treatment, the benefit is unlikely to be sustained for a prolonged time period [4,39]. One method advocated to screen out placebo (or false-positive) responders is to exclude patients who obtain prolonged, discordant pain relief after their diagnostic block. Whereas some patients may experience prolonged relief after MBB with local anesthetic secondary to a resetting of pain thresholds, most should not [24,40,41]. A second possible reason no difference was found in outcomes is because once a certain analgesic threshold is exceeded, a host of different confounding factors diminish the relationship between the degree of pain relief experienced after blocks and the extent l-z joint arthropathy S.P. Cohen et al. / The Spine Journal 8 (2008) 498–504 contributes to back pain. Secondary factors that may contribute to pain relief or the lack thereof after diagnostic MBB include the use of sedation and/or supplemental analgesics, excessive use of superficial anesthesia, a patient’s motivation to adhere to normal activities and maintain a pain diary after the procedure, accuracy of needle placement, the aptitude of the injector, and a patient’s ability to distinguish between their typical symptoms and procedurerelated pain [1,37,38]. Because the role each of these factors plays varies on a case-to-case basis (ie, an anxious patient might be more likely to increase their pain medications the morning of the procedure or require higher dosages of local anesthesia), it is difficult to control for one, let alone all of these variables. Whether the same factors that account for the high false-positive rate of uncontrolled l-z blocks are also responsible for the imprecise correlation between short-term pain relief after diagnostic MBB and denervation outcomes has yet to be determined. There are several flaws in this study that need to be addressed. First and foremost is its retrospective nature, which fails to control for the plethora of potential confounding factors. Yet controlling for these confounding variables would not simulate realistic conditions, whereby anxious patients do often preemptively self-medicate with sedatives and/or analgesics, and patients who experience severe procedure-related pain usually are administered larger quantities of local anesthetic. Collecting data retrospectively also tend to result in more excluded subjects than prospective analyses and less accurate measurements. However, the large size of this study and multiple sources of data might conceivably serve to mitigate these effects. Another shortcoming is that pain relief was classified categorically instead of as a continuous variable. This was deemed appropriate because of slight variations in the way pain relief was assessed and diaries interpreted (ie, in some cases patient comments were used to qualify numerical pain scores) by the responsible attending physician. For obvious reasons, we also did not perform RF on patients who experienced less than 50% pain relief after the diagnostic blocks, although some of these patients would have likely responded to treatment. Whereas classifying the response categorically may be more accurate in some cases, it prevents the use of a receivers operating characteristic curve to determine an optimal cutoff value for RF lesioning. Future studies should attempt to address this deficiency. Finally, some may criticize the lack of controlled blocks as a major flaw in this study. Whereas the use of double blocks may have eliminated some placebo responders and consequently resulted in a higher success rate, the total number of successful outcomes would have also probably been less. However, the presence of a purer target population may obviate the need to establish an optimal cutoff point for definitive treatment. The goal of establishing an ideal threshold is only relevant because of the low success rates reported when single diagnostic blocks are used. 503 Thus, the 50% cutoff advocated here cannot be extrapolated when the more stringent diagnostic criterion of a concordant response to controlled or comparative local anesthetic blocks is used. In conclusion, the results of this study suggest that using more stringent pain relief criteria when selecting candidates for l-z joint RF denervation is unlikely to improve outcomes, but may lead misdiagnosis and the withholding of a potentially beneficial treatment. Although our findings revealed 50% pain relief to be a better cutoff point than 80% pain relief, the optimal threshold cannot be determined from this study. Moreover, the results of this study should not be misconstrued as evidence to support the efficacy of facet joint RF denervation, for which better-designed, controlled studies are needed. Prospective studies classifying the response to MBB as a continuous rather than a categorical variable, and that subject a wide range of responders to RF treatment, are necessary to best determine the ideal cutoff point for facet joint denervation. References [1] Cohen SP, Raja SN. Pathogenesis, diagnosis and treatment of lumbar zygapophysial (facet) joint pain. Anesthesiology 2007;106:591–614. [2] Geurts JW, Van Wijk RM, Stolker RJ, Groen GJ. Efficacy of radiofrequency procedures for the treatment of spinal pain: a systematic review of randomized clinical trials. Reg Anesth Pain Med 2001;26:394–400. [3] Gallagher J, Petriccione di Vadi PL, Wedley JR, et al. Radiofrequency facet joint denervation in the treatment of low back pain: a prospective controlled double-blind study to assess its efficacy. Pain Clinic 1994;7:193–8. [4] van Kleef M, Barendse GA, Kessels A, Voets HM, Weber WE, de Lange S. Randomized trial of radiofrequency lumbar facet denervation for chronic low back pain. Spine 1999;24:1937–42. [5] Leclaire R, Fortin L, Lambert R, Bergeron YM, Rossignol M. Radiofrequency facet joint denervation in the treatment of low back pain. A placebo-controlled clinical trial to assess efficacy. Spine 2001;26: 1411–7. [6] van Wijk RM, Guerts JW, Wynne HJ, Hammink E, Buskens E, Lousberg R. Radiofrequency denervation of lumbar facet joints in the treatment of chronic low back pain: a randomized, double-blind, sham lesion-controlled trial. Clin J Pain 2005;21:335–44. [7] Stovner LJ, Kolstad F, Helde G. Radiofrequency denervation of facet joints C2-C6 in cervicogenic headache: a randomized, double-blind, sham-controlled study. Cephalalgia 2004;24:821–30. [8] Cohen SP, Larkin T, Chang A, Stojanovic M. The causes of falsepositive medial branch blocks in soldiers and retirees. Mil Med 2004;169:781–6. [9] Cohen SP, Hurley RW, Christo PJ, Winkley J, Mohiuddin MM, Stojanovic MP. Clinical predictors of success and failure for lumbar facet radiofrequency denervation. Clin J Pain 2007;23:45–52. [10] Lau P, Mercer S, Govind J, Bogduk N. The surgical anatomy of lumbar medial branch neurotomy (facet denervation). Pain Med 2004;5: 289–98. [11] Dreyfuss P, Baker R. Reply to Leclaire et al. Spine 2002;27:556–7. [12] Bogduk N. International Spinal Injection Society guidelines for the performance of spinal injection procedures. Part 1: Zygapophysial joint blocks. Clin J Pain 1997;13:285–302. [13] Manchikanti L, Singh V. Review of chronic low back pain of facet joint origin. Pain Physician 2002;5:83–101. 504 S.P. Cohen et al. / The Spine Journal 8 (2008) 498–504 [14] Barnsley L. Percutaneous radiofrequency neurotomy for chronic neck pain: outcomes in a series of consecutive patients. Pain Med 2005;6: 282–6. [15] McDonald GJ, Lord SM, Bogduk N. Long-term follow-up of patients treated with cervical radiofrequency neurotomy for chronic neck pain. Neurosurgery 1999;45:61–7. [16] Lord SM, Barnsley L, Wallis BJ, McDonald GJ, Bogduk N. Percutaneous radio-frequency neurotomy for chronic cervical zygapophyseal-joint pain. N Engl J Med 1996;335:1721–6. [17] Dreyfuss P, Halbrook B, Pauza K, Joshi A, McLarty J, Bogduk N. Efficacy and validity of radiofrequency neurotomy for chronic lumbar zygapophysial joint pain. Spine 2000;25:1270–7. [18] Sapir DA, Gorup JM. Radiofrequency medial branch neurotomy in litigant and nonlitigant patients with cervical whiplash: a prospective study. Spine 2001;26:E268–73. [19] Wallis BK, Lord SM, Bogduk N. Resolution of psychological distress of whiplash patients following treatment by radiofrequency neurotomy: a randomised, double-blind, placebo-controlled trial. Pain 1997;73:15–22. [20] Sanders M, Zuurmond WW. Percutaneous intra-articular lumbar facet joint denervation in the treatment of low back pain: a comparison with percutaneous extra-articular lumbar facet denervation. Pain Clinic 1999;11:329–35. [21] Aprill C, Bogduk N. The prevalence of cervical zygapophyseal joint pain. A first approximation. Spine 1992;17:744–7. [22] Lord SM, Barnsley L, Wallis BJ, Bogduk N. Chronic zygapophysial joint pain after whiplash: a placebo-controlled prevalence study. Spine 1996;21:1737–44. [23] Schwarzer AC, Aprill CN, Derby R, Fortin J, Kine G, Bogduk N. The false-positive rate of uncontrolled diagnostic blocks of the lumbar zygapophysial joints. Pain 1994;58:195–200. [24] Manchikanti L, Pampati V, Fellows B, Bakhit CE. The diagnostic validity and therapeutic value of lumbar facet joint nerve blocks with or without adjuvant agents. Curr Rev Pain 2000;4:337–44. [25] Schwarzer AC, Aprill CN, Derby R, Fortin J, Kine G, Bogduk N. The relative contributions of the disc and zygapophyseal joint in chronic low back pain. Spine 1994;19:801–6. [26] Manchikanti L, Pampati V, Fellows B, Bakhit CE. Prevalence of lumbar facet joint pain in chronic low back pain. Pain Physician 1999;2: 59–64. [27] Kaplan M, Dreyfuss P, Halbrook B, Bogduk N. The ability of lumbar medial branch blocks to anesthetize the zygapophysial joint. A physiologic challenge. Spine 1998;23:1847–52. [28] Bogduk N, Holmes S. Controlled zygapophysial joint blocks: the travesty of cost-effectiveness. Pain Med 2000;1:24–34. [29] Stojanovic MP, Zhou Y, Hord D, Vallejo R, Cohen SP. Single needle approach for multiple medial branch blocks: a new technique. Clin J Pain 2003;19:134–7. [30] Stojanovic MP, Dey D, Hord ES, Zhou Y, Cohen SP. A prospective crossover comparison study of the single-needle and multiple-needle techniques for facet-joint medial branch block. Reg Anesth Pain Med 2005;30:484–90. [31] North RB, Han M, Zahurak M, Kidd DH. Radiofrequency lumbar facet denervation: analysis of prognostic factors. Pain 1994;57: 77–83. [32] Carette S, Marcoux S, Truchon R, et al. A controlled trial of corticosteroid injections into facet joints for chronic low back pain. N Engl J Med 1991;325:1002–7. [33] Fujiwara A, Tamai K, Yamato M, et al. The relationship between facet joint osteoarthritis and disc degeneration of the lumbar spine: an MRI study. Eur Spine J 1999;8:396–401. [34] North RB, Kidd DH, Zahurak M, Piantadosi S. Specificity of diagnostic nerve blocks: a prospective, randomized study of sciatica due to lumbosacral spine disease. Pain 1996;65:77–85. [35] Kellegren JH. On the distribution of pain arising from deep somatic structures with charts of segmental pain areas. Clin Sci 1939;4: 35–46. [36] Dreyfuss P, Schwarzer AC, Lau P, Bogduk N. Specificity of lumbar medial branch and L5 dorsal ramus blocks. A computed tomography study. Spine 1997;22:895–902. [37] Ackerman WE, Munir MA, Zhang JM, Ghaleb A. Are diagnostic lumbar facet injections influenced by pain of muscular origin? Pain Pract 2004;4:286–91. [38] Manchikanti L, Damron KS, Rivera JJ, et al. Evaluation of the effect of sedation as a confounding factor in the diagnostic validity of lumbar facet joint pain: a prospective, randomized, double-blind, placebo-controlled evaluation. Pain Physician 2004;7:411–7. [39] Hansen BJ, Meyhoff HH, Nordling J, Mensink HJ, Mogensen P, Larsen EH. Placebo effects in the pharmacological treatment of uncomplicated benign prostatic hyperplasia. The ALFECH Study Group. Scand J Urol Nephrol 1996;30:373–7. [40] Marks RC, Houston T, Thulbourne T. Facet joint injection and facet nerve block: a randomized comparison in 86 patients with chronic low back pain. Pain 1992;49:325–8. [41] Nash TP. Facet jointsdintra-articular steroids or nerve block? Pain Clinic 1990;3:77–82.