Juch 2017

Research
JAMA | Original Investigation
Effect of Radiofrequency Denervation on Pain Intensity

Among Patients With Chronic Low Back Pain
The Mint Randomized Clinical Trials
Johan N. S. Juch, MD; Esther T. Maas, PhD; Raymond W. J. G. Ostelo, PT, PhD; J. George Groeneweg, PT, PhD;
Jan-Willem Kallewaard, MD; Bart W. Koes, PhD; Arianne P. Verhagen, PT, PhD; Johanna M. van Dongen, PhD;
Frank J. P. M. Huygen, MD, PhD; Maurits W. van Tulder, PhD
Supplemental content
IMPORTANCE Radiofrequency denervation is a commonly used treatment for chronic low CME Quiz at
back pain, but high-quality evidence for its effectiveness is lacking. jamanetwork.com/learning
and CME Questions page 88
OBJECTIVE To evaluate the effectiveness of radiofrequency denervation added
to a standardized exercise program for patients with chronic low back pain.
DESIGN, SETTING, AND PARTICIPANTS Three pragmatic multicenter, nonblinded randomized

clinical trials on the effectiveness of minimal interventional treatments for participants with
chronic low back pain (Mint study) were conducted in 16 multidisciplinary pain clinics in the
Netherlands. Eligible participants were included between January 1, 2013, and October 24,
2014, and had chronic low back pain, a positive diagnostic block at the facet joints
(facet joint trial, 251 participants), sacroiliac joints (sacroiliac joint trial, 228 participants),
or a combination of facet joints, sacroiliac joints, or intervertebral disks (combination trial,
202 participants) and were unresponsive to conservative care.
INTERVENTIONS All participants received a 3-month standardized exercise program and

psychological support if needed. Participants in the intervention group received
radiofrequency denervation as well. This is usually a 1-time procedure, but the maximum
number of treatments in the trial was 3.
MAIN OUTCOMES AND MEASURES The primary outcome was pain intensity (numeric rating
scale, 0-10; whereby 0 indicated no pain and 10 indicated worst pain imaginable) measured
3 months after the intervention. The prespecified minimal clinically important difference was
defined as 2 points or more. Final follow-up was at 12 months, ending October 2015.
RESULTS Among 681 participants who were randomized (mean age, 52.2 years; 421 women
[61.8%], mean baseline pain intensity, 7.1), 599 (88%) completed the 3-month follow-up, and
521 (77%) completed the 12-month follow-up. The mean difference in pain intensity between
the radiofrequency denervation and control groups at 3 months was −0.18 (95% CI, −0.76
to 0.40) in the facet joint trial; −0.71 (95% CI, −1.35 to −0.06) in the sacroiliac joint trial;
and −0.99 (95% CI, −1.73 to −0.25) in the combination trial.
CONCLUSIONS AND RELEVANCE In 3 randomized clinical trials of participants with chronic low
back pain originating in the facet joints, sacroiliac joints, or a combination of facet joints,
sacroiliac joints, or intervertebral disks, radiofrequency denervation combined with a
standardized exercise program resulted in either no improvement or no clinically important
improvement in chronic low back pain compared with a standardized exercise program alone.
The findings do not support the use of radiofrequency denervation to treat chronic low back
pain from these sources. Author Affiliations: Author
affiliations are listed at the end of this
article.
TRIAL REGISTRATION trialregister.nl Identifier: NTR3531
Corresponding Author: Esther T.
Maas, PhD, School of Population and
Public Health, University of British
Columbia, 2206 E Mall, Vancouver,
BC V6T 1Z3, Canada (esther.maas
JAMA. 2017;318(1):68-81. doi:10.1001/jama.2017.7918 @ubc.ca).
68 (Reprinted) jama.com
© 2017 American Medical Association. All rights reserved.
Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

Radiofrequency Denervation on Pain Intensity in Patients With Low Back Pain Original Investigation Research
L
ow back pain causes more disability than any other
condition and has major social and economic con- Key Points
sequences. 1-3 In the Netherlands (16.5 million resi-
Question What is the effectiveness of radiofrequency
dents) the cost of low back pain was estimated at €3.5 billion denervation added to a standardized exercise program for patients
(US $3.9 billion) in 2007, and the majority of the costs were with chronic low back pain?
attributable to patients with chronic low back pain. In the
Findings In 3 randomized clinical trials including 681 participants
United States (326 million residents), the costs of low back
with chronic low back pain originating from the facet joints,
pain have not been recently estimated; however, a study by sacroiliac joints, or a combination of these or the intervertebral
Dieleman et al4 evaluated health care spending from 1996 to disks, radiofrequency denervation combined with exercise
2013 in the United States and estimated the health care spend- compared with exercise alone resulted in either no significant
ing on low back and neck pain at $87.6 billion. difference in pain intensity, or a difference smaller than the
Potential sources of low back pain of the spinal column prespecified minimal clinically important difference after 3 months.
include the facet joints, sacroiliac joints, and intervertebral Meaning The study findings do not support the use of
disks. These sources of pain were classified as mechanical radiofrequency denervation for chronic low back pain originating
low back pain.5,6 Radiofrequency denervation is a commonly from these sources.
used treatment in pain clinics for chronic low back pain.
In the United States, facet joint or sacroiliac joint interven-
tions in Medicare recipients increased from approximately obtained from all participating pain clinics. All participants
425 000 interventions in 2000 to 2.2 million interventions in gave written informed consent.
2013.7 Radiofrequency denervation aims to prevent the con- In 16 multidisciplinary pain clinics in the Netherlands,
duction of nociceptive impulses through the use of an elec- pain specialists consecutively screened participants with
tric current that damages the pain-conducting nerve. The chronic low back pain. Inclusion criteria were pain consid-
effectiveness of radiofrequency denervation has not been ered to be related to the facet joint, sacroiliac joint, or a com-
consistently demonstrated. However, there is consensus bination of the facet joint, sacroiliac joint, or intervertebral
among anesthesiologists that minimal interventional proce- disk; aged 18 to 70 years; and no improvement in symptoms
dures such as radiofrequency denervation are effective for after conservative treatment. Medical history and clinical
patients with mechanical low back pain.5 Systematic reviews examination followed a standard format and were performed
and multidisciplinary clinical guidelines concluded that by experienced clinicians to determine the likely source of
there is evidence of very low to moderate quality supporting the pain. To be considered for a diagnostic sacroiliac joint
the effectiveness of radiofrequency denervation in clinical block, at least 3 of 6 provocation tests (compression test; dis-
practice for patients with chronic low back pain.5,8-10 traction test; Flexion, Abduction, and External Rotation
The aim of this study was to evaluate whether radiofre- [FABER] test; Gaenslen test; thigh thrust test; Gillett test) had
quency denervation in addition to a standardized exercise to have positive results.12,13 Participants with suspected iso-
program is more effective than the standardized exercise pro- lated facet joint pain or isolated sacroiliac joint pain received
gram alone for patients with chronic mechanical low back pain. a diagnostic anesthetic block prior to randomization and
were only randomized if the diagnostic block was positive.
Participants with a suspected combination of sources of pain
were randomized based on participant history and physical
Methods examination prior to receiving the diagnostic blocks.
Study Design and Participants This choice was made for ethical reasons. It would be unethi-
The Cost-Effectiveness of Minimal Interventional Proce- cal to give participants in the study multiple diagnostic
dures for Patients with Chronic Low Back Pain (Mint) blocks (ie, a facet joint diagnostic block, a sacroiliac joint
study 11 was an initiative to evaluate minimally invasive diagnostic block, and a provocative discography) before
treatments for patients with spinal column–related chronic treatment. Furthermore, it is common practice in Dutch pain
low back pain, consisting of 4 trials and an observational clinics for participants with chronic low back pain due to
study (participants who did not want to be randomized or facet joints, sacroiliac joints, or intervertebral disks (based on
who did not meet the inclusion criteria for the trials were history taking and physical examination) to start with 1 diag-
asked to participate in the observational study, where they nostic block. If the diagnostic block was positive, the inter-
received usual care). The full protocol is available in vention was provided. If the diagnostic block was negative,
Supplement 1. One trial was designed to evaluate radiofre- then another block was provided. If the second diagnostic
quency denervation for pain from the intervertebral disks. block was positive, the intervention was provided. If the sec-
This trial was prematurely terminated because of a lack of ond diagnostic block was negative, the clinician provided a
eligible participants. The other 3 trials are presented in this third block. All participants were considered candidates for
article: (1) the facet joint trial, (2) the sacroiliac joint trial, intervention based on history taking and physical examina-
and (3) the combination trial (facet joint, sacroiliac joint, or tion. For this reason, participants were randomized and
the intervertebral disk). The Medical Ethics Committee of included in the combination trial after history taking and
the Erasmus University Medical Centre in Rotterdam physical examination, if the pain physician suspected that
granted ethical approval. Local research governance was the pain originated from more than 1 source.
jama.com (Reprinted) JAMA July 4, 2017 Volume 318, Number 1 69

Research Original Investigation Radiofrequency Denervation on Pain Intensity in Patients With Low Back Pain
Exclusion criteria for all trials were pregnancy, severe psy- Radiofrequency Denervation
chological problems (determined with psychological ques- Within 1 week after the first exercise session the intervention
tionnaires), involvement in work-related conflicts or claims; group received radiofrequency denervation. The technical de-
body mass index (BMI; calculated as weight in kilograms di- tails of the radiofrequency denervation procedures are in-
vided by height in meters squared) higher than 35; or antico- cluded in the eAppendix in Supplement 2.19-22
agulant drug therapy or coagulopathy.
Co-Interventions
Diagnostic Blocks In both treatment groups, participants were asked to refrain
For the facet joints,14 a 22-gauge needle was inserted to the pos- from co-interventions during the intervention period of 3
terior primary root of the spinal nerve (medial branch) under months (duration of the standardized exercise program).
C-arm fluoroscopy. L3-4, L4-5, and L5-6 were selected for di- Co-interventions that were not allowed included (but were
agnostic blocks. The lateral image was checked to confirm the not limited to) surgery; manual therapy; chiropractic therapy;
correct position of the needle, after which 0.5 mL of 2% lido- a change in current, back pain–related medication; or newly
caine was injected. prescribed medication. Analgesics were not prescribed, but
For the sacroiliac joints,14 a 25-gauge needle was inserted over-the-counter medication was allowed. Co-interventions
3 mm to 10 mm laterally of the sacral foramina S1-3 under or recurrence of the radiofrequency denervation was allowed
fluoroscopy. The correct depth of the needle was confirmed after the intervention period of 3 months. These interven-
laterally, after which 0.5 mL of 2% lidocaine was injected. tions were recorded. Psychological care was not considered a
The dorsal ramus of L5 was also blocked as described in co-intervention and was provided when needed to partici-
the Spinal Intervention Society guidelines using 0.5 mL of pants in either treatment group.
2% lidocaine.
The blocks were considered positive if the participant re- Outcomes
ported 50% or more pain reduction within 30 to 90 minutes The primary outcome was pain intensity, measured on an
after the block. 11-point numerical rating scale (NRS; a score of 0 indicates no
The current standard for diagnosing discogenic pain is pres- pain; 10 indicates worst pain imaginable) 3 months after the
sure-controlled provocative discography using strict criteria intervention.23
and at least 1 negative control level.15 Secondary outcomes were global perceived recovery,16
participant satisfaction17 (both measured by the 7-point,
Randomization and Masking categorical Global Perceived Effect scale; a score of 1 indi-
Participants were randomized using a computerized random cates fully recovered; 4 indicates no change; 7 indicates
number generator (Alea II, Netherlands Cancer Institute- worse than ever), functional status (measured by Oswestry
Antoni van Leeuwenhoek Hospital), accessed through a Disability Index [ODI]; a score of 0 indicates no restrictions in
password-protected website and maintained independently. daily activities; 100 indicates most restrictions in daily
Randomization was performed at the individual level by activities),24 health-related quality of life (measured by the
means of block randomization (block size = 4), prestratified 3-level EuroQol 5D Health Questionnaire [EQ-5D-3L]; a score
for pain clinic. Participants were allocated (1:1) to receive of 0 indicates worst imaginable health state; 1 indicates best
either radiofrequency denervation with a standardized exer- imaginable health state), 25 general health (measured by
cise program (intervention group) or a standardized exercise RAND 36-Item Health Survey [Rand-36], a score of 0 indi-
program alone (control group). cates lowest general health score; 100 indicates highest gen-
Participants and caregivers were not blinded. The Dutch eral health score),26 and chronic pain experiences (measured
Ministry of Health, Welfare, and Sport requested a pragmatic by the West Haven-Yale Multidimensional Pain Inventory; a
trial in which existing, commonly applied treatment options score of 0 indicates lowest score; 6 indicates highest score).27
would be compared. Data handling, analysis, and interpreta- The minimal clinically important difference in pain for par-
tion of results were conducted blind to treatment allocation. ticipants with chronic low back pain was estimated at 2 points
All participants were sequentially assigned unique numbers. or more of the 10-point NRS, a difference of 20 points on the
Participants’ expectations and satisfaction16,17 were mea- 100-point ODI, and between 0.09 and 0.28 points on the
sured to evaluate a possible risk of bias due to a nonblinded EQ-5D-3L utility score between 0 and 1.28,29 No minimal clini-
study design. cally important differences are known for the other second-
ary outcomes.
Interventions All outcome measures were registered using web-based
Standardized Exercise Program questionnaires, which were sent at baseline and 3-, 6-, 9-, and
All participants received a program based on the Dutch physi- 12-month follow-up. Pain intensity, global perceived recov-
cal therapy guidelines18 in 1 of 102 participating physical ery, and health-related quality of life were also assessed at
therapy practices. The 8- to 12-hour programs focused on qual- 3-week follow-up and 6-week follow-up.
ity of movement and behavior, and took place during a 3-month
intervention period. More details are available in the study pro- Sample Size Calculation
tocol, which is available in Supplement 1. If necessary, partici- A clinically relevant mean difference of 2 points or more on
pants were referred to psychological care. the NRS28 for pain intensity (SD, 4) was used for the sample
70 JAMA July 4, 2017 Volume 318, Number 1 (Reprinted) jama.com

Figure 1. Flow of Patients Through Enrollment in the 3 Randomized Clinical Trials
10 592 Potential participants
5168 Included in observational studya
5424 Patients asked to participate in 1 of 3

trials based on suspected source of pain
2133 Asked to participate in the facet joint trial 2498 Asked to participate in the sacroiliac trial 793 Asked to participate in the combination trial
1882 Excludedb 2270 Excludedb 591 Excludedb

1202 Declined participation 1666 Declined participation 298 Psychological problemsc
258 Negative diagnostic facet joint block 202 Negative diagnostic sacroiliac 52 BMI >35
277 Psychological problemsc joint block 102 Aged >70 y
52 BMI >35 257 Psychological problemsc 139 Other
93 Aged >70 y 47 BMI >35
83 Aged >70 y
15 Other
See Figure 2 See Figure 2 See Figure 3

A B C
251 Randomized 228 Randomized 202 Randomized
b
BMI indicates body mass index (calculated as weight in kilograms divided by Participants not eligible for participation due to 1 positive exclusion criterion or
height in meters squared). more could be included in the observational study.
a c
Observational study was performed alongside randomized clinical trials; Participants were excluded based on psychological problems, assessed by
results from the observational study are not reported in this article. validated questionnaires.
size calculation. With a power of 0.9, a 2-sided α of .05, and groups. Data were compared between complete and incom-
a correlation of 0.5 for repeated measurements, 85 partici- plete cases to identify possible selective dropout.
pants per group were needed. Anticipating potential study Treatment success for the global perceived recovery was
withdrawal (20%), a minimum of 204 participants per trial defined as “much recovery” or “complete recovery.” In post
was needed. hoc analyses, treatment success in pain reduction was de-
fined as either more than 30% or 2 points reduction or more
Statistical Analyses on the NRS pain scale.
Effects were estimated using a maximum likelihood estima- No adjustments for multiple comparisons were made.
tion for longitudinal mixed-effects model, under “missing Findings for the secondary outcomes should be interpreted
at random” assumptions, including a term for pain clinic, if as exploratory.
necessary, based on the likelihood ratio test.30 We used a In 2 sensitivity analyses, participants marked as partici-
generalized linear mixed model (logit link) for the post hoc pants who had protocol violations, and participants who re-
analysis of treatment response for dichotomized outcomes. ceived radiofrequency denervation during follow-up were ex-
The same multilevel structure was used for both models. All cluded from the analyses. Additionally, data were compared
analyses were conducted in accordance with the intention- between complete and incomplete cases. We used MLwiN
to-treat principle. software (University of Bristol), version 2.22, for the effects
Regression coefficients or odds ratios (ORs) with 95% CIs models (2-sided significance P < .05).
were calculated; ORs were converted to relative risks (RRs)
using the method of Zhang et al31: RR = OR/[(1 − prevalence
in control group) + (prevalence in control group × OR)].
We adjusted for the outcome parameter at baseline, and age,
Results
sex (self-reported), BMI, education, smoking, marital status, In total, 251 patients were included in the facet joint trial, 228
back pain complaint history, and participant expectations. patients in the sacroiliac joint trial, and 202 in the combina-
The effect of interest was the time × treatment interaction. tion trial (Figure 1, Figure 2, and Figure 3). The 681 random-
Regression coefficients can be interpreted as mean differ- ized participants had a mean age of 52.2 years, 421 partici-
ences between interventions compared with baseline. Addi- pants were women (61.8%), and the mean baseline pain
tionally, we calculated the number needed to treat and the intensity was 7.1 on the NRS scale. Another 5168 patients were
unadjusted risk differences as absolute differences between included in the observational part of Mint study.

Figure 2. Flow of Patients Through the Facet Joint and Sacroiliac Joint Trials
Continued From Figure 1

Facet joint trial A Sacroiliac joint trial B
251 Randomized 228 Randomized
125 Randomized to intervention group 126 Randomized to control group 116 Randomized to intervention group 112 Randomized to control group
121 Received radiofrequency 92 Received exercise program 110 Received radiofrequency 69 Received exercise program
denervation as randomized as randomized denervation as randomized as randomized
3 Received sacroiliac joint 22 Did not complete exercise 81 Received Palisade 18 Did not complete exercise
Palisade radiofrequency program radiofrequency treatment program
treatmenta 12 Unknown completion 23 Received cooled 25 Unknown completion
1 Did not receive treatment radiofrequency denervation
6 Received SIMPLICITY III
101 Completed exercise program denervation
18 Did not complete exercise 3 Received facet joint
program radiofrequencya
6 Unknown completion 3 Did not receive treatment
89 Completed exercise program
21 Did not complete exercise
program
6 Unknown completion
117 Completed baseline visit 116 Completed baseline visit 109 Completed baseline visit 104 Completed baseline visit
8 Did not complete baseline visit 10 Did not complete baseline visit 7 Did not complete baseline visit 8 Did not complete baseline visit
3 Wrong contact information 4 Wrong contact information 2 Wrong contact information 3 Wrong contact information
4 No response 5 No response 5 No response 4 No response
1 Died 1 Unsure about participating 1 Technical difficulties
108 Completed 3-wk follow-up 101 Completed 3-wk follow-up 94 Completed 3-wk follow-up 86 Completed 3-wk follow-up
17 Did not complete 3-wk follow-up 25 Did not complete 3-wk follow-up 22 Did not complete 3-wk follow-up 26 Did not complete 3-wk follow-up
1 Died 1 Comorbidity 14 No treatment details 3 Technical difficulties
8 No treatment details 10 No treatment details 10 No treatment details
2 Withdrewb
119 Completed 6-wk follow-up 118 Completed 6-wk follow-up 107 Completed 6-wk follow-up 95 Completed 6-wk follow-up
6 Did not complete 6-wk follow-up 8 Did not complete 6-wk follow-up 9 Did not complete 6-wk follow-up 17 Did not complete 6-wk follow-up
1 Died 1 Comorbidity 1 No treatment details 3 Technical difficulties
1 No treatment details 1 No treatment details 3 No treatment details
3 Withdrewb
119 Completed 3-mo follow-up 114 Completed 3-mo follow-up 110 Completed 3-mo follow-up 88 Completed 3-mo follow-up
6 Did not complete 3-mo follow-up 12 Did not complete 3-mo follow-up 6 Did not complete 3-mo follow-up 24 Did not complete 3-mo follow-up
1 No internet 1 Wrong contact information 1 Wrong contact information 6 Wrong contact information
1 Died 1 No treatment details 3 Technical difficulties
1 Comorbidity 6 Withdrewb
2 Unmotivated 1 Wrong contact information 2 Wrong contact information 3 Wrong contact information
1 Died 1 Unsatisfied 3 Technical difficulties
1 No treatment details 7 Withdrewb
1 Comorbidity
1 Unmotivated 1 Wrong contact information 1 Wrong contact information 3 Wrong contact information
1 No internet 18 No response 12 No response 19 No response
3 No time 1 No treatment details 2 Withdrewb 2 Technical difficulties
13 No response 1 Comorbidity 10 Withdrewb
1 Died
3 No time 1 Wrong contact information 1 Wrong contact information 3 Wrong contact information
18 No response 1 No treatment details 12 No response 19 No response
1 Died 19 No response 2 Withdrewb 2 Technical difficulties
2 Unmotivated 11 Withdrewb
1 Comorbidity
125 Included in intention-to- 126 Included in intention-to- 116 Included in intention-to- 112 Included in intention-to-
treat analysis treat analysis treat analysis treat analysis
a
Participants received RF treatment other than their randomized assignment.
b
Study withdrawals were not cumulative.

Figure 3. Flow of Patients Through the Combination Trial
Continued From Figure 1

Combination trial C
202 Randomized
103 Randomized to intervention group 99 Randomized to control group

67 Received radiofrequency denervation 71 Received exercise program as randomized
as randomized 28 Did not complete exercise program
26 Positive facet joint block
21 Positive sacroiliac joint block
21 Positive facet and sacroiliac joint block
67 Did not receive radiofrequency denervation
1 Positive diagnostic facet joint block
35 Negative diagnostic joint block
93 Completed exercise program
10 Did not complete exercise program
98 Completed baseline visit 89 Completed baseline visit

5 Did not complete baseline visit 10 Did not complete baseline visit
1 Wrong contact information 3 Wrong contact information
2 No response 7 No response
1 No complaints
1 Comorbidity
77 Completed 3-wk follow-up 56 Completed 3-wk follow-up

26 Did not complete 3-wk follow-up 43 Did not complete 3-wk follow-up
18 No treatment details 29 No treatment details
1 No complaints
1 Other treatment
1 Comorbidity
90 Completed 6-wk follow-up 82 Completed 6-wk follow-up

13 Did not complete 6-wk follow-up 17 Did not complete 6-wk follow-up
1 No complaints
1 Other treatment
1 Comorbidity
88 Completed 3-mo follow-up 80 Completed 3-mo follow-up

16 Did not complete 3-mo follow-up 19 Did not complete 3-mo follow-up
1 No complaints
1 Other treatment
1 Comorbidity

1 No complaints
1 Other treatment
1 Comorbidity

13 No response 3 No treatment details
4 No treatment details 24 No response
1 No complaints
1 Other treatment
1 Comorbidity

16 No response 3 No treatment details
4 No treatment details 29 No response
1 No complaints
1 Other treatment
1 Comorbidity
103 Included in intention-to-treat analysis 99 Included in intention-to-treat analysis

Table 1. Baseline Characteristics of Participants With Chronic Low Back Pain
Facet Joint Trial Sacroiliac Joint Trial Combination Trial

Intervention Control Intervention Control Intervention Control
Characteristicsa (n = 125) (n = 126) (n = 116) (n = 112) (n = 103) (n = 99)
Age, mean (SD), y 52.98 (11.48) 52.60 (10.79) 51.58 (10.94) 51.13 (12.22) 50.80 (11.33) 53.31 (10.35)
Women, No. (%) 65 (55.56) 60 (51.72) 87 (74.35) 79 (75.96) 64 (65.31) 66 (74.15)
BMI, mean (SD) 26.77 (5.17) 27.62 (4.27) 26.73 (4.17) 26.76 (4.53) 26.84 (3.82) 26.43 (4.25)
Smoker, No. (%) 34 (29.05) 34 (29.05) 29 (26.61) 31 (29.81) 23 (23.46) 26 (29.21)
Education level, No. (%)b
Low 57 (48.72) 64 (55.17) 59 (54.13) 53 (50.96) 52 (53.06) 43 (48.31)
Moderate 35 (29.99) 34 (29.31) 32 (29.36) 32 (30.76) 33 (33.67) 32 (35.96)
High 21 (17.95) 16 (13.79) 18 (16.51) 18 (17.31) 12 (12.24) 14 (15.73)
Married or living with a 93 (79.49) 98 (84.48) 85 (79.61) 82 (79.61) 66 (67.35) 68 (76.40)
partner, No. (%)
Having a paid job, No. (%) 64 (54.70) 66 (56.80) 66 (60.55) 50 (48.07) 48 (48.97) 44 (48.44)
History of back pain,
median (IQR), mo
Time since first 146.00 100.33 97.33 65.08 120.58 97.33
experience with (49.75-267.67) (36.5-186.30) (37.51-228.12) (27.08-144.21) (37.32-222.04) (32.33-192.58)
low back pain
Time since first 31.33 26.73 30.33 24.33 36.50 32.33
current episode with (12.17-103.42) (10.54-73.00) (12.17-76.03) (12.17-66.58) (12.17-121.67) (8.00-97.19)
low back pain
Origin of back pain, No.
Facet and sacroiliac joint 69 70
Facet and disc 18 18
Sacroiliac joint and disc 6 1
Facet and sacroiliac 3 6
joint and disc
Unknown 7 4
CEQ score, mean (SD)c
Credibility 21.36 (3.92) 19.47 (5.49) 21.36 (4.51) 19.88 (5.31) 20.10 (4.70) 17.07 (5.99)
Expectancy 18.97 (4.59) 17.36 (5.20) 18.75 (4.99) 18.23 (5.31) 16.88 (5.78) 14.38 (6.24)
Pain intensity score in the 7.14 (1.38) 7.19 (1.29) 7.17 (1.65) 7.06 (1.43) 7.19 (1.43) 7.43 (1.41)
past week, mean (SD)d
Functioning score, 35.07 (14.66) 34.39 (12.24) 38.07 (14.07) 33.70 (14.43) 39.06 (14.03) 37.20 (13.74)
mean (SD)e
Quality-of-life score, 0.52 (0.26) 0.54 (0.26) 0.50 (0.27) 0.56 (0.27) 0.49 (0.28) 0.52 (0.28)
mean (SD)f
c
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided A higher score indicates more credibility in the effectiveness of treatment or
by height in meters squared); CEQ, credibility expectancy questionnaire. higher expectations about the treatment (score range, 0-27).
a d
Results are presented of the 233 participants in the facet joint trial, Measured by numeric rating scale (score range, 0-10); a higher score indicates
207 participants in the sacroiliac joint trial, and 187 participants in the more severe pain intensity.
combination trial who had complete baseline data. e
Measured by Oswestry Disability Index (score range, 0-100); a higher score
b
Education levels: low indicates preschool, primary school, or lower secondary indicates worse functioning.
school; moderate indicates higher secondary school or undergraduate; high f
Measured by EuroQol-5D (score range, 0-1); a higher score indicates better
indicates tertiary, university, or postgraduate. quality of life.
Facet Joint Trial Baseline characteristics were comparable across groups

Study Participants (Table 1). However, participants in the intervention group had
Between January 1, 2013, and June 3, 2014 (the inclusion pe- a first low back pain episode 12 years prior compared with 8
riod for the facet joint trial), 931 participants received a diag- years prior in the control group.
nostic facet joint block. Patients with a negative result for the Complete data on pain intensity, functional status, and
diagnostic facet joint block (n = 258) were excluded. Patients global perceived recovery after 3 months was obtained from
with psychological problems (n = 277), older than 70 years 233 participants (93%). Complete outcome data on all fol-
(n = 93), or with a BMI higher than 35 (n = 52) were followed low-up points during the year were obtained from 179 partici-
up in the observational study. The inclusion criteria were met pants (71%). Participants with complete data were older, more
by 251 participants for the facet joint trial and were random- often nonsmokers, were more likely to have a partner, had a
ized to the intervention group (n = 125) and control group higher BMI, and had low back pain complaints for a longer pe-
(n = 126) (Figure 1 and Figure 2). riod (eTable 1 in Supplement 2).

Table 2. Pain Intensity Score (Primary Outcome)a,b Among Participants With Chronic Low Back Pain
Intervention Group, Control Group, Between-Group Difference,
Overall Effect Mean (95% CI) Mean (95% CI) Mean (95% CI)c P Value
Facet joint trial, 125 126
No. of participants
Overall −0.08 (−0.50 to 0.34) .71
3 wk 5.17 (4.73 to 5.61) 5.92 (5.58 to 6.26) −0.41 (−1.02 to 0.19) .18
6 wk 5.19 (4.76 to 5.61) 5.90 (5.53 to 6.26) −0.38 (−0.96 to 0.20) .20
3 mo 5.01 (4.59 to 5.43) 5.44 (5.03 to 5.85) −0.18 (−0.76 to 0.40) .55
6 mo 4.61 (4.18 to 5.04) 4.84 (4.38 to 5.30) −0.04 (−0.63 to 0.56) .91
9 mo 4.66 (4.20 to 5.00) 4.73 (4.24 to 5.22) 0.19 (−0.41 to 0.80) .53
12 mo 4.49 (4.00 to 4.97) 4.44 (3.94 to 4.94) 0.47 (−0.14 to 1.07) .13
Sacroiliac joint trial, 116 112 Abbreviation: NNT, number needed
No. of participants to treat.
Overall −0.40 (−0.83 to 0.03) .07 a
Measured by numeric rating scale
3 wk 4.96 (4.51 to 5.40) 6.00 (5.59 to 6.41) −0.96 (−1.63 to −0.29) .005 (score range, 0-10); a higher score
6 wk 5.22 (4.81 to 5.64) 5.69 (5.31 to 6.08) −0.53 (−1.17 to 0.10) .10 indicates more severe symptoms.
b
3 mo 4.77 (4.31 to 5.24) 5.45 (4.94 to 5.95) −0.71 (−1.35 to −0.06) .03 The overall effect measures provide
information over the total follow-up
6 mo 4.50 (4.01 to 4.98) 4.78 (4.24 to 5.31) −0.12 (−0.77 to 0.53) .73 time of 12 mo, instead of the
9 mo 5.03 (4.55 to 5.51) 4.97 (4.39 to 5.56) 0.16 (−0.51 to 0.83) .64 time × treatment effects.
c
12 mo 4.65 (4.16 to 5.13) 4.84 (4.30 to 5.38) −0.07 (−0.74 to 0.60) .83 Values presented (for mean
differences) are model estimates of
Combination trial, 103 99
No. of participants linear mixed-effects models with a
random intercept, and adjusted for
Overall −0.21 (−0.76 to 0.35) .47
outcome at baseline and age, sex,
3 wk 5.45 (4.95 to 5.95) 6.40 (5.91 to 6.89) −0.65 (−1.47 to 0.17) .12 body mass index, education,
6 wk 5.37 (4.89 to 5.85) 6.09 (5.65 to 6.52) −0.40 (−1.14 to 0.34) .29 smoking, marital status, back pain
complaint history, and participant
3 mo 4.77 (4.25 to 5.30) 5.94 (5.42 to 6.45) −0.99 (−1.73 to −0.25) .01
expectations. Regression
6 mo 4.92 (4.39 to 5.44) 4.95 (4.35 to 5.54) 0.33 (−0.53 to 1.09) .39 coefficients can be interpreted as
9 mo 5.01 (4.47 to 5.56) 5.25 (4.65 to 5.86) −0.05 (−0.82 to 0.73) .90 mean differences between
interventions at a certain follow-up
12 mo 4.85 (4.24 to 5.46) 4.38 (3.73 to 5.03) 0.69 (−0.10 to 1.49) .09
point compared with baseline.
Twelve participants in the control group received radio- remained similar (eTable 3 in Supplement 2). After 3 months
frequency denervation within the first 3 months and were of follow–up, 31 control group participants received radiofre-
marked as participants who had protocol violations. Ten par- quency denervation. The analyses were repeated excluding
ticipants (8%) in the intervention group and 11 participants (9%) participants receiving the intervention after the 3-month
in the control group received psychological care during the intervention period; this did not alter the results either
3-month intervention period. (eTable 4 in Supplement 2). The complete case analysis
No treatment-related adverse events were reported dur- showed no significant between-group differences for pain
ing the 1-year follow-up. intensity, functional status, and global perceived recovery at
3 months (eTable 5 in Supplement 2).
Intention-to-Treat Analyses
The mean difference for the primary outcome pain intensity Sacroiliac Joint Trial
at 3 months was −0.18 (95% CI, −0.76 to 0.40). Results on all Study Participants
other follow-up points are shown in Table 2 and Table 3. The Between January 1, 2013, and July 1, 2014 (the inclusion
mean difference for functional status at 3 months was −2.45 period for the sacroiliac joint trial), 832 participants received
(95% CI, −5.53 to 1.03); the RR for global perceived recovery a diagnostic sacroiliac joint block. Patients with a negative re-
at 3 months was 1.35 (95% CI, 0.81 to 2.05). Other follow-up sult for the diagnostic sacroiliac joint block (n = 202) were
points and secondary outcomes are shown in eTable 2 in excluded. Patients with psychological problems (n = 257),
Supplement 2. older than 70 years (n = 83), or a BMI higher than 35 (n = 47),
or other reasons for not participating in the trial (n = 15) were
Post Hoc Analyses of Treatment Response followed up in the observational study. The inclusion criteria
No significant differences between the groups were found were met by 228 participants for the sacroiliac joint trial and
when success was defined as more than 30% or 2 points re- were randomized to the intervention group (n = 116) and the
duction or more in pain at 3 months (Table 4). control group (n = 112) (Figure 1 and Figure 2).
Baseline characteristics were comparable across groups
Sensitivity Analyses (Table 1). However, the first episode of low back pain in the in-
When participants with protocol violations were excluded tervention group was 97 months before inclusion compared
from the analysis, the interpretation of the outcomes with 65 months in the control group.

Table 3. Secondary Outcomes Among Participants With Chronic Low Back Paina
Between-Group
Intervention Group, Control Group, Difference, Risk Difference
Overall Effect Mean (95% CI) Mean (95% CI) Mean (95% CI)b P Value (95% CI) NNT
Functioning Scorec,d
No. of participants
Overall 0.04 (−3.02 to 3.10) .98
3 mo 26.03 (23.01 to 29.06) 28.67 (26.06 to 31.84) −2.45 (−5.93 to 1.03) .17
6 mo 25.38 (22.45 to 28.30) 27.15 (24.07 to 30.23) −0.60 (−4.13 to 2.92) .74
9 mo 25.74 (22.74 to 28.73) 24.52 (21.49 to 27.54) 2.26 (−1.29 to 5.82) .21
12 mo 24.59 (21.39 to 27.79) 25.04 (21.77 to 28.31) 1.48 (−2.09 to 5.06) .42
Sacroiliac joint trial, 116 112
No. of participants
Overall 0.42 (−2.99 to 3.82) .81
3 mo 27.72 (24.50 to 30.95) 29.09 (25.47 to 2.71) −4.20 (−8.39 to −0.00) .05
6 mo 25.99 (22.91 to 29.05) 24.99 (21.45 to 28.52) 0.07 (−4.16 to 4.30) .97
9 mo 28.40 (25.05 to 31.75) 23.45 (20.00 to 6.91) 4.45 (0.14 to 8.77) .04
12 mo 27.29 (23.89 to 30.69) 24.49 (20.74 to 28.23) 2.11 (−2.25 to 6.47) .34
No. of participants
Overall 1.90 (−2.96 to 6.76) .44
3 mo 28.00 (24.65 to 31.35) 33.63 (29.88 to 37.37) −4.66 (−10.21 to 0.89) .10
6 mo 30.24 (26.14 to 34.34) 28.61 (24.80 to 32.43) 4.44 (−1.18 to 0.06) .12
9 mo 30.73 (26.83 to 34.63) 28.70 (24.48 to 32.91) 3.55 (−2.17 to 9.26) .22
12 mo 31.20 (27.20 to 35.20) 24.67 (20.88 to 28.45) 6.44 (0.61 to 12.26) .03
Global Perceived Recoverye
No. With Treatment No. With Treatment Relative Risk P Value Risk Difference NNT
Success/Total No. (%) Success/Total No. (%) (95% CI)f (95% CI)
No. of participants
3 wk 32/108 (29.63) 5/101 (4.95) 5.41 (2.29 to 10.34) <.001 24.68 (15.08 to 34.27) 4
6 wk 35/119 (29.41) 11/118 (9.32) 2.71 (1.37 to 4.68) .005 20.09 (10.37 to 29.81) 5
3 mo 43/119 (36.13) 27/114 (23.68) 1.35 (0.81 to 2.05) .24 12.45 (0.81 to 24.09) 8
6 mo 46/113 (40.70) 39/108 (36.11) 1.04 (0.64 to 1.12) .85 4.59 (−8.21 to 17.41) NA
9 mo 41/106 (38.67) 42/105 (40.00) 0.81 (0.48 to 0.57) .35 −1.33 (−14.50 to 11.86) NA
12 mo 44/103 (42.71) 40/102 (39.22) 0.90 (0.55 to 1.33) .65 3.49 (−9.95 to 16.96) NA
Sacroiliac joint trial, 116 112
No. of participants
3 wk 28/94 (29.78) 9/88 (10.23) 2.83 (1.39 to 4.89) .01 19.55 (8.35 to 30.77) 5
6 wk 43/110 (39.09) 10/95 (10.53) 3.71 (2.00 to 5.74) <.001 28.56 (17.55 to 39.58) 4
3 mo 43/110 (39.10) 19/88 (21.59) 1.87 (1.13 to 2.71) .02 17.51 (4.97 to 30.03) 6
6 mo 46/103 (44.66) 29/88 (32.95) 1.26 (0.83 to 1.84) .21 11.71 (−2.03 to 25.44) NA
9 mo 36/101 (35.64) 25/78 (32.05) 1.13 (0.67 to 1.70) .62 3.59 (−10.35 to 17.54) NA
12 mo 49/102 (48.03) 24/76 (31.78) 1.46 (0.92 to 2.02) .10 16.25 (2.20 to 30.72) NA
No. of participants
3 wk 17/77 (22.07) 4/56 (7.14) 2.23 (0.73 to 5.52) .15 14.93 (3.48 to 25.40) 6
6 wk 25/90 (27.77) 7/82 (8.54) 2.41 (0.99 to 4.90) .05 19.23 (8.19 to 30.30) 5
3 mo 30/88 (34.09) 13/80 (16.25) 1.99 (0.99 to 3.37) .06 17.84 (5.06 to 30.63) 5
6 mo 30/85 (35.29) 28/75 (37.33) 0.76 (0.39 to 1.30) .36 −2.04 (−16.97 to 12.90) NA
9 mo 29/82 (35.36) 21/68 (30.88) 1.11 (0.57 to 1.82) .73 4.48 (−10.61 to 19.57) NA
12 mo 26/75 (34.66) 22/61 (36.06) 0.91 (0.46 to 1.52) .76 −1.40 (−17.65 to 14.76) NA
c
Abbreviations: NA, not applicable; NNT, number needed to treat. Measured by Oswestry Disability Index (score range, 0-100); a higher score
a
The other secondary outcomes are presented in eTable 2 in Supplement 2. indicates worse functioning.
d
b
Values presented (for mean differences) are model estimates of linear The overall effect measures provide information over the total follow-up time
mixed-effects models with a random intercept, and adjusted for outcome of 12 mo, instead of the time × treatment effects.
e
at baseline and age, sex, body mass index, education, smoking, marital status, Measured by the Global Perceived Effect scale (range, 1-7); a score of 1 to 2
back pain complaint history, and participant expectations. Regression indicates success.
coefficients can be interpreted as mean differences between interventions f
Relative risk was estimated based on the method of Zhang et al.31
at a certain follow-up point compared with baseline.

Table 4. Successful Treatment Effects for Pain Intensity by Study Among Participants With Chronic Low Back Pain
Intervention Group, Control Group,
No.With Treatment No. With Treatment Relative Risk Risk Difference
Success/Total No. (%) Success/Total No. (%) (95% CI)a P Value (95% CI) NNT
Facet Joint Trial
Pain intensity
reduction >30%
3 wk 40/102 (39.22) 27/100 (27.00) 1.33 (0.80 to 1.97) .25 12.22 (−0.65 to 25.08) NA
6 wk 45/112 (40.17) 36/114 (31.57) 1.13 (0.70 to 1.63) .59 8.60 (−3.86 to 21.06) NA
3 mo 52/114 (45.61) 40/111 (36.03) 1.16 (0.76 to 1.60) .46 9.58 (−3.20 to 22.36) NA
6 mo 60/108 (55.56) 53/105 (50.47) 1.02 (0.71 to 1.33) .88 5.09 (−8.31 to 18.47) NA
9 mo 52/102 (50.98) 50/102 (49.02) 1.09 (0.75 to 1.42) .60 1.88 (−11.76 to 15.68) NA
12 mo 47/100 (47.00) 53/99 (53.53) 0.78 (0.50 to 1.09) .16 −6.53 (−20.40 to 7.33) NA
Pain intensity
reduction ≥2 points
3 wk 56/102 (54.90) 44/100 (44.00) 1.17 (0.81 to 1.53) .36 10.90 (−2.81 to 24.61) NA
6 wk 57/112 (50.89) 47/114 (41.23) 1.09 (0.74 to 1.46) .65 9.66 (−3.27 to 22.60) NA
3 mo 64/111 (57.65) 52/111 (46.85) 1.07 (0.75 to 1.39) .68 10.80 (−2.25 to 23.87) NA
6 mo 68/108 (62.96) 61/105 (58.09) 1.00 (0.73 to 1.25) .98 4.84 (−8.25 to 17.98) NA
9 mo 56/102 (54.90) 58/102 (56.86) 0.90 (0.62 to 1.17) .47 −1.96 (−15.59 to 11.66) NA
12 mo 55/100 (55.00) 55/99 (55.56) 0.76 (0.49 to 1.05) .11 −0.56 (−14.37 to 13.26) NA
Sacroiliac Joint Trial
Pain intensity
reduction >30%
3 wk 41/90 (45.56) 16/83 (19.27) 2.35 (1.45 to 3.32) .001 26.29 (12.94 to 39.62) 4
6 wk 43/104 (41.35) 25/91 (27.47) 1.49 (0.94 to 2.18) .08 13.88 (0.69 to 27.05) 7
3 mo 48/105 (45.71) 29/84 (34.52) 1.33 (0.87 to 1.81) .16 11.19 (−2.74 to 25.13) NA
6 mo 50/99 (50.51) 42/85 (49.41) 1.01 (0.69 to 1.34) .94 1.10 (−13.40 to 15.58) NA
9 mo 39/98 (39.79) 33/76 (43.42) 0.88 (0.54 to 1.27) .53 −3.63 (−18.39 to 11.14) NA
12 mo 48/97 (49.48) 31/75 (41.33) 1.15 (0.75 to 1.56) .48 8.15 (−6.79 to 23.09) NA
Pain intensity
3 wk 56/90 (62.22) 30/83 (36.14) 1.68 (1.25 to 2.05) .002 26.08 (11.68 to 40.47) 4
6 wk 59/104 (56.73) 40/91 (43.95) 1.29 (0.97 to 1.59) .08 12.78 (−1.18 to 26.73) NA
3 mo 62/105 (59.05) 40/84 (47.61) 1.25 (0.94 to 1.52) .11 11.44 (−2.80 to 25.66) NA
6 mo 61/99 (61.61) 47/85 (55.29) 1.12 (0.85 to 1.35) .37 6.32 (−7.94 to 20.59) NA
9 mo 51/98 (52.04) 41/76 (53.95) 0.96 (0.68 to 1.22) .76 −1.91 (−16.85 to 13.04) NA
12 mo 57/97 (58.76) 41/75 (54.67) 1.04 (0.76 to 1.30) .77 4.09 (−10.83 to 19.03) NA
Combination Trial
Pain intensity
reduction >30%
3 wk 23/75 (30.67) 7/48 (14.58) 2.39 (1.08 to 4.16) .03 16.09 (1.64 to 30.53) 6
6 wk 32/88 (36.36) 21/72 (29.17) 1.16 (0.63 to 1.84) .60 7.19 (−7.34 to 21.73) NA
3 mo 43/86 (50.00) 19/72 (26.38) 1.92 (1.19 to 2.65) .01 23.62 (8.94 to 38.28) 4
6 mo 36/82 (43.90) 38/68 (55.88) 0.77 (0.44 to 1.11) .19 −11.98 (−27.94 to 3.98) NA
9 mo 38/81 (46.91) 26/61 (42.62) 1.05 (0.62 to 1.52) .83 4.29 (−12.21 to 20.79) NA
12 mo 37/75 (49.33) 32/56 (57.14) 0.86 (0.52 to 1.21) .47 −7.81 (−25.02 to 9.40) NA
Pain intensity
3 wk 32/75 (42.67) 12/48 (25.00) 1.67 (0.89 to 2.57) .10 17.67 (1.04 to 34.26) 5
6 wk 44/88 (50.00) 33/72 (45.83) 0.96 (0.58 to 1.37) .83 4.17 (−11.38 to 19.71) NA
3 mo 48/86 (55.81) 28/72 (38.88) 1.32 (0.85 to 1.79) .20 16.93 (1.53 to 32.32) 5
6 mo 49/82 (59.76) 43/68 (63.23) 0.91 (0.59 to 1.19) .54 −3.47 (−19.10 to 12.14) NA
9 mo 48/81 (59.25) 34/61 (55.73) 0.98 (0.62 to 1.31) .91 3.52 (−12.91 to 19.95) NA
12 mo 41/75 (54.67) 37/56 (66.07) 0.80 (0.50 to 1.10) .21 −11.40 (−28.16 to 5.35) NA
Abbreviations: NA, not applicable; NNT, number needed to treat.

a
Relative risk was estimated based on the method of Zhang et al.31

Complete data on pain intensity, functional status, and assessments were obtained from 89 participants (44%) on
global perceived recovery after 3 months were obtained from the effect measures. Participants with complete data had
198 participants (87%). Complete outcome data on all follow back pain complaints for a longer period, but were simi-
low-up points during the year were obtained from 134 partici- lar for all other demographic characteristics (eTable 1 in
pants (59%). The participants with complete data were older, Supplement 2).
more often nonsmokers, were more likely to have a partner, Two participants in the control group received radiofre-
and had low back pain complaints for a longer period (eTable quency denervation, and 2 participants did not receive any
1 in Supplement 2). treatment. In the intervention group, 11 participants did not
Seven participants in the control group received radiofre- receive or it was unknown if they received the standardized
quency denervation within the first 3 months and were marked exercise program. These 14 participants were considered par-
as participants who had protocol violations. Seven partici- ticipants who had protocol violations. Eight participants (8%)
pants (6%) in the intervention group and 6 participants (5%) in the intervention group and 10 participants (10%) in the con-
in the control group received psychological care during the trol group received psychological care during the 3-month in-
3-month intervention period. tervention period.
There was 1 registered treatment-related complication (va- In the intervention group, 35 participants had negative
sovagal reaction to treatment). results for diagnostic blocks and did not receive radiofre-
quency denervation. These participants were still included
Intention-to-Treat Analyses in the intention-to-treat analyses. The diagnostic block had
The mean difference for the primary outcome pain intensity a positive result for 68 participants, of whom 25 received
at 3 months was −0.71 (95% CI, −1.35 to −0.06). Results on all facet joint radiofrequency denervation, 21 sacroiliac joint
other follow-up points are shown in Table 2 and Table 3. The radiofrequency denervation, 21 received a combination
mean difference for functional status at 3 months was −4.20 of radiofrequency denervation treatments (facet and sacro-
(95% CI, −8.39 to −0.002); the RR for global perceived recov- iliac joint radiofrequency denervation), and 1 participant did
ery at 3 months was 1.87 (95% CI, 1.13 to 2.71). Other fol- not receive radiofrequency denervation despite a positive
low-up points and secondary outcomes are shown in eTable result for the diagnostic block.
2 in Supplement 2. One complication was recorded during the 1-year follow-up
in the intervention group: a hematoma, causing extra pain.
Post Hoc Analyses of Treatment Response The participant completely recovered.
No significant differences between the groups were found
when success was defined as more than 30% or 2 points re- Intention-to-Treat Analyses
duction or more in pain at 3 months (Table 4). The mean difference for the primary outcome pain intensity
at 3 months was −0.99 (95% CI, −1.73 to −0.25). Results on all
Sensitivity Analyses other follow-up points are shown in Table 2 and Table 3. The
When participants who had protocol violations were mean difference for functional status at 3 months was −4.66
excluded from the analysis, the interpretation of the out- (95% CI, −10.21 to 0.89); the RR for global perceived recovery
comes remained similar (eTable 3 in Supplement 2). After at 3 months was 1.99 (95% CI, 0.99 to 3.36). Other follow-up
3 months of follow–up, 41 control group participants re- points and secondary outcomes are shown in eTable 2 in
ceived radiofrequency denervation. Excluding these from the Supplement 2.
analysis did not change the long-term results (eTable 4 in
Supplement 2). The complete case analysis showed no sig- Post Hoc Analyses of Treatment Response
nificant between-group differences for the primary outcomes When success was defined as 30% pain reduction (RR, 1.92
at 3 months than participants without complete data (eTable [95% CI, 1.19 to 2.65]), there was a statistically significant dif-
5 in Supplement 2). ference at 3 months favoring the intervention group (Table 4).
Combination Trial Sensitivity Analyses

Study Participants Excluding participants who had protocol violations from the
Between January 1, 2013, and October 24, 2014 (the inclusion analysis slightly increased the contrast between the groups,
period for participants in this trial), 793 participants were eli- as significantly more people in the intervention group recov-
gible for this trial. The inclusion criteria were met by 202 par- ered based on global perceived recovery after 3 months
ticipants, and those participants were randomly assigned to (RR, 2.07 [95% CI, 1.02 to 3.43]) (eTable 3 in Supplement 2).
the intervention (n = 103) and control group (n = 99). All rea- After 3 months follow-up, 31 control group participants
sons for exclusions are presented in the flow charts (Figure 1 received radiofrequency denervation. The analyses were
and Figure 3). repeated without participants receiving the intervention
Baseline characteristics were comparable in both groups after the 3-month intervention period; this resulted in only
(Table 1). minor differences (eTable 4 in Supplement 2). The complete-
Complete data on pain intensity, functional status, and cases analysis showed no significant between-group differ-
global perceived recovery after 3 months were obtained ences for the primary outcomes at 3 months (eTable 5 in
from 168 participants (83%). Complete data on all follow-up Supplement 2).

were used. Controversy concerning the ideal threshold value

Discussion of pain reduction in the diagnostic blocks exists. A 50% cut-
off was most frequently used in previous studies38 and in clini-
In 3 trials, the effects of radiofrequency denervation for par- cal practice. Performing 2 or more independent diagnostic
ticipants with chronic low back pain due to facet joints, sac- blocks will decrease the false-positive rate, but increase the
roiliac joints, or a combination of the facet joints, sacroiliac number of false-negative blocks.38 Furthermore, a clinical trial
joints, or intervertebral disks in addition to a standardized ex- showed that multiple blocks are not cost-effective.38
ercise program were compared with a standardized exercise Fourth, the generalizability of the results might be reduced
program alone. The 2 trials assessing radiofrequency dener- by the large number of people excluded for psychological prob-
vation for the sacroiliac joints and a combination of the facet lems. In the Netherlands, participants visiting a pain clinic often
joints, sacroiliac joints, or intervertebral disks showed a sta- have long-lasting persistent low back pain. A large number of
tistically significant but not clinically important improve- these participants have psychological problems. These partici-
ment in pain intensity 3 months after the intervention. No clini- pants were excluded from this study because in daily practice
cally important or statistically significant differences between they are not considered candidates for radiofrequency denerva-
the groups were shown in the trial assessing radiofrequency tion and will be referred to psychological treatment.
denervation for facet joint pain. Only small or no effects were Fifth, in all 3 trials, some control group participants re-
found for all secondary outcomes. ceived radiofrequency denervation after the 3-month inter-
Based on this study, radiofrequency denervation is not rec- vention period (25% in the facet joint trial, 35% in sacroiliac
ommended and should be performed only in a research set- joint trial, and 31% in the combination trial) and some inter-
ting. Patients with chronic low back pain who show no im- vention group participants received a second radiofrequency
provement in symptoms after conservative treatment have no denervation (8% in the facet joint trial, 17% in the sacroiliac
clear alternative therapies that have been shown to be effec- joint trial, and 15% in the combination trial). This could have
tive. Future research regarding the diagnosis and treatment for influenced the long-term outcomes. However, sensitivity analy-
low back pain in participants with chronic low back pain is nec- ses without these participants showed similar results.
essary and should focus on better participant selection (be- Sixth, in the sacroiliac joint trial, there was a higher drop-
cause there remains a possibility that radiofrequency dener- out in the control group. This could potentially have biased the
vation could be beneficial on a subset of participants) and long-term results.
improvement of the treatment techniques. Seventh, in the combination trial, not all participants in
the intervention group received radiofrequency denerva-
Strengths and Limitations tion, because they did not respond to the diagnostic block or
Strengths of these trials are the large sample sizes and strati- provocative discography.
fied randomization that allowed for well-balanced study Eighth, more missing data were found in the combina-
groups, and the use of outcome measures as recommended by tion trial compared with the other 2 trials. This is a potential
the core outcome set for low back pain research.32 In addition limitation, but because of the relatively large number of
to the primary time point at 3 months, a follow-up of 12 months dropouts at 12 months, the complete case analysis might also
was included. be biased. Although we did not define differences between
This study has several limitations. First, different radiofre- the complete-case analysis and the intention-to-treat analy-
quency denervation techniques (cooled radiofrequency dener- sis using all data, it is possible that completers are different
vation, Palisade, and Simplicity III) were used in the sacroiliac from noncompleters, which could have biased the results of
joint trial.33-36 However, the groups were too small for a sub- the complete-case analyses.
group analysis. Second, because the aim of the study was to pro- Ninth, we assessed multiple outcomes and made no ad-
vide evidence of the added value of radiofrequency denerva- justment for multiple comparisons, which could have re-
tion in a multidisciplinary setting, as done in daily practice, sulted in some statistically significant findings by chance.
participants and clinicians were not blinded. Evidence sug-
gests that treatment effects for subjective outcomes may Comparison With the Literature
be overestimated when outcome assessors (ie, participants, if Recent systematic reviews have evaluated the association of
outcomes are self-reported) are not blinded.37 However, the radiofrequency denervation with isolated pain sources and
magnitude of this bias is unknown. The lack of blinding was a showed evidence of low to moderate quality for associations
significant limitation of the trials, and it is possible that radio- of facet joint radiofrequency with small positive effects on
frequency denervation could even be harmful, but the lack of pain and functional status compared with placebo or steroid
blinding may have made the treatment effect seem null. Also, injections.9-11 There is very low to moderate quality evidence
the short-term differences in global perceived recovery for facet and conflicting evidence for sacroiliac joint radiofrequency de-
joint and sacroiliac joint radiofrequency denervation in ab- nervation and radiofrequency denervation in the interverte-
sence of a difference in functional status might be the result of bral disk.5,9-11 In the trials included in these reviews, partici-
a nonspecific effect due to the nonblinded study design. pants had a baseline pain score of 1 point lower compared with
Third, a reference standard for diagnosing facet joint or sac- this trial, and the radiofrequency denervation groups de-
roiliac joint pain is not available.14 In this pragmatic study, di- creased more (to 3.3 of 10) than the placebo groups (to 5.0 of
agnostic tests that are commonly applied in clinical practice 10). In the Mint study, participants in both groups decreased

in pain, but both groups continued to have a higher pain level joints, or a combination of facet joints, sacroiliac joints, or
compared with other similar trials. intervertebral disks, radiofrequency denervation combined
with a standardized exercise program resulted in either no
improvement or no clinically important improvement in
chronic low back pain compared with a standardized exer-
Conclusions cise program alone. The findings do not support the use of
In 3 randomized clinical trials of participants with chronic radiofrequency denervation to treat chronic low back pain
low back pain originating in the facet joints, sacroiliac from these sources.
ARTICLE INFORMATION Society for Anesthesiology, and the Dutch health 5. van Zundert J, Hartrick C, Lataster A, Huygen F,
Accepted for Publication: June 6, 2017. insurance companies. Mekhail N, van Kleef MPJ. Evidence-Based
Role of the Funder/Sponsor: The funders played Interventional Pain Practice: According to Clinical
Author Affiliations: Department of Diagnoses. Oxford, UK: Wiley-Blackwell; 2011.
Anesthesiology, Erasmus University Medical Centre, no role in the design and conduct of the study;
Rotterdam, the Netherlands (Juch, Groeneweg, collection, management, analysis, and 6. Bogduk N. The anatomical basis for spinal pain
Huygen); Department of Health Sciences, Faculty interpretation of the data; and preparation, review, syndromes. J Manipulative Physiol Ther. 1995;18(9):
of Earth and Life Sciences, Vrije Universiteit or approval of the manuscript or decision to submit 603-605.
Amsterdam, Amsterdam, the Netherlands the manuscript for publication. 7. Manchikanti L, Hirsch JA, Pampati V, Boswell MV.
(Maas, Ostelo, van Dongen, van Tulder); Additional Contributions: We thank Merel Utilization of facet joint and sacroiliac joint
The EMGO+ Institute for Health and Care Research, van Raamt, PT, and Robert van Cingel, PhD interventions in Medicare population from 2000 to
Amsterdam, the Netherlands (Maas, Ostelo, (both from Sports Medical Centre Papendal 2014: explosive growth continues! Curr Pain
van Dongen, van Tulder); Department of [SMCP]), for their help in developing the exercise Headache Rep. 2016;20(10):58.
Epidemiology and Biostatistics, Vrije Universiteit program. SMCP received financial support 8. Koes BW, van Tulder M, Lin C-WC, Macedo LG,
Medical Centre Amsterdam, Amsterdam, the from the research grant for the development McAuley J, Maher C. An updated overview of
Netherlands (Ostelo); Department of of the protocol. We also thank all participants, clinical guidelines for the management of
Anesthesiology, Rijnstate hospital, Velp, the physiotherapists, and everyone working on the non-specific low back pain in primary care. Eur
Netherlands (Kallewaard); Department of General Mint study at the participating pain clinics. Spine J. 2010;19(12):2075-2094.
Practice, Erasmus University Medical Centre, The participating pain clinics received
Rotterdam, the Netherlands (Koes, Verhagen). reimbursement from the research grant for their 9. Maas ET, Ostelo RW, Niemisto L, et al.
extra study-related administrative tasks. Radiofrequency denervation for chronic low back
Author Contributions: Drs Maas and Juch had full pain. Cochrane Database Syst Rev. 2015;10(10):
access to all of the data in the study and take Additional Information: The following were in CD008572.
responsibility for the integrity of the data and the charge of participant recruitment and treatment
accuracy of the data analysis. Drs Juch and Maas per clinic: Bernard Lo, MD (Diakonessenhuis 10. Henschke N, Kuijpers T, Rubinstein SM, et al.
are shared first author, listed in alphabetical order. Utrecht/Zeist); Harmen Pelleboer, MD (Medisch Injection therapy and denervation procedures for
Concept and design: Juch, Maas, Ostelo, Centrum Alkmaar); Jasper Kal, MD, PhD, FIPP (Onze chronic low-back pain: a systematic review. Eur
Groeneweg, Kallewaard, Koes, Verhagen, Huygen, Lieve Vrouwe Gasthuis Amsterdam); Dennis Spine J. 2010;19(9):1425-1449.
van Tulder. Harhangi, MD (Havenziekenhuis Rotterdam); 11. Maas ET, Juch JN, Groeneweg JG, et al.
Acquisition, analysis, or interpretation of data: Juch, Renate Munnikes, MD, FIPP (Maasstad ziekenhuis Cost-effectiveness of minimal interventional
Maas, Koes, van Dongen, Huygen, van Tulder. Rotterdam); Willem-Jan Hofsté, MD (St Antonius procedures for chronic mechanical low back pain:
Drafting of the manuscript: Juch, Maas, Ostelo, ziekenhuis Nieuwegein); Katja Bürger, MD (Alrijne design of four randomised controlled trials with an
Groeneweg, Koes, Verhagen, Huygen, van Tulder. ziekenhuis Leiderdorp); Michel Wagemans, MD economic evaluation. BMC Musculoskelet Disord.
Critical revision of the manuscript for important (Reinier de Graaf Gasthuis Delft); Mettje Hanje, MD 2012;13:260.
intellectual content: All authors. (Medisch Spectrum Twente Enschede); Michel 12. Szadek KM, van der Wurff P, van Tulder MW,
Statistical analysis: Juch, Maas, van Dongen. Terheggen, MD (Rijnstate ziekenhuis Velp); Gilbert Zuurmond WW, Perez RSGM. Diagnostic validity of
Obtained funding: Koes, Huygen, van Tulder. Tjiang, MD (Amphia ziekenhuis Breda); Maurice criteria for sacroiliac joint pain: a systematic review.
Administrative, technical, or material support: Juch, Giezeman, MD (Isala klinieken, Zwolle); Joep J Pain. 2009;10(4):354-368.
Maas, Kallewaard, Huygen. Scholten, MD (Gelre ziekenhuis Apeldoorn); Eric
Supervision: Ostelo, Groeneweg, Huygen, van van den Bosch, MD (Medisch Centrum 13. Hancock MJ, Maher CG, Latimer J, et al.
Tulder. Leeuwarden); Antal van der Gaag, MD (Catharina Systematic review of tests to identify the disc,
ziekenhuis Eindhoven); Hans Aukes, MD (Albert SIJ or facet joint as the source of low back pain. Eur
Conflict of Interest Disclosures: All authors have Spine J. 2007;16(10):1539-1550.
completed and submitted the ICMJE Form for Schweitzer ziekenhuis Sliedrecht).
Disclosure of Potential Conflicts of Interest. Dr 14. Bogduk N. Practice Guidelines for Spinal
Ostelo reports receiving grant funding from the REFERENCES Diagnostic and Treatment Procedures. Hinsdale, IL:
Netherlands Organisation for Scientific Research 1. Martin BI, Deyo RA, Mirza SK, et al. Expenditures International Spine Intervention Society; 2004.
and Scientific Association Physiotherapy. Dr van and health status among adults with back and neck 15. Kallewaard JW, Terheggen MA, Groen GJ, et al.
Tulder reports receiving funding to his institution problems. JAMA. 2008;299(6):656-664. 15: Discogenic low back pain. Pain Pract. 2010;10
from professional organizations (eg, the Royal 2. Vos T, Flaxman AD, Naghavi M, et al. Years lived (6):560-579.
Dutch Association for Physiotherapy; the with disability (YLDs) for 1160 sequelae of 289 16. Devilly GJ, Borkovec TD. Psychometric
Netherlands National Chiropractic Association and diseases and injuries 1990-2010: a systematic properties of the credibility/expectancy
the European Chiropractic Union); travel expenses analysis for the Global Burden of Disease Study questionnaire. J Behav Ther Exp Psychiatry. 2000;
by the organizing professional organizations when 2010 [correction appears in Lancet. 31(2):73-86.
speaking at conferences; and honoraria for 2013;381(9867):628]. Lancet. 2012;380(9859):
reviewing grant proposals from Swedish and 17. Bombardier C. Outcome assessments in the
2163-2196. evaluation of treatment of spinal disorders:
Canadian governmental grant agencies. He has not
received any honoraria or travel expenses from the 3. Katz JN. Lumbar disc disorders and low-back summary and general recommendations. Spine
industry. No other disclosures were reported. pain: socioeconomic factors and consequences. (Phila Pa 1976). 2000;25(24):3100-3103.
J Bone Joint Surg Am. 2006;88(suppl 2):21-24. 18. Hoeijenbos M, Bekkering T, Lamers L,
Funding/Support: This study was funded by grant
171202013 from the Netherlands Organization for 4. Dieleman JL, Baral R, Birger M, et al. Hendriks E, van Tulder M, Koopmanschap M.
Health Research and Development, by the Dutch US spending on personal health care and public Cost-effectiveness of an active implementation
health, 1996-2013. JAMA. 2016;316(24):2627-2646.

strategy for the Dutch physiotherapy guideline for 26. Hays RD, Morales LS. The RAND-36 measure of 33. Ferrante FM, King LF, Roche EA, et al.
low back pain. Health Policy. 2005;75(1):85-98. health-related quality of life. Ann Med. 2001;33(5): Radiofrequency sacroiliac joint denervation for
19. Cohen SP, Hurley RW, Buckenmaier CC III, 350-357. sacroiliac syndrome. Reg Anesth Pain Med. 2001;26
Kurihara C, Morlando B, Dragovich A. Randomized 27. Lousberg R, Van Breukelen GJP, Groenman NH, (2):137-142.
placebo-controlled study evaluating lateral branch Schmidt AJM, Arntz A, Winter FAM. Psychometric 34. Patel N, Gross A, Brown L, Gekht G.
radiofrequency denervation for sacroiliac joint pain. properties of the Multidimensional Pain Inventory, A randomized, placebo-controlled study to assess
Anesthesiology. 2008;109(2):279-288. Dutch Language Version (MPI-DLV). Behav Res Ther. the efficacy of lateral branch neurotomy for chronic
20. Cosman ER Jr, Gonzalez CD. Bipolar 1999;37(2):167-182. sacroiliac joint pain. Pain Med. 2012;13(3):383-398.
radiofrequency lesion geometry: implications for 28. Ostelo RWJG, Deyo RA, Stratford P, et al. 35. Burnham RS, Yasui Y. An alternate method of
palisade treatment of sacroiliac joint pain. Pain Pract. Interpreting change scores for pain and functional radiofrequency neurotomy of the sacroiliac joint:
2011;11(1):3-22. status in low back pain: towards international a pilot study of the effect on pain, function, and
21. Schmidt PC, Pino CA, Vorenkamp KE. Sacroiliac consensus regarding minimal important change. satisfaction. Reg Anesth Pain Med. 2007;32(1):12-19.
joint radiofrequency ablation with a multilesion Spine (Phila Pa 1976). 2008;33(1):90-94. 36. van Tilburg CW, Schuurmans FA, Stronks DL,
probe: a case series of 60 patients. Anesth Analg. 29. van der Roer N, Ostelo RW, Bekkering GE, Groeneweg JG, Huygen FJ. Randomized
2014;119(2):460-462. van Tulder MW, de Vet HC. Minimal clinically sham-controlled double-blind multicenter clinical
22. Gauci CA, Jankowiak B. Manual of RF important change for pain intensity, functional trial to ascertain the effect of percutaneous
Techniques: A Practical Manual of Radiofrequency status, and general health status in patients with radiofrequency treatment for sacroiliac joint pain:
Procedures in Chronic Pain Management. Seattle, WA: nonspecific low back pain. Spine (Phila Pa 1976). three-month results. Clin J Pain. 2016;32(1):921-926.
CoMedical; 2011. 2006;31(5):578-582. 37. Kahan BC, Cro S, Doré CJ, et al. Reducing bias in
23. Downie WW, Leatham PA, Rhind VM, Wright V, 30. Twisk JWR. Applied Multilevel Analysis: open-label trials where blinded outcome
Branco JA, Anderson JA. Studies with pain rating A Practical Guide for medical researchers. Cambridge, assessment is not feasible: strategies from two
scales. Ann Rheum Dis. 1978;37(4):378-381. UK: Cambridge Univ Press; 2006. randomised trials. Trials. 2014;15(1):456.
24. Fairbank JC, Pynsent PB. The Oswestry 31. Zhang J, Yu KF. What’s the relative risk? 38. Cohen SP, Huang JHY, Brummett C. Facet joint
Disability Index. Spine (Phila Pa 1976). 2000;25(22): a method of correcting the odds ratio in cohort pain—advances in patient selection and treatment.
2940-2952. studies of common outcomes. JAMA. 1998;280 Nat Rev Rheumatol. 2013;9(2):101-116.
(19):1690-1691.
25. Brooks R. EuroQol: the current state of play.
Health Policy. 1996;37(1):53-72. 32. Chiarotto A, Deyo RA, Terwee CB, et al. Core
outcome domains for clinical trials in non-specific
low back pain. Eur Spine J. 2015;24(6):1127-1142.

Juch 2017

Uploaded by

Copyright:

Available Formats

Juch 2017

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Juch 2017

Uploaded by

Copyright:

Available Formats

Research

JAMA | Original Investigation

Effect of Radiofrequency Denervation on Pain Intensity

DESIGN, SETTING, AND PARTICIPANTS Three pragmatic multicenter, nonblinded randomized

INTERVENTIONS All participants received a 3-month standardized exercise program and

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

jama.com (Reprinted) JAMA July 4, 2017 Volume 318, Number 1 69

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

70 JAMA July 4, 2017 Volume 318, Number 1 (Reprinted) jama.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

Figure 1. Flow of Patients Through Enrollment in the 3 Randomized Clinical Trials

10 592 Potential participants

5168 Included in observational studya

5424 Patients asked to participate in 1 of 3

1882 Excludedb 2270 Excludedb 591 Excludedb

See Figure 2 See Figure 2 See Figure 3

jama.com (Reprinted) JAMA July 4, 2017 Volume 318, Number 1 71

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

Continued From Figure 1

72 JAMA July 4, 2017 Volume 318, Number 1 (Reprinted) jama.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

Figure 3. Flow of Patients Through the Combination Trial

Continued From Figure 1

103 Randomized to intervention group 99 Randomized to control group

98 Completed baseline visit 89 Completed baseline visit

77 Completed 3-wk follow-up 56 Completed 3-wk follow-up

90 Completed 6-wk follow-up 82 Completed 6-wk follow-up

88 Completed 3-mo follow-up 80 Completed 3-mo follow-up

85 Completed 6-mo follow-up 75 Completed 6-mo follow-up

80 Completed 9-mo follow-up 68 Completed 9-mo follow-up

77 Completed 12-mo follow-up 61 Completed 12-mo follow-up

103 Included in intention-to-treat analysis 99 Included in intention-to-treat analysis

jama.com (Reprinted) JAMA July 4, 2017 Volume 318, Number 1 73

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

Table 1. Baseline Characteristics of Participants With Chronic Low Back Pain

Facet Joint Trial Sacroiliac Joint Trial Combination Trial

Facet Joint Trial Baseline characteristics were comparable across groups

74 JAMA July 4, 2017 Volume 318, Number 1 (Reprinted) jama.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

jama.com (Reprinted) JAMA July 4, 2017 Volume 318, Number 1 75

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

76 JAMA July 4, 2017 Volume 318, Number 1 (Reprinted) jama.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

Abbreviations: NA, not applicable; NNT, number needed to treat.

jama.com (Reprinted) JAMA July 4, 2017 Volume 318, Number 1 77

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a Weill Cornell Medical Library User on 07/03/2017

Combination Trial Sensitivity Analyses

78 JAMA July 4, 2017 Volume 318, Number 1 (Reprinted) jama.com

© 2017 American Medical Association. All rights reserved.