The efficacy of interventions for LBP

Two recent systematic reviews have highlighted major gaps in the literature with regards to studies of different non-invasive treatments for NSLBP (Cashin et al 2022) and interventional approaches to musculoskeletal conditions generally, and specifically for lumbar spine related comnditions (Dutra et al 2026). I posted some remarks about the Dutra study on X and responded to a question. Here is my response:

"I am analyzing the paper, with focus on the spinal pain RCTs. There seems to be many studies missing. Of the 6 lumbar spine studies selected the most recent one from 2012. Albert et al 2013 (placebo vs antibiotics) is not included. Baten et al also not there. MacVicar et al 2021 comparing RF with placebo is not there. Wallis Lord & Bogduk 1999 on facet joint RF in WAD cases is absent. IDET for lumbar discogenic pain is effective (Freeman et al 2005 and Pauza et al 2004), yet the 1999 report (Barendse et al) says it is not. Patel 2012 shows effectivess of lateral branch RF for SIJ pain. The other studies compared different treatments for NSLBP which (predictably) found the treatments ineffective. The SR process itself is highly suspect IMO. useless treatments dilute and ultimately wash away the established efficacy of treatments to specific patho-anatomically defined sources of pain. This SR and Meta-analysis seems to miss the important fact from their own data, that in low back pain studies, treatments targeting clearly identified patho-anatomic diagnoses ARE effective in the RCTs they cite. I would appreciate any one else's perspective on that"

I decided to use Claude AI to do some research for me. Here os my question:

"you able to identify published RCTs that compare a specific treatment with placebo in cohorts where a solid criterion standard of diagnosis was used to select a study sample. Do this for low back pain and neck pain please"

This is Claude AI's response:

"Diagnostic Rigor in Placebo-Controlled Spinal Pain RCTs: A Review Prompted by Dutra et al. 2026

Prepared for Mark, 18 August 2026Context: Reviewing the diagnostic-selection methodology of the low back pain (LBP) trials in Dutra MNP, Canez MS, Pereira BR, Dias JB, Luza LP, Araujo FX. "Invasive Procedures for Most Musculoskeletal Conditions Are Not Better Than Placebo: A Systematic Review With Meta-Analysis." Musculoskeletal Care 2026;24:e70244.

The core question

Dutra et al. concluded that invasive spinal procedures (radiofrequency denervation, intradiscal thermal therapy, vertebroplasty) show no clinically meaningful benefit over placebo. That conclusion is only as strong as the diagnostic method used to select the patients in the underlying trials. If a trial enrolled patients as having "facet joint pain" or "discogenic pain" without a validated criterion-standard diagnostic test, a null result may reflect a diluted, heterogeneous sample (many patients who never had the target pathology) rather than a truly ineffective treatment. This review checks that, trial by trial, for the six low back pain studies Dutra included, and separately searches for neck pain equivalents (none were included in Dutra's review).

Part 1 — The six LBP trials in Dutra et al., and how each diagnosed its cohort

Study	Country	Intervention vs. sham	Diagnostic method for cohort selection	Rigor
Leclaire et al. 2001, Spine	Canada	Radiofrequency facet (medial branch) denervation	Single lidocaine block, assessed informally by referring family physicians ("24 h relief" reported non-quantitatively); no confirmatory second block	Weak. A later systematic review (Cohen 2007-era literature) notes ~94% of screened back-pain patients qualified as "facet pain" candidates — far above the ~15% prevalence facet pain is expected to have in unselected LBP populations — and describes this trial as having "major conceptual flaws" in patient selection
van Kleef et al. 1999, Spine	Netherlands	Radiofrequency lumbar facet denervation	Single intra-articular/medial-branch block, patients stratified only into "good" vs. "equivocal" responders; no second, confirmatory block with a different-acting anesthetic	Weak-moderate. Single block, no comparative confirmation
Nath et al. 2008, Spine	Sweden	Percutaneous lumbar zygapophyseal (facet) joint neurotomy	Reported in secondary literature as requiring a positive response on three diagnostic blocks before randomization	Moderate — more blocks required than Leclaire/van Kleef, but available secondary sources don't confirm whether this used the differential-anesthetic (short- vs. long-acting) "comparative block" design that defines the true criterion standard (see Part 3). Would need primary-source verification if the exact protocol matters to you
Barendse et al. 2001, Spine	Netherlands	Percutaneous intradiscal radiofrequency thermocoagulation	Provocative discography (positive concordant pain response)	Moderate, contested. Discography is the conventional reference test for "discogenic" pain but is itself not a validated gold standard — see caveat below
Freeman et al. 2005, Spine	Australia	Intradiscal electrothermal therapy (IDET)	Provocative discography	Same caveat as above
Pauza et al. 2004, Spine J	USA	IDET	Provocative discography	Same caveat as above

Discography caveat: Carragee and colleagues' "gold standard evaluation" work on provocative discography found high false-positive rates, particularly in patients with chronic pain, prior surgery, or psychosocial risk factors — meaning even the "best-diagnosed" LBP trials in Dutra's set (the three IDET/discogenic trials) rested on a reference test whose own specificity is disputed in the pain-medicine literature.

Part 2 — Does diagnostic rigor actually change the measured treatment effect? Yes, substantially.

A stratified systematic review of lumbar medial branch radiofrequency neurotomy (MacVicar et al., Pain Medicine 2020) pooled outcomes by exactly the diagnostic-rigor gradient described above:

·                     Single block, ≥50% pain relief required: 26% achieved ≥50% relief at 6 months

·                     Dual (comparative) blocks, ≥50% relief required: 49%

·                     Dual comparative blocks, 100% (complete) relief required — the strict International Spine Intervention Society (ISIS)/Bogduk criterion standard: 56%

In other words, the weaker the diagnostic selection, the smaller the measured treatment effect — a textbook dose-response relationship between diagnostic rigor and apparent efficacy. Leclaire (single, informal block) and van Kleef (single block) sit at the weak end of this gradient; both are exactly the trials most likely to understate a true treatment effect through sample dilution.

A related randomized trial by Cohen et al. (Anesthesiology 2010) directly compared 0, 1, and 2 diagnostic-block paradigms before radiofrequency ablation and found dual blocks raised clinical success from ~33–39% to ~64%, though at higher cost per successful outcome — so the field is not unanimous that more blocks are always worth it operationally, even though the diagnostic-accuracy gradient itself is well established.

Part 3 — What the actual "criterion standard" looks like, and where it has been used in a placebo-controlled RCT

The validated reference standard for zygapophyseal (facet) joint pain, developed and psychometrically validated by Bogduk's group, is the comparative (controlled) diagnostic block protocol: two separate blocks, performed on different days, using local anesthetics of different expected duration (short-acting lidocaine, long-acting bupivacaine), with a positive diagnosis requiring concordant relief — short relief after lidocaine, longer relief after bupivacaine — and typically complete (100%) relief of the index pain on both occasions. This is the ISIS practice standard, and it was formally validated for the cervical spine by Barnsley & Bogduk (Clin J Pain 1993) and Lord et al. (Pain 1995, "Comparative local anaesthetic blocks in the diagnosis of cervical zygapophysial joint pain").

That validated cervical protocol was then used, essentially unmodified, as the entry criterion for what is widely regarded as the best-designed placebo-controlled RCT in the interventional spine literature:

Lord RS, Barnsley L, Wallis BJ, McDonald GJ, Bogduk N. "Percutaneous Radio-Frequency Neurotomy for Chronic Cervical Zygapophyseal-Joint Pain." New England Journal of Medicine 1996;335(23):1721–1726. — PubMed / NEJM

·                     Only patients with complete relief of pain on comparative diagnostic blocks were randomized to radiofrequency neurotomy vs. a sham procedure.

·                     Result: median time for pain to return to 50% of pre-treatment severity was 263 days in the active group vs. 8 days in the sham group — a large, statistically and clinically significant effect.

·                     This trial is routinely cited (including in the "Evidence and Controversies" literature on facet radiofrequency treatment) as the exemplar of what happens when the criterion-standard diagnostic protocol is actually used in a sham-controlled design: a clear positive result, in contrast to the largely null lumbar trials that used weaker single-block selection.

For low back pain, no equivalent sham-controlled RCT using the full comparative dual-block (ISIS) criterion standard appears to exist. The searches above turned up:

·                     Prospective cohort/audit studies from Bogduk-affiliated groups (e.g., lumbar medial branch neurotomy audits in New Zealand) that do use the rigorous dual-comparative-block protocol and report much higher success rates (consistent with Part 2's gradient) — but these are not placebo-controlled.

·                     The largest and most recent lumbar RF trial, the MINT trial (Juch et al., JAMA 2017;318(1):68–81 — PubMed), used a single block with only 50% relief required, and — importantly — its comparator was a standardized exercise program, not a sham procedure (RF-plus-exercise vs. exercise alone). That's why it wasn't eligible for Dutra's placebo-controlled review, and it also doesn't meet the "solid criterion standard" bar you're asking about.

·                     Cohen et al.'s diagnostic-paradigm RCT (Anesthesiology 2010) is a trial of diagnostic strategies, not a treatment-vs-placebo trial.

So the honest answer for low back pain is: the best-diagnosed cohorts (dual comparative blocks, ISIS criteria) have only been studied in uncontrolled or active-comparator designs; the placebo-controlled RCTs (the ones Dutra pooled) all used weaker single-block or discography-based selection. This is a genuine, citable gap in the evidence base — it's not simply that "invasive procedures don't beat placebo," but that the trials capable of showing they do (rigorous diagnosis + sham control combined) largely haven't been run for the lumbar spine.

Part 4 — Neck pain: what else exists beyond Lord 1996

Dutra et al.'s review contains no neck pain trial at all — their MSK conditions were limited to lateral epicondylitis, shoulder impingement, SLAP lesions, meniscal tears/knee pain, knee osteoarthritis, vertebral fractures, low back pain, and sacroiliac joint pain. So for neck pain, Lord 1996 stands as the key positive example of rigorous-diagnosis-plus-sham-control design. Other neck/cervicogenic-headache RF trials found in this search were weaker on one or both criteria:

·                     Wallis, Lord, Bogduk 1997 — a psychological-outcomes sub-study of the same Lord cohort (not an independent treatment RCT).

·                     Stovner et al. 2004 (cervical RF for cervicogenic headache, ISRCTN07444684) — PMC — used clinical diagnostic criteria only (Sjaastad criteria for cervicogenic headache, no diagnostic blocks for inclusion) and was not placebo-controlled (active RF vs. active conservative treatment arms) — fails both the diagnostic-rigor and placebo-control tests, and found no significant difference between arms.

·                     Haspeslagh et al. 2006 (cervicogenic headache, RF vs. sham) — turned up in the neck-pain radiofrequency meta-analysis as a sham-controlled trial, but I was not able to confirm its diagnostic-selection methodology from available sources; worth checking the primary paper directly if this one matters to you.

Bottom line

1.                  Among Dutra et al.'s six LBP trials, diagnostic rigor ranges from poor (Leclaire — informal single block, near-100% "positive" rate) to moderate (Nath — three blocks, but comparative-block design unconfirmed; Barendse/Freeman/Pauza — discography, itself a contested reference standard). None used the full dual-comparative-block ISIS criterion standard.

2.                  There is a demonstrated dose-response relationship (MacVicar et al. 2020) between diagnostic rigor and measured treatment effect in lumbar facet RF neurotomy — meaning the weakly-diagnosed trials in Dutra's pool are the ones most susceptible to false-negative (no-better-than-placebo) results from sample dilution.

3.                  The one RCT in this literature that combines a validated criterion-standard diagnosis with a true sham control is Lord et al. 1996 (NEJM) — for the cervical spine — and it found a large, clinically meaningful benefit of radiofrequency neurotomy over placebo.

4.                  No equivalent trial (rigorous diagnosis + sham control) appears to have been done for the lumbar spine — the rigorously-diagnosed lumbar cohorts are cohort studies without a placebo arm, and the placebo-controlled lumbar RCTs all used weaker diagnostic selection. This is a legitimate, citable limitation of Dutra et al.'s "no better than placebo" conclusion as applied to low back pain specifically.

Sources

·                     Dutra et al. 2026, Musculoskeletal Care (uploaded PDF)

·                     Cochrane: Radiofrequency denervation for chronic low back pain (Maas et al. 2015)

·                     Radiofrequency Treatment of Facet-related Pain: Evidence and Controversies

·                     van Kleef et al. 1999, Spine — PubMed

·                     Leclaire et al. 2001, Spine — PubMed

·                     Systematic Review of the Effectiveness of Lumbar Medial Branch Thermal Radiofrequency Neurotomy, Stratified for Diagnostic Methods and Procedural Technique (MacVicar et al., Pain Medicine 2020)

·                     Cohen et al. 2010, Anesthesiology — Multicenter cost-effectiveness study of 0/1/2 diagnostic blocks

·                     Evidence for Diagnostic Blocks Prior to Radiofrequency Ablation of Innervation to the Lumbar Facet Joints — ASRA

·                     Lord et al. 1996, NEJM — Percutaneous Radio-Frequency Neurotomy for Chronic Cervical Zygapophyseal-Joint Pain

·                     Comparative local anaesthetic blocks in the diagnosis of cervical zygapophysial joint pain (Lord et al., Pain 1995)

·                     Juch et al. 2017, JAMA — MINT Randomized Clinical Trials

·                     A Systematic Review and Meta-analysis of the Effectiveness of Radiofrequency Neurotomy in Managing Chronic Neck Pain

·                     Randomised controlled trial of cervical radiofrequency lesions for cervicogenic headache (Stovner et al.)

·                     A gold standard evaluation of the "discogenic pain" diagnosis as determined by provocative discography (Carragee et al.)

These issues are real and there is no simple answer.

1.            Dutra MNP, Canez MS, Pereira BR, Dias JB, Luza LP, de Araujo FX. Invasive Procedures for Most Musculoskeletal Conditions Are Not Better Than Placebo: A Systematic Review With Meta-Analysis. Musculoskeletal Care 2026;24(3):e70244. DOI: 10.1002/msc.70244.

2.            Cashin AG, Furlong BM, Kamper SJ, et al. Analgesic effects of non-surgical and non-interventional treatments for low back pain: a systematic review and meta-analysis of placebo-controlled randomised trials. BMJ Evid Based Med 2025. DOI: 10.1136/bmjebm-2024-112974.