Main Points

• Vertebroplasty is probably more effective than sham or usual care for vertebral compression fractures for reducing pain and improving function in older (Medicare-eligible) populations, but benefits are small. Benefits are smaller in sham compared with usual care controlled trials and larger in trials of patients with more acute symptoms.
• Kyphoplasty is probably more effective than usual care for vertebral compression fractures for reducing pain and improving function in older (Medicare-eligible) populations, but has not been compared against sham.
• Cooled radiofrequency denervation is probably moderately more effective for reducing pain and improving function than sham for sacroiliac pain in younger populations and similarly effective versus conventional radiofrequency for presumed facet joint pain and piriformis corticosteroid injection for piriformis syndrome may be similarly effective versus sham for pain at 1 week, but more effective for reducing pain at 1 month. These interventions were evaluated in younger (non-Medicare-eligible) populations, but findings can probably be applied to older populations.
• Research is needed to determine the benefits and harms of other interventional procedures addressed in this report. Ideally, future trials of interventional procedures should enroll older, Medicare-eligible populations, utilize sham controls, evaluate function as well as pain, include rigorous evaluation of harms, evaluate longer-term outcomes, and evaluate how benefits and harms according to demographic, clinical, and technical factors.

Background and Purpose

The purpose of this systematic review is to evaluate the effectiveness and harms of selected interventional procedures for acute and chronic pain in the Medicare population. The review focuses on procedures which are not currently covered for by the Centers for Medicare & Medicaid Services (CMS) but are relevant for and have potential utility for use in the Medicare population, or procedures that are covered by CMS but for which there is important uncertainty or controversy regarding use.

Methods

Electronic databases (Ovid^® MEDLINE^®, PsycINFO^®, the Cochrane Central Register of Controlled Trials, and the Cochrane Database of Systematic Reviews) were searched through April 12, 2021 for relevant publications. Searches were supplemented by reviewing reference lists and a Federal Register Notice.

Randomized controlled trials (RCTs) of populations undergoing the designated interventional procedures for the specified conditions versus usual care, no treatment, placebo, or sham were selected using predefined criteria and dual review. Observational studies were eligible for assessment of rare, serious adverse events. This review focused on 10 interventional procedures for specific conditions:

1. Vertebral augmentation procedures (vertebroplasty and kyphoplasty) for pain due to vertebral compression fracture
2. Cooled radiofrequency denervation for degenerative back or hip pain and pulsed radiofrequency denervation for degenerative back pain
3. Intradiscal and facet joint platelet-rich plasma for presumed discogenic back pain
4. Intradiscal stem cells for presumed discogenic back pain
5. Intradiscal methylene blue for presumed discogenic back pain
6. Intradiscal ozone for radicular low back pain or nonradicular, presumed discogenic back pain (protocol modification to include intradiscal ozone plus corticosteroid)
7. Sphenopalatine block for trigeminal neuralgia or headache
8. Occipital stimulation for headache
9. Piriformis injection (local anesthetic, corticosteroid, and/or botulinum toxin) for piriformis syndrome
10. Peripheral nerve stimulation for ulnar, median, or radial neuropathy

The main outcomes were pain and function, and additional outcomes were quality of life, emotional function, global improvement, and harms. Outcomes were analyzed at 1 to 2 weeks, 2 to 4 weeks, 1 to 6 months, 6 to 12 months, and 12 months and longer. Meta-analyses were conducted for vertebroplasty versus no vertebroplasty (sham or usual care) and effects on pain, function, quality of life, and harms; analyses were conducted to assess how the control type, duration of symptoms, and other factors impacted findings. Otherwise, meta-analyses were not conducted due to small number of studies, methodological limitations, and study heterogeneity. The magnitude of effects was classified as small, moderate or large using previously defined criteria, and strength of evidence was assessed.

Results

The review included 37 RCTs on the comparative effectiveness of interventional therapies for acute and chronic pain. Evidence was most robust for vertebroplasty, followed by kyphoplasty and radiofrequency denervation, and limited for other interventions. Evidence on vertebroplasty and kyphoplasty was highly relevant to populations eligible for Medicare, based on mean age of over 65 years in the trials. For other interventions, patients were younger and populations eligible for Medicare for reasons other than older age were not addressed. Main findings (focusing on effects on pain and function) are summarized by interventional procedure.

Limitations

We excluded non-English–language articles and did not search for studies published only as abstracts. We did not conduct statistical and graphical methods for assessing for small sample effects (a potential marker for publication bias) due to small numbers of trials and heterogeneity in study design methods, patient populations, and outcomes.

The evidence base had important limitations. For vertebroplasty, trials varied with regard to patient selection criteria (e.g., duration of pain), technical factors (e.g., volume of PMMA), and sham interventions (e.g., sites of local anesthetic infiltration). In addition, the usual care interventions were not well standardized or defined. Pain and function were the most commonly reported outcomes, with limited evidence on quality of life, health status (e.g., Short-Form 36 Health Survey [SF-36]), mood, analgesic (including opioid) use, and other outcomes. Data on harms were relatively sparse and inconsistently reported. The trials were not designed to evaluate how benefits and harms varied in subgroups defined by demographic, clinical, or technical factors. Data on long-term (≥1 year) outcomes was relatively limited.

For the other interventional procedures evaluated in this report, the major limitation was the small numbers of trials, with important methodological shortcomings (e.g., high attrition, lack of intent-to-treat analysis, baseline group differences, small sample sizes, inadequate or unclear randomization or allocation concealment methods, open-label design, and use of unvalidated outcome measures) in almost all eligible studies.

Implications and Conclusions

Vertebroplasty is probably effective at reducing pain and improving function in older patients with vertebral compression fractures, but benefits were small (Table A). Effects of vertebroplasty were reduced in sham versus usual care controlled trials and larger in trials of patients with more acute symptoms. However, it is not possible to attribute differences entirely to the control type used, given substantial other differences across trials with regard to duration of pain, PMMA volume, requirement for bone edema on magnetic resonance imaging (MRI), and other factors. Furthermore, there were not statistically significant interactions between control type and effects on pain intensity at other time points, there was heterogeneity among the sham-controlled trials, and there is controversy regarding potential therapeutic effects associated with different sham procedures. To address outstanding questions regarding vertebroplasty, future trials should ideally include sham as well as usual care control groups and include patients with hyperacute (e.g., <3 weeks) and acute (e.g., 3 to 6 weeks) symptoms. Trials that include sham interventions with and without periosteal local anesthetic could also help clarify whether the sham treatment itself is associated with therapeutic benefits. Kyphoplasty is probably more effective than usual care for vertebral compression fractures in older patients (Table A). However, an important limitation of the evidence is the absence of sham-controlled trials of kyphoplasty. Until such evidence becomes available, kyphoplasty may be considered as an alternative to vertebroplasty, particularly in patients with more vertebral body collapse, as the purpose of kyphoplasty is to help restore vertebral body morphology.

Cooled radiofrequency denervation is probably more effective than sham denervation for sacroiliac pain, cooled radiofrequency may be as effective as conventional radiofrequency for presumed facet joint pain, occipital nerve stimulation may be more effective than usual care for headache, and piriformis corticosteroid injection may be more effective than sham for piriformis syndrome (Table A). Evidence on harms was limited, but lead migration was common following occipital nerve stimulation placement. Although evidence on these interventions was limited to younger (below the age for Medicare eligibility) populations, there is no obvious reason that findings would not apply to older patients. Evidence on the other interventions and conditions addressed in this review is sparse and insufficient, and additional research is needed to determine benefits of harms (Table A). To ideally inform Medicare coverage decisions, future trials of interventional procedures should enroll older, Medicare-eligible populations, utilize sham controls, evaluate function as well as pain, include rigorous evaluation of harms, evaluate longer-term outcomes, and evaluate how benefits and harms according to demographic (age, sex, race/ethnicity), clinical (pain severity, pain duration, use of opioids, psychiatric or medical comorbidities), or technical (dose, intensity, duration, frequency, techniques) factors.

Table A

Interventional pain therapies for acute and chronic pain.