Predictors of outcomes following lumbar intra-articular facet joint injections and medial branch blocks: A scoping review

Abstract

Study design:

A scoping review.

Objective:

We aimed to identify and describe the factors associated with the patient-reported response following lumbar intra-articular facet joint injections or medial branch blocks.

Summary of Background Data:

Facet joint osteoarthritis is among the most common causes of chronic low back pain. Management often includes facet joint intra-articular injection and medial branch blocks (which may be followed by radiofrequency ablation of the nerves innervating these joints). However, the success of these approaches is variable, prompting interest in identifying patient characteristics (imaging features, clinical signs, among others) associated with response to these types of facet injections.

Methods:

We performed a literature search on factors associated with patient-reported outcomes following lumbar facet joint intra-articular injections or medial branch blocks for patients with low back pain published in English or Spanish between 2000 and 2023. We excluded duplicates, papers that did not describe factors associated with outcomes, or those describing other interventions. We collected data on the association of these factors with patient-reported outcomes.

Results:

Thirty-seven studies met the inclusion criteria and were analyzed. These studies evaluated factors such as age, depression, and single-photon emission computed tomography (SPECT), among variables. Age and imaging findings of facet arthropathy were the most frequently described factors. Imaging findings of facet joint arthropathy and positive SPECT were often associated with positive results following intra-articular facet joint injections or medial branch blocks. In contrast, younger age and smoking were frequently associated with less favorable clinical outcomes.

Conclusion:

A myriad of factors were analyzed in the thirty-seven studies included in this review. Imaging findings of facet arthropathy, duration of pain, and positive SPECT were consistently associated with favorable results following facet interventions. Future prospective studies should include control groups. Clinicians may find these results useful in deciding whether to recommend facet injections.

INTRODUCTION

Low back pain is the leading cause of disability worldwide.¹ Nearly 80% of adults experience low back pain in their lifetime.² Low back pain is also the most common type of musculoskeletal pain in older adults leading to functional disability.^3–4 The total costs of chronic low back pain in the United States exceed $100 billion annually.⁵ Pathologic features in persons with low back pain may include disc degeneration, facet arthropathy, annular tears, disc herniations, stenosis of the central canal, lateral recess, or neural foramen, degenerative scoliosis, and spondylolisthesis. Among these morphologic features, facet arthropathy is the most frequently observed finding in patients with axial, non-radicular back pain.⁶

Facet joints (FJs), also called zygapophyseal joints, are synovial articulations in the spine that commonly degenerate as people age. While facet arthropathy is observed in around 60% of patients with low back pain⁸, it is also widely observed in asymptomatic persons. For example, Kim et al. reported that facet arthropathy is demonstrable by computed tomography (CT) imaging in 37% of asymptomatic individuals.⁷ Because imaging evidence may occur without symptoms, the diagnosis of facet-mediated pain is often established with an anesthetic block of the nerves innervating the FJs or intra-articular corticosteroid injection.⁹

The first-line management of FJs pain is usually conservative, including noninvasive measures such as medications, weight loss, and physical therapy.¹⁰ If conservative treatment fails, these patients may benefit from an intra-articular corticosteroid facet joint injection (FJI) or an anesthetic block of the medial branch nerve innervating the facets.¹¹ Medial branch blocks (MBBs) are associated with a higher success rate than intra-articular injections when used as a diagnostic tool before FJ radiofrequency ablation, which may serve as definitive treatment.¹² However, the success of these approaches is variable^13–15, prompting interest in identifying patient characteristics associated with response to FJI or MBBs. No published study has reviewed pertinent literature to identify and compare patient-related factors associated with the outcomes following intra-articular FJI and MBBs.

This study summarizes the literature on factors associated with outcomes of FJ interventions, including intra-articular FJI and MBBs. We aim to describe the most frequently assessed factors associated with patient-reported outcomes following these interventions.

MATERIALS AND METHODS

Search process

We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist.¹⁶ We searched PubMed in October 2022 for articles about factors predictive of outcomes following lumbar FJIs or MBBs published between January 2000 and October 2022. The following Boolean search terms were used: (facet joint) AND ((injection) OR (block)) AND ((predictors) OR (factors) OR (outcomes)). We removed duplicates and studies published before 2000. We updated the search in April 2023.

We excluded reviews, editorials, lectures, case studies, protocols, studies written in languages other than English or Spanish, animal studies, abstracts, publications unrelated to the spine (e.g., knee, shoulder), and retracted papers. We also excluded studies assessing injections performed on FJs from cervical or thoracic levels, publications that did not describe the outcomes following interventions, studies that did not describe predictive factors of lumbar FJJ or MBB, and those assessing response following interventions other than FJI or MBBs (e.g., radiofrequency ablation). One of the authors (JA) screened studies by titles and abstracts in the first screening stage and then by full manuscript to exclude studies that did not meet the abovementioned criteria. When the categorization was unclear, the senior author (JNK) and JA discussed and agreed on its classification. The references of articles included in the analysis were also screened and reviewed using the inclusion and exclusion criteria.

One of the authors (JA) evaluated the included studies using the Van den Heuvel checklist to appraise the sources of evidence critically.¹⁷ The Van den Heuvel bias assessment tool is a checklist of 11 items¹⁷, based on the Downs and Black checklist¹⁸ and additional items from the literature.^19–20 Each item is scored with a “1”, which represents a “yes;” “0,” which means a “no;” and “N/A,” which represents “not applicable to the study.” Each “yes” counts as one point, resulting in a maximum quality score of 11. A quality score of 0 to 6 points is defined as a high risk of bias, and 7 to 11 points as a low risk of bias.¹⁷

Data Extraction and Analysis

For each study included in the analysis, we abstracted data on year of publication, authors, digital object identifier (DOI), type of study, sample size, levels injected, comparator group (if relevant), the reason for intervention, and predictive factors assessed. We also abstracted the follow-up length, the primary outcome measure, the binary indicator for treatment failure (if provided), and the statistical tests used to assess the association of the predictive factor and outcome. The outcomes we reported included patient-reported pain and functional improvement.

RESULTS

Search Results

Figure 1 shows the data extraction results at every screening stage. First, we retrieved 575 titles from our PubMed search. We excluded one duplicate study, leaving 574 studies. Next, we excluded 536 that met at least one of our exclusion criteria, leaving 38 studies assessed by reading the complete manuscripts. Of these, six were excluded, leaving 32 included studies. Finally, we examined the references from these 32 studies and identified 16 additional studies. After eliminating those that met at least one exclusion criterion, 37 studies were left that described factors associated with outcomes following lumbar FJI or MBB.^21–57 We focused our detailed analysis on these 37 studies.

Figure 1. PRISMA-ScR Flow Diagram¹⁶.

Figure 1 represents the criteria for excluding papers, the distribution of studies excluded at each screening stage, and the dissection of the papers included and excluded from our review.

No legend is required for Figure 1.

Overview of the selected studies

Table 1 of the Supplemental Data File describes the general aspects of the included studies. The first study was published in 2003.²¹ Sixteen studies were prospective^{21–29,31–33,48–49,52,57}, twenty were retrospective^{30,34–46, 50–51,53–56}, and one was a randomized controlled trial.⁴⁷ The sample size of the included studies ranged from 12 to 3900 subjects, all with a primary diagnosis of chronic low back pain. Three studies also included subjects with radiculopathy.^34–36 The duration of follow-up for the included studies ranged from one day to one year after the intervention. Twelve studies specified the injected spine levels.^{21,23,28,31,33–35,39–40,43–44,56} Twenty-six studies evaluated factors associated with the response following FJI^21–47, while eleven investigated the response following MBBs.^46;48–57

Bias assessment of selected studies

The Van den Heuvel checklist results are shown in Table 2 of the Supplemental Data File. According to this checklist, all 37 studies had a low risk of bias. The methodology in the included studies was described clearly, except for two studies, which failed to provide adequate descriptions of the selected patients.^27,46 All studies provided estimates of the random variability in the data (e.g., standard deviation). Each study used a consistent method for collecting information from all the subjects, e.g., the same type of magnetic resonance imaging (MRI) scanner and sequences. All included studies, except one ²¹, described and selected a representative study sample, in terms of age and sex, of the population from which they were recruited. The person scoring the MRI scans was adequately trained to do so in eleven of fifteen studies that included MRI readings in their methodology.^{21,32,34–37,40,42,51,53,56} The sample sizes for 11 of the 37 studies were less than 50 subjects.^{21,23,30,31–33,37,42,47,52,55} Each study was assigned a total score based on the number of acceptable items. The highest possible score was 10 or 11, depending on whether the study included MRI scans. Five studies earned a total score of 11 ^{34,40,51,53,56}, 22 earned a score of 10 ^{22,24–26,28–29,32,34–39,41–45,48–50,54,57}, eight earned a score of 9 ^{23,27,30,31,46–47,52,55}, and two earned a score of 8.^21,33

Outcome measures

Pain and functional outcomes were measured using various standardized scales. Fifteen studies used a Visual Analog Scale (VAS) ^{21,23,28–32,43–50,52,54–55} to assess pain; nine used the Numerical Rating Pain Score (NRS)^{24,25,27,37–39,51,56}; seven used the Oswestry Disability Index (ODI)^{30–31,40–41,45,52,54}, which assesses pain and functional status; three used the Patient Global Impression of Change (DGIC)^26–27,42; two used the MacNab criteria^23,31; two used a verbal numerical scale^33,41; and one used the American Academy of Orthopedic Surgeons Musculoskeletal Outcomes Data Evaluation and Management System (AAOS MODEMS) score^22,58, which assesses pain.

Predictive factor assessment

The 37 included studies analyzed a large variety of factors. We classified each factor into four categories: socio-demographic and clinical history, physical exam-related, intervention-related, and imaging-related factors. Table 1 summarizes the number of studies that examined each factor. The socio-demographic and clinical factors included age ^{23,31–32,34–36,38,41,44,51,53,56}, sex ^{24,31,32,34–36,38,41,46,53,56}, and others. The physical exam-related factors included positive provocation tests⁴⁹, among other factors. Intervention-related factors included levels injected^31,38,53, and dose of corticosteroid used^42,43, among others. Lastly, imaging-related factors included FJ arthropathy^{21,31–32,34,36–37,39,43–45,51,53,55–56}, spondylolisthesis^31,34,36,56, and others.

Table 1. Studies that examined each factor.

Table 1 summarizes the number of studies that examined each factor.

SOCIO-DEMOGRAPHIC AND CLINICAL HISTORY FACTORS
Factor	Number of studies
Age	13
Sex	12
History and current pain levels €	10
Body Mass Index (BMI)	6
Smoking	5
Psychiatric medical conditions (e.g., depression, anxiety, or somatization disorder)	4
Prior back surgery	4
History of symptomatology ∑	3
Fibromyalgia	3
Previous opioid use	3
Comorbid conditions (e.g., Hypertension or Diabetes Mellitus 2)	2
Education	2
Weight	1
Height	1
Job Ϡ	1
Sedentary lifestyle π	1
Physical therapy before injection	1
Medication Quantification Scale	1
General health	1
Clinical prediction rules ∞	1
Type of follow-up intervention (telephone versus mailing)	1
Therapisťs attitude	1
TOTAL	75
PHYSICAL EXAM -RELATED FACTORS
Factor	Number of studies
Positive provocation tests §	1
Feeling of joint blocking	1
Pain alleviated by motion	1
Lumbar muscle weakness	1
Supraspinal ligament stretch	1
TOTAL	5
INTERVENTION-RELATED FACTORS
Factor	Number of studies
Levels injected (mean and total)	3
Dose of corticosteroid used	2
Percentage of pain relief from diagnostic block	1
Need for pain-killer medications after facet joint injection	1
Epidural versus facet steroid injections ϕ	1
Epidural contrast spread	1
Additional epidural injection	1
TOTAL	10
IMAGING–RELATED FACTORS
Factor	Number of studies
Facet joint arthropathy	13
Spinal stenosis	4
Singe Photon Emission Computed Tomography	4
Spondylolisthesis	4
Degenerative disc disease	4
Modic changes	2
Type of intervention (type of imaging used)	2
Anatomical defects noted during the intervention ∂	2
Facet synovial cyst (present/absent)	2
Spine deformity	2
Annular tear	1
Disc herniation	1
Neural foraminal stenosis (grade)	1
Sagittal alignment	1
Bone marrow edema (present/absent)	1
Partial Arthrogram (distribution of injected material)	1
Spinopelvic parameters and lumbar intervertebral angles (pelvic incidence, sacral slope, and pelvic tilt)	1
Facet tropism	1
TOTAL	47

^€= Factors in this category include duration of pain, baseline pain scores, and catastrophizing pain scale scores.

^∑= Factors in this category include location of pain, presence of claudication or radiculopathy, and laterality of pain.

^Ϡ= Factors in this category include manual work, active duty, work-related injury, disability, and worker’s compensation.

^π= They defined a sedentary lifestyle as less than 450 metabolic equivalents per week.³¹

^∞=Clinical prediction rules include age over 50 years old, pain is greatest when walking, pain is greatest when sitting, onset of pain is paraspinal, MSPQ scores exceedingly 13 points, extension/rotation pain, and absence of centralization during repeated movement testing.²⁹

^§=This study evaluated Revel’s criteria, which includes four physical examination components: standing flexion, returning from standing flexion, standing extension, and the extension rotation test.⁴⁹

^ϕ= This study assessed the efficacy of facet steroid injection versus epidural steroid injection for patients with lumbar spinal stenosis.³⁶

^∂= These defects include communication through the pars interarticularis (vertically)³⁴, via the retro-dural space through which contrast can be traced to the contralateral joint³⁴, or abnormal arthrogram with bony defects in the pars interarticularis.²⁸

We identified factors assessed by at least three independent studies and summarized the number of studies examining each factor and those that found a significant association between the factor and outcomes in Table 2. Figure 2 summarizes the association between the most frequently evaluated factors and the response direction following FJ intervention.

Table 2.

The most frequently assessed factors per intervention

	Intra-articular corticosteroid injections		Medial branch blocks		Total
Factors	Number of studies that examined that factor	Number of studies that found a statistically significant association between the factor and outcome (p < 0.05)	Number of studies that examined that factor	Number of studies that found a statistically significant association between the factor and outcome (p < 0.05)	Number of studies that examined that factor	Number of studies that found a statistically significant association between the factor and outcome (p < 0.05)
Facet joint arthropathy	9	2	4	2	13	4
Age	9	0	4	2	13	2
Sex	9	0	3	0	12	0
Duration of pain	8	1	1	1	9	2
Body Mass Index	4	1	2	0	6	1
Smoking	3	1	2	0	5	1
Prior back surgery	4	1	0	0	4	1
Spinal stenosis	3	1	1	0	4	1
Positive Single Photon Emission Computed Tomography	3	3	1	1	4	4
Degenerative disc disease	2	1	2	0	4	1
Spondylolisthesis	3	0	1	0	4	0
Depression	1	0	3	1	4	1
Levels injected	2	0	1	1	3	1
Opioid use	2	1	1	1	3	2
Fibromyalgia	2	0	1	0	3	0
Anxiety	1	0	2	1	3	1

Figure 2. The most frequently assessed factors per intervention.

We identified factors assessed by at least three independent studies and summarized the number of studies examining each factor and those that found a significant association between the factor and outcomes in Table 2.

Red = The study found a statistically significant association between the factor and worse outcomes following injection.

Green = The study found a statistically significant association between the factor and better outcomes following injection.

Gray = The study found no statistically significant association between the factor and outcomes following injection.

White = The study did not evaluate the factor.

MBB= The study evaluated factors associated with outcomes following medial branch blocks.

INJ= The study evaluated factors associated with outcomes following intra-articular facet joint injections.

** = They found a statistically significant difference for diagnosis of FJ OA, but no difference between the grades of severity of FJ OA.³⁹

*** = Positive SPECT is associated with positive outcomes if they had failed a prior FJ Injection.⁴³

$ = This study found that facet arthropathy in the setting of other spine pathologies (i.e., spondylolisthesis grades one and two, multiple discopathy, scoliosis, and osteoarthritis of the hip) is associated with better outcomes following FJ injection.⁵⁴

∞ = Patients who underwent bone scanning with Single Photon Emission Computed Tomography and positive facet joint abnormalities, such as facet joint hypertrophy, subchondral sclerosis, and joint space narrowing, in the scan responded better than the comparison group, defined as subjects with positive Single Photon Emission Computed Tomography, but no facet joint abnormalities noted on imaging, and subjects with had a negative Single Photon Emission Computed Tomography.²²

ββ= This study found no association between duration and outcomes following facet joint injection. However, they found that a higher baseline pain, measured by the catastrophizing pain scale, is associated with worse outcomes following injection.³¹

αα= This study found no association between duration and outcomes following facet joint injection. However, they found that a higher baseline pain, measured by the numerical rating scale pain score, is associated with worse outcomes following injection.³⁸

Ƒ=This study found that pain alleviated by motion was a positive predictor of good immediate response, and such pain was also significantly associated with higher grades of facet joint osteoarthritis.⁵⁵

ΩΩ= This study assessed the number and the laterality of the medial branch blocks performed and found that unilateral medial branch block is associated with prolonged pain relief (i.e., positive response).⁵³

A. Socio-demographic, clinical, and physical exam-related factors

Two of the thirteen studies evaluating age as a predictor found a significant association between younger age and less favorable clinical outcomes following MBB.^48,51 None of the twelve studies that assessed sex as a predictor found a significant association between sex and outcomes following FJ intervention. Body mass index (BMI) was examined in six studies.^{23,31,44,46,53} One study identified a significant association between higher BMI and better outcomes, defined as a decrease in at least two or more points on the VAS pain scale.⁴⁶ Two of the five studies assessing smoking as a predictor found a significant association with worse outcomes following the intervention.^31,55 Four studies analyzed prior back surgery as a predictor^,31,38,46, and one found a significant association between prior back surgery and worse outcomes following FJI.³⁸ Two of the ten studies evaluating duration of pain as a predictor found an association between longer pain duration and positive responses following FJI or MBBs. These studies defined positive response as a decrease of at least two or more points in their VAS pain score⁴⁶ and as protracted relief, defined as continued pain relief occurring at least four weeks after the intervention.⁵³ Two of three studies evaluating opioid use as a predictor found a significant association with worse outcomes following FJI and MBB^38,54 None of the three studies investigating fibromyalgia found a significant association with any outcome.^25,31,53 One of four studies evaluating depression as a predictor found an association with worse outcomes following MBB.⁵⁷ Lastly, one of the three studies assessing anxiety as a predictor found an association with worse outcomes following MBB.⁵⁷

B. Intervention and spine imaging-related factors

One of the three studies that assessed the number of levels injected and the laterality of the injections found a statistically significant association with the response following FJ interventions.⁵³ In this study, a unilateral block was associated with better responses to MBB.⁵³ Positive uptake on SPECT was associated with positive outcomes in all four studies evaluating the association between SPECT uptake and response to FJI or MBBs.^22,30,43,52 Spondylolisthesis, diagnosed by radiography, was assessed in four studies, and none found a statistically significant association with outcome.^31,34,36 Only one of the four studies evaluating spinal stenosis found a statistically significant association between severe spinal stenosis and better outcomes following FJI.^30,31,36,51 Facet arthropathy was assessed in thirteen studies. Four found a statistically significant association between more severe facet arthropathy and better outcomes, defined as a greater than fifty percent reduction in functional disability and pain in two.^{21,51,54–55} The studies described facet arthropathy as increased signal intensity changes on MRI²¹, facet hypertrophy⁵⁴, facet osteoarthritis as noted on MRI and radiography,⁵⁵ and facet degeneration, defined as narrowing of the FJ space, the presence of osteophytes, subarticular bone erosions, or subchondral cysts on MRI⁵¹. Lastly, one of the four studies evaluating degenerative disc disease found an association with a positive response following MBB.⁵⁶

DISCUSSION

FJ interventions are performed frequently, including FJI and MBBs. Still, research on predictors of these interventions has not been summarized previously, making it difficult for clinicians to counsel patients about the likelihood that they will respond positively. The purpose of this study was to summarize the literature on the factors associated with response to FJ interventions, including FJI and MBBs. We reviewed 37 articles that quantitatively examined the association between predictive factors and outcomes following these FJ interventions.^21–57

We found that younger age, smoking, prior back surgery, depression, and anxiety were associated with worse outcomes following interventions. Notably, younger age was a consistent negative predictor of response following MBB, whereas smoking was associated with worse outcomes independent of the type of intervention. Our findings are similar to outcomes seen in studies assessing outcomes after hip or knee injections. For example, one study found an association between older age and longer pain relief following hip intra-articular injection.⁵⁹ Another study found that age is associated with worse outcomes following knee intra-articular steroid injection.⁶⁰ Studies evaluating facet arthropathies, such as facet synovitis and FJ hypertrophy, as predictors of response were frequently associated with positive outcomes following FJI or MBB.^21,45,51,55 SPECT uptake, particularly ^99mTc-SPECT, imaging, was associated with positive outcomes in the four studies that examined this relationship (three evaluating intra-articular injections and one evaluating MBBs).^22,30,43,52

Prior studies have shown that SPECT imaging facilitates anatomical visualization and that increased uptake could predict the outcome of imaging-guided anesthetic injections.^61–62 The increased activity on SPECT is also predictive of the associated back pain.^63–64 However, SPECT use has decreased in the last decade due to improved Short Tau Inversion Recovery (STIR) and other fat-suppressing MR modalities. Recent studies have found good correlations between STIR MRI and SPECT findings.^65–66 STIR MR findings are also frequently noted in patients with axial low back pain.^67–68 Thus, further work is needed to clarify if bone marrow or soft tissue edema findings on STIR MRI sequences are associated with outcomes following FJI. If so, MRI STIR sequences might serve as a surrogate for SPECT use in modern clinical practice.

In summary, the associations between the risk factors and outcomes following FJI and MBBs were generally similar. Age and imaging findings of facet arthropathy were the most frequently assessed factors. The most consistent positive predictors were SPECT uptake and imaging findings of facet arthropathy. Conversely, younger age and smoking were frequently associated with less favorable responses to MBBs.

We note several limitations in our review. First, the substantial variability in outcomes assessed and methods used to evaluate associations between predictive factors and outcomes precluded a formal meta-analysis. Further, variability in outcome measures and analytic methods precluded characterizing these associations with effect measures, such as odds or risk ratios. Thus, we used a statistically significant association (p < 0.05) to designate an association as meaningful. Some statistically significant associations are not clinically meaningful in studies with large samples. In smaller studies, some effects that fail to reach statistical significance may be clinically meaningful. However, the outcomes and analytic approaches used were too heterogenous to classify studies based on the effect measures. In addition, only eight of the included studies used multivariate analyses to identify independent predictors of failure^{11, 29, 33, 37, 40, 45–46, 51}. Thus, further work is needed to clarify the independent contributions of these factors to outcomes.

Additionally, there was variability among the study designs. For example, 16 of the 38 studies included had a retrospective design^40–56, and 11 studies examined a sample of fewer than 50 subjects^{21, 24, 33, 37–40, 44, 50, 56, 57}, limiting the power of each study to demonstrate statistically significant differences. Further, the included studies examined two types of interventions: FJI and MBBs, hindering comparing outcomes between different interventions. Also, we cannot comment on clinically relevant factors other than those described in these studies, such as CT scan findings, since none of the studies evaluated CT imaging findings as prognostic factors. Lastly, most of the included studies failed to include control groups. Thus, while clinicians can use these findings to predict which patients undergoing facet joint interventions will likely respond to therapy, the findings cannot predict whether a patient’s outcome would be better following intervention than if they did not pursue the intervention. Future studies should focus on prospective methods that include control groups to evaluate how patient characteristics affect pain relief following injection and non-invasive treatment.

Despite these shortcomings in the literature, the included studies assessed a wide range of factors, enabling us to identify numerous factors potentially predictive of response to FJ interventions. Understanding these factors may help advance the field by contributing to precision medicine and guiding decision-making.

KEY POINTS.

• We found thirty-seven studies that describe factors associated with the response following intra-articular facet joint corticosteroid injections or MBBs.

• Of all the factors assessed in our review, age and imaging findings of facet arthropathy are the most frequently described in the literature.

• Imaging findings of facet joint arthropathy and positive SPECT are the most consistent predictor of positive outcomes following facet joint intra-articular corticosteroid injections or MBBs.

• Conversely, younger age and smoking were frequently associated with poor results following MBBs and intra-articular corticosteroid injections, respectively.

Data access:

No databases were used in this review other than PubMed.