Do Epidural Injections Provide Short- and Long-term Relief for Lumbar Disc Herniation? A Systematic Review

Abstract

Background

As part of a comprehensive nonsurgical approach, epidural injections often are used in the management of lumbar disc herniation. Recent guidelines and systematic reviews have reached different conclusions about the efficacy of epidural injections in managing lumbar disc herniation.

Questions/purposes

In this systematic review, we determined the efficacy (pain relief and functional improvement) of the three anatomic approaches (caudal, lumbar interlaminar, and transforaminal) for epidural injections in the treatment of disc herniation.

Methods

We performed a literature search from 1966 to June 2013 in PubMed, Cochrane library, US National Guideline Clearinghouse, previous systematic reviews, and cross-references for trials studying all types of epidural injections in managing chronic or chronic and subacute lumbar disc herniation. We wanted only randomized controlled trials (RCTs) (either placebo or active controlled) to be included in our analysis, and 66 studies found in our search fulfilled these criteria. We then assessed the methodologic quality of these 66 studies using the Cochrane review criteria for RCTs. Thirty-nine studies were excluded, leaving 23 RCTs of high and moderate methodologic quality for analysis. Evidence for the efficacy of all three approaches for epidural injection under fluoroscopy was strong for short-term (< 6 months) and moderate for long-term (≥ 6 months) based on the Cochrane rating system with five levels of evidence (best evidence synthesis), with strong evidence denoting consistent findings among multiple high-quality RCTs and moderate evidence denoting consistent findings among multiple low-quality RCTs or one high-quality RCT. The primary outcome measure was pain relief, defined as at least 50% improvement in pain or 3-point improvement in pain scores in at least 50% of the patients. The secondary outcome measure was functional improvement, defined as 50% reduction in disability or 30% reduction in the disability scores.

Results

Based on strong evidence for short-term efficacy from multiple high-quality trials and moderate evidence for long-term efficacy from at least one high quality trial, we found that fluoroscopic caudal, lumbar interlaminar, and transforaminal epidural injections were efficacious at managing lumbar disc herniation in terms of pain relief and functional improvement.

Conclusions

The available evidence suggests that epidural injections performed under fluoroscopy by trained physicians offer improvement in pain and function in well-selected patients with lumbar disc herniation.

Electronic supplementary material

The online version of this article (doi:10.1007/s11999-014-3490-4) contains supplementary material, which is available to authorized users.

Introduction

Epidural injections are one of the most common nonsurgical treatments for lumbar disc herniation [8, 22, 69, 71, 100, 103, 119]. Epidural injections are administered by accessing the lumbar epidural space by multiple routes—caudal, transforaminal, and interlaminar [1, 7, 8, 17, 21, 22, 69, 71, 100, 103, 119]. Even though all three modalities deliver medication into the epidural space, there are important differences among the approaches. The interlaminar entry is considered to deliver the medication close to the assumed site of pathology, but the transforaminal approach is considered the target-specific modality requiring the smallest volume to reach the primary site of pathology. In contrast, caudal epidurals are considered the safest and easiest, with minimal risk of inadvertent dural puncture, and are the preferred modality in postsurgery syndrome, even though they require relatively high volumes and have an alleged lack of specificity to the assumed site of pathology. Consequently, therapeutic effectiveness also varies among the three approaches [8, 69, 71, 100].

Numerous systematic reviews and guidelines performed by various groups of authors have reached different conclusions about the level of evidence for the effectiveness of epidural injections [5, 7, 8, 21–23, 31, 66, 69, 71, 79, 100, 103, 105, 106, 109, 113, 119]. Some authors have concluded against the effectiveness and medical necessity of epidural injections in managing pain and improving function in patients with disc herniation and radiculitis [5, 21–23, 103, 119]; however, these have been challenged [69, 76], with multiple trials showing significant improvement in pain and function [7, 8, 31, 66, 69, 71, 100, 106, 109, 113]. Some of these systematic reviews and the trials included were criticized for flaws in the assessment of trials, combining trials with variable designs, and designating active controlled trials as placebo control [22, 35, 45, 69, 70, 103, 119]. A recent systematic review by Pinto et al. [103] analyzed all epidural injection approaches compared with a placebo in patients with sciatica, offered criticism of the past reviews and guidelines [7, 8, 70, 72, 100], and criticized some reviews [5, 21, 23, 31] for summarizing the evidence from placebo-controlled trials without considering the size of the treatment effect. Pinto et al. [103] have concluded that there was high-quality evidence showing that epidural steroid injections have small, short-term effects on leg pain and disability compared with placebo in patients with sciatica but no effect in the long term. However, this systematic review by Pinto et al. [103] also had multiple deficiencies [65, 67]. Since the systematic review by Pinto et al. [103] was based on a methodologic rather than a clinical perspective [78, 80], which we find challenging, our review focused on the clinical aspects with appropriate methodologic quality assessment.

Thus, in this systematic review, we determined the short- and long-term efficacy (pain relief and functional improvement) of the three anatomic approaches (caudal, lumbar interlaminar, and transforaminal) for epidural injections in the treatment of disc herniation, considering only placebo-controlled and active-controlled trials.

Materials and Methods

Search Strategy and Criteria

We performed a literature search from 1966 to June 2013 utilizing data from PubMed, Cochrane library, US National Guideline Clearinghouse, previous systematic reviews, and cross-references. The search strategy emphasized low-back and lower-extremity pain, disc herniation, and radiculitis treated with caudal, interlaminar, or transforaminal epidural injections. Search terms included chronic low-back pain, lower-extremity pain, disc herniation, nerve root compression, radiculitis, epidural injections, caudal, interlaminar, and transforaminal.

Study selection was achieved as recommended by Preferring Reporting Items for Systematic Reviews and Meta-Analyses [64] (Fig. 1). We reviewed all studies providing appropriate management and with outcome evaluations of 3 months or longer and statistical evaluations. Only randomized controlled trials (RCTs) were included for analysis, either placebo or active controlled. The true definition of placebo is to inject an inactive substance into an inactive structure; for the purposes of this review, we have utilized an injection of placebo into the epidural space or over the nerve root by any approach as placebo [48, 49, 54, 82]. The duration of symptoms of the study participants was classified as subacute (6–12 weeks) and chronic (> 12 weeks). For this evaluation, only the studies utilizing patients with chronic symptoms or patients with a mixture of subacute and chronic symptoms were considered. Studies including radiculitis secondary to spinal canal or foraminal stenosis, postsurgery syndrome, or disc herniation in postsurgery syndrome and chemical radiculitis were not included in this review. However, trials with the inclusion of multiple groups of patients were included if they had at least 25 patients with disc herniation in each group.

Fig. 1.

A flow diagram shows the study selection process.

Overall, there were 66 RCTs for consideration. As we wanted to include only RCTs of high or moderate quality for analysis, we assessed the methodologic quality of these 66 studies using the Cochrane review criteria for RCTs [35]. The criteria include assessment of adequate randomization, treatment allocation concealment, appropriate blinding, description of incomplete outcome data with dropouts and intent-to-treat analysis, study being free of suggestions of selective outcome reporting, and other sources of potential bias including baseline characteristics, cointerventions, compliance, and timing of the outcome measurements in all groups. The methodologic quality was graded as high if nine to 12 criteria were fulfilled, moderate if six to eight criteria were fulfilled, and low if zero to five criteria were fulfilled; trials fulfilling fewer than six criteria were excluded. Thirty-nine of the 66 trials were excluded due to multiple reasons [2, 11–17, 24, 27, 28, 30, 36–39, 41, 42, 47, 52, 53, 58, 60, 61, 68, 92–94, 98, 99, 108, 110, 114, 116, 118, 122, 124, 126, 130] (Appendix 1; supplemental materials are available with the online version of CORR^®). Twenty-three trials had high or moderate methodologic quality and were included in our analysis [1, 3, 4, 18, 25, 29, 32, 40, 50, 51, 55–57, 63, 85, 86, 89, 91, 95–97, 104, 107, 111, 112, 121, 128], with four duplicate publications [55 and 56, 85 and 86, 89 and 91, 111 and 112]. Fifteen of these studies were high quality and eight were moderate quality. Ten of the 16 fluoroscopic trials (Table 1) and four of the seven nonfluoroscopic trials (Table 2) were of high quality.

Table 1.

Methodologic quality assessment of fluoroscopic randomized trials.

Study	Randomization adequate	Concealed treatment allocation	Patient blinded	Care provider blinded	Outcome assessor blinded	Dropout rate described	All randomized participants analyzed in the group	Reports of the study free of suggestion of selective outcome reporting	Groups similar at baseline regarding most important prognostic indicators	Cointerventions avoided or similar	Compliance acceptable in all group	Time of outcome assessment in all groups similar	Score
Manchikanti et al. [85, 86]	Y	Y	Y	N	N	Y	Y	Y	Y	Y	Y	Y	10/12
Ackerman and Ahmad [1]	N	N	N	N	N	Y	Y	Y	Y	Y	Y	Y	7/12
Dashfield et al. [29]	Y	Y	Y	N	N	Y	Y	Y	Y	N	Y	Y	9/12
Manchikanti et al. [89, 91]	Y	Y	Y	Y	N	Y	Y	Y	N	Y	Y	Y	10/12
Lee et al. [63]	Y	N	N	N	N	Y	N	Y	Y	Y	Y	Y	7/12
Rados et al. [107]	Y	N	N	N	N	Y	Y	Y	Y	Y	Y	Y	8/12
Kim and Brown [57]	Y	Y	Y	N	N	Y	N	Y	Y	Y	Y	Y	9/12
Amr [3]	Y	Y	Y	Y	N	Y	N	Y	Y	Y	Y	Y	10/12
Ghahreman et al. [40]	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	11/12
Karppinen et al. [55, 56]	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	11/12
Riew et al. [111, 112]	U	U	Y	N	Y	N	Y	Y	Y	Y	Y	Y	8/12
Jeong et al [51]	U	U	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	9/12
Park et al. [97]	N	U	Y	N	U	Y	Y	Y	Y	Y	Y	Y	8/12
Tafazal et al. [121]	Y	Y	Y	N	N	Y	Y	Y	Y	Y	Y	Y	10/12
Cohen et al. [25]	Y	Y	Y	Y	Y	N	N	Y	N	Y	N	Y	8/12
Ng et al. [96]	Y	Y	Y	N	Y	Y	Y	Y	N	Y	Y	Y	10/12

Y = yes; N = no; U = unclear.

Table 2.

Methodologic quality assessment of randomized trials without fluoroscopy.

Study	Randomization adequate	Concealed treatment allocation	Patient blinded	Care provider blinded	Outcome assessor blinded	Drop-out rate described	All randomized participants analyzed in the group	Reports of the study free of suggestion of selective outcome reporting	Groups similar at baseline regarding most important prognostic indicators	Co-interventions avoided or similar	Compliance acceptable in all group	Time of outcome assessment in all groups similar	Score
Iversen et al. [50]	Y	Y	U	N	U	Y	N	Y	N	Y	N	Y	6/12
Murakibhavi and Khemka [95]	Y	Y	N	N	U	Y	Y	U	Y	N	Y	Y	7/12
Dilke et al. [32]	U	U	Y	N	Y	Y	U	Y	Y	Y	Y	Y	8/12
Pirbudak et al. [104]	Y	Y	Y	Y	U	Y	U	Y	Y	Y	Y	Y	10/12
Arden et al. [4]	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	11/12
Carette et al. [18]	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	11/12
Wilson-MacDonald et al. [128]	Y	Y	Y	N	Y	Y	Y	Y	N	Y	Y	Y	10/12

Y = yes; N = no; U = unclear.

We graded the evidence for each approach using five levels of best evidence synthesis, which summarized the results according to the Cochrane rating system based on the quality and the outcome of the studies [119]. The levels of evidence were distinguished as follows: (1) strong evidence: consistent findings among multiple high-quality RCTs; (2) moderate evidence: consistent findings among multiple low-quality RCTs or one high-quality RCT; (3) limited evidence: one low-quality RCT; (4) conflicting evidence: inconsistent findings among multiple trials; and (5) no evidence: no RCTs. The evidence of all three approaches for epidural injections was strong for the short term and moderate for the long term. Further, the results were considered consistent if 60% of the trials showed results in the same direction.

At least two of the review authors (RMB, FJEF) independently, in an unblinded standardized manner, performed each search and methodologic quality assessment. The primary authors of manuscripts were not involved in the methodologic quality assessment. All searches were combined to obtain a unified strategy. Any disagreements between reviewers were resolved by a third author (JAH) and consensus.

Outcome Measures

The primary outcome measure was pain relief, defined as 50% or more reduction in pain or at least a 3-point decrease in pain scores in at least 50% of patients. The secondary outcome measure was functional improvement, defined as 50% reduction in disability or 30% reduction in the disability scores.

The injection therapy was considered effective, either with a placebo control or active control, if there was a difference in effect for the primary outcome measure in a statistically significant manner at the 5% level. Outcomes were considered at 3 months, 6 months, and 1 year and in the short term (< 6 months) and long term (≥ 6 months).

Meta-analysis was considered if more than two RCTs were homogeneous for mode of administration. However, we found that there was insufficient homogeneity for meta-analysis.

Results

Four of five total trials (three fluoroscopic, two nonfluoroscopic) and all three fluoroscopic trials showed efficacy of caudal epidural injections in managing lumbar disc herniation, with strong evidence in the short and moderate evidence for the long term (Table 3). The three fluoroscopic trials [1, 29, 86] included two high-quality trials [29, 86] and one moderate-quality trial showing mid-term efficacy [1]. Of the nonfluroroscopic trials, one moderate-quality nonfluoroscopic trial [95] showed efficacy and one moderate-quality ultrasound trial did not [50].

Table 3.

Description of study characteristics of randomized controlled epidural trials assessing the efficacy of caudal epidural injections in lumbar disc herniation.

Study	Study characteristics	Methodologic quality score	Number of patients	Interventions	Outcome measures	Pain and function improvement			Improvement			Comment(s)
						3 months	6 months	12 months	Short term (≤ 6 months)	Long term
										> 6 months	≥ 12 months
Manchikanti et al. [85, 86] (2012, 2011)	RA, AC, F	10/12	Total = 120	Lidocaine vs lidocaine mixed with steroid Number of injections = 1–5	NRS, ODI, employment status, opioid intake	62% vs 72%	72% vs 73%	67% vs 72%	Y	Y	Y	Double-blind study showing improvement
Ackerman and Ahmad [1] (2007)	RA, AC, F	7/12	Total = 90 Caudal = 30 IL = 30 TF = 30	Methylprednisolone + saline Number of injections = 1–3	Numeric pain score (0–10), rating of pain relief, ODI, BDI, contrast dispersion pattern	Caudal = 57% IL = 60% TF = 283%	Caudal = 57% IL = 60% TF = 83%	NA	Y	Y	NA	Study showing mid-term improvement
Dashfield et al. [29] (2205)	RA, AC, F	9/12	Total = 60 Caudal = 30 Endoscopy = 30	Lidocaine with triamcinolone Number of injections = 1	Pain relief, SF-MPQ, HADS scores	SI	SI	NA	Y	Y	NA	Study showing mid-term improvement
Iversen et al. [50] (2011)	RA, PC, UL	6/12	Total = 116	Saline or triamcinolone acetonide with saline Number of injections = 2	ODI, EQLS, VAS	N	N	N	U	U	U	Study with numerous deficiencies with flawed design and without local anesthetic showing lack of improvement
Murakibhavi and Khemka [95] (2011)	RA, AC, B	7/12	Group A = 50 control conservative management Group B = 52 caudal epidural steroid injections Total = 102	Conservative management or caudal epidural steroid injections	VAS, ODI, BDI, NPI	Group A = 32% Group B = 92%	Group A = 24% Group B = 86%	NA	Y	Y	NA	Moderate-quality study showing short-term improvement

RA = randomized; AC = active control; F = fluoroscopy; PC = placebo control; UL = ultrasound; B = blinded; IL = interlaminar; TF = transforaminal; NRS = Numeric Rating Scale; ODI = Oswestry Disability Index; BDI = Beck Depression Inventory; SF-MPQ = Short-Form McGill Pain Questionnaire; HADS = Hospital Anxiety and Depression Scale; EQLS = European Quality of Life Scale; NPI = Numerical Pain Intensity; SI = significant improvement; N = no; NA = not applicable; Y = yes; U = unclear.

Five fluoroscopic trials [1, 3, 63, 89, 107] showed efficacy of lumbar interlaminar epidural injections in managing lumbar disc herniation, with strong evidence for the short term and moderate evidence [3, 89] for the long term. Two of these trials were high quality [3, 89] and three moderate quality [1, 63, 107] (Table 4). Of the four nonfluoroscopic studies (all high quality), two showed efficacy [104, 128] and two did not [4, 18]. However, the level of evidence was strong and moderate based on fluoroscopic trials only, whereas a combination of fluoroscopic and nonfluoroscopic trials yielded moderate evidence, and when only nonfluoroscopic or blind trials were considered, the evidence was conflicting, with inconsistent findings among multiple trials.

Table 4.

Description of study characteristics of randomized controlled epidural trials assessing the efficacy of lumbar interlaminar epidural injections in lumbar disc herniation.

Study	Study characteristics	Methodologic quality score	Number of patients	Interventions	Outcome measures	Pain and function improvement			Improvement			Comment(s)
						3 months	6 months	12 months	Short term	Long term
									(≤ 6 months)	> 6 months	≥ 12 months
Manchikanti et al. [89, 91] (2012, 2010)	RA, AC, F	10/12	Total = 120 Local anesthetic = 60 Local anesthetic and steroids = 60	Xylocaine® or Xylocaine® with nonparticulate Celestone® Number of injections = 1–5	NRS, ODI, employment status, opioid intake, SI ≥ 50% of NRS scores and ODI scores	72% vs 82%	63% vs 85%	67% vs 85% or 80% vs 86% in successful group	Y	Y	Y	Study with long-term followup showing improvement
Lee et al. [63] (2009)	RA, AC, F	7/12	Total = 93 IL = 34 TF = 59	Lidocaine with triamcinolone Number of injections = 1–3	NRS, PSI	SI in both groups	SI in both groups	SI in both groups	Y	NA	NA	Study with short-term followup showing improvement
Rados et al. [107] (2011)	RA, AC, F	8/12	Total = 64 IL = 32 TF = 32	Lidocaine with methylprednisolone Number of injections = 1–3	VAS, ODI	53% vs 63%	53% vs 63%	NA	Y	Y	NA	Study with short followup showing improvement
Kimand Brown [57] (2011)	RA, AC, F	9/12	Total = 60 Methylprednisolone = 30 Dexamethasone = 30	Depo-medrol® or dexamethasone with bupivacaine Number of injections = 1–2	VAS, pain medication intake, emergency room visits	NA	NA	U	NA	NA	NA	Relatively small study with undetermined results
Amr [3] (2011)	RA, AC, F	10/12	Total = 200 Steroid = 100 Steroid + ketamine = 100	Triamcinolone plus preservative-free ketamine and 0.9% saline Number of injections = 1	Pain scores, Oswestry Low Back Pain Disability Questionnaire	SI in ketamine group	SI in ketamine group	SI in ketamine group	Y	Y	Y	Study with long-term followup showing improvement
Ackerman and Ahmad [1] (2007)	RA, AC, F	7/12	Total = 90 Caudal = 30 IL = 30 TF = 30	Steroid and saline with local anesthetic Number of injections = 1–3	Pain relief	Y	Y	NA	Y	Y	NA	Study showing mid-term improvement
Dilke et al. [32] (1973)	RA, PC, B	8/12	Total = 100 Epidural = 50 Interspinous = 50	Methylprednisolone in normal saline or interspinous ligament Number of injections = 1–2	Pain relief, analgesic consumption, changes in straight leg raising, or neurologic signs	Y	NA	NA	Y	NA	NA	Study showing improvement
Pirbudak et al. [104] (2003)	RA, AC, B	10/12	Total = 92 Epidural = 46 Epidural + amitriptyline = 46	Betamethasone and bupivacaine or with addition of amitriptyline Number of injections = 1–3	VAS, ODI	SI in both groups	SI in both groups	SI in both groups	Y	Y	Y	Study showing improvement
Arden et al. [4] (2005)	RA, PC, B	11/12	Total = 228 Steroid group = 120 Placebo group = 108	Triamcinolone and bupivacaine or normal saline into interspinous ligament Number of injections = 3	ODQ, pain relief, VAS, SF-36	NSI	NSI	NSI	N	N	N	Study showing lack of improvement
Carette et al. [18] (1997)	RA, PC, B	11/12	Total = 158 Methylprednisolone = 78 Placebo = 80	Normal saline vs Depo-medrol® and procaine Number of injections = 1 to 3	VAS, ODI	NSI	NSI	NSI	N	N	N	Inappropriate blinded placebo study showing lack of improvement
Wilson-MacDonald et al. [128] (2005)	RA, AC, B	10/12	Total = 60 Intramuscular = 34 Epidural = 26	Intramuscular injection or epidural bupivacaine with methylprednisolone Number of injections = 1–2	Oxford Pain Chart, ODI	SI in the treatment group	U	U	Y	U	U	Small study showing short-term improvement

RA = randomized; AC = active control; F = fluoroscopy; PC = placebo control; B = blinded; IL = interlaminar; TF = transforaminal; NRS = Numeric Rating Scale; ODI = Oswestry Disability Index; SI = significant improvement; PSI = Patient Satisfaction Index; ODQ = Oswestry Disability Questionnaire; NA = not applicable; Y = yes; NSI = no significant improvement; U = unclear; N = no.

Seven fluoroscopic trials showed efficacy of transforaminal epidural injections in managing lumbar disc herniation, with strong evidence for the short term and moderate evidence for the long term (Table 5). Three of these were high-quality trials [40, 51, 121] and four moderate-quality trials [1, 97, 107, 112]. Two high-quality fluoroscopic trials showed lack of efficacy [25, 55].

Table 5.

Description of study characteristics of randomized controlled epidural trials assessing the efficacy of lumbar transforaminal epidural injections in lumbar disc herniation.

Study	Study characteristics	Methodologic quality score	Number of patients	Interventions	Outcome measures	Pain and function improvement			Improvement			Comment(s)
						3 months	6 months	12 months	Short term	Long term
									(≤ 6 months)	> 6 months	≥ 12 months
Cohen et al. [25] (2012)	RA, PC, F	10/12	Total = 84 Saline = 30 Corticosteroid = 28 Etanercept = 26	Steroids, etanercept, or saline Number of injections = 1–2	Pain relief, ODI, global perceived effect	Steroids = 50% Etanercept = 42% Saline = 43%	Steroids = 29% Etanercept = 38% Saline = 40%	Steroids = NA Etanercept = NA Saline = NA	N	N	NA	Although this was a well-conducted study, it was not a true placebo study. Even though there was no significant difference, the authors concluded epidural steroid injections may provide short-term pain relief for some. The included patients had subacute sciatica.
Ghahreman et al. [40] (2010)	RA, PC, F	12/12	Total = 150 5 groups with 28, 37, 27, 28, 30	Steroids with saline vs local anesthetic vs intramuscular steroids vs Intramuscular saline Number of injections = 1–3	At least 50% pain relief	TF saline = 19% TF local anesthetic = 7% TF epidural = 54%	NA	NA	Steroids = Y Local anesthetic and saline = N	NA	NA	First study of its nature with a true placebo evaluation showing short-term improvement
Karppinen et al. [55, 56] (2001)	RA, PC, F	11/12	Total = 160 Methylprednisolone-bupivacaine = 80 Saline = 80	Sodium chloride solution or methylprednisolone (40 mg) and bupivacaine (5 mg) Number of injections = 1	VAS, ODI, Nottingham Health Profile, cost, physical examination	NA	SI in both groups	SI in both groups	U	U	U	An ineffective or inappropriate placebo design, without applicable results
Jeong et al. [51] (2007)	RA, AC, F	9/12	Total = 193 G = 104 PG = 89	0.5 mL bupivacaine hydrochloride and 40 mg 1 mL triamcinolone Number of injections = 1	VAS	PG = 88.4% G = 70.9%	PG = 60.4% G = 67.2%	NA	Y	Y	NA	Moderate-quality study showing mid-term improvement
Riew et al. [111, 112] (2006, 2006)	RA, AC, F	8/12	Total = 55 Bupivacaine = 27 Bupivacaine + steroid = 28	Bupivacaine 0.25% or bupivacaine with 6 mg betamethasone Number of injections = 1–4	North American Spine Society Outcome Instrument, operative treatment considered as failure of injection treatment	NA	NA	33% vs 71% (avoided surgery)	Steroids = Y Local anesthetic = U	Steroids = Y Local anesthetic = U	Steroids = Y Local anesthetic = N	Surgery avoided in 33% of bupivacaine group and 71% in steroid group
Ng et al. [96] (2005)	RA, AC, F	11/12	Total = 86 Bupivacaine = 43 Bupivacaine + steroid = 43	Bupivacaine only or bupivacaine with methylprednisolone Number of injections = 1	VAS, ODI, change in walking distance, claudication, satisfaction of the outcome	Bupivacaine = 47.5% Bupivacaine + steroid = 41.5%	NA	NA	Steroids = Y Local anesthetic = N	NA	NA	Small study with short-term followup showing improvement
Lee et al. [63] (2009)	RA, AC, F	7/12	Total = 93 IL = 34 TF = 59	Interlaminar vs transforaminal epidural injections 4 mL (TF) Number of injections = 1–3	NRS, PSI	Roland Pain Score: TF = 3.34–1.59 IL = 3.25–1.57	NA	NA	Y	NA	NA	Study with short-term followup showing improvement
Ackerman and Ahmad [1] (2007)	RA, AC, F	7/12	Total = 90 Caudal = 30 IL = 30 TF = 30	Steroid and saline with local anesthetic Number of injections = 1–3	Pain relief	Caudal = 57% IL = 60% TF = 83%	Caudal = 57% IL = 60% TF = 83%	NA	Y	P	NA	Study showing mid-term improvement
Park et al. [97] (2010)	RA, AC, F	7/12	Total = 106 Dexamethasone = 53 Triamcinolone acetate = 53	Dexamethasone or triamcinolone acetate with lidocaine Number of injections = 1	VAS, Short MPQ, ODI	Dexamethasone = 40% Triamcinolone = 71%	NA	NA	Y	NA	NA	Short-term improvement with triamcinolone compared to dexamethasone
Rados et al. [107] (2011)	RA, AC, F	8/12	Total = 64 IL = 32 TF = 32	IL vs TF Number of injections = 1–3	VAS, ODI, disability scores	TF = 53% IL = 75%	TF = 53% IL = 75%	NA	Y	Y	NA	Study showing mid-term improvement
Tafazal et al. [121] (2009)	RA, AC, F	10/12	Total = 76 Bupivacaine = 34 Bupivacaine + steroid = 42	Bupivacaine with methylprednisolone Number of injections = 1–3	VAS, ODI, LBOS, Modified Somatic Perception Questionnaire, MZD	VAS and ODI changes: Bupivacaine = 24.3 and 13.8 Bupivacaine + steroid = 27.4 and 13.6	Y	NA	Y	Y	Y	Study showing long-term improvement

RA = randomized; PC = placebo control; F = fluoroscopy; AC = active control; G = ganglionic; PG = preganglionic; IL = interlaminar; TF = transforaminal; ODI = Oswestry Disability Index; NRS = Numeric Rating Scale; PSI = Patient Satisfaction Index; MPQ = McGill Pain Questionnaire; LBOS = Low Back Outcome Score; MZD = Modified Zung Depression; NA = not applicable; SI = significant improvement; Y = yes; N = no; U = unclear.

Discussion

Lumbar disc herniation is readily diagnosed and the most common indication for surgical intervention in the lumbar spine [43, 127]. While it is generally believed that the course and prognosis of acute sciatica secondary to disc herniation are favorable [59, 125], some patients continue to have persistent and disabling symptoms for 2 years or longer [34, 123] and many undergo surgery [43, 117, 127]. Therapeutic epidural injections utilized as a nonsurgical management for chronic persistent disabling disc herniation and radiculitis have become increasingly popular [46, 81, 83, 84]. In fact, Manchikanti et al. [83], in an assessment of utilization characteristics of epidural injections in the fee-for-service Medicare population, found a 130% growth per 100,000 Medicare patients from 2000 to 2011 [81, 83, 84]. The lumbar transforaminal approach had a breathtaking increase of 665% in that same period. In this systematic review, we determined the efficacy of the three anatomic approaches to epidural injections—caudal, lumbar interlaminar, and transforaminal—in the treatment of disc herniation and radiculitis with placebo-controlled and active-controlled interventions.

The limitations of this review include the paucity of high-quality literature for each modality, despite 23 trials meeting inclusion criteria in lumbar disc herniation for three modes of administration. In addition, we were unable to perform a meta-analysis due to a lack of homogeneity among the RCTs for mode of administration. Further, the majority of the evidence was obtained from active-controlled trials, specifically for long-term improvement with all three approaches. Active-controlled trials compare two different procedures or drugs. Thus, some may consider this as a weakness. The majority of the analytical flaws arrive as methodologists consider one of the drugs as placebo and compare between the two groups rather than baseline to followup periods. Thus, the strengths of active-controlled trials include their comparative evaluation, which has become pivotal in modern evidence-based medicine [77]. These studies provide insight into the effectiveness of local anesthetic with or without steroids and various methods such as interlaminar, caudal, and transforaminal. Placebo-controlled trials are also extremely important and constitute a limitation to our analysis. Specifically, it is desirable to have studies with their placebo group designed with appropriate inclusion of injection of placebo solution into a nonactive structure, which has not been included in the majority of the studies and is a debated issue [9, 10, 18–20, 40, 50, 55, 69, 70, 78, 103]. Further, placebo interventions have been misinterpreted based on the solution injected and the location of the injection. This has reached such proportions that methodologists have continued to consider even a local anesthetic injection or any solution injected into an active structure as placebo even though these invariably result in a multitude of effects, with some of the effects being therapeutic [6, 8–10, 18, 19, 33, 40, 48, 50, 54, 55, 62, 70, 71, 73–75, 78, 80, 86–91, 100–103, 115, 120]. In designing the studies, apart from active-controlled design, the reviewers must be cognizant of the multiple effects of placebo; nocebo; Hawthorne effect; natural course of the disease, which is not applicable in these chronic patients; and regression to the mean, which has been extensively discussed in reference to placebo, nocebo, and pure and impure placebos [48, 54]. Thus, this may be considered as a weakness but not a flaw unless active-controlled trials are misinterpreted. Some may also consider not utilizing all RCTs irrespective of their size and quality as a deficiency; however, we believe that is one of the strengths of this systematic review.

We found high-quality evidence showing that epidural injections have short-term benefits in terms of alleviating pain and disability of lumbar disc herniation and moderate long-term effects for all three approaches. The evidence was stronger for short-term effects compared to long-term effects. The results of our systematic review are similar to those of some previous reviews [8, 69, 71, 76, 100]. Of importance is the systematic review by Pinto et al. [103], which partially agreed with our review in reference to short-term but not long-term improvement. Pinto et al. [103] concluded epidural steroid injections have a small, short-term effect on leg pain and disability compared to placebo in patients with sciatica, but without significant effect in the long term. Our review differs from that of Pinto et al. [103], which has been criticized for its deficiencies [65, 67]. Pinto et al. [103] utilized methodologic quality assessment criteria, based on an instrument developed for physiotherapy, which has not been validated for interventional techniques [65, 67] and differs substantially from criteria developed by the Cochrane review group [35, 119]. Additional deficiencies of that systematic review include the inclusion of a multitude of heterogeneous studies that were labeled homogeneous. The authors indicated that the studies were homogeneous based on the fact that reviewers have decided that local anesthetic injection was a placebo. We believe that such a methodology invalidates the entire concept of meta-analysis of homogeneous studies. Pinto et al. [103] also did not take into consideration the varying effects of placebo and nocebo, impure placebo, and the effects of injecting not only various active solutions but also inactive solutions into active structures [20, 48, 49, 54, 69–71, 78, 82].

The evidence appears to be somewhat stronger for caudal epidural injections. Only one placebo-controlled trial by Iversen et al. [50] utilizing ultrasound imaging showed a lack of efficacy. However, this trial has numerous deficiencies, including flawed design, poor selection criteria, poor inclusion criteria, and no injection of local anesthetic, leading to only moderate quality in methodologic quality assessment. The only study assessing long-term improvement of 2 years, including 120 patients with local anesthetic with or without steroids, was an active-controlled trial that showed sustained improvement over a 2-year period, with repeat procedures as medically necessary and indicated, both in reducing low-back and lower-extremity pain and improving function [79, 85, 86].

For lumbar interlaminar epidural injections, there were two fluoroscopic studies [3, 89] and one blind trial [104] assessing long-term effectiveness that showed improvement. A major study by Carette et al. [18], published in 1997, showed no efficacy with a blind approach. In this assessment, the authors utilized sodium chloride solution in the epidural space, which may have significant effects beyond the placebo effect [8, 9, 44, 71, 82, 85, 86, 89, 91, 100, 129]. Of the trials showing improvement, all were actively controlled [3, 89, 104]. However, true placebo studies with injection into the intraspinous ligament showed contradicting results. Efficacy was shown by Dilke et al. [32] in 1973 and lack of efficacy was shown by Arden et al. [4] in 2005.

In reference to lumbar transforaminal epidural injections, there was only one long-term study with 1-year followup in a relatively small number of patients (total 55) with an active-controlled design. There were multiple well-performed studies showing no efficacy. Among the placebo studies [25, 40, 55], Ghahreman et al. [40] used a true placebo design injecting sodium chloride solution into a muscle and over the nerve root, whereas Cohen et al. [25] and Karppinen et al. [55] utilized a rather impure placebo design injecting sodium chloride solution into the neural space over the nerve root.

There was only one trial, that of Ackerman and Ahmad [1], comparing all three approaches; however, this was only of moderate quality and reported improvement with a 6-month followup, with superior results for the transforaminal approach compared to the interlaminar and caudal approaches. One additional trial by Lee et al. [63] compared interlaminar versus transforaminal epidural injection with a short-term followup and showed basically similar results with both approaches.

In this analysis, utilizing appropriate methodology, based on Cochrane review criteria for quality of evidence for RCTs [119], we found strong evidence for epidural injections in managing chronic pain of disc herniation in the short term and moderate evidence in the long term. This may be in contrast to some of the previous reviews while in agreement with others. Previous authors have concluded lack of efficacy of epidural injections over the long term, inappropriately utilized local anesthetic as placebo, and performed meta-analysis that often yielded misleading results because local anesthetics have yielded results that were similar to steroids [22, 69, 78, 86, 91, 103, 119]. While the strength of our review is that it provided qualitative evidence, we were unable to provide quantitative information (as might be done through pooling data in meta-analysis) because of excessive heterogeneity among the trials identified by the systematic review. Thus, to facilitate meta-analysis, future RCTs should focus on study design with long-term followup and appropriate outcome parameters, which will make it more likely that their results will be suitable for pooling and make them more likely to be included in a meta-analysis. These results can guide shared decision making in which patients are informed about up-to-date evidence and probable outcomes in a balanced manner [26].

Electronic supplementary material

Below is the link to the electronic supplementary material.