Background

Burden of the problem

Chronic non-cancer pain (CNCP), defined as pain persisting beyond the period of normal healing (typically ≥3 months), is commonly experienced by a large proportion of adult population. In the USA, according to 2008 estimates, approximately 100 million adults were affected by CNCP with associated treatment costs ranging from $560 to $635 billion (US). This is greater than annual treatment costs of heart disease, cancer, or diabetes [1]. In the UK, 53 % of seniors reported that CNCP was the most important factor impacting their quality of life [2]. Chronic low back pain (CLBP) has been noted to be the leading cause of years lived with disability, with neck pain as the fourth common cause [3]. CNCP accounts for a large number of physician encounters in the elderly and aging population. In the USA alone, the proportion of CLBP patients requesting care from a health provider increased from 73.1 to 84.0 %, between 1992 and 2006 [4]. Conditions characterized or defined by the presence of pain accounted for five of the leading ten conditions associated with the most years lived with disability [5]. Common peripheral (non-axial) CNCP conditions include pain in upper or lower limbs, knee and shoulder joints, headache, carpal tunnel syndrome, epicondylitis, fascitis and others.

Treatment considerations

A clear understanding of the underlying pathogenesis of CNCP is important to drive treatment decisions; however, despite decades of research, the mechanisms underlying most CNCP conditions are unclear [5]. CNCP conditions involve a complex interplay of active pain pathways leading to sensory signaling, along with peripheral and central sensitization as a result of nervous system changes induced by temporary and long-term neuroplasticity. Changes resulting from sensitization lead to modulation of pain signals, which vary from patient to patient. Most CNCP conditions do not have definitive treatment, and physicians have limited options for symptomatic management. Analgesic medications are not always effective and need to be given on a daily basis; they also involve a risk of drug interaction especially in the elderly adult population. Corticosteroid injections (CSI) are commonly used to manage many CNCP conditions, despite variability in their clinical presentation and underlying pathophysiology. For neuraxial pain, epidural steroid injections (ESI) are the most commonly performed interventions, and in the USA alone, the number of ESIs administered to Medicare recipients essentially doubled between 2000 and 2004 (from 740,845 to 1,437,962 procedures/year) [6]. CSIs are also widely used as injections to knee [7], shoulder [8], carpal tunnel [9], nerve blocks for occipital nerve [10], suprascapular nerve [11], tendinous injections for medial and lateral epicondyle [12], and even for simple trigger point injections [13].

Percutaneous injections using steroids and local anesthetics

The rationale for injecting local anesthetics (LA) is to block sensory signals from the region being injected. Although often used for diagnosis—because it should only lead to a temporary blockade—an injection of LA has the potential to decrease sensitization, which is a feature of chronic or persistent pain, thereby possibly prolonging the treatment effect of LA beyond its pharmacological duration of action [14]. Historically, many investigators have published on the successful use of epidural injections of saline and LA. Cathelin. F, Pasquier and Leri, and Sicard published reports as early as 1901 [15]. Later, Evans (1931) published a successful report using procaine and saline in 22 of 40 patients [16]. Cyriax published multiple manuscripts referring to safe use of epidural injections using only LA and steroids in more than 20,000 cases [17, 18]. In fact, up until the 1950s, the injectate used for epidural injections in sciatica consisted of LA and saline. The first recorded use of steroids in epidural space was by Lievre et al. in 1953 [19]. Discovered by Philip Hench in the 1940s [20], glucocorticoids are potent anti-inflammatory agents. Since then, steroids have been injected for nearly every chronic pain indication [21]. In clinical practice, steroids are typically combined with LA, with or without saline [15]. This practice stems from the hope that the addition of steroid can lengthen the treatment effect [22]. For most CNCP conditions (except for evidently inflammatory conditions such as rheumatoid arthritis), there is no evidence that CSI are disease-modifying agents [23]. Whether steroids have any direct effect on pain generation or transmission is not clear. There is some experimental evidence demonstrating suppression of ectopic discharge in neuromas [24]. Pre-clinical experiments suggest that steroids may reduce neuropathic pain; however, a paradoxical effect of increased pain in some patients has also been shown [25]. Surprisingly, there are few clinical studies that have attempted to elicit the mechanism behind steroid’s effect on chronic pain other than as an anti-inflammatory [26]. Thus, the use of steroid for interventions in CNCP lacks clear rationale.

Harms of steroid injections

Corticosteroids can have systemic and local adverse effects. Systemic side effects are largely dependent upon the patient’s physiology, injected dose, and systemic absorption. Theoretically, these systemic effects can occur when the dose of steroids injected exceeds the rate of endogenous steroid production of about 20 mg per day of hydrocortisone or its equivalent and potentially cause suppression of pituitary adrenal axis, hypercorticism, Cushing’s syndrome, osteoporosis, avascular necrosis of bone, steroid myopathy, weight gain, fluid retention, and hyperglycemia [21]. Although these risks are unlikely with a single injection, in clinical practice most patients receive multiple CSI injections at frequent intervals, and it is not uncommon for patients to suffer from multiple chronic pain conditions treated with steroid injections by different physicians, without necessarily accounting for the systemic effects of steroid injected elsewhere. Local adverse effects include soft tissue atrophy, depigmentation [27], and alopecia [28]. Beyond the above recognized adverse effects, the neuraxial steroid injections (epidural, facet joint) risk causing rare, but catastrophic neurologic injuries such as stroke and spinal cord injury [29].

Limitations of previous reviews

Existing systematic reviews of injection therapy for CNCP conditions have not specifically compared LA vs. LA and steroid injections. Further, some reviews have considered LA injection to be a neutral agent, and combined this intervention with other biologically inactive control agents such as saline injections [12, 30]. More recently, within the last quarter of 2015, two additional reviews have been published. Manchikanti et al. reviewed all spinal injections targeted at epidural, facet joint/nerve, and disc pain conditions. Due to the existing heterogeneity, a meta-analysis was not considered feasible. All RCTs using an active control design were included. They observed that LA alone was as equally effective as LA with steroid, except in disc herniations [18]. The review by Chou et al. also focussed only on epidural injections for radiculopathy and spinal stenosis, without any restriction for the agent used and duration of chronic pain. They combined LA along with saline as placebo comparator. Their results showed a small effect favoring the use of steroids only in radiculopathy for short-term reduction in pain and function [31]. To illustrate these issues, we reviewed the systematic reviews published in English within the last 5 years, focusing on injection therapies using CSI for CNCP conditions. We identified the findings and limitations of those reviews, apart from identifying whether they included trials that addressed our review question of comparing LA and steroid with LA alone for CNCP (Table 1). Twenty seven out of forty-two reviews included trials comparing LA and steroid with LA alone [6, 12, 18, 30–53], but only some reviews have made clear observations on the comparison between LA vs. LA and steroid [18, 33, 38, 41, 45, 48, 51]. Very few conducted meta-analyses [12, 31–33, 35, 39, 41, 45, 48, 51]. The question of whether the injected steroid has any clinical benefit beyond that achieved from LA alone is an area of active debate, as existing studies across various clinical categories have shown equivalent effects comparing LA and steroid vs. only LA [54–57]. Findings from our review will help inform physicians, healthcare providers, and CNCP patients on whether the addition of steroid gives any meaningful benefit to LA injection alone.

Table 1 Limitations of existing reviews identified by a systematic search of literature
Full size table

Objectives

Our primary objective is to perform a systematic review and meta-analysis to assess the effectiveness of steroid + LA injections compared to LA injections, for pain relief among CNCP patients. Secondary objectives of our review are

  • To assess the effect of steroid + LA injections compared to LA injections alone on the duration of effect of pain relief.

  • To assess the effect of steroid + LA injections compared to LA injections alone on six core domains, described as core outcome measures by the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT) [58]. The domains other than reduction of pain include physical functioning; emotional functioning; participant ratings of global improvement and satisfaction with treatment, symptoms and adverse effects, and participant disposition.

Study question

Among adults with CNCP, does LA with corticosteroid injection (with or without saline) offer better pain relief when compared to injection with LA alone (with or without saline), as assessed at a time closest to 4 weeks after treatment?

Methods

Our review protocol has been registered with PROSPERO (registration number PROSPERO 2015:CRD42015020614). This protocol has been prepared for publication according to PRISMA-P guidelines [59].

Eligibility criteria

Participants

We will include adult (≥18 years of age) patients with CNCP, and exclude patients with a known inflammatory cause of pain such as rheumatoid arthritis, ankylosing spondylitis, systemic lupus erythematosus and gouty arthritis, and pain of generalized nature such as fibromyalgia or chronic fatigue syndrome. If a trial involves a mix of cancer and CNCP, or adult and pediatric patients, we will include the study only if they report outcomes separately for our study population of interest, or if at least 90 % of the trial patients are >18 years with CNCP.

Studies

Parallel design, randomized controlled trials (RCTs) will be eligible for our review. We will exclude trials with crossover design, N of 1 trials, and non-therapeutic trials.

Interventions

Eligible studies must randomize patients to receive LA + steroid or LA only, with or without saline, administered as percutaneous injection. We will exclude trials in which injections involve any additional agent (e.g., Hyaluronidase, dextrose, plasma), and any injection which involves a co-interventional procedure (e.g., radiofrequency treatment, epidurolysis).

Information sources

We will search the following electronic databases, from its inception till our search date: EMBASE, MEDLINE, and the Cochrane Central Registry of Controlled Trials (CENTRAL). Our search will not be limited to language. We will record the time and date of the literature search performed on each database. As a supplementary search, we will search the WHO clinical trial registry (http://apps.who.int/trialsearch/Default.aspx), and clinical trial registry (https://clinicaltrials.gov/), to look for any registered studies, which fulfill our eligibility criteria, and crosscheck for published results. Unpublished, but completed, study results will be requested from the authors or investigators. To further ensure comprehensiveness, we will review the bibliographies of relevant reviews published in English over the last 5 years (shown in Table 1).

Search strategy

The search will be performed using a sensitive strategy, prepared by an experienced librarian (RC), for each specific database, based in part on a comprehensive list of CNCP indications for which CSI are commonly utilized. The search terms will include possible indications, along with terms identifying corticosteroids, and LAs (Additional file 1).

Study records

Data management

We will conduct our review using an online software tool specially optimized to conduct systematic reviews—DistillerSR (https://distillercer.com/).

Study screening and selection

Studies fulfilling our eligibility criteria will be selected through a two-level screening process using standardized forms, applied through DistillerSR. At each stage, paired reviewers trained in health research methods will screen for studies for eligibility, independently and in duplicate. The first level will be done on titles and available abstracts of identified citations. For citations judged as potentially eligible, full-text article screening will be done. To ensure consistency, reviewers will perform a calibration exercise, before beginning with screening. Reviewers will be asked to resolve disagreement by consensus, or if a discrepancy remains, through discussion with an arbitrator (HS). A quadratic kappa statistic will be calculated as a measure of inter-observer agreement, independent of chance regarding study eligibility and interpreted as almost perfect agreement (0.81–0.99), substantial agreement (0.61–0.80), moderate agreement (0.41–0.60), fair agreement (0.21–0.40), slight agreement (0.01–0.20), <0 as less than chance agreement [60].

Data collection process

Reviewers working in pairs will extract the required data from each included study using a standardized data extraction form using DistillerSR (https://distillercer.com/). This form will be piloted between all pairs of reviewers for consistency and accuracy. To assist with the data extraction, a detailed instruction manual will be provided along with each relevant form.

Data items

Data abstracted will include study characteristics including risk of bias items, demographic information, participant flow through the study, and outcomes on continuous and binary measures captured on six core domains as recommended by the IMMPACT statement guidelines [58].

Outcomes and prioritization

We will consider pain relief as our primary outcome. We will also capture other outcomes (as guided by IMMPACT) including reporting of adverse effects. We will also prioritize the use of intention to treat analysis (ITT). We will only pool data across trials if there are three or more studies contributing to an outcome domain. It is recognized that presentation of relative effects will facilitate interpretation of treatment effects, and clinicians generally find dichotomous presentation of continuous outcomes more useful [61]. In the following section, we will explain the method of analysis as considered for pain relief. We will perform a similar analysis, as appropriate, for other IMMPACT outcomes.

Data synthesis and analysis of outcomes

We expect that all included trials will have captured a measure of pain relief, and this domain could be expressed in any of the following types: outcomes reported as binary (successful or not successful), outcomes reported in ordinal categorical scale (mild, moderate and severe), and outcomes reported in various continuous outcome scales. For the primary outcome, we will consider the outcomes reported closest to 4 weeks after the study interventions. Pain relief measured at this time will give us indication of treatment efficacy beyond the effects of placebo. For the primary analysis, we will use a complete case analysis with ITT. Analysis and synthesis will be done using revman (review manager) 5.3. Using random effects model for pooling, we will calculate the risk ratio (to be interpreted as the risk of having success) for dichotomous outcomes and weighted mean difference (WMD) for continuous outcomes converted into 0–10 (11 point) numerical rating scale (NRS) for pain. Further, we will dichotomize all included patients into proportion of successful patients for pain relief, to be able to report risk ratio across all included studies. However, dichotomising continuous outcomes will result in a loss of statistical power. Hence, we will also report the WMD. To achieve this, we will be performing the following methodological steps.

  1. 1.

    Converting outcomes reported in various continuous scales to a reference instrument scale of 0 to 10 NRS: According to IMMPACT, the 11-point NRS measure of pain intensity is recommended as a core outcome measure in clinical trials of chronic pain treatments. Furthermore, the 0–10 NRS is preferred over the VAS (visual analog scale) by patients and clinicians for its relative simplicity and ease of administration [58].

  2. 2.

    Converting continuous outcomes from 0–10 NRS into dichotomous data. A reduction of two points on the 0–10 NRS could be considered as the threshold to establish minimally important difference (MID), reflecting an average reduction of pain by 30 % [58, 62]. For studies, which report only end scores, we will consider a threshold of 4 or less in a 0–10 NRS, to dichotomize the patients. A score of 4 or less is considered as mild pain [63].

  3. 3.

    Converting categorical measurement results to a dichotomous data. The categories of ordinal scale commonly used in pain studies could include mild, moderate and severe pain. It has been shown that these categories correspond to the following thresholds of 0–10 NRS. Mild pain 0–4, moderate pain 5–7, and severe pain 8–10 [63]. Any change from a higher ordinal category to a lower ordinal category would be considered as a successful outcome for patients reported in a particular study, allowing us to dichotomize the study outcomes into success and failures for each arm.

  4. 4.

    Imputation for participants treated as lost to follow up for continuous and dichotomous outcomes (for sensitivity analyses). We will consider patient LTFU subsequent to randomization as missing for data analysis, and will be explored further for imputation, if it is >5 %. For trials in which the authors report total missing participant data only, without specifying at what stage the participants were missing, we will consider the total sample size and the actual sample size included for final analysis and assume that missing data were equally distributed between the arms. For trials in which the authors reported imputed analysis only, we will use the imputed results for the meta-analysis. We will perform imputational strategies as described by Ebrahim et al. [64] and Akl et al. [65], for continuous measures and dichotomous measures, respectively.

For secondary outcomes, we will pool the outcomes for other IMMPACT domains using a random effects model only if there are three or more studies for a particular domain. We will report the pooled outcome as WMD or standardized mean difference (SMD) as appropriate.

Risk of bias assessment and identification

Each included study will be assessed for risk of bias at the study level, using the Cochrane risk of bias tool, based on the components of random sequence generation, allocation concealment, blinding of participants, blinding of outcome assessment, incomplete outcome data, and selective outcome reporting. We will use a modified Cochrane risk of bias instrument, with response options of “definitely yes,” “probably yes,” “probably no,” and “definitely no.” We will assign trials in the “definitely yes” and “probably yes” categories a high risk of bias and those in the “probably no” and “definitely no” categories a low risk of bias. These items for each study will be extracted, in duplicate, during the data extraction stage, using the DistillerSR web tool. Any disagreement on the risk of bias item scoring will be noted and arbitrated by the primary investigator (HS). Agreement on risk of bias scoring will be assessed on a component-by-component basis, using quadratic kappa weighting and interpreted as stated before. No study will be excluded based on its risk of bias. We will contact study authors if limitations in reporting lead to uncertainties in eligibility, risk of bias, or outcome.

Assessment of heterogeneity

Statistical heterogeneity will be calculated using Cochrane’s Q test, with a threshold of p value at 0.10, and I 2 statistic to describe the percentage variability in individual effect estimates that could be due to true differences between the studies rather than a sampling error.

Subgroup and sensitivity analysis to explore heterogeneity

The following a priori hypotheses would be considered for subgroup analyses, and as possible reasons for unexplained heterogeneity in the pooled estimates. The subgroup analysis will only be done if there are more than two studies per subgroup.

  1. 1.

    Clinical category level subgroup analysis. Although, the underlying rationale that adding steroids to LA may not provide any meaningful improvement to pain reduction is consistent across all clinical indications and trials, there is potential for clinical heterogeneity in these conditions because of the variation in underlying pain pathways, pain sensitization, and hence the possibility of varying effects of steroids. Considering this, we will perform subgroup analysis of pain outcomes based on target structures injected. The seven clinical categories considered are as follows.

    1. 1.

      Peripheral joint: hip, knee, shoulder joint injections, and other joint injections.

    2. 2.

      Spinal joint/facet nerve (medial branch block)/ spinal disc: Facet joint or disc injections.

    3. 3.

      Spinal nerve or epidural or intrathecal: Epidural injections including transforaminal epidural.

    4. 4.

      Peripheral nerve: various nerves including occipital nerve, suprascapular, and median nerve.

    5. 5.

      Autonomic ganglia: injections to stellate ganglia, celiac plexus, lumbar sympathetic block.

    6. 6.

      Soft tissue injection: injections for lateral or medial elbow ligaments, subacromial bursa, or plantar fascia.

    7. 7.

      Trigger point or intramuscular: injections to trigger point, piriformis muscle, or any other intramuscular injections.

  1. 2.

    Larger effects toward steroids could be observed in studies with components of higher risk of bias; we will conduct this subgroup analysis on a risk of bias component-by-component basis, only if there is considerable variability within the risk of bias component.

  2. 3.

    Sensitivity analysis for LTFU will be conducted as described previously.

  3. 4.

    Studies which involved treatment interventions as a series, rather one specific injection (for example, three steroid injections over 1 month). We anticipate that the direction of effect would favor the steroid treatment due to systemic additive effect.

Addressing potential biases

Publication bias will be assessed using a funnel plot, if there are more than ten studies included in a meta-analysis. Further, it will be explored by the test of Egger [66]. Selective outcome reporting is difficult to identify when the study protocols are not available or published. For our review, we will consider the possibility of selective outcome reporting, when the outcomes are described in the Methods section but not identified or reported in the results section of the same study report [67].

Interpretation and reporting

We will report our findings as risk ratios, WMD, or SMD for individual outcomes, along with their 95 % confidence intervals. Within each clinical category (subgroups), similar findings will be reported. We will also report the outcome for pain relief as a single pooled estimate of risk ratio by dichotomising the continuous outcomes. This method has been suggested as appropriate and meaningful for interpretation in systematic reviews of pain studies [54]. We will also report the findings in measures of relative risk reduction and absolute risk reduction. Rating of quality of evidence will also be done using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach. This will enable us to report our study findings in the form of “summary of findings” table and allow us to evaluate the certainty in effect estimates.

Discussion

Despite the several advances in the field of chronic pain, we still do not fully understand the nature and mechanisms causing CNCP. Although steroids are being used in nearly every injection or interventional therapy, the reasons for its use are unclear. Our comprehensive systematic review will assess the clinical benefits of steroids when used along with LA, by comparing it with the effects of LA alone. This evaluation of both the effectiveness and the adverse events will include patient-important outcomes as outlined in IMMPACT recommendations, thereby enhancing comparability and external validity. Our results will facilitate evidence-based management of patients with chronic non-cancer pain and identify key areas for future research.

The review as such will be limited by the included studies and their quality. As highlighted above, we will attempt to decrease the potential for bias by performing reasonable subgroup and sensitivity analysis, including imputation for missed outcomes.