Background

Pain radiating from the spine into the leg is known as sciatica or radicular pain [1, 2]. Most patients with sciatica consulting in primary care improve over a number of weeks but about a third of patients can continue to struggle with pain and disability at 12 months [3]. Patients who do not improve may be referred to specialist spinal services for further investigations (e.g. imaging), and additional management options including spinal injections and/or consideration for surgical options such as discectomy.

Current clinical guidelines [4,5,6] underline the need for better evidence to support sciatica treatment decision-making. This study focuses on spinal injections for patients with sciatica and the term epidural steroid injection (ESI) is used to describe any type of spinal injection (including local anaesthetic and corticosteroid), for disc-related sciatica for reducing radicular leg pain.

A recent randomised controlled trial comparing the effectiveness of transforaminal ESI (TFESI) versus surgical microdiscectomy on pain and disability outcomes in patients with sciatica showed that TFESI was similarly effective to surgery [7]. However a review of studies of ESIs for radicular/ sciatica pain found injections may have a small short-lasting effect in reducing leg pain and disability [8]. Despite the common use of ESI in sciatica management it is not known which patients are more likely to have a good outcome following ESI, nor which patients are unlikely to benefit. Patients who do not improve with ESI may undergo repeat injections or experience delays proceeding to surgery.

Our recent literature review [9] included 15 studies that identified forty-two predictors of outcome (prognostic factors) following ESI from patient, clinical and imaging characteristics, routinely collected in clinical practice. Overall the quality of the studies was low. No prognostic factor was found to be consistently associated with outcomes following ESI. The review demonstrated that there is little, and low quality evidence to guide practice in terms of factors that predict outcomes in patients following ESI for disc-related sciatica.

The aim of this study was to generate consensus on potential predictors of outcome (prognostic factors) following ESI for disc-related sciatica to inform future research. Two approaches were used to generate consensus; a small consensus workshop followed by a modified Delphi survey.

Methods

The Conducting and Reporting Delphi Studies (CREDES) recommendations [10] are used to report the Delphi survey. Ethical approval was obtained from the Keele University Faculty of Medicine Research Ethics Committee (FMREC Reference No MH-200,160).

Expert workshop

A workshop was arranged with a multidisciplinary panel of experts who consider ESI as a management option for disc-related sciatica in the United Kingdom (UK) healthcare setting e.g. spinal surgeons, specialist spinal physiotherapists, rheumatologists, interventional pain specialists. Two of the experts are members of the research team (KK, JO’D).

The workshop panel were asked to generate a list of items that might be potentially associated with outcome following ESI based on their expert opinion. They voted on each item as to whether it should be included in the Delphi survey as a potential predictor of good or poor outcome following ESI for disc related sciatica (Yes/No). Items that did not receive any yes votes were not included in the Delphi survey.

Delphi study

To reduce the long-list generated in the consensus workshop to a ‘short-list’ of predictors, an electronic two-stage Delphi survey (using ‘Health Survey’), was carried out with invited experts. The research team used their clinical and academic networks to identify potential participants locally, nationally and internationally with clinical and/or research expertise in the field. A minimum of one criterion of either clinical or research expertise in the field was needed (see Table 1).

Table 1 Inclusion criteria for experts invited to participate in the Delphi survey
Full size table

As an introduction to the Delphi survey, a written case scenario was included, of a patient with clinical and MRI findings of disc-related sciatica leg pain, potentially suitable for referral for an ESI, to help guide the participants’ reasoning (see online resources 1 for case scenario and definition guidelines). When considering each item, participants were instructed to think of outcome as either a good outcome (significant reduction in leg pain) or a poor outcome (no significant reduction in leg pain).

For each potential predictor item, participants were asked “Do you think this item is important enough to be included in a future cohort study as a potential predictor of outcome following an ESI?” A 6-point Likert scale from ‘not at all important’ to ‘extremely important’ was used, any number less than 5 in a Likert scale may compromise reliability and validity [11].

In round one, participants were also invited to identify any additional potential predictor items not included in the survey. After the first round, statements that were not considered important by the Delphi respondents were dropped. All items from round one that reached consensus as either important or unimportant were not included in round two and participants were asked to re-rate the remaining statements and were show the breakdown of answers from all respondants in Round one. They were also asked to rate newly suggested items from Round one. Figure 1 depicts stages of the modified Delphi study.

Fig. 1
figure 1

Stages of the Delphi consensus study

Full size image

Sample size

Based on previous Delphi studies on LBP related topics [12, 13]., the target final sample size was a minimum of 30 participants. Assuming 60% response to each of two rounds, the sample size to be invited initially was approximately 90.

Analysis

Participants were asked to rate again the items from round one that were of unclear importance, or for which there was disagreement and lack of consensus (Table 2).

Table 2 Consensus criteria
Full size table

In order to give an overall ranking to items, a scoring system was assigned to the 14 items that reached consensus from rounds one and two and the remaining items from round two. Five points were assigned to “extremely important” down to zero points for “not at all important”. Each item score was totalled and ranked from highest to lowest score.

All items that reached consensus as important from rounds one and two and the higher scoring ranked remaining items from round two were prioritised for reporting as results of this study.

Results

The consensus workshop group comprised four consultant spinal orthopaedic surgeons, one consultant in pain medicine and two consultant spinal physiotherapists all based in the United Kingdom. Using items investigated in the literature and the expert group adding their own items, a list of 70 potential predictor items was generated. After voting, nine items did not receive any votes to be included in the Delphi survey The remaining 61 items were grouped in domains: (i) health, (ii) previous episodes, (iii) medication, (iv) pain, (v) physical function, (vi) psychosocial, (vii) work, (viii) clinical assessment findings, (ix) MRI findings and (x) injection factors and carried forward to the Delphi survey. All items generated by the workshop group are presented in online resources 1, views of the Delphi studies are available in online resources 2.

E-mail invites were sent to 90 professionals to participate in the Delphi survey. The response rates were 49% (n = 44) and 75% (n = 33) for rounds one and two of the Delphi process, respectively. Demographics of the participants are detailed in Table 3.

Table 3 Demographics of Delphi survey participants
Full size table

Round one

In round one seven items reached consensus by at least 70% of participants as a potential predictor of outcomes of ESI for sciatica (leg pain greater than back pain, leg pain intensity, patient treatment expectations, duration of sciatica symptoms, depression, fear avoidance beliefs, type of injection). One item (gender) reached consensus not to be included. Participants suggested 28 additional potential predictors.

Round two

In round two, participants rated items from round one of unclear importance, or for which there was disagreement and lack of consensus and the 28 newly suggested items from round one (a total of 81 items). There was no disagreement on any item in round two. Seven additional items reached consensus as a potential predictor by at least 70% of participants (anxiety, presence of neuropathic pain features, long-term opioid use, distress and somatization, previous history of lumbar surgery, self-efficacy, type of disc herniation). The remaining items that did not reach > 70% consensus from round two were ranked (see methods). The top 25 potential predictors reached by consensus and subsequent ranking are shown in Table 4.

Table 4 Ranked top 25 potential predictor of outcome following ESI
Full size table

Discussion

This is the first study that used a consensus approach to gain agreement on items that could be potential predictors of outcome following ESI for sciatica leg symptoms. An expert workshop followed by an online two-stage Delphi survey helped to identify over 80 potential predictor items across ten domains spanning the biopsychosocial model and these were ranked in order of importance by experts. Looking broadly at the top 25 items that achieved consensus as potentially important predictors, they covered the domains of pain and physical function, health and medication, psychosocial items, clinical assessment findings, imaging findings and injection items.

Comparison of results to other studies

Seven of the top 25 items were pain items. For the highest ranked item of leg pain greater than back pain, studies show that clinical subgroups of sciatica patients, whose back pain is greater than leg pain, fare worse in terms of their outcomes over one year [14]. Current leg pain intensity and duration of symptoms were ranked second and fourth respectively in the Delphi survey. This is unsurprising as current clinical guidelines recommend ESIs in acute and severe sciatica [4] and patients with higher leg pain intensity are more likely to be offered ESI in clinical practice. Having bilateral leg pain and whether the pain is constant or intermittent is not typically looked at and was not addressed in the prognostic factor studies literature and would seem to reflect the impact and severity of pain. However the direction of improvement is unknown.

Previous systematic reviews of prognostic factors for lumbar discectomy surgery for sciatica patients have reported mixed results [15]. Some studies showed that higher leg pain is associated with better outcome, while others did not show a relationship. Low level evidence supports duration of leg pain intensity preoperatively not being associated with outcomes post-surgery. The largest prognostic factor study of patients with sciatica in primary care investigated associations between patient, clinical and imaging characteristics and disability outcomes [16]. The study identified a weak association of shorter pain duration with improvement but this was not sustained at 12 months. In that primary care cohort, only two factors were associated with a poor outcome – identifying symptoms with the condition (identity) and believing symptoms would last a long time (timeline). In the same cohort, patients with persistent neuropathic pain features e.g. unpleasant sensations in the skin such as tingling or pain descriptors such as electric shock, had worse outcomes over three years than those without [17]. Hence exploring its relevance in patients listed for an ESI could shed interesting light on prognosis. Sleep disturbance is a known risk factor for developing chronic high impact widespread pain [18] but its relevance to outcome post-ESI has not been studied.

Seven psychosocial items were ranked in the top 25 potential predictors of outcome. Treatment expectations (ranked number 3) have been shown to play an important role in predicting clinical outcome after lumbar injections [19]. However the study population were patients with low back pain without mention of radicular leg pain/ sciatica thus the influence of treatment expectations on outcome post-ESI for sciatica is unknown. Anxiety and depression can predict unfavourable outcome in low back pain and general musculoskeletal pain [20] however there is limited exploration of their role in predicting outcome in ESI.

The rationale from the consensus expert group to include use of longer term opioids (ranked number 10) was because of its association with increased sensitivity to painful stimuli, known as opioid induced hyperalgesia [21]. Medication use as a prognostic factor for outcome post-ESI was investigated in two studies, one looked at high dose opioid use, the other general medication use (nothing, non-narcotic, tramadol, to stronger narcotic medications). Neither found any association with pain reduction or changes in medication use after ESI [9].

The results from this Dephi study favoured straight leg raise test and location of pain as items from clinical assessment to include in a future cohort study. Straight leg raise test is recognised as a diagnostic criterion for sciatica [22] however its association with outcome post-ESI has not been established. Location of pain is thought to reflect the level of spinal nerve root involved, and should tally with the MRI findings. Perhaps a mismatch would result in a less favourable outcome from the injections.

Seven potential MRI findings were considered in the Delphi study but only two were ranked in the top 25 potential predictors; type of disc herniation (e.g. protrusion/ extrusion) and any associated spinal stenosis/degenerative changes at the segment affected by the herniation. Type of injection was ranked seventh in overall importance (e.g. ultrasound guided versus radiology guided). Many consider ‘radiology guided’ injections as the “gold standard” [23] but there is no convincing evidence of the superiority of ‘guided’ over ‘unguided’ injections [4, 24] and no studies have examined it an a potential predictor of outcome.

Implications of the results

There is no clear evidence from the literature as to what factors or characteristics might predict outcome in sciatica patients post-ESI. Prognostic factor studies to date are generally poor quality and none have used a consensus method to identify the most important prognostic factors. The results of this consensus study have directly informed the predictors included in a planned prospective cohort study called the POiSE study: Predictors of outcome in sciatica patients following an epidural steroid injection in the UK National Health Service [25]. There will be a limit on how many factors can be included in the statistical analysis. To help determine the key factors to include in the data collection within the future cohort, a consensus-derived “short-list” of key potential outcome predictors was needed. The results of this consensus study have directly informed the predictors to be prioritised for inclusion in the prospective cohort study.

Strengths and limitations

Strengths of this Delphi consensus study include carrying out a previous systematic review and a multidisciplinary workshop process to generate a long list of items for the Delphi survey.

The majority of Delphi respondants were UK based and all consensus workshop members were UK based which may limit the generalizability of the results to settings in other countries.

At the start of the Delphi, participants were asked to focus on predictors specific to response to an ESI, however reviewing the top ranked factors chosen by the experts it could be argued that many of the identified predictors (e.g. the psychosocial factors) are likely to be related to outcome after any treatment rather than specific to ESI.

Conclusion

This study generated a long list of potential predictors that could be associated with change in leg pain following an ESI injection. This reflects the range of opinions and uncertainty around what experts think could be potentially influencing outcomes in patients following ESI. The on-line survey generated agreement on 14 items which are being used in a multi centre prospective cohort study investigating prognostic factors of outcome in patients having an ESI for sciatica.