The clinical effectiveness of intra-articular corticosteroids for arthritis of the lower limb in juvenile idiopathic arthritis: a systematic review

Background Juvenile Idiopathic Arthritis (JIA) commonly affects joints of the lower limb including the knee, ankle, subtalar and other foot joints. Intra-articular corticosteroid injections (IACIs) are considered to be effective for short-term relief of synovitis, however, there appears to be a significant lack of published evidence from comparative effectiveness studies. The aim of this study was to identify and critically appraise the evidence for the efficacy of lower limb IACIs in children/adolescents with JIA. Methods Studies were identified in databases Medline, Embase, CINAHL, AMED, PEDro, the Cochrane Library and TRIP, with no date restrictions. The primary search terms ‘juvenile idiopathic arthritis’, ‘lower limb’, ‘knee’; ‘ankle’, ‘foot’ and ‘intra-articular steroid injections’ and related synonyms were used to develop a comprehensive pragmatic literature search strategy. Included studies were quantitative longitudinal design such as randomised controlled trials, pseudo-randomised and non-randomised experimental studies, cohort studies, and case-control studies. All outcomes measures were subject to analysis. Quality assessment was conducted using the Cochrane Collaboration criteria with additional criteria for sample population representativeness, quality of statistical analysis and compliant intervention use and presence of co-interventions. Qualitative data synthesis was conducted for the outcome domains. Meta-analyses were not possible as multiple randomised controlled trials for outcome measures were not available. Levels of evidence were assigned to each outcome measure. Results The inclusion criteria were met by twenty-one studies. One study had high quality for internal validity and nine studies had high quality for external validity. No studies had high quality for both internal and external validity. Four outcome domains were identified. There was weak evidence for IACIs decreasing clinical signs and symptoms in the lower leg, improving joint range of motion, decreasing leg length discrepancy, and for imaging techniques detecting the effects of IACIs. Conclusions There is some weak evidence for the efficacy of IACIs improving certain outcome measures. However, there is also some inconclusive evidence due to a lack of quality studies. More high quality evidence is necessary to definitely determine the efficacy of IACIs for JIA in the lower leg.


Background
Juvenile Idiopathic Arthritis (JIA) is the most common rheumatic disorder in children, affecting approximately 1 in 1000 Australian children [1]. Lower limb arthritis is common in JIA regardless of disease classification, with the knee joint affected most frequently at disease onset (40-60% of cases) followed by the ankle (20-30% of cases) [2,3]. Persistent synovitis in these joints is associated with pain, reduced joint ranges of motion, deformities, gait disruption, and poor functional status [4][5][6][7]. Moreover, radiographic progression of the disease occurs early, and if not addressed may result in permanent joint destruction and poor functional outcomes [8,9].
The primary aim of medical management in JIA is to suppress inflammation prior to the development of irreparable joint damage and functional limitation [10]. Recent improvements in the medical management of JIA have led to a greater emphasis on early diagnosis of JIA, early detection of synovitis, and subsequent prompt and aggressive systemic medical intervention via disease-modifying and/or biologic regimens to control the inflammatory disease process [11]. The introduction of such regimens have led to a relative increase in the number of patients achieving disease improvements characterised by low disease activity or remission [12]. However, in spite of promising signs of improvement, children may experience refractory disease, characterised by continued persistent synovitis, and/or serious adverse effects including hepato-toxicity and immuno-deficiency following administration of methotrexate and etanercept [13].
Intra-articular corticosteroid injections (IACIs) are often used as the main form of management in milder oligoarticular JIA, and also as an adjunct to systemic therapy in other JIA subtypes where localised synovitis is restricted to a single or a small number of joints [13,14]. IACIs are administered to induce rapid relief of symptoms through resolution of localised synovitis, and in some cases may prevent the need for escalation of systemic therapy and associated increased risk of serious side-effects [14]. It is acknowledged that IACIs are also associated with adverse events, however, these may be relatively mild compared to those associated with systemic disease-modifying antirheumatic drugs (DMARDs) or biologics. These may include subcutaneous atrophy, hypopigmentation and/or post-injection pain [14,15]. Several inconsistencies have been highlighted in studies of IACIs such as variable injection techniques, inclusion criteria, post-injection procedures, outcome measures, and follow-up period [14,16,17].
Previous systematic reviews have been conducted but they are now outdated and new studies of IACIs have been published since they were completed. The Cochrane Collaboration recommends that systematic reviews are updated biannually, and the most relevant and recent review was published seven years ago [18]. Two systematic reviews have been conducted to evaluate the evidence for IACIs in JIA; one which focused on IACIs for temporomandibular joint (TMJ) arthritis only [16]; and another which reviewed studies on lower limb IACIs for children with JIA and/or adults with rheumatoid arthritis (RA). However, ultimately they did not include any studies of JIA [17]. The recent systematic review of IACIs for TMJ arthritis [16] may not be directly relevant to this work as IACIs administered in the joints of the lower limb appear to be less efficacious than those administered in upper limb and non-weight-bearing joints [19]. Moreover IACIs administered to the knee joints appear to be more efficacious than those administered to the small joints in the foot, where the response is less predictable [19]. Two other reviews of the overall medical management of JIA adopted systematic approaches to conduct their literature search, but their restrictive inclusion criteria retrieved only two small randomised studies concerning IACIs and only limited conclusions were drawn [10,20]. One other systematic review attempted to appraise the evidence for knee joint IACIs for a range of arthritis conditions including JIA, osteoarthritis (OA) and RA [21]. However, methodological problems were identified with this review including a single joint focus (knee), adult arthritis predominance for studies included, and violations of the detailed inclusion criteria. Narrative reviews have also been conducted and have provided useful summaries of the literature concerning IACIs in JIA [14,22]. However, such reviews are often reinforced with the authors' expert opinions and clinical experience and as a result may be vulnerable to bias [23].
Most research concerning IACIs efficacy to date appears to focus predominantly on the knee joint only, or a combination of several joints which differ in terms of anatomical complexity, and may include weight-bearing/ non-weight-bearing joints which could impact on IACI efficacy. The reviews which have been published to date fail to account for the efficacy of IACIs for synovitis relief in other important lower limb joints, and thus the findings and conclusions drawn from these reviews are not necessarily generalizable. Lastly, it should be noted that standard clinical practice has changed in recent years with a greater emphasis being placed on image-guidance to aid the administration of intra-articular injection therapies [24][25][26]. This advancement of clinical practice may have led to improvements in the effectiveness of IACIs on the basis of improved steroid placement accuracy and thus fewer reports of adverse events [24][25][26]. The emergence of further evidence, introduction of new technologies and changing clinical practices has raised important new questions concerning lower limb IACI efficacy in JIA. Accordingly, the aim of this review is to systematically identify and critically evaluate the clinical effectiveness of IACIs administered to joints of the lower limb in people with JIA.

Search strategy
A detailed electronic database search of the literature was performed using the following databases: Medline, Embase, CINAHL, AMED, PEDro, the Cochrane Library, and TRIP. Reference lists of eligible studies identified by the electronic database search were searched by hand to identify studies not initially found in the electronic search. Personal reference lists of the reviewers were also hand searched to identify any studies known to the reviewers that may not have been identified by the initial search strategy. The following search terms 'juvenile idiopathic arthritis', 'lower limb', 'knee', 'ankle', 'foot', and 'intra-articular steroid injections' and related synonyms were used to develop a comprehensive pragmatic literature search strategy. Standard MeSH terms were utilised where possible or an appropriate text word was adopted. Boolean operators (such as 'explode', 'OR', '*', '$' and ' AND') were used. The full search strategy is outlined in Additional file 1.

Study inclusion criteria
Studies included were of a quantitative longitudinal design, such as randomised controlled trials (RCTs), pseudorandomised and non-randomised experimental studies, cohort studies, and case-control studies. Studies of single cases were excluded. Studies reporting children, adolescents and adults who have JIA were included. Studies of IACIs as the single primary intervention, in combination with systemic medications, and/or in combination with nonpharmacological interventions were included. All outcome measures/variables were subject to analysis. Literature was limited to full text published articles available in English due to language barriers of both independent reviewers. No restriction was imposed on year of publication.
Study abstracts found from the electronic database search and by hand searching were reviewed for eligibility based upon the above inclusion criteria. Two independent reviewers (HJ and KH) selected abstracts which appeared to meet the inclusion criteria. Full text articles of the selected abstracts were obtained and compared against the inclusion criteria. The full text, original research articles meeting the inclusion criteria were assessed for quality.

Quality assessment
Quality assessment was conducted by two independent reviewers (HJ and KH) using an adaption of The Cochrane Collaboration tool for assessing risk of bias as previously described [18,27]. For randomised studies the internal validity criteria included sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and selective outcome reporting. Whereas for non-randomised studies, adapted criteria were adopted that did not include criteria that was relevant to randomised studies only. Further internal validity criteria were also included to determine presence of cointerventions, intervention compliance, and if proper statistical analysis was conducted. External validity criteria were also included to determine restrictiveness of the inclusion and exclusion criteria, and to determine if the sample population was representative of the general population with JIA. For a study to achieve a high quality score overall, all included domains had to be scored independently as high quality by both reviewers. A study that was independently rated with ≥1 domain scored as low quality was rated as low quality overall. Any disagreement was resolved by a third reviewer (GJH).

Data extraction/evidence grading
The data extraction/evidence grading system was agreed upon by the co-authors a priori and is outlined as follows. Due to a lack of multiple RCTs for specific outcomes measures, meta-analysis was not possible and as such qualitative data synthesis was conducted. Once extracted data was analysed and synthesised, an evidence rating was assigned according to previously published criteria [28] ( Table 1). Once studies were rated for quality, they were grouped according to the outcomes measured and then the evidence grading system was adopted to assign levels of evidence for each outcome. This evidence grading criteria considers both quality and quantity of studies as well as the consistency of the findings for each outcome [28].

Results
Using the detailed search strategy, a total of 568 articles were retrieved. The inclusion criteria were met by twenty-one studies (see Figure 1), and a description of each study is presented in Table 2. The majority of the studies were observational in nature, with only one RCT identified for inclusion. Therefore, meta-analyses were not possible as were originally intended. One study had high quality internal validity and twenty had low quality internal validity. Nine studies had high quality external validity and twelve had low quality external validity. There were no studies with both high quality internal and external validity (Table 3). From the assessed studies, nine outcome domains were identified which included tenderness/pain, swelling, synovitis, effusion, hyperthermia, sustained 'clinical response', joint range of motion (ROM), length discrepancy (leg length discrepancy or joint circumference), and imaging detectable outcomes. Leg length discrepancy (LLD) was defined as unequal length of the lower limb from the knee distally, as a result of knee synovitis and/or tibial growth disturbances. Qualitative syntheses of results are outlined below and presented in Table 4.                [35] Knee: 0% detectable after 1 week and 1 month (no significance value given) One patient experienced focal cutaneous atrophy and a possible ruptured popliteal cyst Inconclusive due to lack of studies Synovial hypertrophy Laurell et al., [24] Talocrural joint = 87% regression vs. 13% no effect Local subcutaneous atrophy in 3 patients at 4 injection sites.

Tenderness/pain
Three observational studies investigated measures of tenderness/pain in the lower leg joints, inclusive of the knee and ankle [24,35,38]. While all studies showed a reduction in pain after utilising IACIs, no significance values were given [24,35,38]. A weak level of evidence was found for IACIs decreasing lower leg tenderness/pain. Weak evidence for IACIs decreasing tenderness/pain was also found specifically in the knee (2 studies) and inconclusive evidence in the ankle due to the lack of studies (1 study).

Swelling
Two observational studies investigated a reduction of swelling in the knee [35,38]. While no significance values were reported, there was weak evidence of IACIs decreasing swelling.

Synovitis
Three observational studies investigated a reduction of synovitis in lower limb joints inclusive of the knee ankle, STJ, MTPJ and IPJ [40][41][42]. Inter-quartile ranges (IQR) were only available from one study reporting the overall mean relapse time of 0.5 yrs (IQR 0.3-1.3 yrs) compared with the mean remission time of 0.9 yrs (IQR 0.6-1.9 yrs) [40]. No significance values were given for the other studies [41,42]. A weak level of evidence was found for IACIs decreasing synovitis in lower leg joints. Specifically, a weak level of evidence was found for decreasing synovitis in the knee, ankle and STJ as two studies showed a reduction. There was inconclusive evidence for all other lower limb joints including MTPJ and IPJ.

Effusion
Only one observational study investigated a reduction of effusion in the knee [38] and no significance values were given. There was inconclusive evidence for IACIs decreasing synovitis in the knee due to lack of studies.

Hyperthermia
Only one observational study investigated a reduction of hyperthermia in the knee [35] and no significance values were given. There was inconclusive evidence for IACIs decreasing hyperthermia in the knee due to lack of studies.

Sustained clinical response
Thirteen observational studies investigated a sustained clinical response post-injection in the lower leg joints, inclusive of the knee, ankle and foot [19,24,26,31-34, 36,37,39,42,44,46]. Sustained clinical response included measures of pain, oedema, erythema and general inflammation, and was determined over periods ranging from 1-66 months post-injection, depending on joints investigated. Four studies showed statistically significant sustained clinical response [26,34,37,46]. Nine studies also found sustained clinical response in relation to duration of improvement and reduction/resolution of pain, inflammation, limitation of ROM, joint tenderness, swelling, temperature, joint effusion, deformity, synovitis, contracture and gait abnormalities. However, these results were derived from descriptive statistical analyses only [19,24,[31][32][33]36,39,41,44]. Two studies found statistically significant sustained clinical response in the knee [34,37]. Six studies also found sustained clinical response in the knee. However, no significance values were given as data was presented using descriptive statistics only [31][32][33]36,39,44]. One study found statistical significance for the knee and ankle collectively [46]. Sustained clinical response was also found in three studies for the ankle [19,24,42], three studies for the STJ [19,26,42], and one study for the midfoot [19]. However, no significance values were given. Overall, there was weak evidence for IACIs decreasing clinical signs and symptoms in lower leg joints as thirteen studies showed a sustained clinical response post-injection.

Joint ranges of motion
One RCT and four observational studies investigated joint ROM post-injection. The efficacy of IACIs was investigated for the STJ [25], knee [30,35,38] and ankle [24]. However, an odds ratio was only reported by one study for a mean duration of improvement of 1.2 yrs (SD ±0.9 yrs) [25]. Weak evidence was shown for IACIs increasing ROM in lower leg joints collectively. The RCT showed an increase in knee flexion following administrations of triamcinolone hexacetonide (TH) and betamethasone (BM) [30]. TH increased knee flexion more than BM [43]. Two observational studies also showed improvement of knee ROM [35,38] and weak evidence was found for IACIs increasing ROM in the knee. One observational study showed increased foot inversion and eversion [25] and ankle joint ROM normalisation/partial improvement [24]. Overall, there was inconclusive evidence for IACIs improving joint ROM in the STJ and ankle.

Leg length discrepancy
One RCT and two observational studies investigated length discrepancies (LLD) [30,43,45]. Two studies specifically investigated LLD [43,45]. While both studies showed a decrease in the differences between limb length, only one was tested statistically and achieved statistical significance [43]. The other study only used descriptive statistics [45]. A weak level of evidence was found for decrease in lower LLD. One study also found a decrease in knee joint circumference; however, this was not statistically tested [30]. This particular study demonstrated that there was a decrease in knee joint circumference with use of TH or BM [30] but TH had a greater reduction compared to BM. Overall, there was inconclusive evidence for IACIs reducing knee circumference due to an insufficient number of studies.

Disease activity outcomes detected by imaging
Four observational studies investigated the ability of imaging techniques to detect the efficacy of IACIs on JIA disease activity [24,35,38,42]. These studies showed IACIs led to clinical effects on disease activity as detected by imaging techniques when compared to clinical assessment. However, these differences were not compared using inferential statistics. Overall, there was weak evidence for the use of imaging techniques detecting the effects of IACIs. Specifically, two studies investigated the effects of IACIs using MRI to effusion [35,38]. Three studies investigated the effects of IACIs using MRI to detect pannus [31,34,44]. There was weak evidence for IACIs decreasing MRI detectable effusion and pannus. One study investigated MRI for detecting the effects of IACIs reducing popliteal cysts, destruction of articular cartilage/bone, destruction of meniscus, bone marrow oedema and avascular necrosis [35]. One study investigated MRI for detecting the effects of IACIs reducing the uptake of contrast medium [38]. However, there was inconclusive evidence for these subdomains due to a lack of studies. Overall, the level of evidence was weak due to two studies showing the effects of IACIs using MRI. Two studies investigated the effects of IACS with musculoskeletal ultrasound imaging [24,35]. One study found decreases in signs of joint effusion and/or pannus and popliteal cyst [24], and one study found decreases in signs of synovial hypertrophy and synovial hyperaemia [35]. Therefore, the level of evidence assigned was weak for IACIs decreasing ultrasound detectable signs and symptoms, and inconclusive evidence for each specific subdomain.

Discussion
Weak levels of evidence were found for IACIs decreasing tenderness/pain, swelling, synovitis, clinical signs and symptoms, increasing joint (ROM) in lower limb joints and medical imaging disease-related outcomes such as joint effusion, synovial hypertrophy and erosion. Inconclusive evidence was present for IACIs reducing effusion, hyperthermia, LLD and knee circumference. The majority of studies were low quality observational studies with one low quality RCT, and as such these conclusions are based upon weak evidence only. Additionally, the majority of the studies identified by the search strategy presented results using descriptive statistics only, leading to uncertainty regarding the effectiveness of IACIs versus a suitable comparator.
The results of this contemporary review demonstrate similar issues raised by previous reviews concerning the lack of conclusive evidence in support of IACIs improving clinical outcomes [16,17]. The adopted methodology of the most relevant previous systematic review appeared to be logical in restricting the inclusion criteria to RCTs in an attempt to increase the quality of evidence retrieved. However, at the time, there was a lack of published RCTs available for inclusion in this review and it subsequently concluded there was insufficient research to ascertain the effectiveness of IACIs for JIA. Observational studies were not considered by the authors as part of the inclusion criteria for that review, and as such potentially important studies were excluded which may have provided at least a weak level of evidence concerning IACI efficacy. In contrast, our review included observational studies and found weak evidence to suggest that sustained clinical response and improved joint ROM can be achieved following administration of IACIs to the joints of the lower limb in children with JIA.
A narrative literature review which adopted systematic search criteria on all medical treatment modalities for JIA including IACIs has previously been conducted [10]. The authors concluded there was significant evidence of efficacy for the use of IACIs in persistent oligoarticular JIA and moderate efficacy of IACIs for the use in polyarticular JIA based upon the findings of three studies [30,46,47]. However, this was not a true systematic review as the appraisal strategy did not adopt rigorous methods of quality assessment or evidence grading, and their conclusions were based solely upon effect sizes greater than 50% of patients. The lack of evidence grading introduces ambiguity to the conclusion of such reviews.
Our systematic review has utilised a specific quality assessment criteria with emphasis on the internal and external validity of the included studies [18]. Our study showed that while there was only one observational study with high internal validity [42], nine observational studies showed high external validity [19,24,26,32,34,35,40,44,46]. This may be explained by the fact that recruitment was largely conducted alongside standard practice within paediatric rheumatology outpatient clinics, thus increasing the clinical relevance of the results. This external validity criterion was also necessary to assess whether the study methodologies were pragmatic, as opposed to explanatory [48]. The pragmatic nature of these studies meant their conclusions were more likely to be applicable to the general population of children/adolescents with JIA to whom IACIs may be a possible treatment option.
Due to the lack of multiple RCTs in this area, no metaanalysis could be conducted. This has implications for future research as high quality RCTs and meta-analyses are necessary to inform evidence-based clinical practice. Indeed a pragmatic RCT of IACIs versus a suitable comparator for the relief of localised synovitis could have sufficient external validity to inform and influence current practice and health policy in future. However, it is recognised that several clinical practice guidelines recommend use of IACIs in spite of the lack of high quality research evidence supporting their use [10,20]. As such, it is acknowledged that a placebo-or sham-control arm for a parallel trial of IACIs may not be possible due to ethical restraints surrounding the withholding of standard care. A contextual problem with the lack of robust RCTs to evaluate IACIs at present exists with regards to recent changes to the Australian Medicare Benefits Schedule [49]. Evaluations of the cost-effectiveness of interventions are normally embedded within RCTs. At present, and to the best of the authors' knowledge, an evaluation of the costeffectiveness of IACIs in JIA has yet to be conducted. This raises an important issue; because presently people with arthritic conditions in Australia are potentially required to self-fund the cost of each individual IACI they receive [49], and doctors cannot yet provide sufficient guidance regarding this treatment based on robust evidence of clinical-and cost-effectiveness.
Two studies are frequently cited in reviews of IACIs in JIA and clinical practice guidelines to draw conclusions concerning this intervention [46,47]. These prospective, randomised studies focus on the comparative efficacy of commonly used steroid preparations triamcinolone hexacetonide (TH) and triamcinolone acetonide (TA), concluding that TH was superior. However, a major limitation of these studies was that a non-treatment, placebo, or sham control arm by which to compare the effects of the IACI interventions was not possible due to ethical constraints. Moreover, participants included were not on stable medication for an appropriate period prior to enrolment, so both studies were vulnerable to confounding bias. New evidence has emerged since previous reviews have been published; however, limitations in these studies are still evident. The well documented restrictions of the commercial availability of TH resulted in this preparation being replaced by TA in at least one study included in this review [25]. A lack of standardisation of IACI interventions was identified amongst studies retrieved by our search criteria which may be problematic. Frequently studies do not describe their injection technique, which can vary between individual institutes [22]. Moreover, studies included in this review have utilised varying post-injection management strategies including physiotherapy rehabilitation [31], immobilisation [31,34], and a period of non-weight-bearing or avoidance of physical activity [26,37,40,45]. Despite this, other studies included did not describe a post-injection strategy [24,35,38,42,43]. These approaches appear to lack consensus, and at present there is a lack of evidence available to support a specific post-injection procedure.
A novel finding from this review was that several studies employed image-guidance to aid their IACIs [24][25][26]38,42]. Ultrasonography-guided injections also have been identified as a useful imaging technique for IACIs as they permit the real-time guidance of the needle and allow confirmation of correct deposition of steroid [24,35]. These studies concluded the use of imaging techniques ensured correct needle placement and guided IACIs were superior to standard 'blinded' injection and aspiration techniques. These studies also raise further questions surrounding imageguidance concerning whether or not image-guidance can reduce the rate of complications associated with IACIs. At present there have been no RCTs comparing guided and blindly administered IACIs in JIA.
The conclusions from this review contribute to the evidence base in support of current JIA management practices through aggressive and early intervention to reduce disease activity [11]. This highlights inconsistencies and evidence gaps concerning the efficacy of IACIs in JIA. It also provides valuable additional and contemporary information through a robust and systematic approach relative to previous reviews. While investigating the efficacy of IACIs, post-injection procedure was included in the study description as a contextual factor. This did not occur in previous reviews. Not all studies included a post-injection technique. Those studies that did include post injection procedure are described in Table 3. However, the lack of consistency among the described injection procedures makes it difficult for comparison to make a consensus. This is an area which requires further research.
There were a number of limitations for this systematic review that should be acknowledged. Language was limited to English which is generally not recommended, but is often difficult to overcome [50]. There were two identified studies which had to be excluded based on language barriers (Spanish and Russian languages) that could have potentially had an impact on the findings of this review. It is acknowledged that a number of studies were identified by hand-searching that were not identified using the detailed search strategy. The original search strategy was reviewed to investigate whether or not it was adequate. The minor problem with the search strategy was subsequently attributed to an indexing issue within the databases that could not be overcome by altering the search strategy. However, in spite of this limitation, the pragmatic nature of the reviewing strategy permitted important and relevant studies which met the inclusion criteria to be identified by hand.

Conclusions
In conclusion, there is conflicting and inconclusive evidence for the efficacy of lower limb IACIs for JIA. While there is enough quantity among the current body of knowledge to provide weak evidence in some outcome domains for generalised efficacy of IACIs administered to the lower limb, efficacy for duration of relief of synovitis in specific lower limb joints remains inconclusive. It is for this reason we conclude that more high quality studies are needed to definitively determine the efficacy of lower limb IACIs for improving outcomes associated with the JIA disease process.