A systematic review and meta-analysis regarding the use of corticosteroids in septic arthritis

Background Chondral damage is one of the major sequelae of septic arthritis; occurring even after prompt treatment of a septic joint. Subsequent loss of joint function can have a significant impact on a patient’s quality of life. Corticosteroids are known to have beneficial effects on the rate and extent cartilage destruction in arthritis through a variety of mediators such as synovial RANKL expression, mast cells and pro-inflammatory cytokines. Investigation into sepsis at other sites has suggested improved outcomes with corticosteroid use despite the theoretical risks. This study therefore set out to review current literature with regards to a possible beneficial effect for corticosteroids in Septic Arthritis. Methods A computerised search of the databases MEDLINE and CINAHL was conducted during November 2014 using the EBSCOhost web search engine in order to identify research articles relating to the use of corticosteroids in the treatment of septic arthritis. The search strategy revealed 223 unique articles which were subjected to inclusion/exclusion criteria assessment. 6 articles were selected for study inclusion. These consisted of 3 human studies (2 double-blind randomised controlled trials & 1 double-blind non-randomised controlled trial), and 3 animal studies (3 non-blinded non-randomised controlled trials). Quantitative synthesis (meta-analysis) was only possible regarding two primary outcomes for two of the included studies – time to normalisation of CRP and duration of IV antibiotic therapy. Results All current published evidence in humans is focused upon children. Overall results did however reveal a consensus between these studies for a reduced duration of symptoms and a reduction in inflammatory markers. Animal data suggested a protective effect on the articular cartilage with the addition of corticosteroids to antibiotic therapy. No article noted an adverse effect associated with steroid use. Findings were consistent with systematic reviews of corticosteroid use in other bacterial infections. Conclusions Despite the promising outlook, issues’ regarding generalisability of results and a lack of large randomised controlled trial data necessitates further assessment of the safety and efficacy of steroid use in adults before treatment recommendations can be made. Long term safety data and the determinations of the optimum route, dose and timing of corticosteroids are also required.


Background
Despite modern antibiotic therapy Septic Arthritis still carries a significant associated morbidity (31.6 %) and mortality (11.5 %) [1]. The higher incidence of infection in those of a younger age enhances the magnitude of this issue. Joint contamination typically occurs through haematogenous spread but direct inoculation can occur with intra-articular procedures. Clinical presentation is usually of a monoarthritis with systemic sepsis. The knee is most commonly affected, accounting for approximately 50 % of cases [2].
A vast array of pathogenic organisms have been identified; with gram positive cocci such as Staphylococcus and Streptococcus are the most commonly seen [3]. Once colonisation has occurred rapid proliferation of bacteria ensues initiating an acute inflammatory response. There is activation of Toll-like receptors with substantial release of pro-inflammatory cytokines such as Interleukin (IL)-1 beta, Tumour Necrosis Factor (TNF)-alpha, IL-17 and IL-6 [3]. These molecules stimulate osteoclast differentiation and bone reabsorption as well as matrix metalloproteinase (MMP) release leading to subsequent bone and cartilage degradation [4]. Presence of a joint effusion secondary to increased synovial fluid production increases intra-articular pressure which mechanically impedes blood/nutrient supply to the joint and further exacerbates cellular damage [2]. There is rapid progression and necrosis/degradation of chondrocytes within the first 24 h of onset [5,6].
Corticosteroids are one well-established method of reducing joint inflammation, having being used in rheumatoid arthritis for decades [7]. Pharmacodynamic effects are mediated through binding to glucocorticoid receptors; stimulating targeted protein upregulation (in particular Lipocortin). There is subsequent inhibition of Phospholipase A2 (a potent intracellular producer of prostaglandins, free radicals and leukotrienes) as well as a deleterious effect on several pro-inflammatory cytokines such as IL-1, Interferon (IFN)-γ and TNF-alpha. Other functions include inhibition of elastase and collagenase production, decreased T-cell number and reduced synovial MMP-1 expression [8].
The use of glucocorticoids in infection has classically been contra-indicated due to a theoretical risk of worsening sepsis. Recent evidence regarding the use of corticosteroids alongside antibiotics in infection at other sites has however brought this suggestion into question. Studies into severe sepsis [9], severe pneumonia [10], bacterial meningitis [11] and acute pyelonephritis [12] have all shown improved outcomes associated with steroid use.
Current research therefore suggests evidence for both a positive effect of glucocorticoids on joint destruction in arthritis, and improved clinical outcomes in infection when combined with antibiotic therapy. It is hypothesised that the combination of corticosteroids and antibiotics could be of benefit when considering Septic Arthritis. This study therefore sets out to examine the available evidence on this research topic in the form of a systematic review and meta-analysis.

Methods
A computerised search of the databases MEDLINE and CINAHL was conducted during November 2014 using the EBSCOhost web search engine. This identified research articles relating to the use of corticosteroids in the treatment of septic arthritis. Search criteria included combinations of the terms "steroids", "hydrocortisone", "corticosteroids", "dexamethasone", "triamcinolone", "prednisolone", "glucocorticoids", "septic arthritis" and "joint infection" via a Boolean/Phrase strategy applied to the full text article and abstract. An outline of the full electronic search strategy is shown in Table 1.
Search restrictions included English language text with a publication date 1963-2014. Further potentially relevant literature was identified through bibliographical analysis of identified articles and using an internet search engine (Google). Initial screening of potential articles was performed through an independent abstract review. Relevant articles were then discerned from those initially identified with a thorough independent full text analysis. Inclusion and exclusion criteria were used to ensure any conclusions drawn from the literature were valid and based on a sound methodical construct. Inclusion criteria comprised: interventional articles relating to the concomitant use of corticosteroids and antibiotics in the treatment of septic arthritis; presence of a control group or cohort; clearly stated inclusion and outcome criteria; assessment of results using a validated appropriate statistical test and ethical approval. Exclusion criteria included: No full text availability and review articles.
Quality assessment was performed via the Cochrane Collaboration's risk of bias tool. No formal review protocol was produced prior to the assessment process. All research was conducted in accordance with current international ethical standards [13].
Data extraction was carried out via a manual independent assessment of each article. Variables which were sought included study design, study aim, sample size, subject demographics, intervention, primary outcome measures, results, and author's conclusions. Meta-analysis of pooled outcome measures was performed where possible (≥2 homogeneous randomised controlled studies with at least one matched primary outcome variable) using the Cochrane collaboration's RevMan 5.3 tool [14].
The written report has been produced in concordance with Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [15].

Results
The search strategy revealed 223 unique articles which were subjected to inclusion/exclusion criteria. Further abstract analysis identified 11 of these as potential candidates for review. A total of 6 articles were selected for study inclusion. A flowchart diagram of the complete study selection process is shown in Fig. 1 of 3 human studies in children (2 double-blind randomised controlled trials & 1 double-blind non-randomised controlled trial), and 3 animal studies (3 non-blinded non-randomised controlled trials). No interventional human adult studies were available for analysis. All studies showed a positive effect in at least one primary outcome for the addition of corticosteroids to antibiotics in the treatment of septic arthritis. A total of 349 participants were involved in the analysis (209 humans, 57 rabbits and 83 mice). The mean age of human participants was 5.64 years. Individual study characteristics and results are detailed in Table 2.   Quantitative synthesis (meta-analysis) was only possible regarding two primary outcomes for two of the included studiestime to normalisation of CRP (Fig. 2) and duration of intra-venous (IV) antibiotic therapy (Fig. 3). A risk of bias table detailing the author's judgements about each risk of bias item for each included study is shown in Fig. 4.

Discussion
There was a significant lack of adult human interventional trial data available. All studies did however show a positive effect in at least one primary outcome for the addition of corticosteroids to antibiotics in the treatment of septic arthritis. These improvements were shown across a variety of cellular, histological and clinical levels. Meta-analysis regarding normalisation of CRP and duration of IV antibiotics was performed with statistically significant trends favouring the addition of Dexamethasone vs. placebo for both. A lack of available studies and participant numbers to provide substantial analysis were however noted. No study documented adverse events associated with glucocorticoid use. These results are in keeping with those examining the addition of corticosteroids to antibiotics at other anatomical sites. Whilst these results are promising, there is a lack of large randomised controlled trial evidence with long term follow up and safety data in adult human subject's. As such there is currently insufficient evidence to recommend corticosteroids alongside antibiotics for the treatment of septic arthritis.
The studies identified during the review process were of varying methodological quality. There were two doubleblind randomised controlled trials identified during the search process [16,17]. Both of these used a placebo controlled design to provide evidence for a positive effect in both biological and clinical parameters for dexamethasone in addition to antibiotic therapy. The numbers involved were however small, and there were some flaws identified in the methodological design of both studies. For example both articles failed to document if there was blinding of outcome assessment; determine whether appraisal was performed on an intention-to-treat or per-protocol; perform any formal safety analysis; and account for the impact of surgical intervention on outcomes. In addition despite documentation of duration of hospitalisation as a primary outcome in Harel et al. 2011 [16] the results and statistical analysis were not included in the published article. This study also failed to perform any statistical assessment of long term outcomespossibly due to significant participant attrition at 12 months follow up.
The other human study by Arti et al. 2014 [18] also found a positive effect for Dexamethasone therapy alongside antibiotics with regards to number of days of hospitalisation, clinical outcomes, and laboratory findings. This study had a number of major methodological flaws such as lack of randomisation, failure to compare baseline characteristics between groups and no long term follow up. It was therefore felt to provide minimal additional information to the previous literature.
The remaining studies focused upon the use of corticosteroids alongside antibiotics in an animal model. These allowed for direct analysis of histopathological parameters not easily assessed in humans due to ethical concerns. All studies showed an improvement in arthritis severity with the addition of glucocorticoids to antibiotics. Despite the poor methodological quality of the research and the lack of direct applicability to humans these articles do provide valid scientific rationale for the use of corticosteroids in septic arthritis.
There are still however a number of unanswered questions regarding the use of corticosteroids in Septic Arthritis. Importantly the concerns coming from a previously identified association between the use of glucocorticoids and an increased risk of infection [19] have not yet been fully addressed. Although no study assessed during this analysis identified any safety issues during follow up this was not formally evaluated at any point. A recently published abstract did report safety during a retrospective analysis regarding the use of corticosteroids in septic arthritis with no side effects noted except for 3 cases of symptom rebound after completion of the steroid course. This is a previously established side effect of steroid withdrawal [20]. It is also important to consider the significant long term effects of steroid use such as osteoporosis, diabetes, raised blood pressure, weight gain, mood changes and gastric irritation. It is felt however that the duration and dose of corticosteroids used in this scenario (such as the 4 days of IV Dexamethasone used in Odio et al. 2003 [17] and Harel et al. 2011 [16]) would be unlikely to have any significant long term effect. The optimum route, duration and dose of glucocorticoids need to be determined as well. Comparisons between intramuscular and intraarticular corticosteroids in the Stricker et al. 1996 [21] study did appear to favour an increased effect with the parenteral route. No study has examined the oral route of administration.
This study has also highlighted issues regarding the generalisability of current evidence. So far all studies have focused upon a child population and therefore the impact on and adult cohort is still unknown. It can be hypothesised that known differences in immunology between varying age groups [22] could impact on the glucocorticoid response.

Conclusions
The current evidence base points strongly towards a beneficial effect for corticosteroids in septic arthritis. This is consistent with scientific rationale and data regarding glucocorticoid use in sepsis at other anatomical sites. There is however insufficient evidence to make treatment recommendations at present. Further large scale randomised controlled trials including adult subjects; safety profiles and long term efficacy are required for this promising therapeutic option.