Juvenile Idiopathic Arthritis in South Australia

Background: Juvenile Idiopathic Arthritis (JIA) is a chronic rheumatic disease that shows variation in subtype distribution by geographic regions. The aim of this study was to create a JIA registry, which could describe the epidemiology and characteristics of JIA patients in South Australia (SA). Methods: Prospectively collected data from JIA patients at the Women’s and Children’s Hospital in Adelaide, between August 2019 and March 2020 were reported. All children and young people diagnosed with JIA according to the International League of Associations for Rheumatology (ILAR) classication criteria were eligible for inclusion. Data including demographics, complications/comorbidities, medications, disease activity as well as patient-reported data using Childhood Health Assessment Questionnaire (CHAQ) were documented. Non-identied data were extracted from this registry and utilised in descriptive analysis and comparative studies. Results: There are currently n=112 JIA patients in this registry, including n=11 incident cases (9.8%), with a predominance of female (n=75, 67.0%). The median disease duration was 3.6 years (interquartile range 1.3-7.6). The most common subtype was persistent oligoarthritis (n=35/112, 31.3%), followed by rheumatoid factor-negative polyarthritis (n=32/112, 28.6%), extended oligoarthritis (n=23/112, 20.5%), enthesitis-related arthritis (ERA; n=12/112, 10.7%), systemic onset arthritis (n=5/112, 4.5%), rheumatoid factor-positive polyarthritis (n=3/112, 2.7%) and psoriatic arthritis (n=2/112, 1.8%). Complications were documented in n=40/112 (35.7%) patients, with n=21/94 (22.3%) observed to have uveitis, and musculoskeletal complications observed in n=10/112 (8.9%). Methotrexate intolerance was present in n=10/79 (12.7%) patients. Forty-nine out of 112 (43.8%) patients had at least one comorbidity, with anxiety/depression (n=5/112, 4.5%) and asthma (n=5/112, 4.5%) being the most common. Around half of patients (n=52/112, 46.4%) had clinically inactive disease at enrolment and most (n=85/112, 75.9%) reported no to low functional disability. Non-steroidal anti-inammatory drugs have been almost universally used (n=100/112, 89.3%), and n=26/112 (23.2%) have received treatment with biologics. This study reports the baseline characteristics of our We conducted our study was for included anxiety and depression (n = 5, 4.5%), asthma (n = 5, 4.5%), diseases of the skin and subcutaneous tissue (including paronychia, eczema and psoriasis, n = 4, 3.6%), autism spectrum disorder (n = 4, 3.6%), iron deciency/anaemia (n = 4, 3.6%), developmental delay (n = 4, 3.6%), eye disease (n = 3, 2.7%), IBD (n = 2, 1.8%), coeliac disease (n = 2, 1.8%), Type 1 Diabetes (n = 2, 1.8%), chondromalacia (n Sinding-Larsen Syndrome (n Epilepsy (n and short stature (n 1.8%). Other comorbidities were also seen in very few patients, including but not limited to Lymphoedema (n = 0.9%), Hypothyroidism (n = 1, 0.9%), congenital heart disease (n = 1, 0.9%), Cardio-facio-cutaneous Syndrome (n = 1, 0.9%), Sever's disease (n = 1, 0.9%), disease (n = 1, 0.9%), and hypertrichosis (n = 1, 0.9%).


Study design
The JIA registry in SA is a single-centre, observational and prospective registry with a view to improving the long-term clinical care and patient outcomes. This study reports the baseline characteristics of our SA registry cohort. We conducted our study with the approval of the Women's and Children's Health Network Human Research Ethics Committee (HREC/19/WCHN/59) and in accordance with the Declaration of Helsinki. Written informed consent was obtained for all participants.

Patient cohort and recruitment
This cross-sectional study has captured and analysed baseline patient data and JIA outcomes in SA, in the period between August 2019 and March 2020. Children and young people with JIA were recruited during a visit to the Paediatric Rheumatology outpatient clinic at the Women's and Children's Hospital (WCH), SA, Australia. The WCH is the major paediatric hospital in SA and provides a state-wide paediatric rheumatology service.
The identi cation of JIA cases was made by three paediatric rheumatologists and the rheumatology nurse in the clinic setting based on the International League of Associations for Rheumatology (ILAR) criteria [23]. Incident cases were de ned as those enrolled within six months of initial diagnosis. Prevalent cases older than 18 yrs were excluded, as this is the age of transition to Adult services in our centre.
In some countries, many relevant studies and data are available, while in others, data are scarce or non-existent. The prevalence and incidence of JIA have been found to differ in different ethnic groups and regions [5,7,8], the knowledge of which is crucial for local and worldwide understanding of disease epidemiology and management. The lowest prevalence (3.43 per 100,000) was reported in Egypt and the highest (196 per 100,000) was reported in South America [9,10]. In the US, the prevalence of JIA from different published studies has ranged from 1.6 to 86.1 per 100,000 [11,12]. Studies presented a relatively higher prevalence rate in developed countries, varying from 16 to 150 per 100,000 [13]. However, it should be noted that the current low reported prevalence in developing countries may be related to limited data sources [14]. Published incidence rates for JIA in European countries ranged between 3.5 and 21.7 per 100,000 children under 16 years old [15][16][17][18][19].
In Australia, there are no published studies describing the incidence of JIA. A community-based prevalence study in JIA by Manners et al (1996), reported a prevalence of 400 per 100,000 in a cohort of 2,241 12-year-old children in Perth, Western Australia [20]. The 2004-05 Australian Institute of Health and Welfare (AIHW) reports a prevalence of 100 to 400 per 100,000 Australian children. Moreover, the point prevalence was estimated to exceed 400 per 100,000 among children aged 10-15 years in their data [14]. Both estimates were far higher than found in other published studies [9][10][11][12][13][15][16][17][18][19].
The only current JIA registry in Australia -The Childhood Arthritis Risk Factor Identi cation Study (CLARITY) has the primary aim of analysing DNA and environmental data in order to identify risk factors for JIA [21]. An inception cohort study with comprehensive epidemiological data for newly diagnosed JIA patients was performed between 2010 and 2014, to help initial understanding of the demographics, treatment and outcomes of Australian children [22]. Their results have shown a high similarity in patient demographics, disease course and treatment with cohorts in North America and Europe [22].
In this study, we aimed to create the rst JIA registry in SA which describes the epidemiology and characteristics of the SA JIA cohort. This population-based registry will help us improve patient care and assist in the development of collaborations with other national and international registries by harmonising datasets, sharing data and identifying shared challenges.
Once patients were identi ed as eligible to participate in the registry, they or at least one parent/carer provided written informed consent before the collection of any data. Patients or families unable to provide informed consent were excluded.

Data collection
Upon obtaining consent, treating clinicians were asked to complete a paper-based questionnaire regarding JIA subtype and relevant clinical data. Data including but not limited to demographics, JIA subtype, disease onset, disease duration, complications/comorbidities, medications used as well as investigation results and the clinical Juvenile Arthritis Disease Activity Score (cJADAS) [24] were documented. Patients/carers were asked to complete the paper-based Childhood Health Assessment Questionnaire (CHAQ) [25] at each clinic visit. All data were recorded prospectively in the JIA registry.

Registry development
The registry has been manually curated and stored in a password protected database, accessible only to the principal investigator (MM). Additionally, a newly generated study ID number is applied to each participant so that patient data are kept con dential but re-identi able. To minimise the potential misclassi cation of patients, a 6-month period has been provided to ensure an accurate clinical diagnosis, prior to the nalisation of statistical analysis. Furthermore, datachecking has been performed by the principal investigator (MM) to ensure data consistency, and any changes to the stored data are recorded in a log le.

De nitions of disease activity states
The cJADAS has been shown to have good construct validity in clinical practice in the description of disease activity in JIA [26]. In accordance with Consolaro et al. in 2014 [24], children with inactive systemic arthritis, rheumatoid factor negative (RF-) and rheumatoid factor positive (RF+) polyarthritis, or extended oligoarticular onset disease (EO) were included in the functional polyarthritis category. The functional oligoarthritis category included children with persistent oligoarticular onset disease (PO). Children classi ed in other ILAR subsets were assigned to the functional oligoarthritis or polyarthritis categories based on the number of affected joints (≤ 4 or > 4 joints, respectively) within the rst six months of diagnosis.

Statistical analysis
All analyses were calculated from the extracted data at the rst registry visit and were conducted using SAS 9.4 (SAS Institute Inc., Cary, NC, USA). In Table 1, categorical variables are presented as frequencies and percentages, and continuous variables are presented as mean ± standard deviation (SD) or median (interquartile range, IQR), as appropriate. The Shapiro-Wilk test was used to check the assumption of normality for a variable, thus deciding whether the parametric or nonparametric test was correspondingly carried out. The Wilcoxon Rank Sum Test was used to compare the differences between two groups and the Spearman's coe cient was used to explore the correlations between two variables. In Table 2 and Table 3, the outcomes, including disease activity, functional disability and medication exposure, are reported by frequency and percentage in all patients and within ILAR subtypes. The comparisons with other published data were performed via MedCalc for Windows, version 19.2.6 (MedCalc Software, Ostend, Belgium). Results with p values less than 0.05 were considered as statistically signi cant.

Disease course
At the time of recruitment, the median active joint count was zero (IQR 0-0.8), the median physician global assessment (PGA) was zero (IQR 0-1.1) and the median patient global evaluation (PGE) was 0.5 (IQR 0-2.4) for all JIA patients (Table 2).
Among n = 101 patients with established disease, the median number of active joints was zero (IQR 0-0), with n = 80/101 (79.2%) achieving an active joint count of zero at enrolment, with median PGA of zero (IQR 0-1.0). In contrast, n = 11 incident patients had a median of one active joint count (IQR 0-2.0), with a higher median PGA of 1.2 (IQR 0-1.9).
Based on the aforementioned cJADAS cut-offs [24], n = 18 (16.1%) children had high disease activity at the time of recruitment, with n = 10 of them diagnosed with PO and one with EO. Six had RF-polyarthritis and one had RF + polyarthritis. Patients with EO had signi cantly longer (p = 0.016) disease duration [median 7.5 yrs (IQR 3.4-9.9)] than those with PO [median 3.9 yrs (IQR 1.2-6.4)].
The median CHAQ score (disability index) was 0.1 (IQR 0-0.6); most (n = 61/112, 54.5%) had no or only mild disability. The distribution of the CHAQ score was highly skewed (mean 0.37) [37]. When the analysis of CHAQ score was limited to those within subtypes, no signi cant differences were found.
The median age at onset among patients with functional oligoarthritis (n = 40, 35.7%) and patients with functional polyarthritis (n = 72, 64.3%) was 3.6 yrs (IQR 1.8-8.9) and 6.2 yrs (IQR 2-9.9), respectively. There was a signi cant difference in age at recruitment between functional oligoarthritis and polyarthritis categories of JIA (p = 0.04). There was no signi cant difference in age at diagnosis across the two categories (p = 0.25). The median time from symptom onset to diagnosis was 0.3 yrs (IQR 0.2-0.9) and there was no signi cant between-group difference (p = 0.571, Mann-Whitney U Test).
Of those patients with uveitis in our cohort, one had a cataract removed surgically and one had previously developed glaucoma. Patients with systemic onset disease, ERA, PsA and RF + polyarthritis had no uveitis. In this registry, musculoskeletal complications occurred in n = 10/112 (8.9%) patients, with the majority (n = 8/10, 80%) in the functional polyarthritis category. Musculoskeletal complications in this registry included micrognathia in four, xed exion deformity in three, erosive disease in two, joint damage in one, a pars defect in one, and muscle wasting in one patient.
Associations between disease activity and patient-reported data There was a signi cant positive correlation between CHAQ score and cJADAS (Spearman R = 0.446, p < 0.001), and between patient/carer-reported pain visual analogue scale (VAS) and cJADAS (Spearman R = 0.700, p < 0.001), which indicates a consistency of patient-and clinician-assessed disease activity. A negative correlation was also found between disease duration and PGA (Spearman R=-0.206, p = 0.029).

Discussion
In the SA cohort, we have developed a JIA registry and here we describe our cohort in detail. Our ndings regarding sex distribution, disease activity and patient-reported data support the data of comparable populations in some geographical regions such as North America and Europe [39][40][41][42][43][44]. However, there are discrepancies in JIA subtype distribution and complications such as uveitis and MTX intolerance between our cohort and those in other published studies [22,30,42,45,46].
The majority of the population of SA is Caucasian. However, ethnicity data were not recorded in this registry. The main subtype distribution differences between our cohort and other published studies were seen in PsA, RF-polyarthritis and EO (Table 4) [39-41, 46, 47]. In the present study, PO was the most widely distributed subtype (n = 35/112, 31.3%), which is consistent with the United Kingdom (UK; n = 638/1415, 45.1%) and four Nordic countries (i.e. Denmark, Finland, Sweden, and Norway, n = 132/440, 30%) [39,40,47]. However, North American, Latin American and South African children are more likely to develop polyarticular onset JIA [42,46,48]. PsA, the least common subtype in our study, is very uncommon in all regions, including Asia. Among Asian children, a particularly higher frequency of ERA was observed (around 36%, p < 0.001) than in other areas [49,50], whereas around ten percent of patients had ERA: 10.7% in this study, 11.1% in Nordic countries (p = 0.904 when compared with this study) and 5.4% in the UK (p = 0.0208 when compared with this study). The frequency of systemic onset JIA was the highest in Asia, North America and Latin America, and had similar frequencies in Australia and Europe (4.5% in this study versus 4.1% in Nordic and 6.8% in the UK). Compared to North America, the rate of functional polyarthritis (as per Consolaro et al. [24]) was signi cantly lower in our registry (n = 72/112, 64.3%) versus n = 895/1192 (75%) in the Childhood Arthritis and Rheumatology Research Alliance registry (CARRA, p = 0.0135) [42]. The differences in subtype distribution remain unexplained, but may relate partly to small patient numbers in our study to date. Among Australian children, there was a signi cant difference in the frequency of EO compared with CLARITY data (p < 0.001) [22]. It can be partly explained by the fact that 13% of patients from the CLARITY cohort had undifferentiated arthritis at enrolment with further classi cation awaited after 12-month review.
Different sex ratios according to ethnicity and regions have been shown in other epidemiological studies of JIA. Even though there is no clear gender distribution trend in African and Asian patients [8], our data have a similar pro le to that of European and North American regions which is predominantly female [39,40,42]. Among Australian children, the sex ratio in this study is also comparable with CLARITY data, in which 67% versus 62% of patients were female (p = 0.416) [22].
In a cross-sectional PRINTO study describing JIA cohorts from various geographic areas, Western Europe, Southern Europe, and North American cohorts were found to have very similar median disease duration of around 4.1 years. In our study, n = 101 patients with established disease had a similar cJADAS compared to those in Southern Europe and North America (p = 0.4657 and p = 0.1275) [41]. However, a signi cantly higher median cJADAS was observed in the Western European cohort (p = 0.0086) [41]. Similarly, fewer Western European JIA patients had inactive disease (p = 0.0204). More Australian children's PGAs were rated as zero by the physician than Western European (p = 0.005) and North American cohorts (p = 0.013). However, fewer Australian children/parents reported PGE as zero than did those Southern European (p < 0.001) or North American (p = 0.001). In addition, Australian children/parents were more likely to score the pain VAS > 0, compared to those in Southern Europe (p < 0.001) and North America (p = 0.0418) [41].
Disease outcomes for newly diagnosed patients in this registry were compared with those in other cohorts. In our cohort, n = 11 children with newly diagnosed JIA had a median PGA of 1.2, a median PGE of 1.3, a median CHAQ score of 0.4, and a median pain VAS of 3.0. Similar results were assessed in Nordic incident cohorts of 423 patients (p > 0.05 for all, except p = 0.0291 for pain VAS) [43]. In comparison with the 295 newly diagnosed JIA patients from the recent CARRA study, SA incident patients had a similar median joint count (p = 0.37) and median CHAQ score (p = 0.9), despite small patient numbers [42]. However, a signi cantly lower PGA was observed in our cohort than in North American children [median 1.2 (IQR 0-1.9) versus median 3 (IQR 1.5-5), p = 0.0232]. These data suggest that Australian children with newly diagnosed JIA may have lower disease activity despite similar physical function and Joint-related symptoms than those in North America. Other patient-reported data, such as PGE and pain VAS, were not recorded in the CARRA study. In our study, the median cJADAS for the 11 incident patients was 4.9 (IQR 0.1-10). Our incident cohort had a similar disease activity compared with the Canadian inception cohort study which reported a median cJADAS of 6.5 (IQR 4-10) at enrolment (p = 0.2707) [44].
Worldwide, the prevalence of JIA-associated uveitis is reported to be up to 30% [1,41]. In the SA registry, n = 21/94, 22.3% of patients had uveitis, which is similar with Nordic (n = 89/440, 20.2%) and signi cantly higher than German (n = 406/3271, 12.4%, p = 0.0045) and North American (n = 94/1175, 8%, p < 0.001) populations [40,42,45]. Asian and African populations have signi cantly lower rates of uveitis, compared with other areas. This may be because fewer patients had oligoarthritis and were positive in ANA, both of which are strong predictors of uveitis [51]. The lowest rates of n = 19/379 (5%) and n = 44/726 (6%) were recorded in India and Japan (p < 0.001 for both, when compared with this registry) [41,52]. In our SA cohort, Uveitis was detected in n = 4/31 (12.9%), n = 7/27 (25.9%) and n = 10/22 (45.5%) of the PO, RFpolyarthritis and EO populations, respectively. This result is in accord with the highest uveitis risk in EO subtypes in the large cohort study of 3271 patients in Germany [45]. In our study, both active and previous episodes of uveitis were recorded at enrolment.
Musculoskeletal complications occurred in n = 10/112 (8.9%) patients in the current study, which included joint deformity, damage, erosive disease and muscle wasting. No comparisons could be made because of the difference in methods of assessment. For example, joint damage rate ranged from 9.7-14.7% in European and North American countries, measured by Juvenile Arthritis Damage Index (JADI)-Articular > 0 [41].
In a recent systemic JIA cohort study in Turkey [27], the frequencies of growth retardation and MAS were 11.3% (n = 19/168) and 11.9% (n = 20/168), respectively. In the German Biologics JIA Registry (BIKeR), 4.5% of their systemic JIA patients experienced MAS [53]. However, it is di cult to make a comparison with our data because of the comparatively much smaller number of our systemic JIA patients, which included one patient with growth retardation and one patient with MAS. Notably, some common complications in other populations were not found in our cohort. We did not see amyloidosis or hip replacement in the SA population [54,55].
Among n = 79 patients treated with MTX in this study, n = 10/79 (12.7%) had MTX intolerance, with n = 2/79 (2.5%) patients experiencing MTX-induced liver function test derangement. A higher proportion of MTX intolerance (gastrointestinal and psychological symptoms) and elevated liver enzymes was found in a recent UK study: n = 151/577 (26%, p = 0.0099) and n = 56/577 (10%, p = 0.0298), respectively [30]. Selection bias my account for some of this difference because we included all patients treated with MTX regardless of whether or not they were receiving other DMARDs or biologic therapy, whereas the UK study included patients who were treated with MTX alone. However, we found no differences between this study and the German Methotrexate Registry in the frequency of elevated liver enzymes; n = 2/79, 2.5% versus n = 15/411, 3.6% (p = 0.6227) [56]. To our knowledge, comorbidities have not been reported in more recent JIA cohort studies and so no comparisons with our cohort could be made.
The main strength of our study is that we have recruited both newly diagnosed JIA patients as well as those with established disease, which reduces selection bias. Retrospective collection of data from patients with established disease is also important as it provides the ability to document current disease activity as well as disease management. In this study, we have obtained detailed patient data from a retrospective review of the case notes as well as recording real-time data such as physician assessment, patient evaluation, medications and blood test results and prospective data.
Consultation with the treating physicians and strict adherence to the ILAR criteria have enhanced the integrity and validity of our data, which will also facilitate future local and international collaborations.
Compared to the European and North American cohort studies [39][40][41][42][43]56], our study has a smaller sample size (n = 112). This is mainly because of the relatively short period of time since registry inception, with data from a single centre. We will continue to collect prospective data in our registry with ongoing patient recruitment of both newly diagnosed and established cases of JIA. This will assist in ongoing improvement of patient care as well as providing more accurate data to estimate the prevalence and incidence of JIA in South Australia. Another limitation is that the complications and comorbidities were extracted retrospectively from medical records, at times without a clear record date of onset. Some children had therefore already recovered from the complication or comorbidity at the time of recruitment. However, collection of these data are important in the assessment of disease activity and outcomes.

Conclusions
This is the rst JIA registry study conducted in SA. The epidemiological data and characteristics of JIA patients, such as subtype distribution, sex distribution, disease course and activity, major complications and comorbidities, medication exposure and patient-reported data were recorded. We found that the JIA cohort in SA is female-dominated with PO as the most common subtype, consistent with other European and Australian studies. Most prevalent patients had good disease control and had no or mild disability. One-third of the patients had complications and comorbidities occurred a greater proportion (43.8%). NSAIDs were used in most patients and biologics are the least-used, which is consistent with our subtype distribution. Patient-reported functional disability and pain were positively correlated with clinical assessment of disease activity. Further longitudinal data collection and comparisons will add to the current description of the SA JIA