Outcome in Juvenile Idiopathic Arthritis

A Population-Based Study From Sweden

Elisabet Berthold; Bengt Månsson; Robin Kahn


Arthritis Res Ther. 2019;21(218) 

In This Article


We have investigated the long-term outcome of JIA using this well-defined, population-based cohort of 251 children from southern Sweden with a validated diagnosis of JIA, all diagnosed in the era of biologic treatment. The mean annual incidence rate for JIA was 12.8 per 100,000 children. A considerable part of the children (10.8%) still develop uveitis, and the majority of the follow-up years (60.0%) are spent in a state of active disease. However, only 4.4% of the children require serious orthopedic corrective surgery such as arthrodesis, osteotomy, or arthroplasty.

We have found an annual incidence rate of 12.8 (95% confidence interval 11.3–14.5) per 100,000 children. This is a lower number than in previously published studies from Sweden and the Nordic countries (15/100,000).[6] However, our case collection process, with the collection of ICD codes covering arthritic and psoriatic diagnoses in childhood from the local diagnosis register as well as from the NBHW, diminishes the possibility of selection and referral bias.

Our incidence rate is also close to the incidence rate of 14/100,000, presented in a prospective population-based study carried out in the southeastern part of Norway, with almost an equal amount of children at risk as in our study area of southern Sweden. However, the purpose of the Norwegian study was to study the incidence of all arthritides in childhood.[21] The incidence rate of JIA published from Olmsted County, MN, 10.3/100,000,[7] is also close to our number. However, as we move down to southern Europe, the incidence rates of 6.9/100,000 in Catalonia (Spain)[9] and 3.1/100,000 in Alsace (France)[10] are distinctly lower than in our cohort. When looking at our results together with these selected previously published incidence rates, we bring validation to the suggestion that there seems to exist a north-south gradient in the incidence of JIA in Europe.[22] Genetic and environmental factors, as well as infectious agents, are risk factors for developing JIA, and a geographical gradient explains the combined impact of these factors.

With the retrospective approach of our study, we have the challenge of working with information already stated in the medical review, making it sometimes difficult to strictly apply the exclusion criteria stated in the ILAR definitions to the characterization of a patient. As presented in the "Methods" section, we used the presence of RF at one occasion as an inclusion criterion for patients who otherwise met the criteria for polyarticular disease, but not as an exclusion criterion in patients with a clinical manifestation of oligoarticular disease. This might give an overestimation of the RF+ group and a lesser portion of patients in the uJIA group, and there is also a risk of underestimation of patients in the uJIA group due to missing information about heredity. However, the diagnosis distribution in our cohort is similar to that in the Canadian ReAACh-out cohort, a large prospective JIA cohort for the purpose of studying the outcome in JIA patients. The ReAACh-out cohort comprises approximately 40% oligoarticular disease and 25% polyarticular disease but a larger amount of ERA (14.2% vs. 8.8%) and a smaller amount of uJIA (10% vs. 16.3%) than in our cohort.[23] The diagnostic subgroups in our cohort are also similar to the distribution in the Nordic cohort[6] except that they have classified 22% of the patients (vs. 16.3% in our study) as uJIA, suggesting that our possible lack of hereditary information is of some importance. The demographic information in our cohort is consistent with that in the cohorts mentioned above. In the ReAACh-out cohort, the median time from disease start to diagnosis was 4.3 months,[23] as compared to 5.0 months in our study and 6.6 months in the Olmsted county cohort.[7] A new set of classification criteria for JIA has been proposed.[24] How this will influence the outcome in the subgroups is uncertain. However, it could be of interest to see what impact these new criteria will have on the outcome in our current subgroups as well as to validate the criteria in this cohort.

The characterization of the medical treatment in our cohort shows that almost all children (98%) are prescribed NSAID. However, not only continuous treatment is registered but also NSAID prescribed to be taken in case of disease activity. Intra-articular corticosteroid injections were also an often-used treatment option in this cohort (78.9%), consistent with the current treatment recommendations. Due to the risk of growth retardation and other adverse effects with systemic glucocorticoids, intra-articular injections are favorable. Despite this fact, 42.6% of the total cohort is prescribed systemic glucocorticoids, but on 60.7% of these occasions, it was prescribed for a shorter duration than 2 months. A positive discovery was that 43.4% of the patients experience years of medication, six out of ten of the children with persistent oligoarthritis. Our results on medical treatment are in line with the results from another published prospective cohort from the Nordic countries (except for Iceland) with children included during 1997–2000.[25] As many as 96.1% of the children in this cohort were treated with NSAID and 74.1% received intra-articular corticosteroid injections. The treatment with methotrexate and TNF inhibitors was somewhat more unusual than in our cohort, 48.4% vs. 60.6% for methotrexate and 17.5% vs. 23.9% for TNF inhibitors, which we interpret as an effect of the 10-year difference of the inclusion periods.

Even though these patients all have been diagnosed with JIA in the treatment era of biological DMARDs, the presence of uveitis is 10%, the need for serious orthopedic joint corrective surgery is 4.4%, and 60% of the follow-up years are with active disease. The presence of uveitis in our cohort is however half of the prevalence in the Nordic cohort study with a corresponding number of median follow-up years,[18] which also in this respect interprets as an effect of the 10-year difference of inclusion period and the more common use of DMARDs in our cohort. There were no cases of uveitis in the RF+, systemic, or psoriatic subgroups in our cohort as well, which raises the question of the need for regular ophthalmologic controls in these groups of children. Most cases of chronic uveitis occur in the first 3 years of disease, but our data show that it still can develop after 9 years, at a time when Swedish regular ophthalmologic controls are sparse to once or twice per year, or even finished for most patients. Thus, we conclude that there is no absolute time or age limit to the end of risk for developing JIA-associated chronic uveitis.

It also seems that children with JIA today are in need of joint corrective surgery to a lesser extent than 20 years ago,[16] as a suggestive proof of more frequent use of methotrexate and the biologic DMARDs being effective in diminishing long-term effects of the disease. However, the children in our cohort have spent the majority of their time with inflammatory activity in the joints or eyes. This might partly be an effect of our decision to present the swollen and tender joint count as the total number of affected joints in the 66/68 joint count index that year, i.e., 1 minor arthritis at 1 check-up visit is considered as disease activity that year, but it is also close to the finding that only 47.5% of JIA patients achieve inactive disease at a median of 6.5 years.[13] Thus, a surprisingly large amount of children with JIA still do not achieve inactive disease with the arsenal of treatment options available today, but on the other hand, the functional impact is less evident than 20 years ago. One way to simplify the outcome presentation would have been to use JADAS measures to describe disease activity and disease remission, but this was unfortunately not possible because of the lack of both the parameters of physician global assessment of disease activity and parents/patient global assessment of well-being due to the retrospective study design.

There are limitations to our study. The most important limitation is the retrospective nature, with possible consequences discussed above. On the other hand, the strengths of our study include the population-based approach with minimal or no selection bias and inclusion of patients from all regional healthcare providers. The JIA diagnosis is validated for every patient, and the same training physician has made the validation, also diminishing inclusion bias. It is interesting to point out that as many as 32% of the cases were excluded as they were misdiagnosed as JIA. Of course, a part of the cases were arthritides diagnosed with an ICD code for JIA and labeled "suspected JIA" and later in the medical history reclassified as for example post-infectious arthritis, but a considerable part of these excluded cases turned out to be other conditions incorrectly coded and registered as JIA. In the study of incidence and prevalence of JIA in California (USA), Harrold et al. reviewed a random sample of medical records registered as JIA in order to develop the best case-finding algorithm for the study purpose. Out of the 97 selected records, 69% were determined to have JIA.[26] Thus, when using population-based cohorts of a disease without validation of the individual cases for the purpose of studying causality, the risk of diagnosis misclassification has to be taken into account when making conclusions on the result. The excluded number due to misclassification was higher than expected in our study.