In this study we showed that individuals with JIA diagnosed in the biological era in the population-based south-Swedish JIA cohort have an increased risk of being diagnosed with a comorbid autoimmune condition compared to age- and sex-matched references from the general population. ANA positive disease was associated with a further increased risk. Our results suggest a need of laboratory screening for autoimmune conditions, particularly coeliac disease and hypothyroidism, within the first 7 years after JIA diagnosis and urges physicians to investigate symptoms indicative of comorbid autoimmune disease in individuals with JIA, especially among those with ANA positivity.

In our study, the risk of a comorbid autoimmune disease was elevated the first years after diagnosis. The shift in the risk around 7 years after JIA diagnosis does not only have clinical significance, but is also interesting when comparing to the occurrence of iridocyclitis in JIA. It is well known that the risk of iridocyclitis, or uveitis, is increased in the patient group of ANA positive disease. Furthermore, iridocyclitis usually develops the first years after JIA diagnosis, and mostly during the first 5–7 after JIA onset [1, 22]. That the risk of both comorbid autoimmune disease and iridocyclitis seems to be elevated during the same time periods raises interesting questions about the pathogenesis of JIA and warrants future studies.

Individuals with manifest JIA were diagnosed with coeliac disease more than twice as often as the age- and sex-matched reference population. The presence of coeliac disease in our cohort is in line with previously published Swedish point prevalence in JIA of 2.8% by Öman, et al. In their study, they screened JIA patients with autoantibodies against tissue transglutaminase and confirmed the diagnosis with small intestine biopsies. Two out of six cases had asymptomatic coeliac disease. Their finding of 2.8% did not support screening with antibodies against tissue transglutaminase in individuals with JIA, since their point prevalence was close to the described prevalence in the general population. The authors did however not compare the results to an age- and sex-matched reference population, which might had changed their conclusion [11]. The frequency of 2.6% in our study is lower than reported numbers from Italy [14, 16], although the risk of coeliac disease was increased among JIA cases compared to age- and sex-matched non-JIA controls from the general population [16]. The prevalence of 0.7% coeliac disease in the reference population was lower than expected, likely due to our exclusion of individuals with the studied autoimmune conditions prior to inclusion. However, the same exclusion criteria were applied to the individuals with JIA and the somewhat low prevalence among the references emphasize that they should not be considered as general population. Undiagnosed coeliac disease might lead to severe adverse effects such as malnutrition and poor growth. The result from our study shows increased risk of coeliac disease after JIA diagnosis and indicates that screening routines for coeliac disease should be considered in the clinic.

Hypothyroidism was diagnosed in 3.3% of the individuals in our JIA cohort, which was significantly more often compared to the reference population within the first 7 years after JIA diagnosis. The presence of hypothyroidism in our study was lower than in a previously published Italian study (10.1%) [14], but higher than numbers reported from Finland (0.7%) [17]. Our restriction to only include autoimmune conditions diagnosed post JIA diagnosis, contrary to the Italian study [14], may have contributed to different prevalence of hypothyroidism between the two study populations. More importantly, in the Italian study the patients were actively screened for autoimmune comorbidity with blood test, thereby possibly diagnosing also subclinical, asymptomatic patients. Hypothyroidism is also more common in the adult than paediatric population and since our cohort is diagnosed 2000–2010, only a minor portion of the individuals were in their 30’ies at the end of the study period. A longer follow-up period can potentially increase the prevalence. In line with the Pharmachild registry study exploring predictors of autoimmune thyroid disease [18], individuals with ANA had an increased risk for hypothyroidism in our study.

Our finding of increased risk of autoimmune disease in ANA positive individuals with JIA compared to ANA negative is supported by other studies showing increased presence of ANA in autoimmune thyroid disease [23, 24]. The mechanism behind the occurrence of ANA is not known, and ANA can be considered a general marker of autoimmune processes since many autoimmune diseases share common genes and pathophysiological processes. The increased risk in the ANA positive group might also reflect the paediatric distinction of this JIA subgroup compared to the other JIA subgroups with adult counterparts. This hypothesis is supported by the significantly increased risk of autoimmune comorbidities in the newly suggested JIA subgroup with ANA positive disease with early onset in our study.

Interestingly, our study further emphasises a specifically increased risk of coeliac disease in individuals with ANA positive disease. To our knowledge, the association with ANA to the risk of comorbid coeliac disease in other autoimmune diseases has been analysed in two previous studies, indicating no increased risk of coeliac disease in ANA positive JIA [25] or ANA positive autoimmune thyroid disease [23].

Our finding of elevated, however not statistically significant, risk for autoimmune conditions in the group treated with DMARD within the following calendar year after diagnosis was interesting. This group was investigated as a surrogate marker for more severe disease, and represent individuals with more inflammation possibly also affecting other organs, but can also reflect certain genetic risk factors. Previous studies exploring the risk of autoimmune conditions in individuals with DMARDs are scarce with conflicting results and some suggest lower risk of comorbid autoimmune conditions after DMARD treatment [15, 26], whereas other have found an increased risk [27]. Our study investigated those autoimmune conditions which are most relevant in the clinic and those conditions we most often see in our clinic in this age group, excluding inflammatory bowel disease and psoriasis which are included as part of the JIA diagnosis.

Our study has some limitations. Diagnosis codes for comorbid autoimmune conditions were collected from the regional healthcare register SHR with data from 1998. The SHR did not have full regional coverage of diagnosis codes registered in outpatient care, especially in primary care, until 2004 [21]. However, we believe the number of missed autoimmune cases to be few, since paediatric autoimmune conditions are primarily diagnosed in specialized healthcare facilities in Sweden. However, the lack of full coverage may have contributed to underestimation as well as overestimation of comorbid autoimmune diagnosis in JIA, depending on whether the missing diagnosis code was registered prior to or after JIA diagnosis.

Another limitation is that the autoimmune diagnoses were not confirmed by a review of the medical records or with data on prescribed insulin, levothyroxine, or gluten free nutriments. Unfortunately, this was not possible for our study. To reduce the risk of overestimation as bias, we required that an individual had a registered autoimmune diagnosis at two separate outpatient visits or at one inpatient healthcare visit. The choice of two registered codes in outpatient care has previously been investigated by a medical record review with high validation of correct diagnostics for autoimmune diseases [28].

Furthermore, we only have annual information on DMARD treatment, and therefore lack exact information of the time relation between start of DMARD treatment and diagnosis of autoimmune comorbidity, as well as objective markers of inflammation such as active joint count, and the results on the risk of DMARD treatment within the following calendar year after JIA diagnosis as a marker for more severe disease must therefore be interpreted with caution.

An alternative explanation for the higher presence of autoimmune disease in our JIA cohort compared to references is that individuals in the JIA cohort have regular healthcare visits, increasing the risk of further diagnostics. To reduce this risk, autoimmune comorbidities registered within 90 days after JIA diagnosis were excluded due to the chance of finding such conditions during arthritis diagnostics. Moreover, the references were also selected on having at least one healthcare visit during follow-up, hence not excluding individuals with other autoimmune (not included in our statistical analyses) or inflammatory conditions, where comorbidities might be detected during laboratory diagnostic at disease onset.

Further, a larger cohort would have enabled additional subgroup analysis, possibly identifying other predisposing factors for autoimmune comorbidities in JIA than ANA positive disease, that also needs to be further evaluated as potential risk factors in a larger cohort.

There are also strengths to this study. We have a well-defined study population with validated cases of JIA, including all disease severities from mild to erosive and disabling, and all states of disease activity. The cases chosen for this study have all been diagnosed in the biologic era, making the results applicable to individuals with JIA diagnosed today. Finally, our study provides longitudinal results on cumulative incidence of autoimmune comorbidities.