Juvenile idiopathic arthritis (JIA) is a severe chronic disease in childhood and adolescence frequently leading to handicap, decreased quality of live and high costs [1]. The prevalence varies across regions from 0.007 to 0.401% [2]. Recent estimates for the US approximated that juvenile idiopathic arthritis (JIA) affects about 294,000 children [3]. The majority of them suffer from oligoarticular arthritis (OA JIA) [3, 4].

The knowledge about possible risk factors is crucial for disease prevention. While genetic and environmental factors might play a role in the development of JIA knowledge about its aetiology is still limited [1]. Some of the environmental risk factors for JIA are similar to those described for respiratory allergies (i.e., allergic rhinitis, atopic asthma), e.g. maternal smoking during pregnancy [5], having no siblings, having more affluent parents, living in Northern Europe [6–8]. Furthermore, a positive association between JIA and respiratory allergies has been found in some studies [9, 10] while others have described an inverse [11–13] or no association [14].

For allergic diseases, some studies have suggested that the development of atopic diseases is facilitated by a decreasing level of microbial exposure in early life ("hygiene hypothesis) [15, 16]. Consistent with the hygiene hypothesis, subjects living in rural areas and having contact to farm animals in the 1^st year of life have been shown to have a lower prevalence of respiratory allergies (for review see [17]). The burden of microbial compounds (endotoxins, β-glucans) found in these environments may be responsible for the inverse association [18–21].

Based on these assumptions, we hypothesized that farm animal contact in infancy might also reduce the risk of juvenile autoimmune disease. We and others have previously shown that infant farm animal contact might be inversely associated with inflammatory bowel disease and systemic lupus erythematosus (SLE). In contrast, we were unable to find such an association for type 1 diabetes mellitus (DM1) [22–24]. Nielsen and colleagues [8] have previously reported that children living in urban areas are at increased risk of JIA. However, only 9 of 220 patients lived on a farm and as the study was not designed for this question a more detailed analysis could not be done.

The aim of the study reported here was to test the hypothesis that early childhood farm animal contact might protect from JIA.

Our study provides evidence that place of living as well as contact with animals during the first year of life, one of the most important factor protecting from respiratory allergies, might at least not strongly be associated with OA JIA in children. Likewise, none of the other factors typically considered to be markers of hygiene were associated with OA JIA. These findings are consistent with our results for diabetes mellitus 1 (DM1) [22] but contradict results for IBD [23, 28].

OA JIA was inversely related to allergic rhinitis which is accordance with previous studies and appears to relate to differences in the genetic make-up [11–13]. Our study further confirmed known risk factors for JIA like higher level of parental education and female sex [5, 8, 29, 30]. We could not confirm the findings by Nielsen and colleagues [8] that children living in urban areas are at increased risk of JIA. We could also not confirm maternal smoking during pregnancy as a risk factor for JIA [5]. This might be due to the choice of the control group. Low birth weight and in-utero exposure to maternal tobacco smoke are known risk factors for strabismus [31, 32]. Choosing the right control group is the main concern in any case-control study [33]. Access to our OA JIA cases was given through a specialized clinic which covers a large area. Therefore, the use of population-based controls from all communities where cases were located was not feasible. Instead we used hospital based controls who underwent treatment in referral centers with a similar coverage area. We therefore considered controls to be equally likely to be treated in the same hospital had they been diagnosed with JIA. Due to limited funding we had to use controls who participated in our previous study on animal contact and IBD [23]. As the field work was done in parallel using the same questionnaire we do not consider this a problem. However, as no data on JIA among control parents were available we could not adjust for parental JIA. While genetics is a predictor of JIA [1] we do not consider it a confounder in the association under study.

One may only speculate why infant animal contact might reduce the risk for respiratory allergies and IBD as well as SLE but is associated neither with DM1 nor with OA JIA. One reason might be attributed to similarities in the NOD2/CARD15 polymorphism found for respiratory allergies, IBD and SLE but neither for DM1 nor for OA JIA [34–37]. Therefore, gene-environment interactions might be one factor that has to be considered interpreting the disparities found for IBD, SLE, DM1 and OA JIA in our studies.

The advantage of our study is the relatively large number of cases and controls that could be included. The proportion of subjects living in rural areas as well as the number of participants with regular contact with farm animals during infancy was reasonable. However, due to the limited number of patients with OA JIA the power was slightly lower than in our previous study on IBD [23] but higher than in a similar study on animal contact and DM1 [27].

A detailed questionnaire instrument with mainly standardized questions was used [26, 27]. In addition, response among cases and controls was high. Nevertheless, one potential source of selection bias could be that about 17% of controls had moved since the time of strabismus surgery and could thus not be contacted. We do not consider this factor to be associated with the exposure under study. However, we do not have any information to assess the exposure under study in this group.

The selection of OA JIA patients from a specialized center might have biased the study population towards selection of more severe cases that required treatment in such a referral center.

Due to a relative low number of children newly diagnosed with OA JIA, we could not restrict our study to incident cases [29]. Likewise, children mostly undergo strabismus surgery at around age 6 years. In result, recall bias might have taken place. As the parents of cases and controls were most likely not aware of a potential association under study no major bias is anticipated to result from this selection of patients.

We did not match cases and controls because matching results in a loss of potential study subjects and thus reduces the efficiency of a study. Furthermore, matching may even result in greater difficulty in controlling for additional confounders [38]. Our controls were significantly older than cases and it might be that that the results of controls were less reliable as more time passed by since infancy. However the difference between cases and controls was only two years. Furthermore, we could show that age was not associated with infant exposure to animals. Therefore, we do not consider that this age difference might have biased our findings. The age difference has to be considered when interpreting the inverse association between allergic rhinitis and case status. Nevertheless, the latter finding was consistent across age groups and robust to adjustment for age.

In conclusion, our results imply that animal contact during infancy, one of the main factors protecting against childhood allergies, does not lower the risk of OA JIA.

KR has participated in all parts of the study. DW and DP were involved in the organization of the study, and contributed to the design of the survey as well as the interpretation of the findings. ST made contributions to draft the manuscript. EvM, RH and HM were involved in the idea and the design of the study, contributed to the interpretation of the results. All authors read and approved the final manuscript.