Clinical and psychosocial stress factors are associated with decline in physical activity over time in children with juvenile idiopathic arthritis

This study was part of a multi-site, longitudinal cohort study of Canadian children with newly diagnosed JIA (BBOP; Biologically-based Outcome Predictors in JIA). BBOP aimed to identify the inter-relationships of biologic, environmental, and lifestyle factors as predictors of childhood arthritis outcomes [15]. We analyzed BBOP data to characterize the trajectory of PA levels in JIA.

Participants had a clinical assessment and completed all study questionnaires at baseline, 12 and 24 months. Where developmentally appropriate, children completed their own self-report questionnaires. Otherwise, they were completed by a parent.

In compliance with the Helsinki Declaration, the study was approved by the Biomedical Research Ethics Board, University of Saskatchewan: #07–86 and by the research ethics boards at each of the other 10 participating sites. Parents and, as appropriate, children provided written informed consent; children provided assent, as applicable.

Children with a new diagnosis of JIA were enrolled at 11 Canadian pediatric rheumatology centers between March 2008 and January 2011. Enrollment criteria included: (i) consenting participants who met International League of Associations for Rheumatology (ILAR) JIA classification criteria [16] and (ii) were diagnosed within 6 months of symptom onset. As physical activity assessment tools were applicable only for participants aged 6 to 16 years, younger BBOP participants were not included. The cohort comprised participants from each of the seven ILAR JIA subtypes. The aim was to recruit sufficient numbers for each category rather than to strive to achieve a typical JIA subset distribution. To achieve this, only participants with polyarthritis or systemic JIA (the least prevalent subtypes) were eligible during the first 6 months of enrollment after which any JIA subtype was eligible. Participants who were unable to communicate in English or French were excluded from the study.

The reference group for PA measures for this BBOP study analysis were the 154 Canadian children (82 females and 72 males) who participated in the University of Saskatchewan’s Pediatric Bone Mineral Accrual Study (1991–97, entry age 8 to 15 years) [17].

Statistical analyses were performed with SPSS Statistics Professional, version 23, R, version 3.2.2., and MATLAB R2019a. The data had 21% missing values, which were imputed by Amelia II, R package which imputes missing values employing an expectation maximization algorithm [31, 32]. Age- and sex-specific z-scores for the PAQ-C/A were generated for 8 to 15-year-old participants using normative data from the Saskatchewan Pediatric Bone Mineral Accrual Study [33]. Mann–Whitney U test was used to examine the differences between girls’ and boys’ PA scores. The Kruskal-Wallis test was used to assess the difference between PA scores of the seven JIA subtypes. The associations between PAQ-C/A and clinical disease activity (JADAS-71), function (CHAQ), SES and measures of psychosocial stress (JAQQ psychosocial domain, SLEC, and Hassles scale) were assessed at baseline using Spearman correlations. A linear mixed model for panel data was used to estimate the association between PA over time with age, sex, and disease activity measures including number of active joints, ESR, c-reactive protein (CRP), and JADAS-71 (fixed effect). The measures of psychosocial stress and function were considered as variables with random effects. The best model was selected using a theoretical likelihood ratio test.

We found that Canadian children with new onset JIA and moderate disease activity, pain and functional impairments have lower PA levels than healthy Canadian children [17, 33]. Our findings differ from recent reports that children with new onset JIA and low disease activity have similar PA levels to healthy children [2].

Consistent with the literature in healthy children, PA levels in our study were lower in girls and declined with age [1, 35]. We did not find differences in PA levels among JIA subtypes, but this could be explained by small patient numbers within certain subtypes. Previous studies have shown lower PA in rheumatoid factor positive polyarticular JIA [5] and enthesitis related arthritis [36]. It is notable that in our study the majority of children had polyarticular JIA, due to designed enrollment bias during the first 6 months of the study (only patients with polyarthritis and systemic arthritis were recruited during the first 6 months) and the exclusion of children younger than age 6 (oligoarthritis commonly presents at a younger age). This may partly explain the lower levels of PA in our cohort. It is also possible that our reference population PA levels (1991–1997) are not directly comparable. Children are reported to engage in less active commuting, high school physical education, and outdoor play but more organized sport than previous generations [37]. However despite changes in how children are active, long-term trends in PA over time appear to be stable. A multi-national study of trends in PA from 2002 to 2010 across 32 countries from Europe and North America reported a slight increase in 11-, 13- and 15-year-olds meeting the 60 min of daily PA recommendation (17.0% in 2002 and 18.6% in 2010) [38].

We found a negative correlation between PA and disease activity (JADAS-71), and this correlation increased from baseline to 12-month and 24-month follow-up, despite decreasing disease activity over time. These findings are in keeping with the existing literature which reports a negative relationship between PA and disease activity [5] as well as lower PA in JIA populations with good disease control [5, 6].

PA levels in our cohort progressively declined following diagnosis, with lower PA at 12 months and further decline in PA at 24 months. This pattern of declining PA with age is consistent with healthy children. Our results can be interpreted in the context of the large body of literature which reports lower levels of PA in children with established JIA [4‐6]. Our longitudinal models show that in addition to age-related declines in PA, lower PA in children with JIA is associated with disease duration and disease activity, as measured by the JADAS-71, ESR and active joint counts. However, changes in disease activity do not solely explain the decline in PA. In keeping with our results, Norgaard and colleagues reported that disease-related factors account for a low percentage of the variance in PA levels of children with JIA [5]. As such, non-disease related factors must play a large role in the observed decline in PA during the first 2 years after diagnosis. A new diagnosis of JIA may lead to a change in family, peer and community supports’ (teachers, coaches) attitudes towards PA and sport participation. Concerns about disease flares, increased risk of injury or underperformance may limit a child’s PA. A recent study from Denmark found children with JIA had less participation in sport and less consistent participation in school-based physical education activity. Ninety percent (N = 62) reported pain in joints (81%) and/or muscles (30%) but non-arthritis related factors including shortness of breath/side stitches (30%), lack of competency in specific activities (30%), and lack of support from teachers (25%) were also reported as reasons for decreased school PA participation [39].

Our study is the first to explore the joint relationship of psychosocial stress and SES with PA in children with JIA. Consistent with our hypothesis, higher levels of disease-specific psychosocial stress were associated with lower PA but this was apparent only with the disease-specific JAQQ psychosocial domain, and not with generic measures, and only in univariate tests, not in LME. Although the literature consistently reports that stress has a negative effect on PA in healthy populations [40], our patients apparently distinguished between arthritis-related and non-arthritis-related psychosocial stress. Alternatively, the arthritis-specific instrument was more sensitive; but in either case, our observation underscores the importance of using disease-specific measures. The lack of independent effects of the JAQQ psychosocial in LME may be due to possible interactions among variables or an inadequate sample size.

Personal factors of resilience, self-efficacy and support may mitigate the negative effects of stress and may explain the modest correlation relating PA to the JAQQ psychosocial scale as well as the lack of effect of generic stresses. Studies in healthy children and adolescents and those with disabilities, including JIA, have shown that parental support is positively correlated to PA [41]. In addition, Seid and colleagues reported that parental emotional distress, social support and child self-efficacy are related to changes in health-related quality of life and physical function in newly diagnosed children with JIA [13]. Children and parents who are habitually active as a family may increase or maintain their PA to cope with stress while families that are less habitually active may decrease their PA during stress. We did not capture parent PA level in our study and cannot test this hypothesis.

Our inability to relate SES to PA in children with JIA in our study may be due to small patient numbers in the face of the multiple categories within SES variables. Further studies with larger populations are required to clarify the effect of SES.

Our results highlight that PA is a complex behavior influenced by many factors. Family and child stress may affect PA level, either positively or negatively. The influence of psychosocial stress and the role of resiliency, motivation and self-efficacy in facilitating PA in children with JIA require further study.

The strengths of our study include prospective longitudinal evaluations, using specific longitudinal trajectory models, of PA levels and the relationship of PA to disease activity and psychosocial stress; and the inclusion of both generic and disease-specific instruments. We acknowledge several limitations of our study. JIA is a heterogeneous disease and our cohort did not include children younger than age 6 (because measurement tools for PA and stress were not available for younger children). Our cohort is also skewed towards more severe disease (as a consequence of our recruitment protocol) and therefore is not representative of a usual JIA clinic population. Thus, our results may not be applicable to younger children and all subtypes of JIA. It may be difficult to compare our results to other studies, however, our goal was to generate novel information about PA levels and JIA category, disease activity and measures of SES and psychosocial stress. We used self-report measures of psychosocial stress and PA. While self-report measures have inherent bias, all the measures reported are commonly cited in pediatric research. We did not capture PA intensity or absolute time spent engaged in PA and are thus unable to comment on whether this cohort of children is meeting current PA recommendations of 1 h of moderate vigorous PA per day [42]. We did not measure resilience, motivation or self-efficacy, all of which may influence the effect of stress on PA. These will be important to consider in future studies. We did not capture medication side effects or fatigue, both of which may affect PA. The majority of our patients had moderate to high SES and we were unable to relate SES to PA. Further studies with larger populations and measurements of multiple PA domains (leisure, transportation, occupation, and housing) are needed to clarify the effects of SES. Finally, small patient numbers may be responsible for apparent lack of effects of certain variables.