Abstract
Objective Juvenile idiopathic arthritis (JIA) is a chronic rheumatic disease that causes joint inflammation and pain. Previous studies have indicated affected mental health and increased risk of psychiatric conditions among patients with JIA. We aimed to explore differences in psychiatric morbidity between children with JIA and their peers. We further studied if parental socioeconomic status (SES) influences the association between JIA and the risk of psychiatric morbidity.
Methods We used a matched cohort design to estimate the association between JIA and psychiatric disease. Children with JIA, born between 1995 and 2014, were identified in Danish national registers. Based on birth registers, we randomly selected 100 age- and sex-matched children per index child. Index date was the date of the fifth JIA diagnosis code or the date of matching for reference children. End of follow-up was the date of psychiatric diagnosis, death, emigration, or December 31, 2018, whatever came first. Data were analyzed using a Cox proportional hazard model.
Results We identified 2086 children with JIA with a mean age at diagnosis of 8.1 years. Children with JIA had a 17% higher instantaneous risk of a psychiatric diagnosis when compared with the reference group, with an adjusted hazard ratio of 1.17 (95% CI 1.02-1.34). Relevant associations were found only for depression and adjustment disorders. Stratifying our analysis for SES showed no modifying effect of SES.
Conclusion Children with JIA had a higher risk of psychiatric diagnoses compared to their peers, especially diagnoses of depression and adjustment disorders. The association between JIA and psychiatric disease did not depend on parental SES.
Juvenile idiopathic arthritis (JIA) is a chronic inflammatory disease of unknown etiology with symptom onset before the age of 16 years, and is characterized by joint inflammation and pain, which eventually leads to joint destruction if untreated.1,2
In general, children with chronic diseases often experience lower quality of life due to different factors, such as pain, discomfort, and frequent hospital visits.3-5 The rates of psychiatric problems have been a major subject for decades, and many studies have supported a higher risk of especially depression and anxiety disorders in children and adolescents with chronic diseases.6-9
The heterogeneity of chronic diseases during childhood, however, is large, and JIA stands out as a disease dominated by joint pain and fatigue, even during periods of remission.1,2 In accordance with this, there has been a demand for quality studies on the incidence of psychiatric morbidity and modifying factors among children with JIA.10 Unfortunately, most studies have been questionnaire studies using a cross-sectional study design.11
In review, Fair et al emphasized that the rates of depression and anxiety in children with JIA are comparable to those of children with other chronic diseases, but considerably higher when compared to those of healthy children.11 Socioeconomic status (SES) was pointed out as a possible effect modifier of the association between JIA and depression, but not discussed further. In general, parental SES has been shown to be inversely correlated to mental health and psychopathology in childhood and youth.12-16 This association has also been found in children with other chronic diseases, where lower SES was especially associated with lower quality of life.17 However, to our knowledge, no study has investigated this relationship in a large cohort of patients with JIA.
A recent register-based cohort study from Finland found that especially girls with JIA had a higher risk of psychiatric morbidity compared to healthy controls.18 A correlation between age at JIA-diagnosis and type of psychiatric morbidity was shown, but unfortunately, the study included only patients with JIA receiving medical treatment. In contrast, a recent study from southern Sweden found no increased risk of depression and anxiety among children with JIA compared to the general population.19 These conflicting results call for further quality studies—preferably studies that encompass a broad population-based approach using national registers and a long timeline. In our current study, the Danish registers allowed us to evaluate all children diagnosed with JIA in Denmark born from January 1, 1995, onwards, regardless of treatment, and compare them to a large, matched control group.
The aim of the present study was therefore to explore differences in psychiatric morbidity between children with JIA and their peers in the Danish population. We further examined whether parental socioeconomic factors, age at diagnosis, or disease severity influenced the association between JIA and the risk of psychiatric morbidity. The knowledge from our study can hopefully help clinicians and healthcare workers focus resources and energy on certain subgroups of patients that may need extra attention.
METHODS
Setting. The study population included children born in Denmark and registered in the Danish Medical Birth Register between January 1, 1995, and December 31, 2014.20 To be included in the study, children had to reside in Denmark until JIA diagnosis or index date. Information on migration and death was obtained from Statistics Denmark.
Exposure. We used a matched cohort design. Information on JIA was obtained from the Danish National Patient Registry (DNPR).21 To be included as a patient with JIA, the children had to fulfill the criteria of 5 hospital visits (in- or outpatient contacts) registered with International Classification of Diseases, 10th revision (ICD-10) codes within the categories DM08 and DM09 before the age of 17 years and before December 31, 2016.22 We chose the limit of 5 visits with JIA diagnosis codes to increase the probability of including only children correctly diagnosed with JIA and to exclude observational diagnoses of JIA. Since JIA is a chronic disease that involves multiple visits to the hospital, we assumed that no child with a correct diagnosis of JIA would have only 1 or 2 visits. Therefore, 5 visits was chosen as the best cut-off point to increase the specificity of JIA at the cost of a possible small decrease in sensitivity. The index date was set as the date of the fifth hospital visit with a diagnosis code of JIA. Children with JIA with a psychiatric diagnosis (any ICD-10 code in the F category) before the index date (n = 79) were excluded from this study.
Disease severity was analyzed according to treatment. Data on redeemed prescriptions of methotrexate (MTX) were obtained from the Danish National Prescription Registry. Data on in-hospital treatment with MTX or biologic therapy were obtained from the DNPR.
From the DNPR, data on the diagnosis of uveitis (ICD-10 code DH20* or DH30*) were registered as this is a common comorbidity of JIA.
Matching. For each child with JIA, 100 age- and sex-matched children (reference group) with no JIA diagnosis and no psychiatric diagnoses before the index date were randomly selected from the background population. Age was matched on birth year and month.
Outcome. The outcome of interest was psychiatric disease. Information hereof was obtained from the Danish Psychiatric Central Research Register (DPCR), including information on both in- and outpatient contacts to Psychiatric Hospitals in Denmark.23 The psychiatric diagnoses included in the study were ICD-10 codes DF0-DF9, except for DF7, which covers mental retardation. We analyzed all psychiatric diagnoses as 1 group, as well as specific subgroups of psychiatric diseases. Subgroups were depression (DF32 + DF33), anxiety (DF40 + DF41), obsessive-compulsive disorder (OCD; DF42), eating disorders (DF50), and reaction to severe stress and adjustment disorders (DF43). We chose these specific psychiatric diseases as we found these were most likely to be associated with a chronic condition like JIA. Some children may have had a psychiatric disease not leading to hospital contact. To identify these children for a subanalysis, we included information on reimbursement of psychiatric medication. The Danish National Prescription Registry has recorded information on all redeemed prescriptions in Denmark since 1995, including international Anatomical Therapeutic Chemical (ATC) coding.24,25 We included redeemed prescriptions of all psychiatric medicine with ATC codes N05 and N06, excluding benzodiazepine (N05BA and N05CD) and melatonin-agonists (N05CH) as these groups of medications are also given on indications other than psychiatric, and would contaminate our data. For specific subanalysis of depression, we included information on medicine with ATC code N06A.
Covariates. From Statistics Denmark, information on parental educational level, family income, and registered partner status of the primary caretakers the year before the index year was obtained. If data on the year before the index year were missing, it was replaced by data from the index year to reduce the amount of missing data.
Parental psychiatric disease (any DF ICD-10-code) before the index date was based on information from the DPCR. Unfortunately, we were able to include only information from 1995 onwards. This means that parents diagnosed with psychiatric conditions before 1995 were not correctly grouped unless they had hospital visits because of psychiatric disease after 1995.
Equivalent family income was grouped into income tertiles based on the income distribution of the background population—incomes of 0 Danish kroner or negative were registered as missing. Educational level was grouped into 3 groups based on the parent with the highest completed level: “short,” meaning primary school (UNESCO’s International Standard Classification of Education [ISCED] level 1-2); “middle,” meaning high school, vocational education, and similar shorter education (ISCED level 3-5); and “long,” meaning tertiary education at a bachelor level or higher (ISCED level 6-8).26 Family type was divided into 2 groups according to the number of caretakers living with the child: single caretaker if living with 1 caretaker or nonsingle caretaker if living with > 1 caretakers. Age at inclusion was divided into 3 groups (< 6 years, 6-12 years, and ≥ 13 years).
Statistical analysis. Children were followed from the index date until the end of follow-up, registered as the date of first psychiatric diagnosis, migration, death, or end of follow-up on December 31, 2018, whatever came first. If a reference child fulfilled the criteria of JIA during the follow-up time, they were censored from the reference group at the time of diagnosis and were matched with 100 children randomly selected from the background population and contributed with risk time in the JIA group from then on.
We used a Cox proportional hazards model to estimate hazard ratios (HRs) and 95% CIs for psychiatric disease, comparing children with JIA with the reference children. The underlying time scale was time since index date measured in years. The primary analyses included family income, parental education, family structure, and parental psychiatric diagnosis as covariates. In our adjusted analysis, we included only participants with information on all variables. To study whether parental socioeconomic factors modified the association between JIA and psychiatric disease, we stratified this analysis by family income, parental educational level, and family type. We also stratified the analysis according to sex and age at index. We chose not to adjust our stratified analyses due to a lack of power because of low numbers in some groups. A test of the hypothesis of no difference in HRs between stratified groups was performed.
Kaplan-Meier cumulative incidence curve for psychiatric diagnosis as a function of age was reported. Point estimates of the difference between groups at different ages were calculated with 95% CIs.
A sensitivity analysis of the risk of psychiatric diagnosis when having severe JIA (treatment with MTX or biologic therapy) was performed using a Cox proportional hazards model, including the severity of JIA as a time-dependent variable. The patients with JIA contributed with risk time in the JIA group from baseline until the date of receiving either MTX or biologic therapy; from then they contributed with risk time in the severe JIA group until the end of follow-up. Sensitivity analysis of the risk of psychiatric diagnosis among children with JIA and uveitis was performed using the same statistical methods as for severe JIA.
Further, in a sensitivity analysis, we included information on the redemption of psychiatric prescription medicine in the definition of the outcome. In another sensitivity analysis, we included all children with at least 1 ICD 10 code within the categories DM08 and DM09 in the JIA group.
All analyses were performed using Stata 17.0 (StataCorp).
Ethical considerations. The study was approved by the Danish Data Protection Agency through the Aarhus University comment agreement (journal no. 2016-051-000001, sequential no. 2197). Danish register-based studies do not need approval from an ethics committee.
RESULTS
A total of 2086 children with JIA fulfilled the inclusion criteria. The included children with JIA had a mean age at diagnosis of 8.1 (SD 4.5) years with 62.70% being female.
During follow-up of the children with JIA, 963 (46.16%) were registered with a procedure code of MTX or had a redemption of MTX, 577 (27.66%) had a procedure code of biologic treatment, and 476 (22.82%) were registered as having had a joint injection. Of the children with JIA, 260 (12.46%) were diagnosed with uveitis.
No essential differences were observed comparing the distribution of family income tertiles, parental educational level, parental psychiatric diagnoses, or percentage of single parents when comparing children with JIA with their matched references (Table 1).
Baseline data, socioeconomic status, and number of children diagnosed with psychiatric conditions during follow-up.
A total of 207 (9.92%) children in the group of children with JIA were diagnosed with a psychiatric diagnosis during follow-up compared to 17,620 (8.45%) in the reference group. The unadjusted HR was 18% (95% CI 3-37) higher of any psychiatric diagnosis among the children with JIA compared to the reference group. Adjusting for socioeconomic variables and parental psychiatric disease did not change the results, with an adjusted HR (aHR) of 1.17 (95% CI 1.02-1.34; Table 2). When analyzing specific groups of psychiatric conditions, higher HRs of depression and adjustment disorders were seen among children with JIA (Table 2).
HRs for psychiatric diagnosis among children with JIA and the reference group.
Regarding age, there were no differences (0.48% points, 95% CI −0.79 to 1.76) between children with JIA and the reference group on the proportion with psychiatric diagnoses at age 10 (Figure 1). At ages 15 and 20, children with JIA more often had a psychiatric diagnosis compared with reference children. At age 15, the difference was 2.21% points (95% CI 0.40-4.01) and at age 20 this was increased to 2.65% points (95% CI 0.27-5.02). Children with JIA who were diagnosed with arthritis before school age had the same HR of psychiatric diagnosis as the reference cohort. In contrast, children with JIA diagnosed from school age and onwards had higher HRs of psychiatric diagnosis compared with references (Figure 2, Table 3).
Kaplan-Meier of the cumulative incidence of psychiatric diagnosis with age as time-varying variable. JIA: juvenile idiopathic arthritis.
Kaplan-Meier of the cumulative incidence of psychiatric diagnosis with age as time-varying variable stratified by age at index date. (A) Age group < 6 years. (B) Age group 6-12 years. (C) Age group ≥ 13 years. JIA: juvenile idiopathic arthritis.
HRs for any psychiatric diagnosis according to JIA (stratified analyses).
Stratifying our analyses for parental socioeconomic factors, the association between being diagnosed with JIA and any psychiatric diagnosis was strongest for children living with nonsingle caretakers, parents belonging to the middle-income group, and the highest educational group. However, socioeconomic factors showed no statistically significant effect modification (Table 3). Further, there was no significant difference between female and male children on the HRs of psychiatric diagnoses.
By comparing children with JIA divided on disease severity, we found almost no difference in HR for psychiatric diagnoses. The adjusted hazards were aHR 1.17 (95% CI 0.97-1.40) and aHR 1.22 (95% CI 0.99-1.51) for children with nonsevere JIA and children with severe JIA, respectively. Children with JIA with and without uveitis showed similar risks of psychiatric diagnoses. Adjusted hazards were aHR 1.19 (95% CI 1.03-1.18) for children with JIA and no uveitis and aHR 1.16 (95% CI 0.71-1.89) for children with JIA and uveitis.
We observed a more than 20% higher number with a psychiatric condition in both children with JIA and reference children when including the usage of psychiatric redemption medication (Table 4). We observed a stronger overall association between JIA and psychiatric disease when including information on psychiatric medicine. Looking at depression alone, estimates remained almost unchanged though with narrower CIs.
HRs of psychiatric diagnoses including psychiatric prescription medicine.
Sensitivity analysis using only 1 diagnosis code to define JIA led to the same conclusions as in our main analysis (Supplementary Tables S1-3, available from the authors upon request).
DISCUSSION
We observed differences between children with JIA and their matched peers in the HRs of psychiatric diagnoses. Particularly, higher HRs were seen among children with JIA when examining for depression and adjustment disorders. We found no differences in the HRs between children with JIA and references for anxiety, eating disorders, OCD, or behavioral disorders. However, numbers with OCD and eating disorders were small, hence the power to study these outcomes was low. The association between JIA and psychiatric disease was not modified by parental SES.
The rates of treatment with MTX and biologics in the current study are in accordance with the percentages presented by Glerup et al in their Nordic 18-year follow-up study.27 Children treated with MTX or biologics have a more severe JIA course and would thus be expected to have a higher risk of developing a psychiatric disease. We found the association to be strongest for children with severe JIA, but not much stronger than was found for nonsevere JIA. We found no difference in the hazard risks between children with JIA with and without uveitis.
A high risk of depression among children with JIA has been reported in previous studies using different screening tools, as previously reviewed by Fair et al.11 In contrast, the newly published Swedish register-based study by Berthold et al found no increased risk of depression among patients with JIA.19 Only 4 studies in the review of Fair11,18,28-30 included a healthy reference group, of which 1 study found no differences in depressive symptoms between groups.30 What separates our study from most previous studies is the use of national registers of psychiatric diagnosis codes.
Cartwright et al describe in their qualitative study that living and being diagnosed with JIA is experienced as “a journey of adaptation that was often turbulent,”31 and Min et al describe that most children with JIA experience their illness as life-changing, demanding a large amount of adaptation.32 This view is supported by Kyllönen et al18 and our results, collectively showing a higher risk of adjustment disorders among children with JIA. However, child psychiatrists might also be more likely to choose these diagnosis codes if children already carry a JIA diagnosis, and therefore have a condition to adjust to.
Parental SES has previously been shown to be associated with general child health,12,16,33,34 whereas varying results have been reported on whether parental SES is associated with prevalence and severity of chronic conditions in children. Social support has been shown to predict fewer adjustment problems in children with JIA.35 In our study, we found no clear modification of SES on the association between JIA and psychiatric diagnosis. The lack of modification of SES might be because of the Danish healthcare system, which provides free treatment for all.
In accordance with the findings of Kyllönen et al,18 we found a tendency toward the strongest association between JIA and psychiatric diseases among children with onset of JIA in their teens. We also found children diagnosed with JIA from school age to have a higher HR of psychiatric disease compared to their peers, whereas children diagnosed before the age of 6 did not differ from their peers. This is possibly explained by the fact that children diagnosed at a very early age will tend to form and adapt their coping strategies around the disease, treatments, and check-ups in a better way, partly because the individual reflective self at this age is still underdeveloped.36 For children diagnosed from school age onwards, however, adaption to the diagnosis and its treatment must occur not only at a time in life with a developed reflective self but also during a time with higher demands in school and social life.37,38 In addition, children and adolescents diagnosed at this age have a memory of a disease-free life before the diagnosis.
Our results are in line with the general finding of an increased risk of developing psychiatric disease through school life and adolescence.39 For children with JIA at the age of 16 to 18, transition from pediatric to adult care therefore must be handled with due diligence, since these patients are at a vulnerable stage in their lives.40-42
A strength of our study is the use of national databases for identifying both children with JIA and outcomes. Since all children with JIA are diagnosed at hospitals in Denmark, the nationwide registers cover the total population of children with JIA. Further, we had a very low number of dropouts because of death or migration; hence we do not expect our findings to be because of selection bias.
A limitation of our study is that we did not follow all the children further into adulthood. We analyzed our data using HRs, which take into account the varying follow-up time. However, we were not able to include information on whether the children with JIA continued to have active disease into adulthood, thereby influencing their risk of developing a psychiatric disease.
Typically, children with JIA follow a course of multiple outpatient visits over several years. Hence, we do not expect children having only, for example, 1 or 2 registered JIA contacts to actually have JIA. We therefore included only patients with JIA with at least 5 registered JIA diagnosis codes in our main analysis. The number of children with JIA decreased by almost 1000 children when restricting to 5 diagnosis codes instead of 1, and this reduction was only in a very few cases because of insufficient follow-up time. We are not aware of studies validating the JIA diagnosis in the DNPR, but our strategy may have increased the specificity. Children with fewer than 5 contacts and 1 of the diagnosis codes might have been selected at random for the control group, which could have increased the risk of bias toward the null hypothesis. However, none of the reference persons had any JIA diagnosis before the index date (data not shown). Additional sensitivity analyses, in which children categorized as having JIA already at their first hospital contact, did not change our conclusions.
The DPCR has been shown to have good accuracy in children and adolescents.43 Still, a limitation of our study was the possibility of underdiagnosing psychiatric conditions, which might be more likely in the reference group. Children with JIA have regular contacts with the pediatric departments, which could increase their chance of being registered or investigated for a psychiatric condition. In Denmark, investigation for psychiatric disease at a child psychiatry department requires a referral from either the child’s general practitioner or the public school psychologists. Still, children with no JIA and a psychiatric condition might have a higher risk of being missed. Thus, underdiagnosing of children without JIA may have led to an overestimation of the association between JIA and psychiatric diagnoses. To try to include less severe psychiatric diseases, we included information on redeemed psychiatric medication in our sensitivity analysis. This identified almost exclusively patients aged 18 years or above (data not shown), which is in accordance with the age at which general practitioners can prescribe medication to less severe psychiatric conditions. This strengthened the results for overall psychiatric disease but did not change the results for depression. Children with JIA often develop anxiety for needles,44 which requires treatment from a psychologist. In Denmark, private psychologists are the primary filter by which many children with mental challenges are seen. Unfortunately, we have no information on contacts to the private psychologist sector in the Danish registers, and hence we might have missed some children with anxiety or adjustment disorders in our analyses.
We decided not to adjust our stratified analyses for other covariates due to lack of power. However, as adjustment did not change the results in the main analysis, we do not expect influence from other socioeconomic variables to explain the results in our stratified analyses.
As our study was performed in the welfare state of Denmark, which has free access to health care, the generalizability is most likely only to countries with healthcare systems like that of Denmark. Further, we were able to report data only on severe cases of mental illnesses and are not able to make conclusions on less severe psychiatric illnesses.
In conclusion, children with JIA had higher HRs of psychiatric diagnoses compared to their peers, especially for children diagnosed with JIA from school age onwards, and specifically for the diagnoses of depression and adjustment disorders. The association between JIA and psychiatric disease did not depend on parental SES. We suggest that the risk of psychiatric morbidity in JIA becomes an integral point of attention in the clinical care of these patients.
Footnotes
This work was supported by the Danish Rheumatism Association (grant nos. R209-A7445 and R198-A7079 to MJP).
CH has received lecture fees from MSD, AbbVie, and Novartis. BWD has received lecture fees from Eli Lilly, AstraZeneca, and Pfizer, and is a member of the Advisory Board of AstraZeneca, UCB, and Boehringer Ingelheim. The remaining authors declare no conflicts of interest relevant to this article.
- Accepted for publication May 19, 2023.
- Copyright © 2024 by the Journal of Rheumatology
