To document perceived adherence to treatment (taking medications and performing exercises) in patients with juvenile idiopathic arthritis (JIA) over a 1-year period and to identify related factors.
To document perceived adherence to treatment (taking medications and performing exercises) in patients with juvenile idiopathic arthritis (JIA) over a 1-year period and to identify related factors.
We surveyed parents of patients with JIA at the Montreal Children's Hospital and British Columbia's Children's Hospital in Vancouver. Parents were asked to respond to a series of questionnaires every 3 months over a 12-month period. Perceived adherence was evaluated on a 100-mm visual analog scale (VAS) in the Parent Adherence Report Questionnaire (PARQ). Parental coping, distress, child function, disease severity and duration, perceived helpfulness of treatment, problems encountered, and sociodemographic data were also assessed.
The mean age of our sample of 175 children was 10.2 years; mean age at diagnosis was 6.1 years and mean disease duration was 4.1 years. Perceived adherence to medications was consistently high, with average adherence at baseline, 3, 6, 9, and 12 months being 86.1, 91.7, 90.4, 92.0, and 88.8, respectively, on the PARQ VAS. Perceived adherence to exercise was lower but remained steady, with corresponding means of 54.5, 64.1, 61.2, 63.0, and 54.3, respectively. Using generalized estimating equation analysis, factors associated with higher perceived adherence to medications included perceived helpfulness of medications and lower disease severity; those associated with higher perceived adherence to exercise were younger age of the child, child involvement in responsibility for treatment, and higher perceived helpfulness of the treatment.
Belief in helpfulness of treatment is associated with higher parental perceived adherence to treatment.
Although adherence to treatment in several childhood chronic diseases has been documented, few studies have examined adherence over time. Adherence to treatment may decrease over time in patients with chronic conditions such as epilepsy or human immunodeficiency virus (1–3), but little is known regarding adherence over time in patients with juvenile idiopathic arthritis (JIA).
JIA is a common childhood chronic disease (4) that comprises several subtypes of chronic arthritis starting in a child younger than 16 years of age (5). In terms of prognosis, more than 50% of young persons with JIA will have ongoing active arthritis into adulthood (6, 7). Treatment may be complex, often including several types of medications to control inflammation and reduce joint destruction; physical therapy to relieve pain, maintain range of motion, and strengthen muscles surrounding affected joints; and splinting of joints to prevent deformity.
Poor parental adherence to treatment regimens may compromise potential benefits of conventional treatments (5, 8–10). Some parents believe that the side effects of medications outweigh the potential benefits and are reluctant to use them (11). Also, when their children's joints are swollen and painful, some parents fear that exercise and splinting of affected joints might be harmful; as a result, many parents are resistant to guide their children in using these treatments (12).
Severity of disease or its progression over time might influence whether parents follow medical advice (11, 13–17). Therefore, there may be a complex relationship between chronicity and severity and poor adherence with recommended therapeutic regimens. Poor communication and limited parental comprehension of the medical condition have also been shown to be associated with diminished adherence (18).
There are only a few reports that explore adherence in JIA. In a cross-sectional study, Sturge et al reported adherence to medications to be 92% in the polyarticular group and 68% in the oligoarticular group, whereas adherence to physical therapy and wearing splints was only ∼30% in both groups (19). Poor adherence to medical treatment was associated with increased absence from school and psychiatric problems in children with JIA (19). Parents have reported that exercises were more problematic than medication and splint wearing (12, 20). These studies, however, did not explore the factors influencing these findings, nor did they document adherence over time. These factors are important if adherence to treatment is to be maximized in this population.
Methods for measuring adherence such as provider estimates and patient/caregiver report are considered to be clinically feasible but may overestimate adherence (21). Perceptions of providers and caregivers regarding adherence may differ substantially (22).
A theoretical model proposed by the World Health Organization (WHO) includes the following 5 dimensions/groups of factors that may be related to adherence: social and economic factors (e.g., age, socioeconomic status, distress, coping), health care team and system-related factors (e.g., patient-provider relationship, perception of adherence by the provider), condition-related factors (e.g., severity, duration of disease), therapy-related factors (e.g., complexity of the regimen, side effects), and patient-related factors (e.g., knowledge, attitudes, beliefs, perceptions, and expectations) (3). In JIA, disease onset at an earlier age and longer treatment duration have been correlated with poor adherence to salicylate therapy, independent of current disease severity (23). A recent study of adults with rheumatic disease indicated the following as barriers to adherence: fear of side effects, financial problems, difficulty navigating the health system, perceived inefficacy of treatment, and complex dosing and treatment regimens (24). Other studies have suggested that better adherence may be associated with younger age in children (25) and a better patient-provider relationship in adults (26, 27). The objective of this study was 1) to document longitudinal perceived adherence to treatment (taking medications and doing exercises) in patients with JIA over a 1-year period, 2) to determine whether the professional's (rheumatologist or physical therapist) perception of adherence was associated with the parent's perception, and 3) to describe factors associated with adherence to treatment in children with JIA.
Parents of patients with JIA who attended the JIA clinic at 2 pediatric hospitals (Montreal Children's Hospital, McGill University Health Centre and British Columbia's Children's Hospital) were eligible for the study. Parents were approached to participate if they had a sufficient comprehension of either English or French and if their children were currently undergoing treatment.
Parents of children with JIA completed the following questionnaires repeatedly over a period of 12 months: 1) study questionnaire package that included the Parent Adherence Report Questionnaire (PARQ ; addressing questions about adherence to treatment), a demographics questionnaire, a questionnaire on the use of complementary and alternative medicine, and a question that addressed problems encountered with the health care system (completed at 3-month intervals); 2) the Symptom Checklist-90-Revised (29), which evaluated caregivers' psychological status (completed at baseline, 6 months, and 12 months); 3) the Coping Health Inventory for Parents (CHIP) (30, 31), which documented caregivers' coping with their child's illness (completed at baseline, 6 months, and 12 months); and 4) the Child Health Assessment Questionnaire (CHAQ) (32), which described the child's functioning/disability (completed at baseline, 6 months, and 12 months). Clinicians (physicians, physical therapists, and occupational therapists) also completed a form at each clinical visit indicating prescribed treatment and opinions regarding patient adherence. Clinical information, obtained from the patient's chart at each clinic visit, included type of JIA, active joint count, date of diagnosis, and prescribed treatments. The study was approved by the institutional review boards of the Montreal Children's Hospital, British Columbia's Children's Hospital, and the University of British Columbia.
Perceived adherence to treatment was evaluated as the caregivers' response, using a 100-mm visual analog scale on the PARQ, to the question of how often they followed treatment recommendations as prescribed by the health care provider in the past 3 months. This was asked for each type of treatment (medications, exercises). The PARQ has been shown to have satisfactory construct validity and test–retest reliability (intraclass correlation coefficients [ICCs] ranged between 0.60 for medications and 0.88 for exercises) (28). Caregivers were also asked about perceptions of helpfulness of each type of treatment, who was responsible for the treatment (parent, child, or both parent and child), and whether they encountered any problems with respect to the medical treatment of their child. Problems were defined as organizational (e.g., long waits, difficulty getting appointments, seeing different doctors each time, etc.), communication related (e.g., difficulty in understanding clinician's explanations and instructions, etc.), and treatment related (e.g., side effects, difficulty with the treatment, etc.). The treating professional (physician and/or physical therapist) was also asked to indicate his or her perception of the patient's adherence to treatment on a 100-mm visual analog scale (a separate form was completed by the health care professional at each visit).
Clinical information was abstracted from charts (which contained standardized forms for the inclusion of diagnosis, disease duration, active joint counts, and child's age and sex) and demographic information (e.g., socioeconomic status) was procured from the study questionnaire. Psychological distress was measured with the Symptom Checklist-90-Revised, a widely used and validated 90-item self-report measure of symptoms (9 symptom clusters) experienced during the past week. The clusters are summed to create a global severity index. Clinically important distress corresponds to T scores ≥63 (29). Test–retest reliability ICC scores range between 0.47 and 0.69 and internal consistency Cronbach's alpha range between 0.72 and 0.91 (33).
Parental coping with the child's illness was measured with the CHIP (30, 31), a self-report measure of parents' perceptions of their response to management of family life when a child is seriously or chronically ill. Coping behaviors are grouped into 3 subscales: maintaining family integration, cooperation, and an optimistic definition of the situation; maintaining social support, self-esteem, and psychological stability; and understanding the medical situation through communication with other parents and consultation with the medical staff. Higher scores are indicative of higher utilization of specific coping behaviors. The CHIP has good internal consistency, with alpha values >0.70 for the 3 subscales and fair concurrent validity as evidenced by significant correlations with independent measures of family environment and changes in children's health.
Child functioning/disability was measured by the CHAQ disability index (DI) (32). The CHAQ DI is scored between 0 and 3, with 0–1 indicating mild to moderate disability, 1–2 indicating moderate to severe disability, and 2–3 indicating severe to very severe disability (34). The CHAQ has excellent reliability and validity and good discriminative properties. Furthermore, mean scores for parents and children are highly correlated, suggesting that parents can reliably report for their children (35).
Descriptive statistics were used to characterize the cohort in terms of adherence to medication and exercise, JIA diagnostic category, age, sex, socioeconomic status, and disease severity and duration. Adherence was categorized into tertiles, and agreement between parent and clinician perceptions of adherence was compared using weighted kappa coefficients.
Generalized estimating equations (GEE) were used to describe adherence and related factors over the four 3-month intervals: baseline to 3 months, 3–6 months, 6–9 months, and 9–12 months. The GEE method is an extension of generalized linear models that accommodates the correlated nature of repeated data collected per observation in longitudinal studies. In our GEE analysis, we used independent correlation structure and the cumulative logit link function because the outcome variable had a multinominal distribution. Results were based on empirical standard error estimates. The odds ratios calculated from this model indicated the association between each explanatory variable (factor) and the outcome. The factors that were considered were based on the WHO model (3) and included social factors (age, sex, parental distress, parental coping), economic factors (maternal education level, family income), factors related to the health care team (professional's perception of adherence), parental perceived problems (organizational, communication, and treatment related), condition-related factors (disease duration, severity, active joint count, CHAQ disability index), and patient beliefs (parental perceived helpfulness of treatments and distribution of responsibility for treatments). Adherence questions were asked at the end of each 3-month period (questions referred to the past 3 months, but were posed at the end of that 3-month period), whereas severity was assessed at the beginning of each 3-month period. Because adherence referred to the preceding 3-month interval, the independent variables (e.g., parental distress, parental coping, and severity of disease) referred to the beginning of (severity) or some time during (parental distress, coping) the 3-month interval. The logic for this method was that factors associated with adherence have to precede the parent-assessed outcome to be relevant. All analyses were performed using SAS software (SAS Institute, Cary, NC).
There were 235 patients who consented to participate: 144 of 157 who were approached in Montreal and 91 of 97 approached in Vancouver. However, only 182 participated in the study: 120 from Montreal and 62 from Vancouver (nonrespondents did not return the questionnaires, even after 3 reminders). Seven patients were further excluded from the Montreal sample because they did not have active disease over the year of study and did not undergo any active treatment (either medication or exercise). Therefore the final sample size was 175 patients. There were no differences between participants and nonparticipants in age at disease onset or age at recruitment; however, mean active joint count was higher in participants (1.8 versus 0.6; P = 0.001). General characteristics of our sample are described in Table 1. A total of 69% of the sample were female and the mean ± SD age was 10.2 ± 4.4 years (range 2–18 years). In terms of type of disease, the percentages for oligoarthritis, polyarthritis, systemic arthritis, enthesitis-related arthritis, psoriatic arthritis, and other arthritis were 44.6%, 20.6%, 9.1%, 10.3%, 9.7%, and 5.7%, respectively.
|Age, mean ± SD (range) years||10.2 ± 4.4 (2.0–18.0)|
|Active joint count, mean ± SD (range)||1.8 ± 3.7 (0.0–29.0)|
|Disease duration, mean ± SD (range) years||4.1 ± 3.6 (0.08–15.6)|
|CHAQ disability index, mean ± SD (range)||0.3 ± 0.4 (0.0–1.9)|
|Parental distress (t-GSI), mean ± SD (range)||50.5 ± 12.3 (30.0–81.0)|
|Coping scores (CHIP), mean ± SD (range)|
|Family integration (n = 157)||38.4 ± 9.0 (6.0–51.0)|
|Social support (n = 155)||33.3 ± 11.7 (2.0–57.0)|
|Medical information (n = 157)||16.5 ± 6.1 (3.0–27.0)|
|Female sex||120 (68.6)|
|Cultural background (n = 168)|
|French Canadian||65 (38.7)|
|English Canadian||62 (36.9)|
|High school or less||58 (39.5)|
|More than high school||89 (60.5)|
|Annual income in Canadian $ (n = 127)|
|Any (n = 168)||69 (41.1)|
|Organization (n = 168)||46 (27.4)|
|Communication (n = 168)||7 (4.2)|
|Treatment (n = 168)||34 (20.2)|
|Responsible for the treatment (medication)|
|Parents only||118 (78.7)|
|Child only||21 (14.0)|
|Child and parents||11 (7.3)|
|Responsible for the treatment (exercise)|
|Parents only||86 (65.7)|
|Child only||27 (20.6)|
|Child and parents||18 (13.7)|
|Perceived helpfulness of medication|
|Perceived helpfulness of exercise|
Parent-reported adherence varied between 86.1 and 92.0 for medications, and was considerably lower for exercise (54.3–64.1). The physician and physical therapist ratings ranged from 89.5 to 91.7 and 34.9 to 50.4, respectively. The values for medication adherence were quite consistent between parents and physicians, whereas parents reported a higher exercise adherence than did the physical therapist. There were only small fluctuations in reported adherence over the various intervals in the 1-year period. Specific values for perceived adherence over each 3-month interval by caregivers and health professionals are listed in Table 2.
|Baseline||136; 86.1 ± 26.0||117; 89.5 ± 17.8||122; 54.5 ± 31.6||39; 50.4 ± 29.8|
|Month 3||92; 91.7 ± 16.9||78; 91.7 ± 16.4||84; 64.1 ± 29.6||26; 43.6 ± 33.3|
|Month 6||86; 90.4 ± 19.1||77; 90.8 ± 18.2||78; 61.2 ± 29.2||31; 43.5 ± 30.2|
|Month 9||68; 92.0 ± 14.7||60; 91.0 ± 17.8||56; 63.0 ± 28.4||18; 34.9 ± 26.0|
|Month 12||73; 88.8 ± 19.8||67; 90.1 ± 20.0||67; 54.3 ± 29.9||17; 43.9 ± 36.1|
We evaluated agreement between the parent's assessment and the clinician's assessment of adherence. Because adherence was not normally distributed, we opted to categorize adherence into tertiles (based on the parent's perceptions of the first interval) and compared parent and clinician adherence using weighted kappas. The weighted kappa for parent–physician agreement beyond chance (first interval, measured as adherence in the past 3 months at 3 months postbaseline) for extent of adherence to medication was extremely low at 0.26 (95% confidence interval [95% CI] 0.07, 0.45), and the weighted kappa for parent–physical therapist agreement on adherence to exercise was 0.44 (95% CI 0.16, 0.71). Percent agreement for adherence over the first 3 months was 51.4% and 59.1% for medication and exercise, respectively. Agreements at each of the intervals are described in Table 3; however, it must be noted that a low number of therapists completed the adherence to exercise assessments and these agreements are based on this limited sample and are therefore considered tentative.
|Kappa (95%CI)||0.12 (−0.02, 0.27)||0.06 (−0.25, 0.37)|
|Kappa (95%CI)||0.26 (0.07, 0.45)||0.44 (0.16, 0.71)|
|Kappa (95%CI)||0.19 (−0.01, 0.38)||0.07 (−0.23, 0.37)|
|Kappa (95%CI)||0.17 (−0.03, 0.38)||–|
|Kappa (95%CI)||−0.02 (−0.21, 0.17)||0.61 (0.21, 1.0)|
We used 2 separate GEE analyses to investigate factors related to perceived adherence to medication and exercise (Table 4). Adherence to medication was higher for participants who perceived medication to be highly beneficial, as well as those who had children with lower disease severity (as measured by active joint count). Adherence to exercise was higher for participants who perceived exercise to be a highly beneficial treatment for JIA, for participants with younger children, and for participants with a child who was involved in the responsibility for his or her treatment.
|Younger age (<13 years)||0.88 (0.43, 1.81)||3.90 (1.71, 8.87)†|
|Male sex||1.62 (0.83, 3.15)||1.48 (0.75, 2.94)|
|Low parental distress (t-GSI ≤ 63)||0.67 (0.27, 1.66)||1.35 (0.67, 2.72)|
|Low social support (less than median)||1.02 (0.58, 1.79)||0.84 (0.45, 1.59)|
|Low family income (<75,000)||0.78 (0.41, 1.47)||1.64 (0.77, 3.51)|
|No perceived problem||0.99 (0.56, 1.73)||0.73 (0.41, 1.30)|
|Disease duration <3 years||1.49 (0.80, 2.80)||0.67 (0.34, 1.32)|
|Lower disease severity (AJC = 0)||1.87 (1.10, 3.16)†||0.95 (0.55, 1.66)|
|Low disability (CHAQ disability index ≤ 1)||0.90 (0.48, 1.67)||1.52 (0.84, 2.78)|
|Patient or caregiver related/beliefs|
|Low||Reference category||Reference category|
|Moderate||1.38 (0.68, 2.80)||0.92 (0.44, 1.93)|
|High||5.15 (2.67, 9.93)†||3.18 (1.44, 7.04)†|
|Child involvement in treatment||0.75 (0.36, 1.55)||2.12 (1.02, 4.39)†|
Caregiver perceptions of adherence to pharmacologic treatment were consistently higher than their perceptions of adherence to prescribed exercise. Both perceptions tended to increase slightly and then return to the baseline measure. Hayford and Ross also reported substantially better adherence to medication than exercise (95% versus 67%) in children with JIA (20), as did Sturge et al (92% versus 30%) (19). In a study evaluating perceptions regarding adherence in JIA, parents indicated that exercises and splint wearing were more problematic than taking medications (12).
Although mean perceptions of the extent of adherence to medication were similar for caregivers and physicians, agreement was extremely low. With respect to adherence to exercise, physical therapists had consistently lower perceptions than caregivers and agreements were low. Perhaps parents tended to overreport adherence (somewhat similar to a social desirability bias). In contrast, parents have more opportunity to observe their children than clinicians and may be more accurate in their assessment of adherence. Nevertheless, we did not find that clinicians tended to be biased by rating patients with better outcomes (both lower disease activity and disability score) as being more adherent (data not shown). Another possibility is that parents believe that they are adhering to treatment, although they actually may not be following treatment recommendations (22). This situation may occur if there is a lack of understanding or poor communication between the health care provider and the family.
The factor that was most highly associated with perceived adherence was belief that the treatment was helpful. This finding was true for both medication and exercise, as those who perceived a high degree of helpfulness of the treatment were much more likely to adhere to that treatment. This finding underscores the importance of patients' beliefs affecting adherence. Explanations about the treatments, how they work, and why they are important may potentially influence these beliefs. Educational intervention has been shown to be beneficial in improving adherence for younger children with JIA (9). A more recent randomized controlled trial indicated that adherence to treatment in JIA was significantly improved by a combined behavioral and educational intervention (36). In adults with various chronic diseases, understanding of their condition has been positively related to adherence (16, 37, 38), as have patients' beliefs in the benefits of the therapeutic regimen (17, 24, 39). Tuckett et al (37) found that 77% of patients who believed in the physician's diagnosis and treatment plan were committed to following the recommended course of action, compared with 50% of patients who did not concur with the physician. Patient adherence to treatment appears to be related to the ability of the physician to answer patients' questions, and patients tend to continue seeing physicians who they perceive to be more caring and open to communication as well as those who have more participatory styles (27, 38). Although we did not measure motivation, Kyngas et al reported that motivation and support from physicians, family, and friends were factors supporting high adherence in adolescents with chronic disease (39).
Children who had a lower number of active joints (less severe disease) tended to adhere more to taking medications. It is possible that patients who adhere to treatment reduce their disease activity, and therefore continue to adhere. Even though disease activity was measured prior to measuring adherence, these patients did not have recent-onset disease. Because they had had the disease for some time, a lower degree of disease severity may have been associated with past adherence, thereby favoring continuing adherence. This may, in part, explain the discrepancy between our findings and those of Rapoff et al (40) who found that higher active joint count was associated with adherence in patients with new-onset JIA. In a study of children with epilepsy, Mitchell et al (1) found that adherence to medication was not associated with seizure severity.
We found no association between socioeconomic status and adherence in JIA, unlike a previous report (40). This may be due to universal health insurance in Canada, which mandates equal access to health care for all Canadians.
Older children were less adherent to exercise. Other researchers have also reported that older children/adolescents have lower adherence to treatment than younger children (41–43). In the general population, older children tend to exercise less and be involved less in physical activity than younger children (44). Reasons for this may include lower motivation, lack of good time management, and other commitments such as part-time work (45).
Finally, having the child involved in his or her treatment was associated with adherence to exercise. Treadwell et al reported that sharing of responsibilities between children with sickle cell disease and their parents supported higher adherence (46). Adherence to treatment was increased in patients who participated in their own care (47, 48). Jensen and Lorish (49) have suggested a process model for patient–practitioner collaboration whereby the cooperation on an exercise regimen is mediated by the patient's belief system and requires a therapeutic process of mutual inquiry, problem solving, and negotiation between the therapist and patient.
There are several limitations to the present study. First, self-reported adherence may be biased because parents may overestimate adherence and there may be some social desirability bias. Although adherence to medication was very high, adherence to exercise was not, which may indicate that parents were actually quite truthful regarding perceived adherence. In a previous study, we verified agreements regarding adherence between children and parents and found fair agreement for medication (ICC 0.30) and high agreement for exercises (ICC 0.80) (50). Second, participants included in the study may have been those who were most likely to adhere. This would represent a selection bias, and there may have been an overestimation of adherence in our study. Loss to followup over the 1-year period (each 3-month interval assessment of adherence) may also add to this limitation, because perhaps those who were less adherent would tend to drop out. This may be one reason for the relatively stable adherence levels over time. If, in fact, our sample represents individuals who would be considered to have high adherence to treatment, factors associated with adherence in this group appear similar to those described by other researchers in treatment adherence in other diseases or conditions (17, 26, 41–43, 46, 51). Third, recall may pose a problem, although our previous work comparing patient/caregiver logs and the PARQ indicated a high level of validity of the 3-month measure of adherence on the PARQ (28). Finally, this study addressed factors associated with adherence to treatment; however, we did not evaluate factors associated with patterns of adherence (either continual adherence, nonadherence, increasing adherence, or decreasing adherence). This will be addressed in a future analysis looking at factors associated with trajectories of adherence.
The conclusions drawn from this study were that perceived adherence to medication in JIA was high, whereas perceived adherence to exercise was only fair. Agreements (beyond chance) regarding perceived adherence for exercise and medication were low between parents and clinicians. Because patients' beliefs in helpfulness of treatment are highly associated with higher levels of adherence, better explanations regarding the treatment and the way it works may be extremely important. Also, involving children as partners in their own treatment may favor higher levels of adherence. These strategies may be even more important among older children to maximize adherence while ensuring that their independence is respected. Finally, there is a need to incorporate these and other strategies to improve adherence to exercise in children with JIA.
Dr. Ehrmann Feldman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Ehrmann Feldman, De Civita, Dobkin, Malleson, Duffy.
Acquisition of data. Ehrmann Feldman, Dobkin, Malleson, Duffy.
Analysis and interpretation of data. Ehrmann Feldman, De Civita, Meshefedjian.
Manuscript preparation. Ehrmann Feldman, De Civita, Malleson, Meshefedjian, Duffy.
Statistical analysis. Ehrmann Feldman, Meshefedjian.
Recruitment document preparation. De Civita.
Instrument development. Dobkin.