Disease control and health-related quality of life in juvenile idiopathic arthritis
To examine variability in health-related quality of life (HRQOL) in children with juvenile idiopathic arthritis (JIA) experiencing no or minimal clinical symptoms, and in a subgroup with polyarticular JIA treated with biologic agents for 12 months.
We defined 3 samples using a database of patients ages 2–18 years with JIA (n = 524; patient visits [PV] = 2,354): visits (PV = 2,155) with no or minimal clinical symptoms on at least 1 of 4 measures (active joint count, pain, physician global disease rating, Childhood Health Assessment Questionnaire); visits (PV = 941) with no or minimal symptoms on all 4 measures; and children (n = 31) with polyarticular JIA treated with biologic agents for 12 months. HRQOL was measured using the Pediatric Quality of Life Inventory (PedsQL) and the percentage of patients with suboptimal HRQOL was determined.
In PV with a PedsQL score, suboptimal HRQOL by self-report occurred in 362 (20.6%) PV with at least 1 indicator of minimal symptoms, and in 64 (7.9%) PV with all 4 measures indicating minimal symptoms (519 [25.7%] and 95 [10.7%], respectively, by parent report). For children with polyarticular JIA treated for 12 months with biologic agents, 7 (25.9%) patients by self-report and 10 (35.7%) patients by parent report were in the suboptimal range of HRQOL.
A substantial percentage of patients with JIA who report no or mild clinical symptoms experience suboptimal HRQOL. This is also true for patients with polyarticular JIA treated with biologic agents for 12 months. Although disease activity and clinical symptoms are related to HRQOL, considerable unexplained variation in HRQOL exists. HRQOL needs to be assessed independently regardless of clinical status.
Juvenile idiopathic arthritis (JIA), the most common childhood rheumatic disease, has no cure and can result in significant pain and physical disability. Because JIA can influence virtually all aspects of a child's and his or her family's life, clinicians and researchers increasingly recognize that improved health-related quality of life (HRQOL) is a key treatment goal for JIA (1). The importance of HRQOL as a primary outcome is echoed by the US Food and Drug Administration (2), the Centers for Disease Control and Prevention (3), and the World Health Organization (WHO) (4).
HRQOL refers to how an individual feels about certain aspects of their life with respect to their health or health condition. HRQOL consists, at a minimum, of physical, mental, and social health dimensions, originally delineated in 1948 by the WHO (4). Although researchers have documented worse HRQOL for children with JIA compared with healthy controls (5–7), others have found minimal differences between these groups in psychosocial outcomes such as parental distress, family functioning, social support (8), and HRQOL (9). Despite these equivocal findings, HRQOL is known to be associated with medical variables in children with JIA. Recent studies showed that better self-report (10) and parent proxy-report (5, 6) of child HRQOL is associated with better parent ratings of child functional ability, pain intensity, well-being, physician rating of disease activity, and erythrocyte sedimentation rate (ESR). These relationships are in the small to medium effect size range (11). Whereas functional ability, physician rating, and ESR were related to the physical functioning scales of the HRQOL measures, ratings of pain and well-being were related to the psychosocial functioning scales. Given the well-documented lack of concordance between self- and proxy-report in the adult and pediatric literature (12, 13), it is important to assess both child self-report and parent proxy-report HRQOL whenever possible.
Implicit in the measurement of HRQOL is the notion that medical interventions can affect not only clinical parameters (e.g., pain, joint count) but also more distal outcomes, such as HRQOL. Although much research has focused on understanding how biologic and physiologic variables affect disease outcomes, far less is known about how clinical parameters and therapeutic interventions are related to HRQOL. This knowledge gap is particularly relevant in the context of new biologic therapies used in the treatment of JIA due to potential long-term safety issues (currently unknown) and the high costs of the therapies. These new therapies have resulted in statistically significant mean increases (small to medium effect sizes) in HRQOL for adults with rheumatoid arthritis (RA) (14, 15). However, the results of these trials highlight a key issue: substantial unexplained variation in HRQOL outcomes between patients with similar disease control (i.e., dissociation between traditional medical outcomes and HRQOL outcomes). For example, in these recent RA clinical trials, the SDs for HRQOL within a treatment group are sometimes larger than the average HRQOL improvement, and 5–15% of the treatment group may actually show a worsening in HRQOL (14, 15). Although biologic agents represent great potential for improved HRQOL in children with JIA, such improvement is by no means assured.
Given the growing understanding of HRQOL as an important outcome, the traditional focus of therapeutics on medical outcomes, and the relative lack of knowledge regarding how clinical parameters and therapeutic interventions affect HRQOL in children with JIA, there is a need to understand parent and child perceptions of HRQOL for children with minimal symptoms and for children treated with biologic agents. The current study addresses this need by describing variability in HRQOL among children with JIA. We hypothesized that although clinical functioning would be correlated with HRQOL (5, 6, 10), significant variation in HRQOL would be found in groups of children with JIA experiencing minimal clinical symptoms, as well as for a subgroup with polyarticular JIA treated with biologic agents (14, 15).
PATIENTS AND METHODS
This study was based on a clinical database prospectively collected by clinical protocol in the pediatric rheumatology clinic at Cincinnati Children's Hospital Medical Center between 2003 and 2007. All children ages 2–18 years old (n = 524) who presented for evaluation and treatment of JIA and completed the Pediatric Quality of Life Inventory (PedsQL) during their visit (patient visits [PV] = 2,354) were included in the analysis.
We examined HRQOL for 3 nonexclusive subsets of this database, selected in order to create more homogeneous clinical samples for assessing HRQOL variation. The first subset consisted of visits with a PedsQL score at which patients had minimal symptoms on at least 1 of 4 clinical characteristics (n = 500, PV = 2,155); the second, visits with a PedsQL score at which patients had minimal symptoms on all 4 clinical characteristics (n = 308, PV = 941); and the third, children with polyarticular JIA who had a visit with a PedsQL score within 30 days of having been treated with biologic therapies for 12 months (n = 31, PV = 31), irrespective of symptom control. For this last subsample, we determined presence or absence of biologic therapy rather than an ordinal scale based on dose, because studies of tumor necrosis factor inhibitors have shown similar profound effects on disease control in JIA, and because there are no comparative studies among biologic agents to support a dose-response relationship (16–18).
Note that these samples are subsets of each other, not unique subsets, i.e., in 2,155 PV the patient reported minimal symptoms on at least 1 of 4 measures (a subset of the 2,354 total PV in the analysis); in 941 PV the patients had minimal symptoms on all 4 measures simultaneously (a subset of the 2,155 PV); in 402 PV the patients had a score of 0 on all 4 measures simultaneously (a subset of the 941 PV); and in 31 patients the biologic therapy criteria were met (a subset of the 524 patients).
HRQOL was measured using the PedsQL generic core scales (19) and rheumatology module (7, 20, 21). The PedsQL consists of parallel forms for children (ages 5–18 years) to report on their own HRQOL and for parents of children ages 2–18 years to report on their child's HRQOL. The PedsQL generic core scales and rheumatology module have been shown to be reliable, valid, sensitive to disease severity, and responsive to change in JIA (7). We defined optimal HRQOL as a score no less than 1 minimum clinically important difference (MCID) below the mean of healthy children on the PedsQL generic core total scale. The MCID is the smallest difference in a domain of interest that patients perceive to be beneficial and that would mandate, in the absence of troublesome side effects and excessive costs, a change in the patient's care management (22). The PedsQL MCID is 4.4 points (22) (larger than the minimal detectable difference in this sample of 0.90 = [1.96 × square root of 2 × SEM] ), and the mean score of healthy children is 83 points (19, 22). Thus, a PedsQL generic core total scale score <78.6 is defined as suboptimal.
Medical variables were collected as part of the standard of clinical care. These included disease duration; JIA subtype, active joint count, and patient pain measured using a visual analog scale (VAS; where 0 = no pain and 10 = very severe pain) by patient report or parent proxy (24); physician global assessment of disease activity using a VAS (where 0 = inactive disease and 10 = very active disease) (25, 26); and physical function, measured using the Childhood Health Assessment Questionnaire (C-HAQ) disability index (27, 28). This index is calculated as the unweighted average of 30 questions in 8 domains covering major aspects of daily living over a 1-week period, and it yields a score between 0 (no disability) and 3 (most severe disability).
Minimal symptoms was defined as active joint count ≤2 joints, patient (or parent) pain rating ≤2, physician global assessment of disease activity ≤2, and C-HAQ disability index ≤0.13 (mild disability) (29).
Patient demographic and disease characteristics were summarized on their first visits. Generalized estimating equation (GEE) analyses were conducted to examine the association between clinical characteristics and the PedsQL, taking into account the clustered data structure. All GEE analyses included patient's age, sex, race, ethnicity, age at onset of disease, and duration of disease as covariates. For subsets of patients with minimal symptoms on at least 1 clinical characteristic and on all 4 clinical characteristics, the number of patient visits, number of patients, corresponding PedsQL means, and variance between and within patients were reported. In addition, using 78.6 as the cut point, the number and percentage of patient visits and patients falling below optimal level were determined. All analyses were carried out using SAS software, version 9 (SAS Institute, Cary, NC). Analysis of this deidentified data set was approved by the Institutional Review Board at Cincinnati Children's Hospital Medical Center.
The sample consisted of 524 children with a total of 2,354 PV. Mean ± SD age upon entering the study was 10.00 ± 4.67 years (range 2–18 years). Mean ± SD age at diagnosis was 5.53 ± 4.15 years (range 0–16 years). The sample was predominantly female (387 [73.9%] patients) and white (475 [90.6%] patients). Mean ± SD disease duration was 4.42 ± 4.26 years and was categorized into the following subtypes for 524 patients: oligoarticular extended (46 [8.8%] patients), oligoarticular persistent (184 [35.1%] patients), polyarticular rheumatoid factor (RF)–negative (202 [38.5%] patients), polyarticular RF-positive (27 [5.2%] patients), psoriatic arthritis (2 [0.4%] patients), systemic JIA (56 [10.7%]), and enthesitis-related arthritis (2 [0.4%] patients). Subtypes were missing on 5 patients. The average number of visits per patient was 4.49. Of PV with at least 1 indicator of minimal symptoms, 1,756 self-report and 2,020 parent-report PedsQL scores existed. For visits with all 4 measures indicating minimal symptoms, the denominators were 810 and 891 for self- and proxy-report, respectively.
The results of GEE analyses are summarized in Table 1 by type III chi-square test and the corresponding P values. After adjusting for patient age, sex, race, ethnicity, age at onset of disease, and duration of disease, and accounting for the clustering of data with patients, all clinical characteristics have significant effects (P < 0.001) on PedsQL by both parent and child report. When all clinical characteristics were included, all characteristics remained significantly associated with parent report, but only the pain VAS score and C-HAQ score remained significantly associated with child report of PedsQL.
Table 1. Generalized estimating equation analysis of clinical characteristics on the PedsQL generic core scale and rheumatology module scores*
|Generic core scale|| || || || |
| Active joint count||26.61||< 0.0001||15.52||< 0.0001|
| Pain rating||86.46||< 0.0001||74.65||< 0.0001|
| Physician global assessment||47.13||< 0.0001||31.81||< 0.0001|
| C-HAQ||58.24||< 0.0001||37.66||< 0.0001|
|Rheumatology module|| || || || |
| Active joint count||35.05||< 0.0001||28.59||< 0.0001|
| Pain rating||107.00||< 0.0001||99.25||< 0.0001|
| Physician global assessment||75.63||< 0.0001||57.06||< 0.0001|
| C-HAQ||62.68||< 0.0001||44.49||< 0.0001|
Suboptimal HRQOL by self-report occurred in 362 (20.6%) PV with at least 1 indicator of minimal symptoms and in 64 (7.9%) PV with all 4 measures indicating minimal symptoms (in 519 [25.7%] PV and 95 [10.7%] PV, respectively, by parent report). Descriptive statistics (number of PV, number of patients, mean, between- and within-patient variances) and the percentage of PV and patients below an optimal level of HRQOL (i.e., <78.6 points) on the PedsQL generic core scales by child self-report and parent-proxy report, stratified by 4 clinical characteristics, are shown in Table 2.
Table 2. Number of visits and patients, mean ± SD PedsQL generic core total scale score, and number and percentage of visits and patients below the optimal PedsQL score, stratified by clinical characteristic*
|Active joint count|| || || || || || |
| 0|| || || || || || |
| Patient visits||1,135||87.07 ± 20.88||260 (22.91)||1,007||88.84 ± 18.38||181 (17.97)|
| Patients||388||85.2 ± 7.85||103 (26.55)||354||86.91 ± 6.71||76 (21.47)|
| 1|| || || || || || |
| Patient visits||314||86.07 ± 15.58||80 (25.48)||275||88.61 ± 14.59||54 (19.64)|
| Patients||191||85.72 ± 6.39||48 (25.13)||170||87.26 ± 6.49||35 (20.59)|
| 2|| || || || || || |
| Patient visits||224||83.01 ± 17.78||73 (32.59)||187||85.06 ± 15.03||53 (28.34)|
| Patients||152||82.28 ± 8.84||53 (34.87)||130||83.98 ± 9.11||41 (31.54)|
|Pain rating (0–10)|| || || || || || |
| 0|| || || || || || |
| Patient visits||948||91.49 ± 15.47||105 (11.08)||817||93.12 ± 12.76||62 (7.59)|
| Patients||310||89.57 ± 6.67||48 (15.48)||282||91.04 ± 5.23||32 (11.35)|
| 1|| || || || || || |
| Patient visits||318||86.03 ± 15.1||80 (25.16)||277||88.1 ± 13.39||56 (20.22)|
| Patients||187||85.29 ± 8.56||55 (29.41)||166||86.95 ± 6.83||36 (21.69)|
| 2|| || || || || || |
| Patient visits||206||82.76 ± 15.37||69 (33.50)||173||85.16 ± 14.01||48 (27.75)|
| Patients||143||81.7 ± 7.07||49 (34.27)||120||83.9 ± 7.48||38 (31.67)|
|Physician global assessment (0–10)|| |
| 0|| || || || || || |
| Patient visits||821||89.05 ± 17.26||151 (18.39)||740||90.66 ± 15.33||95 (12.84)|
| Patients||313||87.45 ± 7.6||70 (22.36)||287||89.16 ± 6.38||42 (14.63)|
| 1|| || || || || || |
| Patient visits||484||84.93 ± 19.09||134 (27.69)||432||87.44 ± 16.45||96 (22.22)|
| Patients||249||83.33 ± 7.76||80 (32.13)||227||86.01 ± 6.91||59 (25.99)|
| 2|| || || || || || |
| Patient visits||264||82.08 ± 17.55||91 (34.47)||215||84.24 ± 16.77||62 (28.84)|
| Patients||178||81.93 ± 7.21||62 (34.83)||147||83.59 ± 7.15||44 (29.93)|
|C-HAQ score (0–3)|| || || || || || |
| 0|| || || || || || |
| Patient visits||1,195||91.49 ± 15.18||143 (11.97)||1,085||92.28 ± 14.21||108 (9.95)|
| Patients||348||89.86 ± 6.62||53 (15.23)||331||89.49 ± 5.69||53 (16.01)|
| >0, ≤0.13|| || || || || || |
| Patient visits||223||82.84 ± 13.14||68 (30.49)||197||86.12 ± 12.28||43 (21.83)|
| Patients||158||81.96 ± 9.64||55 (34.81)||139||85.07 ± 9.03||35 (25.18)|
|Minimal symptoms on all 4 characteristics|| || || || || || |
| All characteristics 0|| || || || || || |
| Patient visits||380||94.5 ± 8.97||21 (5.53)||349||94.81 ± 8.41||14 (4.01)|
| Patients||184||93.28 ± 5.67||16 (8.70)||168||93.89 ± 4.57||8 (4.76)|
| All characteristics ≤2‡|| || || || || || |
| Patient visits||511||90.34 ± 13.92||74 (14.48)||461||91.55 ± 12.85||50 (10.85)|
| Patients||230||88.86 ± 7.24||43 (18.70)||213||89.91 ± 6.2||30 (14.08)|
As disease functioning worsened, child- and parent-reported HRQOL also worsened. Although mean PedsQL scores were in the optimal range for the group as a whole, a substantial proportion of children with minimal symptoms on each characteristic experienced scores considered to be suboptimal. Among all PV when active joint count was zero, 260 (23%) PV and 27% (n = 103) of patients had suboptimal HRQOL based on parent-proxy report, whereas 181 (18%) PV and 22% (n = 76) of patients had suboptimal HRQOL based on child self-report. Similar results were seen for pain VAS rating, physician global assessment, and C-HAQ scores. When considering minimal symptoms on all 4 characteristics simultaneously (0 < active joint count ≤ 2; 0 < pain rating ≤ 2; 0 < physician global assessment ≤ 2; and 0 < C-HAQ score ≤ 0.13), 74 (15%) PV and 19% (n = 43) of patients by the parent proxy-report, and 11% PV and 14% (n = 30) of patients by child self-report, indicate suboptimal HRQOL. Even when all 4 symptoms were absent, 21 (6%) PV and 9% (n = 16) of patients by parent proxy-report, and 14 (4.0%) PV and 5% (n = 8) of patients by child report, indicated suboptimal HRQOL.
Suboptimal HRQOL exists also for children with polyarticular JIA treated for 12 months with biologic therapies. We identified 31 patients with polyarticular JIA in the database who had been treated with biologic agents for at least 12 months and who had a visit with a PedsQL score within 30 days of completing 12 months of treatment. The mean ± SD PedsQL generic core scale score by child self-report (n = 27) was 85.4 ± 16.3, and by parent proxy-report (n = 28) was 81.3 ± 17.0. Seven (25.9%) child self-report scores and 10 (35.7%) parent proxy-report scores fell in the suboptimal range. The clinical characteristics of these patients at the time of these visits are summarized in Table 3.
Table 3. Clinical characteristics for patients (n = 31) with polyarticular JIA within 30 days of 12 months of treatment with biologic agents*
|Active joint count|| |
| 0||17 (54.8)|
| 1||5 (16.1)|
| ≥2||9 (29.1)|
|Patient pain rating|| |
| 0||11 (35.5)|
| 1||7 (22.6)|
| ≥2||13 (41.9)|
|Physician global assessment|| |
| 0||12 (38.7)|
| 1||10 (32.2)|
| ≥2||9 (29.1)|
| 0||15 (48.4)|
| >0, ≤0.13||3 (9.7)|
| >0.13||13 (41.9)|
This study demonstrates substantial unexplained variance in HRQOL for children with JIA experiencing minimal clinical symptoms and for children with polyarticular JIA treated with biologic agents for 12 months. Although better clinical scores are associated, on average, with higher HRQOL, many children with no or very mild clinical symptoms continue to experience suboptimal HRQOL. Moreover, the new class of biologic agents, despite providing profound improvement in joint inflammation, often fails to produce optimal HRQOL in children with polyarticular JIA. One of every 4 children treated with biologic agents for at least 12 months reports suboptimal HRQOL, and this proportion increases to 1 of every 3 patients by parent proxy-report.
It is unclear how change in clinical parameters is related to improvement in HRQOL. It is therefore imperative to understand how to maximize the effects of these new treatments on HRQOL in JIA, e.g., to achieve the largest gain in HRQOL possible. This contribution would be significant because improving HRQOL is a fundamental goal of the US health care system, and this knowledge will enable clinicians and researchers to maximize HRQOL gains in JIA patients by tailoring existing interventions, initiating additional interventions, and targeting other modifiable factors. For example, if studies were to find that functional status is a more important determinant of HRQOL than active joint count, clinicians could tailor clinical interventions by choosing treatment interventions that maximize improved functional status. Similarly, if studies were to find that gains in HRQOL are stronger in children who have adequate social support, clinicians could counsel patients or families during discussion of treatment options and, perhaps, refer a child to a behavioral specialist to strengthen the child's social support. Future research needs to understand both medical and nonmedical drivers of between-patient differences in HRQOL outcomes, therefore enabling clinicians to maximize HRQOL for children with JIA and to maximize the effect of JIA treatment on HRQOL.
This study had several limitations. We relied on a retrospective analysis of existing data and only had access to PedsQL total scores for the generic core scales and the rheumatology module. Although it would be useful to examine the relationship between symptoms and subscale scores for the PedsQL, we were unable to do this with the existing data set. Our outcome variable, HRQOL, is based on parent- and child-report and is subject to acquiescence, responder bias, adaptation coping, response shift, maturity, and other bias. However, the PedsQL has been subjected to rigorous testing of its reliability and validity, and it is unlikely that such bias is systematic. In terms of generalizability, our sample was predominantly female and white, which is similar to the epidemiology of JIA (30, 31). However, further studies with higher proportions of males and nonwhites would be useful in examining generalizability. Moreover, although this retrospective analysis of a clinical database represents all patients seen in the only subspecialty clinic and regional referral center in the region and is therefore likely to be generalizable to this catchment area, it is not known to what extent these results would be generalizable to patients in other clinics or to children with this diagnosis not treated in subspecialty clinics.
Results of this study have important implications for clinical practice. Although many patients with JIA who report no or mild clinical symptoms clearly are functioning in the optimal range of HRQOL, there is a significant percentage who are not. Although disease severity is clearly associated with HRQOL, it is not a proxy for this measure. HRQOL needs to be assessed independently regardless of clinical or treatment status.
The authors would like to thank Cynthia Rutherford and Lukasz Itert for their assistance in data collection and management, and T. Brent Graham, MD, Murray H. Passo, MD, and Janalee Taylor, RN, for facilitating patient recruitment and contact.
Dr. Seid 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. Seid, Brunner, Lovell.
Acquisition of data. Seid, Brunner, Lovell.
Analysis and interpretation of data. Seid, Opipari, Huang, Lovell.
Manuscript preparation. Seid, Opipari, Huang, Brunner, Lovell.
Statistical analysis. Seid, Opipari, Huang.