Study design
This is a prospective cohort study with a matched pair design conducted between July 2016 and November 2017.
Participants
Participants with JIA (JIA group) were recruited by their clinician in collaboration with the local pediatric outpatient rheumatology clinic and an external outpatient rheumatology clinic. Participants who gave consent / assent were contacted by phone by the research team for a screening interview and to book a testing session. Healthy control participants (CON group) were recruited using an online research study portal and through friends and family of participants. Inclusion criteria for participants with JIA included: age 10-20 years old; a diagnosis of JIA by a physician with bilateral or unilateral knee involvement; and active or inactive disease at the time of testing. Patients were not eligible if systemic symptoms were present, if medications had changed during the three weeks prior to testing, or if they had active ankle joint involvement. The CON group had no history of JIA or other rheumatological diseases. Exclusion criteria for all participants included: contraindications as indicated on the Physical Activity Readiness Questionnaire for Everyone (PAR-Q+) [20]; previous lower extremity musculoskeletal injury within 3 months prior to testing that resulted in time loss (work, school, or sport); diagnosis of any other arthritides; pregnancy. Ethics approval was granted by the Conjoint Health Research Ethics Board (Ethics ID: REB15-3125).
Data Collections
This study was performed as part of a larger study on the secondary consequences of JIA including dynamic balance, fitness, movement biomechanics and body composition outcomes. Data collections for each study participant were conducted on two separate days. On day 1 participants with JIA were assessed by a pediatric rheumatologist (SB) to record disease history and disease activity and by a pediatric physiotherapist (JB). All participants completed surveys on demographics and knee function and completed a series of testing stations: anthropometrics; dynamic balance; aerobic capacity; and movement biomechanics. Following day 1, participants were provided with an ActiGraph GTX+ (ActiGraph Inc., USA) to measure physical activity and were instructed on the appropriate use. Participants then recorded physical activity for seven consecutive days. Day 2 testing was conducted one week after day 1 testing and involved a body composition assessment by a trained researcher using dual-energy absorptiometry (DXA, Hologic QDR 4500A, Hologic Inc., USA).
Outcomes
Physical activity was measured using the body worn ActiGraph. The ActiGraph has demonstrated excellent validity and classification accuracy for MVPA in adolescents with and without disability [21]. Participants were asked to wear the device for seven consecutive days during waking and sleeping hours and to follow their regular schedule and activities during this time. The ActiGraph was waist mounted on the right hip using an elastic strap. Participants completed an activity monitor log to record physical activity during non-wear times; i.e. when the ActiGraph was removed for showers or swimming. Participants recorded their average rate of perceived exertion (RPE) using the Pictorial Children’s Effort Rating Table (PCERT) for any activities completed without the ActiGraph [22]. Time spent in any activities for which participants recorded an RPE of 5-10 was included in minutes of MVPA. This is a conservative non-wear time estimate, in line with the definition of MVPA by the Canadian 24-hour Movement Guidelines [23]. If RPE was not recorded, participants were contacted for follow up. If RPE could not be obtained, Ridley’s Compendium of Energy Expenditures for Youth was used to estimate RPE [24] and MVPA was defined as ≥4 metabolic equivalents [25]. The total MVPA duration was then divided by the number of valid days and expressed as the average time spent in MVPA per day over the wear period.
MVPA was analyzed using ActiLife (v6.13.3, ActiGraph Inc., USA). Data were collected at 30Hz, at 10 second epochs. Age-specific algorithms by Evenson et al. [25,26] were chosen due to their superior ability to predict energy expenditure in youth across physical activity intensity levels [25]. Here, moderate activity was defined as ≥2296 counts/minute and <4011 counts/minute and vigorous activity as ≥4012 counts/minute [25]. Physical activity data were included if the participant had worn the device for 5-7 days, including at least one weekend day, and for >10 hours per day. If the participant wore the device for >7 days, the days with the longest wear time were included for analysis. Every effort was made to include information on self-reported activities and participant follow up was performed to clarify physical activity participation if self-report data did not correspond with ActiGraph data.
Aerobic capacity was determined using the relative peak oxygen consumption (VO2 peak; mL/kg/min) during an incremental maximal fitness test using a cycle ergometer (Ergoline GmbH, Germany). Participants started at 0W and resistance was increased in 20W intervals every 2 minutes until volitional fatigue, a plateau in maximal heart rate, and/or a reported RPE of 10 on the PCERT scale [22]. Respiratory gases were collected and analyzed using the COSMED K5 portable metabolic system (COSMED, Italy).
Adiposity was determined using DXA and expressed as a fat mass index [FMI; fat mass/height (kg/m2)]. Participants were asked to lie supine on the scan bed and remain still for the duration of the scan. The DXA was calibrated prior to each scan as per the manufacturer’s recommendations and all procedures were consistent with the official positions of the International Society of Clinical Densitometry [27].
Dynamic balance was assessed using a triple single leg hop test (TSLH). This test evaluates neuromuscular control, force generating capacity, and knee stabilization [28,29] and has shown moderate correlation to the Global Rating Scale (r=0.44) [30] and Lower Extremity Functional Scale (r=0.26) [31]. Participants performed a practice trial followed by two test trials in which participants performed three consecutive single-leg hops with the goal of jumping as far as possible [32]. For a trial to be included, the landing following the last jump had to be solid without excessive movement or twisting of the foot. The maximum distance across trials was recorded for each leg and expressed with respect to % leg length.
Further, self-reported physical disability was assessed using the Childhood Health Assessment Questionnaire (CHAQ) [33]. The CHAQ captures the health and functional status of patients across eight domains using a score of 0 and 3 (higher scores indicate greater functional disability). Disease activity was determined by clinical exam and scored using the Juvenile Arthritis Disease Activity Score (cJADAS10) [34]. cJADAS is a sensitive continuous score of disease activity developed for use with individuals with JIA [34]. cJADAS comprises measures of active joint count (10 joints), physician global assessment of disease activity, and evaluation of the child’s wellbeing.
Data Analysis
Matched pairs were assigned based on closest age in months (≤18 months difference between pairs) and sex. Statistical analyses were performed using R (R Core Team, Austria). Demographics data were summarized using medians, and minimum and maximum values for numerical data, and frequencies and percentages for categorical data by group (JIA, CON), sex (male, female) and matched pairs. Pair differences between individuals in the JIA and CON groups for primary (MVPA, Aerobic Fitness) and secondary (FMI, TSLH) outcomes were graphed to assess for normality and outliers. To evaluate differences between normally distributed matched pair outcomes, t-confidence intervals (CI) corrected by the Bonferroni Correction ([1-alpha/(2*number of tests)]% CI) were reported. If the pair differences were not normally distributed, the medians and CIs ([1-alpha/(2*number of tests)]% CI) were determined using the Hodges-Lehmann method. A statistical difference was determined if the CI excluded zero. Further, the effect of sex on the primary and secondary outcomes were examined using medians and first and third quartiles for each study group and the respective pair differences, due to the small sample.