To compare sleep disturbances and neurobehavioral function in children with juvenile idiopathic arthritis (JIA) to age- and sex-matched control children.
To compare sleep disturbances and neurobehavioral function in children with juvenile idiopathic arthritis (JIA) to age- and sex-matched control children.
Children (n = 116) ages 6–11 years with (n = 70) and without (n = 46) JIA and their parents participated. Parents completed questionnaires on sleep habits, sleep behavior, and school competence of their children; children completed computerized neurobehavioral performance tests.
Compared to control children, children with JIA had a statistically significant (P < 0.001) greater mean overall sleep disturbance score and higher scores on 6 of 8 subscales (all P < 0.03) of the Children's Sleep Habits Questionnaire (CSHQ). There were no group differences on neurobehavioral performance test scores. However, regardless of group, children with an overall CSHQ score above an established cutoff for clinically significant sleep disturbances had slower mean simple reaction time (t = −2.2, P < 0.03) and mean 5-choice reaction time (t = −2.3, P < 0.02) compared to those below the cutoff score. The CHSQ overall sleep disturbance score predicted reaction time (P < 0.009) after controlling for age, intelligence quotient, medication, and group.
Children with JIA have more parent-reported sleep disturbances, but performed as well as control children on a series of standardized computer tests of neurobehavioral performance. Children with more disturbed sleep had slower reaction times.
Sleep in America's youth is an issue of growing concern. An estimated 25% of children in the US have some type of sleep disturbance, ranging from sleep disorders (e.g., primary snoring, obstructive sleep apnea) to behavioral disorders (e.g., behavioral insomnia of childhood) (1). Disturbed sleep in children has been associated with daytime sleepiness, poor neurobehavioral performance, and problematic behaviors (e.g., hyperactivity, decreased attention span, distractibility, impulsivity) (2–5).
Children with juvenile idiopathic arthritis (JIA) report poor sleep quality and daytime sleepiness, and parents of children with JIA report symptoms of sleep-disordered breathing (SDB) and daytime sleepiness (6–8). Polysomnographic measures of arousals, awakenings, arousal-associated periodic extremity movements, and indices of SDB provide objective evidence of disturbed sleep in JIA (2, 9, 10). Habitual snoring has been considered benign, but recent findings suggest associations between snoring and behavioral disturbances, poor school performance, cognitive deficits (2, 11–15), and disturbed sleep (arousals, night awakenings, and delta sleep instability) (16). We recently reported that 19% of the sample of children with JIA had sleep latencies of <10 minutes, which is clinically indicative of excessive daytime sleepiness. We also found that after controlling for age, intelligence quotient (IQ), medication, and pain, indices of disturbed sleep were inversely related to reaction time and sustained attention (2). Disease status (active versus inactive) was unrelated to neurobehavioral performance. This observation is consistent with findings from a previous study of cognitive function that showed no differences in tests of memory, fine motor performance, and sustained attention between children with systemic rheumatoid arthritis (17) and healthy children. Disturbed sleep and daytime sleepiness could negatively affect neurobehavioral and school performance in JIA, but few studies have been reported. In the present study, we sought to compare sleep habits, parent-reported sleep disturbances, and neurobehavioral and school performance in children with JIA to age- and sex-matched control children.
Approval for this study was obtained from the Institutional Review Board at the Seattle Children's Hospital in Seattle, Washington. From April 2004 through August 2007, a convenience sample of 70 children with JIA (53 girls) and 46 age- and sex-matched control children (30 girls) ages 6–11 years and their parents were enrolled in this study. Children were excluded if they had a diagnosis of active systemic JIA, a psychiatric condition, attention deficit hyperactivity disorder (ADHD), diabetes mellitus, asthma, or cancer; a family history of narcolepsy in a first-degree relative; or a handicap that would interfere with neurobehavioral performance testing. The mean disease duration for children with JIA was 3.6 years. Of the 70 children with JIA, 37% (n = 26) had oligoarticular disease, 57% (n = 40) had polyarticular disease, and 5.7% (n = 4) had inactive systemic disease. Fifty-six percent (n = 39) had active arthritis (defined as inflammation of one or more joints with swelling, limited range of motion, or tenderness [≥1 on a scale of 0–10]) and 44% (n = 31) had inactive arthritis (defined as a lack of inflammation, limited range of motion, or tenderness [0 on a scale of 0–10]) (18).
Parents completed the CSHQ, a 45-item retrospective report that assesses bedtimes, wake times, sleep behaviors, and sleep problems over a typical week (19). The CSHQ yields a total sleep disturbance score and 8 subscale scores (i.e., bedtime resistance, sleep-onset delay, sleep duration, sleep anxiety, night awakenings, daytime sleepiness, parasomnias, and SDB). Parents rated the frequency of each item on a 3-point scale ranging from “usually” (5–7 times per week) to “rarely” (0 or 1 time per week). Higher scores indicate greater overall sleep disturbance and a score of >41 has been established as a cutoff indicative of clinically significant sleep disturbance (19). The CSHQ has adequate internal consistency, test–retest reliability, and validity, and has been used in previous studies of children with JIA and healthy school-age children (6–8, 19). Reliability of the total sleep disturbance scale in this sample was α = 0.91.
The WASI was used to estimate IQ (20). The WASI was administered to each child by a trained pediatric neuropsychometrician. The WASI full-scale IQ score was calculated from the scores of all 4 WASI subscales and used as the variable to estimate general intelligence.
Tests from the CANTAB were used to measure neurobehavioral performance (21). The CANTAB consists of a series of tasks that index 3 behavioral domains: 1) working memory/planning, 2) visual memory, and 3) visual attention. A series of computerized tests are presented visually on a color monitor with an attached touch screen and a press pad to record responses to stimuli (22). For each test the child was seated in a chair with a foot rest directly opposite the computer monitor. A large “X” in red tape was placed on the computer desk to designate the location of the press pad. The test battery was administered in the afternoon upon arrival to the sleep laboratory. The test battery took approximately 35–45 minutes for each child to complete. Testing took place in a separate room next to the sleep laboratory and each child's performance was monitored by one of the sleep laboratory staff. Parents were not permitted in the room during testing sessions.
CANTAB tests measuring reaction time, movement time, and sustained visual attention were used in this study. Movement was measured with the motor screening test. Reaction time was measured with 1-choice and 5-choice reaction time tests and match to sample visual search (MTS) tests. Sustained visual attention was measured with the rapid visual information processing (RVP) test.
The motor screening test was used to orient the child to the use of the touch screen and also measured movement time. A series of crosses is shown at different locations on the touch screen and the subject must touch the center of the cross on each trial. Movement time in msec was averaged for the trials for each child and reported as the mean for each group.
Reaction time was measured with 1-choice and 5-choice stimulus conditions. In these tests, a child holds a press pad down, releases the press pad when a yellow circle is seen on the screen, and must touch the yellow circle on the screen as quickly as possible. In the first stimulus condition, the child identifies a single object (i.e., a yellow circle); in the second stimulus condition, the subject identifies a yellow circle from among 5 object choices. Simple reaction time (RTI) and 5-choice reaction time (RTI-5) were calculated as the time (msec) from perception (release of press pad) to touch of the appropriate stimulus on the screen and averaged for the trials for each child. Both RTI and RTI-5 were reported as the mean (msec) for each group.
MTS is a test of reaction time that involves a visual search strategy to accurately identify a specific object. In this test there is a speed versus accuracy tradeoff and the test results can be used as an indicator of impulsivity in reaction to the stimuli. An abstract pattern within a red square is displayed in the middle of the computer screen. After a brief delay, a varied number of similar patterns are shown in boxes surrounding the red square in the middle of the screen. The child must determine which of these patterns matches the one in the middle of the screen and touch the appropriate box. The number of patterns displayed around the red square varies from 1, 2, 4, or 8 with each stimulus presentation. MTS percent correct was calculated based on the number of correctly identified targets of a possible 48 total presented. The increase in time necessary to correctly identify a target presented from among 2 choices versus 8 choices was recorded in msec and reported as reaction time (MTS latency change 2–8). MTS latency change 2–8 was averaged for each child and reported as the mean for each group.
RVP measures sustained visual attention. A white square appears in the middle of the computer screen and digits from 2 to 9 are singularly presented in a pseudorandom order at a rate of 100 digits per minute. A child must detect any of 3 possible target sequences (i.e., 2-4-6, 4-6-8, or 3-5-7) and push the press pad when the third number in the target sequence is presented. After a practice session, target sequences are presented 32 times. Variables of interest were calculated: probability of a hit (the proportion of correct responses given when a target sequence is presented) and probability of a false alarm (the proportion of responses when no target sequence is presented). A probability of a hit value close to 1.0 means the child made nearly 100% correct responses. A probability of a miss value close to 1.0 means the child made close to 100% inappropriate responses or false alarms. RVP probability of a hit and RVP probability of a false alarm were averaged for trials for each child and reported as the mean for each group.
The school competence subscale of the CBCL was used as a subjective measure of neurobehavioral performance. This 113-item survey is a parental assessment of behavioral and emotional problems as well as school functioning and performance. The school competence subscale includes items assessing performance in academic subjects, grade retention, special services, and school problems. The raw score is converted to age- and sex-adjusted T scores based on published norms. The reliability and validity of the CBCL is well established in school-age children with and without chronic conditions (23, 24).
Data were analyzed using SPSS for Windows, version 15.0. Data analyses were blocked into conceptual categories and then analyzed (e.g., demographics, sleep disturbances, CANTAB neurocognitive tests scores, and school competence for group differences). Each category was considered a separate analysis with significance set at P values less than 0.05 (2-sided). The first set of analyses addressed group differences in all study variables between children with JIA and control children and an independent t-test or chi-square test was used. Second, we compared the proportion of children above to those below the clinical cutoff score on the CHSQ (raw score of >41) for overall sleep disturbance. Third, we examined the relationships among sleep variables, CANTAB test variables, and school competence with a series of bivariate correlations controlling for age, IQ, and medications. Medications children received anytime during the study were scored as “yes” or “no” and classified into categories: 1) nonsteroidal antiinflammatory drugs, 2) corticosteroids, 3) disease-modifying antirheumatic drugs (methotrexate, leflunomide), 4) tumor necrosis factor α receptor inhibitors (etanercept, adalimumab, infliximab), 5) other (i.e., vitamins), and 6) none. Finally, a linear regression analysis was performed to identify potential predictors of neurocognitive performance. We explored how much of the variance in neurocognitive performance test scores was explained by overall sleep disturbances, controlling for age, IQ, medications, and group.
The clinical characteristics of the children are shown in Table 1. The mean ± SD age for the entire sample was 8.5 ± 1.9 years. Compared to the control children, children with JIA were taking more medications (χ2 = 27.2, P < 0.001) and had a lower IQ (t = 2.6, P < 0.009). There were more girls in the sample because the disease is more prevalent in girls compared to boys.
|JIA (n = 70)||Controls (n = 46)||95% CI|
|Age, mean ± SD years||8.5 ± 1.9||8.5 ± 1.8||−0.74, 0.68|
|WASI, mean ± SD score|
|Verbal||105 ± 11||109 ± 11||−0.33, 8.7|
|Performance†||106 ± 14||113 ± 15||2.0, 13.2|
|Total score†||106 ± 12||112 ± 13||1.6, 11.4|
|Ethnicity, no. (%)|
|White||59 (86.8)||30 (62.5)|
|Asian||2 (2.9)||10 (20.8)|
|Other||7 (10.3)||8 (17.4)|
|Sex, no. (%)|
|Girls||53 (76)||30 (65)|
|Boys||17 (24.2)||16 (35)|
|Medications, no. (%)|
|NSAIDs†||35 (50)||2 (4.3)|
|Corticosteroids†||19 (27.1)||0 (0)|
|DMARDs†||37 (52.9)||0 (0)|
|TNFα inhibitors†||12 (17.1)||0 (0)|
|Other (herbal supplement)†||26 (37.1)||5 (11)|
|None||8 (11.6)||39 (85)|
|Children's Behavior Checklist: social competence, mean ± SD score‡||46.3 ± 7.5||49.3 ± 6.4||0.27, 5.7|
The average parent-reported bedtime for the sample was 9:04 PM and the average wake time was 7:14 AM on school days. Table 2 shows parent report of average weekday bedtime and wake time, and sleep problems between children with JIA and control children. Significant group differences were found between children with JIA and control children for 6 of 8 subscales reflecting specific sleep problems and the overall sleep disturbance score (Table 2). Of the children (n = 70) with a mean overall sleep disturbance score above the CSHQ cutoff score, 71% (n = 50) had JIA.
|JIA group (n = 68), mean ± SD||Control group (n = 46), mean ± SD||P|
|CSHQ sleep times|
|Weekday bedtime, hours||9:03 PM ± 0:43||9:04 PM ± 0:43||0.92|
|Weekday wake time, hours||7:16 AM ± 0:48||7:11 AM ± 0:34||0.58|
|Sleep duration, hours||9:27 ± 0:54||9:36 ± 0:52||0.37|
|Bedtime resistance||7.8 ± 2.2||7.5 ± 1.8||0.57|
|Sleep-onset delay†||1.6 ± 0.7||1.2 ± 0.4||0.001|
|Sleep duration||4.0 ± 1.4||3.7 ± 1.1||0.17|
|Sleep anxiety||5.8 ± 2.0||5.0 ± 1.4||0.02|
|Night awakenings||4.0 ± 1.3||3.3 ± 0.7||0.001|
|Parasomnias||9.3 ± 1.8||7.9 ± 1.1||0.001|
|Sleep-disordered breathing||2.3 ± 0.6||2.1 ± 0.4||0.03|
|Daytime sleepiness||13.0 ± 3.5||11.0 ± 3.3||0.004|
|Total sleep disturbance score||45.0 ± 7.3||39.1 ± 4.9||0.001|
Scores on neurobehavioral performance tests are shown in Table 3. There were no group differences on neurocognitive performance tests between children with JIA and control children. However, in the total sample, significant differences were found for RTI (t = −2.2, P < 0.03) and RTI-5 (t = −2.3, P < 0.02) in children who had an overall sleep disturbance score above (RTI: mean ± SD 419 ± 152 msec, RTI-5: mean ± SD 460 ± 177 msec) compared to those below the CHSQ cutoff score (RTI: mean ± SD 370 ± 76 msec, RTI-5: mean ± SD 402 ± 77 msec).
|CANTAB variables||JIA group (n = 68), mean ± SD||Control group (n = 46), mean ± SD||P|
|MOT movement time, msec||899.2 ± 165.4||861.2 ± 209.8||0.28|
|RTI simple reaction time, msec||404.1 ± 143||393.1 ± 109||0.66|
|RTI 5-choice reaction time, msec||446.7 ± 168.0||423.7 ± 118.1||0.43|
|MTS, % correct||96.1 ± 3.8||95.5 ± 5.2||0.48|
|MTS latency change (2–8), msec||2,492.4 ± 1,723.2||2,783.7 ± 1,484||0.35|
|RVP probability of hits (range 1–10)||0.31 ± 0.21||0.35 ± 0.23||0.38|
|RVP probability of false alarms (range 1–10)||0.05 ± 0.11||0.04 ± 0.06||0.38|
Parents of children with JIA reported a lower mean CBCL school competence score (t = 2.2, P < 0.03) compared to parents of the healthy controls. No differences were found for CBCL school competence between children who scored below (<41) and above the CHSQ cutoff score for overall sleep disturbance.
We conducted a series of partial correlations on the entire sample between sleep variables, CANTAB test scores, and the CBCL school competence subscale score, controlling for age, IQ, and medications. Overall sleep disturbance scores were positively and significantly correlated with longer RTI and RTI-5 (r = 0.24, P = 0.03 and r = 0.37, P < 0.001, respectively). RTI-5 was inversely correlated with school competence (r = −0.26, P = 0.02), but there was no significant correlation between mean overall sleep disturbance score and mean school competence.
Based on these empirical findings, we regressed overall sleep disturbance score on measures of RTI-5 using age, IQ, medications, and group as control variables. Age, IQ, and medication were entered in the first step, study group (JIA or control) was entered in the second step, and school competence and overall sleep disturbance score were entered in the third step (Table 4). Age, IQ, medications, study group, school competence, and overall sleep disturbance accounted for 33% of the variance in reaction time (F[2,91] = 4.90, P < 0.001). Age and overall sleep disturbance were individual significant predictors accounting for 7.3% of the variance in RTI-5 (Table 4).
|Unstandardized β||SE β||Standardized β|
|RTI 5-choice reaction time|
|Total sleep disturbances||5.04||2.33||0.22|
To our knowledge, the findings from this study are the first report of an assessment of parental-reported sleep disturbances in combination with validated tests of neurobehavioral performance in school-age children with JIA compared to age- and sex-matched control children. Children with JIA had significantly higher overall sleep disturbance score and higher scores on sleep-onset delay, sleep anxiety, night awakenings, parasomnias, SDB, and daytime sleepiness subscales compared to healthy children. These findings are similar to those of previous studies in JIA (6–8). Further, 70% of the children who scored above the cutoff score for clinically significant sleep disturbance had JIA, which implies a high prevalence of disturbed sleep in children with JIA.
Children with JIA may have significant daytime sleepiness. In this study, children had self-reported daytime sleepiness similar to previous studies in JIA (7, 8), but reported higher scores compared to previous studies of children with SDB (19). We have found significant physiologic evidence for daytime sleepiness in JIA based on multiple sleep latency tests. In a previous study, 19% of the sample fell asleep in <10 minutes and another 23% fell asleep between 11 and 15 minutes (2). Daytime sleepiness could be an underrecognized consequence of sleep disturbances in JIA and is often overlooked in clinical care. In children, sleepiness manifests as inattention and hyperactivity (25–27), symptoms that are similar to those of ADHD. Some children may be misdiagnosed with ADHD when in fact the underlying problem is daytime sleepiness secondary to unrecognized sleep disorder. Given that the etiology of daytime sleepiness in JIA is not well understood, further research is warranted both on subjective and objective measures of daytime sleepiness in JIA and in comparison to age- and sex-matched control children.
Disturbed sleep has been shown to negatively affect neurobehavioral performance in children (28–30). Compared to control children, we anticipated that children with JIA might show neurobehavioral performance deficits because of prior reports of disturbed sleep, but this was not observed. Children with JIA and control children did not differ on any of the neurobehavioral performance tests. Further, sleep disturbance predicted slower reaction time after controlling for age, IQ, medications, and study group. Alternatively, our findings may be attributed to parental perception of sleep. Compared to parents of children with JIA, parents of control children may not perceive the sleep of their children to be “poor” or “disturbed,” but in actuality, their children may not be getting adequate sleep. Likewise, parents of children with JIA may perceive the sleep of their children to be “poor” or “disturbed,” but in actuality, they may have adequate sleep with little disturbance.
However, children with JIA are at risk for sleep disturbance and sleep disorders that could adversely affect neurobehavioral performance. In the present study, 70% of the children with high CSHQ sleep disturbance scores were in the JIA group. In a recent study from our laboratory, the apnea/hypopnea index, a critical indicator of SDB, and wake bouts, a measure of sleep fragmentation, were inversely associated with measures of attention and reaction time (2). Similar indices of SDB (e.g., snoring, apneas, hypopneas, hypoxia, and respiratory-related arousals) have been associated with sleep fragmentation, daytime sleepiness, and neurobehavioral deficits (e.g., poor school performance, inattention, and behavior problems such as hyperactivity) (28). In addition, children with JIA may have temporomandibular joint involvement (31–33), which leads to reduced mandibular growth and is a risk factor for SDB. Further research is needed to ascertain the prevalence of SDB and its effects on neurobehavioral and school performance in JIA.
Total IQ scores for children with JIA and healthy children were within the normal range, but significant group differences were found for performance IQ and total IQ. Compared to IQ scores from a study of German children with systemic JIA, in the present study the average performance IQ and total IQ scores were slightly higher in children with JIA (17). This finding may be attributed to the small number of children with systemic JIA in the current study. The IQ scores for our children with JIA were above the population average of 100 (34) and similar to previous studies of IQ in children with and without mild SDB (35, 36). IQ tests reflect a composite score of a child's performance across a variety of specific tasks, but they fail to address memory, an important domain of cognitive functioning (34). Additional studies should include teacher and parent reports of school performance in conjunction with objective measures of neurocognitive performance and IQ.
In conclusion, there is a paucity of data on sleep disturbances, daytime sleepiness, and neurocognitive performance in children with JIA. Most of the research on disturbed sleep and neurobehavioral performance has been conducted in healthy school-age children or children with SDB. Given observations in this study and our previous findings (2, 37), additional research is needed to ascertain the prevalence of sleep disorders and the impact on neurobehavioral and school performance in JIA in comparison to age-matched healthy controls. Untreated sleep disorders and sleep disturbances place children at considerable risk for adverse daytime dysfunction and poor health outcomes. Behavioral problems (hyperactivity, impulsivity), memory problems, poor school performance, daytime sleepiness, increased school absenteeism, slower growth, and lower quality of life have been linked to sleep disorders and disturbances (38–41). Despite these adverse effects, sleep disorders continue to be overlooked in clinical care (42), especially in children living with a chronic condition such as JIA. Adequate and good-quality sleep is essential for health and normal growth and development in school-age children. Sleep disorders and daytime sleepiness in JIA may play a major role in cognitive and physical development and manifestations of disease-related symptoms. Children with JIA may be vulnerable, not only to delays in growth and development, but also to cognitive and social delays, and may be at greater risk for developing long-term functional and neurobehavioral complications.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Landis 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 conception and design. Ward, Metz, Archbold, Lentz, Wallace, Landis.
Acquisition of data. Ward, Metz, Archbold, Lentz, Wallace, Landis.
Analysis and interpretation of data. Ward, Ringold, Metz, Lentz, Wallace, Landis.
The authors thank the children and families who helped with this research. We thank Linda Peterson, Research Coordinator, and the staff in the rheumatology clinic for recruiting the participants. We thank Hieke Nuhsbaum for data entry and Salimah Man, Yuen Song, Tuyet Nguyen, Sarah Shapiro, and Whitney Jewell for helping with data collection and processing.