- Email: [email protected]
Attention Problems of Very Preterm Children Compared with Age-Matched Term Controls at School-Age
Attention Problems of Very Preterm Children Compared with Age-Matched Term Controls at School-Age Jorrit F. de Kieviet, MSc1, Ruurd M. van Elburg, MD, PhD2,3, Harrie N. Lafeber, MD, PhD2, and Jaap Oosterlaan, PhD1 Objectives To clarify the severity, specificity, and neurocognitive underpinnings of attention problems in very preterm children.
Study design A sample of 66 preterm (<32 weeks gestation), mean (SD) age 7.5 (0.4) years, and 66 age-matched term controls participated. Symptoms of inattention were assessed using parent and teacher-rated questionnaires, and neurocognitive measures included speed and consistency in speed of information processing, lapses of attention (tau), alerting, orienting, and executive attention, as well as verbal and visuospatial working memory. Group differences were investigated using ANOVA, and Sobel tests were used to clarify the mediating role of neurocognitive impairments on attention problems. Results There was a large decrease in visuospatial working memory abilities (P < .001, d = .87), and medium increases in tau (P = .002, d = 0.55) as well as parent and teacher ratings of inattention (range d = 0.40-0.56) in very preterm children compared with term peers. Tau and visuospatial working memory were significant predictors of parent (R2 = .161, P < .001 and R2 = .071, P = .001; respectively) and teacher (R2 = .152, P < .001 and R2 = .064, P = .002; respectively) ratings of inattention, and completely explained the effects of very preterm birth on attention problems. Conclusions Increased lapses of attention and poorer visuospatial working memory fully account for the attention problems in very premature children at school-age. (J Pediatr 2012;161:824-9).
I
mproved perinatal care has increased survival rates of very preterm (<32 weeks of gestation) infants. However, very preterm birth has negative consequences on normal maturation processes in the brain,1 which consequently result in neurocognitive and behavioral problems in very preterm children in childhood.2-4 At school-age, attention problems are among the most prominent behavioral problems reported in very preterm children.5,6 These attention problems may have a detrimental impact on school performance and social function in very preterm children, however, the exact nature is still poorly understood. Attention is a multifaceted construct and involves a range of neurocognitive abilities. Besides maintaining consistent fast and accurate levels of information processing,7 brain networks of orienting, alerting, and executive attention,8 as well as working memory abilities are involved.9 Brain networks of orienting, alerting, and executive attention are crucial in (1) the ability to orient to; (2) the ability to maintain focus; and (3) ability to shift attention between events, respectively, and have been extensively studied in neurobiologic research.8 Furthermore, proper working memory abilities are essential in guiding attention by holding in mind relevant information for the task at hand.9,10 Importantly, impairments in all these neurocognitive abilities involved in attention have been demonstrated to play a pivotal role in attention problems of children diagnosed with attention deficit/hyperactivity disorder (ADHD).10,11 However, despite the potential for new treatment opportunities,12-14 the roles of these neurocognitive functions in attention problems of very preterm children have not been investigated. To elucidate the neurocognitive underpinnings of attention problems in very preterm children, we studied: (1) the severity and specificity of attention problems in very preterm children using both parent and teacher ratings of inattention at school age; (2) the differences between very preterm children and term peers on neurocognitive abilities involved in attention; and (3) the possibility that these neurocognitive functions mediate the relationship between very preterm birth and attention problems at school age.
Methods A sample of 66 very preterm (<32 weeks gestation) children and 66 age-matched term controls participated. All 66 very preterm children had participated in
ADHD ANT CBCL IQ PDBD SES TDBD TRF
Attention deficit/hyperactivity disorder Attention Network Test Child Behavior Checklist Intelligence quotient Parent Disruptive Behavior Disorders rating scale Social economic status Teacher Disruptive Behavior Disorders rating scale Teacher Report Form
From the 1Department of Clinical Neuropsychology and 2 Department of Paediatrics, VU University Amsterdam, Amsterdam, The Netherlands; and 3Danone Research Centre for Specialised Nutrition, Wageningen, The Netherlands Supported by an unrestricted research grant from Danone Friederichsdorf, Germany, which had no involvement in (1) study design; (2) the collection, analysis, and interpretation of data; (3) the writing of the report; and (4) the decision to submit the paper for publication. R.v.E. is an employee of Danone Research, Centre for Specialised Nutrition. The other authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2012 Mosby Inc. All rights reserved. 10.1016/j.jpeds.2012.05.010
824
Vol. 161, No. 5 November 2012 a randomized controlled trial on enteral glutamine supplementation (0.3 g/kg/d) between day 3 and 30 of life.15 For this study, all very preterm children admitted between September 2001 and July 2003 to the level III neonatal intensive care unit of the VU University Medical Center Amsterdam were eligible for inclusion, and baseline characteristics of the sample have been previously reported.15 Of the 102 infants included in the original study, 89 infants were alive at 1-year follow-up, and 74 were participating in the 6-year follow-up. At 7-8 years of age, parents of all 74 children were contacted and invited to participate in the current study, of which 68 (92%) agreed, and 66 (89%) successfully completed all neurocognitive tasks at the mean (SD) age of 7.5 (0.4) years (Table I). The remaining 2 children were unsuccessful in completing the neurocognitive test battery. There were no differences between the glutamine and the placebo group on any of the dependent variables (all P > .05; all d < 0.49). Age-matched, term born peers from the same classrooms as attended by the very preterm children were invited to participate in the study. Controls were required to be born >37 weeks of gestation without any perinatal complications as reported by their parents, and to attend regular classes and to be free of behavioral and academic difficulties as reported by their teacher. Using this method of recruitment, age-matched controls were recruited for 41 very preterm children. Of the remaining 25 very preterm children, 4 attended special classes and 21 schools were not willing to participate in the recruitment of age-matched controls. Additional age-matched controls (n = 25) were recruited by contacting other schools located in the same area as schools attended by the very preterm children. Social economic status (SES) was determined by classifying the highest level of education in a household on a 4-point scale, with higher scores indicating higher levels of education and corresponding higher SES. All parents completed written informed consent prior to the study, which was approved by the medical ethical committee of the VU Medical Center. Neurocognitive assessments were at the VU University Amsterdam by qualified and
trained testers using standardized instructions. Both parents and each child’s teacher were asked to complete questionnaires addressing behavioral problems including attention problems at home and at school, respectively. Parent ratings of inattention at home were derived from the Attention Problems scale of the Child Behavior Checklist (CBCL)16 and the Inattention scale of the Parent Disruptive Behavior Disorders rating scale (PDBD).17 Teachers completed the teacher equivalents of the parent scales, including the Teacher Report Form (TRF)18 and Teacher Disruptive Behavior Disorders rating scale (TDBD).17 Hyperactivity/impulsivity was measured using the Hyperactivity/Impulsivity scale of the parent-rated PDBD and teacher-rated TDBD. In addition, internalizing behavior problems and externalizing behavior problems were measured using the subscales of Internalizing Behavior and Externalizing Behavior from both the parent-rated CBCL and teacher-rated TRF. All 4 questionnaires are widely used and have excellent psychometric properties.16-18 Scores on the CBCL and TRF were standardized using T-scores with a mean (SD) of 50 (10), and scores on the PDBD and TDBD were standardized with a mean (SD) of 10 (3), with higher scores indicating higher ratings of behavioral problems. Intelligence quotient (IQ) was measured by a short-form of the Wechsler Intelligence Scale for Children-III,19 including the subtests Vocabulary and Block Design. Both subtests correlate strongly (r > .90) with Full Scale IQ. Orienting, alerting, and executive attention were assessed using an adapted version of the Attention Network Test (ANT)8 suitable for the use with young children. In this task, children had to respond as fast as possible to the appearance of a target on the left side or the right side of the screen by pressing a button corresponding to the location at which the target appeared. There were 4 types of trials. Neutral trials contained a neutral cue in the middle of the screen that preceded the target. Orienting trials contained a directional cue in the middle of the screen pointing to the position of the target, which subsequently followed. Executive trials contained directional cues incongruent with the position of the
Table I. Characteristics of very preterm children and term controls Age at follow-up, y (mean SD) SES (mean SD) Sex, male (%) WISC-III full scale IQ (mean SD) Birth weight, g (mean SD) Gestational age, wk (mean SD) Small for gestational age (<10th percentile), n (%) CRIB score (mean SD) Apgar score after 5 min <6, n (%) IVH, n (%) $ 1 infection with positive blood culture, n (%) Sepsis, n (%) BPD, n (%) Cesarean delivery, n (%) Antenatal corticosteroid, n (%)
Preterm (N = 66)
Controls (N = 66)
P value
Effect size (d)
7.5 (0.4) 3.2 (0.7) 33 (50.0) 96.6 (17.6) 1241 (355) 29.3 (1.6) 18 (27.3) 3.0 (2.8) 5 (7.6) 14 (21.2) 41 (62.1) 42 (63.6) 19 (28.8) 37 (56.1) 53 (80.3)
7.6 (0.5) 3.2 (0.8) 28 (42.4) 105.8 (14.4)
.41 .89 .39 .001
0.22 0.01 0.92
BPD, bronchopulmonary dysplasia; CRIB, Clinical Risk Index for Babies; IVH, intraventricular hemorrhage; WISC-III, Wechsler Intelligence Scale for Children-III. Bold number pertains to significant P value (P < .05).
825
THE JOURNAL OF PEDIATRICS
www.jpeds.com
target. Alerting trials contained no cue at all and the target was presented instantaneously. The 4 trial types were randomly presented in 4 blocks of 48 trials. Orienting attention was measured by comparing both speed and accuracy (percentage of errors) of the orienting trials with the neutral trials. Executive attention was measured by comparing both speed and accuracy between the executive trials and neutral trials. Alerting attention was measured by comparing both speed and accuracy between the alerting trials and neutral trials. Speed and accuracy measures of orienting, alerting, and executive attention were included as dependent variables in analysis. Speed and consistency in speed of information processing were assessed using the individual response time distributions derived from all correct ANT trials. An ex-Gaussian distribution model, which was fitted to the individual response time distributions,20 combines a normal distributional shape with an exponential component on the right side of the distribution. This model allows for an accurate determination of average processing speed corrected for extremely slow responses (mu), fluctuations in processing speed (sigma), and the proportion of extremely slow responses assessing lapses of attention (tau) reflected in the exponential component of the reaction time distribution, which has proved its clinical relevance in recent studies.7,21 To minimize the possibility that extremely slow responses pertain to responses resulting from exogenously caused distractions, trials deviating more than three SDs from the mean of the child’s reaction time distribution were removed from the data. Mu, sigma, and tau were calculated in MATLAB (MathWorks, Natick, Massachusetts),20 and included as dependent variables in analysis. Two aspects of working memory were assessed. First, verbal working memory abilities were measured using the Digit Span subtest of the Wechsler Intelligence Scale for Children-III.19 In this task, children had to verbally reproduce dictated series of digits in a forward and a backward condition. Second, visuospatial working memory abilities were measured using an adapted version of a task developed by Nutley et al.22 In this task, children had to reproduce sequences of circles appearing in a 4 4 grid on a touch screen in a forward and a backward condition. Difficulty level was increased during the course of the task by increasing span and by manipulating the position of the stimuli. Two trials were administrated for each difficulty level, and the test was terminated when the child failed to accomplish both trials at a certain difficulty level. Given that the backward condition involves more attentional resources in terms of both holding the information in memory as well as reordering the information, maximal difficulty level of the backward condition for each working memory task were included as dependent variables in analysis. Statistical Analyses All analyses were performed using PASW Statistics 17.0 (SPSS Inc, Chicago, Illinois) on data normalized using Van der Waerden transformation. Missing values on the questionnaires (7%) were replaced by means of expectation maximization.23 Independent t tests were conducted to test group differences on IQ and parent and teacher ratings of 826
Vol. 161, No. 5 inattention, hyperactivity/impulsivity, internalizing, and externalizing behavioral problems. Group differences on the neurocognitive measures were tested using multivariate analysis of variance with group as between-subject factor. If a significant multivariate effect of group emerged, post-hoc ANOVA were conducted to investigate the nature of the multivariate effect. To test whether task manipulation of the ANT was successful, a repeated measure analysis was conducted with condition as within subject factor and group as between-subject factor. To enhance reliability of the behavioral measures of inattention, principal component analysis was used to aggregate the 2 parent ratings of inattention and the 2 teacher ratings of inattention. Neurocognitive measures that significantly discriminated between very preterm children and term peers were entered in a stepwise linear regression analysis to test their predictive value for the aggregated parent and teacher ratings of inattention. Sobel-techniques24 were used to test if predictors significantly mediated the direct pathway of very preterm birth on the aggregated parent and teacher ratings of inattention (Figure; available at www.jpeds.com). Group differences were quantified in terms of effect-sizes (Cohen d)25 with values of 0.20, 0.50, and 0.80 referring to small, medium, and large effects, respectively. For all analyses, a was set at .05.
Results Age, SES, and sex were not different between very preterm children and their term peers (Table I). No differences were present for age (P = .87), SES (P = .25), and sex (P = .23) between both recruiting methods of age-matched peers. Furthermore, no significant effects of sex and being small for gestational age were found on any of the included neurocognitive measures in very preterm children (all Ps > .05, all ds < 0.33). As expected, very preterm children had significantly lower full-scale IQ scores than term peers. Compared with term peers, very preterm children had higher parent and teacher ratings of inattention as assessed with the Attention Problems scale of the CBCL, Inattention scale of the PDBD, Attention Problems scale of the TRF, and Inattention scale of the TDBD (Table II). There were no group differences in terms of hyperactivity/impulsivity, internalizing behavior problems, or externalizing behavior problems, as measured by the CBCL, PDBD, TRF and TDBD, indicating that very preterm birth is specifically related to attention problems. Differences of orienting, alerting, and executive attention between very preterm children and term peers for the dependent measures derived from the ANT were not significant (Table III). However, the inbuilt task manipulations were successful, as shown by a significant condition effect of orienting, alerting, and executive attention on the dependent measures (F [4778] = 107.84, P < .001) using a repeated measure analysis. Very preterm children had significantly lower speed and consistency in speed of information processing compared de Kieviet et al
ORIGINAL ARTICLES
November 2012
Table II. Behavioral parent and teacher ratings Parent rated questionnaires CBCL Internalizing Behavior scale (mean SD)† CBCL Externalizing Behavior scale (mean SD)† CBCL Attention Problems scale (mean SD)† PDBD Inattention scale (mean SD)z PDBD Hyperactivity/Impulsivity scale (mean SD)z Teacher rated questionnaires TRF Internalizing Behavior scale (mean SD)† TRF Externalizing Behavior scale (mean SD)† TRF Attention Problems scale (mean SD)† TDBD Inattention scale (mean SD)z TDBD Hyperactivity/Impulsivity scale mean SD)z
Very Preterm (N = 66)
Controls (N = 66)
P value*
Effect-size d*
50.0 (8.3) 48.8 (9.0) 55.8 (12.6) 11.8 (3.6) 10.9 (1.4)
52.6 (13.2) 47.7 (8.5) 49.2 (9.7) 11.1 (1.3) 10.8 (1.3)
.53 .42 .004 .02 .58
0.11 0.14 0.51 0.40 0.10
48.0 (9.3) 48.5 (10.3) 53.8 (9.4) 12.0 (2.2) 11.3 (1.7)
47.6 (9.5) 48.6 (9.2) 49.4 (5.9) 11.3 (2.9) 11.1 (2.3)
.36 .50 .01 .002 .07
0.16 0.12 0.44 0.56 0.37
Bold numbers pertains to significant P value (P < .05). *Statistics were performed on raw data. †T-scores. zScaled scores (M = 10, SD = 3).
with term peers. Post-hoc ANOVAs showed that group differences in the reaction time distribution were fully accounted for by a greater proportion of extremely slow responses (tau, F [1130] = 10.06, P = .002; d = 0.55). In contrast, average processing speed (mu, F [1130] = 0.28, P = .60; d = 0.09) and fluctuations in processing speed (sigma, F [1130] = 0.04, P = .84; d = 0.04) were not different between both groups. A multivariate ANOVA on the dependent variables derived from the verbal and visuospatial working memory tasks, showed a significant group difference between very preterm children and term peers (Table III). Post-hoc ANOVAs showed that very preterm children had poorer visuospatial working memory abilities, but their verbal working memory abilities were not different from controls. Aggregated parent and teacher ratings of inattention were derived using principal component analysis with varimax rotation, explaining 83% and 93% of the variance in the scores on the Attention Problems scale of the CBCL/TRF and the Inattention scale of the PDBD/TDBD, respectively. A stepwise linear regression indicated that the proportion of extremely slow responses and visuospatial working memory abilities both significantly predicted parent (R2 = .161, P < .001 and R2 = .071, P = .001, respectively) and teacher (R2 = .152,
P < .001 and R2 = .064, P = .002, respectively) ratings of inattention. Sobel’s tests of mediation showed that the direct pathway of very preterm birth on parent and teacher ratings of inattention became nonsignificant after controlling for the effects of the indirect pathway (all P > .08; Table IV), indicating that the increase in attention problems of very preterm children compared with term peers was completely mediated by tau and visuospatial working memory abilities.
Discussion Our study confirms that very preterm birth is associated with a medium sized increase in attention problems as reported by both parents and teacher (range d = 0.40-0.56). Furthermore, we found multiple deficits in the neurocognitive functions involved in attention, including a greater proportion of extremely slow responses (tau) and poorer visuospatial working memory abilities for very preterm children compared with age-matched term peers. Importantly, our findings clearly indicate that behavioral symptoms of inattention in very preterm children are completely explained (mediated)26 by these extremely slow responses and deficits in visuospatial working memory.
Table III. Neurocognitive performance of very preterm children and term controls Very preterm (N = 66) Attention networks Orienting attention, ms (mean SD) Alerting attention, ms (mean SD) Executive attention, ms (mean SD) % of errors orientation (mean SD) % of errors alertness (mean SD) % of errors interference (mean SD) Speed and consistency in speed of information processing Mu, ms (mean SD) Sigma, ms (mean SD) Tau, ms (mean SD) Working memory Verbal working memory (mean SD) Visuospatial working memory (mean SD)
Controls (N = 66)
141 (85) 81 (72) 213 (220) 7.7 (8.8) 8.3 (9.8) 17.8 (17.4)
144 (65) 100 (48) 163 (152) 5.2 (2.3) 5.8 (2.7) 10.5 (8.5)
433 (80) 144 (44) 198 (124)
422 (66) 142 (39) 133 (75)
3.7 (1.4) 4.2 (0.9)
4.0 (1.1) 4.9 (0.8)
F
P value
1.42
.21
Effect-size d 0.02 0.35 0.17 0.26 0.17 0.50
3.48 0.28 0.04 10.06 12.42 2.01 24.98
.02 .60 .84 .002 <.001 .16 <.001
0.09 0.04 0.55 0.25 0.87
ms, milliseconds. Results in bold pertain to the multivariate ANOVA findings.
Attention Problems of Very Preterm Children Compared with Age-Matched Term Controls at School-Age
827
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 161, No. 5
Table IV. Results of mediation analysis for behavioral parent and teacher ratings of inattention of very preterm children Path A a (Sa) Parent Visuospatial working 0.73 (0.15) memory Tau 65.01 (17.82) Teacher Visuospatial working 0.73 (0.15) memory Tau 65.01 (17.82)
Path B P
b (Sb)
<.001
0.43 (0.09)
Path C
P
c (Sc)
P
<.001 0.49 (0.17) .004
3.45
<.001
0.22 (0.18)
.22
<.001 0.004 (0.001) <.001 0.49 (0.17) .004
2.70
.007
0.25 (0.16)
.14
<.001 0.48 (0.16) .003
3.41
<.001
0.22 (0.17)
.22
<.001 0.003 (0.001) <.001 0.48 (0.16) .003
2.32
.02
0.29 (0.17)
.08
0.43 (0.09)
c (Sc)
P
Path C (adjusted for indirect path)
Z
<.001
P
Test of mediation
Path A, effect very preterm birth on neurocognitive functioning; Path B, effect neurocognitive functioning on behavioral ratings of inattention; Path C, effect very preterm birth on behavioral ratings of inattention; Path C (adjusted for indirect path), effect very preterm birth on behavioral ratings of inattention, adjusted for indirect path via path A and path B; a, b, and c, unstandardized path coefficients; Sa, Sb, and Sc, standard errors of the path coefficients. Bold numbers pertains to significant P value (P < .05).
In contrast to the presence of attention problems in very preterm children, parents and teacher did not report more internalizing or externalizing problem behavior compared with term controls. Furthermore, reported behavior problems are similar to the ADHD inattentive subtype,27 as no differences were present in reported hyperactivity/impulsivity between very preterm children and term controls. These findings are in line with the increasing amount of literature showing evidence for a prominent and central role of attention problems in very preterm children.2,3,5 Our findings show an increase in tau in very preterm children, whereas no differences were present in mu, illustrating that the frequently replicated finding of overall slower processing speed in very preterm children28-30 can be traced back to an increase in the proportion of extremely slow responses. In other words, most very preterm children are capable of making fast and adequate responses, but fail to make them continuously, ending up with a slower average response time compared with term peers. These so-called “lapses of attention” translate into short moments of attention loss, which may interfere with daily life both at home and at school in particular in situations that demand continuous concentration for a longer period of time. Furthermore, we confirm the presence of poorer visuospatial working memory skills in very preterm children,28,30,31 showing that these children have more difficulty maintaining visual information in mind for the task at hand than term peers. Previous studies found that slow processing speed and poorer working memory are essential factors involved in poorer executive functioning and poorer academic attainment of very preterm children at school age.30,32,33 These relations may be further unraveled as our results suggest a pathway from lapses of attention and poorer visuospatial working memory abilities to behavioral symptoms of inattention in very preterm children. These attention problems may subsequently be related with their poorer academic attainment.34 For both lapses of attention as well as visuospatial working memory, similar outcomes have been described in ADHD.7,10,21,35 Multiple interventions have been developed to specifically address and improve the underlying neurocognitive dysfunctioning, including neurofeedback13 and working memory training,12 and both interventions effectively reduce behavior problems in children with ADHD. Interest828
ingly, lapses of attention are related to alterations in the default attention network activation and fluctuations of the brain’s arousal system in ADHD, which are involved in maintaining concentration above a certain minimum threshold level.36 Furthermore, dysregulated dopaminergic and noradrenergic arousal systems may modulate visuospatial working memory processes by altering frontostriatal pathways in the brain of ADHD children.37,38 In very preterm children, the mediating role of lapses of attention and visuospatial working memory in behavioral symptoms of inattention, as well as the specific vulnerability of the striatum in hypoxic-ischemic events,39 may reflect a crucial involvement of the underlying brain’s arousal system. As these findings indicate a window of opportunity for interventions, future studies investigating (medical) treatment and involvement of the arousal system in attention problems of very preterm children are warranted. This study has some limitations that need to be taken into account. First, differences in IQ capacities exist between very preterm children and their term peers, which may in part explain differences on the neurocognitive tasks. However, we did not find any evidence for an impact of IQ on differences in neurocognitive task performance in the current study (data available from the first author). Second, group sizes were relatively small in this study. Nevertheless, power was sufficient to detect small to medium-sized differences between very preterm children and term peers, as calculated using G-power software.40 Third, we only included term peers free of behavioral and academic difficulties, whereas a nonselected group of peers might have led to different results. This study showed that very preterm birth is specifically associated with an increase in attention problems as compared with term peers at school age. Furthermore, we clarified the nature of attention problems in very preterm children, showing that lapses of attention and visuospatial working memory abilities fully account for these attention problems. Our findings in very preterm children are similar to those obtained in children with ADHD, suggesting that treatments effective in ADHD might also effectively reduce attention problems in very preterm children. n Submitted for publication Feb 2, 2012; last revision received Apr 5, 2012; accepted May 4, 2012.
de Kieviet et al
ORIGINAL ARTICLES
November 2012 Reprint requests: Jorrit F. de Kieviet, MSc, Department of Clinical Neuropsychology, VU University Amsterdam, Van der Boechorststraat 1, 1081 BT Amsterdam, The Netherlands. E-mail: [email protected]
References 1. Volpe JJ. Brain injury in premature infants: a complex amalgam of destructive and developmental disturbances. Lancet Neurol 2009;1:110-24. 2. Aarnoudse-Moens CS, Weisglas-Kuperus N, van Goudoever JB, Oosterlaan J. Meta-analysis of neurobehavioral outcomes in very preterm and/or very low birth weight children. Pediatrics 2009;2:717-28. 3. Bhutta AT, Cleves MA, Casey PH, Cradock MM, Anand KJ. Cognitive and behavioral outcomes of school-aged children who were born preterm: a meta-analysis. JAMA 2002;6:728-37. 4. de Kieviet JF, Piek JP, Aarnoudse-Moens CS, Oosterlaan J. Motor development in very preterm and very low-birth-weight children from birth to adolescence: a meta-analysis. JAMA 2009;20:2235-42. 5. van de Weijer-Bergsma E, Wijnroks L, Jongmans MJ. Attention development in infants and preschool children born preterm: a review. Infant Behav Dev 2008;3:333-51. 6. Mulder H, Pitchford NJ, Hagger MS, Marlow N. Development of executive function and attention in preterm children: a systematic review. Dev Neuropsychol 2009;4:393-421. 7. Leth-Steensen C, Elbaz ZK, Douglas VI. Mean response times, variability, and skew in the responding of ADHD children: a response time distributional approach. Acta Psychol (Amst) 2000;2:167-90. 8. Fan J, McCandliss BD, Fossella J, Flombaum JI, Posner MI. The activation of attentional networks. Neuroimage 2005;2:471-9. 9. Kane MJ, Bleckley MK, Conway AR, Engle RW. A controlled-attention view of working-memory capacity. J Exp Psychol Gen 2001;2:169-83. 10. Martinussen R, Hayden J, Hogg-Johnson S, Tannock R. A meta-analysis of working memory impairments in children with attention-deficit/ hyperactivity disorder. J Am Acad Child Adolesc Psychiatry 2005;4:377-84. 11. Konrad K, Neufang S, Hanisch C, Fink GR, Herpertz-Dahlmann B. Dysfunctional attentional networks in children with attention deficit/hyperactivity disorder: evidence from an event-related functional magnetic resonance imaging study. Biol Psychiatry 2006;7:643-51. 12. Klingberg T, Fernell E, Olesen PJ, Johnson M, Gustafsson P, Dahlstrom K, et al. Computerized training of working memory in children with ADHD–a randomized, controlled trial. J Am Acad Child Adolesc Psychiatry 2005;2:177-86. 13. Arns M, de RS, Strehl U, Breteler M, Coenen A. Efficacy of neurofeedback treatment in ADHD: the effects on inattention, impulsivity and hyperactivity: a meta-analysis. Clin EEG Neurosci 2009;3:180-9. 14. van der Oord S, Prins PJ, Oosterlaan J, Emmelkamp PM. Efficacy of methylphenidate, psychosocial treatments and their combination in school-aged children with ADHD: a meta-analysis. Clin Psychol Rev 2008;5:783-800. 15. van den Berg A, van Elburg RM, Westerbeek EA, Twisk JW, Fetter WP. Glutamine-enriched enteral nutrition in very-low-birth-weight infants and effects on feeding tolerance and infectious morbidity: a randomized controlled trial. Am J Clin Nutr 2005;6:1397-404. 16. Verhulst F, van der Ende J, Koot H. Handleiding voor de CBCL/4-18. Rotterdam, The Netherlands: Erasmus MC-Sophia Kinderziekenhuis; 1996. 17. Oosterlaan J, Scheres A, Antrop I, Roeyers H, Sergeant J. Vragenlijst voor Gedragsproblemen bij Kinderen (VvGK). Nederlandse bewerking van de Disruptive Behavior Disorders Rating Scale [Dutch translation of the Disruptive Behavior Disorders Rating Scale]. Lisse, The Netherlands: Swets & Zeitlinger; 2000. 18. Verhulst F, van der Ende J, Koot H. Handleiding voor de Teacher’s Report Form (TRF). Rotterdam, The Netherlands: Erasmus MC-Sophia Kinderziekenhuis; 1997. 19. Wechsler D. WISC-III Handleiding. London, United Kingdom: The Psychological Corporation; 2002.
20. Lacouture Y, Cousineau D. How to use MATLAB to fit the ex-Gaussian and other probability functions to a distribution of response times. Tutorials Quant Methods Psychol 2008;1:35-45. 21. Geurts HM, Grasman RP, Verte S, Oosterlaan J, Roeyers H, van Kammen SM, et al. Intra-individual variability in ADHD, autism spectrum disorders and Tourette’s syndrome. Neuropsychologia 2008; 13:3030-41. 22. Nutley SB, Soderqvist S, Bryde S, Humphreys K, Klingberg T. Measuring working memory capacity with greater precision in the lower capacity ranges. Dev Neuropsychol 2010;1:81-95. 23. Gold MS, Bentler PM. Treatments of missing data: a Monte Carlo comparison of RBHDI, iterative stochastic regression imputation, and expectation-maximization. Structural Equation Modeling 2000; 3:319-55. 24. Sobel ME. Asymptotic confidence intervals for indirect effects in structural equation models. Washington, DC: American Sociological Association; 1982. 290-312. 25. Cohen J. Statistical power analyses for the behavioral sciences. 2nd ed. Hillsdale, NY: Erlbaum; 1988. 26. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;6:1173-82. 27. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994. 28. Rose SA, Feldman JF. Memory and processing speed in preterm children at eleven years: a comparison with full-terms. Child Dev 1996;5: 2005-21. 29. Anderson P, Doyle LW. Neurobehavioral outcomes of school-age children born extremely low birth weight or very preterm in the 1990s. JAMA 2003;24:3264-72. 30. Mulder H, Pitchford NJ, Marlow N. Processing speed and working memory underlie academic attainment in very preterm children. Arch Dis Child Fetal Neonatal Ed 2010;4:F267-72. 31. Luciana M, Lindeke L, Georgieff M, Mills M, Nelson CA. Neurobehavioral evidence for working-memory deficits in school-aged children with histories of prematurity. Dev Med Child Neurol 1999;8:521-33. 32. Aarnoudse-Moens CS, Smidts DP, Oosterlaan J, Duivenvoorden HJ, Weisglas-Kuperus N. Executive function in very preterm children at early school age. J Abnorm Child Psychol 2009;7:981-93. 33. Mulder H, Pitchford NJ, Marlow N. Processing speed mediates executive function difficulties in very preterm children in middle childhood. J Int Neuropsychol Soc 2011;1-10. 34. Aarnoudse-Moens CS, Oosterlaan J, Duivenvoorden HJ, van Goudoever JB, Weisglas-Kuperus N. Development of preschool and academic skills in children born very preterm. J Pediatr 2011;1:51-6. 35. Buzy WM, Medoff DR, Schweitzer JB. Intra-individual variability among children with ADHD on a working memory task: an ex-Gaussian approach. Child Neuropsychol 2009;5:441-59. 36. Fassbender C, Zhang H, Buzy WM, Cortes CR, Mizuiri D, Beckett L, et al. A lack of default network suppression is linked to increased distractibility in ADHD. Brain Res 2009;114-28. 37. Arnsten AF. Modulation of prefrontal cortical-striatal circuits: relevance to therapeutic treatments for Tourette syndrome and attention-deficit hyperactivity disorder. Adv Neurol 2001;333-41. 38. Del CN, Chamberlain SR, Sahakian BJ, Robbins TW. The roles of dopamine and noradrenaline in the pathophysiology and treatment of attention-deficit/hyperactivity disorder. Biol Psychiatry 2011;12: e145-57. 39. Lou HC. Etiology and pathogenesis of attention-deficit hyperactivity disorder (ADHD): significance of prematurity and perinatal hypoxichaemodynamic encephalopathy. Acta Paediatr 1996;11:1266-71. 40. Faul F, Erdfelder E, Lang AG, Buchner A. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007;2:175-91.
Attention Problems of Very Preterm Children Compared with Age-Matched Term Controls at School-Age
829
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 161, No. 5
Figure. Model of direct and indirect effects of very preterm birth on attention problems. Besides the direct effect of very preterm birth on attention problems (Path C), an indirect pathway from very preterm birth to attention problems via potential neurocognitive functions involved in attention may be present (Path A and Path B combined).
829.e1
de Kieviet et al
Study design A sample of 66 preterm (<32 weeks gestation), mean (SD) age 7.5 (0.4) years, and 66 age-matched term controls participated. Symptoms of inattention were assessed using parent and teacher-rated questionnaires, and neurocognitive measures included speed and consistency in speed of information processing, lapses of attention (tau), alerting, orienting, and executive attention, as well as verbal and visuospatial working memory. Group differences were investigated using ANOVA, and Sobel tests were used to clarify the mediating role of neurocognitive impairments on attention problems. Results There was a large decrease in visuospatial working memory abilities (P < .001, d = .87), and medium increases in tau (P = .002, d = 0.55) as well as parent and teacher ratings of inattention (range d = 0.40-0.56) in very preterm children compared with term peers. Tau and visuospatial working memory were significant predictors of parent (R2 = .161, P < .001 and R2 = .071, P = .001; respectively) and teacher (R2 = .152, P < .001 and R2 = .064, P = .002; respectively) ratings of inattention, and completely explained the effects of very preterm birth on attention problems. Conclusions Increased lapses of attention and poorer visuospatial working memory fully account for the attention problems in very premature children at school-age. (J Pediatr 2012;161:824-9).
I
mproved perinatal care has increased survival rates of very preterm (<32 weeks of gestation) infants. However, very preterm birth has negative consequences on normal maturation processes in the brain,1 which consequently result in neurocognitive and behavioral problems in very preterm children in childhood.2-4 At school-age, attention problems are among the most prominent behavioral problems reported in very preterm children.5,6 These attention problems may have a detrimental impact on school performance and social function in very preterm children, however, the exact nature is still poorly understood. Attention is a multifaceted construct and involves a range of neurocognitive abilities. Besides maintaining consistent fast and accurate levels of information processing,7 brain networks of orienting, alerting, and executive attention,8 as well as working memory abilities are involved.9 Brain networks of orienting, alerting, and executive attention are crucial in (1) the ability to orient to; (2) the ability to maintain focus; and (3) ability to shift attention between events, respectively, and have been extensively studied in neurobiologic research.8 Furthermore, proper working memory abilities are essential in guiding attention by holding in mind relevant information for the task at hand.9,10 Importantly, impairments in all these neurocognitive abilities involved in attention have been demonstrated to play a pivotal role in attention problems of children diagnosed with attention deficit/hyperactivity disorder (ADHD).10,11 However, despite the potential for new treatment opportunities,12-14 the roles of these neurocognitive functions in attention problems of very preterm children have not been investigated. To elucidate the neurocognitive underpinnings of attention problems in very preterm children, we studied: (1) the severity and specificity of attention problems in very preterm children using both parent and teacher ratings of inattention at school age; (2) the differences between very preterm children and term peers on neurocognitive abilities involved in attention; and (3) the possibility that these neurocognitive functions mediate the relationship between very preterm birth and attention problems at school age.
Methods A sample of 66 very preterm (<32 weeks gestation) children and 66 age-matched term controls participated. All 66 very preterm children had participated in
ADHD ANT CBCL IQ PDBD SES TDBD TRF
Attention deficit/hyperactivity disorder Attention Network Test Child Behavior Checklist Intelligence quotient Parent Disruptive Behavior Disorders rating scale Social economic status Teacher Disruptive Behavior Disorders rating scale Teacher Report Form
From the 1Department of Clinical Neuropsychology and 2 Department of Paediatrics, VU University Amsterdam, Amsterdam, The Netherlands; and 3Danone Research Centre for Specialised Nutrition, Wageningen, The Netherlands Supported by an unrestricted research grant from Danone Friederichsdorf, Germany, which had no involvement in (1) study design; (2) the collection, analysis, and interpretation of data; (3) the writing of the report; and (4) the decision to submit the paper for publication. R.v.E. is an employee of Danone Research, Centre for Specialised Nutrition. The other authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2012 Mosby Inc. All rights reserved. 10.1016/j.jpeds.2012.05.010
824
Vol. 161, No. 5 November 2012 a randomized controlled trial on enteral glutamine supplementation (0.3 g/kg/d) between day 3 and 30 of life.15 For this study, all very preterm children admitted between September 2001 and July 2003 to the level III neonatal intensive care unit of the VU University Medical Center Amsterdam were eligible for inclusion, and baseline characteristics of the sample have been previously reported.15 Of the 102 infants included in the original study, 89 infants were alive at 1-year follow-up, and 74 were participating in the 6-year follow-up. At 7-8 years of age, parents of all 74 children were contacted and invited to participate in the current study, of which 68 (92%) agreed, and 66 (89%) successfully completed all neurocognitive tasks at the mean (SD) age of 7.5 (0.4) years (Table I). The remaining 2 children were unsuccessful in completing the neurocognitive test battery. There were no differences between the glutamine and the placebo group on any of the dependent variables (all P > .05; all d < 0.49). Age-matched, term born peers from the same classrooms as attended by the very preterm children were invited to participate in the study. Controls were required to be born >37 weeks of gestation without any perinatal complications as reported by their parents, and to attend regular classes and to be free of behavioral and academic difficulties as reported by their teacher. Using this method of recruitment, age-matched controls were recruited for 41 very preterm children. Of the remaining 25 very preterm children, 4 attended special classes and 21 schools were not willing to participate in the recruitment of age-matched controls. Additional age-matched controls (n = 25) were recruited by contacting other schools located in the same area as schools attended by the very preterm children. Social economic status (SES) was determined by classifying the highest level of education in a household on a 4-point scale, with higher scores indicating higher levels of education and corresponding higher SES. All parents completed written informed consent prior to the study, which was approved by the medical ethical committee of the VU Medical Center. Neurocognitive assessments were at the VU University Amsterdam by qualified and
trained testers using standardized instructions. Both parents and each child’s teacher were asked to complete questionnaires addressing behavioral problems including attention problems at home and at school, respectively. Parent ratings of inattention at home were derived from the Attention Problems scale of the Child Behavior Checklist (CBCL)16 and the Inattention scale of the Parent Disruptive Behavior Disorders rating scale (PDBD).17 Teachers completed the teacher equivalents of the parent scales, including the Teacher Report Form (TRF)18 and Teacher Disruptive Behavior Disorders rating scale (TDBD).17 Hyperactivity/impulsivity was measured using the Hyperactivity/Impulsivity scale of the parent-rated PDBD and teacher-rated TDBD. In addition, internalizing behavior problems and externalizing behavior problems were measured using the subscales of Internalizing Behavior and Externalizing Behavior from both the parent-rated CBCL and teacher-rated TRF. All 4 questionnaires are widely used and have excellent psychometric properties.16-18 Scores on the CBCL and TRF were standardized using T-scores with a mean (SD) of 50 (10), and scores on the PDBD and TDBD were standardized with a mean (SD) of 10 (3), with higher scores indicating higher ratings of behavioral problems. Intelligence quotient (IQ) was measured by a short-form of the Wechsler Intelligence Scale for Children-III,19 including the subtests Vocabulary and Block Design. Both subtests correlate strongly (r > .90) with Full Scale IQ. Orienting, alerting, and executive attention were assessed using an adapted version of the Attention Network Test (ANT)8 suitable for the use with young children. In this task, children had to respond as fast as possible to the appearance of a target on the left side or the right side of the screen by pressing a button corresponding to the location at which the target appeared. There were 4 types of trials. Neutral trials contained a neutral cue in the middle of the screen that preceded the target. Orienting trials contained a directional cue in the middle of the screen pointing to the position of the target, which subsequently followed. Executive trials contained directional cues incongruent with the position of the
Table I. Characteristics of very preterm children and term controls Age at follow-up, y (mean SD) SES (mean SD) Sex, male (%) WISC-III full scale IQ (mean SD) Birth weight, g (mean SD) Gestational age, wk (mean SD) Small for gestational age (<10th percentile), n (%) CRIB score (mean SD) Apgar score after 5 min <6, n (%) IVH, n (%) $ 1 infection with positive blood culture, n (%) Sepsis, n (%) BPD, n (%) Cesarean delivery, n (%) Antenatal corticosteroid, n (%)
Preterm (N = 66)
Controls (N = 66)
P value
Effect size (d)
7.5 (0.4) 3.2 (0.7) 33 (50.0) 96.6 (17.6) 1241 (355) 29.3 (1.6) 18 (27.3) 3.0 (2.8) 5 (7.6) 14 (21.2) 41 (62.1) 42 (63.6) 19 (28.8) 37 (56.1) 53 (80.3)
7.6 (0.5) 3.2 (0.8) 28 (42.4) 105.8 (14.4)
.41 .89 .39 .001
0.22 0.01 0.92
BPD, bronchopulmonary dysplasia; CRIB, Clinical Risk Index for Babies; IVH, intraventricular hemorrhage; WISC-III, Wechsler Intelligence Scale for Children-III. Bold number pertains to significant P value (P < .05).
825
THE JOURNAL OF PEDIATRICS
www.jpeds.com
target. Alerting trials contained no cue at all and the target was presented instantaneously. The 4 trial types were randomly presented in 4 blocks of 48 trials. Orienting attention was measured by comparing both speed and accuracy (percentage of errors) of the orienting trials with the neutral trials. Executive attention was measured by comparing both speed and accuracy between the executive trials and neutral trials. Alerting attention was measured by comparing both speed and accuracy between the alerting trials and neutral trials. Speed and accuracy measures of orienting, alerting, and executive attention were included as dependent variables in analysis. Speed and consistency in speed of information processing were assessed using the individual response time distributions derived from all correct ANT trials. An ex-Gaussian distribution model, which was fitted to the individual response time distributions,20 combines a normal distributional shape with an exponential component on the right side of the distribution. This model allows for an accurate determination of average processing speed corrected for extremely slow responses (mu), fluctuations in processing speed (sigma), and the proportion of extremely slow responses assessing lapses of attention (tau) reflected in the exponential component of the reaction time distribution, which has proved its clinical relevance in recent studies.7,21 To minimize the possibility that extremely slow responses pertain to responses resulting from exogenously caused distractions, trials deviating more than three SDs from the mean of the child’s reaction time distribution were removed from the data. Mu, sigma, and tau were calculated in MATLAB (MathWorks, Natick, Massachusetts),20 and included as dependent variables in analysis. Two aspects of working memory were assessed. First, verbal working memory abilities were measured using the Digit Span subtest of the Wechsler Intelligence Scale for Children-III.19 In this task, children had to verbally reproduce dictated series of digits in a forward and a backward condition. Second, visuospatial working memory abilities were measured using an adapted version of a task developed by Nutley et al.22 In this task, children had to reproduce sequences of circles appearing in a 4 4 grid on a touch screen in a forward and a backward condition. Difficulty level was increased during the course of the task by increasing span and by manipulating the position of the stimuli. Two trials were administrated for each difficulty level, and the test was terminated when the child failed to accomplish both trials at a certain difficulty level. Given that the backward condition involves more attentional resources in terms of both holding the information in memory as well as reordering the information, maximal difficulty level of the backward condition for each working memory task were included as dependent variables in analysis. Statistical Analyses All analyses were performed using PASW Statistics 17.0 (SPSS Inc, Chicago, Illinois) on data normalized using Van der Waerden transformation. Missing values on the questionnaires (7%) were replaced by means of expectation maximization.23 Independent t tests were conducted to test group differences on IQ and parent and teacher ratings of 826
Vol. 161, No. 5 inattention, hyperactivity/impulsivity, internalizing, and externalizing behavioral problems. Group differences on the neurocognitive measures were tested using multivariate analysis of variance with group as between-subject factor. If a significant multivariate effect of group emerged, post-hoc ANOVA were conducted to investigate the nature of the multivariate effect. To test whether task manipulation of the ANT was successful, a repeated measure analysis was conducted with condition as within subject factor and group as between-subject factor. To enhance reliability of the behavioral measures of inattention, principal component analysis was used to aggregate the 2 parent ratings of inattention and the 2 teacher ratings of inattention. Neurocognitive measures that significantly discriminated between very preterm children and term peers were entered in a stepwise linear regression analysis to test their predictive value for the aggregated parent and teacher ratings of inattention. Sobel-techniques24 were used to test if predictors significantly mediated the direct pathway of very preterm birth on the aggregated parent and teacher ratings of inattention (Figure; available at www.jpeds.com). Group differences were quantified in terms of effect-sizes (Cohen d)25 with values of 0.20, 0.50, and 0.80 referring to small, medium, and large effects, respectively. For all analyses, a was set at .05.
Results Age, SES, and sex were not different between very preterm children and their term peers (Table I). No differences were present for age (P = .87), SES (P = .25), and sex (P = .23) between both recruiting methods of age-matched peers. Furthermore, no significant effects of sex and being small for gestational age were found on any of the included neurocognitive measures in very preterm children (all Ps > .05, all ds < 0.33). As expected, very preterm children had significantly lower full-scale IQ scores than term peers. Compared with term peers, very preterm children had higher parent and teacher ratings of inattention as assessed with the Attention Problems scale of the CBCL, Inattention scale of the PDBD, Attention Problems scale of the TRF, and Inattention scale of the TDBD (Table II). There were no group differences in terms of hyperactivity/impulsivity, internalizing behavior problems, or externalizing behavior problems, as measured by the CBCL, PDBD, TRF and TDBD, indicating that very preterm birth is specifically related to attention problems. Differences of orienting, alerting, and executive attention between very preterm children and term peers for the dependent measures derived from the ANT were not significant (Table III). However, the inbuilt task manipulations were successful, as shown by a significant condition effect of orienting, alerting, and executive attention on the dependent measures (F [4778] = 107.84, P < .001) using a repeated measure analysis. Very preterm children had significantly lower speed and consistency in speed of information processing compared de Kieviet et al
ORIGINAL ARTICLES
November 2012
Table II. Behavioral parent and teacher ratings Parent rated questionnaires CBCL Internalizing Behavior scale (mean SD)† CBCL Externalizing Behavior scale (mean SD)† CBCL Attention Problems scale (mean SD)† PDBD Inattention scale (mean SD)z PDBD Hyperactivity/Impulsivity scale (mean SD)z Teacher rated questionnaires TRF Internalizing Behavior scale (mean SD)† TRF Externalizing Behavior scale (mean SD)† TRF Attention Problems scale (mean SD)† TDBD Inattention scale (mean SD)z TDBD Hyperactivity/Impulsivity scale mean SD)z
Very Preterm (N = 66)
Controls (N = 66)
P value*
Effect-size d*
50.0 (8.3) 48.8 (9.0) 55.8 (12.6) 11.8 (3.6) 10.9 (1.4)
52.6 (13.2) 47.7 (8.5) 49.2 (9.7) 11.1 (1.3) 10.8 (1.3)
.53 .42 .004 .02 .58
0.11 0.14 0.51 0.40 0.10
48.0 (9.3) 48.5 (10.3) 53.8 (9.4) 12.0 (2.2) 11.3 (1.7)
47.6 (9.5) 48.6 (9.2) 49.4 (5.9) 11.3 (2.9) 11.1 (2.3)
.36 .50 .01 .002 .07
0.16 0.12 0.44 0.56 0.37
Bold numbers pertains to significant P value (P < .05). *Statistics were performed on raw data. †T-scores. zScaled scores (M = 10, SD = 3).
with term peers. Post-hoc ANOVAs showed that group differences in the reaction time distribution were fully accounted for by a greater proportion of extremely slow responses (tau, F [1130] = 10.06, P = .002; d = 0.55). In contrast, average processing speed (mu, F [1130] = 0.28, P = .60; d = 0.09) and fluctuations in processing speed (sigma, F [1130] = 0.04, P = .84; d = 0.04) were not different between both groups. A multivariate ANOVA on the dependent variables derived from the verbal and visuospatial working memory tasks, showed a significant group difference between very preterm children and term peers (Table III). Post-hoc ANOVAs showed that very preterm children had poorer visuospatial working memory abilities, but their verbal working memory abilities were not different from controls. Aggregated parent and teacher ratings of inattention were derived using principal component analysis with varimax rotation, explaining 83% and 93% of the variance in the scores on the Attention Problems scale of the CBCL/TRF and the Inattention scale of the PDBD/TDBD, respectively. A stepwise linear regression indicated that the proportion of extremely slow responses and visuospatial working memory abilities both significantly predicted parent (R2 = .161, P < .001 and R2 = .071, P = .001, respectively) and teacher (R2 = .152,
P < .001 and R2 = .064, P = .002, respectively) ratings of inattention. Sobel’s tests of mediation showed that the direct pathway of very preterm birth on parent and teacher ratings of inattention became nonsignificant after controlling for the effects of the indirect pathway (all P > .08; Table IV), indicating that the increase in attention problems of very preterm children compared with term peers was completely mediated by tau and visuospatial working memory abilities.
Discussion Our study confirms that very preterm birth is associated with a medium sized increase in attention problems as reported by both parents and teacher (range d = 0.40-0.56). Furthermore, we found multiple deficits in the neurocognitive functions involved in attention, including a greater proportion of extremely slow responses (tau) and poorer visuospatial working memory abilities for very preterm children compared with age-matched term peers. Importantly, our findings clearly indicate that behavioral symptoms of inattention in very preterm children are completely explained (mediated)26 by these extremely slow responses and deficits in visuospatial working memory.
Table III. Neurocognitive performance of very preterm children and term controls Very preterm (N = 66) Attention networks Orienting attention, ms (mean SD) Alerting attention, ms (mean SD) Executive attention, ms (mean SD) % of errors orientation (mean SD) % of errors alertness (mean SD) % of errors interference (mean SD) Speed and consistency in speed of information processing Mu, ms (mean SD) Sigma, ms (mean SD) Tau, ms (mean SD) Working memory Verbal working memory (mean SD) Visuospatial working memory (mean SD)
Controls (N = 66)
141 (85) 81 (72) 213 (220) 7.7 (8.8) 8.3 (9.8) 17.8 (17.4)
144 (65) 100 (48) 163 (152) 5.2 (2.3) 5.8 (2.7) 10.5 (8.5)
433 (80) 144 (44) 198 (124)
422 (66) 142 (39) 133 (75)
3.7 (1.4) 4.2 (0.9)
4.0 (1.1) 4.9 (0.8)
F
P value
1.42
.21
Effect-size d 0.02 0.35 0.17 0.26 0.17 0.50
3.48 0.28 0.04 10.06 12.42 2.01 24.98
.02 .60 .84 .002 <.001 .16 <.001
0.09 0.04 0.55 0.25 0.87
ms, milliseconds. Results in bold pertain to the multivariate ANOVA findings.
Attention Problems of Very Preterm Children Compared with Age-Matched Term Controls at School-Age
827
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 161, No. 5
Table IV. Results of mediation analysis for behavioral parent and teacher ratings of inattention of very preterm children Path A a (Sa) Parent Visuospatial working 0.73 (0.15) memory Tau 65.01 (17.82) Teacher Visuospatial working 0.73 (0.15) memory Tau 65.01 (17.82)
Path B P
b (Sb)
<.001
0.43 (0.09)
Path C
P
c (Sc)
P
<.001 0.49 (0.17) .004
3.45
<.001
0.22 (0.18)
.22
<.001 0.004 (0.001) <.001 0.49 (0.17) .004
2.70
.007
0.25 (0.16)
.14
<.001 0.48 (0.16) .003
3.41
<.001
0.22 (0.17)
.22
<.001 0.003 (0.001) <.001 0.48 (0.16) .003
2.32
.02
0.29 (0.17)
.08
0.43 (0.09)
c (Sc)
P
Path C (adjusted for indirect path)
Z
<.001
P
Test of mediation
Path A, effect very preterm birth on neurocognitive functioning; Path B, effect neurocognitive functioning on behavioral ratings of inattention; Path C, effect very preterm birth on behavioral ratings of inattention; Path C (adjusted for indirect path), effect very preterm birth on behavioral ratings of inattention, adjusted for indirect path via path A and path B; a, b, and c, unstandardized path coefficients; Sa, Sb, and Sc, standard errors of the path coefficients. Bold numbers pertains to significant P value (P < .05).
In contrast to the presence of attention problems in very preterm children, parents and teacher did not report more internalizing or externalizing problem behavior compared with term controls. Furthermore, reported behavior problems are similar to the ADHD inattentive subtype,27 as no differences were present in reported hyperactivity/impulsivity between very preterm children and term controls. These findings are in line with the increasing amount of literature showing evidence for a prominent and central role of attention problems in very preterm children.2,3,5 Our findings show an increase in tau in very preterm children, whereas no differences were present in mu, illustrating that the frequently replicated finding of overall slower processing speed in very preterm children28-30 can be traced back to an increase in the proportion of extremely slow responses. In other words, most very preterm children are capable of making fast and adequate responses, but fail to make them continuously, ending up with a slower average response time compared with term peers. These so-called “lapses of attention” translate into short moments of attention loss, which may interfere with daily life both at home and at school in particular in situations that demand continuous concentration for a longer period of time. Furthermore, we confirm the presence of poorer visuospatial working memory skills in very preterm children,28,30,31 showing that these children have more difficulty maintaining visual information in mind for the task at hand than term peers. Previous studies found that slow processing speed and poorer working memory are essential factors involved in poorer executive functioning and poorer academic attainment of very preterm children at school age.30,32,33 These relations may be further unraveled as our results suggest a pathway from lapses of attention and poorer visuospatial working memory abilities to behavioral symptoms of inattention in very preterm children. These attention problems may subsequently be related with their poorer academic attainment.34 For both lapses of attention as well as visuospatial working memory, similar outcomes have been described in ADHD.7,10,21,35 Multiple interventions have been developed to specifically address and improve the underlying neurocognitive dysfunctioning, including neurofeedback13 and working memory training,12 and both interventions effectively reduce behavior problems in children with ADHD. Interest828
ingly, lapses of attention are related to alterations in the default attention network activation and fluctuations of the brain’s arousal system in ADHD, which are involved in maintaining concentration above a certain minimum threshold level.36 Furthermore, dysregulated dopaminergic and noradrenergic arousal systems may modulate visuospatial working memory processes by altering frontostriatal pathways in the brain of ADHD children.37,38 In very preterm children, the mediating role of lapses of attention and visuospatial working memory in behavioral symptoms of inattention, as well as the specific vulnerability of the striatum in hypoxic-ischemic events,39 may reflect a crucial involvement of the underlying brain’s arousal system. As these findings indicate a window of opportunity for interventions, future studies investigating (medical) treatment and involvement of the arousal system in attention problems of very preterm children are warranted. This study has some limitations that need to be taken into account. First, differences in IQ capacities exist between very preterm children and their term peers, which may in part explain differences on the neurocognitive tasks. However, we did not find any evidence for an impact of IQ on differences in neurocognitive task performance in the current study (data available from the first author). Second, group sizes were relatively small in this study. Nevertheless, power was sufficient to detect small to medium-sized differences between very preterm children and term peers, as calculated using G-power software.40 Third, we only included term peers free of behavioral and academic difficulties, whereas a nonselected group of peers might have led to different results. This study showed that very preterm birth is specifically associated with an increase in attention problems as compared with term peers at school age. Furthermore, we clarified the nature of attention problems in very preterm children, showing that lapses of attention and visuospatial working memory abilities fully account for these attention problems. Our findings in very preterm children are similar to those obtained in children with ADHD, suggesting that treatments effective in ADHD might also effectively reduce attention problems in very preterm children. n Submitted for publication Feb 2, 2012; last revision received Apr 5, 2012; accepted May 4, 2012.
de Kieviet et al
ORIGINAL ARTICLES
November 2012 Reprint requests: Jorrit F. de Kieviet, MSc, Department of Clinical Neuropsychology, VU University Amsterdam, Van der Boechorststraat 1, 1081 BT Amsterdam, The Netherlands. E-mail: [email protected]
References 1. Volpe JJ. Brain injury in premature infants: a complex amalgam of destructive and developmental disturbances. Lancet Neurol 2009;1:110-24. 2. Aarnoudse-Moens CS, Weisglas-Kuperus N, van Goudoever JB, Oosterlaan J. Meta-analysis of neurobehavioral outcomes in very preterm and/or very low birth weight children. Pediatrics 2009;2:717-28. 3. Bhutta AT, Cleves MA, Casey PH, Cradock MM, Anand KJ. Cognitive and behavioral outcomes of school-aged children who were born preterm: a meta-analysis. JAMA 2002;6:728-37. 4. de Kieviet JF, Piek JP, Aarnoudse-Moens CS, Oosterlaan J. Motor development in very preterm and very low-birth-weight children from birth to adolescence: a meta-analysis. JAMA 2009;20:2235-42. 5. van de Weijer-Bergsma E, Wijnroks L, Jongmans MJ. Attention development in infants and preschool children born preterm: a review. Infant Behav Dev 2008;3:333-51. 6. Mulder H, Pitchford NJ, Hagger MS, Marlow N. Development of executive function and attention in preterm children: a systematic review. Dev Neuropsychol 2009;4:393-421. 7. Leth-Steensen C, Elbaz ZK, Douglas VI. Mean response times, variability, and skew in the responding of ADHD children: a response time distributional approach. Acta Psychol (Amst) 2000;2:167-90. 8. Fan J, McCandliss BD, Fossella J, Flombaum JI, Posner MI. The activation of attentional networks. Neuroimage 2005;2:471-9. 9. Kane MJ, Bleckley MK, Conway AR, Engle RW. A controlled-attention view of working-memory capacity. J Exp Psychol Gen 2001;2:169-83. 10. Martinussen R, Hayden J, Hogg-Johnson S, Tannock R. A meta-analysis of working memory impairments in children with attention-deficit/ hyperactivity disorder. J Am Acad Child Adolesc Psychiatry 2005;4:377-84. 11. Konrad K, Neufang S, Hanisch C, Fink GR, Herpertz-Dahlmann B. Dysfunctional attentional networks in children with attention deficit/hyperactivity disorder: evidence from an event-related functional magnetic resonance imaging study. Biol Psychiatry 2006;7:643-51. 12. Klingberg T, Fernell E, Olesen PJ, Johnson M, Gustafsson P, Dahlstrom K, et al. Computerized training of working memory in children with ADHD–a randomized, controlled trial. J Am Acad Child Adolesc Psychiatry 2005;2:177-86. 13. Arns M, de RS, Strehl U, Breteler M, Coenen A. Efficacy of neurofeedback treatment in ADHD: the effects on inattention, impulsivity and hyperactivity: a meta-analysis. Clin EEG Neurosci 2009;3:180-9. 14. van der Oord S, Prins PJ, Oosterlaan J, Emmelkamp PM. Efficacy of methylphenidate, psychosocial treatments and their combination in school-aged children with ADHD: a meta-analysis. Clin Psychol Rev 2008;5:783-800. 15. van den Berg A, van Elburg RM, Westerbeek EA, Twisk JW, Fetter WP. Glutamine-enriched enteral nutrition in very-low-birth-weight infants and effects on feeding tolerance and infectious morbidity: a randomized controlled trial. Am J Clin Nutr 2005;6:1397-404. 16. Verhulst F, van der Ende J, Koot H. Handleiding voor de CBCL/4-18. Rotterdam, The Netherlands: Erasmus MC-Sophia Kinderziekenhuis; 1996. 17. Oosterlaan J, Scheres A, Antrop I, Roeyers H, Sergeant J. Vragenlijst voor Gedragsproblemen bij Kinderen (VvGK). Nederlandse bewerking van de Disruptive Behavior Disorders Rating Scale [Dutch translation of the Disruptive Behavior Disorders Rating Scale]. Lisse, The Netherlands: Swets & Zeitlinger; 2000. 18. Verhulst F, van der Ende J, Koot H. Handleiding voor de Teacher’s Report Form (TRF). Rotterdam, The Netherlands: Erasmus MC-Sophia Kinderziekenhuis; 1997. 19. Wechsler D. WISC-III Handleiding. London, United Kingdom: The Psychological Corporation; 2002.
20. Lacouture Y, Cousineau D. How to use MATLAB to fit the ex-Gaussian and other probability functions to a distribution of response times. Tutorials Quant Methods Psychol 2008;1:35-45. 21. Geurts HM, Grasman RP, Verte S, Oosterlaan J, Roeyers H, van Kammen SM, et al. Intra-individual variability in ADHD, autism spectrum disorders and Tourette’s syndrome. Neuropsychologia 2008; 13:3030-41. 22. Nutley SB, Soderqvist S, Bryde S, Humphreys K, Klingberg T. Measuring working memory capacity with greater precision in the lower capacity ranges. Dev Neuropsychol 2010;1:81-95. 23. Gold MS, Bentler PM. Treatments of missing data: a Monte Carlo comparison of RBHDI, iterative stochastic regression imputation, and expectation-maximization. Structural Equation Modeling 2000; 3:319-55. 24. Sobel ME. Asymptotic confidence intervals for indirect effects in structural equation models. Washington, DC: American Sociological Association; 1982. 290-312. 25. Cohen J. Statistical power analyses for the behavioral sciences. 2nd ed. Hillsdale, NY: Erlbaum; 1988. 26. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;6:1173-82. 27. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994. 28. Rose SA, Feldman JF. Memory and processing speed in preterm children at eleven years: a comparison with full-terms. Child Dev 1996;5: 2005-21. 29. Anderson P, Doyle LW. Neurobehavioral outcomes of school-age children born extremely low birth weight or very preterm in the 1990s. JAMA 2003;24:3264-72. 30. Mulder H, Pitchford NJ, Marlow N. Processing speed and working memory underlie academic attainment in very preterm children. Arch Dis Child Fetal Neonatal Ed 2010;4:F267-72. 31. Luciana M, Lindeke L, Georgieff M, Mills M, Nelson CA. Neurobehavioral evidence for working-memory deficits in school-aged children with histories of prematurity. Dev Med Child Neurol 1999;8:521-33. 32. Aarnoudse-Moens CS, Smidts DP, Oosterlaan J, Duivenvoorden HJ, Weisglas-Kuperus N. Executive function in very preterm children at early school age. J Abnorm Child Psychol 2009;7:981-93. 33. Mulder H, Pitchford NJ, Marlow N. Processing speed mediates executive function difficulties in very preterm children in middle childhood. J Int Neuropsychol Soc 2011;1-10. 34. Aarnoudse-Moens CS, Oosterlaan J, Duivenvoorden HJ, van Goudoever JB, Weisglas-Kuperus N. Development of preschool and academic skills in children born very preterm. J Pediatr 2011;1:51-6. 35. Buzy WM, Medoff DR, Schweitzer JB. Intra-individual variability among children with ADHD on a working memory task: an ex-Gaussian approach. Child Neuropsychol 2009;5:441-59. 36. Fassbender C, Zhang H, Buzy WM, Cortes CR, Mizuiri D, Beckett L, et al. A lack of default network suppression is linked to increased distractibility in ADHD. Brain Res 2009;114-28. 37. Arnsten AF. Modulation of prefrontal cortical-striatal circuits: relevance to therapeutic treatments for Tourette syndrome and attention-deficit hyperactivity disorder. Adv Neurol 2001;333-41. 38. Del CN, Chamberlain SR, Sahakian BJ, Robbins TW. The roles of dopamine and noradrenaline in the pathophysiology and treatment of attention-deficit/hyperactivity disorder. Biol Psychiatry 2011;12: e145-57. 39. Lou HC. Etiology and pathogenesis of attention-deficit hyperactivity disorder (ADHD): significance of prematurity and perinatal hypoxichaemodynamic encephalopathy. Acta Paediatr 1996;11:1266-71. 40. Faul F, Erdfelder E, Lang AG, Buchner A. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007;2:175-91.
Attention Problems of Very Preterm Children Compared with Age-Matched Term Controls at School-Age
829
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 161, No. 5
Figure. Model of direct and indirect effects of very preterm birth on attention problems. Besides the direct effect of very preterm birth on attention problems (Path C), an indirect pathway from very preterm birth to attention problems via potential neurocognitive functions involved in attention may be present (Path A and Path B combined).
829.e1
de Kieviet et al