Original Article
Behavioral Outcomes at Corrected Age 2.5 Years in Children Born Extremely Preterm Johanna Månsson, MS, Karin Stjernqvist, PhD, Martin Bäckström, PhD ABSTRACT: Objective: This study examined a national cohort of 2.5-year-old children born extremely preterm with respect to behavioral problems from the perspective of parents and whether developmental variables mediated the effects of extreme prematurity on behavioral problems. Methods: As a part of the Extremely Preterm Infants in Sweden Study (EXPRESS), 344 children born before 27 weeks of gestation and 338 control children were given the Cognitive, Language, and Motor Scales of the Bayley-III and the Parent report Child Behavior Checklist/1½-5 (CBCL/1½-5). CBCL/1½-5 assigns 7 syndrome scores, further classified into composite scores for internalizing and externalizing behavior as well as total problems. Group differences in behavioral difficulties and prevalence of clinical problems were calculated. Bayley-III scores were used in regression models to determine if developmental factors mediated the effects of extreme prematurity on behavioral problems, after controlling for sociodemographic factors. Results: Preterm children had significantly higher mean T-scores on internalizing, externalizing, and total problems, as compared with the control subjects, but these were still within the average range. However, the proportion of subjects showing behavioral problems within the clinical range was significantly higher in the preterm group. Levels of cognitive, language, and motor development mediated the between-group differences in behavioral problems. Conclusion: Our findings encourage behavioral assessments during preschool years and emphasize the importance of considering multifactorial pathways of prediction when examining prematurity outcome. (J Dev Behav Pediatr 35:435–442, 2014) Index terms: Bayley-III, behavioral problems, CBCL/1½-5, extremely preterm, mediation analysis.
T
he survival of children born extremely preterm (EPT; ,28 gestational weeks) has increased during the past decades. Investigations of behavioral sequelae consistently show that children born EPT may suffer from a range of difficulties. Depressive symptoms, withdrawn behavior, poorer peer relationships, and increased risks of being diagnosed with attention deficit disorder or attention-deficit hyperactivity disorder are well-known outcomes at school age and adolescence.1–3 Sex differences regarding types of behavioral problems are often found within very preterm (VPT) and EPT populations, wherein boys are more likely to develop externalizing and hyperactivity problems and girls are more likely to develop internalizing problems.4,5 Samara et al1 also found these outcomes when comparing EPT children with same-sex full-term (FT) peers aged 6 years. During
From the Department of Psychology, Lund University, Lund, Sweden. Received January 2014; accepted June 2014. This study was supported by grants from the Swedish Research Council 20063855, 2009-4250, the Crafoord Foundation, the Linnéa and Josef Carlsson’s Foundation, and the “Nils W Svenningsens Stiftelse för Prematurforskning.” Disclosure: The authors declare no conflict of interest. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.jdbp.org). Address for reprints: Johanna Månsson, MS, Department of Psychology, Lund University, Box 213, SE-221 00 Lund, Sweden; e-mail: Johanna.Mansson@psy. lu.se. Copyright Ó 2014 Lippincott Williams & Wilkins
Vol. 35, No. 7, September 2014
preschool years, EPT children are rated by their parents as showing poorer emotional regulation and social competence as well as more somatic problems compared with children born at term. However, less is known about the extent of behavioral challenges at these younger ages.6,7 Some studies have sought to identify factors that influence the occurrence of behavioral problems in children born preterm. Perinatal factors such as prolonged need of artificial ventilation, length of stay in neonatal care units, and cerebral abnormalities have been documented as risk factors.5,6,8 Low socioeconomic status (SES), including environmental factors like limited parental education and economic disadvantages, is related to an increased risk of prematurity as well as behavior problems.9,10 Behavior difficulties are also associated with poor parental well-being.11–13 Distressing experiences during the neonatal period may affect parental perceptions of their child, which in turn influences their child rearing practices.14–16 The relationship between prematurity and behavioral outcome, and how it may be affected by developmental factors, has been explored in studies using mediation analyses. Nadeau et al17 demonstrated that IQ mediated the relationship between prematurity and hyperactivity, whereas neuromotor function mediated the relationship between prematurity and withdrawn behavior. Bayless et al18 showed that cognitive and neuromotor development predicted behavior difficulties better than www.jdbp.org | 435
prematurity per se. Both studies were performed with school-aged children born VPT or EPT. Girouard et al19 reported that language difficulties, as predicted by both neonatal factors and SES, were associated with higher rates of hyperactivity in a group of 5-year-old VPT children. Few studies so far have investigated potential developmental mediators between extreme prematurity and behavioral outcome during early preschool years. The main objective of the population-based Extremely Preterm Infants in Sweden Study (EXPRESS) is to investigate the outcome of EPT birth. The first aim of this report was to assess the extent of behavioral problems at 2.5 years of age within the EXPRESS population compared with a control group of children born FT. Behavioral difficulties during preschool years have been shown to be predictive of similar problems later in life.20,21 Early identification of difficulties is therefore important as it would improve the ability to anticipate special needs and promote early interventions. The second aim was to explore the relationships between EPT birth, cognitive, language and motor development, and behavioral problems. Understanding the predictive pathways of behavioral outcome aids identification of certain risk groups and intervention planning. We hypothesized that EPT children would show significantly more behavioral difficulties compared with FT controls and that the influence of birth status (EPT/FT) would be mediated by developmental factors.
METHODS Participants The Extremely Preterm Group The EXPRESS population consists of all infants born with a gestational age of younger than 27 weeks from April 1, 2004, to March 31, 2007, in Sweden. The incidence of extreme prematurity was 0.33%. Seventy percent (497/707) survived till 1 year of age.22 Six children died before reaching the age of 2.5 years. Thirty children were not eligible for follow-up (mother had a protected identity [n 5 3]; families had moved abroad [n 5 3]; preliminary identity number given at birth did not match [n 5 24]) leaving 461 as possible candidates for follow-up. Five families declined participation, and 57 children were lost to follow-up. Eighty-seven percent (399/461) underwent the assessments used in this study at 2.5 years of age (corrected for the extremely preterm [EPT]). The 5 children who declined participation were all born to non-Nordic mothers and did not have severe neonatal morbidities.23 No significant differences with respect to mean birth weight, gestational age, and rate of male sex were observed between the children who were assessed and those lost to follow-up. Fifty-five (13.8%) were excluded from further analyses in this study due to lack of information on the behavioral and/or developmental assessments. Hence, a total of 344 children born EPT were included in the analyses of this report. There were no differences regarding the prevalence of major neonatal morbidity or parental education level 436 Behavioral Outcomes in Children Born Extremely Preterm
when comparing the excluded and included families. The percentage of non-Nordic mothers was significantly higher in the excluded group: 24.4% versus 9.3% in the included group. The Control Group For each EPT child, a list of 10 potential control children was generated from the Swedish Medical Birth Registry. Selection criteria were singleton birth at term with 5-minute Apgar score .4, and matching for domicile, sex, day of birth, and ethnicity. If the first set of parents declined participation, the second set was approached, and so on, resulting in a total of 366 controls. Twenty-eight (7.7%) were omitted from analyses due to missing information on the behavioral and/or developmental assessments, leaving 338 controls.
Measures Child Behavior Checklist 1½ to 5 Years The Child Behavior Checklist/1½-5 (CBCL/1½-5) questionnaire was used to obtain standardized parental ratings of children’s behavioral functioning. Each item (99 problem items; 1 open-ended item for recording problems not listed) is scored on a 3-point scale (0 5 not true, 1 5 somewhat/sometimes true, 2 5 very true) based on the child’s behavior during the preceding 2 months. Results are expressed in 7 syndrome scales of co-occurring problems designated as emotionally reactive, anxious/depressed, somatic complaints, withdrawn, sleep problems, attention problems, and aggressive behavior. The sum of all problem items is the total problem score (TPS). Additionally, the CBCL/1½-5 syndrome scales can be classified into 2 broad groups of problems: internalizing problems (emotionally reactive, anxious/depressed, somatic complaints, and withdrawn syndrome scales) and externalizing problems (attention problems and aggressive behavior syndrome scales). Higher scores indicate more problematic behavior. The CBCL/1½-5 text version was available in Swedish. Non– Swedish-speaking parents were assisted in their ratings through oral translations by either the examiner or a professional interpreter. Bayley-III Development was assessed using the Bayley Scales of Infant and Toddler Development, 3rd Edition (BayleyIII).24 Bayley-III evaluates developmental functioning of children between the ages of 1 and 42 months through a series of standardized test items. Bayley-III provides 3 index scales: cognitive, language, and motor. The cognitive index scale assesses sensorimotor development, concept formation and categorization, and memory. The language index is composed of the distinct receptive communication and expressive communication subtests. Receptive communication measures verbal comprehension, the understanding of requests, and morphological markers. Expressive communication measures preverbal communication, vocabulary, and syntactic development. The motor index consists of fine motor and gross motor subtests. The fine motor measures functional hand skills Journal of Developmental & Behavioral Pediatrics
and perceptual motor integration. The gross motor examines static positioning, movement of limbs, and balance. Norm-referenced composite scores for the indices are derived from sums of the subtest scale scores (a composite score equivalent is available for the cognitive index scale) and ranges from 40 to 160 with a mean of 100 and SD of 15. Previous reports show that the Bayley-III has psychometric weaknesses that can cause inflated means.25 Using norm-referenced scores may result in underestimations of delay. However, in the EXPRESS, test scores are consistently evaluated on the basis of mean and SD values of the controls.
Procedure The Bayley-III assessments were performed by licensed psychologists. At the time of assessment, parents filled out the CBCL/1½-5 questionnaire. This procedure was administered at the 7 perinatal centers in Sweden. Because psychologists had sometimes previously met the children during their stay at the neonatal care unit, it was not possible for them to be blinded to birth status. The study was approved by the Regional Ethics Review Board at Lund University (Dnr 469/2007). All parents provided written consent for the assessment.
Data Analysis Baseline sociodemographic and neonatal characteristics of the EPTs were compared with the controls using t tests and x2 tests where appropriate. There are currently no Swedish national norm scores for the CBCL/1½-5. Raw scores were therefore converted into standardized T-scores based on control group mean and SD values. The CBCL/1½-5 manual recommends using T-scores ,65 to indicate normal range on the syndrome scales.26 Differences in mean CBCL/1½-5 scores of EPT born children compared with controls were examined using t tests. Results were confirmed by nonparametric Mann–Whitney U tests due to positively skewed distributions of the CBCL/1½-5 scales. Potential interactions of birth status and sex on behavioral scores were explored using factorial analyses of variance. Cohen’s guidelines for effect size were followed wherein a large effect is 0.8, a medium effect is 0.5, and a small effect is 0.2.27 Cutoffs for clinical behavior problems on the internalizing and externalizing composite scales and the TPS are usually set at the 90th percentile.26 In a Swedish follow-up study of 10-year-old children born EPT, CBCL scores above the 90th percentile for the same-sex control subjects were used as cutoffs for clinical range.28 The same procedure was used in this research. Chisquare tests were performed to compare differences in the prevalence of behavioral problems within clinical range between the EPT children and controls. The phi correlation coefficient (f) was calculated to determine the strength of the association between the variables birth status and behavioral problems within clinical range, that is, the effect size. Vol. 35, No. 7, September 2014
To investigate the contributions of different factors to behavioral outcomes, we used mediation analysis as described by Bayless et al18 and Loe et al.29 Mediation analysis explores the mechanism through which a predictor variable affects an outcome measure. In our study, birth status (EPT/full term) was used as an independent variable (IV); CBCL/1½-5 internalizing, externalizing, and TPS were used as dependent variables (DVs); Bayley-III cognitive, language, and motor composite scores were used as potential mediating variables (MVs). Behavioral problems are likely to be associated with social factors. Parental education is a frequently used socioeconomic status indicator in studies investigating preterm outcome and was therefore included as a control variable.9 The variable expressed the highest academic achievement between the 2 parents and was classified in 3 levels: (1) below high school, (2) high school, and (3) above high school (a 2-year university diploma and above). Maternal first-generation immigrant status was also included as a control variable (foreignborn mother). The 3 requirements for mediation analysis were tested using separate regression analyses.30 Additional hierarchical regression analyses were performed to investigate whether the MVs contribute unique variance to the DVs, above and beyond the contribution of the IV. Mediation is considered partial if the contribution of the IV is reduced but still significant, whereas it is total if the IV contribution is reduced to nonsignificance. Results were confirmed using ranked values due to nonnormal distributions of the CBCL/1½-5 scores. For all statistical analyses, SPSS for Windows Version 20.0 (IBM, Armonk, NY) was used.
RESULTS Sociodemographic and Infant Characteristics Sociodemographic and infant characteristics are shown in Table 1. Despite matching attempts, x2 analyses demonstrated that the parental education level was statistically significantly lower and more mothers were foreign born in the extremely preterm (EPT) group compared with the control group. Immigrant mothers were mainly from other Nordic, former East-European, Middle East, or North African countries (Table 1). In both groups, there was a large number of missing values in the parental education variable. This was handled by pairwise exclusions of cases in further analyses.
Behavioral Outcomes The mean age at assessment was 30.1 (corrected) months for the preterm group versus 30.6 (chronological) months for the control group. The planned age range for assessment was 30 6 3 months; 98.7% of the EPT’s versus 97.8% of the full-term (FT) children were tested within this range. Behavioral outcome in EPT children versus FT controls and effect size estimations of © 2014 Lippincott Williams & Wilkins
437
Table 1. Sociodemographic Factors, Infant Characteristics at Birth, Neonatal Morbidity, and Prevalence of Cerebral Palsy Characteristic
EPT FT (n 5 344) (n 5 338) p
Maternal age, mean (SD)
31.4 (5.8)
30.9 (4.9) .309
Parental education, n (%) Below high school
.006 19 (5.5)
7 (2.1)
High school
134 (39.0) 109 (32.2)
High school or higher
138 (40.1) 164 (48.5)
Missing information
53 (15.4)
58 (17.2) .532
61 (17.7)
24 (7.1) .000
Other Nordic country
10 (2.9)
12 (3.6)
Former East-European country
18 (5.2)
2 (0.6)
Middle East or North African country
16 (4.7)
7 (2.1)
Othera
17 (4.9)
3 (0.8)
8 (2.3)
0 (0)
Foreign-born mother, n (%)
Missing information Male sex, n (%)
188 (54.7) 186 (55)
Birth weight, mean (SD), g
780 (170) 3622 (498) .000
Gestational age, mean (SD), wk
24.9 (1.0)
39.5 (1.1) .000
SGA, n (%)b
62 (18.0)
NR
Multiple births, n (%)
55 (16)
NR
Major neonatal morbidities, n (%)c 182 (52.9)
NR
Severe BPD
80 (23.2)
Severe IVH
30 (8.7)
NR
NEC
18 (5.2)
NR
PVL Severe ROP
.921
NR
14 (4.1)
NR
115 (33.4)
NR
CP, n (%)
.000
No
323 (93.9) 338 (100)
Mild CPd
10 (2.9)
0 (0)
Moderate CPe
10 (2.9)
0 (0)
1 (0.3)
0 (0)
f
Severe CP
5 other European, North American, sub-Saharan African, Southeast Asian, South and Middle American countries, New Zealand and Australia. bSGA, birth weight more than 2 SD below the Swedish standard population mean. cMajor neonatal morbidities included severe IVH defined as $Grade 3, PVL, severe BPD defined as $30% oxygen dependency at 36 weeks, severe ROP defined as $Phase 3, and/or NEC. dMild CP defined as ambulant without aid. eModerate CP defined as ambulant with aid. fSevere CP defined as nonambulant. EPT, extremely preterm; FT, full term; BPD, bronchopulmonary dysplasia; CP, cerebral palsy; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; NR, no record; PVL, periventricular leukomalacia; ROP, retinopathy of prematurity; SGA, small for gestational age. aOther
group differences are illustrated in Table 2. Parents in the EPT group reported significantly higher scores on all the CBCL/1½-5 syndrome scales, as well as internalizing and externalizing problems and total problem score (TPS), compared with the control subjects. The EPT children had a significantly higher prevalence of scores within clinical range for internalizing and externalizing problems and TPS compared with the controls (Table 3). The analysis of variance tests revealed no significant interaction effects between birth status and sex on CBCL/ 1½-5 scores (not shown). 438 Behavioral Outcomes in Children Born Extremely Preterm
Mediation Analysis Regression analyses were performed to confirm that the requirements for mediation analysis were met and found that (1) birth status was associated with the Bayley-III composite scores, (2) birth status was associated with CBCL/1½-5 outcome, and (3) the Bayley-III composite scores were associated with CBCL/1½-5 outcome. Results of the preliminary regression analyses are shown in the supporting information (Table S1, Supplemental Digital Content 1, http://links.lww.com/JDBP/A63). Further hierarchical regression analyses were performed in 3 steps. Birth status was entered in the first, foreign-born mother and parental education in the second, and Bayley-III cognitive, language, and motor composite scores in the third step. The CBCL/1½-5 internalizing, externalizing, and TPS T-scores were entered as dependent variable (DVs). Three equations were calculated, 1 for each DV. Results are presented in Table 4. The variables foreign-born mother and parental education each separately contributed to the variance of internalizing problems, partly reducing the contribution of birth status. When Bayley-III composite scores were entered, the contribution of birth status was reduced to nonsignificance. This indicates complete mediation, after controlling for sociodemographics. Parental education significantly contributed to the externalizing score variance, whereas the variable foreign-born mother did not. The contribution of birth status was no longer statistically significant after adding Bayley-III composite scores to the equation, again demonstrating complete mediation. The total regression model only modestly contributed to the externalizing score variance. Similar to the previous equations, the effect of birth status was reduced to statistical nonsignificance when Bayley-III composite scores were entered in the final step, after controlling for the sociodemographic variables. Bayley-III composite scores were entered simultaneously, indicating that cognition alone significantly contributed to the externalizing and TPS variance, whereas cognitive and motor functions both significantly contributed to the internalizing score variance (Table 4) (An additional analysis, using the SPSS AMOS (IBM, Chicago, IL) program, was performed to confirm the results. A latent cognitive variable was measured using the Bayley-III cognitive, language, and motor scores as indicators. A latent behavior variable was measured using the CBCL/1½-5 internalizing and externalizing variables as indicators. All other variables were included in the meditation model as observed variables. Significances of the coefficients were tested with bootstrap confidence intervals. In this model, missing values were replaced with regression weighted imputation. The total effect between birth status and behavior was significant [b 5 .187]; also the indirect effect from birth status through the cognitive latent variable was significant [b 5 2.161]. The direct effect from birth status and behavior was Journal of Developmental & Behavioral Pediatrics
Table 2. Group Differences in CBCL/1½-5 Syndrome and Composite Scale Scores Between EPT Children and FT Controls EPT (n 5 344), T-score
FT (n 5 338), T-score
Mean
SD
Mean
SD
p
Effect Size Cohen’s d
Emotionally reactive
52.0
12.8
50.0
10.0
.025
0.17
Anxious/depressed
53.7
15.6
50.0
10.0
.000
0.28
Somatic complaints
56.3
14.2
50.0
10.0
.000
0.51
Withdrawn
57.1
17.5
50.0
10.0
.000
0.50
CBCL/1½-5 Syndrome scale
Sleep problems
51.6
10.3
50.0
10.0
.035
0.16
Attention problems
55.4
12.2
50.0
10.0
.000
0.48
Aggressive behavior
51.9
11.2
50.0
10.0
.018
0.18
Composite scale Internalizing
56.1
15.0
50.0
10.0
.000
0.48
Externalizing
53.1
11.4
50.0
10.0
.000
0.29
TPS
54.8
13.0
50.0
10.0
.000
0.41
EPT, extremely preterm; FT, full term; CBCL, child behavior checklist; TPS, total problem score.
nonsignificant [b 5 .026]. This clearly supported the mediation hypothesis. The sociodemographic variables were included as covariates. To reach acceptable fit, a number of correlations between birth status and sociodemographics and sociodemographics and the error term of the language variable were added. They did not influence the coefficients related to the mediation hypothesis. The final model had an excellent fit [comparative fit index 5 0.972]). We also investigated as to whether composite scores served as mediators one-by-one. Results indicate that all 3 composite scores partially or completely mediated the relationship (not shown). The contributions of the major neonatal morbidities on CBCL/1½-5 outcome were investigated by including them to the hierarchical regression models. Results are presented in the supporting information (Table S2, Supplemental Digital Content 2, http://links.lww.com/JDBP/ A64). The contribution of birth status on internalizing problems, externalizing problems, and TPS was no longer statistically significant after adding the morbidity variables. No variable by itself was contributing significantly to the hierarchical regression models, except for
Table 3. Group Differences in Prevalence of Behavior Within Clinical Range Between EPT Children and FT Controls
EPT (n 5 344) FT (n 5 338) p
Effect Size (u)
Behaviors within clinical range, n (%)a Internalizing
72 (20.9)
27 (8.0)
.000
0.18
Externalizing
67 (19.5)
31 (9.2)
.000
0.15
TPS
68 (19.8)
30 (8.9)
.000
0.16
range $90th percentile. EPT, extremely preterm; FT, full term; TPS, total problem score. aClinical
Vol. 35, No. 7, September 2014
severe intraventricular hemorrhage to the TPS variance (The regression models were further explored including (1) EPT children without major morbidities and (2) EPT children with major morbidities. Results confirmed that the Bayley-III scores mediate the effect of birth status on CBCL/1½-5 scores, in both groups. Results were less prominent in the EPT children without major morbidities).
DISCUSSION This study compared extremely preterm (EPT) children with full-term (FT) children on the basis of parental ratings of behavioral problems at 2.5 years of age. Group differences were found on all types of problems, with a main focus on the differences with the largest effect sizes, that is, the differences that are presumed to be clinically relevant. EPT children had more somatic problems than controls. This may be expected because EPT birth is associated with severe morbidity and advanced intensive care in the neonatal period, followed by physical problems like eating difficulties and pulmonary complications.3 Children born EPT also had more attentional difficulties than controls. Previous studies show that attention-deficit hyperactivity disorder and attention deficit disorder are among the most frequent psychiatric outcomes of EPT birth at school age. In their study on EPT outcome, Samara et al1 demonstrated that overactivity is explained by cognitive functioning, whereas regulation of attention is not. Hence, attention problems may be a specific feature of development after EPT birth. Johnson et al21 present the increasing indication of an “EPT phenotype,” characterized by inattention, emotional disorders, and social problems. Our results suggest that attention problems may emerge during early preschool years, which highlights the importance of early screening for attention deficit symptoms. © 2014 Lippincott Williams & Wilkins
439
Table 4. Hierarchical Multiple Regression Analyses Estimating Bayley-III Variables as Mediators of the Effect of Extreme Prematurity on Internalizing, Externalizing, and Total Behavioral Problems, Controlling for Sociodemographic Factors R2
DR2
B
b
p
Internalizing Step 1
0.054 0.054
26.01 2.232 .000
Birth status Step 2
0.165 0.111
.000 24.25 2.162 .000
Birth status Foreign-born mother
9.30
.244 .000
24.21 2.187 .000
Parental education Step 3
.000
0.234 0.069
.000 21.12 2.043 .301
Birth status Foreign-born mother
8.47
.222 .000
Parental education
23.59 2.160 .000
Bayley-III score: cognitive
20.160 2.148 .011
Bayley-III score: language
20.066 2.077 .184
Bayley-III score: motor
20.089 2.110 .031
Externalizing Step 1
0.020 0.020
23.07 2.142 .001
Birth status Step 2
0.068 0.048
.000 22.30 2.107 .010
Birth status Foreign-born mother
1.44
.046 .280
23.82 2.207 .000
Parental education Step 3
.001
0.103 0.035
.000 20.609 2.028 .531
Birth status Foreign-born mother
0.822
.026 .541
23.44 2.186 .000
Parental education Bayley-III score: cognitive
20.133 2.149 .017
Bayley-III score: language
20.035 2.050 .425
Bayley-III score: motor
20.018 2.028 .616
Total problems Step 1
0.041 0.041
Step 2
0.133 0.092
Foreign-born mother
5.82
Foreign-born mother Parental education
.169 .000
24.51 2.223 .000
Parental education Birth status
.000 23.34 2.141 .000
Birth status
Step 3
.000 24.78 2.202 .000
Birth status
0.194 0.061
.000 20.779 2.033 .439 5.05
.147 .000
23.99 2.197 .000
Bayley-III score: cognitive
20.167 2.171 .004
Bayley-III score: language
20.049 2.064 .282
Bayley-III score: motor
20.053 2.073 .163
Children born EPT were rated as more withdrawn, showing more symptoms of depression and anxiety as compared with controls. Our results add to the extant 440 Behavioral Outcomes in Children Born Extremely Preterm
findings of others showing that internalizing behaviors are more prominent in preterm children than aggressive and delinquent (i.e., externalizing) behaviors.7,29 Treyvaud et al20 demonstrated that emotional symptoms of 5-year-old children born very preterm were predicted by internalizing problems at 2 years. Internalizing behavior has also been identified as a strong predictor of a psychiatric diagnosis for EPT children at 11 years.21 Recognizing that specific behavioral and emotional challenges during preschool years may increase the risk for later socioemotional problems is crucial for early identification of appropriate interventions. A Swedish study showed that adolescents born EPT reported lower rates of available social interactions than FT peers. They described themselves as engaging less in risk-taking behaviors such as drinking alcohol, possibly explained by them being less socially active.2 Thus, the tendency toward more withdrawn behavior among EPT preschoolers may be linked to longer term patterns of social interactions.7 Part of the differences in socioemotional behavior between EPT and FT children may be explained by family and parental factors. EPT birth is associated with complications that may cause parental mental health problems and affect parental perceptions of the child.8,15,16 Miles and Holditch-Davis14 found that mothers of 3-year-old children born preterm were more protective of their children and less demanding of them compared with siblings. This parenting style was associated with the mothers’ distressing experiences of the neonatal period. Weiss and Seed11 demonstrated that factors within the family, including maternal mental health and poor family functioning, were associated with behavioral problems in a group of 2-year-old children born preterm. Internalizing problems had the strongest relationship to these environmental factors. McManus and Poehlmann12 showed that early exposure to maternal depressive symptoms has a negative influence on preterm children’s cognitive function. In this study, cognitive function was examined as a mediator between prematurity and behavioral outcome. The prevalence of internalizing and externalizing problems and total problem score within clinical range was higher in the EPT group compared with controls. Although the effect sizes of these group differences were modest, the actual numbers of children with clinical difficulties showed a noteworthy difference. Our result confirms findings of other studies, mainly during school age. In this study, we found no interaction effects between birth status and sex on CBCL/1½-5 scores, consistent with findings by Spittle et al.7 Since studies on preterm outcome at older ages often identify sex differences in behavioral outcome, it is possible that these differences emerge and manifest more during early school years. Using a series of hierarchical regression models, we investigated whether developmental variables mediated Journal of Developmental & Behavioral Pediatrics
the effects of extreme prematurity on behavioral problems. Our results suggest that cognitive, language, and motor development fully mediated the relationship between birth status and behavior, above the effect of sociodemographic variables. This indicates that behavioral differences between EPT and FT children can be explained by differences in developmental functioning. Our results therefore support the mediation model, that is, the effect of birth status on CBCL/1½-5 outcome may be accounted for by its effect on Bayley-III outcome, which in turn affects behavioral variables. These results are comparable with other reports demonstrating how the effect of prematurity as a predictor of behavior may be transformed by mediators like IQ and neuromotor functioning.18,19,29 Furthermore, the results indicate that neonatal complications are related to birth status and the CBCL/1½-5 scores, possibly since they affect the developmental variables. Unlike previous studies, it was not possible to differentiate the impact of developmental factors on specific behavioral domains, as the mediators had similar explanatory effects regardless of which domain was used as a dependent variable. Each of the Bayley-III composite scores proved to be significant as a mediator (given the impact of sociodemographic factors). Therefore, the relationship between birth status and behavior may actually be mediated by developmental functioning on a more general level at this age. There were some limitations in this study. The proportion of foreign-born mothers was higher in the EPT families excluded from analyses. A possible explanation was that CBCL/1½-5 text versions were only available in Swedish. It would have been preferable to have versions available in other widely used migrant languages in Sweden. Both Bayley-III and CBCL/1½-5 scores were obtained through similar methods of data collection (i.e., ratings), which may introduce a common method bias. However, they were performed separately by different raters (parents and psychologists, respectively), which might reduce this potential bias. The CBCL/1½-5 and the Bayley-III are developed and normed in the United States, which could have introduced a cultural bias. By including a control group and using the mean and SD values when investigating EPT outcome (instead of US norms), the bias impact was likely reduced. The CBCL/1½-5 outcome showed significant variability with notably high SD values, which may have affected measurement stability. Previous studies, with populations comparable with ours, show similar variability. This could indicate an inherent characteristic of the instrument.3,4 All analyses were confirmed using nonparametric analyses due to skewed distributions. The measure of behavioral problems was obtained by parental ratings only, which introduces a potential rater bias. Some EPT parents may struggle to have others view their child as normal as possible, whereas some emphasize the child as being special and vulnerable.14,15 This may threaten the validity and the reliability of the Vol. 35, No. 7, September 2014
assessment of behavioral problems. Although our results do support a mediation model, there are several other potential mediators than those selected in this study. Future research could benefit from including factors like infant temperament, parental IQ and psychological health, social interaction patterns and family genetic risks, and from using multiple methods to obtain information on child behavior.
CONCLUSIONS This study investigated differences between extremely preterm (EPT) children and full-term controls, with parent-rated behavioral problems as the main outcome variable. The proportion of children showing behavioral problems within a clinical range was significantly higher in the EPT group. Levels of cognitive, language, and motor development mediated the between-group differences in behavioral problems, when controlling for sociodemographic factors. We emphasize the importance of considering multifactorial pathways of prediction when examining EPT outcome. The effect that developmental factors have on behavior should highlight the importance of interventions focusing on both behavioral development and intellectual stimulation in EPT populations. Our findings also encourage behavioral assessments during preschool years, since early signs of deficits may be detected before behavioral problems manifest at school age. In several countries, EPT aftercare is organized in neonatal units by multidisciplinary teams. Practices sometimes vary depending on the individual hospital and sufficient resources. Developing structured follow-ups based on harmonized guidelines is therefore important. In Sweden, a national EPT follow-up program recently started, aiming at early identification of deficits and intervention planning. It includes standardized examinations by pediatricians, psychologists, ophthalmologists, and physiotherapists at 2 years and 5.5 years. Although developmental follow-up is crucial, family-based approaches are likewise important. Reducing parental stress and improving parent–child interactions are related to better socioemotional and behavioral development in children born preterm.13 Hence, programs focusing on both child development and family environment are warranted. ACKNOWLEDGMENTS The authors acknowledge Kari Raivio and Rachel Maddux for reviewing the English language of the article.
REFERENCES 1. Samara M, Marlow N, Wolke D. Pervasive behavior problems at 6 years of age in a total-population sample of children born at ,/5 25 weeks of gestation. Pediatrics. 2008;122:562–573. 2. Hallin AL, Stjernqvist K. Follow-up of adolescents born extremely preterm: self-perceived mental health, social and relational outcomes. Acta Paediatr. 2011;100:279–283. 3. Stoelhorst GM, Martens SE, Rijken M, et al. Behaviour at 2 years of age in very preterm infants (gestational age , 32 weeks). Acta Paediatr. 2003;92:595–601. © 2014 Lippincott Williams & Wilkins
441
4. Potijk MR, de Winter AF, Bos AF, et al. Higher rates of behavioural and emotional problems at preschool age in children born moderately preterm. Arch Dis Child. 2012;97:112–117. 5. Reijneveld SA, de Kleine MJ, van Baar AL, et al. Behavioural and emotional problems in very preterm and very low birthweight infants at age 5 years. Arch Dis Child Fetal Neonatal Ed. 2006;91: 423–428. 6. Arpi E, Ferrari F. Preterm birth and behaviour problems in infants and preschool-age children: a review of the recent literature. Dev Med Child Neurol. 2013;55:788–796. 7. Spittle AJ, Treyvaud K, Doyle LW, et al. Early emergence of behavior and social-emotional problems in very preterm infants. J Am Acad Child Adolesc Psychiatry. 2009;48:909–918. 8. Weiss SJ, Chen JL. Factors influencing maternal mental health and family functioning during the low birthweight infant’s first year of life. J Pediatr Nurs. 2002;17:114–125. 9. Wong HS, Edwards P. Nature or nurture: a systematic review of the effect of socio-economic status on the developmental and cognitive outcomes of children born preterm. Matern Child Health J. 2013; 17:1689–1700. 10. Scott MN, Taylor HG, Fristad MA, et al. Behavior disorders in extremely preterm/extremely low birth weight children in kindergarten. J Dev Behav Pediatr. 2012;33:202–213. 11. Weiss SJ, Seed MS. Precursors of mental health problems for low birth weight children: the salience of family environment during the first year of life. Child Psychiatry Hum Dev. 2002;33: 3–27. 12. McManus BM, Poehlmann J. Parent-child interaction, maternal depressive symptoms and preterm infant cognitive function. Infant Behav Dev. 2012;35:489–498. 13. Huhtala M, Korja R, Lehtonen L, et al. Associations between parental psychological well-being and socio-emotional development in 5-year-old preterm children. Early Hum Dev. 2014; 90:119–124. 14. Miles MS, Holditch-Davis D. Compensatory parenting: how mothers describe parenting their 3-year-old, prematurely born children. J Pediatr Nurs. 1995;10:243–253. 15. Allen EC, Manuel JC, Legault C, et al. Perception of child vulnerability among mothers of former premature infants. Pediatrics. 2004;113:267–273. 16. Jonn-Seed MS, Weiss SJ. Maternal state of mind and expressed emotion: impact of mothers’ mental health, stress, and family satisfaction. J Am Psychiatr Nurses Assoc. 2005;11:135–143.
442 Behavioral Outcomes in Children Born Extremely Preterm
17. Nadeau L, Boivin M, Tessier R, et al. Mediators of behavioral problems in 7-year-old children born after 24 to 28 weeks of gestation. J Dev Behav Pediatr. 2001;22:1–10. 18. Bayless S, Pit-ten Cate I, Stevenson J. Behaviour difficulties and cognitive function in children born very prematurely. Int J Behav Dev. 2008;32:199–206. 19. Girouard PC, Baillargeon RH, Tremblay RE, et al. Developmental pathways leading to externalizing behaviors in 5 year olds born before 29 weeks of gestation. J Dev Behav Pediatr. 1998;19: 244–253. 20. Treyvaud K, Doyle LW, Lee KJ, et al. Social-emotional difficulties in very preterm and term 2 year olds predict specific social-emotional problems at the age of 5 years. J Pediatr Psychol. 2012;37:779–785. 21. Johnson S, Hollis C, Kochhar P, et al. Psychiatric disorders in extremely preterm children: longitudinal finding at age 11 years in the EPICure study. J Am Acad Child Adolesc Psychiatry. 2010;49: 453–463. 22. EXPRESS Group. One-year survival of extremely preterm infants after active perinatal care in Sweden. JAMA. 2009;301:2225–2233. 23. Serenius F, Kallen K, Blennow M, et al. Neurodevelopmental outcome in extremely preterm infants at 2.5 years after active perinatal care in Sweden. JAMA. 2013;309:1810–1820. 24. Bayley N. Bayley Scales of Infant and Toddler Development Technical Manual. 3rd ed. San Antonio, TX: The Psychological Corporation; 2006. 25. Johnson S, Moore T, Marlow N. Using the Bayley-III to assess neurodevelopmental delay: which cut-off should be used? Pediatr Res. 2014;75:670–674. 26. Achenbach TM, Rescorla LA. Manual for the ASEBA Preschool Forms & Profiles. Burlington, VT: University of Vermont, Research Centre for Children, Youth, & Families; 2000. 27. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, MI: L. Erlbaum Associates; 1988. 28. Farooqi A, Hagglof B, Sedin G, et al. Mental health and social competencies of 10- to 12-year-old children born at 23 to 25 weeks of gestation in the 1990s: a Swedish national prospective follow-up study. Pediatrics. 2007;120:118–133. 29. Loe IM, Lee ES, Luna B, et al. Behavior problems of 9-16 year old preterm children: biological, sociodemographic, and intellectual contributions. Early Hum Dev. 2011;87:247–252. 30. Baron RM, Kenny DA. The moderator mediator variable distinction in social psychological-research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986;51:1173–1182.
Journal of Developmental & Behavioral Pediatrics
Behavioral Outcomes at Corrected Age 2.5 Years in Children Born Extremely Preterm Johanna Månsson, MS, Karin Stjernqvist, PhD, Martin Bäckström, PhD ABSTRACT: Objective: This study examined a national cohort of 2.5-year-old children born extremely preterm with respect to behavioral problems from the perspective of parents and whether developmental variables mediated the effects of extreme prematurity on behavioral problems. Methods: As a part of the Extremely Preterm Infants in Sweden Study (EXPRESS), 344 children born before 27 weeks of gestation and 338 control children were given the Cognitive, Language, and Motor Scales of the Bayley-III and the Parent report Child Behavior Checklist/1½-5 (CBCL/1½-5). CBCL/1½-5 assigns 7 syndrome scores, further classified into composite scores for internalizing and externalizing behavior as well as total problems. Group differences in behavioral difficulties and prevalence of clinical problems were calculated. Bayley-III scores were used in regression models to determine if developmental factors mediated the effects of extreme prematurity on behavioral problems, after controlling for sociodemographic factors. Results: Preterm children had significantly higher mean T-scores on internalizing, externalizing, and total problems, as compared with the control subjects, but these were still within the average range. However, the proportion of subjects showing behavioral problems within the clinical range was significantly higher in the preterm group. Levels of cognitive, language, and motor development mediated the between-group differences in behavioral problems. Conclusion: Our findings encourage behavioral assessments during preschool years and emphasize the importance of considering multifactorial pathways of prediction when examining prematurity outcome. (J Dev Behav Pediatr 35:435–442, 2014) Index terms: Bayley-III, behavioral problems, CBCL/1½-5, extremely preterm, mediation analysis.
T
he survival of children born extremely preterm (EPT; ,28 gestational weeks) has increased during the past decades. Investigations of behavioral sequelae consistently show that children born EPT may suffer from a range of difficulties. Depressive symptoms, withdrawn behavior, poorer peer relationships, and increased risks of being diagnosed with attention deficit disorder or attention-deficit hyperactivity disorder are well-known outcomes at school age and adolescence.1–3 Sex differences regarding types of behavioral problems are often found within very preterm (VPT) and EPT populations, wherein boys are more likely to develop externalizing and hyperactivity problems and girls are more likely to develop internalizing problems.4,5 Samara et al1 also found these outcomes when comparing EPT children with same-sex full-term (FT) peers aged 6 years. During
From the Department of Psychology, Lund University, Lund, Sweden. Received January 2014; accepted June 2014. This study was supported by grants from the Swedish Research Council 20063855, 2009-4250, the Crafoord Foundation, the Linnéa and Josef Carlsson’s Foundation, and the “Nils W Svenningsens Stiftelse för Prematurforskning.” Disclosure: The authors declare no conflict of interest. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.jdbp.org). Address for reprints: Johanna Månsson, MS, Department of Psychology, Lund University, Box 213, SE-221 00 Lund, Sweden; e-mail: Johanna.Mansson@psy. lu.se. Copyright Ó 2014 Lippincott Williams & Wilkins
Vol. 35, No. 7, September 2014
preschool years, EPT children are rated by their parents as showing poorer emotional regulation and social competence as well as more somatic problems compared with children born at term. However, less is known about the extent of behavioral challenges at these younger ages.6,7 Some studies have sought to identify factors that influence the occurrence of behavioral problems in children born preterm. Perinatal factors such as prolonged need of artificial ventilation, length of stay in neonatal care units, and cerebral abnormalities have been documented as risk factors.5,6,8 Low socioeconomic status (SES), including environmental factors like limited parental education and economic disadvantages, is related to an increased risk of prematurity as well as behavior problems.9,10 Behavior difficulties are also associated with poor parental well-being.11–13 Distressing experiences during the neonatal period may affect parental perceptions of their child, which in turn influences their child rearing practices.14–16 The relationship between prematurity and behavioral outcome, and how it may be affected by developmental factors, has been explored in studies using mediation analyses. Nadeau et al17 demonstrated that IQ mediated the relationship between prematurity and hyperactivity, whereas neuromotor function mediated the relationship between prematurity and withdrawn behavior. Bayless et al18 showed that cognitive and neuromotor development predicted behavior difficulties better than www.jdbp.org | 435
prematurity per se. Both studies were performed with school-aged children born VPT or EPT. Girouard et al19 reported that language difficulties, as predicted by both neonatal factors and SES, were associated with higher rates of hyperactivity in a group of 5-year-old VPT children. Few studies so far have investigated potential developmental mediators between extreme prematurity and behavioral outcome during early preschool years. The main objective of the population-based Extremely Preterm Infants in Sweden Study (EXPRESS) is to investigate the outcome of EPT birth. The first aim of this report was to assess the extent of behavioral problems at 2.5 years of age within the EXPRESS population compared with a control group of children born FT. Behavioral difficulties during preschool years have been shown to be predictive of similar problems later in life.20,21 Early identification of difficulties is therefore important as it would improve the ability to anticipate special needs and promote early interventions. The second aim was to explore the relationships between EPT birth, cognitive, language and motor development, and behavioral problems. Understanding the predictive pathways of behavioral outcome aids identification of certain risk groups and intervention planning. We hypothesized that EPT children would show significantly more behavioral difficulties compared with FT controls and that the influence of birth status (EPT/FT) would be mediated by developmental factors.
METHODS Participants The Extremely Preterm Group The EXPRESS population consists of all infants born with a gestational age of younger than 27 weeks from April 1, 2004, to March 31, 2007, in Sweden. The incidence of extreme prematurity was 0.33%. Seventy percent (497/707) survived till 1 year of age.22 Six children died before reaching the age of 2.5 years. Thirty children were not eligible for follow-up (mother had a protected identity [n 5 3]; families had moved abroad [n 5 3]; preliminary identity number given at birth did not match [n 5 24]) leaving 461 as possible candidates for follow-up. Five families declined participation, and 57 children were lost to follow-up. Eighty-seven percent (399/461) underwent the assessments used in this study at 2.5 years of age (corrected for the extremely preterm [EPT]). The 5 children who declined participation were all born to non-Nordic mothers and did not have severe neonatal morbidities.23 No significant differences with respect to mean birth weight, gestational age, and rate of male sex were observed between the children who were assessed and those lost to follow-up. Fifty-five (13.8%) were excluded from further analyses in this study due to lack of information on the behavioral and/or developmental assessments. Hence, a total of 344 children born EPT were included in the analyses of this report. There were no differences regarding the prevalence of major neonatal morbidity or parental education level 436 Behavioral Outcomes in Children Born Extremely Preterm
when comparing the excluded and included families. The percentage of non-Nordic mothers was significantly higher in the excluded group: 24.4% versus 9.3% in the included group. The Control Group For each EPT child, a list of 10 potential control children was generated from the Swedish Medical Birth Registry. Selection criteria were singleton birth at term with 5-minute Apgar score .4, and matching for domicile, sex, day of birth, and ethnicity. If the first set of parents declined participation, the second set was approached, and so on, resulting in a total of 366 controls. Twenty-eight (7.7%) were omitted from analyses due to missing information on the behavioral and/or developmental assessments, leaving 338 controls.
Measures Child Behavior Checklist 1½ to 5 Years The Child Behavior Checklist/1½-5 (CBCL/1½-5) questionnaire was used to obtain standardized parental ratings of children’s behavioral functioning. Each item (99 problem items; 1 open-ended item for recording problems not listed) is scored on a 3-point scale (0 5 not true, 1 5 somewhat/sometimes true, 2 5 very true) based on the child’s behavior during the preceding 2 months. Results are expressed in 7 syndrome scales of co-occurring problems designated as emotionally reactive, anxious/depressed, somatic complaints, withdrawn, sleep problems, attention problems, and aggressive behavior. The sum of all problem items is the total problem score (TPS). Additionally, the CBCL/1½-5 syndrome scales can be classified into 2 broad groups of problems: internalizing problems (emotionally reactive, anxious/depressed, somatic complaints, and withdrawn syndrome scales) and externalizing problems (attention problems and aggressive behavior syndrome scales). Higher scores indicate more problematic behavior. The CBCL/1½-5 text version was available in Swedish. Non– Swedish-speaking parents were assisted in their ratings through oral translations by either the examiner or a professional interpreter. Bayley-III Development was assessed using the Bayley Scales of Infant and Toddler Development, 3rd Edition (BayleyIII).24 Bayley-III evaluates developmental functioning of children between the ages of 1 and 42 months through a series of standardized test items. Bayley-III provides 3 index scales: cognitive, language, and motor. The cognitive index scale assesses sensorimotor development, concept formation and categorization, and memory. The language index is composed of the distinct receptive communication and expressive communication subtests. Receptive communication measures verbal comprehension, the understanding of requests, and morphological markers. Expressive communication measures preverbal communication, vocabulary, and syntactic development. The motor index consists of fine motor and gross motor subtests. The fine motor measures functional hand skills Journal of Developmental & Behavioral Pediatrics
and perceptual motor integration. The gross motor examines static positioning, movement of limbs, and balance. Norm-referenced composite scores for the indices are derived from sums of the subtest scale scores (a composite score equivalent is available for the cognitive index scale) and ranges from 40 to 160 with a mean of 100 and SD of 15. Previous reports show that the Bayley-III has psychometric weaknesses that can cause inflated means.25 Using norm-referenced scores may result in underestimations of delay. However, in the EXPRESS, test scores are consistently evaluated on the basis of mean and SD values of the controls.
Procedure The Bayley-III assessments were performed by licensed psychologists. At the time of assessment, parents filled out the CBCL/1½-5 questionnaire. This procedure was administered at the 7 perinatal centers in Sweden. Because psychologists had sometimes previously met the children during their stay at the neonatal care unit, it was not possible for them to be blinded to birth status. The study was approved by the Regional Ethics Review Board at Lund University (Dnr 469/2007). All parents provided written consent for the assessment.
Data Analysis Baseline sociodemographic and neonatal characteristics of the EPTs were compared with the controls using t tests and x2 tests where appropriate. There are currently no Swedish national norm scores for the CBCL/1½-5. Raw scores were therefore converted into standardized T-scores based on control group mean and SD values. The CBCL/1½-5 manual recommends using T-scores ,65 to indicate normal range on the syndrome scales.26 Differences in mean CBCL/1½-5 scores of EPT born children compared with controls were examined using t tests. Results were confirmed by nonparametric Mann–Whitney U tests due to positively skewed distributions of the CBCL/1½-5 scales. Potential interactions of birth status and sex on behavioral scores were explored using factorial analyses of variance. Cohen’s guidelines for effect size were followed wherein a large effect is 0.8, a medium effect is 0.5, and a small effect is 0.2.27 Cutoffs for clinical behavior problems on the internalizing and externalizing composite scales and the TPS are usually set at the 90th percentile.26 In a Swedish follow-up study of 10-year-old children born EPT, CBCL scores above the 90th percentile for the same-sex control subjects were used as cutoffs for clinical range.28 The same procedure was used in this research. Chisquare tests were performed to compare differences in the prevalence of behavioral problems within clinical range between the EPT children and controls. The phi correlation coefficient (f) was calculated to determine the strength of the association between the variables birth status and behavioral problems within clinical range, that is, the effect size. Vol. 35, No. 7, September 2014
To investigate the contributions of different factors to behavioral outcomes, we used mediation analysis as described by Bayless et al18 and Loe et al.29 Mediation analysis explores the mechanism through which a predictor variable affects an outcome measure. In our study, birth status (EPT/full term) was used as an independent variable (IV); CBCL/1½-5 internalizing, externalizing, and TPS were used as dependent variables (DVs); Bayley-III cognitive, language, and motor composite scores were used as potential mediating variables (MVs). Behavioral problems are likely to be associated with social factors. Parental education is a frequently used socioeconomic status indicator in studies investigating preterm outcome and was therefore included as a control variable.9 The variable expressed the highest academic achievement between the 2 parents and was classified in 3 levels: (1) below high school, (2) high school, and (3) above high school (a 2-year university diploma and above). Maternal first-generation immigrant status was also included as a control variable (foreignborn mother). The 3 requirements for mediation analysis were tested using separate regression analyses.30 Additional hierarchical regression analyses were performed to investigate whether the MVs contribute unique variance to the DVs, above and beyond the contribution of the IV. Mediation is considered partial if the contribution of the IV is reduced but still significant, whereas it is total if the IV contribution is reduced to nonsignificance. Results were confirmed using ranked values due to nonnormal distributions of the CBCL/1½-5 scores. For all statistical analyses, SPSS for Windows Version 20.0 (IBM, Armonk, NY) was used.
RESULTS Sociodemographic and Infant Characteristics Sociodemographic and infant characteristics are shown in Table 1. Despite matching attempts, x2 analyses demonstrated that the parental education level was statistically significantly lower and more mothers were foreign born in the extremely preterm (EPT) group compared with the control group. Immigrant mothers were mainly from other Nordic, former East-European, Middle East, or North African countries (Table 1). In both groups, there was a large number of missing values in the parental education variable. This was handled by pairwise exclusions of cases in further analyses.
Behavioral Outcomes The mean age at assessment was 30.1 (corrected) months for the preterm group versus 30.6 (chronological) months for the control group. The planned age range for assessment was 30 6 3 months; 98.7% of the EPT’s versus 97.8% of the full-term (FT) children were tested within this range. Behavioral outcome in EPT children versus FT controls and effect size estimations of © 2014 Lippincott Williams & Wilkins
437
Table 1. Sociodemographic Factors, Infant Characteristics at Birth, Neonatal Morbidity, and Prevalence of Cerebral Palsy Characteristic
EPT FT (n 5 344) (n 5 338) p
Maternal age, mean (SD)
31.4 (5.8)
30.9 (4.9) .309
Parental education, n (%) Below high school
.006 19 (5.5)
7 (2.1)
High school
134 (39.0) 109 (32.2)
High school or higher
138 (40.1) 164 (48.5)
Missing information
53 (15.4)
58 (17.2) .532
61 (17.7)
24 (7.1) .000
Other Nordic country
10 (2.9)
12 (3.6)
Former East-European country
18 (5.2)
2 (0.6)
Middle East or North African country
16 (4.7)
7 (2.1)
Othera
17 (4.9)
3 (0.8)
8 (2.3)
0 (0)
Foreign-born mother, n (%)
Missing information Male sex, n (%)
188 (54.7) 186 (55)
Birth weight, mean (SD), g
780 (170) 3622 (498) .000
Gestational age, mean (SD), wk
24.9 (1.0)
39.5 (1.1) .000
SGA, n (%)b
62 (18.0)
NR
Multiple births, n (%)
55 (16)
NR
Major neonatal morbidities, n (%)c 182 (52.9)
NR
Severe BPD
80 (23.2)
Severe IVH
30 (8.7)
NR
NEC
18 (5.2)
NR
PVL Severe ROP
.921
NR
14 (4.1)
NR
115 (33.4)
NR
CP, n (%)
.000
No
323 (93.9) 338 (100)
Mild CPd
10 (2.9)
0 (0)
Moderate CPe
10 (2.9)
0 (0)
1 (0.3)
0 (0)
f
Severe CP
5 other European, North American, sub-Saharan African, Southeast Asian, South and Middle American countries, New Zealand and Australia. bSGA, birth weight more than 2 SD below the Swedish standard population mean. cMajor neonatal morbidities included severe IVH defined as $Grade 3, PVL, severe BPD defined as $30% oxygen dependency at 36 weeks, severe ROP defined as $Phase 3, and/or NEC. dMild CP defined as ambulant without aid. eModerate CP defined as ambulant with aid. fSevere CP defined as nonambulant. EPT, extremely preterm; FT, full term; BPD, bronchopulmonary dysplasia; CP, cerebral palsy; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; NR, no record; PVL, periventricular leukomalacia; ROP, retinopathy of prematurity; SGA, small for gestational age. aOther
group differences are illustrated in Table 2. Parents in the EPT group reported significantly higher scores on all the CBCL/1½-5 syndrome scales, as well as internalizing and externalizing problems and total problem score (TPS), compared with the control subjects. The EPT children had a significantly higher prevalence of scores within clinical range for internalizing and externalizing problems and TPS compared with the controls (Table 3). The analysis of variance tests revealed no significant interaction effects between birth status and sex on CBCL/ 1½-5 scores (not shown). 438 Behavioral Outcomes in Children Born Extremely Preterm
Mediation Analysis Regression analyses were performed to confirm that the requirements for mediation analysis were met and found that (1) birth status was associated with the Bayley-III composite scores, (2) birth status was associated with CBCL/1½-5 outcome, and (3) the Bayley-III composite scores were associated with CBCL/1½-5 outcome. Results of the preliminary regression analyses are shown in the supporting information (Table S1, Supplemental Digital Content 1, http://links.lww.com/JDBP/A63). Further hierarchical regression analyses were performed in 3 steps. Birth status was entered in the first, foreign-born mother and parental education in the second, and Bayley-III cognitive, language, and motor composite scores in the third step. The CBCL/1½-5 internalizing, externalizing, and TPS T-scores were entered as dependent variable (DVs). Three equations were calculated, 1 for each DV. Results are presented in Table 4. The variables foreign-born mother and parental education each separately contributed to the variance of internalizing problems, partly reducing the contribution of birth status. When Bayley-III composite scores were entered, the contribution of birth status was reduced to nonsignificance. This indicates complete mediation, after controlling for sociodemographics. Parental education significantly contributed to the externalizing score variance, whereas the variable foreign-born mother did not. The contribution of birth status was no longer statistically significant after adding Bayley-III composite scores to the equation, again demonstrating complete mediation. The total regression model only modestly contributed to the externalizing score variance. Similar to the previous equations, the effect of birth status was reduced to statistical nonsignificance when Bayley-III composite scores were entered in the final step, after controlling for the sociodemographic variables. Bayley-III composite scores were entered simultaneously, indicating that cognition alone significantly contributed to the externalizing and TPS variance, whereas cognitive and motor functions both significantly contributed to the internalizing score variance (Table 4) (An additional analysis, using the SPSS AMOS (IBM, Chicago, IL) program, was performed to confirm the results. A latent cognitive variable was measured using the Bayley-III cognitive, language, and motor scores as indicators. A latent behavior variable was measured using the CBCL/1½-5 internalizing and externalizing variables as indicators. All other variables were included in the meditation model as observed variables. Significances of the coefficients were tested with bootstrap confidence intervals. In this model, missing values were replaced with regression weighted imputation. The total effect between birth status and behavior was significant [b 5 .187]; also the indirect effect from birth status through the cognitive latent variable was significant [b 5 2.161]. The direct effect from birth status and behavior was Journal of Developmental & Behavioral Pediatrics
Table 2. Group Differences in CBCL/1½-5 Syndrome and Composite Scale Scores Between EPT Children and FT Controls EPT (n 5 344), T-score
FT (n 5 338), T-score
Mean
SD
Mean
SD
p
Effect Size Cohen’s d
Emotionally reactive
52.0
12.8
50.0
10.0
.025
0.17
Anxious/depressed
53.7
15.6
50.0
10.0
.000
0.28
Somatic complaints
56.3
14.2
50.0
10.0
.000
0.51
Withdrawn
57.1
17.5
50.0
10.0
.000
0.50
CBCL/1½-5 Syndrome scale
Sleep problems
51.6
10.3
50.0
10.0
.035
0.16
Attention problems
55.4
12.2
50.0
10.0
.000
0.48
Aggressive behavior
51.9
11.2
50.0
10.0
.018
0.18
Composite scale Internalizing
56.1
15.0
50.0
10.0
.000
0.48
Externalizing
53.1
11.4
50.0
10.0
.000
0.29
TPS
54.8
13.0
50.0
10.0
.000
0.41
EPT, extremely preterm; FT, full term; CBCL, child behavior checklist; TPS, total problem score.
nonsignificant [b 5 .026]. This clearly supported the mediation hypothesis. The sociodemographic variables were included as covariates. To reach acceptable fit, a number of correlations between birth status and sociodemographics and sociodemographics and the error term of the language variable were added. They did not influence the coefficients related to the mediation hypothesis. The final model had an excellent fit [comparative fit index 5 0.972]). We also investigated as to whether composite scores served as mediators one-by-one. Results indicate that all 3 composite scores partially or completely mediated the relationship (not shown). The contributions of the major neonatal morbidities on CBCL/1½-5 outcome were investigated by including them to the hierarchical regression models. Results are presented in the supporting information (Table S2, Supplemental Digital Content 2, http://links.lww.com/JDBP/ A64). The contribution of birth status on internalizing problems, externalizing problems, and TPS was no longer statistically significant after adding the morbidity variables. No variable by itself was contributing significantly to the hierarchical regression models, except for
Table 3. Group Differences in Prevalence of Behavior Within Clinical Range Between EPT Children and FT Controls
EPT (n 5 344) FT (n 5 338) p
Effect Size (u)
Behaviors within clinical range, n (%)a Internalizing
72 (20.9)
27 (8.0)
.000
0.18
Externalizing
67 (19.5)
31 (9.2)
.000
0.15
TPS
68 (19.8)
30 (8.9)
.000
0.16
range $90th percentile. EPT, extremely preterm; FT, full term; TPS, total problem score. aClinical
Vol. 35, No. 7, September 2014
severe intraventricular hemorrhage to the TPS variance (The regression models were further explored including (1) EPT children without major morbidities and (2) EPT children with major morbidities. Results confirmed that the Bayley-III scores mediate the effect of birth status on CBCL/1½-5 scores, in both groups. Results were less prominent in the EPT children without major morbidities).
DISCUSSION This study compared extremely preterm (EPT) children with full-term (FT) children on the basis of parental ratings of behavioral problems at 2.5 years of age. Group differences were found on all types of problems, with a main focus on the differences with the largest effect sizes, that is, the differences that are presumed to be clinically relevant. EPT children had more somatic problems than controls. This may be expected because EPT birth is associated with severe morbidity and advanced intensive care in the neonatal period, followed by physical problems like eating difficulties and pulmonary complications.3 Children born EPT also had more attentional difficulties than controls. Previous studies show that attention-deficit hyperactivity disorder and attention deficit disorder are among the most frequent psychiatric outcomes of EPT birth at school age. In their study on EPT outcome, Samara et al1 demonstrated that overactivity is explained by cognitive functioning, whereas regulation of attention is not. Hence, attention problems may be a specific feature of development after EPT birth. Johnson et al21 present the increasing indication of an “EPT phenotype,” characterized by inattention, emotional disorders, and social problems. Our results suggest that attention problems may emerge during early preschool years, which highlights the importance of early screening for attention deficit symptoms. © 2014 Lippincott Williams & Wilkins
439
Table 4. Hierarchical Multiple Regression Analyses Estimating Bayley-III Variables as Mediators of the Effect of Extreme Prematurity on Internalizing, Externalizing, and Total Behavioral Problems, Controlling for Sociodemographic Factors R2
DR2
B
b
p
Internalizing Step 1
0.054 0.054
26.01 2.232 .000
Birth status Step 2
0.165 0.111
.000 24.25 2.162 .000
Birth status Foreign-born mother
9.30
.244 .000
24.21 2.187 .000
Parental education Step 3
.000
0.234 0.069
.000 21.12 2.043 .301
Birth status Foreign-born mother
8.47
.222 .000
Parental education
23.59 2.160 .000
Bayley-III score: cognitive
20.160 2.148 .011
Bayley-III score: language
20.066 2.077 .184
Bayley-III score: motor
20.089 2.110 .031
Externalizing Step 1
0.020 0.020
23.07 2.142 .001
Birth status Step 2
0.068 0.048
.000 22.30 2.107 .010
Birth status Foreign-born mother
1.44
.046 .280
23.82 2.207 .000
Parental education Step 3
.001
0.103 0.035
.000 20.609 2.028 .531
Birth status Foreign-born mother
0.822
.026 .541
23.44 2.186 .000
Parental education Bayley-III score: cognitive
20.133 2.149 .017
Bayley-III score: language
20.035 2.050 .425
Bayley-III score: motor
20.018 2.028 .616
Total problems Step 1
0.041 0.041
Step 2
0.133 0.092
Foreign-born mother
5.82
Foreign-born mother Parental education
.169 .000
24.51 2.223 .000
Parental education Birth status
.000 23.34 2.141 .000
Birth status
Step 3
.000 24.78 2.202 .000
Birth status
0.194 0.061
.000 20.779 2.033 .439 5.05
.147 .000
23.99 2.197 .000
Bayley-III score: cognitive
20.167 2.171 .004
Bayley-III score: language
20.049 2.064 .282
Bayley-III score: motor
20.053 2.073 .163
Children born EPT were rated as more withdrawn, showing more symptoms of depression and anxiety as compared with controls. Our results add to the extant 440 Behavioral Outcomes in Children Born Extremely Preterm
findings of others showing that internalizing behaviors are more prominent in preterm children than aggressive and delinquent (i.e., externalizing) behaviors.7,29 Treyvaud et al20 demonstrated that emotional symptoms of 5-year-old children born very preterm were predicted by internalizing problems at 2 years. Internalizing behavior has also been identified as a strong predictor of a psychiatric diagnosis for EPT children at 11 years.21 Recognizing that specific behavioral and emotional challenges during preschool years may increase the risk for later socioemotional problems is crucial for early identification of appropriate interventions. A Swedish study showed that adolescents born EPT reported lower rates of available social interactions than FT peers. They described themselves as engaging less in risk-taking behaviors such as drinking alcohol, possibly explained by them being less socially active.2 Thus, the tendency toward more withdrawn behavior among EPT preschoolers may be linked to longer term patterns of social interactions.7 Part of the differences in socioemotional behavior between EPT and FT children may be explained by family and parental factors. EPT birth is associated with complications that may cause parental mental health problems and affect parental perceptions of the child.8,15,16 Miles and Holditch-Davis14 found that mothers of 3-year-old children born preterm were more protective of their children and less demanding of them compared with siblings. This parenting style was associated with the mothers’ distressing experiences of the neonatal period. Weiss and Seed11 demonstrated that factors within the family, including maternal mental health and poor family functioning, were associated with behavioral problems in a group of 2-year-old children born preterm. Internalizing problems had the strongest relationship to these environmental factors. McManus and Poehlmann12 showed that early exposure to maternal depressive symptoms has a negative influence on preterm children’s cognitive function. In this study, cognitive function was examined as a mediator between prematurity and behavioral outcome. The prevalence of internalizing and externalizing problems and total problem score within clinical range was higher in the EPT group compared with controls. Although the effect sizes of these group differences were modest, the actual numbers of children with clinical difficulties showed a noteworthy difference. Our result confirms findings of other studies, mainly during school age. In this study, we found no interaction effects between birth status and sex on CBCL/1½-5 scores, consistent with findings by Spittle et al.7 Since studies on preterm outcome at older ages often identify sex differences in behavioral outcome, it is possible that these differences emerge and manifest more during early school years. Using a series of hierarchical regression models, we investigated whether developmental variables mediated Journal of Developmental & Behavioral Pediatrics
the effects of extreme prematurity on behavioral problems. Our results suggest that cognitive, language, and motor development fully mediated the relationship between birth status and behavior, above the effect of sociodemographic variables. This indicates that behavioral differences between EPT and FT children can be explained by differences in developmental functioning. Our results therefore support the mediation model, that is, the effect of birth status on CBCL/1½-5 outcome may be accounted for by its effect on Bayley-III outcome, which in turn affects behavioral variables. These results are comparable with other reports demonstrating how the effect of prematurity as a predictor of behavior may be transformed by mediators like IQ and neuromotor functioning.18,19,29 Furthermore, the results indicate that neonatal complications are related to birth status and the CBCL/1½-5 scores, possibly since they affect the developmental variables. Unlike previous studies, it was not possible to differentiate the impact of developmental factors on specific behavioral domains, as the mediators had similar explanatory effects regardless of which domain was used as a dependent variable. Each of the Bayley-III composite scores proved to be significant as a mediator (given the impact of sociodemographic factors). Therefore, the relationship between birth status and behavior may actually be mediated by developmental functioning on a more general level at this age. There were some limitations in this study. The proportion of foreign-born mothers was higher in the EPT families excluded from analyses. A possible explanation was that CBCL/1½-5 text versions were only available in Swedish. It would have been preferable to have versions available in other widely used migrant languages in Sweden. Both Bayley-III and CBCL/1½-5 scores were obtained through similar methods of data collection (i.e., ratings), which may introduce a common method bias. However, they were performed separately by different raters (parents and psychologists, respectively), which might reduce this potential bias. The CBCL/1½-5 and the Bayley-III are developed and normed in the United States, which could have introduced a cultural bias. By including a control group and using the mean and SD values when investigating EPT outcome (instead of US norms), the bias impact was likely reduced. The CBCL/1½-5 outcome showed significant variability with notably high SD values, which may have affected measurement stability. Previous studies, with populations comparable with ours, show similar variability. This could indicate an inherent characteristic of the instrument.3,4 All analyses were confirmed using nonparametric analyses due to skewed distributions. The measure of behavioral problems was obtained by parental ratings only, which introduces a potential rater bias. Some EPT parents may struggle to have others view their child as normal as possible, whereas some emphasize the child as being special and vulnerable.14,15 This may threaten the validity and the reliability of the Vol. 35, No. 7, September 2014
assessment of behavioral problems. Although our results do support a mediation model, there are several other potential mediators than those selected in this study. Future research could benefit from including factors like infant temperament, parental IQ and psychological health, social interaction patterns and family genetic risks, and from using multiple methods to obtain information on child behavior.
CONCLUSIONS This study investigated differences between extremely preterm (EPT) children and full-term controls, with parent-rated behavioral problems as the main outcome variable. The proportion of children showing behavioral problems within a clinical range was significantly higher in the EPT group. Levels of cognitive, language, and motor development mediated the between-group differences in behavioral problems, when controlling for sociodemographic factors. We emphasize the importance of considering multifactorial pathways of prediction when examining EPT outcome. The effect that developmental factors have on behavior should highlight the importance of interventions focusing on both behavioral development and intellectual stimulation in EPT populations. Our findings also encourage behavioral assessments during preschool years, since early signs of deficits may be detected before behavioral problems manifest at school age. In several countries, EPT aftercare is organized in neonatal units by multidisciplinary teams. Practices sometimes vary depending on the individual hospital and sufficient resources. Developing structured follow-ups based on harmonized guidelines is therefore important. In Sweden, a national EPT follow-up program recently started, aiming at early identification of deficits and intervention planning. It includes standardized examinations by pediatricians, psychologists, ophthalmologists, and physiotherapists at 2 years and 5.5 years. Although developmental follow-up is crucial, family-based approaches are likewise important. Reducing parental stress and improving parent–child interactions are related to better socioemotional and behavioral development in children born preterm.13 Hence, programs focusing on both child development and family environment are warranted. ACKNOWLEDGMENTS The authors acknowledge Kari Raivio and Rachel Maddux for reviewing the English language of the article.
REFERENCES 1. Samara M, Marlow N, Wolke D. Pervasive behavior problems at 6 years of age in a total-population sample of children born at ,/5 25 weeks of gestation. Pediatrics. 2008;122:562–573. 2. Hallin AL, Stjernqvist K. Follow-up of adolescents born extremely preterm: self-perceived mental health, social and relational outcomes. Acta Paediatr. 2011;100:279–283. 3. Stoelhorst GM, Martens SE, Rijken M, et al. Behaviour at 2 years of age in very preterm infants (gestational age , 32 weeks). Acta Paediatr. 2003;92:595–601. © 2014 Lippincott Williams & Wilkins
441
4. Potijk MR, de Winter AF, Bos AF, et al. Higher rates of behavioural and emotional problems at preschool age in children born moderately preterm. Arch Dis Child. 2012;97:112–117. 5. Reijneveld SA, de Kleine MJ, van Baar AL, et al. Behavioural and emotional problems in very preterm and very low birthweight infants at age 5 years. Arch Dis Child Fetal Neonatal Ed. 2006;91: 423–428. 6. Arpi E, Ferrari F. Preterm birth and behaviour problems in infants and preschool-age children: a review of the recent literature. Dev Med Child Neurol. 2013;55:788–796. 7. Spittle AJ, Treyvaud K, Doyle LW, et al. Early emergence of behavior and social-emotional problems in very preterm infants. J Am Acad Child Adolesc Psychiatry. 2009;48:909–918. 8. Weiss SJ, Chen JL. Factors influencing maternal mental health and family functioning during the low birthweight infant’s first year of life. J Pediatr Nurs. 2002;17:114–125. 9. Wong HS, Edwards P. Nature or nurture: a systematic review of the effect of socio-economic status on the developmental and cognitive outcomes of children born preterm. Matern Child Health J. 2013; 17:1689–1700. 10. Scott MN, Taylor HG, Fristad MA, et al. Behavior disorders in extremely preterm/extremely low birth weight children in kindergarten. J Dev Behav Pediatr. 2012;33:202–213. 11. Weiss SJ, Seed MS. Precursors of mental health problems for low birth weight children: the salience of family environment during the first year of life. Child Psychiatry Hum Dev. 2002;33: 3–27. 12. McManus BM, Poehlmann J. Parent-child interaction, maternal depressive symptoms and preterm infant cognitive function. Infant Behav Dev. 2012;35:489–498. 13. Huhtala M, Korja R, Lehtonen L, et al. Associations between parental psychological well-being and socio-emotional development in 5-year-old preterm children. Early Hum Dev. 2014; 90:119–124. 14. Miles MS, Holditch-Davis D. Compensatory parenting: how mothers describe parenting their 3-year-old, prematurely born children. J Pediatr Nurs. 1995;10:243–253. 15. Allen EC, Manuel JC, Legault C, et al. Perception of child vulnerability among mothers of former premature infants. Pediatrics. 2004;113:267–273. 16. Jonn-Seed MS, Weiss SJ. Maternal state of mind and expressed emotion: impact of mothers’ mental health, stress, and family satisfaction. J Am Psychiatr Nurses Assoc. 2005;11:135–143.
442 Behavioral Outcomes in Children Born Extremely Preterm
17. Nadeau L, Boivin M, Tessier R, et al. Mediators of behavioral problems in 7-year-old children born after 24 to 28 weeks of gestation. J Dev Behav Pediatr. 2001;22:1–10. 18. Bayless S, Pit-ten Cate I, Stevenson J. Behaviour difficulties and cognitive function in children born very prematurely. Int J Behav Dev. 2008;32:199–206. 19. Girouard PC, Baillargeon RH, Tremblay RE, et al. Developmental pathways leading to externalizing behaviors in 5 year olds born before 29 weeks of gestation. J Dev Behav Pediatr. 1998;19: 244–253. 20. Treyvaud K, Doyle LW, Lee KJ, et al. Social-emotional difficulties in very preterm and term 2 year olds predict specific social-emotional problems at the age of 5 years. J Pediatr Psychol. 2012;37:779–785. 21. Johnson S, Hollis C, Kochhar P, et al. Psychiatric disorders in extremely preterm children: longitudinal finding at age 11 years in the EPICure study. J Am Acad Child Adolesc Psychiatry. 2010;49: 453–463. 22. EXPRESS Group. One-year survival of extremely preterm infants after active perinatal care in Sweden. JAMA. 2009;301:2225–2233. 23. Serenius F, Kallen K, Blennow M, et al. Neurodevelopmental outcome in extremely preterm infants at 2.5 years after active perinatal care in Sweden. JAMA. 2013;309:1810–1820. 24. Bayley N. Bayley Scales of Infant and Toddler Development Technical Manual. 3rd ed. San Antonio, TX: The Psychological Corporation; 2006. 25. Johnson S, Moore T, Marlow N. Using the Bayley-III to assess neurodevelopmental delay: which cut-off should be used? Pediatr Res. 2014;75:670–674. 26. Achenbach TM, Rescorla LA. Manual for the ASEBA Preschool Forms & Profiles. Burlington, VT: University of Vermont, Research Centre for Children, Youth, & Families; 2000. 27. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, MI: L. Erlbaum Associates; 1988. 28. Farooqi A, Hagglof B, Sedin G, et al. Mental health and social competencies of 10- to 12-year-old children born at 23 to 25 weeks of gestation in the 1990s: a Swedish national prospective follow-up study. Pediatrics. 2007;120:118–133. 29. Loe IM, Lee ES, Luna B, et al. Behavior problems of 9-16 year old preterm children: biological, sociodemographic, and intellectual contributions. Early Hum Dev. 2011;87:247–252. 30. Baron RM, Kenny DA. The moderator mediator variable distinction in social psychological-research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986;51:1173–1182.
Journal of Developmental & Behavioral Pediatrics
