کنترل پر زحمت، اظهار احساس مثبت و رفتار مشکلات در کودکان متولد شده زودرس

Abstract The present study focused on the role of high effortful control in the expression of positive emotion and development of behavior problems in children born preterm (mean gestational age = 31.4 weeks). Using data from a prospective longitudinal study, the present study assessed effortful control and behavior problems at 24 and 36 months and positive emotional expression at 24 months in a sample of 173 children born preterm. Less positive emotional expression was associated with higher effortful control for boys but not girls. Higher effortful control was associated with fewer total behavior problems, but this relation was attenuated when socioeconomic assets were included in the model. More socioeconomic assets were associated with fewer behavior problems for both boys and girls and higher effortful control for girls. Socioeconomic assets appear to be an important factor in the development of effortful control and behavior problems in children born preterm regardless of gender, whereas positive emotional expression was important for boys. Future intervention research should examine fostering adaptive levels of effortful control in high-risk populations as a means to facilitate resilience processes.

. Introduction Although infants born preterm (<37 weeks gestation) are more likely to survive than ever before, preterm infants continue to experience elevated morbidity rates such as cognitive delays and behavior problems. Estimates suggest that 50–70% of infants born preterm may develop behavioral difficulties, including externalizing and internalizing problems (Aylward, 2005 and Bhutta et al., 2002Pinto-Martin et al., 2004 and Taylor et al., 2000). These behavioral disturbances in preterm infants may be related to early impairments in self-regulation (e.g., Davis & Burns, 2001), including problems with sustained attention, inhibitory control, and emotion regulation (Berger et al., 2007). One temperament-related construct that reflects emerging self-regulation (Zhou, Chen, & Main, 2013) is effortful control (EC), the capacity to suppress a dominant response in order to express a subdominant response (Rothbart & Bates, 2006). The present study examined the associations among effortful control, positive emotional expression, and children's behavior problems in boys and girls born preterm. The study sought to examine whether high effortful control was associated with less positive emotional expression and higher internalizing behavior problems in children born preterm, and if these associations differed between girls and boys. 1.1. Effortful control and behavior problems Temperament theory describes effortful control as the regulatory aspect of temperament that serves to modulate reactivity (Rothbart & Bates, 2006). Behavior problems are thought to reflect, in part, impairments in the ability to self-regulate. Therefore, children low in EC, including delay of gratification, slowing motor activity, lowering one's voice, suppressing and initiating an activity upon signal, and effortful attention (Kochanska, Coy, & Murray, 2001), are expected to exhibit higher behavior problem scores than children high in EC. In healthy full-term children, low EC predicts higher externalizing behavior problem scores (e.g., Eiden et al., 2009, Kochanska and Knaack, 2003 and Kochanska et al., 2009), especially behavior problems that reflect impulsivity or attentional difficulties (Murray and Kochanska, 2002 and Olson et al., 2005). However, the association between EC and internalizing behavior problems appears more complex. Some studies have found that high EC predicts higher internalizing behavior problem scores (Kochanska et al., 2000 and Murray and Kochanska, 2002), whereas other studies have documented no association (e.g., Dennis, Brotman, Huan, & Gouley, 2007). Moreover, in a sample of low-risk children, Murray and Kochanska (2002) found a curvilinear relationship between EC and behavior problems, in which moderate levels of EC predicted the fewest total behavior problems. Post hoc tests indicated that high EC was associated with higher internalizing behavior problem scores whereas low EC was associated with higher externalizing behavior problems. Few studies have examined links between EC and behavior problems in children born preterm. In one exception, Poehlmann et al. (2010) documented a link between low EC and concurrent symptoms of inattention and Attention-Deficit/Hyperactivity Disorder (ADHD) but not broadband behavior scales. High EC is often conceptualized as an asset or optimal outcome, as children may be protected from developing disorders related to inattention and disinhibition when they excel at suppressing a dominant (or impulsive) response. However, given previous findings regarding EC and internalizing problems in low-risk samples, it is possible that very high levels of EC may represent a tendency toward over-control. Eisenberg and colleagues (e.g., Eisenberg and Morris, 2002 and Eisenberg et al., 2005) have proposed a model regarding three types of control: undercontrol, overcontrol, and optimal control. Overcontrolled individuals are hypothesized to reflect a combination of optimal and less optimal aspects self-regulation. Specifically, high reactive control and low impulsivity are thought to reflect better self-regulation, whereas low effortful attention is thought to reflect poorer self-regulation (Eisenberg, Hofer, & Vaughan, 2007). In addition, overcontrolled individuals are thought to develop internalizing behavior problems and low levels of positive emotion, and they may find it difficult to engage with attractive stimuli or be spontaneous (Eisenberg et al., 2009 and Eisenberg et al., 2007). Following these hypothesized relations, some researchers suggest that higher internalizing problems are associated with low impulsivity and attentional regulation, although the associations are complex (Eisenberg et al., 2009). Studies with preterm children have not examined such possibilities, nor have they examined the relation between positive emotion expression and EC. Optimal effortful control may represent a possible source of resilience for preterm infants (Poehlmann et al., 2010). In order to promote positive development in these vulnerable infants, very high and low levels of effortful control and positive emotion expression and their relations to behavior problems should be investigated. 1.2. Positive emotional expression and effortful control Temperament theory, as described by Rothbart and colleagues (e.g., Rothbart & Bates, 2006), posits two broad temperamental systems: reactive and regulatory. EC is viewed as a regulatory temperament construct, whereas positive affect can be viewed as a reactive temperament construct. EC serves as the “brakes” on the “acceleration” of the temperamental reactive system, which include positive affect and approach (Rothbart, Ellis, & Posner, 2004). According to Rothbart and colleagues, the strength of the brakes and the force of the acceleration can influence each other. High levels of EC can serve to modulate expression of intense emotions; therefore children who exhibit few emotional expressions may also have high EC. High EC may not be adaptive or desirable in all situations, as when it results in difficulty expressing positive emotions. Rothbart and Ellis warn, “The ends achieved through EC may or may not be adaptive ones, however, and when control results in rigid response to social situations, the outcomes may not be favorable ones (1994, p. 452).” Therefore, it is possible that high levels of EC may inhibit children's ability to express positive emotions, even in socially acceptable circumstances. Low levels of positive emotions appear to be associated with high EC in healthy full-term children. For example, Kochanska and Knaack (2003) found that both parent-reported and observed joy was negatively associated with EC scores at 33 and 45, but not 22 months in typically developing children. Similarly, Kochanska et al. (2000) found that children who exhibited less intense joy displays had higher EC scores at 22 and 33 months. Consistent with temperament theory, the authors suggest that EC may serve to modulate intense emotions, and therefore a stronger EC system would be associated with less intense emotional displays. This modulation, however, may not always be optimal when it occurs in excessive amounts, especially in vulnerable children. It is possible that preterm children with high EC may find it difficult to express positive emotions, even when appropriate. Moreover, difficulty expressing positive emotions has been associated with elevated anxiety and internalizing problems (Cole, Zahn-Waxler, Fox, Usher, & Welsh, 1996). Children who demonstrated low positive affect during a laboratory temperament assessment had a frontal EEG asymmetry, which is associated with depression (Shankman et al., 2005). In addition, Lengua, Sandler, West, Wolchik, and Curran (1999) found that higher positive emotionality in children post-divorce was related to fewer depressive symptoms, threat appraisals, and conduct problems, and more active coping. The authors suggest that low levels of positive emotionality are linked to unrewarding experiences and the absence of pleasurable engagement, which then may lead to apathy and depression. However, high positive emotional expression can be associated with maladaptive outcomes as well. For example, high levels of positive emotionality have been linked to externalizing behavior problems and low levels of prosocial behavior, especially when accompanied by low levels of regulation (Eisenberg et al., 1996 and Rydell et al., 2003). It is particularly important to understand children's positive emotional expression in relation to EC in young children born preterm to determine if high EC functions as an asset or possible risk. 1.3. Gender In general, girls score higher on EC tasks than boys. A recent meta-analysis found substantial and significant gender differences in EC, with girls displaying higher EC than boys (Else-Quest, Hyde, Goldsmith, & Van Hulle, 2006). Moreover, boys born preterm display higher rates of ADHD and behavior problems than girls born preterm (Blair, 2002 and Lindstrom et al., 2011). EC may be a particularly important skill for boys born preterm, as it may protect them from developing ADHD and other problems related to inattention and impulsivity. In the present study, we assessed the main effects of gender as well as examining gender as a moderator of the relation between EC and behavior problems in children born preterm. 1.4. Family socioeconomic status and infant birthweight Previous studies have documented the effects of family socioeconomic (SES) risks on child outcomes, including children born preterm (e.g., Brooks-Gunn and Duncan, 1997, Candelaria et al., 2011 and Linver et al., 2002). For example, children from families with fewer socioeconomic assets (e.g., less maternal education) have more health problems and elevated parent-reported behavior problems (e.g., Ruijsbroek et al., 2011). Thus, we examined models with and without SES variables to assess the effect of such risks on relations between EC, positive emotional expressions, and child behavior problems. Building on the research reviewed above, we addressed five research questions using models that incorporate infant birthweight, gestational age, gender, and socioeconomic assets. • In children born preterm, is EC associated with total behavior problems in a curvilinear or linear fashion? • Do preterm children with very high EC show higher internalizing problems and do preterm children with very low EC show higher externalizing problems? • Are EC and positive emotion expression inversely related in preterm children? • Do the associations between EC, positive emotion expression, and behavior problems vary by gender? • Does the inclusion of family SES affect relations among EC, positive emotion expression, behavior problems and gender in children born preterm?

3. Results To address our research questions, curve fit estimations were completed in SPSS 20 and a series of structural equation models (SEM) were tested within Mplus 6.12 (Muthén & Muthén, 1998). SEM was chosen over other analytic approaches because it affords the simultaneous testing of multiple pathways, latent constructs, and the hypothesized associations between variables. The SEM models were specified, identified, and tested for assumption violations prior to model and path estimation and interpretation. Additionally, a full information maximum likelihood (FIML) procedure was used to address missing data. In the Mplus FIML procedure, individual missing data patterns are assessed, and means and covariances for each missing data pattern are calculated to inform the observed information matrix (Arbuckle, 1996 and Kaplan, 2009). The observed information matrix is used to generate estimates (Kenward & Molenberghs, 1998). Addressing missing data via FIML assumes data missing at random (MAR; Little & Rubin, 1989) and is preferable to pair-wise or list-wise deletion (Arbuckle, 1996). Data screening revealed an extreme income for one family (more than 3 standard deviations above the mean) and one family did not report an income. For models including income (within the SES assets index), FIML was used to estimate a relative value for these families. To assess overall EC and behavior problems latent constructs were generated. The EC latent construct contain the 24 and 36 month EC estimates and the behavioral problems constructs contained the 24 and 36 month CBCL scores. 3.1. Curve estimation Our first research question addressed the nature (linear or quadratic) of the association between EC and behavior problems for: (a) the overall sample and (b) for boys and girls. Effortful control and behavior problems were assessed at 24 and 36 months; therefore we assessed the concurrent association between the two constructs at 24 and 36 months (e.g., 24 month EC with 24 month behavior problems). Additionally, we assessed the association between latent constructs of EC and behavior problems. All analyses revealed that a linear association fit the data better when an association was present; therefore, in subsequent models linear estimates were retained (summarized in Table 2). Table 2. Curve estimations between effortful control and total behavior problems. Linear β ŧ Quadratic β ŧ 24 month EC with behavior problems All children F(1,145) = 1.37 −.12 F(2, 144) = 1.23 .09 Male F(1, 72) = .23 −.06 F(2, 71) = .19 .05 Female F(1,71) = 1.54 −.23 F(2, 70) = 1.78 .18 36 month EC with behavior problems All children F(1, 136) = .62 −.08 F(2, 135) = .39 .04 Male F(1, 67) = .09 −.04 F(2, 66) = .05 .01 Female F(1, 67) = .71 −.15 F(2, 66) = .58 .10 Latent EC with behavior problems All children F(1, 171) = 5.11 * −.21* F(2, 170) = 2.73 .06 Male F(1, 90) = 1.57 −.12 F(2, 89) = .78 −.02 Female F(1, 79) = 4.14 * −.37* F(2, 78) = 2.67 .19 Note: EC = effortful control composite. * p < .05. ŧ Standardized beta coefficient. Table options 3.2. Structural equation models Structural equation models were assessed in two steps. First, overall model fit was assessed followed by the interpretation of individual path coefficients. To assess the overall model fit, 3 indices were evaluated, including X2, root mean square error of approximation (RMSEA), and the comparative fit index (CFI). The X2 index is a model of misspecification; therefore, a significant X2 means that the model does not fit the sample data. Because some scholars claim that the exact fit tested in X2 is an unrealistic standard, indices of approximate fit like RMSEA were also assessed. RMSEA tests whether the model fits the population approximately. In RMSEA, .00 is the best possible fit, with higher values indicating poorer fit. Within this study, the CFI compares the specified model to a null model. The null model posits that there are no associations among the variables. CFI ranges from 0 to 1, with higher values indicating better fit. CFI values > .90 are generally interpreted as acceptable model fit. Models were only interpreted if they had acceptable fit across all 3 indices. Separate models were run for internalizing, externalizing, and total behavior problems across: (a) the overall sample and (b) for boys and girls. The base specified model is depicted in Fig. 1. Effortful control and behavior problems were assessed as latent constructs (as detailed above). To aid in the interpretation of EC, and because our hypotheses focused on contrasting children with high and low EC, this latent construct was divided into three groups: children with low EC (z-score below −.25), average EC, and high EC (z-scores above .25) (scored as 1, 2, and 3 respectively). Our final research question focused on family SES factors; therefore, each model was rerun with the addition of family SES assets. In sum, four models were calculated for internalizing, externalizing, and total behavior problems: (a) the base model ( Fig. 1), (b) the base model with the addition of family SES, (c) the base model across gender and (d) the gender model with the addition of family SES. The sections below detail the significant standardized path coefficients for each model (summarized in Table 3). The base specified model for effortfull control (EC), behavior problems, as ... Fig. 1. The base specified model for effortfull control (EC), behavior problems, as indexed by the Child Behavior Checklist (CBCL), and positive emotional expression. Additional models included estimates across gender and the addition of family socioeconomic assets. Figure options Table 3. Structural equation model estimates for total, internalizing, and externalizing behavior problems. Model Total behavior problems Internalizing behavior problems Externalizing behavior problems Base 2a 3b 4c Base 2a 3b 4c Base 2a 3b 4c M F M F M F M F M F M F Behav prob by 24 month CBCL .86** .85** .91** .83** .89** .83** .87** .90** .99** .81** .99** .84** .80** .89** .83** .85** .89** .93** Behav prob by 36 month CBCL .90** .90** .86** .92** .87** .92** .82** .80** .76** .86** .75** .82** .92** .83** .85** .90** .80** .83** Behav prob on Infant prematurity .10 .14 .06 .14 .12 .15 .10 .13 .01 .17 .05 .18 .13 .16*** .19 .09 .25*** .09 Gender .08 .03 .18*** .14 .02 −.03 Family SES −.39** −.34* −.42** −.32** −.26* −.36* −.36** −.36** −.37** Pos. emotion on Infant prematurity .07 .06 .05 .09 .04 .09 .07 .06 .05 .09 .04 .09 .07 .06 .05 .09 .04 .09 Gender −.13 −.13 −.13 −.13 −.13 −.13 Family SES .06 .03 .11 .06 .03 .11 .06 .03 .11 EC groups on Infant prematurity .20*** .17 .29*** .12 .28 .09 .20*** .17 .29*** .12 .28 .09 .20*** .17 .29*** .12 .28 .09 Gender .14 .15 .14 .15 .14 .15 Family SES .26* −.09 .44** .26* .09 .44** .26* .09 .44** EC groups with behav prob −.22* −.13 −.19 −.24 −.15 −.08 −.17 −.09 −.10 −.19 −.07 −.06 −.17 −.08 −.17 −.19 −.12 −.03 Pos. emotion with behav prob .04 .07 .09 −.03 .10 .03 −.02 .00 −.02 −.01 −.02 .04 .09 .13 .14 .05 .16 .11 Pos. emotion with EC groups −.10 −.12 −.32* .16 −.32* .14 −.10 −.12 −.32* .16 −.32* .14 −.10 −.12 −.32* .16 −.32* .14 a The base model with family socioeconomic assets (SES) added. b The base model estimated across gender. c The gender model with family SES added, M = male, F = female, behav prob = behavioral problems latent construct, CBCL = Child Behavior Checklist, pos. emotion = positive emotion expression composite, EC groups = low, average, and high effortful control groups. * p < .05. ** p < .01 *** p < .07. Table options 3.2.1. Total behavior problems Within the base model, children with high EC had fewer total behavior problems (z = −1.94, p = .05; β = −.22). Also within this model, a modest association between infant prematurity and EC emerged (z = −1.84, p = .06; β = .20). Infants within the low EC group tended to be born earlier and weigh less at birth. Infant prematurity was not associated with any of the other variables of interest. However, the addition of family SES to the base model attenuated the association between EC and total behavior problems (z = −1.20, p = .23, β = −.13). Family SES was a robust predictor of EC and behavior problems ( Table 3, Total Behavior Problems Model 2a). Children with fewer SES assets had elevated parent-reported behavior problems and were more likely to be in the low EC group. Family SES did not predict positive emotion expression. Additionally, positive emotion expression was not associated with EC or total behavior problems in the SES model. Within the gender model (Table 3, Total Behavior Problems Model 3b), positive emotion expression was associated with EC for boys (z = −2.36, p < .05; β = −.32) but not girls (z = 1.17, p = .24; β = .16). Boys with high EC exhibited fewer positive emotion expressions. In addition, a small association between infant prematurity and EC was apparent for boys (z = 1.79, p = .07; β = .29) but not for girls (z = .74, p = .46; β = .12). Boys within the low EC group tended to be born earlier and weigh less than boys with within the high EC group. All other tested associations were consistent across boys and girls. The addition of family SES ( Table 3, Total Behavior Problems Model 4c) to the gender model revealed a robust association between SES and EC for girls (z = 2.79, p < .01, β = .44) but not for boys (z = .62, p = .53, β = −.09). Girls within the high EC group had more socioeconomic assets than girls in the low EC group. 3.2.2. Internalizing behavior problems The base model contained a moderate association between gender and internalizing behavior problems (z = 1.88, p = .06, β = .18). Consistent with previous research, parents of girls tended to report higher internalizing problems than parents of boys. No additional paths of interest (e.g., EC with internalizing behavior problems) were significant ( Table 3, Internalizing Behavior Problems Base Model). The addition of family SES to the base model ( Table 3, Internalizing Behavior Problems Model 2a) revealed a robust association between family SES and internalizing problems (z = −3.48, p < .01, β = −.32). Children from homes with more SES assets had fewer internalizing problems. As reported above, family SES continued to predict EC. Within the gender models, family SES predicted internalizing behavior problems for both boys (z = −2.13, p < .05, β = −.26) and girls (z = −2.57, p < .05, β = −.36). Children from families with more socioeconomic assets had fewer parent-reported internalizing problems. As noted above, high EC was associated with lower levels of positive emotion expression for boys but not girls, and more family SES assets were associated with high EC for girls but not boys. 3.2.3. Externalizing behavior problems In the base model, externalizing behavior problems were not associated with EC or positive emotion expression. The addition of family SES revealed a robust association between family SES and externalizing behavior problems (z = −3.94, p < .01, β = −.36), and a small association between infant prematurity and externalizing behavior problems (z = 1.85, p = .06, β = .16). Infants born later and heavier tended to have higher parent-reported externalizing behavior problems. As described above, more family SES assets continued to predict higher EC. The gender models for externalizing behavior problems echoed those of the total behavior problem models with one exception. For boys, infant prematurity was associated with parent-reports of externalizing behavior problems (z = 1.90, p = .06, β = .25). Boys who were born later and heavier had higher externalizing behavior problems. This was not true for girls ( Table 3, Externalizing Behavior Problems Models 4c).