Background

1. THE UNIVERSITY of WISCONSIN MADISON Socio-demographic Risks 24 month Self-Regulation, Behavior Problems and Cognitive Development 9 monthParenting 9 MonthTemperament Infant Birthweight (Control) Maternal Depression (Control) Bayley Scales of Infant Dev. (Control) Infant Gender (Control) • Emerging Self-Regulation in Toddlers Born Preterm: Parenting, Temperament or Differential Susceptibility? • Julie Poehlmann*, A.J. Schwichtenberg*, Rebecca Shlafert, Emily Hahnŧ, Jon-Paul Bianchi*, Rachael Warnerŧ, Brianne Friberg* • *Department of Human Development and Family Studies, University of Wisconsin • t Department of Child Psychology, University of Minnesota • ŧ School of Social Work, University of Wisconsin • Outcome Variables • Children’s Behavior Problems. The Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2000) was used to assess children’s behavior problems at 24 months postterm. The CBCL is a standardized behavior rating scale that is completed by an adult with whom the child lives. The preschool forms list 99 problem behaviors. Mothers rated each problem behavior on a three point scale in reference to the child’s behaviors that occurred during the past two months. Responses were then summed to obtain scores for Internalizing and Externalizing Problem scales that were converted into T-scores on the basis of normative data (Achenbach & Rescorla, 2000). • Self-Regulation: Effortful Control Paradigm. Four components of toddler self-regulation were assessed using 5 tasks adapted from Kochanska et al.’s (2000) effortful control paradigm. • Delay (Snack Delay and Gift-in-Bag) • Effortful attention (Shapes) • Slowing motor activity (Walk-a-Line-Slowly) • Suppressing-initiating activity to signal (Towers) • The Snack Delay task included 4 trials, with successive trials each having longer delay times. Delay to touch the cup (in seconds) was coded for each trial and averaged across all four trials. Longer delay times on this task represent more regulation. Delay tasks were coded by two trained students, with 100% reliability within 1 second. For the Gift-in-Bag task, categorical codes were assigned ranging from 1 (child pulls gift from bag) to 5 (child waited entire 2 minutes without touching the bag). • For the Shapes task, the child’s responses to imbedded animal cards were coded. Children’s answers were scored as incorrect, self-correction, and correct. Higher scores indicated more regulation. • In the Walk-a-line task codes were the average number of seconds the child took to walk a line (across two trials), when asked to walk slowly after a physically exhilarating task. • Finally, the Towers task was coded as a function of the average number of turns the child allowed the experimenter to take. Descriptive statistics and correlations between tasks are shown in Tables 3 and 4. • Hierarchical regression analyses were utilized to test the three competing models. In all analyses, infant birthweight and gender, family sociodemographic risks, maternal depressive symptoms, and infant Bayley MDI scores were entered as controls in the first step. PCERA parenting quality factors and temperament dimensions were entered in the second step (separate analyses for each factor and dimension), and interaction terms were entered in the third step. Partial support for each model emerged. • Model 1: Parenting • In our analyses assessing Model 1, significant direct associations between parenting and 24 month outcomes were identified for PCERA factors 2 and 3, but not factor 1. Infants who experienced more sensitive and less anxious, intrusive, and rigid play interactions with their mothers at 9 months showed longer delay times (B = .28, p < .05), more turn taking (suppressing activity to signal; B = .22, p < .05), and fewer externalizing problems at 24 months (B = -.22, p < .05). In addition, infants who experienced more angry, critical interactions with mothers at 9 months showed less turn taking (B = .24, p < .05), more externalizing behaviors (B = -.22, p < .05), and slightly shorter delay times at 24 months (B = .18, p < .10). However, PCERA parenting scores were not associated with children’s 24 month internalizing problems, slowing motor activity, IQ scores, or attention regulation. • Model 2: Temperament • In our analyses assessing Model 2, we found several significant direct associations between infant temperament characteristics and 24 month outcomes. Infants rated as high in approach at 9 months showed longer delay times (B = .21, p < .05), but more internalizing (B = .37, p < .01) and externalizing (B = .20, p < .05) behavior problems at 24 months. Infants rated lower in distractibility at 9 months engaged in more turn taking (B = -.25, p < .05), had fewer externalizing problems (B = .19, p < .05), and slightly fewer internalizing problems (B = .17, p < .10) at 24 months. Finally, infants rated as higher in activity at 9 months showed significantly more 24 month externalizing problems (B = .28, p < .01), slightly more internalizing problems (B = .15, p < .10) and slightly lower IQ scores (B = -.15, p < .10). Infant temperament did not show direct associations with 24 month attention regulation or motor slowing. • Model 3: Differential Susceptibility • Effortful Attention. Infants with higher birthweights, girls, children from families experiencing fewer sociodemographic risks, and children with higher 16 month Bayley MDI scores performed better on the attention regulation task. In addition, 3 significant and trend level Parenting X Temperament interactions emerged. PCERA Parent Factor 2 interacted with Distractibility, F(8, 99) = 4.79, p <.01, R2 = .28, and Approach, F(8, 99) = 4.05, p < .01, R2 = .25 (Table 5), and PCERA Parent Factor 3 interacted with Distractibility, F(8, 99) = 4.52, p < .01, R2 =.27. • Table 7. Stepwise regression analyses for the Externalizing Behavior outcome with Activity and PCERA Factor 3 (Angry, Critical Parenting). • Internalizing Behaviors. Girls and children of mothers reporting more depressive symptoms at 9 months exhibited more internalizing behaviors at 24 months. Contrary to our model, there were no significant interactions between the PCERA parent factors and temperament dimensions. • Stanford-Binet (5th ed.). Children of mothers with higher receptive vocabulary scores and children who performed better on the 16 month Bayley had higher Stanford Binet scores. In addition, there was a significant PCERA Parent Factor 1 X Activity interaction. However, post-hoc testing revealed that the Stanford Binet scores of children in the high and low activity groups did not differ on the basis of Parent Factor 1 being high or low. • Our results provided partial support for each model in this sample of high risk infants, depending on the outcomes and predictors assessed. • Model 1 – Parenting • For the parenting model, we found that PT LBW infants whose mothers interacted with them in angry or anxious, intrusive, rigid, or insensitive ways showed shorter delay times, less turn taking, and more externalizing problems at 24 months. • Model 2 – Temperament • For the temperament model, we found that PT LBW infants rated higher in 9 month activity exhibited more behavior problems and slightly less optimal cognitive development at 24 months. In addition, infants who were rated as more distractible at 9 months showed less turn taking and more behavior problems at 24 months. Finally, infants rated higher in approach at 9 months showed longer delay times but more behavior problems at 24 months. • Model 3 – Differential Susceptibility • The differential susceptibility model was partially supported for several self-regulation outcomes at 24 months. • Infants rated as highly distractible at 9 months performed worse on the 24 month effortful attention task when they experienced less sensitive and more anxious, intrusive, rigid or angry/critical interactions with their mothers at 9 months, whereas this was not true for infants who were rated low in distractibility. • Infants rated high in approach at 9 months also performed significantly worse on the 24 month effortful attention task when they experienced less sensitive and more anxious, intrusive, and rigid interactions with their mothers at 9 months, whereas this was not true for infants rated low in approach. In contrast, infants rated low in approach at 9 months engaged in slightly more 24 month turn taking when they experienced less angry, critical interactions at 9 months, whereas there was no difference for infants rated high in approach. • Infants rated low in rhythmicity performed slightly worse on the 24 month slowing motor activity task when maternal interactions were less sensitive, more intrusive and anxious, whereas infants rated high in rhythmicity group did not show a difference. • Infants rated low in activity at 9 months exhibited more 24 month externalizing behavior problems when mothers engaged in angry, critical interactions with them, whereas this style of parenting was not related to children’s behavior problems in the high activity group. • Overall, our results suggest that early maternal interaction behaviors characterized by anger or anxiety, intrusiveness, or lack of sensitivity are associated with less optimal toddler self-regulation in this group of high risk infants, even when controlling for degree of infant prematurity, family socioeconomic risks, and maternal depressive symptoms. Moreover, infants who are perceived as having more difficult temperaments (such as high distractibility, low rhythmicity, low approach) appear more vulnerable to these types of negative interactions for several outcomes. This information could be used to screen high risk infants for infant temperament and maternal interaction characteristics to determine problems with “goodness of fit” that may be associated with less optimal toddler outcomes. Then the highest risk dyads could be targeted for preventive interventions. Because each self-regulation outcome was predicted by different variables (and the outcomes were not highly correlated with each other), self-regulation should not be seen as a unitary construct in PT LBW infants. • More than 500,000 babies were born premature in the United States in 2004, and the rate of infants born preterm increased by 14 percent from 1994 to 2004 (March of Dimes, 2006). Preterm low birthweight infants are at risk for compromised development, including self-regulation and behavior problems (e.g., Blair, 2002; Robison & Gonzalez, 1999). As a result, it is imperative that researchers investigate the underlying mechanisms that account for the association between preterm birth and child outcomes to inform preventive interventions. The current study investigated three competing theoretical models of causal pathways leading from child and family risks (i.e., prematurity and socio-demographic risks) to measures of toddler self-regulation (i.e., effortful control and behavior problems). • Transactional development theory (Sameroff & Fiese, 2000) has encouraged researchers to examine the effects of cumulative risk and bidirectional processes that occur between parents and children. On the basis of this theory, we created prematurity and sociodemographic risk indices. Moreover, in light of attachment theory, we proposed a model testing whether parent-child interaction quality mediated the relation between cumulative risks and toddler self-regulation. Researchers who emphasize the importance of child factors, however, contend that temperament is a salient variable that predicts self-regulation and behavior problems (e.g., Rothbart, Posner, & Hershey, 1995). Utilizing this perspective, we posited a second model in which temperament was examined as a mediator of the relation between infant risk and toddler self-regulation. Finally, Belsky’s (2005) model of differential susceptibility, in which child temperament moderates the relationship between parenting quality and children’s social competence, was used to create a third model. In this model, we proposed that temperament would moderate the relation between parent-child interaction quality and toddler self-regulation (Figure 1). Children with difficult temperamental qualities were expected to be more susceptible to parenting influences. • Figure 1. Differential susceptibility model. • Model 1 – Parenting: For infants born preterm or low birthweight, does early parenting quality predict toddler self-regulation, behavior problems and cognitive development? • Model 2 – Temperament: For infants born preterm or low birthweight, do infant temperament characteristics predict toddler self-regulation, behavior problems and cognitive development? • Model 3 – Differential Susceptibility: For infants born preterm or low birthweight, does infant temperament moderate the association between early parenting quality and toddler self-regulation, behavior problems and cognitive development? • This report focused on a subset of infant-mother dyads who are part of a larger longitudinal study. Data were collected at hospital discharge, 9-, 16- and 24-months postterm. Infant birthweight, gender and family sociodemographic characteristics were collected at hospital discharge. At 9 months, infant-mother dyads were videotaped and later coded for parenting quality, and mothers completed a depression screening and infant temperament questionnaire. At 16 months, dyads completed a laboratory visit which included the Bayley Scales of Infant Development. The outcomes of interest (toddler self-regulation, behavior problems, and cognitive development) were assessed at 24 months via Kochanska et al.’s (2000) Effortful Control paradigm, the Child Behavior Checklist, and the Stanford-Binet 5th Edition. Background For this report, data were drawn from 118 families recruited from 3 Wisconsin neonatal intensive care units; however, because of missing data, the sample sizes for analyses ranged from 76 to 118 (n = 76 for most analyses). Infant birth weights ranged from 490g to 2802g with an average birthweight of 1701g (SD = 598). Infant gestational age ranged from 25 to 37 weeks with an average of 31 weeks (SD = 3.25). Sixty-one (80.2%) of the infants were Caucasian, five (6.5%) were African American, one (1.3%) was Asian, one was Latino (1.3%), and eight (10.5%) were more than one ethnicity. Additional sample descriptive statistics are provided in Table 1. Table 1. Sample sociodemographic characteristics (N = 76). SD Temperament Control Variables Bayley Scales for Infant Development (BSID II) – Mental Development Index (MDI). At 16 months postterm, toddlers completed the BSID II, which is a widely used standardized test of infant cognitive skills appropriate for preterm LBW and VLBW infants. It includes items relating to achievement of developmental milestones such as object permanence, problem solving, fine motor skills, and receptive and expressive language abilities. Average MDI scores on the BSID II range from 85 to 115. In our sample MDI scores ranged from 55 to 122 (M = 86.37, SD = 10.76). Center for Epidemiological Studies - Depression Scale (CES-D).Maternal depressive symptoms were assessed at 9 months using the CES-D (Radloff, 1977). The CES-D includes 20 statements that assess depressive symptoms during the past week across seven main areas: sleep disturbance, loss of appetite, psychomotor retardation, hopelessness, helplessness, guilt and depressed mood. CES-D scores ranged from 0 to 43 (M = 9.46, SD = 8.08) and Cronbach’s alpha for the entire measure was .88. Family Sociodemographic Risks. Mothers completed a demographic questionnaire at hospital discharge. An 8 item sociodemographic risk index was created, with one point given for: family income below federal poverty guidelines, both parents unemployed, family using public assistance, single mother, teen mother, 4 or more children in home, and mother or father with less than a high school education. The scores for this index ranged from 0 to 5 with an average of 0.72 (SD = 1.37).Cronbach’s alpha for the sociodemographic index was .75. Predictor Variables Parenting Quality. The Parent Child Early Relational Assessment (PCERA; Clark, 1985) was used to assess parenting quality during 5 minutes of free play in the home when infants were 9 months postterm. The PCERA consists of 65 variables that are rated on a 5 point scale. Inter-rater reliability for this sample ranged from 83% to 97% agreement, with a mean of 88% agreement with a master coder. Initially, we performed a confirmatory factor analysis on maternal variables on the basis of Clark et al.’s (1997) findings focusing on 9 month PCERA free-play interactions. Because the chi square statistics for this analysis indicated poor fit with the data from the present sample, we performed an exploratory factor analysis on parent variables, retaining variables that loaded at .60 or above (Clark, 1985). The parenting factors were named Maternal Affect, Verbal, and Social Connection (Factor 1; α = .95), Anxious, Intrusive, Insensitive, and Rigid Parenting (Factor 2; α = .90), and Angry, Critical Parenting (Factor 3; α = .89). Correlations among the maternal scales ranged from .59 to .82. Higher scores on each factor reflected more optimal parenting. Infant Temperament.Maternal perceptions of infant temperament were assessed at 9 months using the Revised Infant Temperament Questionnaire (RITQ; Carey & McDevitt, 1977). The RITQ contains 95 items that are scored on a 6-point scale and combined to represent nine dimensions. However, only 4 of these dimensions had acceptable internal consistency (see Table 2). Dimension correlations are providedin Table 4. Table 2. Descriptive statistics for PCERA maternal factors (N = 98). Sample • Table 5. Stepwiseregression analyses for the effortful attention outcome with Approach and PCERA Factor 2 (Anxious, Intrusive, Insensitive and Rigid Parenting). • Post-hoc analyses were conducted to tease apart the significant interaction terms (Aiken & West, 1991). Simple regressions were conducted to examine the influence of PCERA parenting factors at high and low levels of each temperament dimensions for the effortful attention outcome. • Analyses revealed that children in the high distractibility group did significantly better on the effortful attention task when mothers interacted with their infants in sensitive, nonintrusive ways. In contrast, children in the low distractibility group showed slightly better attention regulation when mothers showed less sensitivity, more intrusiveness and anxiety during interactions (F change = .09). • We also found that children in the high approach group performed significantly better on the effortful attention task when they experienced more sensitive and less intrusive parenting (F change = .04). In contrast, children in the low approach group did not show differences on the attention regulation task on the basis of maternal sensitivity or intrusiveness (see Figure 2). • Children in the high distractibility group performed slightly worse on the effortful attention task when mothers engaged in angry or critical interactions (F change = .07), whereas children in the low distractibility group did not show a difference in their attention regulation when parenting quality varied. • Figure 2. Interaction between Approach and PCERA Parent Factor 2 for the effortful attention outcome. • Suppressing-initiating activity to signal. None of the controls predicted children’s 24 month turn taking. However, PCERA Parent Factor 3 interacted with Approach at the trend level, F(8, 101) = 1.63, p = .12, R2 = .12 (see Table 6). • Table 6. Stepwise regression analyses for the suppressing-initiating activity to signal outcome with Approach and PCERA Factor 3 (Angry, Critical Parenting). • Post-hoc analyses revealed that children in the low approach group engaged in slightly more turn taking when mothers did not engage in angry, critical interactions (F change = .06). For the high approach group, parenting was not a significant predictor of turn taking (see Figure 3). • Figure 3. Interaction between PCERA factor 3 (Angry, Critical Parenting) and Approach for suppressing initiating activity to a signal (Tower Task). • Slowing Motor Activity. Although none of the controls were significant, the PCERA Parent Factor 2 X Rhythmicity interaction was a significant predictor of slowing motor activity, F(8, 98) =2.55 , p < .05, R2 =.17. Post-hoc analyses revealed that children in the low rhythmicity group did slightly better on the slowing motor activity task when mothers engaged in more sensitive and less anxious-intrusive interactions (F change = .09). For children in the high rhythmicity group, parenting was not a significant predictor. • Delay. Children with higher 16 month Bayley MDI scores demonstrated longer delay times. Contrary to our model, no significant Parenting X Temperament interactions emerged for the delay tasks. • Externalizing Behavior Problems. Children whose mothers reported more depressive symptoms at 9 months exhibited more externalizing behavior problems at 24 months. In addition, a significant PCERA Parent Factor 3 X Activity interaction emerged, F(8, 112) = 4.94, p < .01, R2 = .26 (see Table 7). Summary Measures Results Research Questions Method Discussion This research was supported by grants from the National Institutes of Health and the University of Wisconsin-Madison.