Risk factors for delayed social‐emotional development and behavior problems at age two: Results from the All Our Babies/Families (AOB/F) cohort

Abstract Background and Aims Social‐emotional delays and behavioral problems at preschool age are associated with negative outcomes at school age, including ongoing behavior problems, poorer social functioning, and academic difficulties. Understanding modifiable risk factors for suboptimal development requires consideration of contemporary family circumstances to determine areas for effective early intervention to optimize development. This study aimed to identify risk factors for delayed social‐emotional development and behavior problems at age two among participants of the All Our Babies/Families cohort study. Methods Mothers (N = 1596) completed five comprehensive questionnaires spanning midpregnancy to 2 years postpartum. At child age two, behavior and competence outcomes were measured using the Brief Infant‐Toddler Social and Emotional Assessment. Chi square analysis and multivariable logistic regression modeling was used to identify key risk factors for suboptimal child outcomes. Predicted probabilities for adverse outcomes in the presence of risk were calculated. Results Risk factors for possible delayed social‐emotional development in children included maternal depression at 2 years postpartum (OR 2.46, 95% CI 1.63, 3.72), lower parenting self‐efficacy at 2 years postpartum (OR 2.76, 95% CI 1.51, 5.06), non‐daily play‐based interaction when child was 1 and 2 years old (OR 1.43, 95% CI 1.02, 1.99), child delayed sleep initiation at 2 years of age (OR 1.58, 95% CI 1.05, 2.37), and playgroup non‐attendance between 1 and 2 years postpartum (OR 1.43, 95% CI 1.03, 1.99). Risk factors for possible behavior problems included lower maternal optimism during pregnancy (OR 2.02, 95% CI 1.36, 2.99), maternal depression at 2 years postpartum (OR 2.19, 95% CI 1.46, 3.27), difficulty balancing responsibilities at 2 years postpartum (OR 2.32 95% CI 1.55, 3.47), child second language exposure at 2 years of age (OR 1.88, 95% CI 1.37, 2.58), child delayed sleep initiation at 2 years of age (OR 1.55 95% CI 1.06, 2.26), child frequent night wakings at 2 years of age (OR 2.95 95% CI 2.13, 4.10), and more screentime exposure at 2 years of age (OR 1.85 95% CI 1.34, 2.54). Conclusions This study suggests that addressing maternal mental health and promoting parenting strategies that encourage play‐based interaction, limiting screen time, preventing sleep problems, and engagement in informal playgroups would reduce the risk of behavior and social‐emotional problems.

ents and care providers to implement strategies to remediate problems before school entry. 10 Understanding the factors that contribute to the development of poor social-emotional competencies and behavioral problems in the first few years of life will help mitigate long-term adverse consequences and trajectories associated with suboptimal behavior. From a policy, program, and return on investment perspective, early interventions are cost-effective and have greater positive impact than later remediation. 11,12 Despite evidence about the importance of early identification of children with delayed social-emotional development and behavior problems, 13,14 there is limited Canadian research about family and community factors that may prevent poor outcomes within the context of pregnancy, and the early postpartum period. Assessing a breadth of risk factors helps inform independent effects and potential targets for early intervention. In addition, contemporary information is necessary to determine the magnitude of influence these factors have, within the current social context of dual parent working families, child care, screen time, and time stresses.
The objective of this study was to identify key risk factors for delayed social-emotional development and behavioral problems among 2-year-old children in a community sample of mothers and their children. Data were analyzed from the All Our Babies/Families (AOB/F) cohort study, 15 which contains prospectively collected information on maternal, family, child care, parenting, and community factors related to development. This population-based cohort of women, recruited in Calgary, Canada, is demographically similar to the pregnant and parenting populations of westernized urban centers. 15 These findings will inform and focus policy and programming to improve social-emotional and behavior development in the preschool years.

| Study design
The AOB/F study is an ongoing prospective population-based pregnancy cohort study in Calgary, Canada, that began in 2008. 15 The AOB/F study was developed to investigate the relationships between the prenatal and early life period, and outcomes for infants, children, and mothers. Detailed descriptions of the AOB/F study design and methods have previously been described. 15 Briefly, the cohort was established using a community-based recruitment strategy, involving primary health care offices and a citywide laboratory service (Calgary Laboratory Service). Community posters were implemented to obtain a socioeconomically and ethnically diverse sample of women representative of the parenting population in an urban Canadian center.
Recruitment began in May 2008 and was completed in December 2010. Women were eligible to participate if they were less than 25-week gestation, 18 years of age or older, resided in the greater Calgary area, and were able to complete the written questionnaires in English. Prior to inclusion in the study, all participants provided informed consent. Participants were asked to complete three written questionnaires: (1) at study intake (second trimester), (2) during their third trimester, and (3) at 4 months postpartum. Information was collected on pregnancy history, demographics, lifestyle, health care utilization, physical and mental health, birth outcomes, child health, and parenting. Participants were also asked to consent to the research team accessing their obstetrical and birth records. In total, 3317 women consented to participate and completed at least one questionnaire. Women who consented to be contacted for future research were asked to participate in subsequent follow-up questionnaires when their child was 1 and 2 years old. The design of the 2-year follow-up questionnaire was delayed due to logistical constraints includ-

| Main outcomes
The Brief Infant-Toddler Social and Emotional Assessment (BITSEA) is a concise screening instrument to identify possible behavior problems and delayed social-emotional competencies in children aged 12 to 36 months. 16 It was designed to be completed by parents or caregivers and consists of 42 items. The BITSEA measures two domains, behavior problems (eg, aggression, defiance, over-activity, negative emotionality, anxiety, and withdrawal) and delayed social-emotional competencies (eg, empathy, pro-social behaviors, and compliance). 16 A total behavioral problem score and a total social-emotional competencies score are derived by summing relevant items for each domain subscale. Using the BITSEA standardized scoring cut-offs from the published normative data, children were categorized with possible behavioral problems if they scored at or above the 75 th percentile of normative data, and children were categorized with possible delayed social-emotional competencies if they scored at or below the 15 th percentile of normative data. 16 The BITSEA has shown excellent testretest reliability (behavior problems scale: r = 0.87; social-emotional competence scale: r = 0.85) and good interrater agreement between mothers and fathers (behavior problems scale: r = 0.68; social-emotional competence scale: r = 0.61). 17 Internal consistency measured by Cronbach's alpha was 0.79 for the behavior problems scale and 0.65 for the social-emotional competence scale. 17 Similarly, in the current sample, internal consistency was 0.75 for the behavior problem scale and 0.63 for the social-emotional competence scale. Given the large size of our sample, we used a parent report screening measure to identify children with symptoms of behavioral problems and social emotional difficulties. Screening tools are often used in populationbased studies due to their feasibility and ease of administration compared with more time-consuming and costly diagnostic interviews.
Of note, the use of delayed social-emotional development and behavior problems throughout the manuscript refer to "possible" delays and problems.

| Independent variables
Information on a broad range of potential risk factors collected during pregnancy and the first 2 years postpartum were examined in the current study and were grouped into four broad categories: (1)

| Data analysis
Descriptive statistics were used to describe participants socio-demographics, maternal mental health at 2 years postpartum, and child social-emotional and behavioral functioning at 2 years. Among those eligible for the 2-year questionnaire (n = 2114), socio-demographic characteristics, birth outcomes, and maternal mental health symptoms during pregnancy and at 1 year postpartum were compared between participants and non-participants of the 2-year questionnaire using Pearson's Chi-square Test. An initial bivariate analysis using Pearson's Chi-square test was conducted to identify risk factors for delayed social-emotional development and child behavioral problems at age two. Risk factors identified at the bivariate level based on statistical significance at the P < 0.10 level were considered for inclusion in the subsequent multivariable logistic regression models. A manual stepwise approach was used to build the multivariable models, with blocks of variables being added in sequential steps in the following order: (1) socio-demographic and adverse life event variables, (2) pregnancy and birth outcome variables, (3) 1-year postpartum variables, and (4) 2-year postpartum variables. To best address the research objectives, these blocks were ordered conceptually in temporal order and from least to most modifiable, as per our model building strategy used in previous research in the AOF. [26][27][28] Risk factors within each block were identified for inclusion based on the literature. Statistical significance was set at P < 0.05 for inclusion in the next step of the logistic regression modeling, and parsimonious models were developed at each block. As a final step, we entered each factor that was dropped at previous steps to ensure model robustness. Multivariable logistic regression results are presented as odds ratios (ORs) and 95% confidence intervals (CIs). Predicted probabilities for delayed social-emotional development and behavior problems were calculated for the combination of risk factors identified from the final models.
Stata Version 12.1 was used for the statistical analysis.

| Maternal participant characteristics
The characteristics of mothers who participated in the 2-year questionnaire (n = 1596) are described in Table 1. Compared with women who were eligible for the 2-year questionnaire but did not participate, women who participated in the 2-year questionnaire were more likely to have post-secondary education and household incomes greater than $80 000 CAD and be married or in a common law relationship and Caucasian (all P < 0.05). Two-year questionnaire participants were less likely to both experience symptoms of anxiety during pregnancy and have a preterm birth, than non-participants (all P < 0.05). There were no differences in maternal age, maternal country of birth, prepregnancy mental health history, depressive symptoms during pregnancy, child birthweight, and mental health symptoms at 1-year postpartum between participants and non-participants.

| Child development outcomes at 2 years
At questionnaire completion, children were an average age of 2.03 years (sd = 0.10, range = 1.88-2.84 years). Based on maternal report on the BITSEA, 13% (n = 210/1579) of children were identified with possible delayed social-emotional development and 15% (n = 236/1580) of children with possible behavior problems at age two (Table 1).

| Risk factors for delayed social-emotional development at age two
Bivariate associations between all potential risk factors examined and delayed social-emotional development at age two are summarized in  Table 3). The predicted probability of delayed social-emotional development at age 2 was 65% in the presence of all five risk factors.

| Risk factors for behavior problems at age two
Bivariate associations between all potential risk factors examined and child behavior problems at age two are summarized in Table 4. Multivariable modeling revealed the combination of risk factors most predictive of behavior problems at age two to be having a mother who

| DISCUSSION
Early behavior and social-emotional development is important for children's ongoing behavior, social competence, mental health, and academic success.
Utilizing data from the AOB/F cohort, we identified factors that were associated with delayed social-emotional development and behavior problems at age two, providing insight into beneficial targets for interventions.  Children who did not attend an informal playgroup were more likely to show delays in social-emotional competencies than children who attended informal playgroups. We posit that the impact of attending informal playgroups on child social-emotional development may be twofold: (1) they provide an opportunity for mothers to receive social support, previously identified as a protective factor for child development outcomes 33 and (2) they provide an opportunity for child socialization, which is associated with the development of social competence. 34 Parenting programs and public health professionals could inform parents about the potential benefits of informal playgroups, including opportunities for social support and child socialization, and how these can positively impact child development. Decision makers could encourage communities to provide spaces for parents to congregate to build social support networks, which would enhance well-being for children and families.
Mothers who reported lower levels of optimism during pregnancy were more likely to have a child with behavioral problems at 2 years of age than mothers with higher levels of optimism during pregnancy.
Optimistic individuals adapt to stressors more effectively, 35 and mothers who are optimistic may more easily adapt to their new role  as a parent and the associated challenges of raising a young child.
Optimism has been linked to positive parenting practices, including competence-promoting parenting practices and effective child management. 36 These findings linking maternal optimism to improved coping with stress and positive parenting practices may help to explain how maternal optimism may play a role in children's behavioral development. Although dispositional optimism is a relatively stable trait, research also indicates that, to some extent, optimism can be taught 37 and, therefore, parenting intervention and prevention programs for behavior development in young children could include content aimed at cultivating maternal optimism. 36 Mothers were asked to describe their ability to fulfill family, work, or other responsibilities (eg, volunteer work, household duties, and other children) since giving birth to their child. Mothers who reported "it is difficult most or all of the time" were more likely to have a child with behavior problems at 2 years of age compared with mothers who reported "it is never or sometimes difficult." This concept of challenges and conflict among women's life roles has been well documented in the literature 38 ; however, we did not find any previous studies that directly examined the relationship between balancing family, work, and life roles, and child development.
Mothers who experience more challenges balancing the conflicting demands from their family, work, and life are likely experiencing challenges with coping. Poor maternal coping has been found to be associated with poor child functioning, including behavior problems, poorer child self-regulation, and emotional difficulties. 39 Challenges with balancing roles is also associated with a number of indicators of maternal wellbeing, including increased rates of depression, stress, anxiety, and lower life satisfaction. 40 Supporting new parents with strategies to prevent parents from feeling overwhelmed with balancing their roles and responsibilities provides an opportunity to improve maternal wellbeing, and our results extend the current findings suggesting an opportunity to protect against early child behavioral problems as well. Having at least 1 hour of daily screen time on any type of media (television, DVDs, movies, computer, tablet) was associated with an increased risk of behavioral problems at age two. A longitudinal study has previously reported an association between early childhood television exposure and behavior problems, 41  Children who were exposed to a second language on a regular basis were more likely to have behavior problems at age two than children in single language home settings. The association between exposure to a second language and child behavior has not previously been reported. From the current study, this was a single question that did not ask about the number of hours children were exposed to a second language; therefore, more refined analyses are warranted to investigate this relationship. It is also possible that this variable is a proxy for something else, like cultural influences, which may influence parenting style, discipline strategies, and child behavior.
Although our analyses identified independent risk factors, it is likely that these factors cluster together, and if most or all present, compound the risk of suboptimal social-emotional competence and behavior; this is suggested by the high predicted probabilities for the outcomes when all risk factors are considered present. Further research could examine a risk profile for suboptimal child outcomes in early childhood, and intervention strategies that target a range of risk factors would be beneficial. For example, many of the factors associated with adverse outcomes in preschool children could be addressed through strategies that normalize help-seeking behavior for parents of newborns and encourage use of low-cost community resources. Many parenting programs address content related to efficacy, sleep, mental health, community resources, and child development. Health and child care providers could encourage parents to engage with existing supports and services, which may have the compound benefit of providing parents with skills and strategies for parenting, while creating supportive relationships. Our results also suggest that universal intervention strategies could target maternal depression in early childhood and child sleep hygiene, given that these risk factors were significant for both outcomes. Interaction factors and community engagement were specific to child social-emotional competencies, while maternal coping, and exposure to screens and a second language were specific to child behavior. Further research is warranted to confirm these factors, which would inform targeted interventions.

| Study limitations
Our outcome measure, the BITSEA, which provided measures of children's social-emotional competencies and behavior problems, was based on maternal report; it is a screening instrument, and, therefore, follow-up and more comprehensive evaluation is needed to rule in a diagnosis of a developmental delay or behavioral problem. Multi-informant information would be valuable for identification of at-risk Our analyses are restricted to identifying risk factors for the occurrence of social-emotional delays and behavior problems at 2 years of age. We are, therefore, unable to examine the extent to which these factors increase the risk for poor outcomes across time.
In addition, given that some single items in the BITSEA replicate risk factor constructs (eg, sleep problems) there may be a slight overestimation of the association between factors present at both the exposure and outcome level, and we acknowledge the potential bidirectional relationships between factors measured at the same time point (eg, child behavior and each of parental self-efficacy and worklife balance). Although a similar bi-directionality might exist between maternal depression and child outcomes, we also are ensuring that reporting bias is accounted for, given that mothers with depressive symptoms might overestimate problematic behavior in their children.
There is also potential for reporting bias among mothers with low optimism who may overestimate behavioral challenges in their children.
Finally, there is the potential for selection bias and limits to generalizability, given that participants who completed the 2-year follow-up differed from those eligible, in that they were more likely to have higher education and income, be married, and Caucasian. As such, caution is warranted in the interpretation of the results in the context of families that are more vulnerable.

| CONCLUSIONS
Our study extends prior research by analyzing data from a large-scale, community-based pregnancy cohort, the AOB/F study, which is representative of the parenting and child population in an urban center in Canada. The AOB/F study also provided the unique opportunity to comprehensively examine an extensive number of contemporary factors, as well as socio-demographic, gestational, and birth variables.
This analysis allowed for the evaluation of numerous factors simultaneously, and the predicted probability of poor outcomes for these exposures. A venue for further research in this cohort would be to examine longitudinal trajectories of children identified at risk, as well as mechanisms underlying associations across time and subgroups of families that would benefit from targeted intervention approaches.
Despite this comprehensive approach to considering a large number of risk factors from the prenatal, birth, early infancy, and early childhood periods, our results showed that in general, the factors most predictive of poor child behavior and social-emotional development co-occurred with developmental outcomes at age two. Risk factors co-occur in families and communities, and the high predicted probability of adverse outcomes associated with exposure to multiple risk factors highlights the need to identify families early and invest appropriately for optimal outcomes. 44 The importance of a diverse array of environmental factors can help guide strategies that normalize help seeking and uptake of parenting programs and engagement in community social support opportunities.
Identifying and providing support and intervention to mothers of young children experiencing poor mental health is critical to improving women's mental health and children's psychosocial development. Population-based strategies that support parents of young families in establishing healthy sleep habits, engaging in daily play, attending informal playgroups, and limiting screen time would enhance children's behavior and social-emotional development at age two.