Maternal Self-Efficacy is Protective for Child BMI (but not Mother) BMI among Mexican-Origin Children with Negative Temperament.

Abstract

Objective

Childhood obesity persists as a serious public health concern, particularly among Mexican-origin youth. Teen moms are also at heightened obesity risk. Multiple factors may exacerbate this risk, including stressors associated with parenting. Indeed, difficult child temperaments pose unique parenting challenges, which may also be linked to physical health outcomes in mothers. The purpose of this study was to examine whether the interaction between negative child temperament and parenting self-efficacy is related to the BMI of young children and their adolescent mothers, while controlling for important contextual factors. We also examined which pathways differed for girls versus boys.

Methods

Data were from a longitudinal study spanning five years that included 204 Mexican-origin young mothers and their children (with data collected at birth, age 4 years, and age 5 years). A multi-group structural equation modeling framework was used.

Results

Rate of early childhood obesity was low in comparison to national averages, whereas rate of adolescent mother obesity was notably higher than the national average. Negative child temperament was associated with higher child BMI among those adolescent mothers with low parenting self-efficacy. Among the children with a negative temperament, their mothers’ high parenting self-efficacy may have served as a protective factor against unhealthy child BMI. This significant interaction held for both boys and girls.

Conclusions

Research evaluating the potential effectiveness of interventions that promote parenting self-efficacy during early childhood as a means to reduce rates of obesity among children of adolescent mothers should be conducted.

Introduction

Obesity among children and adolescents persists as a serious public health concern. The current obesity epidemic places one in three children as either overweight or obese and this risk increases markedly if both parents are also overweight or obese.¹ Among adolescents, Latino males and females are significantly more likely to be overweight or obese than their White counterparts² and that same trend holds for younger children as well.²

United States (U.S.)-born children of Mexican-origin have long been at particular risk for having unhealthy BMI status and remain at-risk for other poor health outcomes.³ Exacerbating this is adolescent parenthood, which also places both adolescent mothers and their children at risk for an unhealthy BMI and related chronic health conditions. Indeed, research by Chang and colleagues (2013) demonstrated that teen pregnancy is associated with an increased risk of obesity for the mother.⁴ Further, Latina teens have the highest incidence of teen pregnancies⁵ and Mexican-origin adolescents show the highest rates among all Latina adolescents. Therefore, the need to more clearly understand factors that increase risk for unhealthy BMIs among Mexican-origin adolescent mothers and their children is crucial. Indeed, this is a high-risk subgroup that critically needs our attention. Such efforts would likely help inform our understanding of how to more effectively prevent obesity among Mexican-origin adolescent mothers and their children.

Theoretical Framework

Masten and Coatsworth (1998) offer a useful framework for understanding risk and resilience processes in human development.⁶ Their framework underscores the role of core regulatory and social-emotional skills as necessary precursors to adaptive behaviors and corresponding long-term positive health outcomes in youth.⁶ Broadly speaking, such resilience factors are believed to offset risk. In line with broader public health models on well-being,⁷ Masten and Coatsworth’s work also emphasizes the importance of preventative efforts as protective processes. As such, we expect that protective factors may counter the onset of both poor developmental and health outcomes.⁷ In addition, it is possible that poor developmental outcomes may serve as possible risk factors for health outcomes. As an example, and in this study, child temperamental negativity was examined as a risk factor for children’s and their mother’s BMI.

Child Temperamental Negativity as a Risk Factor

Child temperament refers to relatively stable individual differences in reactivity and regulatory responses.⁸ Research has consistently demonstrated that young children with more negative or difficult temperaments are at risk for poorer behavioral adjustment, particularly in the context of inadequate parenting.⁹ Links between aspects of children’s temperaments and their own weight outcomes have also been identified.¹⁰ For example, research has confirmed the relationship between increased rates of negative reactivity during early childhood and obesity risk, although, the longitudinal relationship in this previous work is unclear.⁹

The types of challenging behaviors during early childhood that are associated with temperamental negativity have also been linked to parents’ health status. For example, parents of children with behavioral challenges have been shown to report significantly poorer physical health¹¹ and to also have dysregulated patterns of the stress hormone, cortisol,¹⁰ which has been linked to weight gain, particularly among women.¹² Given the concerns regarding risk of obesity among teen mothers,⁴ we consider child temperamental negativity as a possible risk factor in the prediction of both mothers’ and children’s BMI.

Parenting Self-Efficacy as a Protective Factor

Despite findings indicating a relation between negative child temperament and child BMI, the few studies that have tested longitudinal relations have revealed more mixed findings.¹⁰ Moderating factors, such as parenting practices, may play a role in explaining these mixed results.¹⁰ There is also a significant body of research linking children’s temperamental negativity with poorer parenting practices, and many studies have additionally demonstrated that the influence of child temperament on developmental outcomes can be moderated by characteristics of the parent.⁸

The unique and challenging circumstances of adolescent motherhood make it particularly important to understand how their readiness to parent may interact with child temperament to also inform their children’s health. There is increasing evidence of the importance of perceived parental efficacy in understanding adolescent mothers’ readiness to parent. Parental self-efficacy¹³ has been significantly linked to parenting behaviors and, in turn, child health and developmental outcomes.¹⁴ Scholars argue that satisfaction and confidence in one’s role as a parent is connected to proficiency¹⁵ and there is evidence linking perceived self-efficacy with positive parenting skills.¹⁵

Importantly, parental efficacy has also emerged as relevant to the feeding practices of parents of young children. For example, research has found that parents with a greater likelihood to use food to soothe their fussy children report significantly lower levels of parenting self-efficacy.¹⁶ Moreover, the use of food to soothe has been positively related to children’s weight status, and this link is strongest for children with more negative temperaments.¹⁶

In addition, previous research has also highlighted the specific challenges Mexican-origin adolescent mothers face in the context of parenting. First, it is important to note that Mexican-origin adolescent females have the highest rates of teenage pregnancy.¹⁷ In addition, previous work has noted an increased risk for mental health problems among adolescent mothers likely due to complexities of the non-normative transition to parenthood during adolescence.¹⁸ Furthermore, scholars have highlighted the complex implications of teenage parenthood for Latina adolescent mothers’ family dynamics.¹⁹ Thus, Mexican-origin adolescent mothers represent an important population on which to focus, particularly with respect to identifying potential protective factors that can support positive development and adjustment among these young mothers and their children. Therefore, and in this study, parenting self-efficacy is examined as a potential moderator of the link between child negative temperament and child BMI.

Child Gende

Differences by gender in terms of both child physical (e.g. BMI) and developmental (e.g. temperament) outcomes have been well established in the literature, and boys appear to be at particular risk. Globally, the prevalence of obesity is higher among boys than girls.^20–21 At the same time, work by Ogden and colleagues²² has identified that significant increases in BMI over time are observed among teen boys, but not teen girls.²² In terms of child temperamental characteristics and the link to parenting, there is also evidence suggesting possible gender differences. For example, scholars argue that parents tend to encourage girls to express fewer externalizing emotional expressions (e.g., anger) and to be more supportive of girls’ expressions of internalized emotions (e.g., sadness).²³

OBJECTIVES

The purpose of this study is therefore to identify how negative child temperament and parenting self-efficacy interact to predict the BMI of i) young children (0–5 years) and ii) their adolescent mothers, while controlling for important contextual factors, including the child’s low birthweight status, nativity, and family income. These contextual factors have been selected knowing that there is an established relationship between low birthweight and children’s subsequent health,²⁴ and between both family income²⁵ and nativity status²⁶ and the health outcomes of Latino children and their mothers.²⁶ Further and because of the known gender differences in health outcomes,¹⁷ this study also aims to examine the aforementioned pathways for girls and boys.

Research Questions

This work specifically seeks to answer the following research questions:

1. What are the rates of overweight and obesity among the children and their mothers in this sample of Mexican-origin adolescent mothers and their young children?

2. Is negative child temperament associated with BMI in either the child and/or their mother? Further, is there a difference in associations by child gender?

3. Does parenting self-efficacy moderate the relationship between negative child temperament and BMI in the child? Further, is there a difference in relationship by child gender?

METHODS

Sample

Data were from a longitudinal study that included 204 Mexican-origin young mothers, their mother figures, and their children.42 The study participants resided in a metropolitan city in the southwestern region of the United States. Families were recruited when adolescents were in their third trimester of pregnancy, and data were subsequently collected annually for five additional years. Data from the present study include demographic data collected during the third trimester of pregnancy, infant birth weight status collected approximately two weeks after birth, and key study variables collected when children were four years of age, and five years of age. The majority of families participated across all six data collection waves (i.e., 88.24% when their child was four years old, and 87.75% when their child was five years old). Mothers were, on average, 16.80 years of age (SD = 1.00) during their third trimester of pregnancy. The majority were attending school (58%), and U.S. born (64%). When their child was four years old, the average family income was $24,773 (SD = $18,077). In terms of household composition, adolescent mothers reported the number of other adults and children living with them; the average was five individuals (SD = 2.57, with a range from 1 (e.g., just her child) to 14 individuals.

Data Collection

Adolescents were recruited from community agencies and high schools in a southwestern metropolitan area. To be eligible to participate, adolescents had to be of Mexican origin, 15 to 18 years of age, currently pregnant, and not legally married. Interviews were conducted in participants’ preferred language, and 61% of adolescents participated in English during the mother’s third trimester of pregnancy. Participants received $50 when their child was four years old, and $60 when their child was five years old. This research was conducted in accord with prevailing ethical principles and all procedures were approved by the university’s Human Subjects Review Board.

Variables

Demographic and control variables. During their third trimester of pregnancy, adolescents reported their nativity (0 = Mexico-born, 1 = U.S.-born). They subsequently reported on infants’ birth characteristics, including gender (0 = boy, 1 = girl), and birth length (centimeters) in a brief phone interview conducted approximately two weeks after the child’s birth. Birth weight was reported in pounds and ounces and subsequently converted to grams in order to be categorized into one of the following birth weight groups (−1 = very low birth weight, 0 = low birth weight, 1 = normal birth weight). Based on reported birth weight, infants were characterized as being low birth weight if they weighed less than 2,500 grams (5 lbs, 8 oz; n = 7), and very low birth weight if they weighed less than 1500 grams (3 lbs, 5 oz; n = 2).

When the child turned four years old, we assessed adolescent-mother figure coresidence status by asking adolescents whether they lived with the mother figure or not, and assessed adolescents’ family income by using self-report data from mother figures who co-resided with adolescents. It should be noted that if no mother figure co-resided with them, these data were still collected by the mother figure, as their estimate served as a close approximation for the adolescents’ available financial resources. Family income was computed by creating a sum of: mother figures’ annual income, funds received from any other individual who contributed to the household, and financial assistance from public assistance and/or food stamps.

Body Mass Index (BMI). Adolescent BMI was assessed when their child turned five years old and based on reported height (in inches) and weight (in pounds). We converted these height and weight measures to meters and kilograms, respectively, and subsequently computed adolescent BMI via the Quetelet index (weight*703/height²).²⁷ Child BMI was also based on their height and weight measurements. Specifically, children’s weight and height were assessed by a trained field interviewer via a home visit when they were four years old and again when they were five years old. Weight measurements were obtained using a calibrated step-on scale in pounds and ounces. Two measures of weight were taken and averaged to ensure accuracy and these values were subsequently converted to kilograms. Height measurements were taken using a wall-mounted measuring tape and recorded in inches and then converted to meters. Child BMI was subsequently calculated via the Quetelet index using these weight (in kilograms) and height (in meters) measurements.²⁷ The Body Mass Index Growth Charts for boys and girls under the age of 5 years and based on measures of BMI via the Quetelet index were used to determine whether the children in this study fell under one of the following categories: underweight, normal weight, overweight, and obese.²⁷ It should be noted that these values in the growth charts are adjusted for gender and age and organized by percentile²⁷ and provide the same results presented in the growth charts used by the CDC to classify weight for children over the age of two years.²⁸ Specifically, children in the 95^th percentile for BMI are considered obese, those in the 85^th percentile are considered overweight, those falling between the 5^th and 85^th percentile are considered normal weight, and those below the 5^th percentile are considered underweight.²⁸ A summary of the BMI frequency distribution for the children and adolescent mothers is presented in Table 1 and Table 2, respectively. BMI was operationalized as a continuous variable in study analyses.

Table 1:

Children’s and Adolescent Mothers’ BMI Frequency Distribution (N = 204)

Wave*	Boys BMI Categories (n = 118; 57.8%)	Sample Distribution (%)	Girls BMI Categories (n = 86; 42.2%)	Sample Distribution (%)	Adolescent BMI Categories (N = 204)	Sample Distribution (%)
W5	Underweight: 0–13.99	24.7%	Underweight: 0–13.749	21.9%	-	-
W5	Normal: 14 – 16.89	71.6%	Normal: 13.75–16.79	68.8%	-	-
W5	Overweight: 16.9 – 17.79	2.5%	Overweight: 16.8–17.99	3.1%	-	-
W5	Obese: 17.8 – 35.0	1.2%	Obese: 18–35.0	6.3%	-	-
W6	Underweight: 0–13.79	11.1%	Underweight: 0–13.49	10.8%	Underweight: < 18.5	2.3%
W6	Normal: 13.8–16.79	76.5%	Normal: 13.5–16.89	73.8%	Normal: 18.5–24.9	41.9%
W6	Overweight: 16.8 – 17.89	6.2%	Overweight: 16.9–18.249	7.7%	Overweight: 25.0–29.9	30.2%
W6	Obese: 17.9–35.0	6.2%	Obese: 18.25–35.0	7.7%	Obese: ≥ 30.0	25.6%

^*Wave 5 = children were, on average, 4 years old (SD = .05; range: 3.82 to 4.19); Wave 6 = children and adolescent mothers were, on average, 5 years old (SD = .50; range: 4.80 to 5.17) and 22 years old (SD = 1.00; range: 20 to 24).

Table 2:

Means, standard deviations, ranges for key study variables and covariates (N = 204)

Variable	Full Sample (N = 204)		Boys (n = 118)		Girls (n = 86)
	Mean (SD)	Range	Mean  (SD)	Range	Mean  (SD)	Range
W1 Adolescent nativitya	0.64 (.48)	0 − 1	.68 (.47)	0 −1	.59 (.49)	0 −1
W2 Child low birth-weight statusb	0.95 (.26)	−1 − 1	.96 (.24)	−1 − 1	.94 (.28)	−1 − 1
W2 Child birth-weight c	7.12 (.99)	3.00 − 9.63	7.12 (1.00)	3.60− 9.63	7.13 (1.00)	3.00− 8.94
W5 Family income	24,773 (18,077)	192− 93,200	25,096 (18,707)	192− 82,752	24,305 (17,260)	1,716− 93,200
W5 Coresidency with mother-figured	0.44 (.50)	0 − 1	.49 (.50)	0 −1	.36 (.48)	0 − 1
W5 Child age	4.02 (.05)	3.82 − 4.19	4.02 (.05)	3.82 − 4.17	4.02 (.05)	3.96 − 4.19
W5 Adolescent parenting efficacy	4.49 (.41)	3.18 − 5	4.53 (.38)	3.32 − 5	4.43 (.46)	3.18 − 5
W5 Child negative temperament	4.37 (.76)	1.55− 6.58	4.46 (.74)	2.42- 6.58	4.26 (.77)	1.55- 5.75
W5 Child BMI	14.70 (1.79)	7.60− 20.85	14.65 (1.67)	9.43− 20.00	14.76 (1.94)	7.60− 20.85
W6 Child BMI	15.49 (1.58)	12.35− 21.72	15.46 (1.45)	12.35− 21.72	15.52 (1.73)	12.71− 20.71
W6 Adolescent BMI	26.67 (5.26)	17.00− 51.58	27.07 (4.99)	18.56− 27.07	26.11 (5.61)	17.00− 51.58

Note.

^a0 = Mx.-born, 1 = U.S.-born.

^b−1 = very low birth weight, 0 = low birth weight, 1 = average birth weight.

^cbirth weight coded into pounds.

^d0 = no, 1 = yes.

Child temperament.

Maternal reports of child temperament were measured using the Child Behavior Questionnaire, Very Short Form (CBQ-VSF).²⁹ For the purposes of the present study, the Negativity factor was used, which is comprised of 12 items reflecting anger, discomfort, sadness, soothability, and fear. Adolescent mothers were asked to describe their children’s behavior over the past six months (e.g., “Gets quite frustrated when prevented from doing something s/he wants to do”, “Tends to become sad if the family’s plans don’t work out.”). Items were scored on a 7-point Likert-type scale from “never” to “always”. The internal reliability of this scale was reported by the authors of the instrument and Cronbach’s alpha was .72.²⁹ In the present study, Cronbach’s alpha was .67 at when the child was four years old.

Parental self-efficacy.

Adolescents’ perceived parental self-efficacy was assessed using the 25-item Postnatal Parental Expectations Survey.^30–31 This measure assesses the degree to which adolescents believe they are able to manage the tasks associated with being a parent (e.g., “knowing how to make your child feel better when he/she is upset”, “knowing what foods your child will and won’t eat”). Item responses were on a 5-point Likert scale, with endpoints of “not at all sure” (1) and “very sure” (5). The internal reliability was tested by the developer of the instrument and Cronbach’s alpha was .97.³¹ It should be noted that this measure was originally designed for new parents/parents of infants.³¹ As the children got older, several items on the measure were adapted by the research team to reflect the activities of parents of young children rather than infants. In the present study, Cronbach’s alpha was .95 when the child was four years old.

Analytic Approach

To address Aim 1, children’s and mothers’ BMI was calculated at 4 and 5 years postpartum (i.e., when child turned 4 and 5 years old). To test Aim 2, the authors used a multi-group structural equation modeling framework in Mplus version 7.2.³² Model fit was considered to be good if the comparative fit index (CFI) was greater than or equal to .95, the root-mean-square-error of approximation (RMSEA) was less than or equal to .05, and the standardized root-mean-square residual (SRMR) was less than or equal to .05.³³ Analyses controlled for adolescents’ nativity status, children’s low birth weight status, age, and gender, family income, and coresidency with adolescents’ own mother figures. Missing data were handled using full information maximum likelihood.³⁴

As a first step, two nested models were compared and the difference between them was examined using the chi-square difference test, examining the ΔCFI, and examining model fit. In this method, the first model allowed all path estimates to be freely estimated across child gender (i.e., unconstrained model), and the second model constrained the path estimates to be equal across gender (i.e., fully constrained model). If the chi-square difference test was significant or the ΔCFI was greater than .01, this suggested that there were significant differences in the model paths based on gender. In the case that a multigroup model was accepted as the better fitting model, subsequent models sequentially constrained paths and tested which paths differed between boys and girls. Non-significant chi-square tests indicate no differences across gender (resulting in constrained paths). Significant chi-square tests indicate paths should be freely estimated (i.e., unconstrained).

Regarding Aim 3, moderation analyses were conducted by mean-centering continuous measures (specifically, child temperament and adolescent parenting self-efficacy), prior to creating interaction terms.³⁵ In the process of testing paths for gender differences, each main effect path, moderator path, and interaction path were constrained at the same time, and the chi-square difference test, ΔCFI, and model fit were examined for the group of constrained paths. This approach was necessary rather than testing for gender differences one path at a time because it was not possible to probe an interaction that had no gender differences (i.e., path was constrained across boys and girls) if the main effect did have gender differences (i.e., path was freely estimated across boys and girls). Thus, paths that included interactions were tested in groups. If the chi-square difference test was not significant, the ΔCFI was less than .01, or model fit did not decrease, it suggested that there were no significant gender differences in the path (or group of paths), and the path (or group of paths) could be constrained to be equal across groups.

Simple slopes analyses were used to decompose significant interaction terms to determine whether individual slopes were significantly different from 0 and aid in the interpretation of statistically significant interactions.³⁶ Significant interactions were probed at one standard deviation above and below the mean of adolescent parenting efficacy, when the child turned four years old.

RESULTS

The results for each research question are presented below.

Research Question 1: Rates of Overweight and Obesity

Rates of child obesity among this sample (at age five years: 6.2% for boys and 7.7% for girls) were low in comparison to the rate of early childhood obesity nationally among Latino children ages 2–5 years old (18.0% for Latino boys and 15.2% for Latina girls).³⁷ Conversely, the rate of obesity among the adolescent mothers in this sample (25.6%) was found to be higher than 1) the current rate of obesity among Latina adolescent girls nationally (21.3%)³⁷ and 2) the current rate of obesity among adolescent girls nationally (21.0%).¹⁷ Descriptives for the key study variables and controls are presented in Table 1.

Research Question 2: Association between Child Temperament and BMI by Child Gender

An examination of the chi-square difference test, the ΔCFI, and the model fit suggested significant gender differences in study associations among children’s negative temperament, child BMI, and adolescent mother BMI. Specifically, results revealed that the adjusted chi-square difference test was significant [Δ χ2 (∆ df = 15) = 31.17, p = .008]. The ΔCFI between the fully constrained model and the unconstrained model was greater than .01. Furthermore, model fit indices indicated that the fully constrained model demonstrated worse fit [χ² (df = 23) = 39.13, p = .02; CFI = .82; RMSEA = .08 (90% CI: .03 - .13); SRMR = .13] than the unconstrained model [χ² (df = 8) = 7.96, p = .44; CFI = 1.00; RMSEA = .00 (90% CI: .00 - .12); SRMR = .04].

In subsequent models, where paths were sequentially constrained to test for group differences, there were no differences across gender in associations between adolescent mothers’ BMI when their child was five years old, nativity status during the mother’s third trimester of pregnancy, and parenting efficacy, family income, and co-residency when their child was four years old. Additionally, results indicated no significant differences across gender with respect to associations between children’s five year BMI and birth weight, and child negative temperament, adolescent parenting efficacy, and co-residency at five years. All other associations were freely estimated. The final partially constrained model demonstrated acceptable fit: χ² (df = 19) = 21.98, p = 0.30; CFI = 0.97, RMSEA = 0.04 (90% C.I. = 0.00, 0.10); SRMR = .05 (see Figure 1).

Figure 1.

Multi-group examination of children’s negative temperament as predictors of adoelscent mothers’ and children’s BMI (body mass index) with parenting efficacy as a moderator.

Model fit: χ² (df = 19) = 21.98, p = 0.30; CFI = .97, RMSEA = 0.04 (90% C.I. = 0.00, 0.10); SRMR = .05. Estimates for boys are displayed inside parentheses. Estimates for paths constrained to be equal across groups are in bold. Dashed lines indicate non-significant paths. Gray boxes represent study controls. ^a 0 = no, 1 = yes; ^b −1 = very low birth weight, 0 = low birth weight, 1 = average birth weight; ^c 0 = Mx.-born, 1 = U.S.-born. Wave 5 = Child average age was 4 years old; Wave 6 = child average age was 5 years old. *** p < .001. ** p < .01. * p < .05.

A significant gender difference emerged in the association between children’s negative temperament and adolescent mothers’ BMI (when their child was five years old). Specifically, girls’ negative temperament at age four was positively associated with adolescent mothers’ BMI when their child turned five years old (b = .23, SE = .10, p = .02), whereas this association was not significant for boys (b = .06, SE = .21, p = .78). Next, significant gender differences emerged in the associations between adolescent mothers’ BMI (when their child was five years old) and their children’s BMI at the same time point. In particular, adolescent mothers’ BMI was positively associated with children’s BMI for girls (b = .31, p < .001), but not boys (b = .18, p = .11).

It should be noted that study findings revealed gender differences in study control variables. First, the association between children’s BMI at age 4 and their age 5 BMI was stronger for girls (b = .71, p < .001), as compared to boys (b = .29, p = .005). Next, family income was significantly negatively associated with children’s five year BMI for boys (b = −.31, p = .001), but not for girls (b = −.06, p = .12).

Research Question 3: Moderating Role of Parenting Self-Efficacy by Child Gender

In evaluating the moderating role of parenting self-efficacy on the relationship between negative child temperament and BMI, there were no significant gender differences in associations between the main effects, interaction between child negative temperament and adolescent parenting efficacy (i.e., negative temperament X parenting efficacy), and children’s BMI. Next, although the main effects were not significant, the interaction between parenting efficacy and children’s negative temperament emerged as a significant predictor of children’s BMI (b = −.39, p < .001). Simple slopes analysis revealed that there was a significant, negative association between children’s negative temperament and their BMI among children whose mothers reported high levels of parenting efficacy (b = −.65, p = .02). Analyses also revealed a positive and significant association between children’s negative temperament and their BMI among children whose mothers reported low levels of parenting efficacy (b = .95, p < .001; see Figure 2). In line with work by Belsky and Pluess,³⁸ moderation findings illustrate contrastive effects, such that both slopes are significantly different from zero but in opposite directions. Given the relatively uniform distribution of weight status, such that a majority of children were categorized as normal weight when they turned five years old, findings suggest that higher efficacy is protective against high levels of children’s negative temperament.

Figure 2.

Moderation effects of adolescent parenting efficacy on the association between children’s negative temperament and children’s BMI (body mass index). Wave 5 = child’s average age was 4 years old; Wave 6 = child’s average age was 5 years old.

Note. *Denotes significant slope at p < .05; ***denotes significant slope at p < .001.

DISCUSSION

This study utilized data from a sample of Mexican-origin children and their teen mothers to identify the rates of overweight and obesity among the children and their mothers in this sample; to determine whether negative child temperament is associated with BMI in the child; and to subsequently assess the interaction between child temperament and parenting self-efficacy on child and mother BMI. Our findings support the importance of considering children’s and adolescent parents’ social-emotional well-being when engaging in efforts to address their collective health outcomes.

In terms of children’s own health outcomes, our study specifically revealed that the link between temperamental negativity and child BMI was moderated by mothers’ self-efficacy, such that mothers’ high parenting self-efficacy served as a protective factor against unhealthy BMI for children with high negativity. This significant interaction held for both boys and girls. Consistent with previous research, these findings underscore the importance of supporting adolescent mothers to strengthen positive beliefs about their parenting. Further, it may be that even in the context of parenting a child with a negative temperament, adolescent mothers with higher self-efficacy engaged in healthier feeding and soothing practices that were subsequently linked to children’s healthier BMIs.⁹ Conversely, our results also suggested that low self-efficacy among this sample of adolescent mothers is a BMI risk factor particularly for children high in negativity. It is possible that mothers with high self-efficacy and/or mothers who have children with low negativity are more likely to feel comfortable seeking out social support, which may have also contributed to these findings. Indeed, as noted earlier and given recent evidence of the role of family and friend support as one possible BMI protective factor,41 the role of a strong social support network during early childhood and particularly for adolescent mothers is critical. Future work should explore these and other possibilities to better understand mechanisms at an individual, family, and community level leading to children’s chronic health outcomes.

Our findings also revealed that children with more negative temperaments had mothers with higher BMIs, providing further evidence that parents of children with challenging behaviors may themselves have poorer physical health. These findings are significant, given the risks for obesity in the studied population. Indeed, while rates of obesity do not seem to be concerning among the young children in this study, obesity persists among this sample of adolescent mothers. Future work should aim to explain potential mediators and moderators of this link. For example, it may be that the challenges associated with temperamentally difficult children exacerbate adolescent mothers’ heightened risk of overweight status by promoting unhealthy food choices and nutrition, or by impeding their self-care efforts. Or, it may be that increased cortisol levels are contributing to adolescent mothers’ increased risk for being overweight or obese. Given Anzman-Frasca et al.’s (2012) finding that temperamentally negative children may be overfed as a means to soothe them,⁹ it could be that for adolescent mothers in our study, their own food choices may have been associated with efforts to cope with the challenges of their temperamentally difficult children. Also, given recent evidence of the role of family and friend support as one protective factor for the BMIs of adolescent mothers facing contextual stressors,41 it will be important for future research to examine the role of adolescent mothers’ broader support context in explaining this link between challenging child behaviors and adolescent mothers’ well-being.

It is important to note that at age 4 years, rates of underweight among children this sample were high (24.7% for boys and 21.9% for girls). However, and interestingly, the percentage of children who were underweight at age 5 years decreased considerably to 11.1% and 10.8% for boys and girls, respectively. At age five years, we also noted an increase in both rates of overweight and obesity among the children in this sample. While these rates were all considerably lower than current national samples of Latinx children,³⁷ a deeper examination of these trends specifically among children of adolescent mothers and Mexican-origin children of adolescent mothers, would be illuminating and informative.

Implications

Although significant improvements have been observed following the implementation of obesity prevention initiatives, rates of obesity among Latinx persist, with adolescent mothers at unique risk. Our present findings suggest that obesity prevention efforts targeting this demographic subgroup should consider young mothers’ well-being and parenting experiences. Health promotion efforts that focus on improving an adolescent mothers’ parenting self-efficacy (through, for example, providing developmentally appropriate techniques to help young mothers navigate decisions around their children’s food choices, discipline, sleep) would likely have significant implications for the physical health of young mothers. Indeed, future preventive efforts might consider drawing on critical work on home visitation programs as a means of addressing obesity during early childhood and among more low-income populations.³⁹ Previous research has confirmed that in addition to providing more skill-based learning and opportunity for more maternal practice,⁴⁰ the benefit of this kind of direct support for the parents of infants and through early childhood is critical for long-term obesity prevention.³⁹

Given the myriad of stressors experienced by adolescent parents, particularly those from lower socioeconomic backgrounds, obesity prevention efforts aimed at reducing BMI among adolescent mothers of Mexican-origin must be situated in the context of adolescent parenthood. Public health professionals should focus on developing obesity prevention programming that can be delivered efficiently, in settings that accommodate young children, and in the communities in which Latina adolescent mothers live, so that access to and utilization of programs is maximized. Further, developing and implementing preventive efforts that do not detract from their school day would ensure that young mothers develop key health skills without losing out on other limited academic time. In line with work by Anzman-Frasca and colleagues,⁹ our findings also suggest that future research must continue to look at longitudinal outcomes to better understand the relation between child temperament and physical health.

Limitations

Our study is the first, to our knowledge, to simultaneously measure the association between child and parent characteristics on the BMI outcomes of both adolescent mothers and their young children. Nevertheless, there are also limitations of this study that should be considered. First, we offer information about a specific population of Mexican-origin adolescent mothers and their children, and it will be important to test these risk and protective factors in other samples of teen and adult mothers, including those with varying access to prenatal and infant care resources, to increase the generalizability of this work. We also did not find comparable data on obesity prevalence to other racial-ethnic groups of adolescent mothers in the literature and thus drew comparisons to rates of obesity among teens who are not pregnant or parenting. Moreover, our study did not measure adolescent mothers’ BMI during pregnancy, despite research supporting the notion that such measures are predictors of children’s subsequent health status.³¹ Future work would benefit from controlling for the impact of a mother’s BMI on child BMI, as research has established that a mother’s BMI before and during pregnancy is an index of a mothers’ nutritional status as well as her metabolic condition. Relatedly, adolescents’ BMI was assessed when their child turned five years old and based on self-reported height and weight, as opposed to measured BMI. Furthermore, as noted in our methods, we used the CBQ-VSF which allowed the adolescent mothers to describe a range of their children’s behaviors and responses in a range of contexts. This abbreviated measure is a validated instrument with strong psychometric properties. In the present study, our Cronbach’s alpha was found to be slightly lower among our sample than in comparison to the internal reliability presented by the instrument authors, suggesting that the reliability of this measure may have been impacted by factors specific to our sample (e.g., adolescent raters, Latino mothers and children). Future research would benefit from the study of the negativity factor with similar samples, using both the shortened and longer CBQ version. In addition, Stifter et al.’s (2011) finding supports the notion that feeding practices (e.g., using food to soothe) may have been implicated in some of the present study’s findings.¹⁶ Future work should explore these and other possibilities to better explain the processes by which child and parent factors are related to health outcomes, and the ideal targets for prevention.

Conclusions

Efforts to promote aspects of well-being, including parenting self-efficacy, among Mexican-origin teen mothers and their children during early childhood may have significant implications for reducing rates of obesity among children of adolescent mothers. In addition to our current obesity prevention initiatives aimed at addressing such health behaviors as nutrition choices, physical activity, access to whole foods, and sleep, findings from the present study indicate that efforts focused on children’s broader parental context may be beneficial. In the face of the types of stressors associated with parenting a child with temperamental risks, future intervention efforts focused on parenting self-efficacy may have significant implications for the physical health of young mothers and their children.

Significance.

In a longitudinal study of Mexican-origin adolescent mothers and their children, findings demonstrate that maternal self-efficacy serves as a protective factor for child BMI.

What’s Already Known on this Subject?

Mexican-origin adolescent mothers and their children are at heightened risk for increased BMI and related comorbidities. Obesity prevention efforts that address the stressors and developmental needs specific to adolescent mothers and their children may be effective at improving health outcomes among this vulnerable subgroup.

What This Study Adds?

Negative child temperament is more likely to be associated with higher BMI among Mexican-origin adolescent mothers with low parenting self-efficacy. Among young children with a negative temperament, their mothers’ high parenting self-efficacy serves as a protective factor against unhealthy BMI.