Maternal Stress, Depressive Symptoms and Body Mass Index of Adolescents: A Prospective Study

Background Changes in adolescent dietary and sedentary behaviors contribute to adolescence being one of the most vulnerable periods for the development of overweight and overweight-related morbidity. Children become more autonomous during adolescence, particularly in regard to decision-making over their own health behaviors. Despite this expanding autonomy, parents still play a major role in shaping health behaviors of adolescents and can still have inuence on children’s weight outcomes. There is rising evidence that maternal stress and depressive symptoms are associated with young children’s weight outcomes. Longitudinal studies that test whether maternal stress or depressive symptoms may precede the development of adolescents’ weight outcomes are rare. This study aimed to ll this gap. Methods In this longitudinal cohort study data from 336 mothers and adolescents aged 10–14 years was used. Adolescents height and weight were measured, and both parents and adolescents lled in questionnaires on perceived stress and depressive symptoms. Regression analyses were performed in R to examine longitudinal links between stress and depressive symptoms at baseline (T1) and adolescents’ zBMI 6 months later (T2). Results Maternal general perceived stress (β = .14, p = .01) at T1 preceded the development of higher adolescents’ zBMI at T2, after controlling for baseline zBMI and other covariates, whereas maternal depressive symptoms at T1 (β = .04, p = .40) and other domain specic stress did not (maternal nancial stress, maternal stress at work, maternal stress at home). Additionally, lower educational level among adolescents was associated with a higher zBMI at T2 (β = .16, p = .001). Conclusions Results suggest that maternal general stress, but not depressive symptoms, may inuence adolescents’ weight development. Our ndings warrant future investigation on whether and how general stress among mothers, but also fathers, may predict weight increases of their adolescent offspring. The participants in the present study were part of wave 1 and wave 2 of the “G(F)OOD together” research project, a longitudinal study on Dutch adolescents’ and their parents’ health behavior. This ongoing longitudinal study has four waves of data collection: fall 2017, spring 2018, spring 2019 and (as planned before Corona-crisis broke out) spring 2020. This paper reports on the rst two waves. Parental consent was provided for 718 children. In the rst wave 667 adolescents from six secondary schools in the South and the East of the Netherlands participated and in the second wave 688 adolescents participated (95.8% participated in both waves). Because some adolescents were absent at wave 1 due to illness or other appointments, more adolescents actually participated in wave 2. Adolescent boys (W1, n = 314; W2, n = 320) and girls (W1, n = 353; W2, n = 368) were approximately equally represented. Most adolescents were born in the Netherlands (97.5%). All participants attended regular secondary education and were in their rst (i.e., US Grade 7; n = 460) or second year (i.e., US Grade 8; n = 207) (M age = 12.9 years; SD age = 0.7; age range = 10.1 to 14.9) at wave 1 and wave 2. More than half of the participants (57.6%) were in pre-university education, 8.2% of the participants was in higher general secondary education, and 34.2% of the participants was in pre-vocational education.


Background
In the transition from childhood to adolescence, dietary quality and physical activity deteriorate and unhealthy dietary and sedentary habits develop [1]. For example, the consumption of fruit, vegetables, and milk decreases from childhood to adolescence [1], while the consumption of soft-drink and snack consumption increases [1][2][3]. Additionally, screen time and sedentary behavior increase [4,5]. These changes in dietary and sedentary behaviors contribute to adolescence being one of the most vulnerable periods for the development of overweight and overweight-related morbidity [6][7][8]. During adolescence, children become more autonomous, particularly in regard to decision-making over their own health behaviors [9]. However, despite this expanding autonomy, parents still play a major role in shaping health behaviors of adolescents, mainly through modelling, parental rules and the creation of a healthy family environment [9][10][11].
In most households nowadays, mothers are still the most important caregivers [12], and manage most of the day-to-day child-care tasks [13]. They are considered to be the primary gatekeeper of the home food environment [14]. Because women experience greater levels of psychological stress than men [15,16], maternal psychological wellbeing and subsequent parenting behaviors may therefore have important implications for children's development. To date, growing evidence suggests that maternal psychological stress and depressive symptoms are associated with increased risk rates of obesity in children [17][18][19][20][21][22][23], children's decreased consumption of fruits and vegetables, higher child sedentary behavior [24][25][26] and breakfast skipping and children's high consumption of sweetened drinks, less sleep and less outdoor play [27][28][29].
It has been hypothesized that maternal stress and depressive symptoms may in uence children's weight related behaviors through three primary pathways [30]. First, maternal stress may alter mothers' own physical activity, sedentary behavior and dietary intake, which may impact adolescents' behaviors through modelling and household exposure (e.g., less healthy family meals) [31]. Additionally it is known that depressed mothers have di culty providing healthful food choices [32][33][34], and modeling physical activity behaviors relative to non-depressed mothers [11]. There is evidence that changes in maternal feeding styles and patterns due to stress and depression can have a signi cant impact on children's food composition and energy intake by for example preparing convenient but unhealthy meals to help manage time [12,35]. Second, maternal stress and depression may affect parenting behaviors or mother child interactions. Mothers experiencing high levels of stress may spend less time with their children [36] and may be less responsive in their interaction with their children [37]. Maternal depressive symptoms, such as negative affect and inactivity, can in uence child weight related behaviors by directly affecting parenting behaviors, reducing maternal sensitivity to and nurturance of the child's needs [29,38] and by facilitating less positive parent-child interactions and less family cohesion [39][40][41]. Third, maternal stress and depression can directly in uence children's behavior through alterations in the stress response of the child itself [31], responding to maternal stress with an increased biological or psychological stress response. Maternal depression also increases risk for depression in children [42,43]. Through these pathways children of mothers experiencing high levels of stress and depressive symptoms are also at a greater risk for overweight and obesity [17,39].
The majority of studies so far have focused on maternal stress or maternal depression or depressive symptoms exclusively, while longitudinal assessment of both maternal stress and depressive symptoms in relation to childhood and adolescents' BMI are rare. The few longitudinal studies investigating maternal stress or depression and depressive symptoms in relation to childhood overweight present mixed ndings [17, 22-24, 30, 44]. Maternal stress was consistently associated with greater risk for childhood overweight and obesity [17], whereas the association between depressive symptoms and child overweight was found to be more inconsistent [44], varying by gender and age [45] and measure of depression [44]. Additionally, most studies examined these associations in young(er) children. Given the importance of maternal behaviors on the development of adolescent overweight and the few longitudinal studies evaluating these associations in adolescence [22] our aim is to examine the association between maternal stress and depressive symptoms and adolescents' BMI in a longitudinal manner. Unique aspects of this study are the focus on these prospective associations in adolescence and the assessment of both maternal stress and depressive symptoms.

Participants
The participants in the present study were part of wave 1 and wave 2 of the "G(F)OOD together" research project, a longitudinal study on Dutch adolescents' and their parents' health behavior. This ongoing longitudinal study has four waves of data collection: fall 2017, spring 2018, spring 2019 and (as planned before Corona-crisis broke out) spring 2020. This paper reports on the rst two waves. Parental consent was provided for 718 children. In the rst wave 667 adolescents from six secondary schools in the South and the East of the Netherlands participated and in the second wave 688 adolescents participated (95.8% participated in both waves). Because some adolescents were absent at wave 1 due to illness or other appointments, more adolescents actually participated in wave 2. Adolescent boys (W1, n = 314; W2, n = 320) and girls (W1, n = 353; W2, n = 368) were approximately equally represented. Most adolescents were born in the Netherlands (97.5%). All participants attended regular secondary education and were in their rst (i.e., US Grade 7; n = 460) or second year (i.e., US Grade 8; n = 207) (M age = 12.9 years; SD age = 0.7; age range = 10.1 to 14.9) at wave 1 and wave 2. More than half of the participants (57.6%) were in preuniversity education, 8.2% of the participants was in higher general secondary education, and 34.2% of the participants was in pre-vocational education.
Moreover, 777 mothers or fathers provided consent to participate in the study themselves, of which 593 took part in Wave 1 and 586 took part in wave 2 (98.8% in both waves). Both parents of the adolescents were invited to participate in the project, though only in 85 cases, both caregivers actually participated. In case two caregivers participated, we only included the biological mothers (n = 83; excluded: n = 5 nonbiological mothers; n = 80 biological fathers; n = 2 non-biological fathers). Although some fathers participated in the study (N = 148 in rst wave, N = 135 in second wave (91.2% in both waves), we chose to only include data of mothers. According to literature mothers are still the most important caregivers in the family, and are more prone to stress and depression [12,15,16] than fathers. Also, in the remaining sample in which only one caregiver participated, biological mothers participated most often (n = 352 biological mothers; excluded: n = 1 non-biological mother; n = 68 biological fathers, n = 2 non-identi able), resulting in 435 biological mothers. We only included cases with data on the main measures on both waves in both mothers and adolescents, leaving a nal sample of 336 biological mothers.

Procedures
When a secondary school agreed to participate, all adolescents attending the rst and second grade and their parents were invited to participate in this study by means of an active parental consent procedure. A letter describing the four-wave study was mailed to the parents and they were asked to return a (paper or online) consent form indicating whether they agreed to their child participating in the study and if they agreed to participate in the study themselves. Children were rewarded with a small incentive, if at least one of their parents' forms was returned, regardless of whether permission was given. Before participation, adolescents and parents were informed that participation was voluntary, that answers would be processed anonymously, and that they could withdraw from the study at any moment. Inclusion criteria for participants were being enrolled in a high school, being in the rst and second grade of this high school, being pro cient in the Dutch language and parents and children both having given active informed consent. Exclusion criteria for participants were not being pro cient in the Dutch language, attending special education and not having given active (parental) consent.
Adolescents completed an online survey at school during one classroom hour (approximately 45 minutes), and height and weight were measured outside the classroom by trained students. Parents completed an online survey, which took approximately 20 minutes to complete. The questionnaires were administered through Qualtrics Survey Software (Qualtrics, Provo, UT, USA). Children received a small present after completing the survey, and several prizes were ra ed among participating parents. The Institutional Review Board of the Faculty of Social Sciences of the Radboud University, Nijmegen, The Netherlands approved the study protocol (reference number ECSW20170805-516) in 2017.

Depressive symptoms
Maternal depressive symptoms were assessed with the 10-item short version of the Center for Epidemiological Studies-Depression (CES-D) scale. The (shortened) CES-D is widely used and has adequate internal reliability [46]. Respondents rated items on a 4-point Likert scale (rarely or none, to most or all the time). The scale includes positive (I was happy) and negative (I could not get going) items. Higher total CES-D scores re ect greater maternal depressive symptomology. In the current study, Cronbach's alpha for the CES-D was .77 at T1 and .80 at T2.

General Stress
Maternal general stress levels were assessed using the 4 item Perceived Stress Scale (PSS). The PSS is a self-report questionnaire measuring a person's evaluation of stressful situations in the previous 1 month of his or her life. It is a global measure of stress that is simple to use, and there are many studies con rming its reliability and validity in a variety of settings and in multiple languages [47][48][49][50][51][52]. The instrument contains 4 statements which measure how unpredictable and uncontrollable respondents feel their lives are, for example: In the last month, how often have you felt con dent about your ability to handle your personal problems? Respondents rate how often they experience stressful situations on a 5point Likert scale ranging from 'never' to 'very often'. Answers of the 4 items were summed into a total PSS score. The higher the score on the PSS, the greater the respondent perceives that their demands exceed their ability to cope. Cronbach's alpha was calculated to investigate the internal reliability of the

Anthropometrics
Adolescents' height and weight were measured according to protocol [53] by trained research assistants.
Body Mass Index (BMI) was calculated as weight in kilograms divided by height in meters squared. Individual age and gender-speci c BMI standard deviation scores (z-scores) were calculated using a Dutch representative sample of 0-21-year olds [54]. Mothers reported their own height and weight based on which we calculated maternal BMI.

Covariates
Adolescents' level of education (1 = lower general secondary education; 2 = higher general secondary education and pre-university education) and gender of adolescent (0 = boys; 1 = girls), and maternal BMI were included as covariates.

Statistical Analyses
Statistical analyses were conducted using the PASW 20.0 and R software package. Descriptive statistics were used (mean, standard deviations and percentages) to describe the study sample and to investigate population characteristics (see Table 2). To assess whether covariates were confounded with the outcome measure zBMI at T2 a one-way ANOVA was used. Data were analysed using multiple linear regression analyses, with adolescents' zBMI at wave 2 as the outcome measure and adjusting for adolescents' zBMI at baseline and potentially relevant covariates (i.e., educational level of the adolescent, gender, and maternal BMI). First, to examine cross-sectional associations between mothers' wellbeing and adolescent zBMI and the covariates, Pearson's correlation coe cients were calculated. Second, to test whether maternal stress or depressive symptoms may precede the development of child weight over time, multiple linear regression analyses were performed using R software package (R Core Team, 2018). The lavaan package in R was used in order to account for missing values and to deal with potential issues involving the non-normal distributions of the outcome measure and other variables. Separate regressions were performed with depressive symptoms (CES-D score), general stress (PSS score), nancial stress, stress at home or stress at work at T1 as the independent variable, and adolescents' zBMI at T2 as dependent variable. We investigated main effects of maternal depressive symptoms and maternal stress on adolescent's zBMI at wave 2, after correction for baseline zBMI, and the effects of potential covariates (i.e., educational level, gender, and maternal BMI).

Preliminary analyses
To assess whether covariates were confounded with the outcome measure one-way ANOVA was used.

Cross-sectional Associations
Pearson's correlation coe cients between maternal wellbeing, covariates and adolescent zBMI are presented in Table 1, along with descriptive statistics. Signi cant correlations were found between maternal depressive symptoms and most of the stress measures, with exception of the 'stress at work' measure. Mostly non-signi cant correlations were found between maternal wellbeing (i.e., stress or depressive symptoms) and adolescents' zBMI. However, perceived nancial stress by mothers was signi cantly associated with higher zBMI at baseline among adolescents (r = .10, p = .04). Of the covariates, maternal BMI was positively correlated with maternal general stress (r = .14, p = .03) and with maternal depressive symptoms (r = .12, p = .01). Educational level of the adolescent was negatively correlated with maternal depressive symptoms (r=-.17, p = .003).

Longitudinal Associations Between Maternal Mental Wellbeing And Adolescents' Zbmi
Maternal general stress (β = .14, p = .01) preceded adolescents' zBMI over time after correction for baseline zBMI, gender, educational level of the child, and maternal BMI as can be seen in Table 2. This effect was still signi cant in a reduced model without covariates (β = .10, p = .04). Notably, for the other maternal well-being variables (i.e., maternal nancial stress, maternal stress at home, maternal stress at work, maternal depressive symptoms) no signi cant longitudinal associations were found with adolescents' zBMI, neither in a reduced model nor after controlling for covariates. Additionally, lower educational level among adolescents was associated with a higher zBMI at T2 (β = .16, p = .001). The nal model (including all covariates) of maternal general stress explained 33% of the variance of adolescent zBMI.

Discussion
In the past decade, several studies have established the link between maternal well-being and childrens' zBMI [17][18][19][20]22]. However, these studies mostly focused on (early) childhood. During adolescence, particularly early adolescence, parents still play an important role in their children's lives. Maternal stress and depressive symptoms have been linked to barriers to a healthy lifestyle and may reduce pro-active obesity-related parenting practices [18,[55][56][57] such as less healthy meal preparation and less transportation to and less participation in organized sports by their children, and by for example, negatively in uencing mother-child interaction and increasing the risk at modelling possibilities of unhealthy maternal behaviors [12, 28, 30, 31, 35-41, 57, 58]. The present study's aim was to investigate the longitudinal link of both maternal stress and depressive symptoms with adolescents' weight development. We found a small link between maternal general stress and adolescent zBMI over time. We did not nd signi cant associations for maternal depressive symptoms or for domain speci c stressors (maternal nancial stress, stress at home and stress at work).
Maternal general perceived stress may re ect a broader personality construct on how mothers more generally react to stressful situations, whereas stress from the home and work environment probably is more contextual in nature. This may explain why maternal general stress is associated with weight development in adolescents, having more general impact than speci c contextual stress factors. Additionally, adolescents with mothers who experience a lot of general stress may be exposed to less healthy family and peer environments, providing increased opportunities to engage in unhealthy behaviors and may lead to unhealthy weight development.
In contrast, no associations were found between maternal depressive symptoms and adolescent BMI over time. Notably, previous research has already shown some mixed ndings particularly with regard to the link between maternal depressive symptoms and children's weight outcomes. A review reports that chronic depression (depression measured on multiple occasions), but not episodic (depression at a single measurement occasion) depression was found to be associated with a greater risk for child overweight [44]. In our study we measured depressive symptoms at one time point, which may explain why we did not nd any associations. However, it should be noted that other studies focusing on episodic symptoms did sometimes nd a link with children's weight outcomes [59,60], though these studies were most often conducted among younger children.
The link between maternal stress and children's weight outcomes have been repeatedly found among families with younger children [17]. To the best of our knowledge, only one previous longitudinal study found this link among adolescents [22]. Our study adds to these previous studies that diverse stress factors have been examined and that only for one speci c stress factor longitudinal links have been found (i.e., general stress) in an adolescent population. Thus, it might be that, particularly for adolescents, maternal general stress has more impact on (healthy) family life and adolescents' weight development than maternal stress in other domains and also more impact than maternal depressive symptoms. Future studies including both age groups (i.e., younger children and adolescents) may further examine this.
Socioeconomic factors are also known to be of speci c interest in weight development of children and adolescents. We found that maternal nancial stress was associated with higher baseline zBMI scores among adolescents, a nding in line with previous ones [23,24,61]. More remarkably, a lower educational level among adolescents was consistently associated with a higher zBMI at T2 (β = .16, p = .001) in all our models. Our study thus suggests that increases in zBMI are most unfavorable in adolescents with a lower educational level. Previous studies have reported about the educational gap with regard to weight outcomes [62]. It seems that the period of adolescence is a particularly important period because of the autonomy involved in making more independent choices about weight related behaviors. Adolescents with lower educational levels may make more unhealthy choices in their weight related behaviors than highly educated adolescents, and may have less ( nancial) opportunities for healthy weight related behaviors.
The current study had several strengths and limitations. One particular strength is that height and weight of the adolescents were objectively measured. Moreover, the high participation rates and the prospective repeated measurements add to the strengths of this study. A nal strength includes the fact that parents reported on their own mental health. Despite these strengths, also some limitations should be acknowledged. First, maternal mental health was analyzed at one time point only, to assess stability in maternal mental state more time points should be taken into account. Second, the sample consisted of a high percentage of highly educated respondents (57.1%) as well as a high proportion of respondents having a healthy or even low zBMI, possibly in uencing the generalizability of the results. Finally, contextual stress factors (i.e., nancial, home or work-related stress) were only measured with one item.

Conclusions
To conclude, the ndings of the current study suggest that maternal depressive symptoms do not predict any changes in weight development among adolescents. In contrast, adolescents whose mothers experienced more general stress may be at greater risk for increases in zBMI. However, our ndings need replication and warrant future investigation on whether and how general stress among mothers, but also fathers, might predict weight increases of their adolescent offspring.