Introduction

Internalizing symptoms are the most common mental health problems among adolescents in the United States and are associated with adverse outcomes such as impaired functioning, substance use, and poor academic achievement [34,61]. Studies have consistently demonstrated that internalizing symptoms often emerge and exacerbate during early adolescence, a period that coincides with the onset of puberty [44, 53]. Time and again, parenting practices and parental symptomatology are linked to adolescent internalizing symptoms [17, 63]. Yet, fewer longitudinal studies have examined familial predictors of internalizing symptoms during the pre- to post-pubertal transition. From a developmental psychopathology framework, the goal of this manuscript is to investigate the role of familial factors in the development of internalizing symptoms from pre- to post-puberty. We briefly review the relationship between adolescence and internalizing symptoms, followed by a review of family factors such as parental psychopathology, parenting behaviors, and family environment in the development of internalizing symptoms.

Adolescence and Internalizing Symptoms

Pubertal initiation at the beginning of adolescence results in many significant physical, hormonal, and psychological transformations [47]. While undergoing biological changes, adolescents also start exploring their social environment by becoming more engaged with their peers to make sense of their identity, values, and personality while seeking independence from their parents [20]. With these physical and psychosocial changes occurring simultaneously and abruptly, puberty has been frequently associated with emotional distress [47]. Indeed, past work has shown that psychological symptoms increase steeply during puberty, prompting investigations into the link between pubertal development and the onset of internalizing disorders and symptoms [32]. These effects can be particularly pronounced for adolescent females who are more likely to experience increased emotional intensity and reactivity, body dissatisfaction, and peer problems [4, 10, 51]. In the literature, scholars have focused on pubertal timing and pubertal status and their associations with internalizing symptoms. Pubertal timing refers to the extent to which an individual matures earlier or later than their peers [27], whereas pubertal status focuses on the current stage of physical maturation at a given time point (e.g., pre-pubertal, mid-pubertal, late/post-pubertal). Specific associations have been observed between pubertal timing and internalizing symptoms [21], with early pubertal timing being a consistent risk factor for mental health problems in adolescents [68]. Regarding the effects of pubertal stage, a systematic review conducted by Stumper and Alloy [65] found that advanced pubertal stage was associated with depression, especially among White females . There were no increases in risk of depression for males as pubertal stage increased, highlighting the sex differences that occur during adolescence. According to the social deviance model, adolescents are at an increased risk of psychosocial problems when they are considered “off-time” in comparison to their peers [66]. As such, the findings from these studies support that when puberty is untimely (i.e., deviant from their peers), adolescents are at a higher risk for internalizing symptoms.

Family Factors and Internalizing Symptoms

Developmental psychopathology provides a useful framework for understanding the onset and course of mental health symptoms, and for this study, internalizing symptoms [16]. According to this framework, development is a dynamic process that may deviate based on adolescent individual characteristics and environmental factors [34]. Scholars have theorized that environmental factors, such as parent psychopathology, parenting behaviors, and family conflict, may influence the development of psychopathology [11, 60]. It is important to study family factors as predictors of adolescent internalizing symptoms, given that these factors are often modifiable and can be targeted in prevention and intervention efforts, thereby reducing the onset of psychopathology in offspring [17]. For this study, we focused on familial factors specifically on as predictors of adolescent internalizing symptoms. A systematic review conducted by Clayborne et al., [17] found that longitudinal studies examining the associations between childhood parenting practices and adolescent internalizing psychopathology (i.e., depression, anxiety, and internalizing symptoms) remain inconsistent. Further, facets such as parental monitoring, studying cohorts over extended time, discrete developmental periods, and diverse samples remain as important areas for inquiry among the parenting literature.

The intergenerational transmission of internalizing symptoms has been well established, with studies demonstrating that parental internalizing symptoms can be passed down to their children starting from early childhood [5, 19]. Researchers have proposed several mechanisms to explain this relationship, including genetic predispositions, environmental influences, and social factors, which have been well documented in the literature [57]. During adolescence, parental anxiety and parental depressive symptoms strongly correlate with adolescent internalizing symptoms [43, 48]. Parenting styles have been established as a factor in the development and maintenance of internalizing symptoms [45, 46, 56]. Broadly, high levels of parental responsiveness through warmth, support, and monitoring yield positive adolescent outcomes such as positive feelings, ability to overcome problems, and lower internalizing symptoms [3, 59]. Conversely, high levels of parental demandingness through psychological control (i.e., emotional manipulation) and behavioral control (i.e., excessive rules and restrictions) yield negative adolescent outcomes such as negative self-concept, behavioral problems, and increased internalizing symptoms [8, 41, 58]. Family conflict (i.e., interparental conflict and parent-child conflict) is another stressor that can influence the development of internalizing symptoms [24, 70]. Indeed, past studies have found that interparental conflict and parent-child conflict are associated with poor psychosocial adjustment and greater internalizing symptoms [23, 71].

Current Study

The Adolescent Brain Cognitive and DevelopmentSM Study (ABCD Study®) is an extensive longitudinal biobehavioral study throughout the United States that follows youth from ages 9 to 10 into their early adulthood [37]. Using longitudinal panel data, the ABCD Study uniquely allows us to examine how familial factors impact youth outcomes across development. To date, a large body of research suggests that family factors and puberty may influence the development of internalizing symptoms during adolescence. However, the relationship between family factors and internalizing symptoms has yet to be explored longitudinally within the ABCD sample and during the pubertal transition (i.e., pre-to-post puberty). To address this gap, the present study examines how family factors influence the development of internalizing symptoms in adolescents who were prepubertal at baseline and reached late puberty by the 4-year follow-up. By addressing this question, we aim to provide insights into how parenting influences the development of internalizing symptoms within the ABCD sample and how these findings can inform the broader field, as well as prevention and intervention measures.

In the present study, our first aim was to examine whether parents’ internalizing symptoms at baseline predicted adolescents’ internalizing symptoms at the 4-year follow-up. Consistent with past findings throughout adolescence, we hypothesized that parent symptoms would predict adolescent symptoms. Our secondary aim was to test whether certain parenting practices and family environments served as buffers or magnifiers of adolescents’ internalizing symptoms. We expected that greater parental monitoring and warmth would predict lower adolescent internalizing symptoms, and that greater family conflict would predict higher adolescent internalizing symptoms. Lastly, we examined whether biological sex moderated the relationship between family factors and adolescent internalizing symptoms.

Methods

Participants

After approval from the National Institute of Mental Health (NIMH) Data Archive, we downloaded the baseline and 4-year follow-up data from the 6.0 data release of the ABCD Study. The ABCD Study collects annual data from adolescents and their families across 21 research sites throughout the United States. Research sites were selected based on probability sampling to ensure diversity and national representativeness. From 2016 to 2018, research teams recruited families through public, charter, and private school systems. All caregivers provided informed consent, and child participants provided assent. For more detailed information regarding the recruitment, consenting, and design of the study protocol, see Garavan et al., [29]. From the 11,875 youth participants at baseline, 9,739 completed their 4-year follow- up visits by the ABCD 6.0 release. We created a subset of adolescents who were pre-pubescent at baseline (Mage = 9.94 ± 0.6 years) and reached late puberty by the 4-year follow-up (Mage = 14.2 ± 0.7 years), using the Pubertal Development Scale (PDS; [55]). For this measure, a score of 1 indicates pre-puberty, and a score of 5 indicates post-puberty. We included adolescents who had a score of 1 at baseline (pre-pubertal) and a score of ≥ 4 at the 4-year follow-up (late-puberty/post-puberty). Among 5,830 participants who met these inclusion criteria at baseline, 1,335 did not provide enough information on the PDS to compute a total score. Overall, our final analytic sample size was N = 2,276. Notably, the first measurement of the ABCD Study occurred after pubertal initiation for many participants (70.3% of females and 32.6% of males were not prepubertal at baseline). This means that our inclusion criteria capture the information on the role of pubertal onset, but only for late developing females and early developing males. Sample demographics are contained in Table 1.

Table 1 Demographic characteristics of participants who met study inclusion criteria
Full size table

Measures

Descriptives and reliability scores for study measures are found in Table 2.

Table 2 Descriptive statistics for adolescent and parent measures
Full size table

Adolescent Internalizing Symptoms

Adolescent internalizing symptoms were measured from parent reports of the Child Behavior Checklist (CBCL; [1]) at baseline and the 4-year follow-up. The CBCL measures behavioral and emotional problems in children and adolescents between the ages of 6 and 18 years old. The CBCL consists of 112 questions on a 3-point Likert scale ranging from 0 (not true) to 2 (very true or often true), asking how often their child is bothered by each problem. The questions yield 8 syndrome scales and 6 DSM-oriented scales that capture internalizing and externalizing symptoms. For our analyses, we used the raw score from the internalizing problems scale (see Table 2 for descriptive statistics), which consists of the anxious/depressed, withdrawn/depressed, and somatic complaints subscales. Internal consistency was strong in our sample (α = 0.87).

Parent Internalizing Symptoms

Parent internalizing symptoms were measured from the Adult Self Report (ASR; [2]) at baseline. The ASR measures adaptive functioning, psychological symptoms, and substance use. The ASR consists of 126 questions on a 3-point Likert scale ranging from 0 (not true) to 2 (very true or often true), asking how often they are bothered by each problem. The questions yield 8 syndrome scales and 6 DSM-oriented scales, assessing internalizing and externalizing symptoms. In our analyses, we used the raw score from the internalizing problems scale (see Table 2 for descriptive statistics), which consists of the anxious/depressed, withdrawn/depressed, and somatic complaints subscales. Internal consistency was strong in our sample (α = 0.89).

Perceived Parental Acceptance

Perceived parental acceptance was measured using the youth report of the acceptance subscale of the Child Report of Parenting Behavior Inventory (CRPBI; [62]) at baseline. The subscale consists of 5 items on a 3-point Likert scale ranging from 1 (not like him/her) to 3 (a lot like him/her), assessing parental warmth, acceptance, and responsiveness (e.g., “My caregiver makes me feel better after talking over my worries with him/her”). A total acceptance score was calculated using the sum of the 5 items (see Table 2 for descriptive statistics). Higher scores indicate greater perceived parental acceptance. Internal consistency was adequate in our sample (α = 0.71).

Perceived Parental Monitoring

Parental monitoring was measured using the youth report of the Parental Monitoring Survey (PMQ; [38]) at baseline. The survey consists of 5 questions on a 5-point Likert scale ranging from 1 (never) to 5 (always or almost always). The questions capture how much a caregiver knows the whereabouts of their child (e.g., “How often do your parents/guardians know where you are?”). A total parental monitoring score was calculated using the sum of the 5 items (see Table 2 for descriptive statistics). Higher scores indicate greater parental monitoring. Internal consistency was suboptimal in our sample (α = 0.48).

Perceived Family Conflict

Family conflict was measured using the parent report of conflict subscale from the Family Environment Scale (FES-P; [49]) and the child report of conflict subscale from Family Environment Scale (FES-Y; [50]) at baseline. The subscale consists of 9 items scored as either 0 (false) or 1 (true). The questions capture how conflict manifests in their family (e.g., “Family members often criticize each other”). A total family conflict score was calculated by summing the 9 items (see Table 2 for descriptive statistics). Higher scores indicated greater perceived family conflict. Internal consistency was modest in our sample for the FES-P (α = 0.67) and FES-Y (α = 0.69).

Perceived Pubertal Status

Pubertal status was assessed by the Pubertal Development Scale (PDS; [55]) at baseline and the 4-year follow-up. Per contemporary recommendations, we used parent reports for our analyses [14, 64]. This 5-item measure captures perceived physical pubertal development. For both sexes, the measure asks about the child’s growth spurt, skin, and body hair changes (3 items). Moreover, there are biological sex-specific items (2 items) assessing breast development and menarche for females (e.g., Have you noticed that your/your child’s breasts have begun to grow?”) and items assessing facial hair and voice changes for males (e.g., “Have you/your child begun to grow hair on your/his face?”). Items are scored on a four-point Likert scale from 1 (not yet changing) to 4 (changes seem complete), except for the menstruation question, where a 4 indicates menstruation has begun. The PDS categorical score is calculated by summing the 2 sex-specific questions and the body hair question, with the resulting sum indicating the pubertal development stage.

Analytic Plan

While the CBCL and ASR are derived from the Achenbach System of Empirically Based Assessment (ASEBA) and historically have displayed high levels of reliability, several other measures that were used have not shown similar levels of reliability. We evaluated reliability of our study measures and found that this pattern held in our sample as well (see Cronbach’s alpha values in Table 2; [22]). To address unreliability in the CRPBI, PMQ, FES-P, and FES-Y, we first modeled them as latent variables to remove measure unreliability due to imperfect internal consistency. To create these latent variables, confirmatory factor analysis (CFA) was utilized. All CFA-based models reflected a unidimensional factor model for each respective measure. Overall model fit was assessed using the Root Mean Square Error of Approximation (RMSEA), the Comparative Fit Index (CFI), and the Standardized Root Mean Square Residual (SRMR). Model fit cutoff values followed established standards where acceptable fit was considered by values of CFI ≥ 0.95 and RMSEA and SRMR values ≤ 0.08 [33, 36]. In the case of inadequate model fit, we implemented theory-based model modifications that were informed by Lagrange multipliers. For our predictive models, we used the observed scores from the CBCL and ASR and the latent variables for the other multi-item measures pertaining to the CRPBI, PMQ, and FES.

To assess for change associated with family factors, we took an approach similar to hierarchical regression. We first regressed the 4-year follow-up CBCL internalizing sum scores on the baseline CBCL internalizing sum scores, along with sex, age, and study recruitment site. This model served as a base model for comparison. In a second set of models, each family factor measure was added to this model one at a time, in separate respective models. This process allowed us to identify the amount of change in CBCL internalizing scores attributable to each family factor measure above and beyond the initial regression model. In a third set of models, we added a test for moderation by sex to each main effect. For main effects that were measured using latent variables, we used moderated structural equation models [7]. For main effects, effect sizes were calculated based on increased model R2 after each family factor covariate was included, with 0.01, 0.09, and 0.25 reflecting small, median, and large effect sizes, respectively [18] . We also included standardized β estimates for relative comparisons along with 95% confidence intervals. For moderation models, we only included standardized β estimates, because latent variable interaction models cannot include residual covariances for latent variables or provide R2 estimates that are directly comparable to fixed effects models. For these models, we provide standardized β estimates to compare relative effects to the β estimates that resulted from the main effect models.

All models were implemented using Mplus 8.6. Base models and main effect models employed mean- and variance-adjusted weighted least squares (WLSMV) estimation to adjust for the ordinal nature of several study measures [52], while robust maximum likelihood was used for moderation models. No items had greater than 0.13% of data missing for participants who met inclusion criteria.

Results

CFA models demonstrated an acceptable fit to the PMQ (CFI = 0.99; RMSEA = 0.02; SRMR = 0.01), CRPBI (CFI = 0.98; RMSEA = 0.07; SRMR = 0.04), and the FES-Y (CFI = 0.97; RMSEA = 0.04; SRMR = 0.05) data, with all CFA loading estimates significant. The initial CFA model fit for the FES-P data did not achieve an acceptable fit (CFI = 0.90; RMSEA = 0.08; SRMR = 0.10). After identifying FES-P items with substantial overlap in item content, we accounted for correlated residuals in the model that could account for this overlap. This resulted in a model with acceptable fit to the FES-P data (CFI = 0.95; RMSEA = 0.07; SRMR = 0.07; see Supplementary Materials for more detail on model fitting, and Supplementary Tables S1-S4 for loading estimates).

Bivariate correlations among study variables can be found in Supplementary Table S5. For testing study hypotheses, in the initial base regression model, higher adolescents’ internalizing symptoms at baseline were significantly associated with higher adolescents’ internalizing symptoms at the 4-year (β = 0.5 [0.47, 0.53]; overall R2 = 0.27). We then added relevant predictors to this base model; Table 3 presents the results of the main analyses that follow the initial model. Higher levels of parental internalizing symptoms (β = 0.1 [0.06, 0.14]; ΔR2 = 0.01) and parent-reported family conflict at baseline (β = 0.06 [0.01, 0.10]; ΔR2 < 0.01) were significantly associated with higher adolescent internalizing symptoms at the 4-year follow-up, with a small effect size for parental internalizing symptoms. Higher parental monitoring (β=-0.06 [-0.11, -0.01]; ΔR2 < 0.01) was significantly associated with lower adolescent internalizing symptoms at the 4-year follow-up. Parental acceptance (β=-0.02 [-0.07, 0.03]; ΔR2 < 0.01) and the child report of family conflict (β=-0.01 [-0.06, 0.03]; ΔR2 < 0.01) were not significantly associated with adolescents’ internalizing symptoms at the 4-year follow-up. No sex-based interaction tests for moderation were significant. These include interactions between sex and the ASR (β = 0.10 [-0.01-0.07], CRPBI (β = 0.02 [-0.03-0.09], PMQ (β=-0.05 [-0.11-0.25], FES-Y (β = .-0.02 [-0.03-0.09], FES-P (β = 0.01 [-0.04-0.11].

Table 3 Results for longitudinal prediction of change in adolescent internalizing symptoms by family factors, after accounting for recruitment site and baseline age and sex
Full size table

Discussion

The ABCD Study, with its large-scale, longitudinal design and nationwide data collection, provides an unprecedented opportunity to examine environmental factors that may contribute to adolescent mental health outcomes. While previous studies suggest that family factors may be associated with adolescent mental health, few have explored this relationship longitudinally from pre- to post-puberty. The present study is one of the first to examine the associations between family factors and adolescent internalizing symptoms longitudinally using ABCD data, one of the largest samples to date. As hypothesized, we found that higher levels of parents’ internalizing symptoms and parent-reported family conflict at baseline were significantly associated with higher levels of adolescent internalizing symptoms at the 4-year follow-up. The youth report of family conflict was not significantly associated with adolescent internalizing symptoms. Additionally, higher levels of parental monitoring were significantly associated with lower levels of adolescent internalizing symptoms at the 4-year follow-up. Contrary to our hypotheses, parental acceptance at baseline was not significantly associated with adolescent internalizing symptoms at the 4-year follow-up. Lastly, there were no moderating effects of adolescent sex on the associations between family factors and adolescent internalizing symptoms.

First, higher levels of parents’ internalizing symptoms longitudinally predicted higher levels of adolescent internalizing symptoms. This finding aligns with extensive research linking parent and adolescent psychopathology [15, 63]. One hypothesized pathway is that parent psychopathology elevates parental stress, straining the parent-adolescent relationship and contributing to more negative parenting, which elicits adolescent psychopathological symptoms [63]. Genetic predisposition may also play a role, with adolescents inheriting vulnerabilities that increase their risk of developing similar emotional symptoms [39, 42]. While we did not directly test these specific pathways, these interpretations may shed some light and explain our findings.

Second, higher levels of parent-reported family conflict were associated with higher levels of adolescent internalizing symptoms. This finding is unsurprising, as the family environment is a critical social context that contributes to adolescent mental health outcomes [9]. When family environments are cohesive, this can result in positive outcomes such as improved mood and quality of life [28]. In contrast, when family environments are high in conflict, this can lead to negative mental health outcomes such as negative mood and psychopathological symptoms [24, 67]. Youth-reported family conflict was not significantly related to adolescent internalizing symptoms. Initial bivariate correlations revealed a small, positive, and significant relationship at 4-year follow-up (see Table S5); however, after accounting for baseline symptoms, age, sex, and study site, the association was attenuated and no longer statistically significant. It is well known that informant discrepancies exist between parent and child reports of psychological phenomena. Havermans et al., [35] found that, while children are as reliable as adults on family conflict measures, they often underreport it. Since the conflict measure used in the ABCD Study broadly assesses family conflict rather than parent-child conflict specifically, it is plausible that parents hide conflicts from their children, leaving children unaware and unable to report them [35].

In addition, higher levels of parental monitoring were associated with lower levels of adolescent internalizing symptoms. This finding supports other studies examining parental monitoring as a protective factor for internalizing symptoms [13, 40, 69]. Parents who know their child’s whereabouts may be more involved in their child’s life, providing greater support and attunement to their child’s mental health [12]. Moreover, as the parent-child relationship changes during adolescence, youth seek more distance from their parents. With this newfound independence, adolescents may subject themselves to more risky situations and peers that can be detrimental to mental health outcomes. Having parents who allow independence while also setting limits shows the adolescent that the parent is both concerned and interested in their whereabouts. This promotes an internalized sense of care and attention for the child, which may reduce the development of internalizing symptoms [12].

It is worth noting that, while parent symptomatology, family conflict, and parental monitoring were associated with adolescent symptoms, the effect sizes were small (ΔR2 = 0.01), therefore limiting the extent to which conclusions can be drawn about these predictors. Nonetheless, this is consistent with contemporary thinking about the role of small effects, which are common across psychological science, given that psychological phenomena are often determined by a multitude of causes, with any individual factor likely to have only a small effect [31] .

Interestingly, we did not find an association between perceived parental acceptance and adolescent internalizing symptoms. We hypothesized that these two variables would be related, based on prior research showing that higher levels of parental warmth are associated with lower levels of adolescent internalizing symptoms [30, 59]. However, we did not find this result in our sample. A possible explanation is that other facets of parenting, such as monitoring, autonomy-granting, or harsh parenting compared to acceptance, may be more predictive of internalizing symptoms during adolescence [25, 56]. Research suggests that associations of warmth and internalizing symptoms tend to be stronger in concurrent studies than in longitudinal studies [56]. Thus, parental acceptance may not be as influential on internalizing symptoms over longer periods of development [26]. The null effect of parental acceptance may also reflect shared variance with other family factors. Since we did not test mediation among our predictor variables, potential indirect pathways may exist and should be explored in future research.

While there is evidence that adolescent sex can moderate the links between parenting and adolescent psychopathology [6], there was no statistically significant evidence supporting adolescent sex as a moderator in our sample. While these findings were somewhat surprising given some of the prior background literature, other ABCD-based studies have also found that family environment effects are robust across both genders, without observed moderation [64] It also is possible that other constructs related to our predictors, such as maladaptive emotion regulation strategies and the development of cognitive coping strategies, are more closely associated with sex differences in adolescent internalizing symptoms.

Several study limitations should be considered. First, the ABCD baseline sample primarily consists of early-developing males and late-developing females [14]. This limitation applies to all ABCD studies examining pubertal development, where information regarding the role of puberty is unavailable because of no prepubertal measurement available. Secondly, puberty is a complex developmental period with a myriad of specific subfactors that can influence the development of internalizing symptoms, such as pubertal timing and tempo (i.e., how advanced pubertal timing is). As this study focused on family factors across the pubertal transition, rather than examining puberty as a moderator, future research is also needed to test whether pubertal timing and tempo influence the effects of family factors on internalizing symptoms during adolescence. Third, the data available do not permit bidirectional influence, which the literature has shown is likely in the association between parent and child psychopathology [54]. All measures employed were self-report measures. Multi-informant approaches, especially those that include youth self-reports for symptomatology, could better elucidate constructs and provide a more reliable assessment. Finally, while the ABCD study retention rates remain strong, some participants did not complete the 4-year follow-up assessments. Because of problems with model convergence when using modern missing-data approaches commonly used in structural equation modeling, we acknowledge that missing data may affect study estimates.

Conclusion

This study is one of the first to examine the role of family factors in the longitudinal development of adolescent internalizing symptoms, using a large sample and across the pubertal transition. Consistent with previous research, family factors such as parent psychopathology and parenting practices are related to the development of internalizing symptoms. Findings from this study can inform prevention and intervention efforts by highlighting the potential value of reducing family conflict and addressing parental mental health to reduce the risk of adolescent internalizing symptoms. Additionally, parental monitoring may serve as a protective factor, emphasizing the importance of parenting practices in supporting adolescent mental health. Overall, our findings reflect modifiable targets that can improve youth well-being, and future research can further build upon it by seeing how these factors function in the context of broader structural and demographic variables.