Anxiety and depression are among the most prevalent mental disorders in youth (1). The estimated worldwide point prevalence of any anxiety and depressive disorder in youth is 6.5 and 2.6%, respectively (2), with a higher prevalence among girls than boys (3, 4). These disorders may impede development and well-being (1, 5, 6) and are associated with numerous negative mental health outcomes (7, 8).

Robust empirical data exist on the prevalence of anxiety and depression (hereafter internalizing disorders) in both population-based youth samples (2, 9) and clinical populations (4, 10). While research on the prevalence of internalizing disorders among youth in foster care is scarce, a meta-analysis estimated a pooled prevalence of 11% for any depressive and 18% for any anxiety disorder among youth in the child welfare system (11). Importantly, the comorbidity of mental disorders is high among youth in foster and residential care (12, 13). Comorbidity is found to be a negative predictor of treatment outcomes for internalizing disorders in clinically referred youth (10). However, there is a dearth of research on the profiles of depressive symptoms and anxiety symptom subtypes among youth in general and especially among youth in foster care. This leaves us unaware of whether the likelihood of experiencing various symptom profiles is similar or dissimilar across different groups of youths. Knowledge on load and distribution of symptoms across a wide range of internalizing problems among youth in foster care may have clinical implications for the conceptualization, assessment, and treatment outcomes of internalizing problems among at-risk youth, such as those in foster care.

Studies investigating the comorbidity of anxiety and depression are usually variable-centered, focusing on the associations of the variables and assuming homogeneity within the sample. Fewer studies have applied person-centered approaches like latent class analysis (LCA) or latent profile analysis (LPA), differentiated by the former examining categorical variables and the latter continuous variables. Person-centered analyses are based on underlying (latent) classes or subgroups, and these characteristics may be present as meaningful patterns or profiles (14); here the most likely profile membership of each youth is estimated (15). Person-centered analyses, such as LPA and LCA, are particularly helpful for the identification of possible similarities and dissimilarities across several internalizing symptoms in a heterogenous sample.

Only a few studies have conducted LPA/LCA on internalizing symptom subtypes and profiles (16); these have found that the profiles differ mainly on symptom levels rather than the shape of the symptom profile (17, 18). One exception is the LCA study of Burstein et al. (19) of sex- and age-specific structures and comorbidity in lifetime anxiety disorders among US adolescents (N = 2,539; M = 15.2 years; range, 13–18). The authors identified classes of anxiety subtypes, where one specific anxiety subtype (e.g., separation anxiety or social phobia) peaked in symptom level among youth in a population-based sample. In the same study, they found few classes of multiple anxiety disorders consisting of same-level symptoms. These latter classes were associated with more clinical severity, impairment, and comorbidity; sex differences were found in a variety of anxiety subtypes but not across age groups (19). The study of Burstein et al. (19) focused on lifetime anxiety disorders; however, retrospective responses may have influenced their findings. Furthermore, other studies using LCA on general adolescent samples mainly find anxiety and depression symptoms with profiles differentiated by level of symptoms, such as low, moderate, and high/severe symptom profiles (17, 20). Clinically referred children and youth have been found to have higher levels of internalizing symptoms compared to non-referred peers in the general sample (20). Nevertheless, these studies are limited to general population youth samples or compare a clinical with a general population youth sample; studies including samples of youth in foster care are therefore lacking.

To our knowledge, no other studies have applied LPA on internalizing symptoms among youth in foster care, compared to clinical and general population youth samples. There is reason to believe that youth in foster care have a higher likelihood of belonging to profiles with a higher symptom level of internalizing symptoms compared to same-age general population youth samples. However, it is less certain as to whether they have a higher likelihood of belonging to specific symptom profiles compared to same-age clinical youth samples. Youth in foster care have often experienced several out-of-home care placements (21) and might have multiple maltreatment experiences prior to placement (22). Hence, compared to their clinical peers, youth in foster care are more likely to have been exposed to parental psychopathology and negative parenting behaviors, both associated with increased risk of child psychopathology (23). Furthermore, adverse care experiences are one of the risk factors shared between depression and anxiety (24). This raises the question of whether youth in foster care run a higher risk of belonging to specific profiles of internalizing symptoms compared to their non-placed peers.

Various studies have identified possible mechanisms for the associations between maltreatment and anxiety—for example, changes in brain functioning (25), neuropsychological functioning (26), sleep (27), and cognition (28). These studies do not explore differences in anxiety subtypes. However, an increase of certain anxiety symptom subtypes could be expected among foster youth compared to a general population youth sample. A previous study of youth in foster care found that social and generalized anxiety were the most frequent anxiety symptom subtypes. Exposure to sexual abuse was the only maltreatment type associated with an increase in the level of all anxiety symptom subtypes, whereas neglect yielded increased symptom levels for all anxiety subtypes, except separation anxiety and fear of physical injury (i.e., simple phobia). Physical/emotional abuse increased the symptom level for social anxiety and OCD (29). These findings indicate the relevance of investigating subtypes of anxiety, as this may be affected by the specific maltreatment experiences of the youth and manifest itself in the shape of the symptom profile for foster youth. Increased knowledge on these matters is also needed to discuss potential adjustments and improvements in interventions and treatment programs offered to youth in foster care.

If youth in foster care are more likely (1) to be comparable to a general population youth sample and (2) belong to a high symptom profile, one may hypothesize that interventions targeting internalizing problems work for all youth across care conditions. However, if clear group differences appear, standard interventions may need to be tailored to the individual needs of youth in foster care.

In summary, knowledge is lacking regarding the level and pattern of internalizing symptom subtypes among youth in foster care compared to their peers in clinical settings and in the general population.

The main objectives of this study are 2-fold: first, to examine the symptom profiles of self-reported anxiety and depression across three unique samples (youth in foster care, youth in clinical treatment for internalizing problems, and a general population youth sample); and second, we investigated whether these symptom profiles were predicted by affiliation to group (i.e., their specific sample) and age and whether there were potential interactions of sex.

This study used data from three unique samples investigating internalizing symptoms in youth: a foster care sample (30), a clinical sample (31), and a general population youth sample (32). The total sample consisted of 2,005 youth (53.1% girls) with a mean age of 13.96 years (SD = 1.19; range, 11–18).

This study investigated the symptom profiles of anxiety and depression in three samples of 11–18-year-olds: youth in foster care, youth clinically referred to specialized mental health services for treatment of anxiety and depression, and a general population youth sample. By using LPA, we identified three distinct profiles: one with low, one with medium, and one with high levels of internalizing symptoms, which also differed somewhat in shape. The latent symptom profiles were predicted by group affiliation: that is, when compared to the general population youth sample, youth in foster care were more likely to belong to the high symptom profile. Youth in the clinical sample were more likely to belong to the medium and high symptom profiles compared to youth in both the general population youth sample and the foster care sample. Furthermore, sex and age were related to symptom profiles, with girls being more likely to belong to the high symptom profile. The likelihood of having higher symptoms increased with age.

Our finding of three symptom profiles differentiated between low, medium, and high is consistent with previous findings: mainly that there are differences in levels rather than the shape or pattern of internalizing symptoms. Wadsworth et al. (20), for example, identified three levels of symptom profiles of anxiety and depression among both non-referred and clinically adolescents (20). Furthermore, van Lang et al. (17) found that groups of adolescents with both anxiety and depression had high, moderate, and low symptom levels (17). Others have also concluded that high levels of one specific type of anxiety were generally not concurrent with low levels of other anxiety disorders (18, 48).

In the present study, we found that youth in foster care reported high symptom levels compared to general youth; however, compared to clinical youth, they showed an overall lower symptom load (Table 4). Youth in the general population sample were identified as more likely to belong to the low symptom profile and the clinical youth sample to the high symptom profile. This is in line with other studies in which youth in general populations have been found to have lower symptoms of anxiety and depression and youth clinically referred with internalizing problems have higher symptom levels (20, 49). As noted earlier, however, similar studies on youth in foster care are lacking.

Wadsworth et al. (20) reported a smaller gap between the sexes in pre-adolescent years for anxiety and depression problems, while in adolescence this gap increased. Other studies have confirmed increased symptoms of depression (50) and certain anxiety subtypes such as generalized anxiety and social anxiety during adolescence (49). This is in line with our findings that internalizing symptom levels increased with older age.

Our finding that youth in foster care are more likely to belong in the high symptom profile compared to the general population youth sample strengthens the argument that internalized symptoms in youth entering foster care should be assessed. While elevated levels of internalizing symptoms may reflect a history of maltreatment and relational problems among youth in foster care (29), these may also reflect how demanding the transition into a new home may be for these youth. As they must establish new relationships away from their primary carers (51, 52), these challenges may trigger or maintain symptoms of anxiety or depression in youth in foster care. The development of internalizing problems in youths should also be viewed within the caregiving context. Several aspects should be considered here. Placement instability is often a reported concern, as it may contribute to disrupting the developmental growth and resilience in youths (53). Through the experiences of safe and caring adults, youths will strengthen their relational abilities in forming new relationships with other adults and peers. As youth in foster care often have a background of maltreatment, adapted family interventions may be necessary. The supervision of foster parents, focusing especially on how to provide care for children and youth who independently do not seek this, is found effective in reducing internalizing problems among children and youth in foster care (54).

Even when high levels of internalizing symptoms are reported, youth in foster care are often not referred to Child and Adolescent Mental Health Services (CAMHS) (30). This may be due to internalizing problems often being more concealed compared to other mental health symptoms in youth, for example, externalizing problems. Thus, these may be difficult for caregivers, caseworkers, or health care practitioners to detect. Factors within the youth themselves may also hinder their anxiety or depression from being noticed—for example, youth may lack mental health literacy or see their long-term mental health symptoms as normal. Furthermore, coping with other life challenges—like a new foster home where they must get to know, trust, and build a relationship with foster parents and other family members—may feel more imperative than their mental health challenges (53, 54). Research has also identified barriers for help-seeking behaviors among young people, in general, which include fear of stigma and preference for self-reliance (55). These are barriers that might be especially relevant for youth in foster care due to care experiences that increase the risk of a reserved relational strategy such as impaired ability to seek and accept comfort (22). Youth in foster care may therefore benefit from an initial screening process for internalizing symptoms.

Our findings regarding sex differences in symptom levels among youth in the clinical sample are in line with other studies showing that girls in clinical samples report more internalizing problems than boys (19, 20). In the foster care sample, sex differences were not found for internalizing problems during childhood [6–12 years (12)] but seem to have emerged during their teenage years as found in the present study. These findings are consistent with studies on general and clinical youth populations confirming that, with increasing age, sex differences appear in the prevalence of internalizing problems (56–58). Furthermore, in the present study, girls in foster care were more likely to have medium or high symptom profiles compared to boys in both the general population sample and in the foster care sample. Girls in foster care also seemed to have a profile more similar to clinically referred boys in that both groups were more inclined to belong to the high symptom profile. However, boys in the general and foster care samples seemed to have higher probabilities of belonging to the low symptom profile. These findings indicate that there are important differences in symptom load across a specter of internalizing problems depending on sex among youth in foster care.

Our findings have implications for what the dominant focus of mental health services and foster parents should be. Regarding internalizing symptoms, a thorough assessment of girls in foster care is needed. A previous study reported sex differences in maltreatment exposures such as physical/emotional abuse, neglect, and sexual abuse among youth in foster care, with girls being more exposed (29). Furthermore, in this same study, girls in foster care were found to have higher anxiety symptoms when subjected to all three maltreatment types and higher depression symptoms when exposed to neglect compared to boys. This may indicate a greater vulnerability to internalizing symptoms among girls in foster care compared to boys (29).

Although we find clear differences regarding the levels of symptom profiles, fewer differences in the shape of the symptom profiles were identified. According to Dozois et al. (24), the psychopathology of parents seems to be a risk factor affecting a broad range of internalizing problems. This is related to our previous findings where different maltreatment types were associated with a broad range of internalizing symptoms (29). These findings may partly explain why, in the present study, we do not identify profiles with large differences in shape and that youth in foster care did not have a higher risk of belonging to a specific symptom profile compared to their peers in the clinical and population-based samples.

The most profound difference between the shapes of the three profiles, however, is the increase in panic/agoraphobia symptoms in the high symptom profile, whereas in the low and medium symptom profiles panic/agoraphobia was the rarest anxiety symptom subtype of anxiety. One might speculate that panic attacks may be more easily triggered by the other anxiety subtypes—e.g., agoraphobia, separation anxiety, and social anxiety—if the level is high enough. Panic attacks are typically characterized by incidents of intense fear and are accompanied by intense physiological symptoms (like heavy breathing, lightheadedness, and increased pulse) as well as cognitions like one is “about to die” or “become insane” (43). Research shows that impairing symptoms lead to clinical treatment, possibly due to the attention these receive from the caregivers of youth or other significant adults (44). Other studies show that panic attacks relate to help-seeking behavior due to the impairment they cause (45, 46): this may explain why the clinical youth had a higher likelihood of belonging to the high symptom profile, which has a higher panic/agoraphobia than the low symptom profile, to which youth from the general population sample are more likely to belong.

In the present study, the symptom subtype of generalized anxiety also had a relatively more pronounced increase in the high symptom profile of internalizing symptoms. Generalized anxiety covers a range of indiscriminate fears, where even worrying may be believed to be dangerous (47). It is also associated with physiological symptoms like headaches, muscle tension, and sleep problems over an extended period (43). These symptoms may influence important life areas such as school achievements and social relations. If this is the case, the symptoms may be impairing and may be a prioritized condition for admission to the CAMHS. However, whether panic and generalized anxiety are more impairing than other internalizing sub-scales (e.g., social anxiety or depressive symptoms) is not clear and needs further examination.

In the present study, sex and group affiliation predicted three symptom levels (low, medium, and high symptom profiles) across subtypes of internalizing symptoms. However, in line with Burstein et al. (19), we found a “complex” profile (our high symptom profile) to which girls in foster care and youth in clinical treatment had a higher probability of belonging. The present findings emphasize that girls, compared to boys, in foster care have a relatively high symptom load, but without major spikes in specific subtypes of internalizing symptoms. Moreover, their levels of depressive symptoms follow their levels of anxiety symptom subtypes and do not seem to be separate from the different anxiety symptom subtypes. This is in line with research indicating that anxiety is a predictor of depression (59, 60). It may seem that those suffering from internalizing symptoms have overall high symptom levels across subtypes, also including depressive symptoms. The finding that levels of anxiety and depressive symptoms follow each other in terms of symptom load across the three profiles may indicate that a transdiagnostic understanding of internalizing problems is important even if the profiles differed somewhat in shape (24, 61).

A strength of the present study is its comparison of three unique youth samples: a foster care sample, a general population youth sample, and a clinical sample. Each of the three studies, from which the sample of the present study was drawn, had methodological strengths that offer insights into the expression of youth anxiety and depressive symptoms. The total sample is large and heterogenous, generating solid results across different groups of youth, and with sufficient variance to run LPA. The total sample consists of youth who have completed identical measures of anxiety and depressive symptoms, with well-documented psychometric properties. Methodologically, LPA has the advantage of identifying latent profiles across groups and classifying the probability of membership of youth in these profiles compared to the others.

Although the present study has important strengths, it also has some limitations. One limitation concerns the final model decision. Although the fit indices pointed toward a four-profile solution, we decided on a three-profile solution. This was justified through our aims of choosing a model with profiles unique in their shape and level. Our decision was somewhat subjective as the fit indices were not consistent, and we concluded that adding more profiles would only lead to differences in level and not shape. However, our choice of a three-profile solution is also supported. The four-, five- and six-profile plots are shown in Figure 3 and reveal that increasing profiles almost exclusively contribute with changes in level. In the four-, five-, and six-profile solutions, the smallest classes consist of fewer than 5% of the participants, which is not considered ideal (47). Furthermore, the age range of the total sample (11–18) may be considered too wide and could possibly affect the analyses. We tested this by including only youth between ages 11–15 years old across the three samples in the LPA. However, the results were very similar and thus did not impact the results.

Another limitation is the difference in time points for the three periods of data collection: in particular, the large time difference between the assessment of the foster care sample (2016–2017) and the clinical sample (2008–2010). Possible time trends and events of social significance may impact the mental health of young people. Indeed levels of internalizing problems seem to have increased (especially for older girls) in the 21st century, worldwide (62), and in Norway (63, 64). Norwegian youth may have been influenced by the financial crisis in 2008, thus affecting the scores of the clinical sample; however, research indicates varying results regarding associations between socioeconomic status and mental health problems among youth in Norway (64). It is also possible that the internalizing symptom levels in the foster youth and the general population youth sample could have been influenced by the terrorist attacks in Oslo and Utøya in 2011: these seemed to particularly affect the mental health of Norwegian youth who had experienced prior adversity (65, 66).

A final limitation is that the response rate for the general population youth sample was relatively low.

The elevated risk among youth in foster care, particularly girls, to belong in the high-level profile of internalizing symptoms indicates that health services and child-protective services should be alert to the presence of internalizing symptoms and offer treatment when necessary. However, being in foster care per se is not sufficient to warrant access to assessment and treatment for internalizing disorders. An early-step care approach, with screening as a first step, could identify youths in foster care in need of more thorough assessment and treatment.

As at-risk youth may suffer more privately of internalizing symptoms, lacking the stability of concern of a caregiver, standardized early assessments and interventions may be even more crucial for children and youth in foster care. Timely assessment is essential to identify possible risk factors in the characteristics of a child (e.g., temperamental inhibition) and history (e.g., maltreatment exposures) as well as parental rearing styles and other factors conceivably maintaining the internalizing problems into adulthood (67). Considering that youth are in a critical neurobiological and socio-developmental phase in life, this may be a pivotal time for impacting further trajectories of internalizing problems (68).

An early assessment of risk factors, following the progression of the internalizing problems over time, could help the clinician to offer targeted effective interventions to the current stage of problems as early as possible (69). In a framework of a clinical staging model, clinicians integrate the biological, social, and contextual aspects of youth into their clinical and diagnostic understanding of the youth and how these factors may impact their problems (70). The treatment and interventions are consequently more personalized to the specific needs of the youth (71). A first international consensus statement for transdiagnostic clinical staging for youth mental health is proposed, stating its principles and assumptions (72). Although promising, the adoption and evaluation of such a framework remains to be seen for the foster care population, as this group is not automatically referred to mental health services.

However, it is unknown if youth in foster care with internalizing problems benefit from adjusted treatment approaches. Possible adjustments could be content-wise, for instance, by added focus on treating trauma and stress-related symptoms or enhancing specific skills such as problem-solving, relational functioning, or augmentation of specific procedures and/or increased duration of treatment. Youth in foster care might also benefit from treatment approaches identical to evidence-based methods found for youth who are clinically referred. Independent of group affiliation, the youth in all three samples reported a broad specter of internalizing symptoms, indicating the need of treatments targeting different internalizing symptoms rather than treatment focusing specifically on a single subtype of internalizing symptoms. The transdiagnostic perspective proposes that there is more flexibility in contributing more universal treatment interventions and prevention strategies across specific disorders (61).

Apart from individual treatment of internalizing symptoms, interventions involving the close family relations of the youth have been proven helpful (73). Thus, interventions should be adapted to the needs of the individual youth as well as their families (54). An example of this is the Norwegian intervention program, “The Scaffolders” (our translation) (74), where clinicians systematically work with youth in foster care with complex problems and specifically aims to cooperate with the foster families as well as with the child-protective services. Although the project appears promising, it is not fully evaluated. Another family-based intervention, named “Alternatives for Families: a Cognitive–Behavioral Therapy,” involves the family of children exposed to abuse and other traumatic events to facilitate change in destructive relational patterns (75). However, interventions targeting children with trauma exposures vary in content, structure (individual or group-based), frequency, and duration and show varying effects (76). More research is needed to evaluate the effects of interventions toward at-risk youth in foster care and their families.

Nonetheless, the results generated from our study cannot inform us if youth in foster care have a qualitatively different symptom profile with a unique set of comorbidities compared to clinically referred youth. Future research is encouraged to investigate these matters to ensure that youth in foster care receive treatment tailored to their needs.

The latent profile analyses on self-reported depressive symptoms and anxiety symptom subtypes across three youth samples (youth in foster care, youth in clinical treatment, and general population youth sample) show three symptom profiles. These profiles are characterized by differences in level and shape.

Prior exposure to maltreatment may be a leading factor in why youth in foster care are at a higher risk of belonging to the high symptom profile. Moreover, features within the foster care system itself may unintentionally contribute to a further increase in mental health problems in general, including internalizing problems—for instance, the consequences of a match or mismatch between the characteristics, needs, and capacities of the youth and the foster parents may affect a “failed” or “successful” placement (77, 78). The experience of foster parents, with follow-up and support from the child-protective services is identified to be essential in reducing the risk of placement disruption (78). Other systemic biases may therefore be placement instability and a high turnover among case workers, which can contribute to feelings of loneliness and a low or insecure sense of belonging. These may be altogether contributive factors for why youth in foster care are at a higher risk of belonging to the high symptom profile compared to their non-clinical peers.

It is critical that the services focus on the broad assessment of internalizing symptoms and early intervention, aiming to assess risk and protective factors in youth in foster care. Intervening early in the course of possible internalizing problems is crucial to prevent further disease progression. A transdiagnostic clinical staging model could be a promising framework for the management of these matters for youth in foster care. Overall, the findings indicate that current treatments may have to be tailored to the needs of youth in foster care.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by The Norwegian Regional Ethics Committee West. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

SL, GW, and BH conceived the idea of this research project. SL initiated the methodological framework. YM, KB, GW, BH, ML, and SL contributed to the conceptualization and the design of the study. YM under the supervision of KB, conducted the statistical analyses and interpreted and presented the results in this manuscript. YM wrote the original main draft and was the main contributor to the present manuscript. GW, BH, KB, and SL contributed to the manuscript. ML critically commented on a draft. GW and BH took part in the data collections for the clinical and general population youth sample. ML and SL took part in the data collection for the foster care sample. All authors approved the submitted version of the manuscript.

GW reports receiving consulting fees and lecture fees from Medice. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.