Hostname: page-component-848d4c4894-xfwgj Total loading time: 0 Render date: 2024-06-18T08:37:03.293Z Has data issue: false hasContentIssue false

Population heterogeneity in developmental trajectories of internalising and externalising mental health symptoms in childhood: differential effects of parenting styles

Published online by Cambridge University Press:  31 March 2023

Ioannis Katsantonis*
Affiliation:
Psychology, Education and Learning Studies Research Group, Faculty of Education, University of Cambridge, 184 Hills Rd, Cambridge CB2 8PQ, UK
Jennifer E. Symonds
Affiliation:
School of Education, University College Dublin, Roebuck Offices Belfield, Dublin 4, Ireland
*
Author for correspondence: Ioannis Katsantonis, E-mail: ik388@cam.ac.uk
Rights & Permissions [Opens in a new window]

Abstract

Aims

Multiple studies have connected parenting styles to children's internalising and externalising mental health symptoms (MHS). However, it is not clear how different parenting styles are jointly influencing the development of children's MHS over the course of childhood. Hence, the differential effects of parenting style on population heterogeneity in the joint developmental trajectories of children's internalising and externalising MHS were examined.

Method

A community sample of 7507 young children (ages 3, 5 and 9) from the Growing Up in Ireland cohort study was derived for further analyses. Parallel-process linear growth curve and latent growth mixture modelling were deployed.

Results

The results indicated that the linear growth model was a good approximation of children's MHS development (CFI = 0.99, RMSEA = 0.03). The growth mixture modelling revealed three classes of joint internalising and externalising MHS trajectories (VLMR = 92.51, p < 0.01; LMR = 682.19, p < 0.01; E = 0.86). The majority of the children (83.49%) belonged to a low-risk class best described by a decreasing trajectory of externalising symptoms and a flat low trajectory of internalising MHS. In total, 10.07% of the children belonged to a high-risk class described by high internalising and externalising MHS trajectories, whereas 6.43% of the children were probable members of a mild-risk class with slightly improving yet still elevated trajectories of MHS. Adjusting for socio-demographics, child and parental health, multinomial logistic regressions indicated that hostile parenting was a risk factor for membership in the high-risk (OR = 1.47, 95% CI 1.18–1.85) and mild-risk (OR = 1.57, 95% CI 1.21–2.04) classes. Consistent (OR = 0.75, 95% CI 0.62–0.90) parenting style was a protective factor only against membership in the mild-risk class.

Conclusions

In short, the findings suggest that a non-negligible proportion of the child population is susceptible to being at high risk for developing MHS. Moreover, a smaller proportion of children was improving but still displayed high symptoms of MHS (mild-risk). Furthermore, hostile parenting style is a substantial risk factor for increments in child MHS, whereas consistent parenting can serve as a protective factor in cases of mild-risk. Evidence-based parent training/management programmes may be needed to reduce the risk of developing MHS.

Type
Original Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
Copyright © The Author(s), 2023. Published by Cambridge University Press

Introduction

Persuasive evidence suggests that parenting styles significantly influence children's developmental outcomes and, especially, their internalising and externalising mental health symptoms (MHS) (Bayer et al., Reference Bayer, Sanson and Hemphill2006, Reference Bayer, Ukoumunne, Lucas, Wake, Scalzo and Nicholson2011b; Williams et al., Reference Williams, Degnan, Perez-Edgar, Henderson, Rubin, Pine, Steinberg and Fox2009; Lewis et al., Reference Lewis, Collishaw, Thapar and Harold2014; Braza et al., Reference Braza, Carreras, Muñoz, Braza, Azurmendi, Pascual-Sagastizábal, Cardas and Sánchez-Martín2015). Although research studies have examined the heterogeneity in developmental trajectories of MHS, the differential effects of parenting style on children's MHS trajectories remain largely overlooked. Preceding empirical evidence suggests that children's MHS trajectories are varying significantly from study to study (Fernandez Castelao and Kröner-Herwig, Reference Fernandez Castelao and Kröner-Herwig2014; Parkes et al., Reference Parkes, Sweeting and Wight2016; Patalay et al., Reference Patalay, Fink, Fonagy and Deighton2016, Reference Patalay, Moulton, Goodman and Ploubidis2017; Vaillancourt et al., Reference Vaillancourt, Haltigan, Smith, Zwaigenbaum, Szatmari, Fombonne, Waddell, Duku, Mirenda, Georgiades, Bennett, Volden, Elsabbagh, Roberts and Bryson2017; Flouri et al., Reference Flouri, Papachristou, Midouhas, Joshi, Ploubidis and Lewis2018; Papachristou and Flouri, Reference Papachristou and Flouri2020), however, most of these studies did not consider parenting styles collectively as predictors of children's trajectory profiles of internalising and externalising MHS.

Heterogeneous developmental trajectories of MHS

Previous research has investigated the developmental trajectories of children's MHS from a person-centred perspective by identifying heterogeneous subpopulations of children with specific characteristics in their MHS development. Most of the studies have usually found that the majority of children's MHS development was exhibiting a stable low trajectory of internalising and/or externalising MHS, in addition to multiple other trajectories that usually described stably increasing or moderate-increasing/decreasing trajectories of internalising and externalising MHS (Patalay et al., Reference Patalay, Deighton, Fonagy and Wolpert2015, Reference Patalay, Moulton, Goodman and Ploubidis2017; Parkes et al., Reference Parkes, Sweeting and Wight2016). Specific trends are emerging from these studies suggesting the existence of more than two distinct developmental trajectories (Miner and Clarke-Stewart, Reference Miner and Clarke-Stewart2008; Parkes et al., Reference Parkes, Sweeting and Wight2016; Patalay et al., Reference Patalay, Fink, Fonagy and Deighton2016; Vaillancourt et al., Reference Vaillancourt, Haltigan, Smith, Zwaigenbaum, Szatmari, Fombonne, Waddell, Duku, Mirenda, Georgiades, Bennett, Volden, Elsabbagh, Roberts and Bryson2017; Flouri et al., Reference Flouri, Papachristou, Midouhas, Joshi, Ploubidis and Lewis2018; Papachristou and Flouri, Reference Papachristou and Flouri2020).

It should be noted that the heterogeneity of the developmental trajectories of MHS may occur due to various reasons. Amongst these reasons may be the mixtures of different developmental periods, the nature of the sampling, the different analytic techniques, the socio-demographic composition of the samples and the measure of MHS used. In the present study, we utilised a national cohort sample of Irish children that covered a range of social-economic backgrounds. Hence, it was expected that some discrepancies between studies would emerge.

Nevertheless, a limitation of some preceding empirical works is that they examined internalising and externalising symptoms separately (Miner and Clarke-Stewart, Reference Miner and Clarke-Stewart2008; Fernandez Castelao and Kröner-Herwig, Reference Fernandez Castelao and Kröner-Herwig2014; Patalay et al., Reference Patalay, Deighton, Fonagy and Wolpert2015; Parkes et al., Reference Parkes, Sweeting and Wight2016). Research studies have noted that internalising and externalising MHS display some comorbidity where a common trait underlies exposure to both domains of psychopathology (Willner et al., Reference Willner, Gatzke-Kopp and Bray2016). Evidence coming from behaviour genetics (Cosgrove et al., Reference Cosgrove, Rhee, Gelhorn, Boeldt, Corley, Ehringer, Young and Hewitt2011; Lahey et al., Reference Lahey, Van Hulle, Singh, Waldman and Rathouz2011) and survey (Goodman et al., Reference Goodman, Lamping and Ploubidis2010; D'Urso and Symonds, Reference D'Urso and Symonds2022; Katsantonis, Reference Katsantonis2022) research suggests that internalising and externalising MHS are strongly correlated. Hence, we aimed to identify common MHS subtypes of internalising and externalising developmental trajectories in order to examine the degree of comorbidity between these broad types of MHS.

Parenting styles predicting developmental trajectories of MHS

Parenting styles are very important for children's typical development since they play a major role in creating an affective climate that serves as the foundation of the parent–child interactions (Darling and Steinberg, Reference Darling and Steinberg1993). They are connected to multiple developmental outcomes, and can be either a risk or a protective factor for children's MHS (Bayer et al., Reference Bayer, Sanson and Hemphill2006, Reference Bayer, Ukoumunne, Lucas, Wake, Scalzo and Nicholson2011b; Lewis et al., Reference Lewis, Collishaw, Thapar and Harold2014; Braza et al., Reference Braza, Carreras, Muñoz, Braza, Azurmendi, Pascual-Sagastizábal, Cardas and Sánchez-Martín2015). Thus, rightly the UNICEF refers to parenting as fundamental for child mental health (United Nations Children's Fund, 2021).

Different parenting styles have been connected to children's MHS in different ways. For example, warm or warm-engaged parenting style was found to be a significant predictor of reduced child MHS (Buschgens et al., Reference Buschgens, van Aken, Swinkels, Ormel, Verhulst and Buitelaar2010; Bayer et al., Reference Bayer, Ukoumunne, Lucas, Wake, Scalzo and Nicholson2011b; Parkes et al., Reference Parkes, Sweeting and Wight2016), whilst over-involved/protective, harsh and disciplinary parenting are predictors of increased MHS (Bayer et al., Reference Bayer, Ukoumunne, Lucas, Wake, Scalzo and Nicholson2011b; Lewis et al., Reference Lewis, Collishaw, Thapar and Harold2014). Similarly, studies have found that hostility was linked with a constellation of affective and behavioural MHS in toddlers (Dwairy, Reference Dwairy2008; Bayer et al., Reference Bayer, Ukoumunne, Lucas, Wake, Scalzo and Nicholson2011b; Schwerdtfeger et al., Reference Schwerdtfeger, Larzelere, Werner, Peters and Oliver2013). Hence, depending on the frequency and the nature of the parenting style, it may be either a substantial risk or protective factor for the development of children's MHS.

Although different types of parenting styles have been linked with greater or lower likelihood for development of internalising and externalising MHS, it is striking that few studies have estimated the differential effects of parenting styles on children's heterogeneous developmental trajectories of MHS. For example, Weeks et al. (Reference Weeks, Cairney, Wild, Ploubidis, Naicker and Colman2014) found that hostile parenting style predicted membership in the onset trajectory of internalising MHS compared to the low-stable trajectory. Likewise, Parkes et al. (Reference Parkes, Sweeting and Wight2016) reported that warm parenting style predicted low likelihood of membership in the trajectory of high and decreasing internalising MHS compared to the stable-low trajectory. Since different parenting styles may co-occur in the same parent (Heberle et al., Reference Heberle, Briggs-Gowan and Carter2015), it is important to collectively explore the impact of various parenting styles on children's MHS development.

Aims and hypotheses

Some evidence gaps were identified regarding the differential effects of parenting styles on children's developmental trajectories of MHS. Specifically, previous studies have not thoroughly examined the effects of parenting styles on the population heterogeneity in MHS trajectories across childhood. Additionally, many preceding studies examined the trajectories of internalising and externalising MHS separately or without considering how the trajectories of internalising MHS may influence the trajectories of externalising MHS, and vice versa. Thus, we aimed to address these gaps by deploying a person-centred approach (Muthén and Muthén, Reference Muthén and Muthén2000) along with robust national longitudinal data. Hence, the following hypotheses guided the present study. It was hypothesised that the analysis of heterogeneity in children's joint MHS trajectories would reveal more than two distinct developmental trajectories (H1). Further, it was expected that the developmental trajectories of internalising and externalising MHS would be correlated (H2). Finally, it was hypothesised that, adjusting for covariates, warm and consistent parenting styles would predict membership in the subpopulations with decreasing trajectories (H3), whilst hostile parenting style would be linked with subpopulations of accelerating trajectories (H4).

Method

Sample and dataset

Data come from three of the main waves of the Growing Up in Ireland infant cohort study (McNamara et al., Reference McNamara, O'Mahony and Murray2020). The cohort study had five waves of data available. At 9 months old, 11 134 children and their families participated, whilst 9793 children (88% response rate) participated at age 3 (wave 2). In total, 9001 children (80.8%) participated in wave 3, whereas 8032 children participated in wave 5 (72.1%). Although another wave of data is available at age 7/8, this wave's data collection was conducted through a postal questionnaire and the response rate was 48% (McNamara et al., Reference McNamara, O'Mahony and Murray2020). Since the sampling weights do not permit representative analysis of all the waves (Quail et al., Reference Quail, O'Reilly, Watson, McNamara, O'Mahony and Murray2019), the longitudinally matched sample size amounted to 7507 children aged 3 years old (wave 2) that were followed at age 5 (wave 3) and at age 9 (wave 5).

Measures

Internalising and externalising MHS

The four negative symptom scales from the Strengths and Difficulties Questionnaire (SDQ) (Goodman, Reference Goodman1997, Reference Goodman2001; Goodman et al., Reference Goodman, Ford, Simmons, Gatward and Meltzer2000) were administered to the cohort. Parent-reported (>98% mothers) scores were used in this study. Despite that the SDQ is a broad screening measure of MHS, the scales show good sensitivity, specificity (Cocker et al., Reference Cocker, Minnis and Sweeting2018) and accuracy in predicting common mental disorders (Goodman et al., Reference Goodman, Ford, Simmons, Gatward and Meltzer2000; Goodman and Goodman, Reference Goodman and Goodman2011). According to the SDQ validation studies, the summed composite of the emotional symptoms and peer problems scales forms the internalising symptoms score, whilst the composite of the conduct problems and hyperactivity scales forms the externalising score (Goodman et al., Reference Goodman, Lamping and Ploubidis2010). All composite scores range from 1 to 10. The scales displayed good reliability across the survey's waves (McCrory et al., Reference McCrory, Williams, Murray, Quail and Thornton2013; McNamara et al., Reference McNamara, O'Mahony and Murray2020).

Parenting styles

Three scales adopted from the Longitudinal Study of Australian Children (LSAC) were indexing parenting styles (McNamara et al., Reference McNamara, O'Mahony and Murray2020). Composite scores on warm (six items – display of affection, awareness of child's needs), consistent (five items – consistent expectations and applications of rules) and hostile (six items – overcontrolling, discipline) parenting styles ranging from 1 to 5 were available. The scales displayed good reliability across the survey's waves (McCrory et al., Reference McCrory, Williams, Murray, Quail and Thornton2013; McNamara et al., Reference McNamara, O'Mahony and Murray2020). Parenting styles from wave 2 were included in the modelling.

Covariates

Drawing upon a developmental and psychopathology multisystem resilience approach (Masten et al., Reference Masten, Lucke, Nelson and Stallworthy2021), three types of covariates, namely socio-demographics, child health and parental health, from wave 2 (see Table 1) were included in the modelling since they have been identified either as risk or protective factors.

Table 1. Sample descriptive statistics for all cohort members

Note: ref, reference category; data were parent-reported by the primary caregiver (>0.98% mothers).

Analysis strategy

Preliminary data management was conducted through Stata 16 (StataCorp., 2019), whilst latent variable modelling and multinomial regressions were conducted with Mplus 8.7 (Muthén and Muthén, Reference Muthén and Muthén2017). A two-step strategy was followed in the modelling. In the first instance, we estimated a linear parallel-process growth model (LGM) with correlated two intercepts and two growth factors defined by the composite internalising and externalising MHS scores. Within-wave residual correlations were estimated per the suggestions of the modification indices. After evaluating the fit of the model, we opted for a person-centred approach to identify population heterogeneity in developmental trajectories of MHS. To this end, a parallel-process growth mixture model (GMM) with equal variances (the Mplus default) across classes was specified and estimated (Muthén and Muthén, Reference Muthén and Muthén2000; Jung and Wickrama, Reference Jung and Wickrama2008). Missing data were handled with full-information maximum likelihood (Enders, Reference Enders2022). After determining the number of latent classes, the most probable class membership was extracted and multinomial regressions were estimated to examine the effects of parenting styles, whilst controlling for the effects of covariates. The longitudinal sampling weight (including non-response adjustment) at age 9 was applied to all inferential analyses.

All models were estimated with the robust maximum likelihood estimator. To evaluate the fit of the LGM, the conventional thresholds in goodness-of-fit indices were utilised. Specifically, values above/close to 0.95 in the Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) in conjunction with values less than 0.06 and 0.08 in the Root Mean Square Error of Approximation (RMSEA) and the Standardised Root Mean Residual (SRMR), respectively, are considered indicators of good fit to the sample covariance matrix (Hu and Bentler, Reference Hu and Bentler1999). To decide on the number of latent classes, we considered a combination of fit indices in GMM, and the clinical and practical interpretability of the classes. Specifically, we consulted the entropy, the Akaike Information Criterion (AIC), the Bayesian Information Criterion (BIC), the sample-adjusted Bayesian Information Criterion (sBIC), the Vuong–Lo–Mendel–Rubin (VLMR) and the adjusted Lo–Mendel–Rubin (LMR) likelihood tests. Entropy values above 0.8 are preferred, whereas the AIC and (s)BIC values are used to compare two nested models for parsimony (Nylund et al., Reference Nylund, Asparouhov and Muthén2007). Given that the BIC is known to persist decreasing with additional classes, the BIC values were plotted to identify the point (‘elbow’) of diminishing returns (Nylund-Gibson and Choi, Reference Nylund-Gibson and Choi2018). Finally, the VLMR and LMR are likelihood ratio tests that compare the model with n–1 classes with the model with n classes and provide p-values. Non-significant likelihood ratio tests indicate that the model with n classes does not offer an improvement in model-data fit compared to the n–1 classes model (Nylund et al., Reference Nylund, Asparouhov and Muthén2007; Ferguson et al., Reference Ferguson, Moore and Hull2020). To ascertain the extent to which the correct number of classes was identified and the best loglikelihood replicated, a large number of 500 random starting values was utilised and more latent classes were extracted.

Results

Sample descriptive statistics, such as means, SDs, or percentages, for key outcomes and covariates are presented in Table 1.

As a reference only, the sample was distributed into the age-appropriate cut-off centiles based on the new fourfold classification of the scores on the SDQ scales in the UK (www.SDQinfo.org). As can be seen in Table 2, most children's scores on all SDQ scales were close to average with fewer children scoring in the higher centiles.

Table 2. Distribution of the cohort sample in each centile cut-off of mental health symptoms based on the SDQ fourfold classification

a Listwise sample size; 80% centile: close to average; 90/92% centile: slightly raised; 95/96% centile: high; 99.99% centile: very high; age-appropriate cut-offs were derived from www.SDQinfo.org.

Heterogeneity in MHS trajectories

An intercept-only model was found to be degraded compared to the intercept-slope model, Satorra–Bentler χ 2(9) = 926.16, p < 0.001. The linear parallel-process LGM (intercept-slope) exhibited great fit to the data, i.e. scaled χ 2(4) = 30.22, p < 0.001, CFI = 0.996, TLI = 0.98, RMSEA = 0.03, SRMR = 0.013. Thus, we could be confident that a linear model of change could be a good approximation for the MHS trajectories. The LGM parameters are presented in the Supplementary materials. Next, we estimated the GMM to identify possible heterogeneity in the trajectories of MHS. The model with three classes of MHS trajectories exhibited significant improvement compared to the two-class model. Moreover, the likelihood ratio tests revealed that the addition of a fourth class did not result in improved model-data fit (p > 0.05). The ‘elbow’ plot of the BIC values showed that adding more than three classes did not improve fit substantially (see Supplementary materials). Additionally, we estimated further models where the variances of the intercept and slope parameters were freely varying between classes. However, the variance–covariance matrices were not positive definite, and thus, the variances were held equal across classes. Keeping in mind that this was a general population sample, we did not expect extreme variation in mental health symptomatology, and thus, the three-class model was retained. Fit indices for the models are presented in Table 3.

Table 3. Fit indices for the parallel-process growth mixture models

Note: ***p < 0.001; **p < 0.01; the bootstrap LRT is not available with sampling weights.

The model revealed that 83.49% (N = 6268) of the children belonged to class 1 (low risk) that was characterised by a flat low trajectory of internalising MHS, and a declining trajectory of externalising MHS. In contrast, 10.07% (N = 756) of the children were members of a high-risk class 2 that was characterised by an accelerating high trajectory of internalising and externalising MHS. Finally, 6.43% (N = 483) of the children belonged to class 3 with MHS trajectories that were slightly improving but still elevated (mild-risk). The estimated trajectories are shown in Fig. 1.

Fig. 1. Estimated internalising (left) and externalising (right) MHS trajectories.

The growth parameters are presented in Table 4. The internalising intercept was negatively associated with the slope of externalising MHS, r = −0.15, p < 0.01. Higher externalising MHS at age 3 were associated with a decelerating externalising MHS development, r = −0.28, p < 0.001. In contrast, children's internalising scores underwent a substantial shift between ages 3 and 9, r = 0.09, p > 0.05. The externalising and internalising intercepts were positively associated indicating that children with higher initial status in internalising MHS exhibited high initial status in externalising MHS, r = 0.49, p < 0.001. Finally, the two slopes were also positively correlated, r = 0.46, p < 0.001, indicating that higher internalising trajectories were associated with higher externalising trajectories.

Table 4. Growth parameters for the parallel-process growth mixture model

Note: ***p < 0.001; **p < 0.01; INT, internalising; EXT, externalising; equal σ 2 across classes.

Effects of parenting styles on MHS classification

Having established three heterogeneous trajectories of MHS, the effects of the three parenting styles on children's MHS trajectories were estimated using multinomial logistic regressions adjusting for the covariates. To ascertain the extent to which multicollinearity may have occurred, the variance inflation factors (VIF) were inspected and found to be less than 6 (mean VIF = 1.84). Thus, no extreme collinearity was diagnosed for the model. The results of the multinomial logistic regressions are presented in Table 5.

Table 5. Multinomial logistic regressions predicting high-risk class 2 and mild-risk class 3 membership v. low-risk class 1 membership

Note: ref, reference category; OR, odds ratio; 95% CI, 95% confidence interval; N = 7507; values in bold reached statistical significance according to the 95% CI. Statistical significance did not vary between the logit and OR coefficients.

Hostile parenting style was a substantial predictor of membership in the high-risk class (OR = 1.47, 95% CI 1.18–1.85) and the mild-risk class (OR = 1.57, 95% CI 1.21–2.04). Warm parenting style did not predict membership in the high-risk (OR = 1.30, 95% CI 0.96–1.75) or the mild-risk (OR = 0.74, 95% CI 0.55–1.01) classes. Although consistent parenting style did not predict membership in the high-risk class (OR = 0.87, 95% CI 0.74–1.02), it was a protective factor against membership in the mild-risk class (OR = 0.75, 95% CI 0.62–0.90). The results for the covariates are discussed in the Supplementary materials.

Discussion

The person-centred approach revealed some extent of heterogeneity in children's MHS trajectories. Contrary to most preceding evidence (Patalay et al., Reference Patalay, Deighton, Fonagy and Wolpert2015, Reference Patalay, Fink, Fonagy and Deighton2016, Reference Patalay, Moulton, Goodman and Ploubidis2017; Vaillancourt et al., Reference Vaillancourt, Haltigan, Smith, Zwaigenbaum, Szatmari, Fombonne, Waddell, Duku, Mirenda, Georgiades, Bennett, Volden, Elsabbagh, Roberts and Bryson2017; Flouri et al., Reference Flouri, Papachristou, Midouhas, Joshi, Ploubidis and Lewis2018; Papachristou and Flouri, Reference Papachristou and Flouri2020), only three joint trajectories of internalising and externalising MHS were identified. Thus, our H1 was retained. In other words, despite a substantial number of previous studies indicating more than three distinct classes of MHS trajectories, the present modelling identified that 10.07% of the children followed a MHS developmental trajectory that was best described as high-risk with characteristics of accelerating high MHS trajectories. Another smaller class of children (6.43%) was characterised by improving yet still elevated trajectories of MHS, and was, thus, defined as mild-risk class. In short, most Irish children (83.49%) did not seem to be at-risk of developing some MHS. The majority of the Irish children's MHS development was following a decreasing externalising trajectory and a stable flat-low internalising trajectory. These findings seem to agree with psychiatric epidemiological evidence suggesting decreasing trends in children's internalising and externalising MHS (Maughan et al., Reference Maughan, Collishaw, Meltzer and Goodman2008; Sellers et al., Reference Sellers, Maughan, Pickles, Thapar and Collishaw2015; Pitchforth et al., Reference Pitchforth, Fahy, Ford, Wolpert, Viner and Hargreaves2019).

As mentioned, most of the preceding evidence did not examine (Miner and Clarke-Stewart, Reference Miner and Clarke-Stewart2008; Fernandez Castelao and Kröner-Herwig, Reference Fernandez Castelao and Kröner-Herwig2014; Patalay et al., Reference Patalay, Deighton, Fonagy and Wolpert2015; Parkes et al., Reference Parkes, Sweeting and Wight2016) or discuss in depth the correlated nature of the MHS trajectories. Previous studies have reported inconclusive results regarding the correlated nature of the internalising and externalising MHS with some studies reporting low or non-significant correlations (Rhee et al., Reference Rhee, Lahey and Waldman2015), whereas other studies showed significant comorbidity across time (Willner et al., Reference Willner, Gatzke-Kopp and Bray2016). Hence, in the current study, we clarified this through a developmental perspective by exploring the association between the growth parameters of the internalising and externalising trajectories. This resulted in a longitudinal estimate of a moderate degree of comorbidity between the development of these broad MHS in this community sample. Thus, H2 was confirmed.

Last but not least, the present study sought to evaluate the impact of parenting styles on the heterogeneity in children's MHS developmental trajectories. Using a developmental and psychopathology multisystem resilience approach (Masten et al., Reference Masten, Lucke, Nelson and Stallworthy2021), we controlled for various other variables that have been linked with children's (mal-)adaptive functioning. After adjusting for several socio-demographic factors and child and parental health, the multinomial regressions illustrated that hostile parenting was a significant risk factor for the development of MHS. In précis, the model revealed that hostile parenting was predicting greater odds of membership in the high- and mild-risk classes of MHS trajectories. These findings are congruent with extant research that found a significant longitudinal effect of hostility and consistency on children's MHS (Bayer et al., Reference Bayer, Rapee, Hiscock, Ukoumunne, Mihalopoulos and Wake2011a; Lewis et al., Reference Lewis, Collishaw, Thapar and Harold2014; Parkes et al., Reference Parkes, Sweeting and Wight2016). In contrast, consistent parenting style was a protective factor only against being at mild-risk for developing MHS. Despite evidence suggesting a protective effect of warm parenting style on children's MHS development (Buschgens et al., Reference Buschgens, van Aken, Swinkels, Ormel, Verhulst and Buitelaar2010; Bayer et al., Reference Bayer, Ukoumunne, Lucas, Wake, Scalzo and Nicholson2011b; Parkes et al., Reference Parkes, Sweeting and Wight2016), its longitudinal effect on MHS was not significant in our modelling. Thus, H3 was partially rejected and H4 was confirmed. The differential contributions of the parenting styles cement the theoretical tenant that several parenting styles may co-exist in a single family (Heberle et al., Reference Heberle, Briggs-Gowan and Carter2015). To the best of our knowledge, no previous study has examined collectively the effects of warm, consistent and hostile parenting styles on children's heterogeneous profiles of MHS trajectories. Regarding the covariates, in both the high- and mild-risk classes of MHS, we found that children's and parental general health, as well as the household structure and parental stress were consistently non-negligible risk factors for MHS. Being a female child predicted greater odds of membership in the high-risk class, whereas higher income was a protective factor against high-risk MHS confirming the social gradient in MHS.

Overall, the present study is distinct from preceding works (Fernandez Castelao and Kröner-Herwig, Reference Fernandez Castelao and Kröner-Herwig2014; Patalay et al., Reference Patalay, Moulton, Goodman and Ploubidis2017; Flouri et al., Reference Flouri, Papachristou, Midouhas, Joshi, Ploubidis and Lewis2018; Papachristou and Flouri, Reference Papachristou and Flouri2020) in the sense that it focused only on the developmental period of childhood. This allowed us to identify distinct classes of MHS developmental trajectories that were influenced only by the developmental characteristics of childhood. Additionally, a substantial part of previous evidence was based on cohort studies in the UK (Parkes et al., Reference Parkes, Sweeting and Wight2016; Patalay et al., Reference Patalay, Moulton, Goodman and Ploubidis2017; Flouri et al., Reference Flouri, Papachristou, Midouhas, Joshi, Ploubidis and Lewis2018; Papachristou and Flouri, Reference Papachristou and Flouri2020), which does not necessarily translate to other contexts. Finally, the present study's results may differ from preceding evidence given the analytic choice of a parallel-process modelling in contrast to previous approaches that modelled internalising and externalising MHS separately as distinct broad types of MHS (Fernandez Castelao and Kröner-Herwig, Reference Fernandez Castelao and Kröner-Herwig2014; Patalay et al., Reference Patalay, Deighton, Fonagy and Wolpert2015, Reference Patalay, Fink, Fonagy and Deighton2016; Parkes et al., Reference Parkes, Sweeting and Wight2016).

Nevertheless, the present work also had some limitations. Specifically, it should be underscored that SDQ is a broad screening tool of children and adolescents’ MHS and does not necessarily target specific symptoms of the most common mental disorders. Moreover, evidence suggests that SDQ may not have high criterion validity in predicting emotional and behavioural disorders (Aydin et al., Reference Aydin, Siebelink, Crone, van Ginkel, Numans, Vermeiren and Westenberg2022). Furthermore, given the limitations of the data at hand we did not have any further indicators of other parenting styles. Additionally, without more waves of data, it was not possible to evaluate models of non-linear growth. Finally, given that the data came exclusively from parent reports/interviews, this may suggest that social desirability may bias somewhat the results, especially, concerning the parenting styles.

Implications for policy and practice

The present findings may have significant implications for policy and practice. For instance, the correlated nature of the developmental trajectories of internalising and externalising MHS indicates that mental health professionals should always screen for both types of MHS. Furthermore, the person-centred modelling indicated that only a small proportion of the general population of young children may be at-risk for increased MHS. However, mental health practitioners should be vigilant to identify those children. Thus, more robust extensive screening may be needed. Most importantly, the present study's results underscored the importance of parenting styles for the identification of children at-risk for developing high/mild MHS. Therefore, we suggest that routine background checks of parenting behaviours should be applied while screening children for possible MHS. Given that hostile parenting style was associated with children's membership in the high- and mild-risk classes with high MHS trajectories, we recommend evidence-based parent training/management programmes to improve positive interactions between parent and child.

Supplementary material

The supplementary material for this article can be found at https://doi.org/10.1017/S2045796023000094.

Data

The Growing Up in Ireland infant cohort datasets, which were utilised in the present study, were accessed and are available via the Irish Social Science Data Archive – www.ucd.ie/issda. The authors have access to anonymised microdata files which include composite variables only.

Financial support

The research leading to these results has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 101008589, COORDINATE – COhort cOmmunity Research and Development Infrastructure Network for Access Throughout Europe. Ioannis Katsantonis is supported by the Alexander S. Onassis Foundation (scholarship ID: F ZR024/1-2021/2022) and the A.G. Leventis Foundation.

Conflict of interest

None.

Ethical standards

The protocols of the Growing Up in Ireland study have received ethical approval by the Research Ethics Committee convened by the Irish Department of Children and Youth Affairs. Written informed consent was obtained from the primary caregiver of each child.

References

Aydin, S, Siebelink, BM, Crone, MR, van Ginkel, JR, Numans, ME, Vermeiren, RRJM and Westenberg, PM (2022) The diagnostic process from primary care to child and adolescent mental healthcare services: the incremental value of information conveyed through referral letters, screening questionnaires and structured multi-informant assessment. BJPsych Open 8, E81. doi: 10.1192/bjo.2022.47CrossRefGoogle ScholarPubMed
Bayer, JK, Sanson, AV and Hemphill, SA (2006) Parent influences on early childhood internalizing difficulties. Journal of Applied Developmental Psychology 27, 542559.CrossRefGoogle Scholar
Bayer, JK, Rapee, RM, Hiscock, H, Ukoumunne, OC, Mihalopoulos, C and Wake, M (2011 a) Translational research to prevent internalizing problems early in childhood. Depression and Anxiety 28, 5057.CrossRefGoogle ScholarPubMed
Bayer, JK, Ukoumunne, OC, Lucas, N, Wake, M, Scalzo, K and Nicholson, JM (2011 b) Risk factors for childhood mental health symptoms: national longitudinal study of Australian children. Pediatrics 128, e865e879.CrossRefGoogle ScholarPubMed
Braza, P, Carreras, R, Muñoz, JM, Braza, F, Azurmendi, A, Pascual-Sagastizábal, E, Cardas, J and Sánchez-Martín, JR (2015) Negative maternal and paternal parenting styles as predictors of children's behavioral problems: moderating effects of the child's sex. Journal of Child and Family Studies 24, 847856.CrossRefGoogle Scholar
Buschgens, CJM, van Aken, MAG, Swinkels, SHN, Ormel, J, Verhulst, FC and Buitelaar, JK (2010) Externalizing behaviors in preadolescents: familial risk to externalizing behaviors and perceived parenting styles. European Child & Adolescent Psychiatry 19, 567575.CrossRefGoogle ScholarPubMed
Cocker, C, Minnis, H and Sweeting, H (2018) Potential value of the current mental health monitoring of children in state care in England. BJPsych Open 4, 486491.CrossRefGoogle ScholarPubMed
Cosgrove, VE, Rhee, SH, Gelhorn, HL, Boeldt, D, Corley, RC, Ehringer, MA, Young, SE and Hewitt, JK (2011) Structure and etiology of co-occurring internalizing and externalizing disorders in adolescents. Journal of Abnormal Child Psychology 39, 109123.CrossRefGoogle ScholarPubMed
Darling, N and Steinberg, L (1993) Parenting style as context: an integrative model. Psychological Bulletin 113, 487496.CrossRefGoogle Scholar
D'Urso, G and Symonds, J (2022) Developmental cascades of internalizing and externalizing problems from infancy to middle childhood: longitudinal associations with bullying and victimization. Journal of School Violence 21, 294311.CrossRefGoogle Scholar
Dwairy, MA (2008) Parental inconsistency versus parental authoritarianism: associations with symptoms of psychological disorders. Journal of Youth and Adolescence 37, 616626.CrossRefGoogle Scholar
Enders, CK (2022) Applied Missing Data Analysis. New York: The Guilford Press.Google Scholar
Ferguson, SL, Moore, EWG and Hull, DM (2020) Finding latent groups in observed data: a primer on latent profile analysis in Mplus for applied researchers. International Journal of Behavioral Development 44, 458468.CrossRefGoogle Scholar
Fernandez Castelao, C and Kröner-Herwig, B (2014) Developmental trajectories and predictors of externalizing behavior: a comparison of girls and boys. Journal of Youth and Adolescence 43, 775789.CrossRefGoogle ScholarPubMed
Flouri, E, Papachristou, E, Midouhas, E, Joshi, H, Ploubidis, GB and Lewis, G (2018) Early adolescent outcomes of joint developmental trajectories of problem behavior and IQ in childhood. European Child & Adolescent Psychiatry 27, 15951605.CrossRefGoogle ScholarPubMed
Goodman, R (1997) The strengths and difficulties questionnaire: a research note. Journal of Child Psychology and Psychiatry 38, 581586.CrossRefGoogle ScholarPubMed
Goodman, R (2001) Psychometric properties of the strengths and difficulties questionnaire. Journal of the American Academy of Child & Adolescent Psychiatry 40, 13371345.CrossRefGoogle ScholarPubMed
Goodman, A and Goodman, R (2011) Population mean scores predict child mental disorder rates: validating SDQ prevalence estimators in Britain. Journal of Child Psychology and Psychiatry 52, 100108.CrossRefGoogle ScholarPubMed
Goodman, R, Ford, T, Simmons, H, Gatward, R and Meltzer, H (2000) Using the Strengths and Difficulties Questionnaire (SDQ) to screen for child psychiatric disorders in a community sample. British Journal of Psychiatry 177, 534539.CrossRefGoogle Scholar
Goodman, A, Lamping, DL and Ploubidis, GB (2010) When to use broader internalising and externalising subscales instead of the hypothesised five subscales on the strengths and difficulties questionnaire (SDQ): data from British parents, teachers and children. Journal of Abnormal Child Psychology 38, 11791191.CrossRefGoogle ScholarPubMed
Heberle, AE, Briggs-Gowan, MJ and Carter, AS (2015) A person-oriented approach to identifying parenting styles in mothers of early school-age children. Infant and Child Development 24, 130156.CrossRefGoogle Scholar
Hu, L and Bentler, PM (1999) Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Structural Equation Modeling 6, 155.CrossRefGoogle Scholar
Jung, T and Wickrama, KA (2008) An introduction to latent class growth analysis and growth mixture modeling. Social and Personality Psychology Compass 2, 302317.CrossRefGoogle Scholar
Katsantonis, I (2022) The mediating role of school and sibling bullying in the relationship between subjective well-being and mental health symptoms. Psych 4, 258268.CrossRefGoogle Scholar
Lahey, BB, Van Hulle, CA, Singh, AL, Waldman, ID and Rathouz, PJ (2011) Higher-order genetic and environmental structure of prevalent forms of child and adolescent psychopathology. Archives of General Psychiatry 68, 181189.CrossRefGoogle ScholarPubMed
Lewis, G, Collishaw, S, Thapar, A and Harold, GT (2014) Parent–child hostility and child and adolescent depression symptoms: the direction of effects, role of genetic factors and gender. European Child & Adolescent Psychiatry 23, 317327.CrossRefGoogle ScholarPubMed
Masten, AS, Lucke, CM, Nelson, KM and Stallworthy, IC (2021) Resilience in development and psychopathology: multisystem perspectives. Annual Review of Clinical Psychology 17, 521549.CrossRefGoogle ScholarPubMed
Maughan, B, Collishaw, S, Meltzer, H and Goodman, R (2008) Recent trends in UK child and adolescent mental health. Social Psychiatry and Psychiatric Epidemiology 43, 305310.CrossRefGoogle ScholarPubMed
McCrory, C, Williams, J, Murray, A, Quail, A and Thornton, M (2013) Growing Up in Ireland: Design, Instrumentation and Procedures for the Infant Cohort at Wave Two (3 Years). Ireland: Department of Children and Youth Affairs, p. 114.Google Scholar
McNamara, E, O'Mahony, D and Murray, A (2020) Design, Instrumentation and Procedures for Cohort ‘08 of Growing Up in Ireland at 9 Years Old (Wave 5). Ireland: Department of Children and Youth Affairs. Available at https://www.growingup.ie/pubs/Cohort08at9-Design-Report-2020-1.pdf.Google Scholar
Miner, JL and Clarke-Stewart, KA (2008) Trajectories of externalizing behavior from age 2 to age 9: relations with gender, temperament, ethnicity, parenting, and rater. Developmental Psychology 44, 771786.CrossRefGoogle ScholarPubMed
Muthén, B and Muthén, LK (2000) Integrating person-centered and variable-centered analyses: growth mixture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research 24, 882891.CrossRefGoogle ScholarPubMed
Muthén, LK and Muthén, BO (2017) Mplus User's Guide, 8th Edn. Los Angeles, CA: Muthén & Muthén.Google Scholar
Nylund-Gibson, K and Choi, AY (2018) Ten frequently asked questions about latent class analysis. Translational Issues in Psychological Science 4, 440461.CrossRefGoogle Scholar
Nylund, KL, Asparouhov, T and Muthén, BO (2007) Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Structural Equation Modeling: A Multidisciplinary Journal 14, 535569.CrossRefGoogle Scholar
Papachristou, E and Flouri, E (2020) Distinct developmental trajectories of internalising and externalising symptoms in childhood: links with mental health and risky behaviours in early adolescence. Journal of Affective Disorders 276, 10521060.CrossRefGoogle ScholarPubMed
Parkes, A, Sweeting, H and Wight, D (2016) Early childhood precursors and school age correlates of different internalising problem trajectories among young children. Journal of Abnormal Child Psychology 44, 13331346.CrossRefGoogle ScholarPubMed
Patalay, P, Deighton, J, Fonagy, P and Wolpert, M (2015) The relationship between internalising symptom development and academic attainment in early adolescence. PLoS ONE 10, e0116821.CrossRefGoogle ScholarPubMed
Patalay, P, Fink, E, Fonagy, P and Deighton, J (2016) Unpacking the associations between heterogeneous externalising symptom development and academic attainment in middle childhood. European Child & Adolescent Psychiatry 25, 493500.CrossRefGoogle ScholarPubMed
Patalay, P, Moulton, V, Goodman, A and Ploubidis, GB (2017) Cross-domain symptom development typologies and their antecedents: results from the UK millennium cohort study. Journal of the American Academy of Child & Adolescent Psychiatry 56, 765776.e2.CrossRefGoogle ScholarPubMed
Pitchforth, J, Fahy, K, Ford, T, Wolpert, M, Viner, RM and Hargreaves, DS (2019) Mental health and well-being trends among children and young people in the UK, 1995–2014: analysis of repeated cross-sectional national health surveys. Psychological Medicine 49, 12751285.CrossRefGoogle Scholar
Quail, A, O'Reilly, C, Watson, D, McNamara, E, O'Mahony, D and Murray, A (2019) A Summary Guide to Cohort ‘08 at 9 Years (Wave 5 of the Infant Cohort at 9 Years). Ireland: Department of Children and Youth Affairs. Available at https://www.growingup.ie/data-documentation/.Google Scholar
Rhee, SH, Lahey, BB and Waldman, ID (2015) Comorbidity among dimensions of childhood psychopathology: converging evidence from behavior genetics. Child Development Perspectives 9, 2631.CrossRefGoogle ScholarPubMed
Schwerdtfeger, KL, Larzelere, RE, Werner, D, Peters, C and Oliver, M (2013) Intergenerational transmission of trauma: the mediating role of parenting styles on toddlers’ DSM-related symptoms. Journal of Aggression, Maltreatment & Trauma 22, 211229.CrossRefGoogle Scholar
Sellers, R, Maughan, B, Pickles, A, Thapar, A and Collishaw, S (2015) Trends in parent- and teacher-rated emotional, conduct and ADHD problems and their impact in prepubertal children in Great Britain: 1999–2008. Journal of Child Psychology and Psychiatry 56, 4957.CrossRefGoogle ScholarPubMed
StataCorp. (2019) User's Guide, Release 16. College Station, Texas (USA): STATA Corporation.Google Scholar
United Nations Children's Fund (2021) The State of the World's Children 2021. New York: UNICEF. Available at https://www.unicef.org/reports/state-worlds-children-2021 (Accessed 9 August 2022).Google Scholar
Vaillancourt, T, Haltigan, JD, Smith, I, Zwaigenbaum, L, Szatmari, P, Fombonne, E, Waddell, C, Duku, E, Mirenda, P, Georgiades, S, Bennett, T, Volden, J, Elsabbagh, M, Roberts, W and Bryson, S (2017) Joint trajectories of internalizing and externalizing problems in preschool children with autism spectrum disorder. Development and Psychopathology 29, 203214.CrossRefGoogle ScholarPubMed
Weeks, M, Cairney, J, Wild, TC, Ploubidis, GB, Naicker, K and Colman, I (2014) Early-life predictors of internalising symptom trajectories in Canadian children. Depression and Anxiety 31, 608616.CrossRefGoogle Scholar
Williams, LR, Degnan, KA, Perez-Edgar, KE, Henderson, HA, Rubin, KH, Pine, DS, Steinberg, L and Fox, NA (2009) Impact of behavioral inhibition and parenting style on internalizing and externalizing problems from early childhood through adolescence. Journal of Abnormal Child Psychology 37, 10631075.CrossRefGoogle ScholarPubMed
Willner, CJ, Gatzke-Kopp, LM and Bray, BC (2016) The dynamics of internalizing and externalizing comorbidity across the early school years. Development and Psychopathology 28(4pt1), 10331052.CrossRefGoogle ScholarPubMed
Figure 0

Table 1. Sample descriptive statistics for all cohort members

Figure 1

Table 2. Distribution of the cohort sample in each centile cut-off of mental health symptoms based on the SDQ fourfold classification

Figure 2

Table 3. Fit indices for the parallel-process growth mixture models

Figure 3

Fig. 1. Estimated internalising (left) and externalising (right) MHS trajectories.

Figure 4

Table 4. Growth parameters for the parallel-process growth mixture model

Figure 5

Table 5. Multinomial logistic regressions predicting high-risk class 2 and mild-risk class 3 membership v. low-risk class 1 membership

Supplementary material: File

Katsantonis and Symonds supplementary material

Katsantonis and Symonds supplementary material

Download Katsantonis and Symonds supplementary material(File)
File 78.4 KB