Introduction
Adolescents with common mental disorders struggle with successfully transitioning into adulthood and experience poor health, social, educational, and economic outcomes (Wagner & Newman,
2012). They are more likely to get disengaged from education and employment as young adults, a status referred to as NEET (not in education, employment, or training) (e.g., Witt et al.,
2019). More importantly, those with subclinical depression and anxiety experience equal functional impairment as those with clinical levels (Wesselhoeft et al.,
2013, Balázs et al.,
2013). They experience long-term consequences in the same areas including becoming NEET (Clarke & Lovewell,
2021). However, there is a significant care gap for this group of young people as the majority do not access mental health services and receive early intervention (Soneson et al.,
2022). There is a need to focus on adolescents with such common mental health problems before they transition from compulsory education (e.g., Cornaglia et al.,
2015). A comprehensive investigation of what factors are associated with later disengagement from education and employment is lacking for this population in the extant literature. This would provide important insights for preventive strategies during secondary schooling. The current study focused on longitudinal associations between adolescent psychosocial factors and NEET status in young adulthood among individuals with common mental health problems.
Adolescents with mental health problems pose a serious and growing issue as a public health challenge. Among secondary school-aged adolescents in England, one in six (17.6%) is identified with a probable mental disorder, an increase from 13.6% in 2017 (Clarke et al.,
2020). Depression and anxiety are the most common mental disorders experienced by young people (Clarke et al.,
2020). Prevalence is higher in female adolescents than males (Balázs et al.,
2013). Adolescents are least likely to seek treatment compared to any other age group and only about half access services in England (Soneson et al.,
2022). One of the reasons is under-identification of needs (Soneson et al.,
2022). They are poorly understood, go unrecognized by the school services, and lack the support they need at an early stage which leads to more severe psychopathology (Baksheev et al.,
2011). Long-term consequences of such problems regarding impaired role functioning have gained increasing recognition and the impact appears to be substantial and greater than that associated with physical disorders (e.g., Mojtabai et al.,
2015).
Adolescents with common mental health problems face unique challenges when transitioning from school to further education and employment. They experience adverse labour market outcomes in adulthood such as reduced employment and earnings (e.g., Fletcher,
2013). They are also more likely to become NEET in young adulthood (e.g., Cornaglia et al.,
2015; Witt et al.,
2019). In the UK, recent figures show that the total number of people aged 18 to 24 years who are in NEET status is 744,000 which represents 13.8% of young adults in the same age range (Office for National Statistics,
2021). There are consequences to being NEET such as health, societal and economic burden to the government and poor health, prolonged unemployment, low wages, and social exclusion to the individual (Arnold & Baker,
2013).
Exploration of risk factors for becoming disengaged from education and employment is necessary to identify targeted preventive strategies for adolescents with common mental health problems during compulsory education. This is crucial not just because of the associated long-term consequences, but also because intervention is difficult once they become disengaged and there is a high recurrence rate (Arnold & Baker,
2013). A young person who enters NEET category once is 7.9 times more likely to become disengaged again (Arnold & Baker,
2013). A simple measure such as the 12-item General Health Questionnaire (GHQ-12) is applicable in secondary school settings to identify those with common mental health problems (Baksheev et al.,
2011). It is also useful for predicting who is at risk of getting disengaged from education and employment (Cornaglia et al.,
2015). Early identification through screening at schools using GHQ-12 rather than relying on help-seeking would aid in preventing development of severe psychopathology and lead to better long-term outcomes (Baksheev et al.,
2011).
Research has mainly focused on background characteristics as risk factors for NEET. Socioeconomic disadvantage, single-parent household, being a young carer or teenage parent, and educational attainment are some of the key risk factors (Dorsett & Lucchino,
2014, Pitkänen et al.,
2021). There are a number of additional important psychosocial risk factors at play. Psychosocial is an umbrella term that describes the intersection and interaction of social, cultural, and environmental influences on the mind and behaviour (American Psychological Association,
2020). A recent systematic review identified that various domains of adolescent psychosocial factors such as behavioural problems and bullying are associated with education and employment status in young adulthood (Tayfur et al.,
2021). Attitudes to school has also been associated with later education and employment status (Duckworth & Schoon,
2012, Spengler et al.,
2018). Since no risk factor occurs in isolation in shaping a developmental outcome (Duckworth & Schoon,
2012), it is important to identify the independent and relative role of these psychosocial factors with disengagement from education and employment for vulnerable youth.
Results
Table
1 shows the general characteristics of the sample stratified by young adult NEET status. Of the current sample, 66.8% were female (
n = 1486) and 68.9% of participants were White (
n = 1,533). Only 5.7% had a caring responsibility within the household (
n = 126) and mean socioeconomic status of the participants were 22.8. In terms of household, the majority came from English speaking (89.4%) and married or cohabiting families (78.4%). It is found that young adults who were in NEET status significantly differed in respect to some characteristics during adolescence. Differences were established in relation to sex (F
adj = 7.02,
p = 0.008); caring responsibility (F
adj = 7.66,
p = 0.006); family composition (F
adj = 15.25,
p < 0.001), and socioeconomic status (t = 7.3,
p < 0.001). Ethnicity (F
adj = 0.594,
p = 0.441) was not significant.
Table 1
Baseline characteristics of the sample stratified by NEET status at age 25–26 years
Sex | Female | 1486 | (66.8) | 1282 | (65.8) | 204 | (73.9) | p = 0.008 |
| Male | 738 | (33.2) | 666 | (34.2) | 72 | (26.1) |
Ethnicityb | White | 1533 | (68.9) | 1347 | (69.2) | 186 | (67.6) | p = 0.441 |
| Other | 688 | (30.9) | 599 | (30.8) | 89 | (32.4) |
Caring responsibility | Yes | 126 | (5.7) | 99 | (5.1) | 27 | (9.8) | p = 0.006 |
| No | 2093 | (94.1) | 1844 | (94.9) | 249 | (90.2) |
| No | 234 | (10.5) | 191 | (9.8) | 43 | (15.6) |
Family composition | Married/ cohabiting | 1743 | (78.4) | 1554 | (80.7) | 189 | (70.8) | p < 0.001 |
| Single/no parent | 449 | (20.2) | 371 | (19.3) | 78 | (29.2) |
| | na | SD | Mean | SD | Mean | SD | Kruskal-Wallis test |
Socioeconomic status | (Range: 1-86) | 22.8 | 16.9 | 21.8 | 16.3 | 30.1 | 19.1 | p < 0.001 |
Correlations between variables are reported in Table
2. The interpretation for the phi coefficient is that between 0.10–0.30 are small, 0.30–0.50 are medium and ≥ 0.50 are large and Cramer’s v for at least 3 categories in either row or column are interpreted as 0.07–0.20 small, 0.20–0.35 medium and ≥0.35 large (Mangiafico,
2016). As for the Epsilon-squared, between 0.01–0.08 are small, 0.08–0.26 are medium and ≥ 0.26 large (Mangiafico,
2016). None of the significant correlations were found to be of large size. Since the goal is to avoid inclusion of explanatory variables correlated at
r ≥ 0.8 (large), the significant correlations were in a range of tolerable inter-correlation for logistic regression analysis (Field,
2013).
Table 2
Correlations between variables
1. Self-esteem | | | | | | | | | | | | | | |
2. Locus of control | 0.13b | | | | | | | | | | | | | |
3. Substance use | 0.16b | 0.06a | | | | | | | | | | | | |
4. Bullying | 0.13b | 0.06a | 0.08a | | | | | | | | | | | |
5. Behavioural problems | 0.12b | 0.08a | 0.35a | 0.08a | | | | | | | | | | |
6. Educational aspirations | 0.06b | 0.04a | 0.13a | 0.03a | 0.15a | | | | | | | | | |
7. Attitudes to school | 0.09c | 0.10c | 0.11c | 0.04c | 0.09c | 0.12c | | | | | | | | |
8. Job aspirations | 0.05b | 0.03a | 0.11a | 0.03a | 0.002a | 0.04a | 0.03c | | | | | | | |
9. Physical activity | 0.07b | 0.01c | 0.08b | 0.06b | 0.05b | 0.03b | 0.04c | 0.08b | | | | | | |
10. Sex | 0.16b | 0.06a | 0.11a | 0.002a | 0.10a | 0.14a | 0.03c | 0.02a | 0.22a | | | | | |
11. Ethnicity | 0.04b | 0.009a | 0.19a | 0.09a | 0.003a | 0.11a | 0.06c | 0.16a | 0.04a | 0.02a | | | | |
12. Socioeconomic status | 0.05c | 0.04c | 0.07c | 0.05c | 0.07c | 0.07c | -0.05d | 0.10c | 0.05c | 0.02c | 0.19c | | | |
13. Family composition | 0.10b | 0.02a | 0.05a | 0.03a | 0.09a | 0.03a | 0.05c | 0.02a | 0.07a | 0.02a | 0.06a | 0.10c | | |
14. Caring responsibility | 0.03b | 0.02a | 0.02a | 0.06a | 0.04a | 0.003a | 0.03c | 0.008a | 0.04a | 0.04a | 0.05a | 0.11c | 0.05a | |
The univariable logistic regression results are presented in Table
3. Results showed that lower self-esteem (COR 2.89, 95% CI 1.89–4.42), external locus of control (COR 1.76, 95% CI 1.01–3.06), behavioural problems (1.58, 95% CI 1.13–2.20), low/uncertain educational aspirations (COR 1.64, 95% CI 1.08–2.50), no job aspirations (COR 1.83, 95% CI 1.24–2.72), and low/no physical activity (COR 2.60, 95% CI 1.68–4.04) in adolescence significantly increased the likelihood of being NEET as young adults whereas positive attitudes toward school (COR 0.96, 95% CI 0.94–0.98) and not being bullied (COR 0.54, 95% CI 0.36–0.79) significantly decreased odds for being NEET at age 25–26 years. Substance use was found to be insignificant at 5% level. Regarding the interpretation of the size of the odds ratios, 1.68, 3.47, and 6.71 are equivalent to Cohen’s
d = 0.2 (small), 0.5 (medium), and 0.8 (large), respectively (Chen et al.,
2010). Therefore, effect sizes were small for behavioural problems, educational aspirations, attitudes to school, and bullying ranging from 0.54 to 1.64. Moreover, effect sizes were medium for self-esteem, locus of control, job aspirations and physical activity odds ratios ranging from 1.76 to 2.89 considering that the range is between 1.68 and 3.47 (Chen et al.,
2010).
Table 3
Logistic regression analysis of adolescent psychosocial factors and young adult NEET status
Self-esteem (Much more than usual vs. not at all) | 2161 | 2.89 | [1.89–4.42] | <0.001 | 1.75 | [1.06–2.89] | 0.029 |
Locus of control (External vs. internal) | 2182 | 1.76 | [1.01–3.06] | 0.045 | 1.93 | [1.08–3.45] | 0.027 |
Substance use (Yes vs. no) | 2100 | 1.35 | [0.98–1.86] | 0.064 | 1.02 | [0.69–1.51] | 0.923 |
Behavioural problems (Yes vs. no) | 2133 | 1.58 | [1.13–2.20] | 0.008 | 0.92 | [0.61–1.38] | 0.676 |
Bullying (No vs. yes) | 2125 | 0.54 | [0.36–0.79] | 0.002 | 0.57 | [0.36–0.90] | 0.017 |
Educational aspirations (Low/uncertain vs. high) | 2224 | 1.64 | [1.08–2.50] | 0.021 | 1.34 | [0.80–2.25] | 0.262 |
Job aspirations (No vs. yes) | 2221 | 1.83 | [1.24–2.72] | 0.003 | 1.70 | [1.09–2.65] | 0.019 |
Attitudes to school (Range: 0-48) | 2224 | 0.96 | [0.94–0.98] | <0.001 | 0.97 | [0.95–0.99] | 0.009 |
Physical activity (Low/none vs. high) | 2223 | 2.60 | [1.68–4.04] | <0.001 | 1.98 | [1.16–3.38] | 0.013 |
To examine whether the observed differences could be explained by variations in sex, ethnicity, socioeconomic status, caring responsibility and family composition, the analysis was adjusted for these covariates. The observed differences did not change except for educational aspirations and behavioural problems. As seen in Table
3, the multivariable logistic regression model showed that after adjustment for background characteristics, lower self-esteem (AOR 1.75, 95% CI 1.06–2.89), external locus of control (AOR 1.93, 95% CI 1.08–3.45), no job aspirations (AOR 1.70, 95% CI 1.09–2.65), and low to no physical activity (AOR 1.98, 95% CI 1.16–3.38) in adolescence significantly increased the likelihood of being NEET as young adults whereas positive attitudes toward school (AOR 0.97, 95% CI 0.95-0.99) and not being bullied (AOR 0.57, 95% CI 0.36–0.90) significantly decreased odds for being NEET at follow-up. While the effect size for bullying and attitudes to school was small, it was medium for self-esteem, locus of control, job aspirations, and physical activity (Chen et al.,
2010). The goodness of fit statistics, discriminatory power, and assumptions were tested to establish the validity of the logistic regression model.
The multiple imputation results are shown in Table
4. After controlling for sex, ethnicity, socioeconomic status, caring responsibility, and family composition, lower self-esteem (AOR 1.76, 95% CI 1.11–2.79), no job aspirations (AOR 1.58, 95% CI 1.05–2.37), and low to no physical activity (AOR 1.90, 95% CI 1.18–3.04) in adolescence significantly increased the likelihood of being NEET as young adults whereas positive attitudes toward school (AOR 0.97, 95% CI 0.95–0.99) and not being bullied (AOR 0.63, 95% CI 0.41–0.96) significantly decreased odds for being NEET at follow-up. The magnitude of the associations was found to be similar to those of complete case analysis. It was small for job aspirations, bullying, and attitudes to school since up to 1.68 is considered small (Chen et al.,
2010). The magnitude was medium for self-esteem and physical activity since the range is between 1.68 and 3.47 (Chen et al.,
2010). The only statistically significant difference between the complete case analysis and the multiple imputation was observed for locus of control. Although it was significant at 5% level in the former, the multiple imputation suggested otherwise (Table
4).
Table 4
Multiple imputation results for the multivariable logistic regression analysis
Self-esteem (Much more than usual vs. not at all) | 1.76 | [1.11–2.79] | 0.016 |
Locus of control (External vs. internal) | 1.37 | [0.76–2.46] | 0.291 |
Substance use (Yes vs. no) | 0.91 | [0.62–1.33] | 0.630 |
Behavioural problems (Yes vs. no) | 1.14 | [0.79–1.66] | 0.481 |
Bullying (No vs. yes) | 0.63 | [0.41–0.96] | 0.033 |
Educational aspirations (Low/uncertain vs. high) | 1.27 | [0.80–2.02] | 0.305 |
Job aspirations (No vs. yes) | 1.58 | [1.05–2.37] | 0.026 |
Attitudes to school (Range: 0-48) | 0.97 | [0.95–0.99] | 0.036 |
Physical activity (Low/none vs high) | 1.90 | [1.18–3.04] | 0.008 |
Discussion
Adolescents with common mental health problems have difficulty transitioning from school to further education and employment and end up being ‘not in education, employment or training’ (NEET) as young adults (e.g., Witt et al.,
2019). There is a significant care gap as the majority of young people do not access mental health services and receive early intervention (Soneson et al.,
2022). Considering the long-term consequences of being NEET, a focus on secondary school students with common mental health problems is necessary before they leave compulsory education (Cornaglia et al.,
2015, Soneson et al.,
2022). Research on to what extent psychosocial factors are associated with later disengagement from education and employment is lacking for this population. This type of knowledge is needed to develop targeted preventive strategies during secondary schooling. In this study, it was possible to examine longitudinal associations between adolescent psychosocial factors and NEET status in young adulthood among a nationally representative sample of secondary school students with common mental health problems.
Main findings of this study were that having lower self-esteem, external locus of control, no job aspirations, and low to no physical activity significantly increased the likelihood of being NEET at ages 25–26 years whereas not being bullied and having more positive attitudes toward school decreased it for adolescents with common mental health problems. These findings remained significant even after controlling for socioeconomic, individual, and family characteristics in adolescence. Educational aspirations were significantly associated with later NEET status, a consistent finding with previous studies (Duckworth & Schoon,
2012), but its significance disappeared when other factors were taken into account suggesting that having more positive attitudes toward school is a better protective factor for avoiding NEET status although the magnitude is small. Moreover, early labour market engagement which reflects higher job aspirations appear to be a relatively stronger factor in order to avoid entering the NEET category. Previous studies have also emphasized that early job engagement is important for being employed later on (Baert et al.,
2017) as well as to avoid becoming NEET (Duckworth & Schoon,
2012). In addition, one of the perceived barriers of service-seeking youth outside the labour force is indeed the lack of work experience (Ose & Jensen,
2017).
Previous studies show that behavioural problems are significantly associated with later NEET status; however, the current study shows that while this is true in the univariable model the association does not hold after adjustment for other factors whereas being bullied appears to be still important which is consistent with the literature (Tayfur et al.,
2021). Similar results are found for self-evaluation factors including but not limited to self-esteem and locus of control, with both factors significantly associated with later NEET status regardless of background characteristics. Although the multiple imputation results for the multivariable model suggested that locus of control is not significant, the role of self-evaluations still appears to be important due to self-esteem. There are very few studies in the literature focusing on self-evaluation factors in the NEET context, but they indicate that low self-esteem and external locus of control are significantly associated with being disengaged from education and employment in young adulthood (Mendolia & Walker,
2015). Internal locus of control has been shown to protect disadvantaged youth against becoming NEET (Ng-Knight & Schoon,
2017). Moreover, one of the perceived barriers to education and employment among service-seeking youth is low self-esteem (Ose & Jensen,
2017).
In this study, physical activity was significantly associated with becoming disengaged from education and employment although this is relatively an overlooked factor in the literature. A recent systematic review has identified only one study for this domain (Tayfur et al.,
2021). It is important to highlight that physical activity had the strongest magnitude of association with later NEET status over and above background characteristics. Participation in sport is recognized as a unique construct for positive psychosocial development (Bedard et al.,
2020). It is important to focus on sport participation considering the benefits of physical activity for mental health as well as social competence of youth which can also elevate self-evaluations (Bedard et al.,
2020). A recent study has showed that higher physical activity at age 15 is positively related to the likelihood of being employed in adulthood suggesting that investments in adolescent physical activity could yield better labour market outcomes (Kari et al.,
2021). Therefore, sports may be an important supplemental component for mental health related interventions in schools to increase life chances. For instance, a recent innovation called the Game-Changer has been developed in Scotland to increase positive outcomes for vulnerable youth through sports demonstrating the potential for similar intervention strategies (Fitzpatrick et al.,
2020).
There is an increasing interest in promoting healthy transitions into adulthood through school-based mental health interventions during adolescence (O’Connor et al.,
2017). Considering the long-term consequences of being disengaged from education and employment, it is crucial to implement early strategies to prevent vulnerable youth entering the NEET category in the first place as the recurrence rate is also high (Arnold & Baker,
2013). As the findings of the current study supports it is preferable to do this during compulsory education given the difficulties of reaching young adults who are in NEET status because they are not attached to a singular institution. Moreover, interventions of re-engagement of NEET individuals and policy efforts have not been very successful (Hutchinson et al.,
2016). It is also suggested that improvement in depressive symptoms among help-seeking young adults do not lead to re-engagement in employment or education (O’Dea et al.,
2016). Therefore, preventive efforts should focus on other risk factors at play in order to deliver the most effective and targeted mental health interventions before adolescents leave the school setting.
Secondary schools are uniquely placed to identify mental health needs and provide care and support (Soneson et al.,
2022). Given that many adolescents experience subclinical difficulties without receiving treatment, schools can deliver early interventions to equip adolescents with common mental health problems with the skills and competencies that they need to handle the subsequent transition period successfully (O’Connor et al.,
2017, Soneson et al.,
2022). School-based mental health provision in the UK is improving and one model of early identification of needs is through screening (Soneson et al.,
2022). Utilizing any gain from early intervention is dependent upon screening to identify those who require it (Arnold & Baker,
2013). This study shows that prevention and promotion of mental health should have a focus on psychosocial factors. GHQ-12 is applicable in secondary school settings and can be used as a method of universal screening (screening
all pupils) (Baksheev et al.,
2011). However, relevant screening tools specific to NEET may be developed and implemented to further improve identification and preventive strategies. This may be used as a method of selective screening (screening
at-risk pupils) among those with common mental health problems. It is suggested that more than half of those at risk for becoming NEET can be identified by age 14 (Arnold & Baker,
2013). Since interventions can be costly, targeted screening for NEET can identify those who will potentially benefit the most (Arnold & Baker,
2013). Therefore, development and implementation of cross-cultural screening tools should be encouraged such as the NEET- Hikikomori Risk Factors scale from Japan which also holds a psychosocial focus (Uchida & Norasakkunkit,
2015). To our knowledge, there is no such scale available in the UK. These developments should encompass a range of psychosocial factors, particularly the ones identified in this study. Mental health services may be supplemented by screening and fostering for a stronger trajectory of psychosocial wellbeing to encourage resiliency towards developmental tasks in the areas of work and education. Building on resiliency of students could help equip them with the assets to face the challenges that lie ahead once they leave the compulsory school setting and transition into adulthood (O’Connor et al.,
2017).
The limitations of this study include that of secondary data analysis such as having no control over the data collected (Trzesniewski et al.,
2011). Consequently, some of the factors are based on a single item which creates a limitation although efforts are made to ensure their effectiveness based on previous studies or indicators such as the Cronbach’s alpha. Moreover, this study could not include self-efficacy–another self-evaluation factor- as part of the interest of psychosocial factors despite its association with later education and employment status (Pinquart et al.,
2003). The frequency of substance use also could not be included although it is suggested that the amount and frequency of use should be considered along with the use of different substances together when looking at longitudinal associations with education and employment (Tayfur et al.,
2021). This may explain the current findings regarding substance use and should be considered in future research. Furthermore, both unemployed (without a job and actively seeking work within the last four weeks and are available to start work within the next two weeks) and inactive (without a job and not seeking work within the last four weeks and/or unable to start work in the next two weeks) young adults are considered to be NEET by definition nationally and internationally (Office for National Statistics,
2021, Eurofound,
2017). Therefore, an individual identified as NEET will always be either unemployed or economically inactive. Studies typically take the approach of classifying someone as NEET for whatever reason as long as they are not in any form of employment or education (e.g., Moore et al.,
2015). So, unemployed and inactive subcategories such as looking after the home or disabled are grouped together (e.g., Roldós,
2014). However, the factors that predict NEET status may differ for those unemployed and those who are inactive. Therefore, it may provide further insight to examine this issue to better understand the heterogeneity within the group of young adults who are disengaged from education and employment. A study has drawn attention to this by distinguishing inactive NEETs from NEET by referring to them as ‘NLFET’ (neither in the labour force nor in education or training) and excluding unemployed youth because they are thought to be still a part of the labour force (Ose & Jensen,
2017). Unfortunately, the current study could not stratify analysis by subgroups of NEET due to the small number of participants in NEET status. However, this may be a good strategy to consider for future research.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.