Bullying Victimisation and Mental Health: The Mediating Role of Peer and Family Social Support

ABSTRACT

Introduction

Adverse childhood experiences have been consistently linked to a range of mental health related difficulties in adulthood. One such experience is bullying, which has been shown to predict psychotic symptoms. This study aimed to explore whether bullying victimisation in early adolescence is associated with certain mental health outcomes (psychosis, depression and cannabis abuse) and whether this is mediated by the individuals' perception of social support from peers and family.

Methods

This study performed secondary data analysis of a prospective cohort study, the Belfast Youth Development Study (BYDS). It used longitudinal data from 2087 participants from ages 12–21 years and performed path analysis (using Mplus V8) between bullying and the above mental health outcomes and mediators.

Results

The results showed a direct path from bullying to psychotic symptoms (estimate = 0.10; SE = 0.0; p = 0.030) as well as an indirect path mediated by peer attachment, parental attachment and depression symptoms (estimate = 0.01; bootstrapped 95% CIs 0.01, 0.02).

Conclusion

Being bullied in childhood is directly associated with a greater risk of having psychotic symptoms as an adult. There is also an indirect path from bullying in childhood to psychosis symptoms as an adult mediated by peer attachment, parental attachment, and depression. Interventions should be aimed at addressing bullying and strengthening social support from peers and family.

1. Introduction

Adverse childhood experiences have been consistently linked to a range of mental health related difficulties in adulthood such as depression and drug abuse (Evans et al. 2020; Schilling et al. 2007). Furthermore, adverse childhood experiences have now been associated with psychosis (Gibson et al. 2016; Inyang et al. 2022; Stanton et al. 2020; Varese et al. 2012).

Most studies on adverse childhood experiences and mental health outcomes including psychosis have focused on sexual, physical or emotional abuse and/or neglect. However, bullying may also be an important category of childhood adversity. A large meta‐analysis of 165 articles that had explored health and mental health associations with childhood bullying suggests causal associations between bullying victimisation and mental health problems such as depression, anxiety, poor general health and suicidal ideation and behaviours (Moore et al. 2017). One study showed a dose‐response relationship between bullying victimisation and depression, with victimisation increasing the likelihood for later depression (Brunstein Klomek et al. 2016).

The relationship between psychotic symptoms and bullying has received limited attention to date. A systematic review identified ten studies that examined the relationship between childhood bullying and psychosis via a prospective research design (Cunningham et al. 2016). It suggested that being bullied does predict the later development of psychotic symptoms. It suggested that being bullied does predict the later development of psychotic symptoms. Importantly, prospective population‐based research has demonstrated that peer victimisation predicts psychotic symptoms as early as age 12 in nonclinical samples (Schreier et al. 2009), and psychotic experiences at age 18 in longitudinal cohort studies, even after controlling for confounders (Wolke et al. 2014). Cunningham et al. (2016) noted that a better understanding of the factors that mediate the bullying/psychosis relationship is needed. Several variables are likely to mediate such a relationship for example parental and peer attachment, depression and cannabis use.

Further longitudinal work has begun to examine potential pathways linking childhood victimisation and psychosis‐like symptoms. For example, findings from the ALSPAC birth cohort suggest that emotional and cognitive processes may partially explain the association between victimisation and later psychosis‐like experiences (Fisher et al. 2012), supporting the need to investigate mediating mechanisms. Moreover, evidence indicates that bullying may have effects on psychotic experiences that are comparable to, or in some cases stronger than, those associated with adult‐inflicted traumas, depending on type, frequency, and developmental timing of exposure (Croft et al. 2019).

Styles of attachment have been associated with a variety of poor mental health outcomes. Adolescents with insecure attachment to parents are more likely to report emotional difficulties such as anxiety and emotional regulation (Brumariu et al. 2012) and depression (Rodrigues et al. 2024). Attachment has also been suggested as a possible mechanism between childhood maltreatment and psychotic like experiences (Sheinbaum et al. 2020). Sitko et al. (2014) reported that associations between specific psychotic symptoms and specific childhood adversities are mediated by attachment styles (Sitko et al. 2014).

Adolescence is a time in which people begin to transfer their attachments to peers and these peers begin to become significant sources of intimacy and social support (Laible 2007). As such, it is hardly surprising that peer attachment has some connection with mental health outcomes such as conduct‐related problems and delinquency (McElhaney et al. 2006) and depression (Nickerson and Nagle 2005), and mediates the relationship between parental attachment and prosocial behaviour (Oldfield et al. 2016). It is also acknowledged in the literature that both parental and peer attachments are likely to interact to predict outcomes (Oldfield et al. 2016).

The association between bullying and mental health symptoms is also important to consider. A meta‐analysis by Moore et al. has shown convincing evidence for a relationship between bullying in children and adolescents and the development of depression (Moore et al. 2017). Similarly, longitudinal studies across two countries have shown that the long‐term mental health consequences of peer bullying in childhood can be as severe as, or even exceed, those associated with maltreatment by adults (Lereya et al. 2015). This has also been shown to be mediated by attachment styles. For example, peer attachment has been found to moderate the relationship between bullying and depression (Yin et al. 2017). Brunstein Klomek et al. (2016) found that young people with both more parental and peer support were less likely to self‐harm after being bullied than those with less support (Brunstein Klomek et al. 2016). Additionally, other studies suggested the importance of depression in the development of psychosis in at risk individuals (Yung et al. 2003; Yung et al. 2004). Evidence has also shown that the presence of depression to be associated with the severity, distress and content of symptoms in psychosis (Hartley et al. 2013).

The recreational use of cannabis has been shown to be a risk factor for the development of psychosis (Large et al. 2011). In particular, high potency cannabis (skunk) and earlier age of use have been shown to be important factors (Di Forti et al. 2015; Dragt et al. 2012). More severe use of cannabis (abuse or dependence) is also important for the transition to psychosis in at‐risk individuals (Kraan et al. 2016). However, other studies have shown that when controlling for confounders, cannabis use was not associated with psychotic‐like experiences (Fonseca‐Pedrero et al. 2020). This highlights the complexity of this relationship and the possibility of other mediating factors playing an important role.

The proposed study is a secondary data analysis of a prospective cohort study; namely, the Belfast Youth Development Study (BYDS). The BYDS followed a group of children longitudinally from ages 12 to 21 years. This study aims to investigate the relationship between bullying and three related mental health outcomes; namely, psychosis, depression and cannabis abuse. Two research questions were considered:

1. Is bullying victimisation in early adolescence associated with psychosis symptoms, depression symptoms and cannabis abuse in young adulthood?

2. Is the relationship between bullying victimisation and mental health outcomes (i.e., psychosis symptoms, depression symptoms and cannabis abuse) mediated by the individuals' perception of social support from peers and family?

2. Materials and Methods

2.1. Design

The present study is a secondary data analysis of a prospective cohort study; namely, the Belfast Youth Development Study (BYDS). The study spanned 7 waves from 2001 to 2011. All secondary schools in 3 areas of Northern Ireland were invited to take part in the survey from 2001 to 2005 (71% of invited schools took part in 2001). Participation numbers fluctuated throughout this period due to dropout and additional young people joining the cohort. Response rates were calculated based on the total eligible to take part at each wave (Higgins et al. 2018). Participating numbers and response rates at each wave were as follows: wave 1 (age 11; in 2001; n = 3834; response rate = 86.9%); wave 2 (age 12; in 2002; n = 4343; response rate = 83.3%); wave 3 (age 13; in 2003; n = 4522; response rate = 86.3%); wave 4 (age 14; in 2004; n = 3965; response rate = 75.5%); wave 5 (age 15; in 2005; n = 3830; response rate = 74.3%); wave 6; (age 16/17; in 2006/07; n = 2335; response rate = 55.7%); wave 7; age 20/21 (in 2010/11; n = 2087; response rate = 49.9%). Overtime, some groups were less likely to participate in the study including males (reducing by 14% points from wave 1 wave 7) and those in receipt of free school meals (reducing by 5% points from wave 1 to wave 5). Extensive details on the BYDS cohort are provided in Higgins et al. (2018).

The BYDS study was reviewed and approved by the then School of Sociology, Social Policy and Social Work, Queen's University Belfast ethics committee. In each school, in each wave, letters were sent to parents explaining the study and they were given the option to ‘opt out’ or remove their child from the study. Children were invited to complete a paper copy of the questionnaire in school and were given the option to refuse to participate. After participants left school this was completed by post.

2.2. Participants

The present study included the 2087 individuals who took part in the BYDS in 2011 (Wave 7). The analytical sample was predominantly female (61%), and 16% were in receipt of free school meals (a proxy measure of low income) at age 12 (Table 1). Of the 2087 wave 7 participants, 1700 fully completed the psychosis measure. From these, 169 (10%) provided responses that are consistent with psychosis.

TABLE 1.

Sample characteristics by psychosis.

		Psychosis	No psychosis	Total
		M (SD)	M (SD)	M (SD)
Bullying victimisation		0.51 (0.73)	0.25 (0.54)	0.29 (0.57)
Peer Attachment		31.71 (15.73)	27.40 (15.85)	27.76 (15.93)
Parental attachment		38.10 (19.22)	32.11 (20.08)	33.47 (20.25)
Cannabis abuse		2.77 (5.49)	0.44 (1.89)	0.68 (2.60)
Depression		9.21 (6.70)	3.71 (4.47)	4.43 (5.17)
Free School Meals (composite version)		0.51 (0.82)	0.33 (0.67)	0.34 (0.68)
		%	%	%
Gender	Male	56%	37%	39%
	Female	44%	63%	61%
FSM (age 12 years)	Yes	28%	15%	16%
	No	72%	85%	84%
Cannabis history	Never	36%	58%	55%
	Infrequent	32%	31%	32%
	Frequent	18%	4%	6%
	Used to	14%	7%	7%

2.3. Procedure

A detailed description of the methodology used in the BYDS is provided in the corresponding cohort paper for the study (Higgins et al. 2018). The path model (see Figure 1) was designed to test our prediction that being bullied in childhood could directly affect mental health (e.g., psychosis symptoms, depression symptoms, and substance abuse) in young adulthood. Indirect paths through parental and peer attachment were also included in the model, to test the mediating role of peer and parental social support. Due to the temporal order of the peer and parental variables, a predictive path from peer attachment (age 13 years) to parental attachment (age 14 years) was also included. Paths were included in the model to show the predicted relationship between the mental health variables; both depression and cannabis abuse were predicted to increase risk of psychosis, and the relationship between cannabis abuse and depression was expected to be bidirectional.

FIGURE 1.

Path model tested. Single headed arrows represent predictive paths, and double headed arrows show non‐directional correlations. Significant paths denoted by an asterisk, and significant indirect paths are shown in bold.

2.4. Measures

The measures included in the bullying victimisation–mental health mediation model were as follows:

2.5. Predictors

Bullying (age 12 years). This was measured by the “other children pick on me or bully me” item from the Strengths and Difficulties Questionnaire (Goodman 1997). This item is scored on a 3‐point scale (not true; somewhat true; certainly true).

2.6. Outcomes

Psychosis (age 21 years). The Psychosis Screening Questionnaire (PSQ, (Bebbington and Nayani 1995)) was used to classify participants according to whether or not they self‐report showing signs of psychosis. This measure includes questions designed to screen for each of the following: hypomanic mood; thought interference; persecution; strange experiences; and hallucinosis. If a participant screens positively in any one of these areas, then they are also screened positively for psychosis. Both specificity and sensitivity of the test are very high (96.9% and 95.3%).

Depression (age 21 years). The Patient Health Questionnaire (PHQ‐9; (Kroenke et al. 2001)) comprises nine items measured on four point Likert scales (‘not at all’ to ‘nearly every day’). The scale range is 0–27 with higher scores being indicative of greater levels of depressive symptoms. Internal consistency of the scale is excellent (Cronbach's alpha = 0.90).

Cannabis abuse (age 21 years). The Cannabis Abuse Screening Test (CAST; (Legleye et al. 2007)) comprises 6 items and aims to measure cannabis abuse. Each item is measured on a 5‐point Likert scale with responses ranging from ‘never’ to ‘very often’. The items are summed to form a scale ranging from 0 to 24 with higher scores indicating greater levels of cannabis abuse (e.g., using before midday; unsuccessful attempts to stop using). BYDS participants were only asked to complete these items if they indicated having used cannabis in the past year, otherwise they were given a score of zero on this scale. The CAST has excellent internal reliability (α = 0.90).

2.7. Mediators

Peer attachment (age 13 years). The Peer sub‐scale of the Inventory of Peer and Parental attachment (Armsden and Greenberg 1987) measures peer attachment on a 0–100 scale. Each of the 25 items uses a 5‐point Likert scale, and higher scores are indicative of worse peer attachment.

Parental attachment (age 14 years). This was measured using the 28‐item parental subscale of the Inventory of Peer and Parental attachment (Armsden and Greenberg 1987), with each item measured on a 5‐point Likert scale. On this 0–112‐point scale, higher values represent worse attachment to one's parents.

2.8. Covariates

Gender: Coded as 0 for male and 1 for female. This was included as a covariate for all mediator and outcome variables.

Free school meals (FSM): Annually, from ages 12 to 16 years participants were asked if they received FSM (yes/no); this was used as a proxy measure of socio‐economic deprivation. The age 12 FSM variable was used as a covariate for the parental and peer attachment variables (ages 13 and 14). A composite FSM variable was formed from the 12 to 16 year versions and coded as never (0); at least one year (1); and all years (2) based on the FSM status of the participants throughout the 5‐year period. The composite FSM variable was used as a covariate for the age 21 years outcomes variable which it related significantly to; namely, depression and cannabis abuse.

Cannabis history (age 21 years). Frequency of cannabis use in the past year was measured by responses to “Thinking about cannabis which of these statements best describes you?” The response options included: Never (0); I have only tried it once (1); I have tried it between 2 and 5 times (2); I use it about once a month (3); I use it about once a week (4); I use it more than once a week (5); I use it everyday (6); I used to use it, but don't smoke it now (7). For the purpose of this analysis code 0 was coded as ‘Never’, 1–3 as ‘Infrequent’, 4–6 as ‘Frequent’ and 7 as ‘Used to’. In the analysis ‘never’ was treated as the reference category. This was included as a covariate of all outcome variables.

2.9. Analysis

Pearson correlations were used to test the relationship between the model variables. The path model was conducted using Mplus Version 8 software (Muthén and Muthén 1998–2010). The model assessed the mediating effect of peer and parental attachment on the bullying victimisation‐psychosis, bullying victimisation‐depression and bullying victimisation‐cannabis abuse relationships. Initially, the models were run with the WLSMV estimator to obtain model fit statistics. Model fit statistics used were chi square values, root mean square error of approximation (RMSEA); comparative fit index (CFI), Tucker‐Lewis index (TLI) and Standardised Root Mean Square Residual; The parameter estimates were then estimated using Maximum Likelihood estimation to handle missing data on the full Wave 7 (n = 2087) sample. The data were assumed to be missing at random as logistic regression models revealed that other variables within the analytical dataset were predictive of missingness (e.g., age 14 parental attachment and age 21 years depression predicted missingness on the psychosis measure at age 21 years). Due to the dichotomous nature of the psychosis outcome variable Montecarlo integration was used. Significance of indirect effects were assessed using bootstrapped‐bias corrected confidence intervals (with bootstrap samples set to 5000); these are particularly important for interpreting indirect effects as they do not rely on distributional assumptions (Hayes 2017; MacKinnon 2008). Indirect effects for which the confidence intervals did not overlap 0 are significant at α < 0.05.

3. Results

Sample characteristics are presented in Table 1 for those with and without self‐reported psychosis. Pearson correlation coefficients for the model variables are shown in Table 2.

TABLE 2.

Correlation matrix for model variables.

	1	2	3	4	5	6	7	8
1. Bullied
2. Peer attachment	0.192***
3. Parental attachment	0.139***	0.307***
4. Depression	0.104***	0.106***	0.229***
5. Psychosis	0.139***	0.078**	0.085**	0.325***
6. Cannabis abuse	0.067**	0.080**	0.121***	0.177***	0.270***
7. Gender	−0.089***	−0.303***	−0.064**	0.064**	−0.115***	−0.167***
8. FSM	0.077**	0.111***	0.064*	0.027	0.079**	0.125***	0.001

Note: One tailed Pearson correlations.

^* p < 0.05.

^** p < 0.01.

^*** p < 0.001.

The path analysis model tested is depicted in Figure1. The model demonstrated acceptable fit across a range of indices: ×2 = 11.42, df = 9, p = 0.25; RMSEA = 0.015; CFI = 0.997; TLI = 0.985; SRMR = 0.028.

The standardised path analysis results are presented in Table 3. Significant relationships between being bullied more frequently and subsequently having worse parental and peer attachment were evident. Poor peer attachment at age 13 years was linked to poor parental attachment one year later. Low levels of peer and parental attachment predicted greater levels of depressive symptoms in young adulthood; comparable significant relationships were not evident for the other mental health outcome variables. The direct paths from bullying victimisation to psychosis and depression were significant, with being bullied more frequently in childhood being associated with higher levels of depression and greater likelihood of self‐reporting psychosis symptoms in young adulthood. The path between bullying victimisation and the other mental health related outcome, cannabis abuse, was non‐significant. Higher levels of depression and cannabis abuse predicted greater likelihood of psychosis, and a significant positive bidirectional relationship between depression and cannabis abuse was found.

TABLE 3.

Summary of path analysis results.

	Standardised beta (Standard error), p value
Model effects
Bullying victimisation → Parental attachment	0.08 (0.03), p = 0.003
Bullying victimisation → Peer attachment	0.17 (0.03), p < 0.001
Bullying victimisation → Cannabis Abuse	0.03 (0.03), p = 0.251
Bullying victimisation →Depression	0.08 (0.03), p = 0.014
Bullying victimisation → Psychosis	0.10 (0.05), p = 0.030
Peer attachment → Parental attachment	0.30 (0.03), p < 0.001
Peer attachment → Depression	0.07 (0.03), p = 0.034
Peer attachment → Cannabis abuse	−0.01 (0.03), p = 0.816
Peer attachment → Psychosis	−0.03 (0.06), p = 0.646
Parental attachment → Depression	0.18 (0.03), p < 0.001
Parental attachment → Cannabis abuse	0.01 (0.02), p = 0.713
Parental attachment → Psychosis	−0.05 (0.06), p = 0.368
Depression → Psychosis	0.38 (0.03) p < 0.001
Cannabis Abuse → Psychosis	0.10 (0.05) p = 0.037
Depression ← → Cannabis Abuse	0.14 (0.03) p < 0.001
Covariates Effects (for Early Adolescence variables, ages 13–14 years)
Gender → Peer attachment	−0.29 (0.02), p < 0.001
Gender → parental attachment	0.06 (0.03), p = 0.026
Free School meals → Peer attachment	0.06 (0.03), p = 0.030
Free school meals → parental attachment	−0.04 (0.03), p = 0.161
Covariates Effects (for Young adulthood variables, age 21 years)
Cannabis history (Infrequent) → Depression	0.12 (0.02), p < 0.001
Cannabis history (Frequent) → Depression	0.11 (0.03), p < 0.001
Cannabis history (Previously used) → Depression	0.04 (0.02), p = 0.057
Cannabis history (Infrequent) → Cannabis abuse	0.06 (0.01), p < 0.001
Cannabis history (Frequent) → Cannabis abuse	0.71 (0.03), p < 0.001
Cannabis history (Previously used) → Cannabis abuse	0.14 (0.03), p < 0.001
Cannabis history (Infrequent) → Psychosis	0.05 (0.05), p = 0.309
Cannabis history (Frequent) → Psychosis	0.08 (0.06), p = 0.159
Cannabis history (Previously used) → Psychosis	0.11 (0.04), p = 0.006
Gender → depression	0.12 (0.02), p < 0.001
Gender → Cannabis abuse	−0.07 (0.02), p < 0.001
Gender → Psychosis	−0.17 (0.05), p = 0.001
Free school meals → Cannabis use	0.07 (0.02), p = 0.001
Free school meals → Psychosis	0.06 (0.05), p = 0.205

The present research also predicted that parental and peer attachment would mediate the effect of bullying victimisation on mental health (Table 4). Evidence of this was found for depression and psychosis, but not cannabis abuse. The significant indirect path to depression went through peer and then parental attachment (Bully → Peer Attachment → Parental attachment → Depression; estimate = 0.08; 95% CI = 0.05, 0.13). The Bully → Peer Attachment → Depression (0.10; 95% CI = 0.01, 0.22) and Bully → Parental Attachment → Depression (0.14; 95% CI = 0.05, 0.26) paths were also significant. The indirect effect path from bullying to psychosis purely through the peer and parental attachment variables was not found. There was, however, an indirect path to psychosis through peer attachment, then parental attachment followed by depression (Bully → Peer Attachment → Parental attachment → Depression → Psychosis; estimate = 0.01; 95% CI = 0.01, 0.02).

TABLE 4.

Summary results for indirect paths.

	Estimate (Bootstrapped bias‐corrected 95% CI)
Indirect effect
Bully →Parental attachment →Psychosis	−0.02 (−0.07, 0.02)
Bully → Peer Attachment →Psychosis	−0.02 (−0.10, 0.05)
Bully → Peer Attachment →Parental attachment → Psychosis	−0.01 (−0.03, 0.01)
Bully → Parental attachment →Depression	0.14 (0.05, 0.26)
Bully → Peer Attachment → Depression	0.10 (0.01, 0.22)
Bully → Peer Attachment → Parental attachment → Depression	0.08 (0.05, 0.13)
Bully → Parental attachment → Cannabis abuse	0.00 (−0.02, 0.03)
Bully → Peer Attachment →Cannabis abuse	−0.01 (−0.05, 0.04)
Bully → Peer Attachment → Parental attachment → Cannabis abuse	0.00 (−0.01, 0.01)
Bully → Peer Attachment → Parental attachment → Depression →Psychosis	0.01 (0.01, 0.02)
Bully → Peer Attachment → Parental attachment →Cannabis abuse →Psychosis	0.00 (−0.00, 0.00)

4. Discussion

4.1. Main Results

This study is the first to explore pathways between bullying in childhood and the development of psychotic symptoms. It has shown that being bullied in childhood is directly associated with a greater risk of having psychosis symptoms as an adult. Bullying victimisation in childhood also exerts an indirect effect on later risk of psychosis. The indirect pathway between bullying in adolescence to psychosis in young adulthood was shown to be mediated by both peer attachment, parental attachment, and depression. This mediation effect is modest.

Being bullied makes it difficult to form strong attachments to peers; consistent with research showing that social adjustment issues arise from being bullied (Hodges et al. 1997; Hodges and Perry 1999). Peer attachment predicted later parental attachment; however, this is most likely an artefact of earlier parental attachment problems generalising to later peer relationships (Meeus 1989). It is a limitation of the study that a measurement of earlier parental attachment is not available.

Individuals with attachment issues may be overwhelmed by the demands of managing increasingly sophisticated attachment relations, resulting in depressive symptoms (Allen and Land 1999). Individuals with depression are more likely to experience psychotic symptoms. Indeed, Garety and colleagues hypothesised that depressive symptoms could lead to the development of psychotic symptoms (Garety et al. 2001).

Lastly, cannabis did show a direct relationship with psychosis, adding to previous evidence (Large et al. 2011). However, it did not play a mediating role between bullying and psychosis within this study.

4.2. Implications

This report highlights the importance of anti‐bullying interventions within schools as a key preventative strategy. Anti‐bullying interventions have been shown to be effective at reducing bullying victimisation (Gaffney et al. 2019). Additionally, given the findings, mental health services should consider routinely enquiring about bullying as part of the assessment process.

Besides tackling childhood bullying, the other key area is parental and peer attachment. There is strong evidence for attachment theory and more recently growing evidence for attachment‐based interventions (e.g., Gregory et al. 2020). However, in theory by targeting parental attachment styles and helping young people move to a more secure attachment, future outcomes are likely to improve.

4.3. Strengths and Limitations

This BYDS study possesses several strong design features that will help us better understand the nature of the relationship between bullying and three related mental health outcomes. Firstly, the study employed a prospective design rather than asking participants to recall childhood experiences retrospectively. Secondly, the large number of time points in the study allows us to select independent, mediating, and outcome variables in a temporal order, thus strengthening conclusions.

Nonetheless, there are some limitations; for example, one such limitation of the study is the measure for bullying. Bullying was only explored using one question from the Strengths and Difficulties Questionnaire. This is a simplistic measure and could omit finer details such as exploring the effect of physical bullying, verbal bullying, and cyberbullying. Similarly, psychosis was measured through the Psychosis Screening Questionnaire (PSQ, (Bebbington and Nayani 1995)) which has been shown to be well validated but could have been improved with further objective clinical assessment. In the absence of clinical diagnostic data we were not able to validate/contextualise the self‐report data, and that given recent development of viable routes for linking research data to health data, future studies should seek to gain consent from participants for such purposes.

In addition, we could not control for the possible influence of preexisting mental health difficulties prior to age 11.

5. Conclusion

Being bullied in childhood is directly associated with a greater risk of having psychosis symptoms as an adult. There is also an indirect path from bullying in childhood to psychosis symptoms as an adult mediated by peer attachment, parental attachment and depression. Further research should focus on areas not measured in this study but likely to be important within the pathway (e.g., adverse childhood experiences and parental mental health problems). Interventions aimed at addressing bullying and strengthening social support from peers and family may well be an important preventative strategy.

Funding

This work was supported by Northern Health and Social Care Trust.

Data Availability Statement

The data that support the findings of this study are available from Queen's University Belfast, School of SSESW. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the author(s) with the permission of Queen's University Belfast, School of SSESW.