The association between transient childhood psychotic experiences and psychosocial outcomes in young adulthood: Examining the role of mental disorders and adult attachment

Abstract Aim Evidence suggest individuals with mental disorders and psychotic experiences (PE), even transient PE, show poorer psychosocial outcomes relative to those with mental disorders. The concept of “attachment” is hypothesized as the mechanism by which people seek support in times of need. This can be measured as discrete styles or as positive (low avoidance/anxiety)/negative (high avoidance/anxiety) dimensions. Adult attachment has previously been examined on PE risk factors, but not outcomes. This study aimed to examine the relationship between transient childhood PE and adult psychosocial outcomes, comparing those with and without mental disorders. Second, to examine the role of adult attachment. Method Participants (n = 103) attended baseline (age 11–13) and 10‐year follow‐up. PE and mental disorders were measured using the Schedule for Affective Disorders and Schizophrenia for School‐aged Children. Attachment and outcomes were measured using self‐report measures. Analysis compared those with PE (with/without mental disorders), and mental disorders without PE, to controls, using linear and Poisson regression. Results PE was associated with lower self‐esteem (β = −2.28, p = .03), perceived social support from friends (β = −2.80, p = .01), and higher stress in platonic relationships (IRR = 1.64). PE and mental disorders were associated with lower self‐esteem (β = −5.74, p = .002), higher stress in romantic (IRR = 1.40) and platonic (IRR = 1.59) relationships, general stress (β = 5.60, p = .006), and mental distress (β = 5.67, p = .001). Mental disorders alone was not associated with any measure. Adult attachment dimensions attenuated some results. Conclusions This paper illustrates the association between transient PE and adult psychosocial outcomes, with & without co‐occurring mental disorders, and demonstrates the role of adult attachment.

Evidence shows that PE, even when transient, are associated with long-term adverse outcomes (Carey et al., 2020;Healy et al., 2018).
Mental disorders and PE are often co-occurring (Healy et al., 2019).
Those who report both PE and mental disorders have worse outcomes than those with only mental disorders, including elevated risk of additional subsequent mental disorders , suicide behaviours (Bromet et al., 2017), and increased use of mental health services (Bhavsar et al., 2021). To date less research has focused on non-clinical outcomes, such as psychosocial outcomes. Trotta et al. (2020) found PE in childhood were associated with poor psychosocial outcomes (loneliness, lower quality of life and higher rates of criminal behaviour) in adulthood, but the wider scope of psychosocial outcomes following PE, are currently under-examined.
Attachment is hypothesized as an innate psychobiological dependency infants have for care givers, which depending on positive/ adverse support from caregivers, develops into positive/negative attachment (Bowlby, 1982). Attachment was originally hypothesized to form in early childhood and remain stable (Ravitz et al., 2010), but more recent evidence has failed to support this (Badovinac et al., 2021;Fearon & Roisman, 2017;Mullen, 2019). Evidence does support that attachment can reliably predict different outcomes; Adult attachment has been linked to differences in self-esteem, mental health difficulties, and response to treatment (Foster et al., 2007;Kuipers & Bekker, 2012;Mullen, 2019). Attachment can be measured as discrete styles, although more recent evidence supports a dimensional approach (Fearon & Roisman, 2017). Child & adult attachment has previously been shown to mediate the relationship between trauma and rates of PE (Sheinbaum et al., 2014(Sheinbaum et al., , 2020. Its role in outcomes following PE, are currently unknown. This paper aims to examine the effect of transient PE on adult psychosocial outcomes, examining the differences between those who report both PE and mental disorders, PE, and mental disorders, compared to controls. Secondly we examined the role of adult attachment.

| Recruitment and participants
Participants were recruited as part of the Adolescent Brain Development study, [for full details see (Kelleher, Murtagh, et al., 2012)].
Briefly, at baseline 211 participants aged 11-13 (mean age 11.7), were recruited from primary schools and invited for in-depth clinical interviews. All participants were invited to return for a follow-up 10 years later. At follow-up, 103 participants returned [for full details see (Carey et al., 2020)]. Only individuals who completed the adult followup assessment are included within this study.

| Demographic and clinical information
At baseline demographic information on age and sex were collected.
Victimization was based on presence of the adverse experiences of physical abuse, sexual abuse or bullying, (Coughlan et al., 2021). Victimization was a dichotomous variable, using binary (present/not) for physical and sexual abuse, bullying was treated as binary using the threshold of above the median score for distress (4/10) for incidents of bullying.

| Psychotic experiences
At baseline, participants completed clinical screenings using the Schedule for Affective Disorders and Schizophrenia for School-aged Children, Present and Lifetime Versions (K-SADS) (Kaufman et al., 1997). At 10-year follow-up participants were interviewed using the psychosis section of the Structured Clinical Interview for DSM-5 (First et al., 2015) & additional to questions from the SOCRATES instrument (Kelleher & Cannon, 2016).

| Mental disorders
At baseline the K-SADS measure (Kaufman et al., 1997) was used to assess Axis 1 mental disorders, measuring current and lifetime morbidity, classified by the DSM-IV of Axis I disorders (American Psychiatric Association, 1994). Only individuals who reported experiences of anxiety and post-traumatic stress disorders, mood disorders, eating disorders, substance disorders, conduct disorder or psychotic disorders were considered to have mental disorders in analysis. Simple phobias were not considered.

| Psychosocial outcomes assessment
All psychosocial measures are self-report measures and were collected at the follow-up interview.

| Social support
The Multidimensional Scale of Perceived Social Support (MPSS) is a 12 item measure of perceived social support good internal reliability (α = 0.84-0.92 (Zimet et al., 1990)).

| Mental distress
The General Health Questionnaire (GHQ) is a measure of psychiatric impairment and strain (Banks et al., 1980;El-Metwally et al., 2018).

Stress
(1) The Perceived Stress Scale (PSS), is a 10 item measure of the degree to which daily life was stressful in the last month (Cohen et al., 1983). PSS shows good test-retest reliability (α = 0.74-0.91) (Lee, 2012).

| Adult attachment assessment
The revised adult attachment scale (RAAS) is an 18-item self-report questionnaire, used to measure adult attachment (Collins & Read, 1990). Attachment dimensions were used within this study. The RAAS shows fair test-retest reliability (α = 0.58) (Ravitz et al., 2010).

| Statistical analysis
All analysis was conducted using RStudio (R Core Team, 2020). To assess the effects of transient childhood PE in the analysis, recurring PE (n = 2) were excluded from all analysis. Included participants were divided into four groups for all subsequent analysis; those with child- 2. Model 2 included the confounders of Model 1 and additionally adult attachment anxiety and avoidance.
Linear regression was used for the continuous measures GHQ-12, MPSS, PSS and the RES. Effect size was reported using the localized measure of Cohen's f 2 measure, f 2 ≥ 0.02, f 2 ≥ 0.15, and f 2 ≥ 0.35 represent small, medium, and large effect sizes, respectively (Selya et al., 2012). The adapted SLES was used as a count measure and data showed a positive skew. Therefore a Poisson regression was used.
Robust estimate and standard errors were calculated (Cameron & Trivedi, 2009). Within this analysis effect size was measured using incident rate ratio (IRR).

| Attrition analysis
No differences between those who returned at follow-up and those who did not were found in sex (χ 2 = 0.01, p = 1.0), age (t = À1,

| Demographic and psychosocial characteristics
In the sample of 103 participants who returned to follow up, two reported recurring PE and were excluded. 29 reported transient PE in childhood, and 58 reported no mental disorders or PEin childhood.
Demographic and psychosocial differences were calculated (Table 1).
Compared to controls, those reporting childhood PE were more likely to be male, and report higher rates of victimization (Table 1). In psychosocial outcomes, PE showed higher levels of stress in daily life and platonic relationships, and lower perceived support from friends (Table 1). No significant differences were found in mental distress, perceived social support from family or significant others, or stress in romantic relationships, between PE and control (Table 1).
This sample was further divided based on those with childhood PE (n = 14; 13.9%), mental disorders (n = 14; 13.9%), both PE and mental disorders (n = 15; 14.9%) and controls (n = 58; 57.4%), and analysis of demographic and psychosocial differences were conducted (Table 2). In demographic measures, there was a significant difference at a group level in victimization but not age or sex (Table 2). Additionally, there were significant group level differences in perceived social stress (Table 2).

| Longitudinal relationship of childhood PEs and mental disorders on individual outcomes
In Model 1 participants reporting PE, with and without mental disorders, were significantly related to lower self-esteem compared to controls, a medium and small effect size, respectively. (Table 3; Figure 1). Only those reporting both PE and mental disorders in childhood were related with higher levels of general stress and mental distress compared to controls, of small and medium effect size (Table 3; Figure 1). Those reporting mental disorders but not PE in childhood showed no association with later individual psychosocial outcomes, but showed a moderate trend of higher stress levels compared to controls (Table 3; Figure 1). When adult attachment anxiety and avoidance were added (Model 2), the model completely attenuated the association between childhood PE and poor adult self-esteem, in both the PE and PE and mental disorders group (Table 3). In those reporting PE and mental disorders, worse general stress relative to controls was similarly attenuated. Mental distress remained significantly higher compared to controls even when adult attachment was accounted for, with a small effect size (Table 3).

| Longitudinal relationship of childhood PE and mental disorders on social outcomes
In Model 1, accounting for sex and victimization, childhood PE and mental disorders was associated with elevated levels of romantic and platonic stress, but no difference was found compared to controls in perceived social support (Table 4; Figure 1). Model 2 including adult attachment dimensions accounted for the difference compared to controls for platonic stress, but not the higher rates of stress in T A B L E 1 Demographic and psychosocial differences between individuals reporting PEs and controls  Figure 1). Model 2 including adult attachment anxiety and avoidance accounted for the perceived social support in friendship, but it continues to show a general trend. The measure of stress in platonic relationship remained stable with the inclusion of adult attachment, and with a moderate incidence rate ratio ( Table 4). The comparison measure of childhood mental disorders with no PE showed no effect on any measure, in either Models 1 or 2. However a trend of higher levels of stress in romantic and platonic relationship was observable, though non-significant, and showed a moderate incident rate ratio in both Models 1 and 2 (Table 4; Figure 1).

| DISCUSSION
This paper set out to examine the role of transient childhood PE on outcomes in later life, accounting for the role of adult attachment.
Two key findings emerged; (1) Transient PEin childhood, with and without mental disorders, were associated with poorer outcomes in adulthood in the areas of: self-esteem, general stress, perceived social support, and stress in romantic and platonic relationships. Mental disorders in childhood was not found to be related directly to any adult psychosocial outcomes.
(2) Including adult attachment into these models significantly attenuated the relationship between childhood PE and certain adult psychosocial measures.
The primary finding of this paper was the long term association between childhood PE on adult outcomes. Transient PE, even in the absence of co-occurring mental disorders, was associated with lower self-esteem and perceived social support from friends, and higher levels of stress in platonic relationships. This study supports indicating that childhood PE are a marker of poor mental health which can show sustained negative trajectories into adulthood (Bhavsar et al., 2021;Bromet et al., 2017;Carey et al., 2020;Healy et al., 2018Healy et al., , 2019Trotta et al., 2020), showing this association is relevant in psychosocial outcomes, in addition to clinical outcomes.
Mental disorders without PE, were not found to be significantly related to any psychosocial measure, but showed a moderate trend in all measures of stress. One explanation for this may be that childhood mental disorders in the absence of PE may represent a less severe trajectory. Previous research has reported findings similar to this study . Alternatively, psychotic experiences have been proposed as a marker of severe psychopathology (Ajnakina et al., 2019;Guloksuz & van Os, 2018;Stochl et al., 2015).
While these participants met criteria for a mental disorder, they perhaps had less severe symptoms, which allowed them to resolve prior to adulthood without this adverse effect.
Psychotic experiences and mental disorders in childhood showed particularly poor outcomes compared to controls. PE and mental disorders in childhood was associated with lower self-esteem and higher rates of general stress, and stress in romantic and platonic relationships in adulthood. This study is in line with previous work, which has shown those with both PE and mental disorders report worse functioning, and higher rates of suicidal ideation, suicidal behaviour, and use of mental health services (Bhavsar et al., 2021;Bromet et al., 2017;Kelleher et al., 2015).  (Gawęda et al., 2018;Sheinbaum et al., 2014Sheinbaum et al., , 2020. This study furthers this research by indicating that in adulthood, high attachment anxiety and avoidance may better explain certain psychosocial outcomes, which were previously suggested to be associated with PEs, such as self-esteem (Gawęda et al., 2012;Hafeez & Yung, 2021