Introduction
According to global estimates, 7–36% of women and 3–29% of men report being victims of sexual abuse during childhood (Pereda et al., Reference Pereda, Guilera, Forns and Gomez-Benito2009). In Australia, the prevalence is 14–20% for women and 3–10% for men (Moore et al., Reference Moore, Romaniuk, Olsson, Jayasinghe, Carlin and Patton2010). The evidence increasingly supports strong associations between childhood sexual abuse and psychotic disorders in later life. Retrospective studies show that a high proportion of people with psychotic disorders have previously experienced sexual abuse (Bendall et al., Reference Bendall, Jackson, Hulbert and McGorry2008; Cutajar et al., Reference Cutajar, Mullen, Ogloff, Thomas, Wells and Spataro2010; Bebbington et al., Reference Bebbington, Jonas, Kuipers, King, Cooper, Brugha, Meltzer, McManus and Jenkins2011; Mauritz et al., Reference Mauritz, Goossens, Draijer and van Achterberg2013). Research has also shown strong associations between schizophrenia, as well as other mental disorders including depression, anxiety, post-traumatic stress disorder, dissociative disorders, substance dependence and self-harm and childhood abuse, including sexual, physical and emotional abuse, and neglect (Van der Kolk et al., Reference Van der Kolk, Roth, Pelcovitz, Sunday and Spinazzola2005; Cutajar et al., Reference Cutajar, Mullen, Ogloff, Thomas, Wells and Spataro2010; Gonzalez et al., Reference Gonzalez, Boyle, Kyu, Georgiades, Duncan and MacMillan2012; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012; Matheson et al., Reference Matheson, Shepherd, Pinchbeck, Laurens and Carr2013; van Nierop et al., Reference van Nierop, Viechtbauer, Gunther, van Zelst, de Graaf, ten Have, van Dorsselaer, Bak and van Winkel2015; Palmier-Claus et al., Reference Palmier-Claus, Berry, Bucci, Mansell and Varese2016). Although it has long been recognised that stressful life events and difficulties contribute to depression and anxiety (see Craig, Reference Craig, Morgan and Bhugra2010), less is known about how relatively common adverse experiences (e.g. parental separation) may contribute to psychoses.
Psychosocial outcomes refer to a person's recovery in areas such as vocational and social functioning (Killackey et al., Reference Killackey, Harvey, Amering, Herrman, Tasman, Kay, Lieverman, First and Riba2015; Stolerman and Price, Reference Stolerman, Price, Stolerman and Price2015). This is an important but neglected area of research (Fossey and Harvey, Reference Fossey and Harvey2001; Misiak et al., Reference Misiak, Krefft, Bielawski, Moustafa, Sasiadek and Frydecka2017). Poor psychosocial outcomes directly affect mental health consumers and their families as well as contributing to a general impact on quality of life (National Mental Health Commission, 2014; Cotter et al., Reference Cotter, Kaess and Yung2015; Poon et al., Reference Poon, Harvey, Mackinnon and Joubert2017). For these reasons, understanding the antecedents of poor psychosocial outcomes is relevant for consumers and families, clinicians, policymakers and researchers alike (Felitti et al., Reference Felitti, Anda, Nordenberg, Williamson, Spitz, Edwards, Koss and Marks1998). Improved understanding of psychosocial stressors both preceding and at illness onset also offers an opportunity for prevention and early intervention. In a previous study, we examined the relationship between childhood abuse and psychosis and reported high rates (31%) of any childhood abuse before illness onset (Shah et al., Reference Shah, Mackinnon, Galletly, Carr, McGrath, Stain, Castle, Harvey, Sweeney and Morgan2014), but our investigation mostly focused on clinical and physical health outcomes, rather than psychosocial outcomes and the recurrence of psychosocial stressors.
Research on psychosocial outcomes among people with psychosis has highlighted that individuals with significant abuse histories often have difficulty with employment, housing and social functioning (Gil et al., Reference Gil, Gama, de Jesus, Lobato, Zimmer and Belmonte-de-Abreu2009; Alameda et al., Reference Alameda, Ferrari, Baumann, Gholam-Rezaee, Do and Conus2015; Cotter et al., Reference Cotter, Kaess and Yung2015). Other studies have drawn a connection between early and subsequent exposure to psychosocial stressors (e.g. victimisation, homelessness, crime and offending) and an increased likelihood of developing a psychotic response (Rosenberg et al., Reference Rosenberg, Lu, Mueser, Jankowski and Cournos2007; Pluck et al., Reference Pluck, Lee, David, Macleod, Spence and Parks2011; Lataster et al., Reference Lataster, Myin-Germeys, Lieb, Wittchen and van Os2012). These findings are interesting but limited because of a reliance on single questions to assess trauma – usually defined as child abuse (physical and sexual) or sexual abuse only; little consideration of potential confounders (for instance, family history of mental illness and cannabis use); examination of a small number of outcomes and limited power with only a small number of individuals experiencing the exposure or the outcome (Gil et al., Reference Gil, Gama, de Jesus, Lobato, Zimmer and Belmonte-de-Abreu2009; Hooven et al., Reference Hooven, Nurius, Logan-Greene and Thompson2012; Berthelot et al., Reference Berthelot, Paccalet, Gilbert, Moreau, Merette, Gingras, Rouleau and Maziade2015; Trauelsen et al., Reference Trauelsen, Bendall, Jansen, Nielsen, Pedersen, Trier, Haahr and Simonsen2015; Cristobal-Narvaez et al., Reference Cristobal-Narvaez, Sheinbaum, Ballespi, Mitjavila, Myin-Germeys, Kwapil and Barrantes-Vidal2016; Gallagher and Jones, Reference Gallagher and Jones2016; Holshausen et al., Reference Holshausen, Bowie and Harkness2016).
Our aim was to address some of the gaps and limitations in previous studies by examining all forms of abuse and other negative experiences in childhood and their relationships with a range of outcome variables, adjusting for a wide variety of potential confounders. These are: age (Bentall et al., Reference Bentall, Wickham, Shevlin and Varese2012), sex (Bentall et al., Reference Bentall, Wickham, Shevlin and Varese2012; Matheson et al., Reference Matheson, Shepherd, Pinchbeck, Laurens and Carr2013), currently married or previously married (Bendall et al., Reference Bendall, Jackson, Hulbert and McGorry2008), presence of children or step children (Bendall et al., Reference Bendall, Jackson, Hulbert and McGorry2008), born in Australia (Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012), highest qualification is a school-level qualification (Bendall et al., Reference Bendall, Jackson, Hulbert and McGorry2008; Bentall et al., Reference Bentall, Wickham, Shevlin and Varese2012; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012), family history of mental illness (Bendall et al., Reference Bendall, Jackson, Hulbert and McGorry2008; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012; Matheson et al., Reference Matheson, Shepherd, Pinchbeck, Laurens and Carr2013), socioeconomic status of the participant's residence (Bentall et al., Reference Bentall, Wickham, Shevlin and Varese2012; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012) and lifetime cannabis use (Janssen et al., Reference Janssen, Krabbendam, Bak, Hanssen, Vollebergh, de Graaf and van Os2004; Bendall et al., Reference Bendall, Jackson, Hulbert and McGorry2008; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012). Specifically, we examined the association between exposure to diverse adverse childhood events and a number of mental health symptoms and diagnoses (Bendall et al., Reference Bendall, Jackson, Hulbert and McGorry2008; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012; Van Nierop et al., Reference van Nierop, Viechtbauer, Gunther, van Zelst, de Graaf, ten Have, van Dorsselaer, Bak and van Winkel2015) as well as important psychosocial outcomes (Rosenberg et al., Reference Rosenberg, Lu, Mueser, Jankowski and Cournos2007; Gilbert et al., Reference Gilbert, Widom, Browne, Fergusson, Webb and Janson2009; Alameda et al., Reference Alameda, Ferrari, Baumann, Gholam-Rezaee, Do and Conus2015). Our primary hypothesis was that any adverse event in childhood (defined as sexual, physical and emotional abuse and neglect, and interpersonal loss) would be associated with clinical and psychosocial impairment. Our secondary hypothesis was that different types of abuse as well as interpersonal loss would also be associated with impairment.
Method
Overview
This study analysed data collected from the 2010 Australian National Survey of Psychosis (Survey of High Impact Psychosis, SHIP) (Morgan et al., Reference Morgan, Waterreus, Jablensky, Mackinnon, McGrath, Carr, Bush, Castle, Cohen, Harvey, Galletly, Stain, Neil, McGorry, Hocking, Shah and Saw2012, Reference Morgan, McGrath, Jablensky, Badcock, Wattereus, Bush, Carr, Castle, Cohen, Galletly, Harvey, Hocking, McGorry, Neil, Saw, Shah, Stain and Mackinnon2014). The survey methodology is described in detail elsewhere (Morgan et al., Reference Morgan, Waterreus, Jablensky, Mackinnon, McGrath, Carr, Bush, Castle, Cohen, Harvey, Galletly, Stain, Neil, McGorry, Hocking, Shah and Saw2012; Shah et al., Reference Shah, Mackinnon, Galletly, Carr, McGrath, Stain, Castle, Harvey, Sweeney and Morgan2014). A cross-sectional survey was conducted across seven catchment sites in five states. Because psychosis is comparatively rare, sampling was done in two stages. A census of people with psychosis was taken from three mutually exclusive groups: people in contact with public mental health services during the census month (March 2010); people not in contact with those services, but in contact with non-government organisations supporting people with mental illness during the census month and people not in contact with either of these two services during the census month but in contact with public mental health services during the preceding 11 months. All those identified in the census were screened using the Psychosis Screen (Jablensky et al., Reference Jablensky, McGrath, Herrman, Castle, Gureje, Evans, Carr, Morgan, Korten and Harvey2000) and 7955 screened positive. From this, a random sample (n = 1825) stratified by age (18–34 years and 35–65 years) were selected and interviewed by trained mental health professionals. The response rate was 44% and there was no evidence of selection bias (Morgan et al., Reference Morgan, Waterreus, Jablensky, Mackinnon, McGrath, Carr, Bush, Castle, Cohen, Harvey, Galletly, Stain, Neil, McGorry, Hocking, Shah and Saw2012). The study was approved by ethics committees at the seven study sites. Participants provided written, informed consent.
Adverse event variables
The primary exposure of interest was the reported occurrence of an adverse event in childhood, defined as experiencing sexual abuse, other abuse (physical abuse, emotional abuse and neglect) or interpersonal loss (loss of a parent, sibling or close relative, parental separation or divorce). Only cases where the adverse event occurred during childhood (⩽18 years) and before the onset of psychotic illness were included.
Our adverse event variables were constructed from responses to a series of questions. Participants were initially asked the following questions: ‘Did your parents separate or divorce when you were a child?’; ‘Did you suffer the loss of a parent when you were a child?’; ‘Did you suffer the loss of a brother or sister when you were a child?’; ‘Did you suffer the loss of another close relative when you were a child?’ and ‘Were there any other distressing or traumatic events in your childhood?’. The age when each event happened was also recorded. These first four questions were used to construct a variable representing the presence or absence of interpersonal loss in childhood.
If participants reported any other distressing or traumatic events, then interviewers took notes of what had happened. These qualitative responses were coded into two main categories: sexual abuse and other types of abuse (defined as physical abuse, emotional abuse, neglect and other forms of abuse), based on Australian guidelines (Australian Institute of Health and Welfare, 2012). Further details about these variables are described elsewhere (Rosenman and Rodgers, Reference Rosenman and Rodgers2004; Shah et al., Reference Shah, Mackinnon, Galletly, Carr, McGrath, Stain, Castle, Harvey, Sweeney and Morgan2014)
We constructed a variable representing the occurrence of any of these adverse events in childhood (0 v. ⩾1). We also separated them out into three variables: sexual abuse (yes, no), physical and emotional abuse and neglect (yes, no) and interpersonal loss (yes, no). This coding allowed for the possibility that an individual could experience more than one type of adverse event. Finally, to understand the effect of each type of adverse event in isolation (in particular, childhood adversity in the absence of abuse), we constructed three hierarchically coded variables. Reported sexual abuse trumped physical and emotional abuse and neglect and, physical and emotional abuse and neglect trumped interpersonal loss, such that individuals could only be counted once in the set of analyses.
Outcome variables
The SHIP interview schedule included a number of embedded instruments (Morgan et al., Reference Morgan, Waterreus, Jablensky, Mackinnon, McGrath, Carr, Bush, Castle, Cohen, Harvey, Galletly, Stain, Neil, McGorry, Hocking, Shah and Saw2012). Symptom profile and mental health diagnoses were determined using the Diagnostic Interview for Psychosis (DIP) (Castle et al., Reference Castle, Jablensky, McGrath, Carr, Morgan, Waterreus, Valuri, Stain, McGuffin and Farmer2006). The DIP allows lifetime, previous year and previous 4 weeks’ ratings of symptoms; diagnoses can be assigned according to ICD-10 (World Health Organisation, 1992) and DSM-IV (American Psychiatric Association, 2000) criteria. The DIP contains selected interview questions and probes from the WHO Schedules for Clinical Assessment in Neuropsychiatry (Wing et al., Reference Wing, Babor, Brugha, Burke, Cooper, Giel, Jablenski, Regier and Sartorius1990) mapped on the OPCRIT (McGuffin et al., Reference McGuffin, Farmer and Harvey1991). DIP scores are fed into the computerised OPCRIT algorithm which produces diagnoses according to, in this case, the 10th edition of the International Classification of Disease (ICD-10) (World Health Organisation, 2015).
Using the DIP, we identified six symptoms or symptom clusters occurring over the lifetime. These were depressive symptoms, mania (elevated or irritable mood), self-reproach, delusions, hallucinations and subjective thought disorder. We used ratings of suicidal thoughts in the previous 4 weeks and self-harm in the previous 12 months. We also identified the presence of two additional symptoms and syndromes: anxiety in the previous 12 months using relevant Schedules for Clinical Assessment in Neuropsychiatry items (SCAN) (Wing et al., Reference Wing, Curtins and Beevor1999; Bolton, Reference Bolton, Morgan and Bhugra2010), and a count of the number of negative symptoms over the previous 12 months (using SCAN items of the Carpenter deficits syndrome) (Kirkpatrick et al., Reference Kirkpatrick, Buchanan, McKenney, Alphs and Carpenter1989).
We identified three diagnoses occurring over the lifetime, all using ICD-10 criteria (World Health Organisation, 1992). These were alcohol and other substance abuse and dependence (excluding cannabis), and diagnoses for affective psychosis and non-affective psychosis. Affective psychosis was defined as a diagnosis of bipolar disorder or mania (both in the presence of psychotic symptoms); non-affective psychosis by a diagnosis of schizophrenia, schizoaffective disorder or delusional disorders and other non-organic psychoses. Diagnoses of substance use and affective psychosis/non-affective psychosis were not mutually exclusive due to possible comorbidity.
Finally, we included five variables measuring psychosocial outcomes. Four of these were based on previous 12 months and were: behavioural and social functioning including role performance (assessed using the Personal and Social Performance Scale (PSP) (Morosini et al., Reference Morosini, Magliano, Brambilla, Ugolini and Pioli2000)); experience of victimisation (ABS Crime and Safety Survey) (Australian Bureau of Statistics, 2005); perpetration of crime or offending (extracted from the section on criminality within the Opiate Treatment Index (Darke et al., Reference Darke, Ward, Hall, Heather and Wodak1991)) and homelessness (Homelessness and Mental Health Survey (Robinson, Reference Robinson2003)). The fifth psychosocial outcome was the presence of a definite psychosocial stressor within 12 months of the first psychotic episode.
PSP scores range from 0 to 100, where 100 represents excellent functioning. The recognised cut-point of 70 or less is used to indicate poor functioning (scores above this value indicate the absence of disability or only mild difficulties). Experience of victimisation was measured using four items (number of times the participant was a victim of a break-in, theft, assault or was threatened with assault) and coded 0 v. ⩾1. Homelessness was measured by the number of homeless days in the previous year (coded 0 v. ⩾1 day). Perpetration of crime was measured using a positive response to the question: ‘in the last 12 months have you committed any crimes such as break and enter, shoplifting, dealing drugs, forging cheques, armed robbery or assault?’ Psychosocial stressors included potentially significant and/or threatening events that occurred during the 12 months preceding onset of psychosis such as: important problems with primary support groups, educational problems, occupational problems, housing problems, economic problems, legal problems and other problems.
Statistical analysis
For our primary analysis we used logistic regression to model the effects of exposure to any adverse event on each of the 18 outcome variables. As a sensitivity analysis, we then repeated the analysis restricted to people with non-affective psychosis (‘sensitivity analysis 1’) and affective psychosis (‘sensitivity analysis 2’). For our secondary analysis we modelled separately the effects of exposure to each of the specific types of abuse or loss (sexual abuse, physical and emotional abuse and neglect, interpersonal loss) on the outcome variables. Finally and as a third sensitivity analysis, we modelled the effect of the three specific types of abuse or loss using the hierarchically coded variables whereby sexual abuse trumped physical and emotional abuse and neglect which trumped interpersonal loss (‘sensitivity analysis 3’). To account for potential confounding between the adverse event variables and the outcome variables, all models included age (18–25, 26–35, 36–45, 46–65), sex, currently married or in a de facto relationship or previously married (yes, no), presence of children or step children (yes, no), born in Australia (yes, no), highest qualification is a school-level qualification (yes, no), family history of mental illness (yes, no), socioeconomic status of the participant's residence (in quintiles) and lifetime cannabis use (yes, no). Socioeconomic status was based on the Index of Relative Socioeconomic Disadvantage (IRSED) (Australian Bureau of Statistics, 2011). We report the effect sizes for the adverse event exposures only.
Results
Characteristics of the sample
Among the 1825 participants, 14% were aged 18–25 years and 28% were aged 46–65 years (Table 1). Most were male (60%), born in Australia (82%), and had a family history of mental illness (58%). Forty-nine percent were married/in a de facto relationship or had been married and 40% had children or step children. Thirty-five percent lived in neighbourhoods of low socio-economic status (lowest two IRSED quintiles). Forty-seven percent met the ICD-10 criteria for schizophrenia, 16% schizoaffective disorder, 18% bipolar disorder or mania (both with psychotic symptoms), 9% severe depression without psychosis, 5% delusional disorder, 4% depressive psychosis, 1% screened positive for psychosis but did not meet full ICD-10 criteria for psychosis. About 9% were classified as dependent drinkers using the AUDIT and over two-thirds had a lifetime history of cannabis use. The prevalence of the outcomes ranged from 9% for homelessness (⩽12 months) to 87% reporting delusions (lifetime). Concerning functional outcomes other than homelessness, 39% reported victimisation, 11% had been involved in some type of crime and offending and 82% reported social dysfunction (all ⩽12 months).
Table 1. Characteristics of participants, n = 1825
Note: Numbers may not sum to 1825 due to missing data. IRSED is Index of Relative Socioeconomic Disadvantage.
Exposure to adverse events
Eighty percent of participants had experienced an adverse event in childhood – 16% had experienced sexual abuse, 17% had experienced other forms of abuse and 72% had experienced interpersonal loss. Parental separation or divorce was the most common interpersonal loss (40%).
Outcomes associated with any adverse event
The occurrence of any adverse childhood event was associated with increased odds of 12/18 symptoms, diagnoses and psychosocial outcomes (Table 2). For instance, those who had experienced a childhood adverse event had 1.68 times higher odds of having depression compared with those who had not experienced an adverse event. The largest effect sizes were for homelessness (odds ratio (OR) = 2.82), participation in crime and offending in the previous 12 months (OR = 1.95) and suicidal thoughts in the previous 4 weeks (OR = 1.90). Of the three diagnoses examined, the only association observed was for alcohol and other substance dependence (OR = 1.39).
Table 2. Logistic regression models estimating the effect of any adverse event on 18 variables measuring symptoms, diagnoses and functioning (primary and sensitivity analyses)
Significant values are in bold.
Note: Models adjusted for age, sex, marital status, parental status, born in Australia, highest qualification obtained, family history of mental illness, socioeconomic status and lifetime cannabis use.
In a sensitivity analysis where the sample was restricted to people with non-affective psychosis, these findings largely held, with any adverse event in childhood still associated with eight outcomes. Most ORs were of similar magnitude to the primary analysis. In a second sensitivity analysis restricted to people with affective psychosis, an adverse event in childhood was associated with only three outcomes. Any adverse event in childhood was associated with lower odds of mania (OR = 0.28) and higher odds of self-reproach (OR = 2.06) and a definite psychosocial stressor within 12 months of first episode (OR = 1.87). Some large ORs were non-significant, suggesting that the reduced sample size used for these analyses may have been underpowered to detect effects.
Outcomes associated with specific adverse events
Table 3 shows the association between three specific adverse events and the outcomes under investigation. Sexual abuse in childhood predicted 10/18 outcomes. Other forms of abuse (physical abuse, emotional abuse and neglect) were associated with 13 outcomes, and experiencing interpersonal loss (loss of a parent, loss of a sibling, loss of a close relative, parental separation or divorce) was associated with four outcomes. Common across all types of adverse events was the association with depression, anxiety and a definite psychosocial stressor within 12 months of first episode.
Table 3. Secondary analysis: logistic regression models estimating the effect of sexual abuse, other abuse and interpersonal loss on 18 variables measuring symptoms, diagnoses and measures of functioning, n = 1793
Significant values are in bold.
Note: Models adjusted for age, sex, marital status, parental status, born in Australia, highest qualification obtained, family history of mental illness, socioeconomic status and lifetime cannabis use.
The relationship between sexual abuse in childhood and the outcome variables was strongest for depression (OR = 2.16), self-harm (OR = 1.79) and a definite psychosocial stressor within 12 months of first episode (OR = 1.66). The effect of other forms of abuse was strongest for homelessness (OR = 1.92), participation in crime and offending (OR = 1.84) and victimisation (OR = 1.80). Interpersonal loss in childhood was associated with homelessness (OR = 1.91), depression (OR = 1.41), anxiety (OR = 1.26) and definite psychosocial stressor within 12 months of first episode (OR = 1.26).
In the third sensitivity analysis where abuse and loss were coded hierarchically, the results for sexual abuse and other forms of abuse were broadly similar to that reported in the non-hierarchically coded analysis (Table 4). However, a different picture emerged for interpersonal loss. This form of adversity (in the absence of the other two) was associated with lower odds of self-reproach (OR = 0.70), negative syndrome (OR = 0.75) and victimisation (OR = 0.82).
Table 4. Sensitivity analysis 3: logistic regression models estimating the effect of sexual abuse, other abuse and interpersonal loss (hierarchically coded) on 18 variables measuring symptoms, diagnoses and measures of functioning, n = 1793
Significant values are in bold.
Note: Models adjusted for age, sex, marital status, parental status, born in Australia, highest qualification obtained, family history of mental illness, socioeconomic status and lifetime cannabis use.
Discussion
This study is unique in using a large representative population of people with psychosis to examine the prevalence of diverse adverse events in childhood and their relationship with symptomatic and psychosocial outcomes. It adds to previous research by collecting systematic data on a wide range of clinical and psychosocial outcomes, some previously unexamined in this context (including homelessness and the presence of a definite psychosocial stressor within 12 months of illness onset).
Four in five Australians living with psychosis reported experiencing some form of adverse event in childhood and approximately one in six reported exposure to childhood sexual abuse on at least one occasion before 18 years of age. However, since ascertainment of abuse histories by retrospective rather than prospective methods (Hooven et al., Reference Hooven, Nurius, Logan-Greene and Thompson2012; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012), or by face-to-face interviews compared with confidential self-report measures (Bendall et al., Reference Bendall, Jackson, Hulbert and McGorry2008), tends to underestimate this type of abuse, we expect this figure may be greater. Our reported prevalence of childhood sexual abuse within this population falls within previously reported ranges (e.g. Gustafson and Sarwer, Reference Gustafson and Sarwer2004; Casey and Nurius, Reference Casey and Nurius2005; Messman-Moore et al., Reference Messman-Moore, Walsh and DiLillo2010). These findings support previously reported increased rates of childhood adversity in people with diagnoses of mental illness when compared to the general population (Morgan and Fisher, Reference Morgan and Fisher2007; Cutajar et al., Reference Cutajar, Mullen, Ogloff, Thomas, Wells and Spataro2010; McCabe et al., Reference McCabe, Maloney, Stain, Loughland and Carr2012; Mauritz et al., Reference Mauritz, Goossens, Draijer and van Achterberg2013).
The other key finding is that more than one type of adversity appears to be associated with the development of psychosis (van Nierop et al., Reference van Nierop, Lataster, Smeets, Gunther, van Zelst, de Graaf, ten Have, van Dorsselaer, Bak, Myin-Germeys and Viechtbauer2014). Our results also suggest that childhood interpersonal loss, occurring in the absence of other forms of abuse, contributes to less severe symptom and functioning impacts within the population of people with psychosis.
A number of functional outcomes were strongly associated with physical and emotional abuse and neglect, especially victimisation, offending and homelessness, adding to a growing body of evidence about the functional sequelae of childhood adversity (e.g. Cantor-Graae, Reference Cantor-Graae2007; Rosenberg et al., Reference Rosenberg, Lu, Mueser, Jankowski and Cournos2007; Gil et al., Reference Gil, Gama, de Jesus, Lobato, Zimmer and Belmonte-de-Abreu2009). As in the Gil study, the strongest associations were with reported abuse other than sexual abuse. This highlights the potentially pervasive negative impact of all forms of childhood adversity in this sub-population and the likely contribution of adverse childhood experiences to psychosocial disability and complex needs among people with psychosis (Harvey et al., Reference Harvey, Brophy, Parsons, Moeller-Saxone, Grigg and Siskind2016).
Subsequent adversities
Our findings demonstrate that the association between childhood adversity and subsequent adversity previously recognised within the general population (e.g. Briere and Elliott, Reference Briere and Elliott2003; Casey and Nurius, Reference Casey and Nurius2005) is also evident in people with psychosis. For all participants, exposure to a definite psychosocial stressor within the 12 months preceding illness onset was high (63%) as was recent victimisation (39%). Those participants reporting any form of childhood adversity (and those who had experienced sexual abuse or other forms of abuse) had a higher rate of reporting a definite psychosocial stressor within 12 months of illness onset and also recent victimisation. These findings are consistent with the possibility that perceived social threat is a key risk factor for psychosis (van Nierop et al., Reference van Nierop, Lataster, Smeets, Gunther, van Zelst, de Graaf, ten Have, van Dorsselaer, Bak, Myin-Germeys and Viechtbauer2014). This also supports cognitive models of psychosis that suggest particular psychotic symptoms may be best conceptualised as dimensional such that expression of these symptoms is linked to exposure to cumulative adverse events beginning in childhood. This may also be associated with increased stress reactivity later in life (Perry et al., Reference Perry, Pollard, Blakey, Baker and Vigilante1995; Garety et al., Reference Garety, Bebbington, Fowler, Freeman and Kuipers2007; Bolton, Reference Bolton, Morgan and Bhugra2010). It is plausible that subsequent experiences of adverse events could be an important contributor in elevating cortisol levels which in turn may affect the severity of depressive and anxiety symptoms and increase the risk of developing psychosis (Garety et al., Reference Garety, Bebbington, Fowler, Freeman and Kuipers2007). This theory is also supported by our finding that depression and anxiety are associated with exposure to any type of childhood adversity.
Positive psychotic symptoms
We found associations between any childhood adversity and positive psychotic symptoms of hallucinations and subjective thought disorder but not delusions. Hallucinations were strongly associated with childhood sexual abuse and we found a relationship between subjective thought disorder and all forms of childhood abuse, as reported elsewhere (Schenkel et al., Reference Schenkel, Spaulding, DiLillo and Silverstein2005; McCabe et al., Reference McCabe, Maloney, Stain, Loughland and Carr2012; Shah et al., Reference Shah, Mackinnon, Galletly, Carr, McGrath, Stain, Castle, Harvey, Sweeney and Morgan2014). While some studies have found a relationship between child abuse and delusions (Scott et al., Reference Scott, Chant, Andrews, Martin and McGrath2007; van Nierop et al., Reference van Nierop, Lataster, Smeets, Gunther, van Zelst, de Graaf, ten Have, van Dorsselaer, Bak, Myin-Germeys and Viechtbauer2014), most find a stronger association between child abuse and hallucinations. Such studies also show a dose–response relationship between the number and types of traumatic events and delusional experiences, such that a combination of child abuse and adult abuse may be a strong predictor of delusional experiences (Read et al., Reference Read, Agar, Argyle and Aderhold2003; Scott et al., Reference Scott, Chant, Andrews, Martin and McGrath2007). In line with other studies, we found an association between any form of childhood adversity and alcohol and other substance dependence. There were no associations between abuse and interpersonal loss and lifetime diagnoses of either affective psychosis or non-affective psychosis, consistent with a number of other reports of a lack of diagnostic specificity (Matheson et al., Reference Matheson, Shepherd, Pinchbeck, Laurens and Carr2013; Shah et al., Reference Shah, Mackinnon, Galletly, Carr, McGrath, Stain, Castle, Harvey, Sweeney and Morgan2014; van Nierop et al., Reference van Nierop, Viechtbauer, Gunther, van Zelst, de Graaf, ten Have, van Dorsselaer, Bak and van Winkel2015). In analyses restricted to people with affective psychosis, we found a different pattern of associations to that observed for people with non-affective psychosis. However, the sample of people with affective psychosis was small making interpretation difficult. Future research could explore this issue further. Our results suggest that there may be a link between event type and the increased likelihood of particular symptomatic responses. Previous studies have also found a dose–response relationship between the severity of sexual abuse in childhood and risk of hallucinations in adulthood (Shevlin et al., Reference Shevlin, Dorahy and Adamson2007; Bentall et al., Reference Bentall, Wickham, Shevlin and Varese2012), however this was not within the scope of our study.
Negative schematic belief systems
Those who had experienced sexual abuse had a higher likelihood of depression, more than twice that of those who had not experienced abuse. Further and consistent with the literature, self-harm and suicidal thoughts were associated with childhood abuse (Schenkel et al., Reference Schenkel, Spaulding, DiLillo and Silverstein2005; Rosenberg et al., Reference Rosenberg, Lu, Mueser, Jankowski and Cournos2007; Read et al., Reference Read, Fink, Tudegeir, Felitti and Whitfield2008; Conus et al., Reference Conus, Cotton, Schimmelmann, McGorry and Lambert2009; Hooven et al., Reference Hooven, Nurius, Logan-Greene and Thompson2012; Shah et al., Reference Shah, Mackinnon, Galletly, Carr, McGrath, Stain, Castle, Harvey, Sweeney and Morgan2014; van Nierop et al., Reference van Nierop, Lataster, Smeets, Gunther, van Zelst, de Graaf, ten Have, van Dorsselaer, Bak, Myin-Germeys and Viechtbauer2014). We also found associations between abuse experiences and self-reproach which add to existing theories regarding emotional sequelae of abuse, including negative schematic beliefs and self-appraisals, and their proposed mediating role in facilitating or maintaining psychosis (Garety et al., Reference Garety, Bebbington, Fowler, Freeman and Kuipers2007; Gracie et al., Reference Gracie, Freeman, Green, Garety, Kuipers, Hardy, Ray, Dunn, Bebbington and Fowler2007). Likewise, these findings suggest avenues for study of conceivable causal pathways and tailored psychological interventions in this population (Mueser et al., Reference Mueser, Rosenberg, Xie, Jankowski, Bolton and Lu2008).
Limitations
The cross-sectional design is a limitation of this study as is the fact that all the individuals who entered our study had a diagnosis of psychosis; we were therefore unable to compare those who had experienced childhood trauma with people who had similar experiences in the general population. Recall bias associated with retrospective reporting of adversity is possible, although retrospective methods tend to bias towards under-reporting rather than over-reporting (Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012). Although carefully operationalised according to recognised definitions (Shah et al., Reference Shah, Mackinnon, Galletly, Carr, McGrath, Stain, Castle, Harvey, Sweeney and Morgan2014), our abuse types were based on a single question. The presence of these was determined from a single open-ended question about other distressing or traumatic events in childhood, with detailed notes taken and responses coded into categories. However, this may have further contributed to underreporting. Further, we recognise the limitations of using dichotomous variables meaning some of the detail of the experience is lost. Finally, our coding scheme meant that we did not distinguish between the occurrence of physical and emotional abuse and neglect (all were coded as other abuse). Understanding how specific types of abuse and neglect in childhood may contribute to the development of mental and psychosocial impairment in adulthood would be a useful focus of future research.
Conclusions
In our sample of people living with psychosis, adverse events in childhood, especially interpersonal loss, were common and suggest a need for appropriate and targeted prevention. Sexual abuse and other types of abuse were less common but may lead to elevated risk of poorer mental health and poorer psychosocial outcomes in adulthood relative to other people with psychosis. An implication of these findings is that mental health professionals should be trained to make routine and sensitive enquiries about adverse childhood events since these may have prognostic implications for mental health outcomes. Further, mental health professionals and researchers should develop and apply the evidence base for trauma-focussed interventions. Longitudinal research designs are warranted to provide stronger evidence about the risk of developing psychosis following adverse events in childhood and the potential role of subsequent psychosocial stressors. These designs should also gather more details about the exposure (e.g. age of occurrence, frequency of adverse event types and victim's relationship to perpetrator), whether disclosure occurred and what response occurred to any disclosure, in order to further improve our understanding of relationships between childhood adversity and psychosis.
Data
The 2010 Survey of High Impact Psychosis is available for approved researchers to access. Contact Professor Vera Morgan, Head: Neuropsychiatric Epidemiology Research Unit, School of Population and Global Health, University of Western Australia, email: vera.morgan@uwa.edu.au.
Acknowledgements
This publication is based on data collected in the framework of the 2010 Australian National Survey of High Impact Psychosis. In 2010, the members of the Survey of High Impact Psychosis Study Group were: V. Morgan (National Project Director), A. Jablensky (Chief Scientific Advisor), A. Waterreus (National Project Coordinator), R. Bush, V. Carr, D. Castle, M. Cohen, C. Galletly, C. Harvey, B. Hocking, A. Mackinnon, P. McGorry, J. McGrath, A. Neil, S. Saw and H. Stain. The Australian National Survey of High Impact Psychosis was funded by the Australian Government Department of Health and Ageing. This report acknowledges, with thanks, the hundreds of mental health professionals who participated in the preparation and conduct of the survey and the many Australians with psychotic disorders who gave their time and whose responses form the basis of this publication.
Author contributions
All authors contributed to the study design. Suzanne Turner and Carol Harvey conducted the review of the literature. Matthew Spittal and Suzanne Turner undertook the statistical analysis. Suzanne Turner wrote the initial draft of the manuscript in collaboration with Carol Harvey and Matthew Spittal. All authors commented on the first manuscript and contributed to writing subsequent versions. All authors approved the final manuscript.
Financial support
Matthew Spittal is a recipient of an Australian Research Council Future Fellowship (project number FT180100075) funded by the Australian Government.
Conflict of interest
None.
Ethical standards
Ethics approvals for the study were obtained from relevant institutional human research ethics committees.
Open access
