Childhood trauma and substance use underlying psychosis: a systematic review.

Background: Schizophrenia spectrum disorders (SSD) are mental diseases caused by a combination of genetic susceptibility and a number of environmental factors. Among these factors, the role of traumatic events suffered in childhood, as well as that of substance use, have been of particular research interest.
Objectives: To conduct a systematic review to clarify whether there is an interaction between childhood trauma and substance use related to the diagnosis or symptoms of SSD. It was also the objective of this review to collate the associations that may exist between the three variables of the study (trauma, substance use and psychosis).
Methods: We conducted a systematic search resulting in 240 articles. We considered all of the original articles that explored childhood trauma and substance use in patients suffering from SSD.
Results: Twenty-three articles were selected for this review. Several of the reviewed papers found associations between childhood trauma and substance use with SSD, as well as interactions between trauma and drug use on SSD. Conclusions: The results suggest that childhood trauma and substance use may be present at the basis of psychosis. This double hit on the pathogenesis could have clinical implications, since each of these impacts could be considered a window of opportunity for the primary prevention of SSD.

Introduction

Schizophrenia spectrum disorders (SSD) are mental diseases caused by a combination of environmental factors and genetic susceptibility (Bernardo et al., 2017; Pelayo-Teran, Suarez-Pinilla, Chadi, & Crespo-Facorro, 2012; van Os, Kenis, & Rutten, 2010). However, the fact that a considerable percentage of identical twins are discordant regarding disease expression confirms a complex pattern of heritability for SSD. Studies suggest a genetic-environment interaction in which some genetic factors influence the onset of psychosis but different environmental experiences may moderate this susceptibility (Cannon, Kaprio, Lonnqvist, Huttunen, & Koskenvuo, 1998; Gottesman, 1994; Kendler, 1983; Kringlen, 2000; Mrazek, 1994). For these reasons, non-genetic factors should be widely considered in the pathogenesis of SSD (Torrey, Bartko, & Yolken, 2012; van Os, Krabbendam, Myin-Germeys, & Delespaul, 2005).

The role of suffering from early trauma as socio-environmental risk for developing SSD has reached a particular research interest. Trauma models of mental disorders have emphasized the effects of stress, particularly during the early stages of life, as a key factor in their subsequent development (Cassidy, 2008). A prior history of childhood trauma is associated with increased odds of experiencing perceptual abnormalities in ultra-high-risk (UHR) subjects (young people who are identified as being at high risk of developing a psychotic illness in the near future) (O’ Connor, Nelson, Cannon, Yung, & Thompson, 2019) and with the onset of SSD (Matheson, Shepherd, Pinchbeck, Laurens, & Carr, 2013; Morgan & Fisher, 2007; Trotta, Murray, & Fisher, 2015).

Moreover, the abuse of substances with psychoactive properties, such as alcohol, amphetamines, cannabis, cocaine, and hallucinogens, has been associated with a greater risk for psychotic experiences in the general population (Kelleher & Cannon, 2011). Epidemiological and biological studies suggest that cannabis is a component that interacts with other factors in the development of schizophrenia or other psychotic disorders (Castle, 2013). For instance, Caspi and colleagues have suggested the role of the catechol-O-methyl-transferase (COMT) gene in the development of SSD when exposed to cannabis (Caspi et al., 2005); Arseneault and colleagues, in a review of studies with well-defined samples, have shown that cannabis use confers an overall twofold increase in the relative risk for later schizophrenia on an individual level, whereas on a population level, the elimination of cannabis use would reduce the incidence of schizophrenia by approximately 8%, assuming a causal relationship (Arseneault, Cannon, Witton, & Murray, 2004). Furthermore, other studies show that cannabis consumption in adolescence may play a role in neurodevelopment processes relevant to SSD, and it may produce changes in the endocannabinoid system (Murray et al., 2017) and dopamine sensitization (D’Souza, Sewell, & Ranganathan, 2009; Volkow, 2016). In any event, the psychosis risk-increasing effects are not explained by only a shared genetic predisposition between SSD and cannabis use (Murray et al., 2017). It is noteworthy that evidence for the causative role of other substances is less systematic, and existing studies were based on only correlations between SSD and substance use (Thirthalli & Benegal, 2006), showing that exposure to several neurotoxic substances during specific sensitive time windows may affect the stages of brain development (Heyer & Meredith, 2017).

There is growing evidence that some of the risk factors for SSD may have a multiplicative effect on diagnoses beyond the associated risk to either factor alone (Pelayo-Teran et al., 2012; Zammit, Lewis, Dalman, & Allebeck, 2010). The ‘double hit’ theory of the pathogenesis of SSD proposes that the illness may involve an early exposure to any adverse factor that produces a latent vulnerability, and when this vulnerability is manifested, individuals become more susceptible to other stressful events that contribute to the development of SSD (Feinberg, 1982; Maynard, Sikich, Lieberman, & LaMantia, 2001). Hence, some authors have hypothesized that childhood trauma may advance existing gene-environment conceptualizations of the cannabis-psychosis link (Houston, Murphy, Shevlin, & Adamson, 2011), and childhood adversity has been postulated as a key factor to include in analyses investigating associations between cannabis and SSD (Murphy, Houston, Shevlin, & Adamson, 2013).

To date, several reviews have been devoted to the study of risk factors for SSD. Some deal with the relationship between childhood trauma and psychosis (Blacker, Frye, Morava, Kozicz, & Veldic, 2019; Trotta et al., 2015), while others deal with substance use and psychosis (Martinotti et al., 2012; Parakh & Basu, 2013). Recently, the review by Stilo & Murray was aimed at the study of different non-genetic risk factors for schizophrenia, showing that there is a cumulative effect among environmental risk factors (Stilo & Murray, 2019). However, according to our knowledge, none of the reviews carried out so far thoroughly examines the influence of the relationship between childhood trauma and substance use on schizophrenia spectrum disorders. Therefore, our objective was to conduct a systematic review of the literature to clarify whether there is an interaction between childhood trauma and substance use related to the diagnosis or symptoms of SSD. It was also the objective of this review to collate the associations that may exist between the three variables of the study: trauma, substance use and psychosis.

Material and methods

Comprehensive search of the literature

We followed Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines for the reporting of this systematic review (Moher, Liberati, Tetzlaff, & Altman, 2010). Searches were conducted by two persons independently (one psychologist and one medical doctor) through an electronic search of scientific journals from the Medline database via PubMed and Embase. The following key terms were used for the search: ‘childhood trauma OR early trauma OR adverse childhood experience’ AND ‘substance use OR substance abuse’ AND ‘psychosis OR schizophrenia’. No restriction was used in the search. We found 231 articles, all of which were published between September 1980 and September 2019. In addition, the reference list of identified studies was also hand-searched to obtain additional articles, and nine articles that met the established inclusion criteria were included.

Eligibility and study selection

Records were selected if they met the following inclusion criteria: i) were reported articles written in English; ii) explored the relationship between childhood trauma, substance use and schizophrenia spectrum disorders (SSD) (SSD: schizophrenia, schizophreniform disorder, brief psychotic disorder, not otherwise specified psychosis, schizoaffective disorder), not being necessary that they had studied all possible relationships or interactions between the three variables; iii) reported their results quantitatively or qualitatively; and iv) were peer reviewed articles.

Records were excluded if they i) were conducted in animals; ii) were case-series studies, case studies, case reports or experimental or quasi-experimental studies; iii) were reviews or meta-analyses; or iv) were books, comments, conference papers, editorials, letters, theses or related publications. No time restriction was imposed in either the computerized or the manual searches.

The full text of each potentially eligible article was read by two researchers before a final decision was reached about its inclusion in the present review. Articles that contained two samples were included in the review when at least one of them fulfilled the inclusion criteria, taking into account only the sample of interest. We did not consider studies with overlapping samples, which means that a sample of subjects was included only once in this review despite appearing in more than one study.

Data extraction and analysis

Relevant data from the selected studies were extracted into a pre-defined, structured extraction table. Data extraction was performed by a single researcher for each included study, and a second individual independently verified the extracted information. The variables for each article included in the review were the year of publication, number of participants, type of childhood trauma, type of substance consumed, diagnosis, psychometric instruments employed to assess childhood trauma, type of substance use, type of psychotic disorder, and symptoms. The results of the included articles were synthesized into four sections: the first was about the association between childhood trauma and substance use; the second was about the association between childhood trauma and SSD; the third was about the association between substance use and SSD; and the last was about the interaction between childhood trauma, substance use and SSD.

Quality assessment

To assess the risk of bias in the different studies, we used the Newcastle-Ottawa Scale (NOS) (Wells et al., 2014). The NOS is an eight-item tool designed to rate methodological aspects of case-control and cohort studies, which includes three domains: study selection, the comparability of cohorts on the basis of the design or analysis, and exposure. The overall score ranges from 0 to 9. Scores <5 were considered low-quality studies (i.e., high risk of bias), scores 5–6 as moderate-quality studies and scores >7 indicated high-quality studies (Stang, 2010).

We used an adapted version of the NOS to evaluate the risk of bias of the cross-sectional studies (Modesti et al., 2016). The three domains included in this scale were study selection, comparability and outcome. The overall score ranges from 0 to 10. Scores <5 were considered low-quality studies, scores 5–6 were considered moderate-quality studies, scores 7–8 were considered good-quality studies, and scores 9–10 were considered high-quality studies (Herzog et al., 2013).

Results

Out of the 240 found articles, 164 were selected based on the search criteria after duplicates were removed. Of them, 130 articles were chosen as potentially relevant after excluding articles for different reasons (1 book chapter, 7 articles in different languages, 24 reviews, and 2 letters to editor). Among them, 107 studies were excluded because they did not explore a direct relationship between childhood trauma and substance use, between trauma in childhood and SSD, between substance use and SSD, or among three variables (childhood trauma, substance use and SSD). The final number of studies included in the review was 23 (see Figure 1). Data from all included studies are summarized in Table 1. Among these articles, we found two studies that included several samples in their results (Konings et al., 2012; van Nierop et al., 2016). In one of them (Konings et al., 2012), only one sample was taken into account since the other was included in other study of this review.

Figure 1.

Flow diagram selection of the study process.

Table 1.

Characteristics of the included studies involving patients with SSD.

Authors	Sample	Trauma Assessment	Type of Trauma	Drug Assessment	Type of Drug	Diagnostic Tool	Diagnosis	Symptoms Assessment	Design of studies	Relationship betweenvariables
Arranz et al., 2018	207	CTQ-SF	Emotional, physical and sexual abuse; emotional and physical neglect	Interview designed by authors	Cannabis	DSM-IV	Psychotic disorder	US	Case/Control	+B*+C*+D*
Morkved et al., 2018	57	CTQ-SF	Emotional, physical and sexual abuse; emotional and physical neglect	DUDIT and KKS	Any drug	ICD-10 and SCL-90-R	Psychotic disorder	PANSS/SCL-90-R	Case/Control	-A+B^
Haahr et al., 2018	191	BBTS	Emotional, physical and sexual abuse	US	Drugs and alcohol	DSM-IV	SSD	PANSS	Cross-sectional	+A
Ajnakina et al., 2018	237	CAECA-Q	Severe sexual abuse, severe physical abuse, separation, and/or loss	AUDIT/CEQ	Cannabis and alcohol	ICD-10	FEP	US	Longitudinal	+A+B^
Tomassi et al., 2017	345	CAECA-Q	Severe sexual abuse, severe physical abuse, separation, and/or loss	Cannabis Experiences Questionnaire	Cannabis, cocaine and heroin	ICD-10	Affective andnon-affective psychosis	US	Cross-sectional	+A+B*
van Nierop et al., 2016	532	CTQ-SF	Emotional, physical and sexual abuse; emotional and physical neglect	CIDI	Any drug	DSM-IV	Psychotic disorder	PANSS	Cross-sectional	+A+B^
Duhig et al., 2015	100	CTQ	Emotional, physical and sexual abuse; emotional and physical neglect	Current and lifetime substance use questionnaire	Alcohol, tobacco, cannabis, and other illicit drugs	ICD-10	FEP	PANSS	Cross-sectional	-A
Barrigon et al., 2015	120	US	Emotional neglect, psychological, physical or sexual abuse	CIDI	Cannabis	SCID-I	Psychotic disorder	PANSS	Case/Control	+B*+C*-D*
Schalinski et al., 2015	62	MACE	Emotional, physical and sexual abuse; emotional and physical neglect	ICD-10	Any drug	ICD-10	SSD	PANSS	Longitudinal	+A+B^
Saddichha et al., 2015	409	CTQ	Emotional, physical and sexual abuse; emotional and physical neglect	MINI Plus	Cannabis/Stimulant	MINI Plus	Psychotic disorder	US	Cross-sectional	+A+C*
Banducci et al., 2014	280	CTQ	Emotional, physical and sexual abuse; emotional and physical neglect	DSM-IV-TR	Alcohol, cannabis, cocaine and opiates	DSM-IV-TR	Psychiatric disorders including psychotic disorder	SCID	Cross-sectional	+A+B^
Ding et al., 2014	189	US	Emotional, physical and sexual abuse	ICD-10	Methamphetamine/other drugs	MINI-Plus	Meth-associated psychosis	SCID	Cross-sectional	+A+B*+D*
Konings et al., 2012	1636	A question from the parental questionnaire	Unspecified	US	Cannabis	CAPE	Psychotic disorder	CAPE	Longitudinal	-A+D*
Ramsay et al., 2011	61	CTQ-SF	Emotional, physical and sexual abuse; emotional and physical neglect	LSUR	Cannabis/alcohol	DSM-IV	FEP	SAPS/SANS	Cross-sectional	+A+B^+C^
Houston et al., 2011	7394	Questionnaire “Domestic Violence and Abuse”	Sexual trauma	SCAN	Cannabis	SCAN	General population(Psychosis)	SCAN	Case/Control	+B*+D*
Cutajar et al., 2010	5365	Child sexual abuse	Child sexual abuse	ICD	Substance abuse	ICD	Axis I (Psychosis)	ICD	Case/Control	+A
Harley et al., 2010	211	K-SADS	Physical and sexual abuse; Domestic violence	K-SADS	Cannabis	K-SADS	General population(Psychosis)	K-SADS	Case/Control	+A+B*+D*
Conus et al., 2010	658	EPFQ	Emotional, physical and sexual abuse	US	Substance use	DSM-IV	FEP	US	Longitudinal	+A
Houston et al., 2008	5877	CIDI	Rape and sexual assault	CIDI	Cannabis	DSM-III-R	General population(Psychosis)	CIDI	Case/Control	-B*-C*+D*
Rosenberg et al., 2007	569	SAEQ/CTS	Emotional, physical and sexual abuse	DALI/SCID	Drugs and alcohol	DSM-IV	SSD	BPRS	Cross-sectional	+A
Compton et al., 2004	18	CTQ	Emotional, physical and sexual abuse; emotional and physical neglect	DSM-IV	Cannabis	DSM-IV	Psychotic disorders	US	Cross-sectional	+A
Scheller-Gilkey et al., 2004	122	CTES	Death of a family member, sexual or physical abuse, exposure to violence and illness or injury	ASI	Drugs and alcohol	US	Schizophrenia	PANSS	Cross-sectional	+A
Neria et al., 2002	426	DSM-III-R (PTSD module)	Any harm or threat to physical integrity	DSM-III-R	Substance abuse	DSM-IV	Psychotic disorders	US	Cross-sectional	+A

Table 1. +: There is relationship; -: There is not relationship; *: Relationship with Diagnosis; ^: Relationship with symptoms or psychotic experiences; A: Relationship between childhood trauma and substance use; B: Relationship between childhood trauma and psychosis; C: Relationship between substance use and psychosis; D: Interaction between childhood trauma, substance use and psychosis; ASI: Addiction Severity Index; AUDIT: Alcohol Use Disorders Identification Test; BBTS: Brief Betrayal-Trauma Survey; CAPE: The Community Assessment of Psychic Experiences; CAECA-Q: Childhood experience of care and abuse-questionnaire; CEQ: Cannabis Experience Questionnaire; CIDI: Composite International Diagnostic Interview; CTES: Childhood Traumatic Events Scale; CTQ: Childhood Trauma Questionnaire; CTQ-SF: Childhood Trauma Questionnaire-Short Form; CTS: Conflict Tacties Scales; DALI: Dartmounth Assessment of Lifestyle Instrument; DSM-III-R: Diagnostic and Statistical Manual of Mental Disorders, 3rd ed. Rev; DUDIT: The Drug Use Disorder Identification Test; EPFQ: Early Psychosis File Questionnaire; FEP: First Episode of Psychosis; KKS: Norwegian national client mapping system; K-SADS: Schedule for Affective Disorders and Schizophrenia for School-Age Children; LSUR: Lifetime Substance Use Recall; MACE: Maltreatment and Abuse Chronology of Exposure; SAEQ: Sexual Abuse Exposure Questionnaire; SCAN: Schedule for Assessment in Neuropsychiatry; SCID-I: Structured Clinical Interview for DSM; SCL-90-R: Symptoms Check List -90- Revised; SSD: Schizophrenia Spectrum Disorders; US: Unspecified Scale.

The methodological quality assessment of each included study was performed according to the NOS tool (see Supplementary Tables 1 and 2). Methodological flaws were found for the selection assessment (5/23), comparability (7/23) and exposure in the case of cohort and case-control studies (1/11) or for the outcome in the case of cross-sectional studies (5/12). Overall, scores ranged from 4 to 9 in the cohort studies and between 5 and 10 in the cross-sectional studies, suggesting moderate- to high-quality studies. One of the studies were classified as having a high risk of bias, ten as moderate risk, and twelve studies as low risk.

Relationship between childhood trauma and substance use

Of the 23 articles selected for review, 19 (83%) directly analysed the association between the two environmental risk factors for psychosis: childhood trauma and substance use. In 16 of them, a positive association was found between both factors: specifically, in nine studies, childhood trauma was related to the use of different substances (Conus et al., 2010; Cutajar et al., 2010; Haahr et al., 2018; Neria, Bromet, Sievers, Lavelle, & Fochtmann, 2002; Rosenberg, Lu, Mueser, Jankowski, & Cournos, 2007; Schalinski, Fischer, & Rockstroh, 2015; Scheller-Gilkey, Moynes, Cooper, Kant, & Miller, 2004; Tomassi et al., 2017; van Nierop et al., 2016); in five studies, childhood trauma was related to cannabis use (Ajnakina et al., 2018; Compton, Furman, & Kaslow, 2004; Harley et al., 2010; Ramsay, Flanagan, Gantt, Broussard, & Compton, 2011; Saddichha, Werker, Schuetz, & Krausz, 2015); one study based its analyses on the relationship between trauma and cocaine use (Banducci, Hoffman, Lejuez, & Koenen, 2014); and another investigated the relationship between trauma and methamphetamine use (Ding, Lin, Zhou, Yan, & He, 2014).

The remaining three studies found no significant relationship. One of them based their analyses on the relationship between childhood trauma and cannabis use (Konings et al., 2012), while the other two studied the relationship between trauma and the use of any substance (Duhig et al., 2015; Morkved et al., 2018).

Relationship between childhood trauma and SSD

Of the total included studies in this review, 13 (57%) discussed the relationship between childhood trauma and SSD. All studies except two (Ajnakina et al., 2018; Schalinski et al., 2015) conducted their analyses cross-sectionally. Eight of these studies showed good methodological quality; in four, the quality was moderate; and one of them was low quality.

There were five studies that showed a positive association between exposure to childhood trauma and the severity of psychotic symptoms (Ajnakina et al., 2018; Banducci et al., 2014; Ramsay et al., 2011; Schalinski et al., 2015; van Nierop et al., 2016). In one of these studies, major methodological flaws were found for the selection and comparability assessment (Schalinski et al., 2015). Three other studies related sexual abuse in childhood with a higher probability of presenting a diagnosis of SSD (Ding et al., 2014; Houston et al., 2011; Tomassi et al., 2017). Barrigon et al. (2015) and Harley et al. (2010) found that the chances of suffering psychosis were 7.3 and 5.2 times higher, respectively, for those who had suffered childhood trauma than for those who had not experienced it, and both works had moderate methodological quality. Additionally, with moderate quality, the Morkved study examined two different patient samples: a sample of patients with SSD and a sample of patients with substance abuse. Although no significant association was found in the first sample, it is worth noting that in the substance abuse sample, there were statistically significant correlations between the psychoticism scale score and the Childhood Trauma Questionnaire (CTQ) score (Morkved et al., 2018). In the study by Arranz et al., using the CTQ scale, they noted that the FEP group was more likely than healthy controls to report higher total scores in the trauma measure (Arranz et al., 2018). In total, twelve studies showed a positive relationship between exposure to childhood trauma and SSD. In contrast to the results described, there was a study that found no relationship between having lived through a traumatic experience in childhood and presenting a psychotic disorder. This last work presented moderate quality in their methodology (Houston, Murphy, Adamson, Stringer, & Shevlin, 2008).

Relationship between substance use and SSD

Five studies (22% of the total included studies) showed data on the relationship between substance use and SSD, and in four of them, a positive relationship was found between the two. All the included studies in this section evaluated their variables cross-sectionally. Two of them presented good-quality methodology, and three presented moderate-quality methods.

The study carried out by Ramsay et al. (2011) found a positive relationship between cannabis and the severity of psychotic symptoms. The work of Saddichha et al. (2015) examined whether there were differences between those who used cannabis and those who used stimulants with regard to SSD, and they found that cannabis users had higher rates of lifetime SSD than stimulant users. Barrigón et al., found that the odds of developing psychosis were 16.5 times higher for subjects who used cannabis than for those who did not use cannabis (Barrigon et al., 2015). The last two described works that presented a moderate methodological quality. Furthermore, Arranz et al. (2018) detected that the relationship of cannabis use and psychosis depended on the amount consumed; they observed that a severe cannabis smoking pattern was associated with psychosis, while moderate use was not. Finally, contrary to these studies, Houston et al. (2008) found no significant effects of early cannabis use on SSD.

Interaction between childhood trauma, substance use and SSD

There were seven studies (30% of the total sample) in which analyses were performed to determine the interaction among the three factors: childhood trauma, substance use and SSD. Of these, six were cross-sectional, and one had a longitudinal design. All studies showed NOS scores that suggested moderate to high quality.

All but one study in this section, studied the effect of the interaction between childhood trauma and cannabis use on SSD. On the other hand, Ding et al. (2014) studied the effect of the interaction between childhood trauma and methamphetamine use on the risk of psychosis and found a significant interaction.

In the studies of Harley et al. (2010) and Konings et al. (2012) a significant interaction between childhood trauma and cannabis use was presented in their psychosis model. Two studies found that childhood sexual abuse trauma and cannabis use showed a significant interaction that contributed to the presence of SSD (Houston et al., 2008, 2011). Both studies introduced socio-demographic variables (sex, age, urbanity, education and employment status) as confounding variables, and the Houston et al. (2011) study also introduced alcohol consumption. In the study by Arranz et al. (2018) the highest levels of exposure to childhood trauma and cannabis use were associated with FEP. Furthermore, the combined effect of risk factors (socio-demographic variables, recent events, childhood trauma and misuse variables) yielded a significant association with psychosis, explaining 49% of its variation.

Finally, it is necessary to highlight that one of the seven studies did not find a significant interaction among the studied variables (Barrigon et al., 2015).

Discussion

The main aims of this review were to evaluate whether there is an interaction between childhood trauma, substance use and psychotic disorder diagnosis or psychotic symptoms in patients with SSD and to collate whether there is a relationship among the three studied variables.

The summarized results can lead us to infer that early trauma or maltreatment and substance use play a role in the diagnosis of SSD or in their symptomatic exacerbation. Overall, the quality of the included studies that address this interaction ranges from moderate to high. A significant two-way interaction was found in 86%, showing a positive relationship.

This plausible two-way interaction may be explained in part by the ‘two-hit’ model of the pathogenesis of SSD. This model proposes that a first factor during early life (‘first-hit’) disrupts central nervous system (CNS) development. This disruption produces a long-term vulnerability to a ‘second-hit’ during adolescence or adulthood, leading to the expression of SSD (Feinberg, 1982; Maynard et al., 2001). In the interaction among trauma, substance use, and psychosis, suffering from childhood trauma may be the first impact that could cause changes in the stress responses, modulating neurotransmitter or neuropeptide systems, or the development of a cognitive framework of susceptibility that may confer brain vulnerability to a second impact, such as substance use during adolescence or young adulthood, which could be a precipitating factor for SSD. Whereas an early risk factor may be related to neurodevelopmental disturbances, such as structural and neurochemical abnormalities in the brain (Bak et al., 2005; Ucok et al., 2007), subsequent environmental risk factors acting later may be considered not only as potential aetiological risk factors but also as precipitants for the illness in vulnerable individuals (Dean & Murray, 2005). In the same vein, Scheller-Gilkey and colleagues suggest that the mechanism of physiological adaptation to stress in response to trauma in early life may lead to long-term stress vulnerability, representing the common aetiological pathway in individuals with co-occurring disorders such as schizophrenia and substance abuse (Scheller-Gilkey et al., 2004). Following this trend, animal studies suggest that early and severe stress can create permanent behavioural and hormonal responses to stress in vulnerable animals (Anisman, Zaharia, Meaney, & Merali, 1998). Therefore, stress exposure at an early age might prompt brain dysfunctions that potentially cause a latent vulnerability to CNS disorders.

Secondarily, when only the association between early trauma exposure and substance use was explored among the included articles, 84% showed a positive association between early trauma and addictive behaviours. The review included four longitudinal studies, and in three of them, there was a significant relationship between childhood adversity and substance use. All included studies have a quality range from moderate to high, except for a longitudinal study, with poor quality in the comparability domain (Schalinski et al., 2015). These results provide support that both environmental factors are correlated between them, independent of the ulterior development or exacerbation of SSD. In this line, previous reviews have already concluded that there is a positive association between childhood or adulthood trauma and substance use (Kittirattanapaiboon, Srikosai, & Wittayanookulluk, 2017; Konkolÿ Thege et al., 2017; Lijffijt, Hu, & Swann, 2014; Simpson & Miller, 2002), but the timing of the association is still unclear as most of the traumatic experiences are retrospectively self-reported. Therefore, longitudinal well-designed studies about this area are highly recommended to establish robust conclusions.

This review also gives consistency to the already well-known relationship between substance use and the diagnosis of SSD. Overall, 80% of the included articles that address this association support a positive association between the two variables. All studies range from moderate to high quality, but none of the studies have a longitudinal design. However, as seen in previous research in this area (Henquet, Murray, Linszen, & van Os, 2005; Khokhar, Dwiel, Henricks, Doucette, & Green, 2018), cannabis use seems to be involved in the majority of associations. Longitudinal studies based on different types of drugs are encouraged to address this issue.

Finally, a positive association between early trauma and psychosis (the diagnosis of SSD or symptom exacerbation in patients with SSD) was found in 92% of the included studies that addressed this relationship. Two of the positive studies have a longitudinal design, and overall, the quality ranges from moderate to high, except in a longitudinal study (aforementioned previously), with a low score in the comparability domain. Therefore, our review confirms previous works that highlight the role of trauma and adverse childhood experiences in psychotic symptoms in patients with SSD (Schalinski et al., 2019) and in symptomatic exacerbation in patients with a first episode of schizophrenia (Misiak, Moustafa, Kiejna, & Frydecka, 2016). For future reviews and meta-analyses, subgroup analysis based on the same population type may partially overcome the heterogeneous results in this area.

This review has some limitations that are worth knowing for the interpretation of the current findings. Importantly, several limitations are inherent to each original study, mainly because different risk factors for SSD may be serious confounders (Mrazek, 1994); however, to partially address this limitation, the confounding variables controlled for in the included studies are referred to in the Results section. Furthermore, the existence of uncontrolled variables such as the age of the trauma, the type of trauma and substance, the length of the exposure, or the disease severity could not always be identified in the included articles; this may explain some of the apparent discrepancies across different studies. It is also worth noting that the most used drug among the analysed studies was cannabis; however, some studies also analysed the relationship between childhood trauma and the use of any type of drug. For future research, it seems necessary to design prospective, instead of retrospective, studies with larger sample sizes and follow-up studies of UHR subjects. These studies may give more information about the influence of early adverse events on the start and course of substance use disorders and on the diagnosis and course of SSD. Among the strengths of this work, it is interesting to note that, to the best of our knowledge, this is the first effort to systematically summarize the interaction among childhood trauma, substance use, and SSD.

In conclusion, knowing the relationship between childhood trauma and substance use with SSD and the interaction among the three factors may have an important implication for public health. Detecting cases of childhood trauma, as well as cases of trauma associated with substance use, could be useful for the primary prevention of some psychiatric diseases. Future studies designed in a prospective manner during childhood and adolescence, prior to the appearance of psychotic symptoms, could shed light on the mechanisms underlying the trauma and substance use relationship and clarify their role in brain development and the vulnerability to SSD while identifying potential points of intervention before psychosis onset.