Introduction
Psychotic experiences (PEs) are reported by many ‘healthy’ individuals who are not diagnosed with a psychotic disorder (Baumeister, Sedgwick, Howes, & Peters, Reference Baumeister, Sedgwick, Howes and Peters2017; McGrath et al., Reference McGrath, Saha, Al-Hamzawi, Alonso, Bromet, Bruffaerts and Kessler2015; van Os, Linscott, Myin-Germeys, Delespaul, & Krabbendam, Reference van Os, Linscott, Myin-Germeys, Delespaul and Krabbendam2009). The well-documented psychosis continuum posits that this is a result of an extended psychosis phenotype (Linscott & van Os, Reference Linscott and van Os2013; van Os et al., Reference van Os, Linscott, Myin-Germeys, Delespaul and Krabbendam2009). According to this hypothesis, subclinical PEs should be differentiated from clinically relevant expressions of psychosis by considering an individual's ‘need for care’.
It has been reported that non-clinical individuals (those with enduring PEs but no diagnosis or need for treatment), despite experiencing hallucinations and other first-rank symptoms (Brett, Peters, & McGuire, Reference Brett, Peters and McGuire2015; Peters et al., Reference Peters, Ward, Jackson, Morgan, Charalambides, McGuire and Garety2016), have fewer negative symptoms, cognitive difficulties and attentional anomalies compared to ultra-high-risk help-seeking and clinically diagnosed groups (Brett et al., Reference Brett, Peters and McGuire2015; Peters et al., Reference Peters, Ward, Jackson, Morgan, Charalambides, McGuire and Garety2016). van Nierop et al. (Reference van Nierop, van Os, Gunther, Myin-Germeys, de Graaf, ten Have and van Winkel2012) reported that, compared to a control group, individuals with self-reported PEs which were not confirmed by a clinical interview (‘false-positive’ group) had higher rates of mood, anxiety and substance use disorders, trauma and negative life events, and poorer physical, mental and social functioning. When compared to those with clinically confirmed PEs (‘true-positive’ group), however, these associations were generally smaller. Additionally, individuals who experience non-clinical PEs are significantly more likely than those without PEs to engage in help-seeking behaviour (e.g. see GP for emotional problems, attend counselling) (DeVylder, Oh, Corcoran, & Lukens, Reference DeVylder, Oh, Corcoran and Lukens2014; Murphy, Shevlin, Houston, & Adamson, Reference Murphy, Shevlin, Houston and Adamson2012). Furthermore, greater perceived control, less distress, more positive (e.g. spiritual) appraisals and normalising responses have generally been reported to distinguish non-clinical from diagnosed groups (Bak et al., Reference Bak, Myin-Germeys, Hanssen, Bijl, Vollebergh, Delespaul and van Osz2003; Baumeister et al., Reference Baumeister, Sedgwick, Howes and Peters2017; Brett et al., Reference Brett, Peters, Johns, Tabraham, Valmaggia and Mcguire2007; Brett, Heriot-Maitland, McGuire, & Peters, Reference Brett, Heriot-Maitland, McGuire and Peters2014; Johns et al., Reference Johns, Kompus, Connell, Humpston, Lincoln, Longden and Larøi2014; Powers, Kelley, & Corlett, Reference Powers, Kelley and Corlett2017).
As yet, little research has considered variation in functioning, symptom expression and aetiological risk at levels that encapsulate and correspond to a broader description and conceptualisation of the psychosis phenotype [i.e. variation from (i) no PEs to (ii) non-distressing PEs to (iii) distressing PEs to (iv) clinically defined psychotic disorder]. This is surprising given how informative evidence of variation at each of these levels could be; determining if functional impairment, symptomology and risk factors vary at each of these ‘levels’ may help us to understand (a) whether impaired functioning is associated with psychosis experience at all levels of the phenotype or whether it is something that is specifically reflective of distressing and clinically captured experiences only, (b) whether the defining characteristics of the phenotype (symptom expression) are stable or variable along the continuum, and if so, how and where, and (c) what factors are responsible for transitions along the continuum and where. In line with previous research on the psychosis continuum and need for care, it was broadly hypothesised that there would be similarities in areas of impaired functioning, symptom expression and risk factors along the continuum. However, it was expected that functioning would be poorer, PEs would be more likely to be endorsed and risk factors would be more strongly associated with psychosis expression moving from the lower to the upper end of the continuum.
Method
Sample
The National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) is a longitudinal survey designed to be representative of the civilian, non-institutionalised adult population of the USA, including the residents of the District of Columbia, Alaska and Hawaii (Grant & Kaplan, Reference Grant and Kaplan2005; Grant, Kaplan, Shepard, & Moore, Reference Grant, Kaplan, Shepard and Moore2003b). Respondents included those living in private households, military personnel living off base and people residing in non-institutionalised group housing. One adult was randomly selected from each dwelling (Grant & Dawson, Reference Grant and Dawson2006).
Wave 1 was conducted between 2001 and 2002. Face-to-face computer-assisted personal interviews (CAPI) were conducted by trained laypersons on 43 093 adults (81.0% response rate; Grant et al., Reference Grant, Kaplan, Shepard and Moore2003b). At Wave 2 (2004–2005), 34 653 eligible respondents were re-interviewed (86.7% response rate). The cumulative response rate for both waves combined was 70.2% (Grant & Kaplan, Reference Grant and Kaplan2005). At both waves, data were weighted, clustered on primary sampling units and stratified to be representative of the US general population on a range of sociodemographic variables, based on the 2000 Decennial Census (Grant & Dawson, Reference Grant and Dawson2006). Descriptions of the survey design and data collection processes are available in greater detail elsewhere (Grant et al., Reference Grant, Kaplan, Shepard and Moore2003b; Grant & Kaplan, Reference Grant and Kaplan2005).
Measures
The Alcohol Use Disorder and Associated Disabilities Interview Schedule – DSM-IV version (AUDADIS-IV; Grant & Dawson, Reference Grant and Dawson2000) is a fully structured, self-report, diagnostic CAPI designed to be administered by clinicians or trained laypersons (Grant & Dawson, Reference Grant and Dawson2000). It assesses both past year and lifetime occurrence of a variety of psychiatric disorders, including personality disorders (Grant et al., Reference Grant, Dawson, Stinson, Chou, Kay and Pickering2003a). The AUDADIS-IV measures of psychiatric disorders have been shown to demonstrate high reliability in general population samples (Grant et al., Reference Grant, Dawson, Stinson, Chou, Kay and Pickering2003a; Ruan et al., Reference Ruan, Goldstein, Chou, Smith, Saha, Pickering and Grant2008).
PEs
Sixteen PEs were drawn from the ‘unusual feelings and actions’ section of the AUDADIS-IV at Wave 2, each of which mapped onto one of three distinct schizotypal personality dimensions; ‘social/interpersonal’, ‘disorganisation’ and ‘cognitive/perceptual’ (see Table 2). Respondents were asked if they had ever experienced each PE (i.e. ‘Most of the time throughout your life, regardless of the situation or whom you were with…’, Yes/No response option). If yes, a follow-up item enquired if any distress or impaired functioning had been associated with that PE [i.e. ‘Did this (experience) ever trouble you or cause problems at work or school, or with your family or other people?’]. Additionally, respondents were asked at Wave 1 if they had ever been diagnosed with schizophrenia or a psychotic disorder/episode by a doctor or other health professional. At Wave 2, respondents were asked if they had been diagnosed since their Wave 1 interview. Information from both waves was used to categorise lifetime psychotic disorder/episode.
PE groups
Based on responses to the abovementioned PE items, individuals were categorised into four groups (unweighted N, weighted %; see Fig. 1):
a) Diagnosed with a psychotic disorder/episode [N = 1205 (3.1%); ‘Diagnosed’]: This group comprised individuals who reported a lifetime diagnosis of schizophrenia or another psychotic disorder or episode. Inclusion in this category was made regardless of respondents' endorsement of PE items.
b) PEs with distress or impairment [N = 3119 (8.6%); ‘PE-Impaired’]: This group comprised individuals who endorsed at least one of the 16 PE items and reported distress/impairment associated with at least one of these experience(s). However, these individuals did not report being diagnosed with a psychotic disorder/episode.
c) PEs without distress or impairment [N = 15 893 (45.4%); ‘PE-Experienced Only’]: This group comprised individuals who had endorsed at least one of the 16 PE items but did not report any distress/impairment associated with any of these experiences, nor had they reported being diagnosed with a psychotic disorder/episode.
d) No psychotic disorder diagnosis nor endorsement of PEs [N = 12 505 (37.6%); ‘No PEs’]: This group comprised individuals who neither reported a lifetime diagnosis of psychotic disorder/episode nor any of the 16 PEs.
Figure 1. Flow diagram detailing categorisation into PE groups: (a) Diagnosed, (b) PE-Impaired, (c) PE-Experienced Only and (d) No PEs.
Individuals who could not be categorised into one of these four groups due to missing data were removed from the analysis [N = 1931 (5.2%)].
Functioning
Impaired functioning was measured at Wave 2 using the Short-Form 12 Health Survey (SF-12v2; Ware, Kosinski, Turner-Bowker, & Gandek, Reference Ware, Kosinski, Turner-Bowker and Gandek2002), a 12-item measure of current functioning and life satisfaction over the past 4 weeks. The SF-12v2 produces norm-based scores across eight subscales: physical functioning, role physical functioning (i.e. how physical health interferes with regular activities), bodily pain, general health, vitality, social functioning, role emotional functioning (i.e. how emotional health interferes with regular activities) and mental health. Scores are standardised and range from 0 to 100 (M = 50, s.d. = 10). Higher scores reflect better functioning (Ware et al., Reference Ware, Kosinski, Turner-Bowker and Gandek2002). The SF-12v2 has been reported as a reliable and valid measure in both general population samples (Kim et al., Reference Kim, Jo, Ahn, Ock, Shin and Park2014; Montazeri et al., Reference Montazeri, Vahdaninia, Mousavi, Asadi-Lari, Omidvari and Tavousi2011) and in those with serious mental health or behavioural difficulties (Huo, Guo, Shenkman, & Muller, Reference Huo, Guo, Shenkman and Muller2018).
Predictor variables
Using the NESARC Wave 2 variables, an attempt was made to match the sociodemographic, environmental and psychological variables used by Peters et al. (Reference Peters, Ward, Jackson, Morgan, Charalambides, McGuire and Garety2016). The following were used as predictor variables in the analysis:
• Sex: Male (1), female (0).
• Age: Six age categories ranging from 24 years old or younger (0) through to being 65 years old or older (5).
• Children: Has no children (1) or at least one child (0).
• Ethnicity: Ethnic background was recoded into a dichotomous variable, which identified respondents as either White (0) or other ethnicity (1).
• Relationship status: Identified respondents as either being married or living with their partner (0) or not married/living with partner (1).
• Education: Has completed high school education (0) or not (1).
• Unemployment: Unemployed (1) or not unemployed (0; i.e. employed, student, retired, etc.).
• Parental mental health: This indicated whether, before age 18, the respondent had a parent/other adult living in the home who had been treated or hospitalised for mental illness (1) or not (0).
• Religious services: Current attendant of places of worship (1) or not (0).
• Importance of religion: Respondents were asked whether religious or spiritual beliefs were important in their daily lives. This was recoded into a binary variable whereby responses of ‘very important’ were coded as 1 and all other responses (somewhat important, not very important and not important at all) were coded as 0.
• Migrant: Identified respondents' parents' country of origin either as outside the USA (1) or at least one parent being from the USA (0).
• Cannabis use: Information from the ‘medicine use’ section of the questionnaire was used to identify if respondents had used cannabis since their last interview at Wave 1 (yes = 1, no = 0).
• Other drug use: Similarly, a separate variable was created to identify if respondents had used any other drug (sedatives, tranquilisers, opioids, amphetamines, cocaine, hallucinogens, inhalants, heroin, other) since their last interview (yes = 1, no = 0).
• Sexual assault: During the ‘traumatic events’ section of the questionnaire, respondents were asked if they were ever sexually assaulted, molested or raped or if they ever experience unwanted sexual activity (1) or not (0).
• Physical assault: A binary variable was created which indicated whether respondents had ever been physically attacked or badly beaten or injured by their parents, their partner or someone else (1) or not (0).
• Mugged: Respondents were asked if they had ever been mugged, held up or threatened with a weapon (1) or not (0).
• Parental neglect: This variable indicated whether respondents had been seriously neglected by either of their parents/guardians before the age of 18 (1) or not (0).
• Family bonding: Five items from the ‘background information’ section of the questionnaire enquired about family bonding and support. Responses were scored on a five-point Likert scale ranging from ‘never true’ to ‘very often true’. Responses were recoded into a binary variable whereby ‘never true’ or ‘rarely true’ indicated a lack of family support (1) and all other responses (sometimes true, often true or very often true) indicated family support (0).
• Major depressive (MD) episode: Respondents who met the diagnostic criteria for major depressive episode within the past 12 months (excluding substance-induced disorders or those due to a medical condition) were categorised yes (1) and no (0).
• Generalised anxiety disorder (GAD): Respondents who met the diagnostic criteria for GAD over the past 12 months (excluding substance-induced disorders or those due to a medical condition) were categorised yes (1) and no (0).
For all variables, ‘unknown’ responses were treated as missing data.
Analytic plan
Firstly, multivariate multiple regression analysis was conducted to assess whether degree of functional impairment differed across the PE groups. This analysis allows several dependent variables (the eight SF-12v2 subscales) to be jointly regressed on all predictor variables. The predictor variables in this model included dummy coded PE group variables (Diagnosed, PE-Impaired and PE-Experienced Only, with the No PE group as the reference category) and control variables (age, sex). The estimated regression coefficient for each variable indicates the mean difference between the specific PE category and the No PE group, and the associated confidence intervals (CIs) were used to indicate statistical significance. This approach was used as (1) all the model parameters are estimated simultaneously thereby avoiding the need for post-hoc adjustment for multiple testing, (2) the use of robust maximum likelihood (MLR) estimation allows for missing data to be handled efficiently by using all available data (Schafer & Graham, Reference Schafer and Graham2002), and third, MLR estimation is robust against deviations from normality and produces unbiased standard errors (West, Finch, & Curran, Reference West, Finch, Curran and Hoyle1995). Next, the proportion of respondents endorsing PEs, across the full sample and within the Diagnosed, PE-Impaired and PE-Experienced Only groups, was examined and χ2 tests of association with pairwise z-tests (Bonferroni adjusted) were conducted to compare proportions across all groups. Finally, three separate multinomial logistic regressions were conducted to assess whether sociodemographic, substance use, trauma, family bonding and diagnostic predictor variables could discriminate between PE group membership. In each analysis, a different reference category was used to ensure that all PE group comparisons were estimated. Analyses were carried out in SPSS v27 and Mplus 8.3 (Muthén & Muthén, Reference Muthén and Muthén2017) using survey design variables.
Results
Sample characteristics are presented in online Supplementary Table S1.
Functioning
Table 1 reports the unstandardised regression coefficients and CIs for the Diagnosed, PE-Impaired and PE-Experienced Only groups (with No PEs as the reference category) along with age and sex, predicting the total score on each of the SF-12v2 subscales. A graded trend was present across all eight SF-12v2 subscales, such that compared to the No PEs group, the Diagnosed group had the poorest functioning, followed by the PE-Impaired then PE-Experienced Only groups.
Table 1. Unstandardised (B) regression coefficients for SF-12v2 subscales (N = 32 714)a
a Weighted, clustered and stratified using survey design variables.
PE prevalence and symptom expression
Overall, 60% of the total sample reported having experienced at least one of the 16 PE items (Table 2). Social/interpersonal PEs were the most commonly endorsed, with almost half the total sample reporting at least one of these experiences. More than a quarter had experienced at least one cognitive/perceptual PE, while less than a fifth reported any disorganised PE. This trend was also present across PE groups. The χ2 tests of association were significant for all 16 PEs, indicating that there was an association between PE endorsement and PE group. Overall, the PE-Impaired group was more likely than the PE-Experienced Only group to endorse all 16 PEs and was also more likely than the Diagnosed group to endorse 15 of the 16 PEs. Moreover, the Diagnosed group was more likely than the PE-Experienced Only group to endorse 13 of the PEs.
Table 2. PE item endorsement across total sample and PE groups*
*Unweighted N, weighted using survey design variables %; χ2 based on weighted, clustered and stratified data; abcgroup comparisons based on weighted only data; differing subscript letters within a row denote item proportions differ significantly from each other at the 0.05 level (Bonferroni adjusted); data missing across PE items 0.6–1.3%; row and column Ns may not total 100% due to missing data.
† Despite the same proportions reported for the Diagnosed and PE-Impaired groups, analysis indicated a statistically significant difference between these groups. We believe that this unusual finding may be as a result of the process of weighting the data.
Regarding endorsement of PE items at the distress/impairment level (see online Supplementary Table S2), χ2 tests revealed that there was a significant association between PE group (i.e. PE-Impaired or Diagnosed) and 14 of the 16 PE items. The PE-Impaired group had a significantly higher proportion of distress endorsement than the Diagnosed group in all of these cases, with the exception of ‘Shadows’. Regarding item count, 77% of the Diagnosed group reported not being impaired by any of the PEs (this included missing responses). A significantly higher proportion of the PE-Impaired group reported being distressed by one (56.1% v. 7.6%), two (21.8% v. 4.9%), three (9.8% v. 2.3%), four (5.2% v. 2.7%) and five or more PEs (7.0% v. 5.9%).
Predicting PE group membership
Table 3 presents the results of the multinomial logistic regression analyses comparing predictor variables at different points along the psychosis continuum. In general, when compared to the No PEs reference group (columns 2–4), when risk was present, it was present across all the PE groups. Having any PE experience (PE-Experienced Only, PE-Impaired or Diagnosed) was associated with being male, non-White ethnicity, being unemployed, not being married/cohabiting with partner, having a parent/other adult living in the home being hospitalised due to their mental health before age 18, recent cannabis and other drug use, experiencing sexual assault, physical assault, being mugged and parental neglect, and past year MD episode and GAD. While these same variables generally tended to differentiate the Diagnosed and PE-Impaired groups from the PE-Experienced Only group (columns 5 and 6), the sociodemographic factors of non-White ethnicity, unemployment, not cohabiting with partner and parent mental health hospitalisation were specific to the Diagnosed group.
Table 3. Multinomial logistic regression analyses predicting PE group membership (N = 32 027)†
PY, past year; SLI, since last interview.
†Weighted, clustered and stratified using survey design variables; *p < 0.05, **p < 0.01, ***p < 0.001. Reference categories: aNo PEs, bPE-Experienced Only, cPE-Impaired. Significant ORs in bold.
Finally, at the upper end of the continuum (column 7), few variables differentiated the Diagnosed and PE-Impaired groups. The Diagnosed group were more likely to be female, older, unemployed, have had a parent hospitalised due to their mental health, and less likely to have experienced physical assault. Overall, ORs varied; in some cases, a clear graded pattern was present such that the highest ORs were associated with the Diagnosed group, reducing for the PE-Impaired group and again for the PE-Experienced Only group (e.g. parental neglect, cannabis use, unemployment). However, in several cases, the PE-Impaired group were at increased risk compared to the Diagnosed group (e.g. other drug use, physical assault, and MD and GAD diagnoses). An overall trend was also present for age in that PE-Impaired group were the youngest while the Diagnosed group were the oldest.
Discussion
The purpose of this study was to explore functional impairment, symptom expression and aetiological risk across the psychosis continuum.
Functioning
Two main findings are noted in relation to functioning. Firstly, as expected, the findings of the current study suggest that individuals who report no distress/impairment associated with their PEs still experience compromised functioning across all domains, compared to individuals who have not experienced PEs. Thus, despite not having a need for care, these individuals experience physical, emotional and social impairments. Age and sex were controlled for in this analysis, therefore, it is possible that these individuals' functional impairments are the result of other factors, such as comorbid emotional or behavioural issues rather than a direct result of their PEs (DeVylder et al., Reference DeVylder, Oh, Corcoran and Lukens2014; Murphy et al., Reference Murphy, Shevlin, Houston and Adamson2012), or that these deficits in functioning (e.g. poor physical health, reduced mobility, social isolation) create an adverse environment which may facilitate the development of PEs. Notably, however, the effect sizes observed for the PE-Experienced Only group were much smaller than those observed for the Diagnosed and PE-Impaired groups. Secondly, despite not necessarily experiencing more PEs (whether distressing or not) than the PE-Impaired group, Diagnosed individuals had significantly lower functioning, globally. These greater deficits in functioning may have acted as a catalyst to receiving a diagnosis (Addington et al., Reference Addington, Farris, Stowkowy, Santesteban-Echarri, Metzak and Kalathil2019; Riecher-Rössler & Studerus, Reference Riecher-Rössler and Studerus2017) or may, in part, be the result of antipsychotic medication side effects (Tandon et al., Reference Tandon, Lenderking, Weiss, Shalhoub, Barbosa, Chen and Castle2020) and the social stigma that accompanies a psychotic disorder diagnosis (Degnan, Berry, Humphrey, & Bucci, Reference Degnan, Berry, Humphrey and Bucci2021).
PE prevalence and symptom expression
Sixty per cent of the sample endorsed at least one PE; this is high compared to other general population studies (12–28%; Kendler, Gallagher, Abelson, & Kessler, Reference Kendler, Gallagher, Abelson and Kessler1996; Nuevo et al., Reference Nuevo, Chatterji, Verdes, Naidoo, Arango and Ayuso-Mateos2012; Pignon et al. Reference Pignon, Schürhoff, Szöke, Geoffroy, Jardri, Roelandt and Amad2018; van Os, Hanssen, Bijl, & Ravelli, Reference van Os, Hanssen, Bijl and Ravelli2000). However, this finding is not necessarily inconsistent with studies which have utilised a schizotypal personality measure; for example, a study of schizotypal personality traits in adolescents reported that 93.6% of the sample endorsed at least one of 22 items (Fonseca-Pedrero, Paíno-Piñeiro, Lemos-Giráldez, Villazón-García, & Muñiz, Reference Fonseca-Pedrero, Paíno-Piñeiro, Lemos-Giráldez, Villazón-García and Muñiz2009). Social/interpersonal experiences were the most likely to be endorsed across all groups, followed by cognitive/perceptual and disorganised PEs. This finding was not unexpected given that firstly, social/interpersonal difficulties are not unique to psychosis symptomology and also social isolation and exclusion have been heavily implicated in theories of psychosis aetiology (e.g. Hoffman, Reference Hoffman2007). Furthermore, non-social PEs that were most commonly reported in the absence of distress/impairment or diagnosis included ‘force’ and ‘strange ideas’; phenomena which may align with spiritual or faith-based experiences and thus may not be considered as distressing.
Despite similar PE expression across groups, the PE-Impaired group were more likely to endorse all PEs individually compared to the PE-Experienced Only group and most PEs compared to the Diagnosed group. They were also more likely to endorse greater numbers of PEs than the PE-Experienced Only group, and unexpectedly, a greater number of impairing and non-impairing PEs compared to the Diagnosed group. Furthermore, only three-quarters of individuals in the Diagnosed group reported ‘lifetime’ endorsement of PEs and only a quarter reported being distressed by one or more PEs. Given that these individuals had been clinically diagnosed with a psychotic disorder/episode, this finding has a number of implications. Firstly, the lower prevalence of most PEs among the Diagnosed group could indicate that, at its most severe, psychosis is characterised by a narrower, more specific phenotypic expression and that only particular PEs are associated with progression to the extreme end of the continuum. The main phenomenologically distinguishing feature associated with the upper end of the continuum (i.e. Diagnosed group) was both endorsement of, and distress associated with, the ‘Shadows’ item. It was the only item for which the Diagnosed group had a significantly greater proportion of experience (7.1% v. 5.5%) and distress (3.5% v. 2.0%) endorsement compared to the PE-Impaired group. Given that auditory and visual hallucinations are the hallmark symptoms for schizophrenia, experiencing or being distressed by this symptom may be more likely to result in a diagnosis than other symptoms.
A second reason for the apparently lower prevalence of PEs in the Diagnosed group might be the failure of the PE measure to capture negative symptoms. In previous research, measures of schizotypy have tended to correspond to measures of positive schizophrenic symptomology (Cochrane, Petch, & Pickering, Reference Cochrane, Petch and Pickering2010; Thomas et al., Reference Thomas, Rossell, Tan, Neill, Van Rheenen, Carruthers and Gurvich2018), but they may be less efficient at measuring negative psychosis symptomology (Cochrane et al., Reference Cochrane, Petch and Pickering2010). Thus, individuals in the Diagnosed group may be in a more advanced stage of their disorder, characterised by a greater number of negative symptoms and cognitive impairments than positive symptoms. The differences between the Diagnosed and PE-Impaired groups could also indicate a lack of insight into symptomology in the former, which is common among individuals diagnosed with schizophrenia (Lincoln, Lüllmann, & Rief, Reference Lincoln, Lüllmann and Rief2007), or the impact of treatment (especially antipsychotic medication) which would be expected to reduce the experience of PEs. Finally, the Diagnosed group were the oldest in the sample whereas the PE-Impaired group was the youngest (see online Supplementary Table S1). Typical age of onset of psychosis is late teens or early adulthood (Kessler et al., Reference Kessler, Amminger, Aguilar-Gaiola, Alonso, Lee and Ustun2007) but patients often have a long history of diagnosed psychosis (Perkins, Gu, Boteva, & Lieberman, Reference Perkins, Gu, Boteva and Lieberman2005) and the probability of being diagnosed and treated presumably increases with age. Hence, some in the PE-Impaired group may be individuals who, in the future, may enter the Diagnosed group.
Predictors of group membership
As expected, the predictor variables operated similarly across the continuum, with risk varying in a quantitative rather than a qualitative way. The trauma variables, diagnostic variables and other drug use variables, in particular, differentiated between all levels of the continuum, with the exception of the extreme end (PE-Impaired v. Diagnosed). Only a few variables distinguished the Diagnosed from the PE-Impaired group at the upper end of the continuum: female sex, older age, unemployment, parental mental health hospitalisation and lower likelihood of having experienced physical assault. The specific link with parental mental health hospitalisation during childhood could be suggestive of a genetic component to experiencing psychosis at a clinical level; however, there may also be a bias toward diagnosing individuals who have a known family history of psychosis. Moreover, most individuals are unemployed in the month prior to first-episode psychosis (Ramsay, Stewart, & Compton, Reference Ramsay, Stewart and Compton2012) and many continue to be unemployed after diagnosis (Marwaha & Johnson, Reference Marwaha and Johnson2004).
A lower likelihood of physical assault may seem inconsistent with the extant evidence base (e.g. Shevlin, Houston, Dorahy, & Adamson, Reference Shevlin, Houston, Dorahy and Adamson2008; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer and Bentall2012); however, the comparison here was between PE experienced with impairment and clinical psychosis. It is notable that those classified as PE with impairment were over one and half times more likely to experience physical assault compared to those classified as PE experience only. It may be the case that diagnosis and associated clinical intervention affords some level of protection from social adversities that have been commonly associated with psychosis in the general population.
While the strength of ORs generally indicated that the predictor variables were more strongly associated with the PE-Impaired over the PE-Experienced Only group, in contrast to the hypotheses, this trend could not clearly be extended to the Diagnosed group. For example, other drug use since last interview, past year MDD and GAD, and sexual and physical assault were more strongly associated with the PE-Impaired than Diagnosed group. As suggested above, this finding could reflect the treatment and support the Diagnosed group may have access to, improving their mood, reducing their recent substance use and offering some protection from adversity.
Limitations
There are several limitations which must be acknowledged. Firstly, the high proportion of PE items endorsed, with and without distress/impairment in the PE-Experienced Only and PE-Impaired group is, in part, a product of the method of categorisation, i.e. that individuals must have endorsed at least one PE to qualify for either group and must have been impaired/distressed by at least one PE to qualify for the PE-Impaired group. Secondly, diagnosed status was self-report and not based on a clinical measure. This group included individuals who may have experienced a single psychotic episode alongside those with a more extensive history of psychosis, therefore heterogeneity within groups is likely.
Thirdly, in the current study, PEs were derived from a schizotypal personality measure. While this measure was a trait-based assessment it still captured experiential accounts pertaining to both thoughts and perceptions. Moreover, use of a schizotypal personality scale as a proxy for experiential assessment is consistent with many other studies. For example, in a recent systematic review on definitions and assessments of psychotic-like experiences (PLEs), Lee et al. (Reference Lee, Chan, Chang, Lee, Hui and Chen2016) showed that a significant proportion of reviewed studies used schizotypal personality measures to investigate PLEs. Furthermore, studies have shown that measures of schizotypal personality provide non-clinical analogues of the heterogeneous symptomatology found in schizophrenia, although, as previously mentioned, they may be more aligned to positive rather than negative psychosis symptomology (Cochrane et al., Reference Cochrane, Petch and Pickering2010). Importantly, a distinction must be acknowledged between the assessment of schizotypal personality, which are usually considered as stable traits across time and other measures of PEs, which adopt a symptom or state approach (Pedrero & Debbané, Reference Pedrero and Debbané2017). Finally, predictor variables were dichotomised for the purpose of the analysis and as such, they tended to capture the presence, rather than severity/frequency of the events (e.g. trauma, drug use).
Conclusion
The current findings have advanced our understanding of the psychosis continuum in three important ways. First, they have shown that impaired functioning is associated with psychosis experience at all levels of the phenotype and that it is not specifically reflective of distressing and clinically captured experiences only. Second, they have shown that symptom expression is variable along the continuum, becoming most pronounced among those experiencing distressing PEs but without a psychosis diagnosis. Third, commonly recognised and evidenced risk factors for psychosis generally operate consistently along the continuum in a graded, incremental way, with a few notable exceptions that may be informative for clinical assessment (e.g. family history of psychosis). While these findings will need to be replicated, the proposed questions and analytic framework highlight the importance of continuum-based interpretations of the psychosis phenotype and afford valuable opportunities to consider how and in what way/context individuals transition from one position to the next.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291723001046
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Competing interests
None.
Ethical standards
Potential NESARC respondents were informed in writing of the nature of the study, the confidentiality procedures that were in place, the intended use for the data and the voluntary nature of their participation (Grant & Dawson, Reference Grant and Dawson2006). The research was approved by the US Census Bureau and the US Office of Management and Budget (Grant & Dawson, Reference Grant and Dawson2006).
Open access
