Correlates of substance use in a large naturalistic cohort of young people with early and emerging psychosis

Data were collected from headspace Early Psychosis (hEP) services delivered at 14 treatment centres across Australia. Treatment for young people diagnosed with UHR or FEP at these sites is delivered based around the early psychosis prevention and intervention centre (EPPIC) model, which comprises 16 core components of care [33]. This is a baseline cohort not yet exposed to treatment, and so many participants are not formally diagnosed nor prescribed medication. Duration of care in these hEP services is recommended as a minimum of 2 years for FEP patients and 6 months for UHR patients.

Data used in this analysis is from the minimum data set collected by hEP clinicians as part of routine baseline data collection at the time of presentation. The study design is outlined in more detail elsewhere [11]. In brief, inclusion criteria of this study include age between 12 and 25 years and admission to one of the hEP services between 19th of June 2017 and the 30th of September 2019 with either a FEP or identified as being at UHR status, defined according to the Comprehensive Assessment of At Risk Mental States (CAARMS) criteria [40]. Data from one service cluster were excluded due to collection issues. Data were collected for each distinct treatment period that participants spent within the service, also known as ‘episodes of care’; however only data from the first presentation were used in these analyses. Participants’ age, self-reported gender, cultural and linguistic diversity status, Aboriginal and Torres Strait Islander status, and engagement with education/employment/training were collected at intake into headspace services. In the current analysis, these data were included as potential confounders only. They have been presented previously [11].

Substance use was assessed using the WHO ASSIST [24]. Non-medical use of alcohol, tobacco, cannabis, amphetamine-type stimulants (ATS), cocaine, hallucinogens, opioids, inhalants, and other substances was self-reported by participants. In the ASSIST, ATS includes any amphetamine derivative stimulant including methamphetamine, amphetamine (and its isomers), and 3,4-methylenedioxymethamphetamine (MDMA). Hallucinogens include lysergic acid diethylamide (LSD), psilocybin and mushrooms containing it, and synthetic hallucinogens such as NBOMes. The questionnaire asks individuals about how often they have used substances in the previous 3 months with potential responses being ‘never’, ‘once or twice’, ‘monthly’, ‘weekly’, or ‘daily or almost daily’. For regression and network analyses, responses were dichotomised to indicate either presence or absence of substance use in the previous 3 months.

Psychiatric symptoms were rated by clinicians according to the extended Brief Psychiatric Rating Scale (BPRS-E) and separated into positive and negative symptom subscales [18]. This scale rates a treating clinician’s perception of the severity of 24 distinct psychiatric symptoms with the total score range from 24 to 168. Two subscales have been used to assess positive and negative symptoms of psychosis. The BPRS-psychosis subscale (with a score range from 4 to 28) comprises four items: suspiciousness, hallucinations, unusual thought content, and conceptual disorganisation to measure the positive symptoms of psychosis. The BPRS-negative subscale (range from 3 to 21) comprises three items: blunted affect, emotional withdrawal, and motor retardation and is used to assess the negative symptoms of psychosis.

The Kessler Psychological Distress Scale (K10, 10 item ranging from 10 to 50) is a self-report measure used to assess subjective psychological distress [4].

Personal and occupational functioning was assessed with the Social and Occupational Functioning Assessment Scale (SOFAS) [21] which is rated from 0 to 100. For example, no interpersonal conflict and good occupational function would be rated at 100 with impairment in functioning denoted by a reduction in score. A score of 50 indicates “serious impairment across social, occupational, or school functioning (e.g. no friends, unable to keep a job)”.

All young people included in this study provided informed consent for their data to be used for service evaluation. Ethics approval was granted by the University of Melbourne human research ethics committee (ref: 2021-20371-13617-3).

In total, data from 609 UHR and 643 FEP participants were included in these analyses. The mean age of young people with UHR status was 17.5 (SD 3.0) and FEP was 19.7 (SD 2.7) years. Cohort characteristics can be seen in Table S1 in Supplementary Information and have been described elsewhere [11]. At baseline, the two groups had similar scores on the SOFAS, but those with UHR status scored significantly higher on psychological distress, with a mean K10 of 31.3 (SD 8.7), vs a mean K10 of 24.5 (SD 8.9) in the FEP group. Positive symptoms were slightly elevated in the FEP cohort with a mean BPRS-E psychosis scale score of 9.9 (SD 5.0), while the mean score was 8.5 (SD 3.4) in the UHR group. Negative symptoms were similar between cohorts (supplementary table S1).

The proportion of patients using each substance category was first compared between cohorts (see Table 1 for UHR and FEP prevalence comparisons). The most common substances used by both cohorts were cannabis and tobacco. ATS, cannabis, tobacco, and any illicit substance were all more commonly used in the FEP than the UHR cohort (p < 0.001) with demographic variables such as age and gender controlled for.

In the FEP cohort, the use of any illicit substance predicted higher levels of positive symptoms (OR = 1.06, 95% CI 1.02–1.11) and lower levels of negative symptoms (OR = 0.92, 95% CI 0.86–0.99) compared to non-use. The same pattern in symptomatology was seen with ATS use (positive symptoms OR = 1.08, 95% CI 1.02–1.12; negative symptoms OR = 0.92, 95% CI 0.86–1.00) and tobacco use (positive symptoms OR = 1.05, 95% CI 1.00–1.09; negative symptoms OR = 0.92, 95% CI 0.87–0.98). For cannabis use, higher levels in positive symptoms was seen in isolation (OR = 1.05, 95% CI 1.01–1.10) without any difference observable in negative symptoms.

In the UHR group, lower levels of negative symptoms were associated with the use of any illicit substance (OR = 0.90, 95% CI 0.82–0.98), with ATS (OR = 0.87, 95% CI 0.76–0.99), and with cannabis (OR = 0.91, 95% CI 0.83–0.99). No significant associations with positive symptom levels were found with the use of any substance in the UHR group. Neither symptoms of psychological distress (K10) nor functioning (SOFAS) were associated with substance use for either the UHR or FEP group (see Table 2).

The high co-use of multiple substances, particularly tobacco and cannabis, is shown in Fig. 1. Network analysis of the interrelations between substances used in the two different treatment groups (Fig. 2) revealed the qualitative differences in patterns of substance use between the UHR and FEP cohorts. Overall substance use in the UHR group showed stronger associations between the use of each drug and the use of another. On the other hand, drug use in the FEP cohort showed slightly weaker associations between each other as a result of diverse choices in polysubstance use in this group (the median number of types of drug used in the FEP group is 3 compared with 2 in the UHR group, see Table S2).

A high prevalence of the past 3-month substance use was observed in both cohorts, with the use of ATS, cannabis, and tobacco all associated with significantly higher levels of positive symptoms in the FEP group, a finding consistent with extant literature [6, 12, 29]. Conversely, in the UHR group, substance use was associated with reduced negative symptoms. There were similar effect sizes and directions in positive and negative symptoms across the different substances, including tobacco. This convergence of associations is expected in the setting of high co-occurrence of use; for example, 20% of the UHR cohort used tobacco and cannabis, but only 9% used cannabis alone. Concomitant with the pronounced overlap in the use of ATS, tobacco, and cannabis is a homogeneity of presentation among those who use them, with each substance tending towards a clinical picture of more florid positive psychotic symptoms with diminished negative symptoms. Effect size was generally diminished in the UHR cohort, where no associations with positive symptoms were observed. No association was found between substance use and functioning nor psychological distress in either cohort.

Network analysis demonstrates particularly strong intercorrelations between the use of ATS and the use of all other studied substances, suggesting that ATS use may be useful as a marker of risky polysubstance use in early psychosis populations, a finding that merits further investigation but is supported by research in the general population [17]. Weaker associations between the use of substances in the FEP group suggest that substance usage is more spread across a variety of substances. On the other hand, people who used ATS, cocaine, and hallucinogens in the UHR cohort were more tightly grouped. This suggests that intervention in substance use in those who are at UHR status could mitigate a trend towards polysubstance use and polysubstance use disorder in those who go on to experience a FEP, with higher rates of polysubstance use in this group represented in Fig. 1.

A significantly larger proportion of the FEP population was found to have used any illicit substance, tobacco, cannabis, and ATS compared to the UHR cohort. Substance use prevalence in the population sampled here falls within prevalence rates seen in extant literature with 44% of the UHR cohort and 58% of the FEP cohort found to use cannabis; in papers analysed in a review on the topic [2], cannabis use prevalence was found to vary between 33 and 54% for UHR groups and 13 and 64% for FEP populations. The next most common substances used were ATS, a finding echoed by a similar Australian FEP cohort [10].

The use of any illicit substance and of ATS were associated with significantly lower negative symptoms in both cohorts and additionally higher levels of positive symptoms for the FEP cohort only. An association between higher levels of positive symptoms of psychosis and substance use is well established [26, 27]. Relationships between negative symptoms of psychosis and substance use are less consistent in the literature, varying with study methodology employed and cohorts analysed.

Frequency and persistence of substance use are both important clinical variables; however, frequency of substance use for an individual can vary substantially alongside changes in life circumstances. Dichotomising substance use to a single Yes/No within the 3 months prior to presentation casts a broader net in identifying use among this large cohort. The significant differences in symptomatology observed with this method are less likely to be due to acute intoxication or withdrawal which may be seen in people who use regularly at the time of assessment. Therefore, it may be more likely to reflect underlying differences in cohorts between those who do and do not use these substances.

No significant correlation between substance use and functioning or levels of psychological distress were seen. Past studies of FEP populations have also not demonstrated any consistent relationship between measures of functioning and substance use [28]. However, persistent substance use has been specifically studied in an FEP population by excluding individuals who sporadically use substances or who cease using after baseline assessment and before follow-up [36]. Using this method, persistent use was found to be associated with poorer functioning. Given that the method used in this current analysis did not discriminate between persistent and sporadic use the absence of any correlation between substance use and functioning is potentially unsurprising.

Tobacco use predicted higher levels of positive symptoms and lower of negative symptoms in the FEP cohort. This association arises in the setting of a high co-occurence of tobacco use and that of other substances, particularly cannabis. A meta-analysis probing symptomatological correlations of cannabis and nicotine in adults with known psychotic illness found elevated positive symptoms in nicotine and cannabis users, but no elevation in positive symptoms for those who used nicotine only [29], suggesting that this association may be due to the cannabis used by tobacco smokers in the FEP cohort.

Network analysis used in this paper demonstrates that ATS use is more strongly associated with the use of all other surveyed substances than any other substance category. The relationship between ATS and polysubstance use in FEP populations has been demonstrated previously with polysubstance use found in 50% of those who used cannabis but 65% of those who used cocaine/amphetamines in a Canadian cohort [28].

It further demonstrates a difference in the pattern of substance use between cohorts. While the UHR group tends to have tighter grouping and stronger associations between the use of “hard” drugs, the FEP group shows more laxity of correlation, with more varied patterns of substance use. This indicates a spreading in the pattern of substance use as age and psychotic illness progresses with more varied pattern of substance use. Increased polydrug use in FEP compared to UHR groups is quantified in Fig. 1 which shows, for example, identical proportions of people in UHR and FEP using cannabis or ATS alone (9% & < 1%, respectively), but a doubling in the proportion using both cannabis and ATS in the FEP group (2% UHR, 5% FEP).

A strength of these analyses is that with a consent rate of 92%, the data are highly representative of young people presenting to these real-world early intervention in psychosis services. Data comprises two well matched cohorts from appreciably similar demographic and geographical backgrounds, permitting effective comparison between the UHR and FEP groups.

Limitations, however, also must be considered. Firstly, young people using a variety of substances were not separated from those using each substance individually. The similarities seen in the symptomatology of groups using a given substance were therefore confounded by the use of other surveyed substances. Research intending to comment on substance-specific effects, for example between tobacco and psychotic illness, could compare nonusers with tobacco-only users and those who use tobacco and other substances. Polydrug use is so high, however, that dissecting out individual substances becomes difficult due to the statistical requirement of large samples. Secondly, results seen here may not be generalisable outside of Australia given that substance use patterns differ by geography. Thirdly, data analysed relied on routine clinical data collection completed by clinicians and may therefore not be as accurate as that collected in research studies, as there were no specific reliability checks. Client under-reporting of substance use may be common. Although the quality of the MDS data is monitored, the data still suffer from quality issues including data that is missing or subject to entry errors. Finally, the cross-sessional nature of the data does not allow us to evaluate how substance use trajectories may impact on development and progression of clinical symptoms and how different risk factors (e.g. demographics) may mediate or modify the association.