Introduction
Psychotic disorders (e.g. schizophrenia, schizoaffective, schizophreniform, delusional or brief psychotic disorder) are a group of heterogeneous syndromes with indeterminate neurobiological mechanisms (Kahn et al., Reference Kahn, Sommer, Murray, Meyer-Lindenberg, Weinberger, Cannon and Insel2015). The onset of a first-episode psychosis (FEP) is usually preceded for many years by several underlying biological processes at both the peripheral level and the central nervous system (Kahn & Sommer, Reference Kahn and Sommer2015). Both FEP and patients with established psychosis are at markedly increased risk for substance use disorders, particularly nicotine dependence (Hartz et al., Reference Hartz, Pato, Medeiros, Cavazos-Rehg, Sobell and Knowles2014; Volkow, Reference Volkow2009).
Tobacco use is common in the general population, with a worldwide prevalence estimated at 17.5% (WHO, 2021) and remains specially high in patients with psychosis (de Leon & Diaz, Reference de Leon and Diaz2005; de Leon, Becona, Gurpegui, Gonzalez-Pinto, & Diaz, Reference de Leon, Becona, Gurpegui, Gonzalez-Pinto and Diaz2002; Lally et al., Reference Lally, Spaducci, Gardner-Sood, Atakan, Greenwood, Di Forti and Gaughran2019; Mallet et al., Reference Mallet, Le Strat, Schurhoff, Mazer, Portalier, Andrianarisoa and Group2017, Reference Mallet, Le Strat, Schurhoff, Mazer, Portalier, Andrianarisoa and Group2019). Eleven percent of smokers in the general population are heavy smokers (i.e. smoking 25 or more cigarettes per day), as compared with 60–80% rates of heavy smokers among patients with psychosis who smoke (Kelly & McCreadie, Reference Kelly and McCreadie1999; Lally et al., Reference Lally, Spaducci, Gardner-Sood, Atakan, Greenwood, Di Forti and Gaughran2019; Zhang et al., Reference Zhang, Chen, Xiu, Haile, He, Luo and Kosten2013). In the case of FEP, approximately 50–70% of patients report using tobacco (Coletti et al., Reference Coletti, Brunette, John, Kane, Malhotra and Robinson2015; Grossman et al., Reference Grossman, Bowie, Lepage, Malla, Joober and Iyer2017; Wade et al., Reference Wade, Harrigan, Edwards, Burgess, Whelan and McGorry2005).
The relationship between tobacco use and earlier age at onset of psychosis is controversial. Several studies have found no evidence of an association between tobacco use and earlier age of onset in FEP patients (Hickling et al., Reference Hickling, Ortiz-Garcia de la Foz, Ayesa-Arriola, Crespo-Facorro, McGuire and Perez-Iglesias2017; Myles et al., Reference Myles, Newall, Curtis, Nielssen, Shiers and Large2012). However, a meta-analysis showed that daily tobacco smoking was associated with earlier age at onset of psychosis (Gurillo, Jauhar, Murray, & MacCabe, Reference Gurillo, Jauhar, Murray and MacCabe2015).
Moreover, previous studies report that patients with schizophrenia show a 5-fold greater likelihood of having a comorbid substance use disorder (SUD) compared with the general population (Regier et al., Reference Regier, Farmer, Rae, Locke, Keith, Judd and Goodwin1990), with two-thirds of patients with schizophrenia and comorbid SUD reporting tobacco use (Depp et al., Reference Depp, Bowie, Mausbach, Wolyniec, Thornquist, Luke and Harvey2015; McCreadie, Reference McCreadie2002). Furthermore, tobacco smokers with schizophrenia extract 1.3 times more nicotine from each cigarette than smokers in the general population (Lally et al., Reference Lally, Spaducci, Gardner-Sood, Atakan, Greenwood, Di Forti and Gaughran2019; Olincy, Young, & Freedman, Reference Olincy, Young and Freedman1997; Williams, Farmer, Ackenheil, Kaufmann, & McGuffin, Reference Williams, Farmer, Ackenheil, Kaufmann and McGuffin1996).
Some studies have found a positive association between tobacco smoking and a diagnosis of schizophrenia in the general population (Hunter, Murray, Asher, & Leonardi-Bee, Reference Hunter, Murray, Asher and Leonardi-Bee2020), with regular tobacco use preceding the onset of psychosis (Beratis, Katrivanou, & Gourzis, Reference Beratis, Katrivanou and Gourzis2001; Kelly & McCreadie, Reference Kelly and McCreadie1999; Kotov, Guey, Bromet, & Schwartz, Reference Kotov, Guey, Bromet and Schwartz2010; Ma et al., Reference Ma, Li, Meng, Du, Wang, Wang and Li2010) and with a dose–response relationship (Saha et al., Reference Saha, Scott, Varghese, Degenhardt, Slade and McGrath2011). Moreover, among patients with psychosis, there is some evidence linking tobacco smoking with higher odds of a diagnosis of schizophrenia than of other non-affective psychoses (King, Jones, Petersen, Hamilton, & Nazareth, Reference King, Jones, Petersen, Hamilton and Nazareth2021).
There is evidence that nicotine alters signaling in the dopaminergic, cholinergic, and glutamatergic neurotransmitter systems, which results in a negative impact on brain development, based on early exposure to this substance (Ferreira & Coentre, Reference Ferreira and Coentre2020; Smith, McDonald, Bergstrom, Ehlinger, & Brielmaier, Reference Smith, McDonald, Bergstrom, Ehlinger and Brielmaier2015). Like cannabis, tobacco is implicated in the pathophysiology of schizophrenia (de Leon et al., Reference de Leon, Becona, Gurpegui, Gonzalez-Pinto and Diaz2002; Kelly & McCreadie, Reference Kelly and McCreadie1999; Moore et al., Reference Moore, Zammit, Lingford-Hughes, Barnes, Jones, Burke and Lewis2007; Zammit et al., Reference Zammit, Thomas, Thompson, Horwood, Menezes, Gunnell and Harrison2009) and the shared genetic, environmental, and socioeconomic factors of cannabis and tobacco use make it difficult to address the separate influence of each substance on the onset of FEP and its clinical course (Agrawal, Budney, & Lynskey, Reference Agrawal, Budney and Lynskey2012; Gage et al., Reference Gage, Hickman, Heron, Munafo, Lewis, Macleod and Zammit2014; Grossman et al., Reference Grossman, Bowie, Lepage, Malla, Joober and Iyer2017).
Based on previous data, we hypothesized that (1) tobacco use would be more frequent in FEP than in controls even after controlling for cannabis use, (2) FEP tobacco users would show an earlier age at onset of psychosis, and (3) would receive a diagnosis of schizophrenia more frequently than non-users after controlling for cannabis use.
Methods
Study design
The European Network of National Schizophrenia Networks Studying Gene-Environment Interactions (EU-GEI) project is a large, international, multisite, observational study. The network recruited participants between May 2010 and April 2015 from 17 catchment areas across six countries (Brazil, France, Italy, the Netherlands, Spain, and the UK). A subset of these incident cases was recruited for a concurrent case-control study, with controls selected from the same catchment areas as the incident cases. The aim of the project was to examine the interactive genetic, clinical, and environmental determinants involved in the development, severity, and outcome of schizophrenia. Further information about the study procedures can be found elsewhere (Gayer-Anderson et al., Reference Gayer-Anderson, Jongsma, Di Forti, Quattrone, Velthorst, de Haan and Morgan2020; van Os et al., Reference van Os, Rutten, Myin-Germeys, Delespaul, Viechtbauer, van Zelst and Mirjanic2014).
Participants
A total of 2774 participants with FEP comprised of the EU-GEI study sample (Gayer-Anderson et al., Reference Gayer-Anderson, Jongsma, Di Forti, Quattrone, Velthorst, de Haan and Morgan2020; Jongsma et al., Reference Jongsma, Gayer-Anderson, Lasalvia, Quattrone, Mule and Szoke2018). Of those, 1130 FEP patients and 1497 healthy controls (HC) consented to participate in the case-control study. We excluded 167 subjects due to lack of data on tobacco consumption, so we analyzed data from 1105 FEP patients and 1355 HC. Within the group of tobacco-user FEP patients (n = 613), from 35 patients data on frequency of use were missing; therefore, we analyzed data from 224 heavy users and 354 non-heavy users. A recruitment flow chart is shown in the online Supplementary material.
The inclusion criteria for FEP patients were: (1) a diagnosis of psychotic disorder based on the International Classification of Diseases, 10th Edition (ICD-10), codes F20–F33 (WHO, 1992), (2) age between 18 and 64 years, and (3) to be users of the mental health services in the catchment areas. We excluded participants whose psychotic symptoms were due to acute intoxication (ICD-10: F1X.5) or organic psychosis (ICD-10: F09) or if they had previously received antipsychotic medication. Volunteers from the same catchment areas were selected for the control sample, using a mixture of random ad quota sampling, resulting in controls representative of local populations with regard to age, gender, and ethnicity. They had to meet the following inclusion criteria: (1) age between 18 and 64 years, (2) to be a resident of the same catchment area as the group of patients at the time of consent, (3) fluency in the site primary language, and (4) absence of current or past psychotic disorder. Further details about the sample recruitment can be found elsewhere (Gayer-Anderson et al., Reference Gayer-Anderson, Jongsma, Di Forti, Quattrone, Velthorst, de Haan and Morgan2020).
Participants gave written informed consent before inclusion in the study. The research ethics committees of each site approved the study.
Measures
The modified version of the Medical Research Council (MRC) Sociodemographic Schedule (Mallet, Reference Mallet1997) was used to collect sociodemographic data. Diagnoses were obtained using clinical interviews and the 90-item computerized Operational Criteria Checklist for Psychotic Illness and Affective Illness (OPCRIT) system (McGuffin, Farmer, & Harvey, Reference McGuffin, Farmer and Harvey1991; Quattrone et al., Reference Quattrone, Di Forti, Gayer-Anderson, Ferraro, Jongsma, Tripoli and Reininghaus2019; Williams et al., Reference Williams, Farmer, Ackenheil, Kaufmann and McGuffin1996) and assigned to one of two categories: (1) schizophrenia and (2) other psychotic diagnoses. Age of onset of psychosis was calculated by subtracting duration of untreated psychosis from age at the time of assessment. The EU-GEI Tobacco and Alcohol Questionnaire was used to collect information about the use of tobacco in the year prior to the assessment and other legal and illegal substances by trained personnel at each site, whose reliability was assessed throughout the study (κ = 0.7) (Quattrone et al., Reference Quattrone, Di Forti, Gayer-Anderson, Ferraro, Jongsma, Tripoli and Reininghaus2019). For purposes of categorizing the severity of tobacco use in this manuscript we assigned ⩾20 cigarettes per day, equivalent to one pack of cigarettes per day, to the heavy use group and anything less to the non-heavy use group, a cutoff value supported in the tobacco literature (Chin, Hong, Gillen, Bates, & Okechukwu, Reference Chin, Hong, Gillen, Bates and Okechukwu2013; Kay-Lambkin et al., Reference Kay-Lambkin, Edwards, Baker, Kavanagh, Kelly, Bowman and Lewin2013; Shelef, Diamond, Diamond, & Myers, Reference Shelef, Diamond, Diamond and Myers2009). We used a modified version of the Cannabis Experience Questionnaire from the EU-GEI project (CEQEU-GEI), and categorized lifetime frequency of use into three groups: (1) never or occasional use (less than once a week), (2) more than weekly use (but less than daily), and (3) daily use (Di Forti et al., Reference Di Forti, Quattrone, Freeman, Tripoli, Gayer-Anderson, Quigley and Group2019). Impairment of functionality was measured with the Global Assessment of Functionality (GAF) (Jones, Thornicroft, Coffey, & Dunn, Reference Jones, Thornicroft, Coffey and Dunn1995).
Statistical analysis
Normality distribution of quantitative outcomes in a large sample was tested following Satorra–Bentler's procedure (Russell, Reference Russell2016; West, Sandler, Pillow, Baca, & Gersten, Reference West, Sandler, Pillow, Baca and Gersten1991) in which a variables' distribution was considered normal if skewness was <2 and kurtosis was <7. Little's missing completely at random (MCAR) test showed data were missing at random (χ2 = 13.9; p = 0.179) and comparisons between included and excluded participants showed no significant differences in sociodemographic and clinical variables (sex, age, years of education, tobacco use, alcohol use, other drug use, age of psychosis onset, and diagnosis). We used means and standard deviations for continuous variables and frequency and percentages for categorical variables. For comparisons between two groups, we used χ2 tests for categorical variables. For continuous variables, we used t tests. For comparisons of more than two groups [those between: (1) tobacco-user FEP, (2) tobacco non-user FEP, (3) tobacco-user controls, and (4) tobacco non-user controls), we used analysis of variance and analysis of covariance (ANCOVA) followed by Bonferroni post-hoc correction. We used η 2 and Cohen's d tests to calculate the effect sizes, as appropriate. Binary logistic regression model analyses were performed to examine the association of tobacco use and frequency of tobacco use with the case and control groups and with the dichotomous variable for diagnosis (schizophrenia v. other psychotic disorders). Linear regression model analyses were used to analyze the association of tobacco use and frequency of tobacco use with age at onset of psychosis.
Two models were conducted for each regression analysis: model 1 was adjusted for: age, sex, country, years of education, and alcohol use (except for the age-of-onset analysis, in which the confounding variable ‘age’ was not included to avoid collinearity). Model 2 additionally included cannabis use frequency [categorized as never or occasional use, used more than once a week or daily use (Di Forti et al., Reference Di Forti, Quattrone, Freeman, Tripoli, Gayer-Anderson, Quigley and Group2019)]. All categorical confounders were included as fixed factors. For logistic regression models we report adjusted odds ratios (AORs) and for linear regression models, we report beta values.
Supplementary ANCOVA (for age at psychosis onset) and binary logistic regression analyses (for case-control status and diagnosis) were used to assess the main and interaction effects of tobacco (yes/no) and cannabis use (yes/no) on the different outcomes. In addition, we conducted ANCOVAs with a Bonferroni correction of the post-hoc comparisons and binary logistic regression analyses including categorical variables reflecting the possible combinations of (i) tobacco and cannabis use [i.e. (1) tobacco only (TO), (2) tobacco and cannabis (T&C), (3) cannabis-only (CO), and (4) nonuse (NU)] and (ii) frequency of tobacco use and cannabis use [(1) heavy-tobacco only (HTO), (2) non-heavy-tobacco only (NHTO), (3) heavy tobacco and cannabis (HT&C), (4) non-heavy tobacco and cannabis (NHT&C), (5) cannabis-only (CO), (6). nonuse (NU)] to assess the effect of the different combined patterns of use on the same outcomes.
Finally, we performed supplementary analyses substituting the covariable ‘country’ for ‘site’ to check if there were any effects related to specific locations and we repeated the main analyses after including the covariable ‘other substance use’. A cut-off value of p < 0.05 was chosen for statistical significance. All statistical analyses were conducted using SPSS 25.
Results
Sociodemographic and clinical description
FEP v. controls
The sample comprised of a total of 1105 (44.9%) patients [613 (55.5%) tobacco users and 492 (44.5%) non-users] and 1355 (55.1%) controls [329 (24.3%) tobacco users and 1026 (75.7%) non-users]. Compared to controls, FEP patients were younger (t = −10.5; p ⩽ 0.001; d = 0.4), more often men [χ2 = 54.6; p ⩽ 0.001, OR 1.8; 95% confidence interval (CI) [1.5–2.1]], and had a lower education level (t = −10.8; p ⩽ 0.001; d = 0.4). Controls were more likely to have been employed (χ2 = 22.6; p ⩽ 0.001, OR 1.9; 95% CI [1.5–2.6]) and more often presented a higher socioeconomic status (χ2 = 205.7; p ⩽ 0.001).
More patients than controls reported having used tobacco (χ2 = 250.6; p ⩽ 0.001; OR 3.9; 95% CI [3.3–4.6]) and in a heavy use frequency (χ2 = 7.6; p = 0.006; OR 1.5; 95% CI [1.1–2.0]). Patients were more frequently TO users (χ2 = 157.6; p ⩽ 0.001; OR 3.4; 95% CI [2.8–4.1]) and T&C users (χ2 = 165.2; p ⩽ 0.001; OR 5.6; 95% CI [4.2–7.4]) than controls. Patients also reported smoking more tobacco cigarettes (t = 3.3; p ⩽ 0.001; d = 0.24) than controls users. We found no difference in type of tobacco used (cigarettes, cigars, pipes, snuff, or chewing tobacco) between FEP and control tobacco users (see Table 1). Further information about the sociodemographic distribution of the site and the use of other drugs is shown in the online Supplementary material.
Sociodemographic and substance use comparisons between the four groups
FEP, first-episode psychosis; HC, healthy controls; GAF, General Assessment of Functioning. Significant results are highlighted in bold.
*Bonferroni test values: a (p ⩽ 0.001); b (p ⩽ 0.001); c (p ⩽ 0.001); d (p ⩽ 0.001); e (p ⩽ 0.001); f (p = 023); g (p = 0.004); h (p = 003); i (p ⩽ 0.001); j (p = 0.017); k (p ⩽ 0.001); l (p = 0.031); m (p ⩽ 0.001); n (p = 0.011); o (p ⩽ 0.001); p (p ⩽ 0.001); q (p ⩽ 0.001); r (p = 0.043); s (p ⩽ 0.001); t (p ⩽ 0.001); u (p = 0.010); v (p ⩽ 0.001); w (p ⩽ 0.001);x(p ⩽ 0.001); y (p = 0.006); z (p = 0.014); aa (p = 0.022); ab (p ⩽ 0.001); ac (p ⩽ 0.001); ad (p ⩽ 0.001); ae (p ⩽ 0.001); af (p = 001); ag (p ⩽ 0.001); ah (p ⩽ 0.001); ai (p ⩽ 0.001); aj (p ⩽ 0.001); ak (p = 0.003); al (p = 0.019); am (p = 0.007); an (p ⩽ 0.001); ao (p = 0.011); ap (p ⩽ 0.001); aq (p ⩽ 0.001); ar (p ⩽ 0.001); as (p ⩽ 0.001); at (p ⩽ 0.001); au (p ⩽ 0.001); av (p = 0.011); aw (p = 0.004); ax (p ⩽ 0.001); ay (p ⩽ 0.001); az (p ⩽ 0.001); ba (p ⩽ 0.001); bb (p ⩽ 0.001); bc (p ⩽ 0.001); bd (p ⩽ 0.001); be (p ⩽ 0.001); bf (p ⩽ 0.001); bg (p = 0.007).
FEP users v. FEP non-users
Bivariate analyses showed that FEP patients with a history of tobacco use were more frequently male (χ2 = 34.5; p ⩽ 0.001; OR 2.1; 95% CI [1.6–2.7]), on average 3.3 years younger (t = 5.1; p ⩽ 0.001; d = 0.3), and received half a year of education less (t = −2.1; p = 0.037; d = 0.02) than FEP non-users. A diagnosis of schizophrenia was more common in FEP tobacco users than in non-users (χ2 = 19.057; p = 0.002) than in people with another diagnosis of psychosis. FEP tobacco users had an earlier onset of psychosis compared with FEP non-users (t = −4.76; p ⩽ 0.001; d = 0.3) (see Table 2).
Comparisons between tobacco user and non-user patients in sociodemographic and clinical variables
FEP, first-episode psychosis; GAF, General Assessment of Functioning; DUP, duration of untreated psychosis. Significant results are highlighted in bold.
a OR and 95% CI were calculated for dichotomous variables.
FEP heavy users v. FEP non-heavy users
FEP patients with heavy-tobacco use were on average 1.6 years younger (t = −1.8; p = 0.075; d = 0.2) and had received almost 2 more years of education less (t = 4.6; p ⩽ 0.001; d = −0.4) than non-heavy-tobacco users. There were no differences in diagnosis or age at onset of psychosis between groups based on frequency of tobacco use (see Table 3).
Comparisons between heavy and non-heavy-tobacco user FEP patients in sociodemographic and clinical variables
FEP, first-episode psychosis; GAF, General Assessment of Functioning. Significant results are highlighted in bold.
a OR and 95% CI were calculated for dichotomous variables.
Rates of tobacco use
FEP v. controls
Logistic regression models showed that tobacco use was 3.3 times higher in the FEP group than in the control group (AOR 3.3, 95% CI [2.7–4.0]). After including the frequency of cannabis use as a covariate in the model there was a slight reduction in the effect size of tobacco use (AOR 2.6, 95% CI [2.1–3.3]). In this model, daily cannabis use was associated with increased odds of FEP (AOR = 3.1, 95% CI [2.3–4.1]) compared with never or occasional use (see Table 4).
Logistic regression including patients and controls with and without cannabis use covariate
AOR, adjusted odds ratio. Significant results are highlighted in bold.
bThe reference group for ‘sex’ was male; for ‘country’ was Brazil. For ‘Cannabis frequency’, the reference group was ‘Never/Occasional use’.
Participants who were TO users (AOR 3.0, 95% CI [2.4–3.7]) and those who were T&C users (AOR 3.9, 95% CI [2.9–5.4]) had higher odds of FEP that NU, with a non-significant trend for differences between the CO and the NU groups (p = 0.051; AOR 0.6, 95% CI [0.3–1.0]) (see the online Supplementary material).
Heavy users v. non-heavy users
Logistic regression models showed that heavy-tobacco users had higher odds of FEP than non-heavy users, even after covarying by the cannabis use [model 1 (AOR 1.7, 95% CI [1.2–2.4]), model 2 (AOR 1.7, 95% CI [1.2–2.4])]. Both HTO (AOR 1.6, 95% CI [1.1–2.4]) and HT&C users (AOR 2.8, 95% CI [1.4–5.9]) showed significantly higher odds of FEP than NHTO users. No significant differences were found between the HTO and the HT&C groups. Heavy users also showed significantly higher odds of FEP than tobacco non-users even after controlling by cannabis use [model 1 (AOR 4.0, 95% CI [3.0–5.4]), model 2 (AOR 3.4, 95% CI [2.5–4.7])] (for more information, see the online Supplementary material). Figure 1 presents the OR of patients and controls for the combined measure of use and frequency of tobacco and cannabis.
ORs of patients and controls for the combined measure of use and frequency of tobacco and cannabis. ORs in model 1 were adjusted for age, sex, country, years of education, and alcohol use and ORs in model 2 were additionally adjusted for cannabis use frequency (categorized as never or occasional use, used more than once a week, or daily use). Error bars represent 95% CIs. OR, odds ratio.
Age of onset
Tobacco use was associated with an earlier age at psychosis onset (β = −2.3; p ⩽ 0.001; 95% CI [−3.7 to −0.9]), but this result became non-significant when cannabis use frequency was included as a covariate (β = −0.4; p = 0.610; 95% CI [−1.1 to 1.9]). Heavy-tobacco use (v. non-heavy use) was not significantly associated with an earlier age of psychosis onset [model 1 (β = −0.5; p = 0.590; 95% CI [−1.3 to −2.3]); model 2 (β = −0.1; p = 0.876; 95% CI [−1.6 to 1.9])]. Supplementary ANCOVA analyses revealed a significant interaction effect of tobacco and cannabis use (F = 18.6; p ⩽ 0.001). Concretely, the group of T&C users had an earlier age at onset than the NU and TO groups, and the CO group presented an earlier age of psychosis onset than the NU and TO groups (see the online Supplementary material). Concerning frequency of use, the NHTO group presented an earlier age of psychosis onset than the NU group; the HT&C group had an earlier age at onset than the HTO, NHTO, and NU groups; the NHT&C group presented an earlier age at onset than the HTO and NU groups; and the CO group showed an earlier age of psychosis onset than the HTO and the NU groups (see the online Supplementary material).
Diagnosis
FEP patients who used tobacco had 1.3 times higher odds of being diagnosed with schizophrenia than other psychotic disorders. When we included frequency of cannabis use in the model, tobacco use was no longer significantly associated with diagnosis [model 1 (AOR 1.3; 95% CI [1.0–1.8]); model 2 (AOR 1.1; 95% CI [0.8–1.5])].
Supplementary logistic regression analyses did not detect a significant interaction effect of cannabis and tobacco use on diagnosis (p = 0.125; AOR −1.6; 95% CI [0.9–3.1]). The group of T&C use presented higher odds of schizophrenia diagnosis than both the TO (AOR 1.5; 95% CI [1.0–2.3]) and NU groups (AOR 1.9; 95% CI [1.3–2.9]) (see the online Supplementary material). Regarding frequency of tobacco use, the HT&C (AOR 2.0; 95% CI [1.1–3.7]) and NHT&C (AOR 1.7; 95% CI [1.1–2.7]) groups presented higher odds of receiving a schizophrenia diagnosis than NU, with no significant differences between the remaining patterns of use; see the online Supplementary material for further details.
Analyses controlled by site found comparable results in terms of the direction and magnitude of the effects (see the online Supplementary material). Analyses controlled by use of other substances showed that tobacco use was 5.7 times higher in the FEP group than in the control group (model 1: AOR 5.7, 95% CI [3.5–9.1]). This effect remained significant after the inclusion of frequency of cannabis use as a covariate in model 2 (AOR 5.4, 95% CI [3.3–8.9]). However, we observed no significant effects of tobacco use on age of psychosis onset and diagnosis (schizophrenia v. other psychosis) regardless of the patterns of tobacco and cannabis use or the frequency of tobacco use (see the online Supplementary material for further information).
Discussion
Tobacco users had more than thrice the odds of experiencing FEP than did non-tobacco users and heavy-tobacco users had almost twice the odds of experiencing FEP than did non-heavy-tobacco users. These results remained significant after controlling for cannabis use. Tobacco use was associated with an earlier age at psychosis onset and higher odds of receiving a diagnosis of schizophrenia. However, the latter two associations were no longer significant after controlling for cannabis use.
As expected, FEP patients smoked tobacco more frequently and more heavily than controls, which is consistent with previous literature (Coletti et al., Reference Coletti, Brunette, John, Kane, Malhotra and Robinson2015; Depp et al., Reference Depp, Bowie, Mausbach, Wolyniec, Thornquist, Luke and Harvey2015; Grossman et al., Reference Grossman, Bowie, Lepage, Malla, Joober and Iyer2017; Kelly & McCreadie, Reference Kelly and McCreadie1999; Lally et al., Reference Lally, Spaducci, Gardner-Sood, Atakan, Greenwood, Di Forti and Gaughran2019; McCreadie, Reference McCreadie2002; Wade et al., Reference Wade, Harrigan, Edwards, Burgess, Whelan and McGorry2005; Zhang et al., Reference Zhang, Chen, Xiu, Haile, He, Luo and Kosten2013). This result was also observed in patients with established psychosis, concretely schizophrenia [with a 60% prevalence of tobacco regular use (Jamal et al., Reference Jamal, Homa, O'Connor, Babb, Caraballo, Singh and King2015)]. However, the increase in the exposure to tobacco, particularly in adolescence (Smith et al., Reference Smith, McDonald, Bergstrom, Ehlinger and Brielmaier2015) is crucial to examine the evidence regarding the early stages of psychosis and increased risk for nicotine dependence (Hartz et al., Reference Hartz, Pato, Medeiros, Cavazos-Rehg, Sobell and Knowles2014; Scott et al., Reference Scott, Matuschka, Niemela, Miettunen, Emmerson and Mustonen2018; Volkow, Reference Volkow2009). In that regard, we observed in our study that heavy-tobacco-only use and the combination of heavy-tobacco and current cannabis use were more frequently observed in FEP patients than in non-users. This result is consistent with previous findings showing that daily tobacco smoking is associated with a greater risk of onset of psychosis (Gurillo et al., Reference Gurillo, Jauhar, Murray and MacCabe2015), this could possibly be explained by signaling alterations produced by nicotine on the dopaminergic, cholinergic, and glutamatergic neurotransmitter systems (Smith et al., Reference Smith, McDonald, Bergstrom, Ehlinger and Brielmaier2015) that may involve interactions between the cholinergic-nicotinic system and the dopaminergic system (Gurillo et al., Reference Gurillo, Jauhar, Murray and MacCabe2015; Mallet et al., Reference Mallet, Le Strat, Schurhoff, Mazer, Portalier, Andrianarisoa and Group2019). However, the cross-sectional methodology of this study and the use of self-reported and 1 year retrospective measures to assess the use of tobacco precludes an inference about causality. Moreover, heavy substance use during critical periods for brain development, such as pregnancy, adolescence, and young adulthood, could be specifically associated with later onset of psychosis in genetically vulnerable individuals (Gage et al., Reference Gage, Hickman, Heron, Munafo, Lewis, Macleod and Zammit2014; McGrath et al., Reference McGrath, Alati, Clavarino, Williams, Bor, Najman and Scott2016; Mustonen et al., Reference Mustonen, Ahokas, Nordstrom, Murray, Maki, Jaaskelainen and Niemela2018; van Os, Kenis, & Rutten, Reference van Os, Kenis and Rutten2010; Zammit et al., Reference Zammit, Thomas, Thompson, Horwood, Menezes, Gunnell and Harrison2009). Besides the hypothesis of considering tobacco per se as a risk factor for psychosis, its impact on the phenotypic expression of the disorder and, consequently, the possibility of inducing a differentiated entity in terms of etiopathogenesis, clinical features, prognosis, response to conventional treatments, and/or preventive interventions, is still inconclusive (Gonzalez-Blanco et al., Reference Gonzalez-Blanco, Garcia-Portilla, Gutierrez, Mezquida, Cuesta, Urbiola and Group2021). Shared genetic risk for schizophrenia and smoking behaviors in a European population have been observed, suggesting the presence of common mechanisms underlying these associations (Peterson et al., Reference Peterson, Bigdeli, Ripke, Bacanu, Gejman, Levinson and Fanous2021). In our sample, tobacco, cannabis, and other drugs were more frequently used in the group of FEP patients, suggesting a potential vulnerability that could make patients more prone to use other substances because of their reinforcing effects on the dopaminergic system, often referred to as ‘the gateway effect’ (Kandel, Reference Kandel2002). An alternative explanation would concern the self-medication hypothesis in FEP, as substance intake may reduce psychotic symptoms (Fang et al., Reference Fang, Wang, Zhu, Lin, Cheng, Zou and Cui2019).
We found a significant difference between the FEP tobacco user and non-user groups in age at onset, but including cannabis covariation diluted the effect. Contrary to our hypothesis, this finding is consistent with previous literature that observed an absence of association between tobacco use and age at onset of psychosis (Gonzalez-Blanco et al., Reference Gonzalez-Blanco, Garcia-Portilla, Gutierrez, Mezquida, Cuesta, Urbiola and Group2021; Hickling et al., Reference Hickling, Ortiz-Garcia de la Foz, Ayesa-Arriola, Crespo-Facorro, McGuire and Perez-Iglesias2017; Myles et al., Reference Myles, Newall, Curtis, Nielssen, Shiers and Large2012), and it is also consistent with previous studies in which cannabis use was related to earlier age at onset of psychosis (Di Forti et al., Reference Di Forti, Sallis, Allegri, Trotta, Ferraro, Stilo and Murray2014; Large, Sharma, Compton, Slade, & Nielssen, Reference Large, Sharma, Compton, Slade and Nielssen2011; Murray et al., Reference Murray, Englund, Abi-Dargham, Lewis, Di Forti, Davies and D'Souza2017). One explanation of the attenuation of the tobacco effects after including cannabis as a covariate may be the differences in the duration assessed with the self-reported measures for each substance. In the case of cannabis, we asked participants to report their lifetime use but in the case of tobacco, we only asked about the last year. This could have influenced the weight of the effect of each of the two substances. Similarly, when we analyzed the association between tobacco users and diagnosis, we primarily observed that FEP patients had 1.3 times higher odds of a diagnosis of schizophrenia, but this association disappeared after controlling for cannabis use. This is consistent with earlier findings that both the earlier onset of psychosis and the more frequent diagnosis of schizophrenia may be explained by the cannabis use itself (Belbasis et al., Reference Belbasis, Kohler, Stefanis, Stubbs, van Os, Vieta and Evangelou2018; Di Forti et al., Reference Di Forti, Sallis, Allegri, Trotta, Ferraro, Stilo and Murray2014; Lowe, Sasiadek, Coles, & George, Reference Lowe, Sasiadek, Coles and George2019). Nevertheless, schizophrenia was more frequently diagnosed in FEP patients who used the combination of tobacco and cannabis than in the group of only-tobacco or non-users. Also, the combination of tobacco (both heavy and non-heavy use) and cannabis use was associated with an earlier age of psychosis onset than heavy-tobacco-only use and non-use. Cannabis-only users showed an earlier age of psychosis onset than non-users or heavy-tobacco users. The fact that only 2.7% of the FEP sample used cannabis-only limits the interpretation of these results. In the present study, we could identify co-use of tobacco and cannabis in FEP patients, but we could not quantify the proportion of tobacco and cannabis in each joint. Most people who use cannabis in Europe include tobacco in a variable proportion (Dekker et al., Reference Dekker, Meijer, Koeter, van den Brink, van Beveren, Kahn and Investigators2012; Gage et al., Reference Gage, Hickman, Heron, Munafo, Lewis, Macleod and Zammit2014; Jones et al., Reference Jones, Gage, Heron, Hickman, Lewis, Munafo and Zammit2018; Rabin & George, Reference Rabin and George2015), either concurrently with cigarettes (co-use) or as a component of cannabis joints (simultaneous use), so cannabis-only use is very rare (Quigley & MacCabe, Reference Quigley and MacCabe2019). Disentangling the independent contributions of each substance is challenging, as most of the information regarding substance use was retrospectively self-reported by participants (Gonzalez-Blanco et al., Reference Gonzalez-Blanco, Garcia-Portilla, Gutierrez, Mezquida, Cuesta, Urbiola and Group2021) and also most studies about cannabis use and its relation with psychosis have not taken into consideration the potential effects of the mixture of cannabis and tobacco in a simultaneous use. We know that smoking cannabis and tobacco together increases the amount of tetrahydrocannabinol inhaled per g (Van der Kooy, Pomahacova, & Verpoorte, Reference Van der Kooy, Pomahacova and Verpoorte2008), thus enhancing the effects of cannabis, which could lead to a higher risk of psychosis. In this regard, the results of a recent study suggest that there is a synergistic effect of tobacco and cannabis on psychosis risk (Jones et al., Reference Jones, Gage, Heron, Hickman, Lewis, Munafo and Zammit2018; Quigley & MacCabe, Reference Quigley and MacCabe2019). In our study we found a significant interaction effect between tobacco and cannabis use on age of onset, but we did not find differences between tobacco-only and tobacco and cannabis use in FEP odds or diagnosis. Prospective studies may help to clarify whether these associations are due to cannabis or due to the combination of tobacco and cannabis.
To our knowledge, this is the first study to examine the association between tobacco use and its frequency of use, with FEP diagnosis, controlling for the use of other substances such as cannabis, alcohol, and other substance of abuse and examining the differences between use of tobacco-only and the combination of tobacco and cannabis in a large, international study dataset consisting of patients with FEP and HC. Some limitations of the study include: first, cross-sectional studies limit assessment of causality. Therefore, prospective measurements are needed to confirm if tobacco use increases the risk of psychotic disorders. Second, the presence and absence, as well as the frequency of tobacco use and other substances of abuse, was self-reported, and this information was not validated by biological samples, like urine analysis or expired carbon monoxide measures, in the case of tobacco smoking severity. However, research has shown that reporting of drug abuse by adults with psychosis is generally accurate (Van Dorn, Desmarais, Scott Young, Sellers, & Swartz, Reference Van Dorn, Desmarais, Scott Young, Sellers and Swartz2012). Third, although we analyzed the independent effect of tobacco-only, tobacco and cannabis co-use, cannabis-only and non-use on the outcomes, this study did not quantify the proportion of tobacco which could have been included in cannabis preparations in cannabis-only users. Fourth, reports about tobacco use in patients and controls were limited to the 12 months prior to enrolment in the study, so we lack information about lifetime tobacco use. Therefore, we are not able to explore the cumulative effect of chronic tobacco use on the odds and age at onset of psychosis. Finally, another limitation of this study is the use of an arbitrary cut-off value to categorize the continuous variable of tobacco use frequency.
In conclusion, tobacco use and heavy-tobacco use are associated with higher odds of FEP. An earlier age at onset and greater odds of receiving a diagnosis of schizophrenia is found in patients who have used both tobacco and cannabis than in non-users. Given these results, tobacco consumption should be thoroughly assessed and treated in clinical practice, especially in concurrent cannabis users and in populations vulnerable to psychosis.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291723000806
Acknowledgments
We thank all of the study participants and our EU-GEI WP2 colleagues.
Financial support
The European Network of National Schizophrenia Networks Studying Gene-Environment Interactions (EU-GEI) Project is funded by grant agreement HEALTH-F2-2010-241909 (Project EU-GEI) from the European Community's Seventh Framework Programme. The Brazilian study was funded by grant 2012/0417-0 from the São Paulo Research Foundation. Dr Rapado-Castro is a Ramon y Cajal Research Fellow (RYC-2017-23144), Spanish Ministry of Science, Innovation and Universities and was supported by a NARSAD independent investigator grant (no. 24628) from the Brain & Behavior Research Foundation. Dr Díaz-Caneja, Parellada, Rapado-Castro, and Dr Arango were partially supported by the Spanish Ministry of Science and Innovation, Instituto de Salud Carlos III (PI15/00723, SAM16PE07CP1, PI16/02012, PI17/00481, PI18/00753, PI19/01024, PI20/00721, JR19/00024, PI21/00701), co-financed by ERDF Funds from the European Commission, ‘A way of making Europe’, CIBER-Consorcio Centro de Investigación Biomédica en Red- (CB/07/09/0023); Madrid Regional Government (S2022/BMD-7216 AGES 3), European Union Structural Funds, European Union Seventh Framework Programme, European Union H2020 Programme under the Innovative Medicines Initiative 2 Joint Undertaking (grant agreement No. 101034377, Project PRISM-2, and grant agreement no. 777394, Project AIMS-2-TRIALS), European Union Horizon Europe, the National Institute of Mental Health of the National Institutes of Health under Award Number 1U01MH124639-01 (Project ProNET) and Award Number 5P50MH115846-03 (project FEP-CAUSAL), Fundación Familia Alonso, and Fundación Alicia Koplowitz. Dr Marta Di Forti is funded by the UK Medical Research Council (MRC) MR/T007818/1. Dr Bernardo acknowledges the support of the Spanish Ministry of Science, Innovation and Universities (PI08/0208; PI11/00325; PI14/00612) integrated into the Plan Nacional de I + D + I and cofinanced by ISCIII-Subdirección General de Evaluación and the Fondo Europeo de Desarrollo Regional (FEDER); the Instituto de Salud Carlos III; the CIBER of Mental Health (CIBERSAM); the Secretaria d'Universitats i Recerca del Departament d'Economia i Coneixement (2017SGR1355), and the CERCA Programme/Generalitat de Catalunya and the Departament de Salut de la Generalitat de Catalunya for the PERIS grant SLT006/17/00345.
Conflict of interest
Dr Arango has been a consultant to or has received honoraria or grants from Acadia, Angelini, Gedeon Richter, Janssen-Cilag, Lundbeck, Minerva, Otsuka, Roche, Sage, Servier, Shire, Schering Plough, Sumitomo Dainippon Pharma, Sunovion, and Takeda. Dr Marta Di Forti has received honoraria for educational seminars by Janssen. Dr Murray has received similar honoraria from Janssen, Lundbeck, Otsuka, Sunovian, and MSD. Dr Díaz-Caneja has received honoraria from Exeltis and Angelini and travel support from Janssen and Otsuka. Dr Bobes has received research grants and served as consultant, advisor, or speaker within the last 5 years for: AB-Biotics, Acadia Pharmaceuticals, Angelini, Casen Recordati, D&A Pharma, Exeltis, Gilead, Janssen-Cilag, Lundbeck, Mundipharma, Otsuka, Pfizer, Reckitt-Benckiser, Roche, Sage Therapeutics, Servier, Shire, Schwabe Farma Ibérica, and has received research funding from the Spanish Ministry of Economy and Competiveness – Centro de Investigación Biomedica en Red area de Salud Mental (CIBERSAM) and Instituto de Salud Carlos III, Spanish Ministry of Health, and the 7th Framework Program of the European Union. Dr García-Portilla has been a consultant to and/or has received honoraria/grants from Angelini, Alianza Otsuka-Lundbeck, Janssen-Cilag, Lundbeck, Otsuka, and Sage Therapeutics. Dr Bernardo has been a consultant for, received grant/research support and honoraria from, and been on the speakers/advisory board of ABBiotics, Adamed, Angelini, Casen Recordati, Janssen-Cilag, Menarini, Rovi, and Takeda. The rest of the authors declare they have no conflict of interest.
