A cross-sectional study was performed that was based on a local research cohort “Deinstitutionalization Amsterdam.” This cohort consisted of a representative sample of adults with a severe mental illness living independently (i.e., not clinically admitted). Participants were recruited from the SMI patient populations of the two largest mental health care institutions located in Amsterdam, where they receive mental health care. Detailed information about the composition of this cohort and the study procedures is described elsewhere [18]. In the current study, we included 281 patients diagnosed with a psychotic disorder (according to the DSM-V criteria). Within this sample, 172 (61.2%) patients had a schizophrenia spectrum disorder (DSM-IV 295.4 and 295.9), 56 (19.9%) were diagnosed with a schizoaffective disorder (DSM-IV 295.7), and 53 (18.9%) had a diagnosis of unspecified psychosis (293.81, 297.1, and 298.8). Inclusion criteria were an age range from 18 to 65 years and proficiency in either Dutch or English. The study was approved by the medical ethical review committee of the Amsterdam University Medical Centre (METC-AMC W16_276 # 16.324). Prior to the interview, participants provided written informed consent concerning both the interview and retrieval of information from patient files. All participants received a financial compensation of 15 euros. The data were obtained by research assistants from the mental health care institutions who interviewed the participants using a series of structured and validated questionnaires on among others substance use and social functioning. Data were collected between December 2017 and February 2020.

All participants provided information on sociodemographic features. A priori, we selected age, gender, and severity of psychopathology as potential confounders as these variables are associated with substance use and several aspects of social functioning [33, 34]. To rate the severity of psychopathology, the Brief Psychiatric Rating Scale (BPRS) was administered [35]. The BPRS is a widely used 24-item tool to measure the presence and severity of several psychiatric symptoms and behavior in patients with a psychiatric disorder by observation and self-reports during a semi-structured interview.

The BPRS consists of five different subscales (positive symptoms, negative symptoms, affectivity, resistance, and activation). In the analyses, the BPRS total score was included as a covariate. The BPRS has been tested for validity and reliability, which appeared to be satisfactory [36].

Statistical analyses were performed using IBM SPSS Statistics version 26. The normality of the distribution of scores on the outcomes of interest was assessed by visual inspection of the histograms, which showed no deviations from normality. Linear regression models were used to evaluate the associations between substance use and social aspects. Each substance was entered as an independent variable with each of the self-reported aspects of social functioning as the dependent variable while correcting for age, gender, and psychopathology. As the independent variables were multi-categorical, dummy variables were created with “no use” being the reference group. As sensitivity analyses, a second set of linear regression analyses, were performed with the independent variables number of cigarettes smoked per day and the amount of cannabis smoked per day (uniformed in grams) instead of the multi-categorical variables. Finally, we explored the potential differences in self-reported aspects of social functioning between no users, single substance users, and polysubstance users (≥two substances). All statistical tests were two-sided, and significance was set at p < 0.05.

Compared to nonsmokers, both intermediate and heavy smoking patients showed lower scores on loneliness (E = −0.580, SE = 0.258, p = 0.025 and E = −0.547, SE = 0,272, p = 0.046, respectively) when correcting for confounders, as shown in Table 2. A post hoc analysis (Supplement 2) revealed a negative association between the number of cigarettes smoked per day and levels of loneliness (E = −0.017, SE = 0.008, p = 0.042). Perceived social support, reported social participation, and stigmatization were not significantly associated with tobacco after adjusting for confounders.Table 2.

Results of regression models evaluating the associations between substances and self-reported aspects of social functioning (N = 281)

Patients who used cannabis daily reported less social participation deficits than patients who did not use cannabis (E = −0.348, SE = 0.145, p = 0.017). In a sensitivity analysis, we explored whether a dose–response relationship existed between the amount of cannabis smoked per day (uniformed in grams) and reported social participation. No significant association was found, as shown in Supplement 3. Neither perceived social support, reported stigmatization, nor loneliness were significantly associated with cannabis use when correcting for confounders (Table 2).

Problematic alcohol use was associated with more perceived social support compared to no-alcohol use (E = 3.152, SE = 1.102, p = 0.005) after adjusting for confounders, as shown in Table 2. Given differences in the literature regarding the cut-off scores for problematic alcohol use [24, 37], we also explored whether this association remained significant when raising the AUDIT cut-off score to 16, indicating severe problematic use [37]. As shown in Supplement 4, outcomes were not affected (E = 3.394, SE = 1.581, p = 0.033). Alcohol use was not significantly related to social participation, stigmatization, and loneliness when correcting for all confounders.

Patients with polysubstance use reported less loneliness compared to no users (E = −0.569, SE = 0.287, p = 0.049) (Supplement 5). No significant differences were found between no users, single substance users, and polysubstance users on all other self-reported aspects of social functioning.

The current study mapped substance use among community-dwelling patients with a psychotic disorder in the Netherlands and examined associations between the use of specific substances (tobacco, cannabis, alcohol) and several self-reported aspects of social functioning.

Tobacco use (intermediate and heavy use) was associated with less self-reported loneliness compared to those who did not smoke. Patients with cannabis use reported more social participation compared to non-cannabis-using patients. Problematic alcohol use was associated with more self-reported social support compared to non-problematic alcohol use. On the subscales of the stigma scale (including self-stigma [disclosure] and experienced stigma [discrimination]), no differences were found between any of the substance user groups. Finally, we found that self-reported loneliness was less prevalent among patients with polysubstance use than among nonusers. In summary, severe use of both cannabis and alcohol seems to be associated with better self-reported scores on some of the outcomes investigated. Concerning tobacco, this relationship was also found in patients with low to intermediate use and heavy use.

The prevalence estimate of tobacco use in our study (55.8%) is consistent with the rate (62%) found in a large population of patients with a psychotic disorder [4]. With respect to cannabis, the prevalence of daily use in our study population (9.6%) was lower compared to rates of cannabis abuse in psychotic patients in general (20%), as shown in a meta-analysis [38]. Concerning alcohol, the prevalence estimate of problematic use of 22.4% obtained in our study was consistent with the literature [39]. With respect to polysubstance use, our prevalence rate of 46.3% was higher compared to rates (30%) found by another study [40]. This could be explained by the fact that the latter study only included patients who misused two or more substances, whereas in our study, patients who used two or more substances were also included. One study [41] suggested that cannabis use can be regarded as a proxy-measure for the use of other drugs. This notion appears to be confirmed in our study, as there were only two patients who used cannabis but none of the other substances.

To the best of our knowledge, no previous studies evaluated the relationship between tobacco use (smoking) and perceived loneliness in patients with a psychotic disorder. Smoking was associated with less perceived loneliness. The direction of this association remains unclear. Based on findings in the general population, where three large studies [42–44] concluded that smoking increased self-reported feelings of loneliness and social isolation, this finding was contrary to our expectations.

One could argue that assessments made by clinicians may differ from those self-reported by patients with psychosis. Indeed, a study indicated that elevated levels of perceived loneliness in individuals with psychosis were not dependent on objective social isolation [12], challenging the assumption that perceived loneliness is solely a consequence of limited social contact in psychosis. In that light, it is of interest to explore in future psychosis studies the relationship between substance use and both self-reported and clinician-rated aspects of social functioning. Furthermore, it is widely recognized that patients with psychosis experience increased feelings of loneliness compared to the general population [45]. For patients experiencing psychosis, smoking was perceived as a way to find relief from loneliness [46, 47] – even though this may not be associated with a reduction in objective social isolation.

To the best of our knowledge, no other studies evaluated the relationship between cannabis and self-reported social participation. The existing literature utilizing clinician-rated scales on social aspects reported conflicting findings. For instance, a study [48] demonstrated that patients without cannabis use disorder showed significant improvements in social functioning over time. This aligns with two other studies reporting that continued cannabis correlated with poorer social functioning [49–51], while others concluded that cannabis use was linked to enhanced global social functioning in individuals with an increased risk for psychosis [52]. Given these conflicting results, future studies could investigate the relationship between cannabis use and both self-reported and clinician-rated aspects of social functioning.

An explanation for the positive relationship between cannabis and more self-reported social participation in our study could also be related to potential differences in the patients’ profile: it might be that cannabis use may precipitate psychosis in individuals who would otherwise not transit to a psychotic disorder [53], compared to patients who converted without using cannabis [54]. This explanation is in line with the vulnerability-stress model, in which cannabis use is a risk factor to pass the threshold and develop a psychotic disorder [55]. Hence, those with cannabis use and a psychotic disorder might be better socially functioning with less severe cognitive impairments than non-cannabis-using patients, who are thought to be more vulnerable due to genetic factors [56]. Another explanation is that cannabis use occurs within a bidirectional social-cannabis context, wherein the craving for cannabis adds active engagement with a specific peer group mutually reinforced each other, resulting in increased cannabis consumption.

With respect to alcohol, we found that problematic alcohol use was associated with higher scores on perceived social support, which was at odds with another study [14], where patients reported more alcohol associated-problems in interpersonal and family relationships compared to non-problematic users. This is in line with Cetty et al [57], who found that hazardous alcohol use predicted lower self-reported scores on social relationships in patients with a first-episode psychosis. Differences in the exact outcome scores (perceived social support vs. the quality of relationships) might play a role in these divergent findings.

It might be that patients with increased social interaction are more at risk for problematic alcohol use as they may go along with individuals with problematic alcohol use. Some support for this explanation comes from a recent study conducted in the general population that found that adolescents with less social interest used alcohol less frequently [58]. Furthermore, a number of psychosis studies have found that alcohol use, including binge drinking [59] and hazardous drinking [57], was associated with lower severity of negative symptoms [60–62]. It has been suggested that impaired rewarding systems in patients with severe negative symptoms could be associated with diminished reinforcing effects of alcohol use [61]. Given the overlap between lower negative symptom severity and higher social support as an outcome in the current study, this could play a role in explaining our findings.

We did not find any significant relationship between any of the substances and perceived stigmatization. It is, nonetheless, important to acknowledge that in one study, patients reported to smoke to relief feelings of stigma by trying to “fit in” [46]. Correspondingly, smoking cessation was associated with fear of being socially ostracized when quitting smoking [63]. However, in the past decades, smoking actually became more associated with stigma [64]. Furthermore, previous literature found that especially those who experience disapproval by others over their inability to give up tobacco feel stigmatized [64]. Patients with a psychotic disorder can be seen as harder-to-treat smokers [65]. Therefore, efforts should be directed toward assisting patients in smoking cessation, not least because quitting has also been associated with reductions in public stigma and discrimination [66] along with decreases of depression, anxiety, and stress [67].

The strengths of the current study are the relatively large sample, representative of patients with psychotic disorders living in the community, and the evaluation of different substances on several self-reported aspects of social functioning, allowing for an in-depth examination of separate relationships between substance use and self-reported aspects of social functioning in patients with a psychotic disorder. However, the current study has several limitations. First, the cross-sectional design of the current study lacks the capability of evaluating the potential causal effects of substance use behavior in relation to several self-reported aspects of social functioning. Second, reverse causality and residual confounding cannot be ruled out due to the cross-sectional analyses and observational design. Third, information concerning the use of medication was missing. Fourth, given the explorative design of the current study, we chose to set the statistical significance at p < 0.05 instead of using a more conservative approach as Bonferroni correction for multiple testing. However, by this decision, we introduced the risk of a type-I error. Fifth, the majority of the study population was male (68.7%) and lower educated (64.9%), although these characteristics are probably representative for the group of patients with longstanding psychotic disorders. Sixth, the assessment of frequency and severity of substance use was based on self-report. Hence, under- or overestimation of substance use cannot be ruled out. Seventh, we did not correct for polysubstance use in our main analyses as dissecting out individual substance use would have drastically decreased the sample size. Nevertheless, we aimed to explore the effects of comorbid use by analyzing no users, single users, and polysubstance users. Finally, the outcome “social participation” was based on one subscale of the SEI-HS that included only four items. This could potentially have led to low content validity.

In the current study, smoking was associated with less perceived loneliness. As the negative health consequences of tobacco are disastrous, patients should be encouraged to stop smoking. As smoking may be a maladaptive coping strategy, it is beneficial to the patient to replace smoking with healthier coping skills. Referring the patient for psychological and behavioral treatment is required.

Therapists should be alert on the possibility that loneliness may be an issue and that individuals who try to stop smoking may need support in this respect. In the Netherlands, referral to a certified coach for psychological smoking cessation interventions is covered by the health insurance and should be offered more consequently to patients with psychosis.

In the current sample, psychotic patients who use cannabis reported betters scores on social participation. Consequently, for some patients, it might be that cannabis use is considered helpful in facilitating social interactions. However, the overwhelming (long-term) negative effects of cannabis have been documented [68], and cannabis reduction had favorable effects on symptoms and social functioning [69]. Therefore, it is important to help patients in finding education, work, or other structured activities, as these are crucial elements for enabling a reduction in cannabis use and, at the same time, finding alternative social participation [70, 71].

Concerning current findings of the association between (heavy) alcohol use and self-reported aspects of social functioning in patients with a psychotic disorder, longitudinal studies are needed to explore causality.

In the current study, polysubstance users reported less loneliness compared to patients who did not use drugs. A review that explored the reasons for polysubstance use [72] concluded that substances are generally combined to improve experiences and to increase/prolong a state of euphoria. It might be that negative feelings of loneliness are masked by a “high” state caused by polysubstance use. Nonetheless, psychosis risk was found to be associated with the cumulative effect of polysubstance use [41]. What is more, polysubstance use is also related to more psychiatric and psychosocial problems [73]. In future research, it might be interesting to explore other social aspects (for example functional or symptomatic recovery) in polysubstance users compared to single users and no users.