Main findings
This study presents a detailed examination of symptom-level associations between common mental disorders (CMD), substance use and severity of dependence. To our knowledge, such analyses are rare [
26], and offer novel information to validate theories about comorbidity and psychopathological processes. Consistent with the wider literature cited above, we found significant and moderate correlations between psychopathology and severity of dependence in a clinical sample of drug and alcohol users, although these were distinct and separate dimensions. Factor analyses indicated that depression and anxiety symptoms loaded onto a single underlying dimension, which converges with prior studies [
51,
52]. Substance use and severity of dependence were strongly associated and mapped onto a single dimension. After controlling for patients’ level of psychopathology represented by their CMD factor scores, we found evidence of statistically significant albeit small correlations between substance use and psychological distress at the level of individual symptoms. In what follows, we describe these patterns of associations and offer a theoretical interpretation of the findings.
The most prominent association pattern was that between severity of dependence (SDS) with depressive rumination, and its negative correlation with feelings of anxiety, uncontrollable worry, restlessness and inability to relax. This inverse pattern may explain the unexpected residual negative correlations between depression and anxiety symptoms (after partialling out their shared psychopathology). We note that these residual anxiety symptoms are similar to the phenomenology of craving and withdrawal [
8,
9]. Further supporting evidence was found in the partial correlations (controlling for psychopathology) between worry about missing a dose (of substance use) and psychomotor agitation, impaired control over substance use and inability to relax, and the finding that heavier heroin and crack users were less prone to worry. It is plausible that more severely dependent respondents were compulsively using substances in a way that mitigated such feelings of restlessness and craving, but this ultimately resulted in negative bouts of rumination, which in turn exacerbated wider aspects of depressive mood.
This deduction from our findings fits within a wider body of research. For example, Franken et al. [
53] carried out a factor analysis of two opiate craving questionnaires in a sample of 102 addiction service users and described a 3-factor solution covering aspects of ‘desire and intention to use’ , ‘negative reinforcement’ and ‘control’. The strongest inter-factor correlation (
r = .39) indicated that ‘desire to use’ substances was driven by a need to suppress worry, tension and thoughts about life problems (‘negative reinforcement’). Furthermore, items from the ‘control’ factor were negatively associated with items from the other 2 domains. Craving is also known to be associated with biased attention to environmental cues related to substance use [
54], which may partly explain the experience of uncontrollable worry (about substance use). On the other hand, perseverant negative thinking (rumination) is often associated with depression [
55‐
57] and has been described as a transdiagnostic factor underlying several mental disorders [
58]. Rumination has been found to predict the onset of depressive symptoms in non-clinical samples [
59‐
61], and it appears to maintain symptoms of low mood and anxiety in depressed patients [
62].
Moreover, we found that two of the substance use measures (heroin and alcohol) were correlated with loss of control over substance use. This is consistent with research indicating that alcohol impairs inhibitory control [
63,
64], although this is less evident for opiates [
64] and instead may reflect a psychological aspect of lower perceived control over heroin use. Cannabis use was correlated with impaired concentration, which reflects the expected psychoactive effects of this substance [
65]. Although robust quantitative research for cannabis-induced neurocognitive deficits is still scarce, there is some evidence that cannabis use impairs the ability to learn and recall new information [
66].
A particularly interesting pattern of non-linear associations were found for alcohol use (Fig.
2). Moderate drinkers had lower mean scores for feelings of irritability compared to non-drinkers and heavy/severe users. In addition, light drinkers reported lower levels of perceived difficulty in abstaining compared to non-drinkers and moderate to severe drinkers. Non-linear associations between alcohol and psychological distress have been reported in numerous studies [
50,
67‐
72]. These studies model associations based on aggregated scores, which may mask more specific symptom-level patterns. The ‘stress buffer’ theory [
73] seems like a plausible explanation for our findings, suggesting that a moderate dose of alcohol use may mitigate feelings of psychological distress [
74]. Still, at higher doses, alcohol may provide less protection from irritability once inhibitory control declines, often giving way to overt aggression particularly in men with heightened irritability [
75]. An alternative explanation may be that the findings are confounded by other sample characteristics [
76], for instance more well-adjusted respondents may cluster in the ‘moderate group’. This latter explanation seems less probable, given that moderate drinkers seem less resilient in their efforts to abstain from alcohol use (as shown in Fig.
1, panel
a).
Strengths and limitations
An important limitation is the relatively small sample size by comparison to epidemiological studies in this area. Our sample size was adequately powered to undertake factor analysis based on Bryant and Yarnold’s criteria [
77], which would require 5 respondents per item (
N = 130). Still, we note that there are divergent views about sample size calculations for factor analysis, and others suggest a minimum of 500 participants [
78]. Other considerations to note about the generalizability of these findings concern the outpatient setting, with a majority of respondents whose primary reason for treatment related to opiate use. Although nearly half of all respondents reported using alcohol, and some at a very severe level, we noted that alcohol use was weakly correlated with only 2 SDS items, and did not load onto factor 2. It may be that including more participants from alcohol detox or inpatient settings could render different patterns of correlation. It is also possible that the method of administration of the SDS measure may have influenced the strength of correlations. SDS was rated for the primary substance of concern, which in some cases could have been a substance other than alcohol (e.g., heroin), and this may have therefore impacted on the strength of observed correlations between alcohol use and SDS. Finally, we also note that the cross-sectional nature of these data limit the possibility of making more certain claims about casual relationships. Our deductions from this sample should therefore be taken as a preliminary investigation of functional links between aspects of psychopathology and addictive behaviours, awaiting further validation in prospective studies with mediational tests.
Considerations for practice and research
The emerging literature on comorbidity suggests that substance induced depression and anxiety symptoms are relatively uncommon, though they may be more conspicuous in addiction treatment settings. Our findings show that after factoring out general psychological distress, essentially no specific covariance between substance use patterns and psychological distress items remains. This is strong evidence that general psychological distress is a moderator of the relationship, but it is unclear from the current study in which direction the causal arrow points.
From this perspective, the common practice of assuming that most CMD symptoms are drug-induced is ethically questionable, especially if such practices hamper timely diagnosis and access to mental healthcare. Psychometrically adequate screening methods are available to detect CMD in substance users [
27,
41,
79,
80], but these methods are not consistently implemented in routine practice. Based on our symptom-level correlations, we propose that diagnostic results may perhaps be enhanced by applying a repeated screening method after a month of watchful wait for heavy cannabis users who show signs of severe disruption to concentration, and heavy drinkers who show increased signs of irritability and hyper-arousal. Residual anxiety symptoms that may reflect craving/withdrawal phenomena should be carefully distinguished from generalised anxiety disorder, possibly by supplementing screening measures with probing questions or interviews. The training of addiction treatment professionals [
81,
82] in the application of such screening methods may be an important focus of future dissemination studies and policies.