Responses to social evaluative stress in regular cannabis smokers

Abstract

Background:

Aspects of the canonical stress response differ in stimulant, opioid, and alcohol users relative to controls, and dysregulated responses to stress may contribute to continued use of these drugs. Little prior research has focused on stress responses in regular cannabis smokers. We assessed responses to a standardized laboratory social stress assay (the Trier Social Stress Task; TSST) in regular cannabis smokers (CANs) compared with controls (CONs).

Methods:

Healthy, non-treatment-seeking adult CANs (⩾4×/week; smoking cannabis as usual) and demographically matched CONs completed the TSST. Outcome measures were subjective mood, heart rate, and salivary cortisol.

Results:

Nineteen CANs (1 female) and 20 CONs (2 female) participated; groups were matched on trauma exposure, sex, race, and age. CANs smoked cannabis 6.4 ± 1.1 days/week. Eight CANs and one CON smoked tobacco cigarettes daily. Overall, the TSST produced expected increases in anxiety, negative mood states, cortisol, and heart rate. CANs had blunted subjective response to stress relative to CONs, but they did not differ in physiological (cortisol and cardiovascular) stress responding.

Conclusion:

These results indicate that CANs have blunted mood responses to social stress, but normative physiological stress responding. Observed differences could be due to residual effects of cannabis, reluctance to endorse negative mood states, or to issues related to identifying (i.e., emotional identification) or feeling (i.e., interoception) stress-related affective states. Further research is warranted to characterize the mechanisms of these differences and assess implications for daily functioning and treatment outcomes.

Introduction

Cannabis is one of the most widely used drugs in the United States (SAMSHA, 2017) and internationally (UNODC, 2018). Past-year prevalence of adult cannabis use in the United States increased from 4.1% in 2001 to 2002 to 9.5% in 2012 to 2013, with a parallel increase in the prevalence of DSM-V cannabis use disorder (Hasin, 2018). Only a subset of cannabis users experience problems with their use. However, demand for treatment for cannabis-related problems is likely to increase in the coming years, as use among adults is expected to continue to grow with changes in the legal status of cannabis and diminishing perceptions of cannabis-related harm (Hasin, 2018). Given the paucity of efficacious interventions for cannabis use disorder (Haney et al., 2019), research focused on the psychological, behavioral, and physiological mechanisms maintaining problematic use of cannabis is needed to inform development of new treatment approaches. One area of interest is stress responding, given that cannabis use is cited as a way to cope with stress in many cannabis smokers (Hyman and Sinha, 2009).

Stressful situations in the environment trigger a characteristic cascade of physiological responses in the human body, including rapid release of corticotropin-releasing hormone (CRH), adrenocorticotropic hormone, and cortisol (CORT). Higher circulating levels of stress-related hormones in turn lead to increased heart rate (HR) and breathing rate, muscle tension, and sweating (McEwen, 2007). In combination, these components of the stress response serve to prepare the organism for rapid reactions to dangers in the environment (i.e., fight, flight, or freeze). Importantly, in humans this response may be triggered by psychosocial stressors (e.g., rejection or negative social evaluation) as well as environmental threats (Kirschbaum et al., 1993). Optimally, stress responding involves the body mounting an adequate acute response (i.e., neither blunted nor exaggerated) to manage a stressor, with rapid post-stressor recovery to the baseline physiological state (McEwen, 2008).

Several converging lines of evidence suggest that stress and stress responding are involved in the development and maintenance of substance use disorders (SUDs). First, in laboratory animals trained to self-administer drugs of abuse in whom this behavior has been extinguished, exposure to stressors such as mild electric shocks reliably reinstates drug-seeking behavior (Stewart, 2000). Similarly, in abstinent human drug users, stressful events are thought to precipitate relapse (Sinha, 2008). Second, many drugs of abuse produce physiological changes that are similar to those that occur during stress. The similarity between drug effects and the effects of stress exposure suggest that, in some people, stress may trigger drug use by mimicking drug effects and acting as a reminder of drug use (Van Hedger et al., 2017). Third, individuals with SUDs have been found to have dysregulated stress responding relative to healthy populations, potentially contributing to drug use as a way of coping with stress (Sinha, 2008).

Although research on stress responding in cannabis users has been limited, atypical stress system functioning has been documented in users of other drugs. Cocaine-dependent individuals show exaggerated subjective responses relative to healthy controls following intravenous exposure to a hormonal stressor (CRH; Brady et al., 2009) as well as a laboratory social stress assay (Moran-Santa Maria et al., 2010). Consistent with this, cocaine dependence is also associated with more negative perceptions of daily life stressors (Back et al., 2008a). Adolescent/young adult methamphetamine users exhibit elevated salivary CORT responses to social stress compared with age-matched controls (King et al., 2010). Dependent prescription opioid users show heightened subjective but not physiological stress responding following an acute social stressor compared with controls (Back et al., 2015). By contrast, recently detoxified alcohol-dependent patients exhibit decreased hormonal stress responding relative to healthy controls after exposure to similar stress (Muehlhan et al., 2017, 2018).

There is also some initial evidence for dysregulated stress responding associated with cannabis use. In adolescents, both lifetime and past-year (⩾5 times) cannabis use—relative to no use—is associated with blunted salivary CORT responses to psychosocial stress, with past-year (⩾5 times) users also showing reduced CORT responding compared with adolescents reporting lifetime use <5 times (van Leeuwen et al., 2011). To date, only one prior study to our knowledge has directly assessed stress reactivity in adult cannabis users relative to non-users, reporting that cannabis users had blunted CORT and subjective responses following an acute stressor comprising both physiological (placing hand in an ice bath) and psychosocial (social evaluation) components (Cuttler et al., 2017). These findings are broadly consistent with the earlier study of stress responding in adolescent cannabis users (van Leeuwen et al., 2011). However, in a recent paper by our group, we reported that the extent of past trauma exposure was associated with heightened CORT and anxiety responses to stress among cannabis smokers (Chao et al., 2018). This indicates that trauma exposure should be explicitly considered in investigations of stress reactivity as a function of cannabis use. Moreover, the study by Cuttler et al. (2017) did not assess for psychiatric diagnoses, only excluding participants who reported having been diagnosed or treated for a psychiatric disorder. Given the high prevalence of stress and anxiety-related conditions among cannabis smokers (Kedzior and Laeber, 2014)—many of which may be undiagnosed—and the known impact of psychiatric conditions on stress responding (Zorn et al., 2017), assessment of stress responses in cannabis users formally assessed to ensure they do not have comorbid psychiatric conditions is needed.

In this study, we examined affective and physiological stress responding in regular (⩾2 cannabis cigarettes/day; ⩾4 days/week in the past month) adult cannabis smokers without comorbid psychiatric disorders relative to controls carefully matched for trauma exposure and demographics. We hypothesized that cannabis users would show blunted stress reactivity compared with controls. Given previous evidence that stress may exacerbate craving for cannabis (McRae-Clark et al., 2011), we also investigated the effects of stress on cannabis craving.

Methods

Participants

Participants were non-treatment-seeking, 21- to 50-year-old male and non-pregnant female cannabis users and demographically matched controls. The cannabis group (CAN) comprised individuals who reported regular cannabis use (at a minimum, an average of ⩾2 cannabis cigarettes/day, smoked ⩾4 days/week during the past month). Regular cannabis use was biochemically verified via positive tetrahydrocannabinol (THC) urine toxicology (5-panel drug screen, All Test North American, Gilbert, AZ) at all visits. Estimates of cannabis use were based on a rate of 1 “blunt” (i.e. hollow cigar filled with cannabis)= 2 cannabis cigarettes (Mariani et al., 2011). CANs could not report regular (>2×/week) use of illicit drugs (e.g., cocaine, ecstasy) or having experienced prior adverse cannabis effects. Controls (CONs) were excluded if they reported use of non-cannabis illicit drugs (>2×/week). Given the high rates of cannabis experimentation in the general population, we allowed CON participants to report up to 10 lifetime cannabis use occasions, with no use in the past year. Absence of current cannabis and other drug use in CONs was verified with negative urine toxicology at all visits.

All participants had to be fluent in English. Females were non-pregnant and using effective contraception. Exclusion criteria were (1) current DSM-IV major depressive episode or generalized anxiety disorder or lifetime panic disorder, obsessive–compulsive disorder, anorexia nervosa, manic episode, or psychotic illness (APA, 1994); (2) daily use of prescription or over-the-counter medications; (3) current medical condition contradicting participation; and (4) current parole or probation. To comply with the requirements of the broader study (see experimental protocol), all participants were right-handed and had no magnetic resonance imaging (MRI) contraindication, and CAN participants were excluded for clinically significant dietary restraint and use of non-smoked cannabis administration methods (i.e., oral or vaporized). All participants provided written informed consent, as approved by the New York State Psychiatric Institute Institutional Review Board.

Experimental protocol

Volunteers completed three to five screening visits, undergoing physical examination and psychiatric assessment (structured clinical interview for DSM-IV; First et al., 2002), electrocardiogram, and laboratory blood and urine chemistry tests. A self-report battery was administered to assess demographics, physical health and drug use history, and symptoms of anxiety (State–Trait Inventory of Cognitive and Somatic Anxiety; Grös et al., 2007; Van Dam et al., 2013) and depression (Beck Depression Inventory-II; Beck et al., 1996). A doctoral-level researcher assessed substance use using a structured interview.

This study was a sub-component of a larger investigation focused on drug-related choice in cannabis smokers. In addition to participating in the outpatient session described in the Trier Social Stress Task section, CAN participants underwent a training session followed by a 6-day inpatient stay involving MRI, behavioral testing, and cannabis administration. CON participants only completed the outpatient session. Data from the MRI, behavioral, and cannabis administration components of the larger study address different hypotheses and will be presented separately.

All participants underwent a social stressor during a single outpatient session, scheduled in the afternoon to control for circadian variations in CORT. Of note, CAN participants were not instructed to alter their cannabis smoking prior to attendance, because the focus of this investigation was on stress reactivity in cannabis users’ “normal” day-to-day state (as opposed to during acute intoxication or withdrawal). At arrival, urine toxicology and breathalyzer (Select S80, BACTrack, San Francisco, CA) measurements were used to verify recent abstinence from substances other than cannabis (no participant tested positive for breath alcohol at the session). Females underwent urine pregnancy testing (Alere hCG Dipstick, Alere, Orlando, FL). Both groups completed the social stress task (described in the Trier Social Stress Task section); subsequently CONs underwent a brief MRI scan (data to be presented elsewhere; CAN participants completed this scan during their inpatient stay). All participants were compensated for their time upon completion of the outpatient session.

Trier Social Stress Task

The Trier Social Stress Task (TSST) is a standardized laboratory assay of social stress responding (Kirschbaum et al., 1993) that has been shown to effectively and safely produce acute transient increases in stress and associated physiological and subjective states in a wide range of populations (Allen et al., 2017). Prior to the stressor, participants are asked to nominate a job that they would like. During the set-up phase for the TSST, baseline measurements are collected (see Assessments). An HR monitor continuously recording the participant’s HR is fitted (Polar RS400 Sports Watch, Oulu, Finland). The 20-min stress phase is then initiated. The acute stressor consists of (1) Introduction: the participant is taken into a testing room in which two “committee members” (confederates) are seated at a table wearing white coats. On the table is a video camera and a voice recorder. The research assistant explains that the participant will be asked to prepare a job interview speech based on their previously nominated job for the committee, who will evaluate them. Participants are told they will be recorded by the voice and video recorders, with data used for behavioral and vocal analysis of signs of stress. Participants are informed that following the speech they will be given a second task by the committee. They are offered an opportunity to ask questions; (2) Preparation: Participants are led to a separate room, where they are left alone to prepare their speech for 10 min; (3) Speech: Each participant is led back to the committee room where they are asked to present their speech (without notes) for 5 min. During their speech, the committee members provide minimal feedback and no verbal or non-verbal encouragement; and (4) Mental arithmetic: Following the speech, the committee informs the participant of the next task, which is a mental arithmetic activity involving serial subtraction of odd numbers from a large starting number (e.g., 17 from 2743, then 17 from 2726). This task continues for 5-min, with the committee informing the participant when they are incorrect and asking them to restart the exercise. Following the TSST, participants stay in the laboratory for 90 min to complete the assessments.

Assessments

Subjective variables.

Mood state was measured via the Profile of Mood States (POMS; McNair and Droppleman, 1971), the state component of the State–Trait Anxiety Inventory (STAI; Spielberger et al., 1983), and Visual Analog Scale (VAS; Folstein and Luria, 1973) questionnaires at baseline, immediately after the TSST (Time 0), then at 15, 30, 60, and 90 min. The 72-item POMS employed yields nine affective subscores: Anxiety, Depression, Anger, Fatigue, Confusion, Vigor, Friendliness, Elation, Arousal. The STAI produces a single state score summed across 20 items reflecting current state anxiety. VAS items consisted of the following descriptors: I feel … upset, nervous, stressed, confident, angered, craving for cannabis with “Not at all” and “Extremely” as anchors.

Cortisol stress response.

Salivary CORT was measured using Salivettes polyester swabs (Sarstedt, Germany). Saliva was collected twice at baseline (the second sample was used in analyses to account for novelty effects) and at, 0, 15, 30, 60, and 90 min after the stressor. Samples were stored at −25°C and assayed by the Analytical Psychopharmacology Laboratory (PI: Thomas A. Cooper) in the Nathan Kline Institute, Orangeburg, NY (Chao et al., 2018).

Cardiovascular stress response.

HR was measured continuously throughout the session using the Polar RS400 Sports Watch (Oulu, Finland), with mean beats per minute calculated for the following epochs: baseline (approx. −35 to −21 min), introduction/preparation (−20 to −11 min), public speaking (−10 to −6 min), arithmetic (−5 to 0 min), and recovery (0–14 min, 15–29 min, 30–44 min, 45–59 min, 60–74 min, and 75–89 min).

Statistical analyses

The main outcomes were subjective (STAI, VAS, POMS) and physiological (CORT, HR) variables. CORT data from one CON were of insufficient quantity for analysis and were therefore excluded. HR data from one CAN and two CONs were lost due to equipment failure. Outliers with absolute z-scores greater than 3.29 were truncated to one increment more or less than the closest non-outlier value. Pre-truncated data were retained if there were no differences in outcome compared with those obtained from truncated data. A single discrepancy in outcome obtained using truncated versus original data was observed for the VAS Stressed variable. Specifically, the interaction between group and time reached statistical significance only in the truncated analyses. Five and one VAS Stressed data points were truncated from CANs and CONs, respectively. All other data were retained in their original form.

Demographics, psychiatric symptoms, and substance use were compared between the CAN and CON groups using chi-square and independent samples t-tests. The overall effects of the TSST on the outcome variables and group differences were analyzed with mixed between–within group analyses of variance (ANOVAs), with group (CAN/CON) as the between-subject factor and time as the within-subject factor, where an interaction between group and time would indicate a differential response to the stressor between the two groups. Interactions were followed up with independent t-tests comparing groups at each time point. The alpha level was set at .05. We used Greenhouse–Geisser corrected degrees of freedom where Mauchley’s test of sphericity indicated a significant departure from sphericity (p < .05). Effect sizes are presented as partial η². Analyses were conducted using IBM SPSS Statistics 23 (IBM, Armonk, NY).

Results

Participants

Table 1 shows the demographics of participants with TSST data (N = 39). Data from 3 CANs were included in a previous manuscript (Chao et al., 2018). Groups were well-matched demographically, with the exception that CANs reported fewer years of education than CONs. CANs also contained more current daily cigarette smokers than the CON group, consistent with the high comorbidity between cigarette and cannabis smoking (Haney et al., 2013). CANs were also more likely to report isolated incidents (i.e., 1–2 times in their lifetime) of use of drugs other than cannabis (see Table 1).

Table 1.

Demographic characteristics.

	CAN (N = 19)	CON (N = 20)
	n (%)	n (%)
Sex, female	1 (5)	2 (10)
Ethnicity, non-Hispanic	16 (84)	15 (75)
Race, AA	14 (74)	13 (65)
Race, Caucasian	2 (11)	3 (15)
Race, other	3 (16)	4 (20)
	Mean (SD)	Mean (SD)
Age	28.8 (5.3)	29.3 (4.8)
Education, years	12.7 (1.5)*	14.8 (2.1)*
Trauma exposure: TAA	1.2 (1.3)	0.7 (1.4)
Depression Sx: BDI	2.6 (4.0)	0.7 (1.4)
Anxiety Sx: STICSA	24.9 (5.9)	23.2 (3.8)
Cannabis use
Days/week	6.4 (1.1)	-
Cannabis cigarettes/day	7.3 (7.6)	-
Cost/week (USD)	100 (68.0)	-
Age of 1st use	15.4 (2.4)	-
Age of daily use	18.3 (4.2)	-
Years of regular use	10.6 (5.0)	-
Alcohol use
⩾weekly drinkers (n)	7 (37)	8 (40)
Occasions/weeka	1.5 (0.6)	1.6 (0.6)
Drinks/occasiona	2.9 (1.3)	2.9 (1.0)
Cigarette use
Daily smokers (n)	8 (42)**	1 (5)**
Days/weeka	7.0 (0.0)	7
Cigs/daya	6.6 (5.2)	4
Other substance use
Lifetime cocaine use (n)	2	0
Lifetime opioid use (n)	8	5
Lifetime meth/amphetamine use (n)	5	0*
Lifetime hallucinogen use (n)	3	0

Note: Data presented as frequency (%) and mean (SD). AA: African American; BDI: Beck Depression Inventory; STICSA: State–Trait Inventory of Cognitive and Somatic Anxiety; Sx: symptoms; TAA: Trauma Assessment for Adults; USD: United States dollars.

^*p < .05.,

^**p < .01.

^aData from those who reported regular (⩾weekly) alcohol use or daily cigarette smoking.

– not applicable.

Stress responding overall

As expected, the TSST increased indicators of stress in participants combined. Main effects of time were seen across all subjective variables (see Table 2), except POMS Fatigue. STAI state anxiety increased during the stress phase, peaked immediately after, and decreased close to baseline, remaining somewhat elevated. POMS Anxiety, Depression, Anger and Confusion, as well as VAS Upset, Nervous, Stressed, and Angered, increased and peaked immediately after the TSST, returning to baseline soon after. POMS Vigor and Arousal declined gradually from baseline over the session, whereas POMS Friendliness and Elation and VAS Confident dropped sharply after the TSST. Of these three variables, only VAS Confident returned close to baseline by the end of the session.

Table 2.

Subjective mood results.

	Time effect			Group effect			Interaction
	df	F	p	df	F	p	df	F	p
STAI	3.0, 97.1	9.5	<.0001	1, 32	0.0	.884	3.0, 97.1	1.6	.185
POMS
Anxiety	2.4, 89.1	16.3	<.0001	1, 37	3.8	.058	2.4, 89.1	2.0	.130
Depression	2.5, 92.4	5.5	.003	1, 37	7.2	.191	2.5, 92.4	2.9	.047
Anger	1.8, 65.9	6.1	.005	1, 37	1.2	.286	1.8, 65.9	0.5	.581
Vigor	3.0, 110.0	5.7	.001	1, 37	0.1	.732	3.0, 110.0	0.4	.747
Fatigue	3.1, 116.4	1.1	.347	1, 37	4.9	.033	3.1, 116.4	0.7	.555
Confusion	2.9, 110.7	7.8	<.0001	1, 37	2.9	.098	3.0, 110.7	2.2	.089
Friendliness	3.2, 117.9	10.3	<.0001	1, 37	0.7	.398	3.2, 117.9	1.4	.244
Elation	3.1, 114.6	6.7	<.001	1, 37	0.4	.511	3.1, 114.6	0.2	.927
Arousal	3.3, 120.6	5.0	.002	1, 37	0.6	.461	3.3, 120.6	0.3	.829
VAS
Upset	2.3, 81.0	7.6	.001	1, 35	1.8	.188	2.3, 81.0	1.0	.372
Nervous	2.0, 70.6	4.7	.012	1, 35	0.0	.853	2.0, 70.6	3.2	.047
Stresseda	1.5, 51.6	6.3	.007	1, 35	3.3	.079	1.5, 51.6	3.6	.048
Confident	2.6, 89.7	4.4	.009	1, 35	0.8	.375	2.6, 89.7	1.4	.261
Angered	2.0, 68.8	6.3	.003	1, 35	1.3	.260	2.0, 68.8	0.4	.657

Note: Results from mixed ANOVA on subjective outcome measures. POMS: Profile of Mood States; STAI: State–Trait Anxiety Inventory—State; VAS: visual analogue scale.

^aResults from truncated data.

There was a main effect of time on CORT (F(1.8, 60.0) = 9.7, p < .001, partial η² = .22), with an increase from baseline during stress, peaking at 15 min and steadily declining thereafter. There was a main effect of time on HR (F(3.6, 98.2) = 44.5, p < .001, partial η² = .62), such that HR increased from baseline, peaking during public speaking and decreasing to baseline after the stress phase.

Stress responding as a function of group

There was an interactive effect of group and time on VAS Nervous (F(2.0, 70.6) = 3.2, p = .047, partial η² = .08) and VAS Stressed F(1.5, 51.6) = 3.6, p = .048, partial η² = .09), such that self-rated nervousness and stress were both significantly higher immediately after the stressor in CONs than in CANs (see Figure 1). There was also an interaction of group and time on POMS Depression (F(2.5, 92.4) = 2.9, p = .047, partial η² =.07), such that depression increased immediately after the TSST in CONs relative to CANs. There was a main effect of group on POMS Fatigue (F(1, 37) = 4.9, p = .033, partial η² = .12) with CONs reporting higher levels of fatigue overall relative to cannabis smokers (Figure 1). There were no other main effects of group, or interaction effects on the subjective variables.

Figure 1.

Subjective stress responding as a function of group. Cannabis users had blunted subjective stress responses relative to controls. Data are means ± SEMs. Asterisks indicate statistically significant difference between groups at individual time points.

There were no interactions or main effects of group on CORT or HR (see Figure 2).

Figure 2.

Physiological stress responding. As expected, salivary cortisol and heart rate increased as a function of stress in the overall group, with no preferential effects by group. Data are means ± SEMs.

Cannabis craving as a function of stress exposure

There was a main effect of group (F(1, 35) = 75.7, p < .001, partial η² = .68) but not time, on reported VAS cannabis craving, with CANs reporting greater overall craving relative to CONs. There was no interactive effect of group and time on craving, indicating that CANs did not increase their cannabis craving reports as a function of stress.

Discussion

This study examined subjective and physiological responding to a standardized social stress assay in regular cannabis users relative to controls. Overall, the TSST produced expected increases in HR, CORT levels, anxiety, and other negative mood states. Cannabis users showed blunted subjective responses to stress compared with controls, with controls reporting elevations in nervousness, subjective stress, and depressed mood immediately after the social stressor—increases in negative affect that were largely absent in cannabis users. By contrast, salivary CORT and HR responding in the cannabis users was similar to that of the healthy controls. Finally, there was no apparent increase in cannabis craving following the stressor in cannabis users.

These findings concur to some extent with the limited existing evidence on stress responding in cannabis users. The only prior study on social stress responding in adult cannabis users compared with controls also documented blunted subjective stress responding as a function of cannabis use (Cuttler et al., 2017). Conversely, that study as well as one in adolescent cannabis users reported blunted salivary CORT responses to stress in cannabis users compared with non-users (Cuttler et al., 2017; van Leeuwen et al., 2011), whereas we observed normative physiological (both CORT and HR) stress responding in cannabis-using adults. As noted previously, our study controlled for potential effects of trauma exposure and psychiatric conditions on stress responding by matching groups on trauma exposure and carefully assessing psychiatric status, excluding volunteers experiencing current depressive episodes or generalized anxiety disorder, as well as lifetime panic disorder, obsessive–compulsive disorder, post-traumatic stress disorder, anorexia nervosa, manic episode, or psychotic illness. It is thus possible that previous findings of abnormal CORT responses in cannabis users relative to controls (Cuttler et al., 2017; van Leeuwen et al., 2011) could be related to the effects of undiagnosed psychiatric illness or differential trauma exposure between groups. We also found no significant increase in cannabis craving following the TSST in cannabis users. In contrast, a previous study reported an increase in craving following the TSST compared with cannabis users undergoing a non-stress condition (McRae-Clark et al., 2011). As already noted, we investigated cannabis users in their normal state and did not require them to abstain from cannabis for set times prior to testing. It is possible that elevated cravings after the TSST in the aforementioned study (McRae-Clark et al., 2011) could have been because participants were instructed to abstain from cannabis on the testing day.

Previous evidence suggests that dysregulation of the stress system in substance users more broadly may differ in direction, with stimulant (King et al., 2010; Moran-Santa Maria et al., 2010) and opioid (Back et al., 2015) users showing exaggerated stress responding, while alcohol (Muehlhan et al., 2017, 2018) and cannabis users appear to have blunted stress responding relative to controls. These findings underline the likely complexity in the mechanisms of these differences, with various factors potentially contributing to observed differences including pre-existing factors, uncontrolled differences in psychiatric comorbidities between drug-using groups and controls, and acute, residual, and chronic drug effects. Moreover, effects associated with cigarette smoking, which is highly comorbid with use of alcohol, cannabis, and illicit substances, cannot be discounted in most studies, including our own, given that the cannabis group contained more current daily cigarette smokers than the controls. However, differences in stress responding as a function of cigarette smoking have been reported in women but not men (Back et al., 2008b), suggesting that this was unlikely to contribute to the current findings in primarily male participants.

To our knowledge, this is the first study to investigate acute social stress responding in regular adult cannabis users compared with controls while controlling for trauma exposure and psychiatric comorbidities. Consistent with our primary hypothesis, cannabis users exhibited blunted stress reactivity compared with controls, although this difference occurred only in subjective, not physiological, domains. As noted, mechanisms of this difference are likely to be complicated and causality cannot be ascertained based on the current design. Previous evidence indicates that acute administration of low doses of oral THC (7.5 mg but not 12.5 mg) to occasional cannabis smokers preferentially reduces subjective social stress responding relative to placebo (Childs et al., 2017). The oral administration route and occasional cannabis-using sample may limit the relevance of these findings to the present study. However, given that participants in our study were in their normal state, and were not asked to abstain from cannabis prior to sessions (although none presented as acutely intoxicated), these previous findings indicate that it is possible that the observed blunting of subjective stress responding was due to residual cannabis effects. Irrespective of causality, these results have functional implications, suggesting that regular cannabis users may not experience normative mood responses to stressors experienced during daily life, potentially impacting on their behavioral responses to environmental challenges. The combination of intact physiological stress responding combined with blunted subjective responses also suggests the possibility that cannabis users may have blunted interoceptive awareness of internal states, as has been suggested to be the case in users of stimulant drugs (Gowin et al., 2014).

As an early investigation of stress responding in cannabis users, this study has limitations. First, cannabis use history was based on self-report, thus it is impossible to know whether, for instance, controls had used more cannabis than reported. However, participants underwent extensive screening and produced urine tests consistent with their self-reports at each of the three to five visits, supporting the accuracy of their drug histories. Second, participants were not required to abstain from cannabis for any particular length of time prior to testing, meaning that direct and residual drug effects cannot be excluded as mechanisms of group differences. However, this study focused on characterizing stress responding in daily life in regular cannabis users (i.e., not intoxicated but also not in withdrawal), which has important clinical and functional implications irrespective of causality. Third, cannabis users were more likely to smoke cigarettes and report use of other drugs than were the controls (see Table 1). In relation to substances other than cannabis and cigarettes, this was lifetime use with limited reported occasions of use and no evidence of current use. Cigarette smoking showed more marked between-group differences, with eight cannabis users versus only one control reportedly smoking tobacco cigarettes daily. While limiting the degree to which causality can be attributed specifically to cannabis use, this is consistent with the high rates of concurrent tobacco smoking in cannabis users more broadly (Haney et al., 2013). Further research is needed to disentangle the complex and apparently clinically significant interactions between tobacco and cannabis smoking (Herrmann et al., 2019). Groups also differed by education levels, with cannabis users reporting fewer years of education than controls. To assess whether this difference had an impact on outcomes, we reanalyzed data including years of education as a covariate. A single effect on outcome was found for POMs Fatigue, such that the main effect of group on fatigue was no longer significant. All other outcomes remained unchanged, suggesting that differences in education did not underlie the main findings of this study. Fourth, the sample was disproportionately male. As a result, findings cannot speak to differences between male and female cannabis smokers, and may not be fully generalizable to female cannabis users. Given previous evidence (Chao et al., 2018), further assessment of stress responding as a function of both cannabis use and gender is warranted. Fifth, participants were non-treatment-seeking individuals without comorbid psychiatric or medical conditions. While allowing for more specificity in assessment of the relationship between cannabis use and stress responding, results may not be generalizable to treatment-seeking cannabis users. However, given that the majority of cannabis users do not seek treatment (Hughes et al., 2016), it is important to investigate this group. Sixth, we did not employ a placebo or no-stress condition, such that we cannot definitely attribute the changes in study endpoints over time to the stressor. However, the time course observed in these data is closely similar to that seen in other studies using the TSST including those that did use a no-stress condition (e.g., Childs et al., 2017), strengthening our interpretation that changes in subjective and physiological measures over the session were largely due to the stressor rather than time or another non-specific factor. Seventh, we did not assess the THC content of the cannabis habitually smoked by the CAN group. We considered this outside of the scope of the study given that the methodology employed did not allow us to infer causality for group differences to direct or residual drug effects. Eighth, we focused resources on recruiting well-matched controls and screening both groups comprehensively, rather than collecting large samples, and HR and salivary CORT data were lost due to technical issues for three and one participants respectively. This is likely to have reduced our ability to detect small effect sizes. Finally, because subjective mood states rely on self-report, it is not possible to conclusively determine whether the differences observed, which were limited to subjective domains, represent “true” differences in affective experiences or a preferential reluctance to endorse negative emotional states in cannabis users. However, groups were well-matched on demographic characteristics, minimizing the potential for systematic between-group differences in willingness to endorse certain mood states.

In conclusion, findings suggest that cannabis users have blunted mood responses to social stress relative to non-users, without differences in physiological stress responding. This study contributes to a growing body of evidence implicating atypical stress responding in the use of a range of drugs, including cannabis. Altered stress responding processes may have impacts on daily functioning and in the context of treatment. Further research elucidating the mechanisms, impact, and time course of atypical stress responding in cannabis users is warranted.