Background

Gambling is prevalent worldwide. While most people who gamble do so recreationally and do not experience significant negative personal consequences, some individuals develop symptoms of disordered gambling, which can progress into clinically defined gambling disorder (GD) [1]. GD is defined in the fifth edition of the Diagnostic and Statistical Manual (DSM-5) as a persistent and recurring pattern of gambling associated with substantial stress and functional impairment. GD is associated with negative long-term outcomes such as poorer health, relationship problems, poor work or school performance, financial issues, and suicidality. “Problem gambling” (PG) describes individuals who exhibit symptoms of disordered gambling that are not identified using formal diagnostic criteria or exist below the clinical threshold [2].

One critical aspect prevalent in GD is high stress, which contributes to the development and maintenance of GD; acute stress may increase gambling urges and gambling behaviour may be used to mitigate stress [3, 4]. Chronic stress may also increase susceptibility to maladaptive gambling and the rewarding value of gambling behaviours [5••, 6, 7]. Relapse likelihood may increase during stressful situations, making recovery more difficult [3]. Furthermore, stress can alter cognitive and neural function over time which could promote gambling fallacies such as the illusion of control [3, 4]. Certain forms of gambling, such as slot machines, rely on unpredictability and uncontrollability factors that may induce acute stress, such that the action of gambling itself can become a cue for stress, even for people who gamble recreationally [8].

Acute stress induces a physiological response mediated by two systems: the hypothalamic-pituitary-adrenal (HPA) axis, and the sympathetic nervous system (SNS) [9]. The HPA axis drives the response from several hormones including corticotropin-releasing hormone (CRH), adrenocorticotropin hormone (ACTH), and cortisol, while the SNS initiates cardiovascular activity to increase heart rate and blood pressure [9, 10]. Other important physiological markers of stress include epinephrine, norepinephrine, and alpha-amylase [10, 11]. The physiological acute stress response maintains homeostatic balance; however, overactivation due to chronic stress disrupts this system over time [7]. Since stress is both a risk factor for developing GD and is enhanced by the disorder, individuals who display maladaptive gambling behaviour likely experience altered physiological acute stress activity. Therefore, the purpose of this review is to examine literature that describes physiological stress activity and reactivity in individuals with PG and GD. We discuss research investigating basal stress markers, acute stress effects, stress physiology during risky decision-making, and therapeutic and pharmaceutical considerations in PG and GD populations.

Methodology

This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [12]. Database searches were done on PubMed, Web of Science, and PSYCInfo. Keywords used included “Stress”, “HPA”, “Cortisol”, “ACTH”, “CrH”, “epinephrine”, and “amylase” coupled with “gambling”, “problem gambling”, “gambling disorder”, and “pathological gambling”. Results were not restricted by year of publication or journal. The original inclusion criteria were that studies: (1) be peer-reviewed and accessible in English, (2) contain primary research involving human subjects (meta-analyses, systematic reviews, abstracts, and commentaries excluded) (3) include a comparison between a PG or GD population and a healthy control (HC) group, and (4) contain a biological measure of neuroendocrine stress function. After the search was completed, our criteria were expanded to include studies examining physiological stress markers in PG and GD without a HC group for comparison. Database searches and screening for studies took place between October 2022 and January 2023, and results were extracted thereafter.

Results

Our initial search generated 1024 results, 18 of which met our inclusion criteria (see Fig. 1). The results are organized across four main sections summarizing study findings on 1) basal and diurnal stress hormone measures; 2) the physiological stress response to gambling and other risky decision-making tasks; 3) direct acute stressor effects on stress hormones; and 4) therapeutic and pharmacological effects on stress hormones in GD. The study details are summarized in Tables 1 and 2.

Fig. 1
figure 1

A flowchart based on Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines representing the number of articles included in the review using set criteria. From: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71. doi: 10.1136/bmj.n71. For more information, visit: http://www.prisma-statement.org/

Full size image
Table 1 Summary of findings on baseline stress markers, therapeutic and pharmacological effects
Full size table
Table 2 Summary of findings on risk- and stress-driven effects on stress physiology
Full size table

Basal Stress Measures

While basal stress measures represent baseline physiological markers, diurnal activity follows a consistent circadian pattern. The “cortisol awakening response” (CAR) refers to the typical increase within 40 minutes of awakening, while the “diurnal cortisol slope” (DCS) is the subsequent decrease until nighttime [13]. The search identified three studies collecting basal stress markers, with one measuring diurnal cortisol activity (Table 1).

Geisel and colleagues [14] found no significant difference in basal levels of ACTH, cortisol, and copeptin (a marker of arginine vasopressin released as a stress hormone) between GD, Internet Gaming Disorder, and HC groups. However, lower cortisol levels were associated with greater GD severity. Another study [15] examined leptin, finding no significant differences between GD and HC groups. Although the GD group displayed a positive association between leptin and copeptin levels, leptin was not associated with ACTH or cortisol in either group, nor was it associated with self-reported gambling urges.

Roy and colleagues [16] measured diazepam binding inhibitor (DBI), a peptide implicated in depression, and CRH in individuals with GD, Major Depressive Disorder (MDD), and HCs. A trending larger range of DBI levels were observed in the HC group compared to the GD group, whereas the range of CRH levels trended greater in the GD group. Both groups displayed positive associations between DBI and CRH, although this correlation in the GD group was driven by individuals who were not depressed.

Wohl and colleagues [17] examined diurnal cortisol rhythms in university students classified as having recreational gambling (RG), PG or GD. All three groups showed the typical diurnal cortisol response, but the CAR was significantly larger in both the PG and GD groups compared to the RG group. Only at 330 minutes post-wakeup was there no difference in cortisol between groups.

Physiological Stress Activity During Risk-Taking

Five studies from our search examined the HPA axis and SNS activity under situations involving risky decision-making (Table 2).

Buchanan and colleagues [18] assessed diurnal cortisol profiles in comparison to performance on the Columbia Card task (CCT) and the Cups Task. Compared to HCs, the PG group displayed a flatter DCS and significantly lower early morning cortisol, but there were no differences right at waking, in the afternoon, or at bedtime. A lower CAR was associated with higher PG severity and riskier decisions in the “hot” CCT (targeting affectively driven decision-making), but not the “cold” CCT (relying on controlled decisions). Conversely, Cups Task performance yielded no significant relationships to cortisol or gambling pathology.

Labudda and colleagues [19] examined salivary cortisol and alpha-amylase (sAA) before and during the Computerized Game of Dice Task (cGDT) in GD versus HC groups. Although the GD group chose more disadvantageous alternatives on the cGDT than HCs, no significant differences in cortisol levels or sAA were identified between groups and no changes were observed throughout the task. In only the GD group, disadvantageous choices on the cGDT were negatively associated with sAA between baseline and task cessation.

Kruger and colleagues [20] examined plasma cortisol and heart rate activity when playing blackjack. In RG, PG, and GD groups, plasma cortisol and heart rate increased at the onset of gambling and declined at the end. These effects were more pronounced in conditions where participants gambled with their own money in a casino (experimental condition), than in a neutral laboratory setting without money (control condition). Furthermore, individuals rated as more impulsive had significantly greater heart rates during the experimental condition, but no cortisol differences.

Two methodologically similar studies by Meyer and colleagues [21, 22] also identified a relationship between physiological stress and real-stakes gambling. Meyer and colleagues [21] found significantly higher salivary cortisol levels and heart rates during the experimental compared to control conditions. GD severity negatively correlated with heart rate in only the control condition, with no effect on cortisol. Meyer and colleagues [22] also found that heart rate and norepinephrine levels increased more significantly in a PG compared to RG group during the experimental condition. Baseline epinephrine was comparatively greater in the PG group and remained elevated during the experimental condition, and ACTH, cortisol, and prolactin showed no significant main group differences.

Cue Reactivity and Cortisol

Paris and colleagues [23] applied an experimental protocol where GD and RG groups were presented videos of other people winning or losing money while gambling using their preferred method (e.g., slot machine, sports betting, etc.) (Table 2). Both groups rated “winning” and “neutral exciting” videos as more exciting than “losing” videos; however, only the RG group displayed cortisol increases following all three video presentations.

Physiological Response to Acute Stress

Three studies examined the physiological response to acute stress (Table 2). Wemm and colleagues [24••] administered the Trier Social Stress Test (TSST), a well-validated psychosocial stressor [25], to PG, heavy smoking, and HC groups; all displayed a cortisol rise, but HCs had a steeper increase compared to the PG group. An exploratory analysis showed that cortisol changes among the PG group were unrelated to gambling urges. Maniaci and colleagues [26] applied a similar protocol with the addition of measuring heart rate variability and also found TSST-increased cortisol levels and altered heart rate variability in both HC and GD groups. Conversely, there were no group differences over time, but the GD group showed a trend of greater salivary cortisol immediately and 20 minutes following task cessation compared to HCs.

Steinberg and colleagues [27] employed a psychological stress paradigm, with “uncontrollable noise” to induce acute stress. Participants with combined PG and Alcohol Use Disorder (PG+AUD) had significantly greater baseline systolic blood pressure (SBP) compared to PG and HC groups. Following acute psychological stress, both gambling groups displayed significantly decreased SBP while HCs had increased SBP. This effect was most pronounced in the PG group after receiving placebo beer, and greatest in the PG+AUD group when participants drank soda.

Therapeutic and Pharmacological Effects on Stress Markers in GD

Various interventions are used to treat GD, such as cognitive behavioural therapy and motivational interviewing [28], both of which are well-validated across many studies, including when they are combined [29]. Five studies extracted in this review explore therapeutic and pharmaceutical effects on stress physiology in GD (Table 1).

Angelo and colleagues [30] studied the efficacy of a short versus long physical activity program among individuals with GD. There were no significant changes in ACTH, cortisol, or prolactin following completion of the program, regardless of its duration. However, trending changes in prolactin levels were significantly associated with self-reported gambling craving.

Four studies measured stress marker changes following drug administration. Ramirez and colleagues [31] found that dexamethasone, an anti-inflammatory drug [32], did not alter cortisol activity in individuals with GD. Pallanti and colleagues [33] identified increased cortisol in both GD and HC groups and comparatively greater prolactin increases in the GD group following administration of a serotonin receptor agonist, meta-chlorophenyl piperazine (m-CPP). Moreno and colleagues [34] reported decreased prolactin in GD, but not HC groups, following clomipramine (serotonin reuptake inhibitor) administration, but no changes in cortisol. Zack and colleagues [35] administered amphetamine which increased cortisol and heart rate in both GD and HC groups, albeit with a blunted increase in cortisol and stronger subsequent heart rate decline in the GD group.

Discussion

This review aimed to examine acute stress physiology in GD and PG groups; however, only three studies met our original inclusion criteria, providing somewhat inconsistent results. Therefore, the criteria were subsequently broadened to include any study examining physiological stress activity in gambling populations.

Physiological Response to Acute Stress

While few studies to date have implemented an acute stressor in PG populations, the 2 studies identified in our review did not detect group differences with controls [24••, 26]. However, Wemm and colleagues demonstrated a blunted cortisol response over time in the PG group, relative to the healthy controls. Notably, the blunted cortisol response did not differ from a smoking control group in the same study.

Steinberg and colleagues [27] extended these findings, as both PG and PG+AUD participants displayed blunted cardiovascular activity compared to HCs following acute stress. Nevertheless, it must be noted that Steinberg and colleagues provided a placebo (non-alcoholic beer) to participants, which may have resulted in an expectancy effect [36]. Altogether, the blunted acute stress reactivity findings further replicate blunted cortisol responses reported in cocaine use disorder [37], opiate use disorder [38], chronic cannabis use [39], alcohol and combined alcohol and stimulant dependency [40, 41], and recently in internet addiction [42]. These findings are important as blunted responses further relate to stress appraisal [43], and behavioural motivations for drugs [44].

The growing literature across addictive disorders suggests that blunted responses may relate to HPA-axis alterations that either pre-exist (e.g., due to chronic stress) or relate to other specific addiction processes, rather than a direct effect of substances. Larger longitudinal studies comparing across groups may shed more light on these mechanisms.

Baseline Stress Physiology

The limited studies examining basal stress activity identified no differences in baseline levels of cortisol, copeptin, ACTH, CRH, and DBI between GD and HC groups [14,15,16]. Nevertheless, most basal samples were collected without control for diurnal fluctuations, thereby making firm conclusions difficult to draw regarding baseline stress activity in this population. Contradictory findings across studies in CAR may be due to the sample ages and chronicity of GD [17, 18], as dampened cortisol activity may be associated with a longer disorder duration [26].

Physiological Stress Activity During Risk-Taking

The finding of a blunted CAR in individuals with GD with more affectively driven risky decisions made on the CCT [18] can be understood in context with the underlying neural circuitry impacting cortisol activity. Previous research elucidates a relationship between cortisol activity and brain regions facilitating emotional processing and executive control; both human and preclinical studies have identified increased activation of the amygdala and prefrontal cortex in response to acute stress [45,46,47]. Rapid hydrocortisone administration, which stimulates cortisol, increases amygdala activity and errors in response to negative-affective stimuli in an emotional Stroop Task [45]. This suggests that endogenous cortisol stimulation activates key regions implicated in emotional processing, which subsequently produces a negative emotional bias and reduces executive control. Disruptions to executive control networks are further supported by Yuen and colleagues [46] who showed that greater plasma cortisol was associated with increased glutamatergic activity in the prefrontal cortex of stressed rats. In humans with GD, a yohimbine infusion generating an acute physiological stress response produces greater left amygdala activation compared to HC groups [47]. Stress-induced cortisol activity therefore impacts emotional processing and executive control through related brain regions, and this might be particularly impacted in people with PG or GD. Emotionally driven disadvantageous decisions made in GD may therefore be mediated by cortisol effects exerted onto these limbic regions.

Relatively consistent physiological findings were observed across four PG studies investigating how gambling with or without risk impacts stress markers. Real-stakes gambling, relative to no-stakes gambling, reliably produced a greater stress response through increased heart rate, norepinephrine, and cortisol [20, 21]. The cortisol and heart rate response were also greater in PG compared to RG groups in real-stakes gambling [22]. While Labudda and colleagues [19] observed no differences in cortisol or sAA between PG and HC groups, this study did not apply a real-stakes gambling task. In sum, this research provides consistent evidence of heightened stress reactivity in PG populations when exposed to gambling that mimic real-life scenarios. Given cortisol interactions with executive control networks described above, an important future research direction is better understanding the arousal and decision-making interactions in specific gambling scenarios.

Paris and colleagues [23] examined stress reactivity to gambling cues, rather than direct participation. Interestingly, these findings showed a blunted cortisol response when observing videos of other people gambling in the PG group compared to HCs. One possible interpretation of these results could relate to the perceived outcome of risk and decisions related to gambling in those with PG. Perhaps watching individuals gamble from a secondary perspective does not produce arousal in the way personal real-stakes gambling does, as the risk of negative consequences is mitigated, similar to in controlled no-stakes gambling. However, this is not well-explored and thus more work should seek to understand the mechanisms that drive increased versus blunted stress and how the context of gambling influences this in individuals with PG and GD.

Therapeutic and Pharmacological Considerations Regarding Stress Markers in GD

Most studies examining therapeutic treatment applications in GD use pharmaceuticals. Angelo and colleagues [30] employed a physical activity program among a treatment-seeking GD group, finding no significant changes in ACTH, cortisol, or prolactin. These findings contrast with research showing that exercise enables individuals with substance use disorders (SUDs) to effectively cope with daily stressors without substances, reduce feelings of tension and fatigue, and decrease stress reactivity [48]. Although Angelo and colleagues [30] did not observe significant changes following the intervention, exercise should not be ruled out as a viable treatment for GD. Other findings show reductions in GD severity and craving following exercise treatment in a GD population, and a program completion rate of 81.63% of participants [49]. Therefore, more work should expand on its effects on stress in GD.

Findings from Ramirez and colleagues [31] also directly contradicted similar research conducted in other clinical populations. There were no identified changes in diurnal cortisol following dexamethasone administration in GD populations. Conversely, similar doses of dexamethasone have successfully blunted cortisol activity in people with heavy smoking or tobacco and individuals with Binge Eating Disorder [50, 51]. Like Angelo and colleagues, Ramirez did not use a HC group for comparison. Thus, while it is unclear why dexamethasone or exercise produced no significant changes observed in other addictive disorders, more research is required to understand their true effects in these populations.

Zack and colleagues [35] found between-group differences in HPA axis and SNS activity, with a blunted cortisol increase and more rapid heart rate decline in the GD group following amphetamine administration. These effects were attributed to lower baseline cardiovascular measures in the GD group compared to controls, potentially indicative of chronic stress effects in this population. Pallanti and colleagues [33] found no between-group differences in cortisol following m-CPP administration, but increased prolactin in only the GD group. Similarly, Moreno and colleagues [34] found no differences in cortisol levels from clomipramine administration between groups, but significantly lower prolactin in the GD group. Since both m-CPP and clomipramine increase the functional activity of serotonin, these results might suggest a relationship between the serotonergic system and GD. Although it is unclear whether serotonergic systems are hyperactive or hypoactive in this population, serotonin appears to induce a disruption to maintenance of homeostatic balance that is facilitated by prolactin [52].

Limitations

The research reviewed in this paper came with significant limitations, including small sample sizes (mostly < 30 participants per group), which reduces the rigor. Another significant limitation is accounting for co-occurring conditions; only 7 of 18 studies explicitly controlled for other psychiatric disorders. SUDs such as nicotine and tobacco use disorders, AUD, and other SUDs frequently co-occur with GD, and the presence of psychiatric conditions significantly increases the odds of disordered gambling [53, 54•, 55]. Altered HPA axis activity has also been observed in people with heavy use of heroin, cocaine, alcohol, and stimulants [40, 41, 56,57,58]. Thus, limited research observing stress reactivity in GD and PG populations could be driven by SUDs that were not controlled in studies in this review. Similarly, mood disorders are common in individuals with GD [53]. Other research has found altered cortisol and heart rate activity in mood disorders such as MDD, MDD with psychotic features, and Bipolar Disorder compared to healthy individuals [59, 60]. Therefore, findings discussed in this review should be understood considering other potential mood conditions that may impact them.

Lastly, these limitations may be further compounded by sex- and gender-related differences; 12 of 18 studies discussed in this review included only male participants. Although males generally have higher GD rates [61], previous research also identifies sex differences in stress physiology. For example, research conducted in MDD and Generalized Anxiety Disorder populations finds that only females displayed different area under the curve in cortisol compared to HC groups [62]. Differences are also observed in SUDs, such that males with combined SUDs have greater cortisol activity in response to psychosocial stress than females [37]. Previous work has even highlighted different stress-driven motivations for gambling between sexes, as females are more likely to report gambling to relieve stress and anxiety than males and often report greater past-week mood disturbances than males [63, 64]. Thus, the current state of the research on stress physiology provides little insight into important sex differences.

Conclusion

This systematic review elucidated a need for more research surrounding stress physiology in gambling populations. Due to greatly limited research exploring acute stress reactivity in GD and PG populations, inclusion criteria were expanded to allow studies generally exploring physiological stress activity. Studies discussed in this review came with significant limitations, including small and mostly male samples with limited control for relevant co-occurring conditions. While it is difficult to draw concrete conclusions from these findings due to their heterogeneity, some themes can be identified. First, the acute stress response may be blunted in PG and GD populations. However, these findings were only consistent across two of three studies, with notable confounds. Nevertheless, findings to date suggest that basal stress measures do not differ between PG/GD populations and healthy groups. Diurnal activity, however, appears more frequently altered amongst gambling populations, but the direction of change is unclear. Failure to control for diurnal fluctuations in studies extracting basal measures may potentially explain null findings. Risky decision-making and gambling may more reliably alter stress activity in PG/GD populations, particularly in real-stakes situations. In neutral lab settings and second-hand viewings, differences are largely mitigated between PG/GD and HC groups. Considering the diverse methodologies employed in treatment intervention studies, general conclusions cannot be drawn from the existing work. Thus, future research should seek to validate acute stress findings, explore their relationships to decision-making and associated neural circuits, examine further basal stress activity, and investigate potential therapeutic options targeting stress systems in GD and PG populations.