An exploratory study of anxiety-motivated gambling in adolescents: Associations with minority status and gambling, health and functioning measures

Abstract

Gambling and anxiety are major public health concerns in adolescents and have been linked to emotion dysregulation and mood-modulating behaviors. While previous studies have shown links between positively reinforcing excitement-motivated gambling, health and functioning measures, and gambling perception and behavioral correlates in adolescents, few studies have examined such relationships relative to negatively reinforcing anxiety-motivated gambling (AMG). This study systematically examined relationships between adolescents reporting gambling to relieve anxiety (compared to those who gambled but did not report AMG) and measures of health/functioning and gambling-related measures. Participants included 1,856 Connecticut high-school students. Chi-square and logistic regression models were conducted. AMG was reported by 6.41% of the sample and was associated with identifying with a minority group (Black, Asian-American, Hispanic), at-risk/problem gambling, more permissive attitudes towards gambling, and higher odds of heavy alcohol, tobacco and other drug use, and violence-related measures. Adolescents with AMG were more likely to report non-strategic gambling, and gambling to escape/relieve dysphoria and due to feeling pressure. Additional between-group differences were found for gambling types, locations, motivations, and partners. Together, AMG may represent a mood-modulating behavior indicative of multiple problematic concerns, suggesting that emotional dysregulation may be an important factor in understanding the relationship between anxiety, problem gambling, and risky behaviors in youth. Additionally, the negative reinforcing motivations to gamble to relieve anxiety may be relevant particularly to adolescents from underrepresented minority racial/ethnic groups, and the specific factors underlying this relationship warrant further investigation.

INTRODUCTION

A meta-analysis estimated that 37.4% of community-dwelling adults exhibiting problem/pathological gambling had an anxiety-disorder (Lorains et al., 2011). Anxiety has been correlated with problem-gambling severity (Medeiros et al., 2016), and problem gambling has been linked to clinically significant anxiety (Barrault and Varescon, 2013). Specifically in adolescents, anxiety may contribute to problem gambling (Shead et al., 2010), with adolescent problem gambling associated with abnormal excitability, disinhibition, and emotional distress (Gupta and Derevensky, 1998).

The Pathways Model (Blaszczynski and Nower, 2002) proposes that positive-reinforcements, including gambling to enhance excitement, and negative-reinforcements, such as gambling to relieve depressed or uncomfortable affective states, may influence participation and development of habitual gambling in adolescents (Gupta et al., 2013). Gambling has been associated with excitement, and with positive-reinforcement from gambling “wins” that produce states of arousal described as akin to “drug-induced highs” (Blaszczynski and Nower, 2002). Engaging in gambling may also alleviate aversive affective states, including anxiety, through negative-reinforcement motivations (Blaszczynski and Nower, 2002). Positively reinforcing excitement-motivated gambling is common among adolescents and has been associated with greater severities of problem gambling, more frequent gambling, and more permissive gambling-related attitudes (Farhat et al., 2021). However, the relationships between negatively-reinforcing anxiety-motivated gambling (AMG) and gambling-related behaviors and attitudes among adolescents are not well understood. Previous literature has suggested that adolescents with elevated anxiety may use gambling to escape and disengage from stressful life-events or problems (Shead et al., 2010), consistent with coping-related motivations to gambling that have been linked to adolescent problem gambling (Grande Gosende et al., 2019). Furthermore, gambling problems have been hypothesized to proceed from ineffective emotion regulation, wherein individuals may misuse gambling to alleviate aversive affective states (Mestre-Bach et al., 2020).

Emotional dysregulation is central to anxiety problems, and associated with affective symptoms in adolescents. Youth with anxiety problems have exhibited more dysregulated expression of worry, sadness, and anger, as well as maladaptive coping strategies to negative emotions (Suveg and Zeman, 2004). Similarly, children with anxiety-disorders, relative to those without, employ less adaptive emotional coping strategies in response to negative life-events (Legerstee et al., 2010). A meta-analysis of 35 studies in children and adolescents showed that maladaptive emotion-regulation strategies, including rumination, avoidance, and suppression, were related to depressive and anxious symptomatology (Schäfer et al., 2017). Dysregulated emotion prospectively increased the likelihood of developing mood disorders in adolescents with a proposed unidirectional effect (McLaughlin et al., 2011; Young et al., 2019). Adolescence is a risk period for anxiety and depression. Emotion regulation involves changes in prefrontal cortical (PFC) regions that continue to develop through adolescents into adulthood (Ahmed et al., 2015; Gogtay et al., 2004; Young et al., 2019). PFC-amygdala functional connectivity may change directionality in adolescence, partly explaining developmental changes in emotional regulation and anxiety (Gee et al., 2013).

Emotion dysregulation has also been associated with more severe gambling problems. A longitudinal study examining the latent profiles of gamblers demonstrated that emotionally vulnerable and dysregulated adolescents with problem gambling had greater anxiety symptoms (Allami et al., 2017). Thus, anxiety and emotion dysregulation are implicated in adolescent gambling and problem gambling.

Problem gambling and anxiety have also been linked to poor health and functioning, and associated risky mood-modulating behaviors. Alcohol, tobacco, marijuana and illicit drug use, depressive symptoms, impulsivity, and violence have been meta-analytically associated with problem gambling in adolescents (Dowling et al., 2017). Longitudinal data demonstrated that adolescents with anxiety-disorders are prospectively 1.98 times more likely to develop substance-use disorders (Wolitzky-Taylor et al., 2012). Similarly, adolescents who used drugs or alcohol had higher frequencies of anxiety symptoms and were more likely to experience anxiety disorders (Wu et al., 2010). Evidence converging across multiple studies has shown that poor physical health, including obesity, is associated with anxiety symptoms in children and adolescents (Wang et al., 2019). However, how gambling perceptions and behaviors, and health/functioning measures may differ between adolescents stratified by whether they gambled to relieve anxiety is not well understood.

Hence, this study examined relationships between problem-gambling severity, gambling characteristics, and health/functioning in adolescents stratified by AMG/non-AMG (NAMG) status. We hypothesized that among adolescents, while controlling for potential confounding effects of demographic variances, AMG relative to NAMG would be associated with: 1) greater problem-gambling severity; 2) more permissive perceptions about gambling; and, 3) measures of poorer health and functioning. Additional exploratory analyses probed differences in gambling characteristics including gambling types, motivations, and partners in adolescents stratified by AMG status. We hypothesized in exploratory analyses that AMG would relate to gambling motivations to escape and to relieve dysphoria and feeling pressure to gamble. Based on previous findings between engaging in non-strategic gambling and anxiety (Barrault et al., 2019) and coping-related gambling motivations in adolescents (Grande Gosende et al., 2019), we also hypothesized that AMG would be associated with non-strategic types of gambling.

METHODS

Recruitment and Sample Characteristics

Data were drawn from a cross-sectional survey of Connecticut high-school students used in prior publications (e.g., (Farhat et al., 2021; Yip et al., 2011; Zhai et al., 2021, 2019)). Recruitment and data collection occurred over one year, and procedures have been previously described (Zhai et al., 2017). Briefly, invitations were sent to Connecticut public four-year and non-vocational or special education high-schools. Targeted recruitment was conducted to ensure adequate representation of the state’s geographic quadrants, and from its three tiers of family-socioeconomic-status-related district reference groups. Further permission was obtained from school boards and/or superintendents as necessary. Passive parental consent and student assent procedures, and study methods were approved by the Yale School of Medicine Institutional Review Board. Parents were mailed letters outlining the study and notified them to contact the school should they wish to decline their child’s/children’s participation. All procedures were performed in accordance with the 1964 Helsinki Declaration and its amendments. Surveys were administered at each school in a single day. Answers were anonymous and confidential, and students were reminded that participation was voluntary. The refusal rate of participation was <1%.

The survey consisted of 154 questions that evaluated demographic characteristics, health/functioning measures, substance use and other risk behaviors. The survey included previously established measures including the Massachusetts Gambling Screen (MAGS), and items from epidemiological surveys of adolescents including the Youth Risk Behavior Surveillance (Eisenmann et al., 2002; Kann et al., 2018). Of the students who completed the survey, 1,856 participants who reported gambling in the past-year completed the question measuring anxiety-gambling and all 12 MAGS questions corresponding to criteria for problem-gambling based on DSM-IV criteria. Surveys that had incomplete items for the analyses herein and questions with incorrectly circled responses were excluded.

Measures

Demographics.

Demographic variables included gender, race/ethnicity, grade-level in school, age, and family structure (e.g., living with one parent). The question and options assessing race/ethnicity were not mutually exclusive. Race/ethnicity was assessed by the item, “What is your racial background? Check all that apply.” African-American/Black, White/Caucasian, Asian, and Other categories were listed, and responses were coded as “yes” or “no” for each option. Sample demographics were consistent with the reports of the 2000 Census of Connecticut residents between ages 14–18 years. All regression models regarding AMG adjusted for potential confounding effects of gender, race/ethnicity, grade, and family structure.

Anxiety-Motivated Gambling.

AMG status was assessed using the question: “In the past year have you experienced a growing tension or anxiety that can only be relieved by gambling?” This item assessing anxiety that was relieved after gambling has been validated and utilized in previous studies to measure of urges and triggers for gambling (Farhat et al., 2020; Slavin et al., 2013; Weinberger et al., 2015; Yip et al., 2011). Students who answered “yes” were considered as having AMG, and those who answered “no” had NAMG. Of 1,856 students in the current study, 119 (6.41%) had AMG and 1,737 (93.59%) had NAMG.

Problem-Gambling Severity.

Problem-gambling severity was assessed using the DSM-IV pathological-gambling criteria subscale within the MAGS, a self-report instrument based on (Shaffer et al., 1994). The subscale contained 12 items corresponding to the 10 criteria for DSM-IV pathological gambling. Responses were coded as “yes” or “no”. Tolerance and withdrawal symptoms were each assessed by two items. Tolerance involved endorsing either betting increasing amounts to achieve the same level of excitement, or that the same amount of gambling had reduced effects. Withdrawal was assessed by endorsement of gambling to avoid withdrawal symptoms (e.g., of restlessness or irritability), or gambling to escape these uncomfortable feelings generated when not gambling. The total criteria were calculated to determine problem-gambling severity, with acknowledgement of either of the two questions each assessing tolerance or withdrawal counting once so that a total score of 0–10 was generated for assessing problem-gambling severity. The MAGS subscales have previously demonstrated adequate reliability (Cronbach’s = 0.83–0.87), predictive validity, and construct validity in adolescents (Shaffer et al., 1994). At-risk/problem-gambling (ARPG) was defined by DSM-IV criteria as previously in adolescents (Farhat et al., 2021; Kundu et al., 2013; Zhai et al., 2021, 2019) and adults (Desai and Potenza, 2008). As the MAGS demonstrated different types of at-risk gambling consistent with the Pathways Model of multiple problem-gambling trajectories (Edgren et al., 2016), a broader criteria for defining APRG was used to sufficiently capture risk variability. Adolescents who reported past-year gambling, but no diagnostic criteria, were classified as having low-risk gambling (LRG), and those with past-year gambling and one or more criteria had ARPG. Of the 1,856 participants, 1,211 (65.25%) had LRG and 645 (34.75%) had ARPG (mean criteria=2.81, sd=2.43).

Gambling Perceptions.

Adolescents endorsed perceptions (important vs. not important) on gambling prevention approaches including: “Checking identification for purchasing lottery tickets,” “Hanging out with friends who don’t gamble.” Participants were also assessed on their perceived parental attitude towards gambling (disapprove/approve/neither) with the question, “How do you think your parents would feel about your gambling, even once or twice, over the next 12 months?” and whether they had any concerns of family members’ gambling (yes/no) with the question, “Has the gambling of a close family member caused you worry or concern?”

Health and Functioning.

Health and functioning were assessed with the following measures: involvement in extracurricular activities (yes/no); grade average (A’s and B’s, mostly C’s, D’s or lower); lifetime tobacco smoking (never, occasionally, regularly), marijuana use (yes/no), other drug use (yes/no), and alcohol use (yes/no); and current (past 30-day) alcohol use (never, light, moderate, heavy) and caffeine use (none, 1–2 per day, 3+ per day). Body-mass indices (BMIs) were also recorded with underweight being ≤18.5, normal-weight being 18.6–24.9, and overweight/obese ≥25. Violence-related measures included: past 30-days weapon carrying (gun, knife, or club) (yes/no); not going to school because of feeling unsafe (yes/no); and being threatened by a weapon (yes/no); and past-12-months involvement in a physical fight (yes/no), and being injured in a physical fight requiring the treatment of a doctor or nurse (yes/no). Depression/dysphoria was measured by the question, “During the past 12 months, did you ever feel so sad or hopeless almost every day for two weeks or more in a row that you stopped doing some usual activities?” Given the link between anxiety and depression, models regarding AMG also controlled for depression/dysphoria.

Gambling Characteristics.

Adolescents reported engagement (yes/no) in strategic, non-strategic, and machine gambling. Engagement in strategic gambling (“games of skill” involving chance) was indicated as previously (Yip et al., 2011) by positively endorsing items including: playing cards (not in a casino), placing a bet with a bookie, betting with friends, betting on video/arcade games, betting on dice, betting on pool, and betting on other games of skill. Engagement in non-strategic gambling (“games of chance” not involving skill) was indicated as previously (Yip et al., 2011) by positively endorsing items including: receiving instant lottery or scratch tickets, receiving other lottery tickets, purchasing instant lottery or scratch tickets, purchasing other lottery tickets, and playing bingo at a church, synagogue, or other public places. Adolescents who endorsed items in both categories were identified as having both strategic and non-strategic gambling. Technology-facilitated gambling is an increasingly important contributor in problem-gambling risk among adolescents (Griffiths, 1999, 1993; Griffiths and Parke, 2010). Hence, engagement in machine gambling was indicated by positively endorsing having played slot, poker, or other gambling machines that utilize strategic (poker machines) or non-strategic (slots) approaches. Adolescents were also asked if they had (yes/no) different gambling behaviors and motivations, including specific gambling locations (internet, school, casino), reasons (excitement/fun, financial, escape/relieve dysphoria, social), and partners (alone, friends, parents, other adults, family, strangers, siblings). Gambling urges were assessed with the question, “Do you ever feel pressure to gamble when you do not gamble?” (yes/no). Adolescents also indicated age of onset of gambling (≤8 years old, 9–11 years old, 12–14 years old, ≥15 years old).

Statistical Analyses.

Statistical analyses were conducted using SPSS 26. Chi-square analyses compared the frequency distributions of demographic characteristics, gambling perceptions, and health/functioning measures between AMG and NAMG groups. To account for multiple comparisons, Bonferroni correction was applied such that chi-square analyses with p≤0.00185 were considered statistically significant. Binary and multinomial logistic regressions models were computed to examine associations between AMG and nominal health and functioning dependent variables, and enabled controlling for potential confounding covariates. Follow-up exploratory analyses examined associations between AMG and gambling characteristics. Odds-ratios (ORs) and 95% confidence intervals (CIs) were computed for regression models, adjusted for gender, ethnicity, grade, family structure, and depression/dysphoria to control for potential confounds, and significance was set at p≤0.05.

RESULTS

Demographics

Frequencies and chi-square results of demographic characteristics between AMG and NAMG groups are shown in Table 1. AMG was acknowledged by 6.41% of the sample. AMG related to identifying as male (χ²=11.15, p=0,001) having other family structure (χ²=61.47, p<0.001) and ARPG (χ²=155.40, p≤0.001). AMG and NAMG adolescents also differed at p≤0.001 on multiple racial/ethnic categories (with AMG related to minority status). Chi-square comparisons of adolescents with complete and incomplete data showed no systematic differences in endorsing racial/ethnic categories of White, Hispanic, Asian, and Other Ethnicity (all p>0.05), but those with incomplete data were more likely to endorse being female (χ²=37.72, p<0.001), Black (χ²=4.60, p=0.03), and younger (χ²=11.42, p=0.003). Demographic variables were included as covariates in regression models to account for potential confounding effects.

Table 1.

Sociodemographic characteristics of adolescents stratified by anxiety-motivated-gambling status

	Non-Anxiety-Motivated Gambling (N=1737)		Anxiety-Motivated Gambling (N=119)
Dependent Variable	N	%	N	%	χ2	p
Gender					11.15	0.001
Male	1052	61.3	87	77
Female	665	38.7	26	23
Race/Ethnicity
White					15.47	<0.001
No	456	26.3	51	42.9
Yes	1281	73.7	68	57.1
Black					26.91	<0.001
No	1565	90.1	89	74.8
Yes	172	9.9	30	25.2
Asian					12.05	0.001
No	1666	95.9	106	89.1
Yes	71	4.1	13	10.9
Hispanic					42.93	<0.001
No	1407	85.1	68	61.3
Yes	247	14.9	43	38.7
Other					1.89	0.169
No	1456	83.8	94	79
Yes	281	16.2	25	21
Grade					4.02	0.259
9th	524	30.3	29	24.6
10th	439	25.4	30	25.4
11th	447	25.8	29	24.6
12th	321	18.5	30	25.4
Age					19.97	<0.001
≤14 years	210	15.8	13	13.4
15–17 years	921	69.4	53	54.6
≥18 years	196	14.8	31	32
Family Structure					61.47	<0.001
One parent	407	23.8	26	22.6
Two parents	1200	70.3	60	52.2
Other	101	5.9	29	25.2
Problem-Gambling Severity					155.40	<0.001
Low-Risk Gambling	1196	68.9	15	12.6
At-risk/Problem Gambling	541	31.1	104	87.4

Gambling Perceptions

Chi-square results for gambling perceptions between AMG and NAMG adolescents are shown in Table 2. AMG adolescents differed from NAMG adolescents at p≤0.001 and chi-squares ranged from 10.31 to 107.46 on all assessed measures of the perceived importance of gambling preventions among teenagers and perceived parental attitudes towards gambling, except for not having access to internet gambling at home (χ²=4.76, p=0.029). In all cases, adolescents with AMG, relative to NAMG, reported more permissive attitudes towards gambling and were more likely to report a concern about a family member with a gambling problem. Follow-up logistic and multinomial regression analyses of gambling perceptions between AMG and NAMG adolescents are shown in Supplemental Table 1. Similarly, AMG adolescents differed from NAMG adolescents on all gambling perception measures except for not having access to internet gambling at home and having concern regarding a family member’s gambling. Additionally, a sum score of gambling perceptions was computed. Biserial correlation, controlling for demographic variables and depression/dysphoria, showed that having AMG inversely associated with total perceptions of gambling preventions as important (r=−0.21, p<0.001), indicating more permissive attitudes towards gambling in the AMG group.

Table 2.

Gambling perceptions in adolescent stratified by AMG status

	Non-Anxiety-Motivated Gambling		Anxiety-Motivated Gambling
Variable	N	%	N	%	χ2	p
Perceived parental perception of gambling					68.66	<0.001
Disapprove	508	34.6	42	42.4
Neither approve nor disapprove	809	55	22	22.2
Approve	153	10.4	35	35.4
Importance for preventing gambling problems in teens
Checking identification for purchasing lottery tickets					74.51	<0.001
Not important	311	19.1	59	54.1
Important	1314	80.9	50	45.9
Hanging out with friends who don’t gamble					34.47	<0.001
Not important	469	29.1	61	56
Important	1141	70.9	48	44
Participating in activities that are fun and free of gambling					74.88	<0.001
Not important	313	19.4	60	54.5
Important	1299	80.6	50	45.5
Fear of losing valuable possessions, close friends, and relatives					107.46	<0.001
Not important	190	11.8	52	47.3
Important	1423	88.2	58	52.7
Advertisements that show the problems associated with gambling					36.10	<0.001
Not important	396	24.8	56	50.9
Important	1202	75.2	54	49.1
Not having access to internet gambling at home					4.76	0.029
Not important	610	38.1	53	48.6
Important	991	61.9	56	51.4
Parent/Guardian strictness about gambling					34.59	<0.001
Not important	360	22.5	52	47.3
Important	1241	77.5	58	52.7
Warning from adults in family					51.22	<0.001
Not important	356	22.3	57	52.8
Important	1241	77.7	51	47.2
Warning from, or listening to, peers					55.23	<0.001
Not important	329	20.6	56	51.4
Important	1267	79.4	53	48.6
Having parents who don’t gamble					46.46	<0.001
Not important	369	23.1	57	52.3
Important	1229	76.9	52	47.7
Learning about the risks of gambling in school					47.79	<0.001
Not important	412	25.8	62	56.4
Important	1183	74.2	48	43.6
Learning about the risks of gambling from parents					52.18	<0.001
Not important	330	20.6	55	50.5
Important	1272	79.4	54	49.5
Learning about the risks of gambling from peers					48.04	<0.001
Not important	380	23.8	59	53.6
Important	1218	76.2	51	46.4
Adults not involving kids in gambling					57.03	<0.001
Not important	320	20.1	56	50.9
Important	1276	79.9	54	49.1
Parent/Guardian not permitting card games (for money) at home					10.31	0.001
Not important	611	38.2	59	53.6
Important	989	61.8	51	46.4
Family Concern					13.73	<0.001
No	1395	87.8	79	75.2
Yes	194	12.2	26	24.8

Health/Functioning Measures

Associations between AMG status and health and functioning measures are shown in Table 3, and the frequencies and percentages for endorsing each measure are shown in Supplemental Table 2. Adolescents with AMG, compared to NAMG, were more likely to endorse having lower grades (D’s or lower: OR=4.94, p<0.001), regular tobacco smoking (OR=4.75, p<0.001), current heavy alcohol use (OR=8.87, p=0.002), lifetime use of marijuana (OR=2.48, p=0.003) and other drugs (OR=5.46, p<0.001), and violence-related measures including weapon-carrying (OR=4.44, p<0.001), feeling unsafe at school (OR=4.70, p<0.001), having been threatened with a weapon (OR=5.15, p<0.001), involvement in a physical fight (OR=2.80, p=0.001), and having been injured in a physical fight (OR=4.77, p<0.001). Of these differences, only marijuana use (p=0.003) did not withstand Bonferroni correction.

Table 3.

Adjusted multi-variable analyses of health and functioning measures related to AMG status

	Anxiety-Motivated Gambling vs. Non-Anxiety-Motivated Gambling
Variable	Odds Ratio	95%CI	p
Academic and Extracurricular
Any extracurricular activities	0.69	0.38 – 1.24	0.22
Grade average
A’s and B’s	Ref.
Mostly C’s	1.05	0.54 – 2.06	0.88
D’s or lower	4.94	2.58 – 9.45	<0.001
Substance Use
Smoking
Never	Ref.
Occasionally	1.31	0.64 – 2.69	0.46
Regularly	4.75	2.47 – 9.14	<0.001
Marijuana Use, ever	2.48	1.36 – 4.52	0.003
Alcohol use, ever	0.62	0.34 – 1.96	0.65
Alcohol use, current
Never regular	Ref.
Light	3.63	0.94 −13.98	0.06
Moderate	2.92	0.72 – 11.86	0.13
Heavy	8.87	2.29 – 34.40	0.002
Other drug use, ever	5.46	3.02 – 9.84	<0.001
Caffeine use
None	Ref.
1–2 per day	0.59	0.28 – 1.21	0.15
3+ per day	1.22	0.62 – 2.42	0.57
Weight
Body Mass Index
Normal	Ref.
Underweight	0.69	0.20 – 2.36	0.55
Overweight/Obese	0.90	0.43 – 1.86	0.77
Violence-Related Measures
Weapon-Carrying	4.44	2.51 – 7.85	<0.001
Felt Unsafe	4.70	2.53 – 8.71	<0.001
Threatened by Weapon	5.15	2.94 – 9.02	<0.001
Physical Fighting	2.80	1.52 – 5.16	0.001
Physical Fighting with Injury	4.77	2.64 – 8.61	<0.001

Note: Odds ratios adjusted for gender, ethnicity, grade, family structure, and depression/dysphoria

Gambling Characteristics

Results from exploratory analyses of how gambling characteristics differed between AMG and NAMG groups are shown in Table 4, and the frequencies and percentages of engagement for each gambling characteristic are shown in Supplemental Table 3. Adolescents with AMG, compared to NAMG, had greater likelihoods of participating in machine gambling (OR=6.48, p<0.001), non-strategic gambling (OR=3.27, p=0.001), and both strategic and non-strategic gambling (OR=2.57, p=0.004). Additionally, AMG adolescents had greater likelihoods of gambling on the internet (OR=6.63, p<0.001), at school (OR=7.70, p<0.001), and at casinos (OR=11.53, p<0.001), for financial reasons (OR=4.50, p<0.001), to escape/relieve dysphoria (OR=3.77, p<0.001), and for social reasons (OR=1.96, p=0.01). They were also more likely to acknowledge having felt pressure to gamble (OR=12.03, p<0.001) and to have gambled alone (OR=13.53, p<0.001), and with other adults (OR=3.15, p<0.001) and strangers (OR=8.57, p<0.001). They were also less likely to report having gambled with friends (OR=0.32, p<0.001).

Table 4.

Adjusted multi-variable analyses of gambling measures related to AMG status

	Anxiety-Motivated Gambling vs. Non-Anxiety-Motivated Gambling
Variable	Odds Ratio	95%CI	p
Gambling Type
Strategic Gambling	0.60	0.21 – 1.68	0.331
Non-Strategic Gambling	3.27	1.58 – 6.75	0.001
Strategic and Non-Strategic Gambling	2.57	1.36 – 4.87	0.004
Machine Gambling	6.48	3.69 – 11.39	<0.001
Gambling Location
Internet	6.63	3.78 – 11.64	<0.001
School	7.70	3.77 – 15.71	<0.001
Casino	11.53	6.34 – 20.98	<0.001
Gambling Motivation
Excitement/Fun	1.53	0.77 – 3.07	0.227
Financial	4.50	2.18 – 9.30	<0.001
Escape/Relieve Dysphoria	3.77	2.18 – 6.52	<0.001
Social	1.96	1.15 – 3.33	0.014
Gambling Urges
Pressure	12.03	6.55 – 22.11	<0.001
Early Gambling
Age of Onset
≤8 years	Ref.
9–11 years	0.43	0.20 – 0.93	0.032
12–14 years	0.19	0.09 – 0.40	<0.001
≥15 years	0.11	0.05 – 0.27	<0.001
Gambling Partners
Alone	13.53	7.36–24.87	<0.001
Friends	0.32	0.18 – 0.55	<0.001
Parents	1.43	0.79 −2.59	0.240
Other Adults	3.15	1.78 – 5.59	<0.001
Family	0.60	0.33 – 1.07	0.084
Strangers	8.57	4.62 – 15.91	<0.001
Siblings	1.39	0.80 – 2.41	0.230

Note: Odds ratios adjusted for gender, ethnicity, grade, family structure, and depression/dysphoria

DISCUSSION

To our knowledge, this is one of few studies to explore the relationships between AMG and sociodemographics, gambling perceptions and attitudes, health and functioning measures, and gambling behaviors and motivations in U.S. adolescents. AMG was acknowledged by a small proportion of the sample (6.41%) and was related to male gender, minority status, older age, and “other” family structure. Consistent with the first hypothesis, AMG was associated with ARPG. In support of the second hypothesis, a greater proportion of adolescents with AMG perceived gambling prevention methods as unimportant, and their parents as approving of gambling. Further, they were more likely to report concerns regarding a family member with gambling problems. Multi-variable regressions testing the third hypothesis demonstrated between-group differences in health/functioning, such that AMG adolescents were more likely to have lower grade averages, regular tobacco smoking, current heavy alcohol use, lifetime drug use, and violence-related measures. Exploratory analyses demonstrated between-group differences for gambling locations, types, motivations, and partners. Consistent with hypotheses, adolescents with AMG had greater likelihoods of gambling to escape/relieve dysphoria and pressure to gamble.

The relatively small percentage of adolescents acknowledging AMG is in contrast to the larger percentage (67%) reporting gambling motivated by excitement-seeking in a separate study from the same survey (Farhat et al., 2021). These findings suggest that positive-reinforcement motivations may predominate in adolescent gambling behaviors, consistent with adolescence being a period high in sensation-seeking (Steinberg et al., 2008). However, whereas excitement-seeking gambling associated with identifying as White (Farhat et al., 2021), AMG was associated with identifying as Black, Hispanic, and Asian-American. Consistent with the Pathways Model (Blaszczynski and Nower, 2002), these results suggest that specific youth from minority groups who gamble may differ on gambling related to negative-reinforcement versus positive-reinforcement mechanisms. While the precise reasons for these findings are not clear, minority groups have often experienced considerable stress (Williams, 2018), which may contribute to affective problems (Thoits, 2010). The extent to which AMG impacts health determinants in adolescent from a minority group warrants additional direct investigation, which may help address health disparities.

Consistent with our second hypothesis, adolescents with AMG were more likely to acknowledge more permissive gambling attitudes and having family members with gambling concerns. The greater likelihood that adolescents with AMG perceived their parents as approving of gambling may promote gambling among youth given the important roles of parental behaviors (Zhai et al., 2017). The views that adolescents with AMG consider gambling-related prevention measures as unimportant suggest that gambling may not be perceived as harmful and may underlie the relationship between AMG and ARPG. Notably, some of the prevention approaches deemed unimportant by AMG adolescents (e.g., those related to specific forms of gambling like lotteries and specific locations like schools) coincided with greater likelihoods of specific forms of gambling (e.g., on lotteries and at school). In light of the permissive attitudes towards gambling in AMG adolescents, and the acceptance of gambling by parental figures, gambling may be perceived as an acceptable outlet and regulator of distressing emotions that may potentiate risk in AMG adolescents. Future studies may utilize longitudinal data to examine the pathways between gambling perceptions in AMG adolescents and their parents, and the risk of problem gambling and other concerns. Further, having family members with gambling problems being linked to AMG suggests the possibility of transgenerational impacts, although this possibility warrants direct examination.

Adolescents with AMG were more likely to acknowledge measures of poorer health and functioning. AMG was associated with poor academic performance (D’s or worse), which contrasts with excitement-seeking gambling motivations that were associated with better academic performance (A’s or B’s) (Farhat et al., 2021). AMG was also linked to substance use. High anxiety in gambling youth has been previously associated with greater severities of alcohol and illicit drug use (Ste-Marie et al., 2006). The self-medication hypothesis postulates that addictive substances may be used as mood-modulators to relieve anxiety, distress, and negative affect (Wolitzky-Taylor et al., 2012). Indeed, among individuals with gambling problems, self-medication-motivated gambling, as opposed to gambling-for-pleasure, was associated with substance-use problems (Li et al., 2013). These data suggest that anxiety may increase the risk of misusing drugs and alcohol within AMG adolescents, which may be similarly aimed at alleviating negative emotions, although this possibility warrants direct examination.

AMG adolescents had greater odds of endorsing violence-related measures. Physical fighting and sexual dating violence and aggression, have been associated with gambling and gambling problems in youth (Zhai et al., 2020). Longitudinal studies have demonstrated that emotion dysregulation prospectively linked to aggression in children and adolescents (Röll et al., 2012). Indeed, violence and aggression have been associated with maladaptive emotion-regulation styles (Bao et al., 2016). Individuals who endorse violence-related measures may be both perpetrators and victims of aggression (Lu et al., 2019). However, the current study did not probe the motivations underlying violence-related measures (e.g. self-defense, perpetration).Together with the findings above, anxious youth may engage a variety of risky behaviors that share mood-altering effects, and transition from more perceived acceptable gambling behaviors to more norm-violating and antisocial behaviors. Follow-up longitudinal research that probes potential directional relationships between gambling perception, AMG, and health-risk behaviors are important to inform problem-gambling prevention efforts in adolescents.

Exploratory analyses revealed that AMG adolescents had greater likelihoods of gambling to escape/relieve dysphoria, which included gambling to calm down and feel good, and gambling as a distraction. Emotion dysregulation has been associated with subsequent anxiety symptoms, especially in children and adolescents, and youth with anxiety-disorders have shown poorer understanding of emotion regulation (Legerstee et al., 2010; Southam-Gerow and Kendall, 2000; Suveg and Zeman, 2004). Dysregulated emotion has been proposed to have a unidirectional effect on the risk of anxiety problems (McLaughlin et al., 2011; Young et al., 2019). Gambling to escape and to improve negative moods were the most frequent gambling motivations among Lithuanian adolescents with problem gambling (Skokauskas and Satkeviciute, 2007). However, here only a small percentage of adolescents endorsed AMG as compared to gambling for excitement (Farhat et al., 2021), suggesting possible jurisdictional differences in gambling motivations that should be investigated further using psychometrically validated instruments like the Gambling Motives Questionnaire (Grande Gosende et al., 2019; Stewart and Zack, 2008). In line with the current findings linking AMG to feeling pressure to gamble, periods of conflict and stress have been linked to gambling in individuals with gambling problems (Wood and Griffiths, 2007). Furthermore, poor engagement of adaptive emotion-regulation strategies associated with pathological gambling (Orlowski et al., 2019). Hence, anxiety and motivations to be distracted from negative emotions may contribute to gambling to alleviate negative affective states. Although emotion dysregulation may be underlying the association between gambling to escape and AMG, further research is necessary to determine how emotion regulation may relate to AMG and its correlates in adolescents.

Between-group differences were found for multiple gambling characteristics. Adolescents with AMG were more likely to gamble alone and with strangers and less likely to gamble with friends and family. Solitary gambling may mediate the pathway between anxiety sensitivity and excessive gambling (Bristow et al., 2018). Indeed, adolescent AMG was associated with gambling on the internet, which may represent a solitary behavior linked to problem gambling (Griffiths, 1999), and internet gambling has been found to weaken the relationship between ARPG and gambling with friends (Potenza et al., 2011). High anxiety may motivate gambling in less anxiety-provoking contexts and increase the risk of asocial forms of gambling. In line with our hypothesis, AMG was associated with non-strategic gambling, which was consistent with findings linking coping-motivated gambling to non-strategic gambling (Grande Gosende et al., 2019).Together, findings suggest links between AMG and engagement in solitary and non-strategic forms of gambling which may be motivated by coping and serve to attenuate anxiety.

While traditional gambling types and formats continue to be present, new gambling and gambling-like opportunities have emerged. Internet gambling has grown (online sports gambling, daily fantasy sports), and loot boxes/crates, considered gambling in some jurisdictions, are in some video games (Delfabbro and King, 2020). Adolescents may be particularly vulnerable to gambling, including online forms (Floros, 2018; Prensky, 2001). As some gambling, including gambling on cards, has declined among adolescents, gambling may be conducted on digital devices (Delfabbro and King, 2020; Emond and Griffiths, 2020; Zhai et al., 2020). During the COVID-19 pandemic, online behaviors including gambling have changed (Håkansson, 2020; Sallie et al., 2021). Future studies utilizing data from a range of time points is important for understanding changes in AMG and gambling behaviors, and may inform development of new preventions of problem-gambling among adolescents affected by anxiety.

Limitations of the current study should be considered. Associations between AMG and outcomes of interest were not compared between ARPG and LRG. However, the sample size may have limited power, and there may not be sufficient resolution to accurately determine differences between problem-gambling groups. AMG status was defined by an individual item. This precluded further assessment of additional aspects of the anxiety related to gambling, including specific dimensions of anxiety, as well as their respective frequency and duration. However, previous studies have employed a similar approach to assess the effects of anxiety-motivated gaming (Garakani et al., 2021) and excitement-motivated gambling (Farhat et al., 2020; Pantalon et al., 2008) in epidemiological samples of adolescents. It is important for further studies with more sophisticated, validated instruments (e.g., ones that can assess dimensions of anxiety continuously, as well as symptoms of anxiety disorders that may motivate gambling to assess AMG) to examine AMG in greater detail. Psychometrically validated instruments that assess motivations for gambling should be considered in this regard (Grande Gosende et al., 2019; Stewart and Zack, 2008). Longitudinal data may be used to determine the stability of AMG. Chi-square results showed no association between grade-level and AMG status, suggesting that the presence of AMG and non-AMG did not vary considerably across high-school ages. Nonetheless, longitudinal studies should examine the transitions between AMG and non-AMG groups during development. Self-report data were not verified by other individuals and may have been subject to over- or under-reporting. However, sensitive information, such as violence, may not be readily disclosed. Despite questions derived primarily from validated measures and broadly used surveys, there was incomplete psychometric information on aspects of the survey. Although the study was conducted in 2006 and examination of relationships with AMG in current gambling environments in a large epidemiological adolescent sample is warranted, the study provides important historical data against which gambling behaviors in more recent gambling contexts may be compared. While the study utilized an epidemiological sample from multiple high schools across Connecticut, generalizability to other jurisdictions may be limited. Furthermore, the cross-sectional data limit interpretations of directional links between measures. Longitudinal approaches may further elucidate mechanisms underlying relationships between anxiety, gambling, and maladaptive behaviors in adolescents.

CONCLUSIONS

This is one of the first studies to examine relationships between AMG, gambling attitudes and behaviors, and health and functioning measures among high-school students. AMG was relatively rare relative to excitement-seeking gambling and was associated with under-represented-minority status and poor academic performance. Adolescents with AMG had greater likelihoods of gambling to escape/relieve dysphoria and to engage in solitary gambling. Additionally, AMG adolescents had greater likelihoods of engaging in other potentially mood-modulating behaviors, including heavy alcohol, tobacco, and other drug use, as well as violence-related measures. Together, these findings highlight the importance of AMG among adolescents, particularly those identifying with minority groups. It may be beneficial for prevention efforts in schools and communities to focus on emotion regulation within anxious adolescents who gamble and experience substance use and violence-related behaviors.