Associations Between Heavy Episodic Drinking and Perceived Social Isolation in U.S. Young Adults by Sexual Orientation

Abstract

Heavy episodic drinking (HED) is prevalent in young adulthood, yet its relationship with psychosocial well-being remains complex. This study examines the association between HED and perceived social isolation among young adults and tests whether this relationship varies by sexual orientation. Using pooled, nationally representative data from the 2022 and 2024 Health Information National Trends Survey (HINTS), this study analyzed adults aged 18–29 (N = 723). Perceived social isolation was measured using the PROMIS Social Isolation Short Form. Weighted multivariable linear regression models assessed interactions between sexual orientation and HED occasions (0 vs. 1+), adjusting for sociodemographic variables and psychological distress. 45.5% reported HED. Lesbian/gay (B = 5.62, SE = 0.58, p < 0.001) and bisexual (B = 1.66, SE = 0.34, p < 0.001) young adults reported higher isolation than straight peers; HED was inversely associated with isolation (B = −1.71, SE = 0.20, p < 0.001). A significant interaction indicated that among lesbian/gay young adults, heavy drinking was associated with lower perceived isolation (B = −5.77, SE = 0.98, p < 0.001). Interventions should account for the social meanings of alcohol use to avoid unintentionally increasing isolation among sexual minoritized populations.

1. Introduction

Heavy episodic drinking (HED), defined as alcohol consumption of 5 or more drinks for males or 4 or more drinks for females within 2 h, is highly prevalent in the United States (U.S.) with 7.9% of adolescents and young adults aged 12–20 reporting HED in 2024 and no significant change in these rates over the past 4 years (Substance Abuse and Mental Health Services Administration [SAMHSA], 2025). Developmental research consistently demonstrates that alcohol use escalates during young adulthood, with HED peaking as young adults navigate transitions related to education, employment, and social belonging (Brown et al., 2008, 2009; Quigley & Marlatt, 1996; Schulenberg & Maggs, 2002). HED in young adulthood has been linked to adverse mental health outcomes, including elevated symptoms of depression and anxiety, as well as increased risk for alcohol-related problems and longer-term substance use disorders (Gates et al., 2016; White & Jackson, 2004). At the same time, alcohol use in this age group is deeply embedded in social contexts, frequently occurring in peer-centered environments where drinking may function as a mechanism for social facilitation, stress reduction, and perceived connection (Brook & Willoughby, 2016; White & Jackson, 2004). This dual role of alcohol use underscores that while HED is associated with significant psychological harm, it may also coincide with greater social engagement and perceived connectedness, complicating its relationship with psychosocial well-being during the transition to adulthood (Mathes Winnicki et al., 2023; Villar et al., 2022).

Perceived social isolation refers to an individual’s subjective sense of social disconnection, loneliness, or lack of belonging and is conceptually distinct from objective indicators of social isolation such as network size, frequency of social contact, or living arrangements. Individuals may report high levels of perceived isolation despite having extensive social networks, underscoring the primacy of perceived social quality over structural measures of social integration (Cacioppo & Hawkley, 2009; Cadigan et al., 2023). Perceived social isolation is particularly salient during young adulthood, a developmental period in which peer affiliation, social acceptance, and identity-affirming relationships are central to psychological well-being and emerging adult functioning. Disruptions to perceived connectedness during this life stage have been linked to elevated psychological distress, including depressive and anxiety symptoms, as well as maladaptive coping behaviors such as increased substance use and alcohol consumption (Cadigan et al., 2023; Rapier et al., 2019). Growing evidence further suggests that perceived social isolation may shape longer-term health risk trajectories by influencing stress processes, emotional regulation, and health behaviors across the life course (Campagne, 2019). In population-based research, perceived social isolation is increasingly assessed using the Patient-Reported Outcomes Measurement Information System (PROMIS) Social Isolation measure, a validated instrument grounded in item response theory that captures subjective social disconnection and allows for standardized comparisons across populations and studies (Hahn et al., 2010). PROMIS-based measures have been widely applied in behavioral and public health research, including studies of young adults and digital social environments, supporting their utility for examining social isolation as a critical psychosocial determinant of health (Primack et al., 2017; Thakur et al., 2025; Whaite et al., 2018).

Sexual minority stress theory provides a central framework for understanding disparities in social and mental health among sexual minoritized individuals by positing that stigma-related stressors, including discrimination, social exclusion, identity concealment, and internalized stigma, operate in addition to general life stressors to undermine psychological well-being (Alessi, 2014; Hoy-Ellis, 2023; Pitoňák, 2017). These chronic stress processes are closely linked to experiences of social disconnection, as sexual minoritized individuals may encounter rejection within families, peer networks, and broader social institutions, or may limit social engagement to avoid stigma exposure (Pachankis et al., 2020). Consistent with this framework, lesbian, gay, and bisexual young adults report higher levels of loneliness, perceived social isolation, and psychological distress compared with their heterosexual peers, disparities that emerge early and persist across developmental stages (Argyriou et al., 2021; Garcia et al., 2020). Sexual minoritized populations are not monolithic; meaningful heterogeneity exists across sexual identity groups, with bisexual individuals often experiencing distinct and, in some cases, heightened vulnerability related to identity invisibility, marginalization from both heterosexual and gay/lesbian communities, and differential coping strategies (McCormack, 2023). Such heterogeneity extends to health behaviors and social contexts, highlighting the importance of disaggregating sexual orientation groups when examining social isolation, mental health, and related risk behaviors among young adults.

A substantial body of research documents elevated alcohol use and heavier drinking patterns among sexual minoritized populations, particularly among young adults and bisexual individuals, who often report higher prevalence of HED and alcohol-related consequences compared with heterosexual peers (Cerezo et al., 2020; Dyar & Kaysen, 2022; Talley et al., 2016). These disparities are frequently framed within minority stress processes, wherein alcohol use may function as a coping strategy for managing stigma-related stressors such as discrimination, identity concealment, and internalized stigma (Kalb et al., 2018; Miller et al., 2026). At the same time, alcohol use among sexual minoritized individuals is deeply embedded in social contexts. Alcohol-centered venues, including bars, clubs, and social gatherings, have historically served as critical spaces for community formation, identity affirmation, and access to social support in the context of broader societal marginalization (Fairlie et al., 2018; Parks & Heller, 2013). Engagement in these settings may therefore reflect social affiliation and community connectedness rather than exclusively maladaptive coping, particularly among young adults navigating identity development and belonging (Cogger et al., 2012; Feinstein et al., 2017). This dual function of alcohol use suggests that the association between HED and perceived social isolation may differ by sexual orientation, with HED potentially coinciding with lower perceived isolation for some sexual minoritized groups while remaining a marker of psychological distress for others, demonstrating the importance of examining effect heterogeneity across sexual identity groups.

Despite extensive research on alcohol use, mental health, and social well-being among young adults and sexual minoritized populations, several critical gaps remain. First, much of the existing literature has focused on psychological distress or substance use outcomes independently, with comparatively limited attention to perceived social isolation as a distinct psychosocial determinant that may both shape and be shaped by alcohol use behaviors. Second, studies examining alcohol use among sexual minoritized populations frequently aggregate lesbian, gay, and bisexual individuals, obscuring important heterogeneity in social contexts, coping strategies, and health behaviors across sexual identity groups. Third, few population-based studies explicitly test whether the association between HED and psychosocial outcomes differs by sexual orientation, despite theoretical and empirical evidence suggesting that alcohol use may function both as a coping response to minority stress and as a mechanism for social affiliation and community connectedness. Finally, much of the existing evidence is derived from convenience samples or pre-pandemic data, limiting generalizability and relevance to contemporary social environments. Addressing these gaps using nationally representative data and analytic approaches that allow for effect modification is essential to advancing understanding of how HED intersects with perceived social isolation across sexual identity groups during young adulthood.

The purpose of this study was to examine the association between HED occasions and perceived social isolation among young adults aged 18–29 in the U.S., and to assess whether this association differs by sexual orientation. Using pooled, nationally representative data from the 2022 and 2024 cycles of the Health Information National Trends Survey (HINTS), this study extends prior research by focusing on perceived social isolation as a key psychosocial outcome and by explicitly testing effect heterogeneity across sexual identity groups. Guided by minority stress theory and social affiliation frameworks, it was hypothesized that sexual minoritized young adults would report higher levels of perceived social isolation than their straight counterparts. It was further hypothesized that HED would be associated with perceived social isolation overall, but that the magnitude and direction of this association would vary by sexual orientation, reflecting differences in social contexts, coping mechanisms, and community connectedness across lesbian/gay, bisexual, and straight young adults.

2. Methods

2.1. Study Design and Data

This study used data from pooled responses from the HINTS 6 (2022) and the HINTS 7 (2024). The HINTS is a nationally representative, cross-sectional survey administered by the National Cancer Institute to monitor how U.S. adults aged 18 years and older seek, access, and use health information, along with related health behaviors (National Cancer Institute, 2025). The survey samples the noninstitutionalized U.S. adult population across all 50 states and the District of Columbia using a stratified, probability-based sampling design. Both survey cycles employ comparable methodologies, enabling pooled analyses across years to increase statistical power and enhance population-level inference. Detailed information regarding sampling procedures, survey administration, and weighting strategies is available in the publicly accessible HINTS methodology documentation. This study met criteria for exemption from institutional review board (IRB) review as it is publicly available and deemed not human subjects research.

Age functioned as the primary inclusion/exclusion variable restricting the analytic sample to those aged 18–29 (n = 1191). The sample was further restricted to only those who drank alcohol in the past 30 days excluding those reporting no alcohol use (31.2%). Missing data accounted for 6.8% of sex, 9.7% of sexual orientation, 7.1% of race/ethnicity, 6.3% of education level, 7.1% of household income, 0.6% of smoking status, 4.0% of Personal Health Questionnaire (PHQ)-4 scores, and 4.2% of the Patient Reported Outcomes Measurement Information System (PROMIS) Social Isolation scores. Primary analyses were conducted using complete-case (listwise deletion) data, resulting in a final analytic sample of 723 respondents. Sensitivity analyses using multiple imputation were conducted to evaluate the robustness of findings to missing data assumptions.

2.2. Measures

2.2.1. Sociodemographic Measures

Participants reported a range of sociodemographic characteristics that were examined in relation to HED occasions (0 vs. 1+). Age was treated as a continuous variable. Biological sex, or sex assigned at birth, was self-reported and categorized as male or female. Sexual orientation was classified as lesbian/gay, straight, or bisexual; respondents identifying as something else were excluded due to small cell sizes. Race and ethnicity were measured using standard federal classification categories and grouped as non-Hispanic (NH) White, NH Black, Hispanic, NH Asian, and NH other (i.e., American Indian, Alaska Native, Pacific Islander, and those who selected multiple races).

Educational attainment reflected the highest level of education completed and was categorized as less than high school, high school graduate, some college or associate degree, and bachelor’s degree or higher. Household income was assessed in six categories: <$20,000; $20,000–<$35,000; $35,000–<$50,000; $50,000–<$75,000; $75,000–<$100,000; and $100,000+. Household composition was captured as the number of adults living in the household and dichotomized as 1 versus 2 + adults. Survey year (HINTS 6 [2022] vs. 7 [2024]) was included as a covariate in pooled analyses to account for potential secular trends and between-cycle differences across survey waves.

2.2.2. Psychological Distress

Symptoms of depression and anxiety were measured using the PHQ-4, an established ultra-brief screening instrument for psychological distress (Kroenke et al., 2009). The PHQ-4 includes four items derived from longer validated scales, with two items capturing anxiety symptoms (General Anxiety Disorder [GAD]-2) and two assessing depressive symptoms (PHQ-2). Respondents indicated how frequently they experienced each symptom during the prior two weeks using a four-point response scale ranging from 0 (“not at all”) to 3 (“nearly every day”). Item responses were summed to create a total score ranging from 0 to 12, with higher values reflecting greater overall distress. Sample items include statements such as feeling nervous or on edge and being unable to control worrying (anxiety items), as well as having little interest or pleasure in activities and feeling down or hopeless (depression items). The PHQ-4 has strong evidence of reliability and validity across diverse populations, commonly applied in both clinical and epidemiologic research as an indicator of overall mental health symptom burden. In the original validation study, internal consistency was reported as α = 0.78 for the full scale, with α = 0.82 for the GAD-2 and α = 0.75 for the PHQ-2 (Löwe et al., 2010). A subsequent systematic review found Cronbach’s alpha values ranging from 0.72 to 0.88 for the total scale, with somewhat lower ranges for the PHQ-2 (0.65–0.81) and GAD-2 (0.74–0.88) (Caro-Fuentes & Sanabria-Mazo, 2024). In the current sample, the PHQ-4 demonstrated excellent internal consistency (α = 0.87).

2.2.3. Perceived Social Isolation

Perceived social isolation was measured using the PROMIS Social Isolation Short Form, which captures individuals’ subjective experiences of social disconnection and loneliness using four items assessing feelings of loneliness, exclusion, and lack of companionship, focusing on how often respondents experience social disconnection in their daily lives (e.g., “I feel left out”, “I feel people barely know me”) (Hahn et al., 2010). Participants rated how often they experienced each feeling using a five-point Likert scale, with response options ranging from “never” to “always.” Higher scores indicate greater perceived isolation.

Raw scores were converted to PROMIS t-scores standardized to the U.S. general population (mean = 50, SD = 10), allowing for comparability across studies and populations. T-scores above 50 represent higher-than-average levels of perceived social isolation, whereas scores below 50 indicate lower perceived isolation. The PROMIS Social Isolation measure is grounded in item response theory and has demonstrated strong reliability and validity despite its brief format, making it well suited for population health and behavioral research examining social disconnection and related health outcomes. The PROMIS Social Isolation measure showed excellent internal consistency in this sample with α = 0.92. Although distress is conceptually related to perceived social isolation and may lie on the causal pathway linking minority stress to isolation (Falk et al., 2026), its inclusion allows for estimation of associations independent of overall psychological distress; however, there is the potential for overadjustment bias. The variance inflation factor between these two variables was 1.1 denoting no collinearity.

2.2.4. Heavy Episodic Drinking

HED was measured using the item: “For males: During the past 30 days, how many times did you have 5 or more alcoholic drinks on one occasion? For females: During the past 30 days, how many times did you have 4 or more alcoholic drinks on one occasion?” Responses were coded as 0 = never and 1 = one or more times for individuals who drank. Those who did not drink any alcohol in the past 30 days (31.2% of the sample) were excluded. This measure captures the frequency of HED but does not assess the social context in which drinking occurs (e.g., social events versus solitary drinking).

2.3. Statistical Analysis

All analyses incorporated the complex sampling structure of the HINTS survey. Person-level final weights were applied to produce estimates representative of the U.S. adult population. These weights account for unequal selection probabilities, survey nonresponse, and alignment with national population benchmarks using iterative proportional fitting procedures. To obtain valid variance estimates, the set of HINTS-provided jackknife replicate weights was used, consistent with the survey’s stratified, two-stage sampling design. Prior to analysis, the survey design was specified by assigning the final person-level weight as the probability weight and incorporating all 100 replicate weights over both cycles for variance estimation.

Weighted descriptive statistics, including frequencies, proportions, means, and jackknife standard errors, were calculated to summarize sample characteristics. Descriptive analyses were conducted for the overall sample and stratified by reported number of HED occasions. Group differences in categorical variables were evaluated using Rao–Scott adjusted chi-square tests, while differences in continuous measures (age, PHQ-4 scores) were assessed using weighted one-way analysis of variance (ANOVA).

Primary weighted, multivariable linear regression models were estimated to examine the association between HED occasions and perceived social isolation. Perceived social isolation, operationalized as the PROMIS Social Isolation t-score, was modeled as a continuous outcome. All models accounted for the complex HINTS survey design through the application of person-level sampling weights and replicate weights. An interaction term between sexual orientation and HED occasions (0 vs. 1+) was included to assess whether the association between HED and social isolation differed by sexual orientation. Models were adjusted for sociodemographic covariates and survey year. Adjusted regression coefficients and 95% confidence intervals are reported.

The interaction between sexual orientation and HED on social isolation was also examined. Sexual orientation was categorized into three groups (straight, lesbian/gay, bisexual), and HED occasions were dichotomized (0 vs. 1+ occasions). Both predictors were dummy-coded, with “straight” as the reference group for sexual orientation and “0 occasions” as the reference group for heavy drinking. The interaction was parameterized by including product terms for (Lesbian/gay × 1+ HED) and (Bisexual × 1+ HED), allowing assessment of whether the association between HED and social isolation differed for sexual minoritized respondents relative to straight peers.

Significant interactions were interpreted using estimated marginal means and simple effects. Predicted PROMIS Social Isolation t-scores were calculated for each combination of sexual orientation and HED status while holding covariates at their mean values. Simple effects analyses examined differences in isolation scores between respondents with 0 versus 1+ HED occasions within each sexual orientation group. To facilitate interpretation of within-group differences, additional models were estimated in Appendix A 

Table A1. Weighted and adjusted linear regressions predicting PROMIS Social Isolation t-score of adults in the United States aged 18–29, 2022 and 2024, with interactions releveled by sexual orientation (N = 723).

	Model 1				Model 2
			95% CI				95% CI
	B	SE	LL	UL	B	SE	LL	UL
Intercept	54.85 ***	1.16	52.56	57.15	45.44 ***	1.01	43.43	47.45
Age	−0.11 ***	0.03	−0.16	−0.06	−0.11 ***	0.03	−0.16	−0.06
Sex (ref. = Male)
Female	1.44 ***	0.22	1.01	1.88	1.44 ***	0.22	1.01	1.88
Sexual identity
Lesbian/Gay	Ref.				−6.74 ***	1.14	−9.01	−4.47
Straight	−7.88 ***	0.78	−9.44	−6.33	−0.97	0.55	−2.06	0.12
Bisexual	−6.68 ***	0.88	−8.43	−4.93	Ref.
Race/ethnicity (ref. = NH White)
Non-Hispanic Black	−3.54 ***	0.63	−4.79	−2.29	−3.54 ***	0.63	−4.79	−2.29
Hispanic	−1.78 ***	0.24	−2.25	−1.30	−1.78 ***	0.24	−2.25	−1.30
Non-Hispanic Asian	1.55 ***	0.49	0.58	2.52	1.55 ***	0.49	0.58	2.52
Non-Hispanic Other	4.81 ***	0.52	3.77	5.84	4.81 ***	0.52	3.77	5.84
Education level (ref. = college graduate or more)
Less than high school	−1.70 **	0.60	−2.89	−0.51	−1.70 **	0.60	−2.89	−0.51
High school graduate	0.34	0.30	−0.26	0.94	0.34	0.30	−0.26	0.94
Some college	−0.12	0.27	−0.65	0.42	−0.12	0.27	−0.65	0.42
Household income (ref. = $100,000+)
<$20,000	2.11 ***	0.52	1.09	3.13	2.11 ***	0.52	1.09	3.13
$20,000–<$35,000	−1.75 ***	0.49	−2.71	−0.78	−1.75 ***	0.49	−2.71	−0.78
$35,000–<$50,000	1.00	0.52	−0.02	2.03	1.00	0.52	−0.02	2.03
$50,000–<$75,000	−0.76	0.51	−1.76	0.25	−0.76	0.51	−1.76	0.25
$75,000–<$100,000	−0.21	0.48	−1.16	0.74	−0.21	0.48	−1.16	0.74
Adults in household (ref. = 2+)
1	0.42	0.26	−0.09	0.94	0.42	0.26	−0.09	0.94
HINTS Cycle (ref. = 7[2024])
6 (2022)	0.79 *	0.30	0.19	1.39	0.79 *	0.30	0.19	1.39
Smoking Status (ref. = never)
Current	2.73 ***	0.31	2.11	3.35	2.73 ***	0.31	2.11	3.35
Former	−1.02 **	0.38	−1.78	−0.27	−1.02 **	0.38	−1.78	−0.27
Personal Health Questionnaire (PHQ)−4 score	2.11 ***	0.05	2.00	2.22	2.11 ***	0.05	2.00	2.22
Heavy episodic drinking (HED) occasions (ref. = 0)
1+	−7.31 ***	0.95	−9.19	−5.42	−0.57	0.53	−1.62	0.49
Interaction (HED = 1+ × sexual orientation)
HED = 1+ × lesbian/gay	Ref.				−6.74 ***	1.14	−9.01	−4.47
HED = 1+ × straight	5.77 ***	0.98	3.84	7.71	−0.97	0.55	−2.06	0.12
HED = 1+ × bisexual	6.74 ***	1.14	4.47	9.01	Ref.

Note: B = unstandardized regression coefficient; SE = jackknife standard error; CI = confidence interval; LL = lower limit; UL = upper limit; PROMIS = Patient-Reported Outcomes Measurement Information System; * p < 0.05, ** p < 0.01, *** p < 0.001.

with sexual orientation re-leveled such that lesbian/gay and bisexual respondents served as the reference groups in separate models, allowing direct estimation of the association between HED and social isolation within each sexual minoritized subgroup. All analyses accounted for the complex HINTS survey design using the svy prefix in Stata 19.5 (StataCorp, LLC., College Station, TX, USA), incorporating person-level final weights and 100 jackknife replicate weights for valid variance estimation.

Due to the small sizes of the lesbian/gay (n = 39) and bisexual (n = 38) subgroups, particularly in the context of interaction analyses, statistical power to detect small-to-moderate effects was limited. Detectable effect size calculations indicate that only large effects (Cohen’s d ≈ 1.0) could be reliably detected within these subgroups.

As a sensitivity analysis, missing data were addressed using multiple imputation by chained equations. Continuous variables were imputed using linear regression and categorical variables using multinomial logistic regression. Twenty imputed datasets were generated and combined using Rubin’s rules. The same survey-weighted regression models, including the sexual orientation × HED interaction, were re-estimated in the imputed data to assess the robustness of findings to missing data assumptions. Results from the multiple imputation analyses are reported in Appendix A 

Table A2. Weighted and adjusted linear regression models using multiple imputation to predict PROMIS Social Isolation t-scores among U.S. adults aged 18–29, 2022 and 2024 (N = 742).

	Model 1				Model 2
			95% CI				95% CI
	B	SE	LL	UL	B	SE	LL	UL
Intercept	43.99 ***	2.77	38.56	49.43	43.78 ***	2.77	38.33	49.22
Age	−0.09	0.10	−0.29	0.10	−0.09	0.10	−0.29	0.10
Sex (ref. = Male)
Female	1.08	0.63	−0.16	2.32	1.09	0.63	−0.14	2.33
Sexual identity (ref. = straight)
Lesbian/Gay	2.97 *	1.30	0.41	5.53	4.59 *	1.84	0.99	8.19
Bisexual	2.23 *	0.86	0.54	3.92	2.41 *	1.21	0.05	4.78
Race/ethnicity (ref. = NH White)
Non-Hispanic Black	0.53	0.93	−1.30	2.36	0.59	0.93	−1.24	2.42
Hispanic	−0.06	0.72	−1.47	1.36	−0.06	0.72	−1.48	1.36
Non-Hispanic Asian	2.37 *	1.10	0.21	4.54	2.50 *	1.11	0.32	4.67
Non-Hispanic Other	3.64 *	1.47	0.74	6.53	3.61 *	1.47	0.72	6.51
Education level (ref. = college graduate or more)
Less than high school	−0.39	1.85	−4.03	3.24	−0.34	1.85	−3.98	3.30
High school graduate	−0.63	0.97	−2.54	1.28	−0.55	0.98	−2.47	1.37
Some college	−0.72	0.75	−2.20	0.76	−0.67	0.76	−2.15	0.82
Household income (ref. = $100,000+)
<$20,000	0.88	1.33	−1.74	3.50	0.80	1.34	−1.82	3.43
$20,000–<$35,000	−0.94	1.42	−3.72	1.85	−1.00	1.42	−3.79	1.79
$35,000–<$50,000	0.68	1.29	−1.86	3.21	0.59	1.29	−1.95	3.13
$50,000–<$75,000	0.73	1.20	−1.64	3.09	0.66	1.21	−1.71	3.03
$75,000–<$100,000	−0.49	1.21	−2.87	1.88	−0.50	1.21	−2.88	1.87
Adults in household (ref. = 2+)
1	0.63	0.74	−0.82	2.09	0.67	0.74	−0.79	2.13
HINTS Cycle (ref. = 7[2024])
6 (2022)	0.50	0.68	−0.83	1.82	0.55	0.68	−0.78	1.88
Smoking Status (ref. = never)
Current	1.55	1.11	−0.63	3.72	1.60	1.11	−0.58	3.77
Former	0.82	1.15	−1.45	3.09	0.81	1.16	−1.46	3.08
Personal Health Questionnaire (PHQ)−4 score	1.97 ***	0.10	1.78	2.16	1.97 ***	0.10	1.77	2.16
Heavy episodic drinking (HED) occasions (ref. = 0)
1+	−1.39 *	0.62	−2.60	−0.18	−1.15	0.69	−2.50	0.21
Interaction (ref. = HED = 1+ × straight)
HED = 1+ × lesbian/gay					−3.19	2.58	−8.25	1.88
HED = 1+ × bisexual					−0.37	1.65	−3.61	2.87

Note: B = unstandardized regression coefficient; SE = standard error; CI = confidence interval; LL = lower limit; UL = upper limit; PROMIS = Patient-Reported Outcomes Measurement Information System; * p < 0.05, *** p < 0.001.

.

Predicted marginal means and 95% confidence intervals are reported to convey precision, and findings, including the observed interaction between sexual orientation and heavy episodic drinking, are interpreted as preliminary and hypothesis-generating, warranting replication in larger samples of sexual minoritized young adults.

3. Results

3.1. Sample Distribution and Bivariate Comparisons

The weighted mean age was 23.6 years in the analytic sample of 723 young adults, and the jackknife standard error (SE) was 0.05. Just over half of respondents identified as male (52.1%), and the majority identified as straight (84.7%), followed by bisexual (10.8%) and lesbian/gay (4.5%) identities. Slightly more than half of the sample identified as NH White (53.3%), with the remainder identifying as Hispanic (22.3%), NH Asian (11.0%), NH Black (8.6%), or another NH race/ethnicity (5.0%). Approximately one-third of respondents had completed a bachelor’s degree or higher, and household income was broadly distributed across categories. Most respondents lived in households with two or more adults (88.4%).

Nearly half of the sample reported one or more HED occasions (45.5%;

Table 1. Sample distribution of U.S. adults aged 18–29 and comparison by number of heavy episodic drinking (HED) occasions (0 vs. 1+), 2022 and 2024 (N = 723).

	Full Sample	HED Occasions *		p-Value †
		None (n = 394)	1+ (n = 329)
	M (SE) or n (%)	M (SE) or %	M (SE) or %
Age	23.6 (0.05)	23.3 (0.06)	24.0 (0.06)	<0.001
Sex				<0.001
Female	460 (47.9)	50.3	49.7
Male	263 (52.1)	59.8	40.2
Sexual orientation				0.012
Lesbian/gay	39 (4.5)	61.8	38.2
Straight	576 (84.7)	55.1	44.9
Bisexual	108 (10.8)	53.8	46.2
Race/ethnicity				<0.001
Non-Hispanic White	336 (53.3)	50.6	49.4
Non-Hispanic Black	94 (8.6)	72.6	27.4
Hispanic	202 (22.3)	48.7	51.3
Non-Hispanic Asian	60 (11.0)	71.9	28.1
Non-Hispanic Other	31 (5.0)	68.6	31.4
Education level				<0.001
Less than high school	21 (6.1)	37.0	63.0
High school graduate	100 (18.4)	64.7	35.3
Some college	202 (41.4)	59.3	40.7
College graduate or more	400 (34.1)	48.5	51.5
Household income				<0.001
<$20,000	112 (18.3)	47.7	52.3
$20,000–<$35,000	75 (9.9)	54.0	46.0
$35,000–<$50,000	114 (14.6)	60.3	39.7
$50,000–<$75,000	153 (19.4)	64.7	35.3
$75,000–<$100,000	188 (27.1)	59.4	40.6
$100,000+	81 (10.8)	35.0	65.0
Adults in household				0.170
1	157 (11.6)	57.1	42.9
2+	566 (88.4)	55.0	45.0
HINTS Cycle				<0.001
6 (2022)	432 (66.8)	63.7	36.3
7 (2024)	291 (33.2)	38.3	61.7
Smoking Status				<0.001
Current	62 (9.5)	9.8	90.2
Former	52 (7.1)	42.1	57.9
Never	606 (83.5)	61.7	38.3
Personal Health Questionnaire (PHQ)-4 score	3.3 (0.04)	3.12 (0.05)	3.5 (0.06)	<0.001

Note. * Comparisons of weighted means (Ms) with jackknife standard errors (SEs) for continuous variables and weighted percentages for categorical variables. † p-value represents survey weighted one-way analysis of variance (ANOVA) for continuous variables and the Rao-Scott Chi-square test of weighted proportions for categorical variables. HINTS = Health Information National Trends Survey.

). Significant differences by HED status were observed across multiple sociodemographic characteristics. Individuals reporting one or more HED occasions were older on average than those reporting none (p < 0.001). HED at least one time in the past 30 days was also more prevalent among females than males (49.7% vs. 40.2%, p < 0.001) and varied by sexual orientation (p = 0.012), with lower prevalence observed among lesbian/gay (38.2%) respondents compared with straight (44.9%) and bisexual (46.2%) respondents.

HED occasions differed significantly across racial and ethnic groups (p < 0.001), with higher prevalence among Hispanic (51.3%) and NH White (49.4%) respondents and lower prevalence among NH Black (27.4%) and NH Asian (28.1%) respondents. Educational attainment was also associated with HED status (p < 0.001), with higher prevalence among those with less than a high school education (63.0%) and those with a college degree or higher (51.5%). Household income demonstrated a graded association with HED (p < 0.001), with the highest prevalence observed among respondents in the highest income category (65.0%). The number of adults in the household was not significantly associated with HED occasions.

HED prevalence differed by survey cycle, with a greater proportion of respondents reporting one or more HED occasions in 2024 compared with 2022 (p < 0.001). In bivariate comparisons of psychosocial measures, individuals reporting one or more HED occasions had significantly higher psychological distress scores as measured by the PHQ-4 (p < 0.001).

3.2. Weighted Linear Regressions

Table 2. Weighted and adjusted linear regressions predicting PROMIS Social Isolation t-scores of adults in the United States aged 18–29, 2022 and 2024 (N = 723).

	Model 1				Model 2
			95% CI				95% CI
	B	SE	LL	UL	B	SE	LL	UL
Intercept	44.68 ***	0.78	43.14	46.22	44.24 ***	0.79	42.66	45.81
Age	−0.12 ***	0.03	−0.18	−0.07	−0.11 ***	0.03	−0.16	−0.06
Sex (ref. = Male)
Female	1.42 ***	0.22	0.99	1.85	1.44 ***	0.22	1.01	1.88
Sexual identity (ref. = straight)
Lesbian/Gay	5.62 ***	0.58	4.47	6.77	7.88 ***	0.78	6.33	9.44
Bisexual	1.66 ***	0.34	0.99	2.33	1.21 *	0.48	0.26	2.15
Race/ethnicity (ref. = NH White)
Non-Hispanic Black	−3.64 ***	0.63	−4.90	−2.38	−3.54 ***	0.63	−4.79	−2.29
Hispanic	−1.76 ***	0.24	−2.23	−1.29	−1.78 ***	0.24	−2.25	−1.30
Non-Hispanic Asian	1.37 *	0.48	0.41	2.32	1.55 ***	0.49	0.58	2.52
Non-Hispanic Other	4.87 ***	0.52	3.84	5.90	4.81 ***	0.52	3.77	5.84
Education level (ref. = college graduate or more)
Less than high school	−1.61 *	0.59	−2.78	−0.44	−1.70 *	0.60	−2.89	−0.51
High school graduate	0.16	0.29	−0.43	0.74	0.34	0.30	−0.26	0.94
Some college	−0.19	0.27	−0.72	0.34	−0.12	0.27	−0.65	0.42
Household income (ref. = $100,000+)
<$20,000	2.12 ***	0.51	1.10	3.14	2.11 ***	0.52	1.09	3.13
$20,000–<$35,000	−1.67 ***	0.48	−2.63	−0.71	−1.75 ***	0.49	−2.71	−0.78
$35,000–<$50,000	1.18 *	0.52	0.15	2.21	1.00	0.52	−0.02	2.03
$50,000–<$75,000	−0.58	0.52	−1.61	0.45	−0.76	0.51	−1.76	0.25
$75,000–<$100,000	−0.18	0.48	−1.13	0.77	−0.21	0.48	−1.16	0.74
Adults in household (ref. = 2+)
1	0.42	0.25	−0.08	0.92	0.42	0.26	−0.09	0.94
HINTS Cycle (ref. = 7[2024])
6 (2022)	0.73 *	0.30	0.14	1.33	0.79 *	0.30	0.19	1.39
Smoking Status (ref. = never)
Current	2.78 ***	0.31	2.16	3.40	2.73 ***	0.31	2.11	3.35
Former	−0.99 *	0.38	−1.73	−0.24	−1.02 *	0.38	−1.78	−0.27
Personal Health Questionnaire (PHQ)−4 score	2.11 ***	0.06	2.00	2.21	2.11 ***	0.05	2.00	2.22
Heavy episodic drinking (HED) occasions (ref. = 0)
1+	−1.71 ***	0.20	−2.10	−1.32	−1.54 ***	0.21	−1.96	−1.11
Interaction (ref. = HED = 1+ × straight)
HED = 1+ × lesbian/gay					−5.77 ***	0.98	−7.71	−3.84
HED = 1+ × bisexual					0.97	0.55	−0.12	2.06

Note: B = unstandardized regression coefficient; SE = jackknife standard error; CI = confidence interval; LL = lower limit; UL = upper limit; PROMIS = Patient-Reported Outcomes Measurement Information System; * p < 0.05, *** p < 0.001.

presents weighted, adjusted linear regression models estimating associations with perceived social isolation, measured by PROMIS Social Isolation t-scores, among U.S. adults aged 18–29. Model 1 included main effects only, while Model 2 added an interaction between sexual orientation and HED to assess effect modification. In Model 1, several sociodemographic factors were significantly associated with social isolation. Older age was associated with lower social isolation scores (B = −0.12, SE = 0.03, p < 0.001), whereas females reported higher isolation than males (B = 1.42, SE = 0.22, p < 0.001). Compared with straight respondents, lesbian/gay (B = 5.62, SE = 0.58, p < 0.001) and bisexual individuals (B = 1.66, SE = 0.34, p < 0.001) reported significantly higher social isolation. NH Black (B = −3.64, SE = 0.63, p < 0.001) and Hispanic respondents (B = −1.76, SE = 0.24, p < 0.001) reported lower social isolation than NH White respondents, while NH Asian (B = 1.37, SE = 0.48, p < 0.05) and NH other respondents (B = 4.87, SE = 0.52, p < 0.001) reported higher isolation. Lower household income was associated with higher social isolation, particularly among respondents earning less than $20,000 annually (B = 2.12, SE = 0.51, p < 0.001) compared to those earning over $100,000. Interestingly, lower educational attainment (less than high school) was associated with lower isolation scores relative to college graduates (B = −1.61, SE = 0.59, p < 0.05). Survey cycle was also significant with respondents in the 2022 cycle reporting slightly higher isolation than those surveyed in 2024 (B = 0.73, SE = 0.30, p < 0.05). Compared to never smokers, current smokers reported significantly higher perceived social isolation (B = 2.78, SE = 0.31, p < 0.001), and former smokers reported significantly lower social isolation (B = −0.99, SE = 0.38, p < 0.05). Psychological distress was strongly and positively associated with social isolation, with each one-point increase in PHQ-4 score corresponding to a substantial increase in PROMIS Social Isolation t-score (B = 2.11, SE = 0.06, p < 0.001). HED was inversely associated with social isolation; respondents reporting one or more HED occasions had significantly lower isolation scores (B = −1.71, SE = 0.20, p < 0.001) compared with those reporting none.

Model 2 results were largely consistent with Model 1 after inclusion of the interaction term (Table 2). The main effect of HED remained significant, though slightly attenuated (B = −1.54, SE = 0.21, p < 0.001). A significant interaction was observed between HED occasions and lesbian/gay sexual identity. Among lesbian/gay respondents, reporting one or more HED occasions was associated with substantially lower social isolation compared with straight respondents who reported HED (B = −5.77, SE = 0.98, p < 0.001). In contrast, the interaction between HED and bisexual identity was not statistically significant, indicating that the association between HED and social isolation did not differ between bisexual and straight respondents.

Figure 1 presents model-adjusted predicted PROMIS Social Isolation t-scores and 95% confidence intervals from Model 2 by sexual orientation and HED status among U.S. adults aged 18–29. Social isolation is shown separately for lesbian/gay, straight, and bisexual respondents, with light grey bars representing no HED and dark grey bars representing one or more HED occasions. The figure illustrates the interaction between sexual orientation and HED, with predicted values indicating that HED was associated with lower social isolation scores specifically among lesbian/gay respondents. Among straight respondents, scores were modestly higher among those reporting no HED, though differences were substantially smaller than those observed for lesbian/gay respondents. Among bisexual respondents, predicted scores were similar across drinking groups, with overlapping confidence intervals, indicating little variation by HED status. Simple effects analyses confirmed that the inverse association between HED and social isolation was unique and most pronounced in the lesbian/gay group.

3.3. Sensitivity Analyses

To interpret the interaction within subgroups, models were re-leveled so that each sexual orientation group served as the reference (Appendix A Table A1). Predicted marginal means and within-group differences indicate that HED was associated with lower perceived social isolation only among lesbian/gay respondents (B = −6.74, SE = 1.14, p < 0.001), while no significant associations were observed among bisexual or straight respondents.

Multiple imputation was used to account for the missing data listwise deleted from the complete case analysis. In the complete-case analysis, the interaction between HED and sexual orientation was significant for lesbian/gay participants, indicating lower social isolation scores among those reporting HED compared with straight participants (B = −5.77, SE = 0.98, p < 0.001), while the interaction for bisexual participants was not significant (B = 0.97, SE = 0.55, p = 0.078; Table 2). After accounting for missing data through multiple imputation, the interaction terms for both lesbian/gay and bisexual participants were no longer statistically significant (B = −3.19, SE = 2.58, p = 0.217; B = −0.37, SE = 1.65, p = 0.822, respectively; Table A2). This attenuation likely reflects increased uncertainty due to the inclusion of cases with previously missing data, which widened confidence intervals and reduced statistical significance. The direction of the interaction effects remained consistent with the complete-case estimates, suggesting that the observed differences may still exist but are less certain once missing data are properly addressed.

4. Discussion

Using nationally representative data from the 2022 and 2024 cycles of the HINTS, this study examined associations between HED and perceived social isolation among U.S. young adults aged 18–29, with a particular focus on differences by sexual orientation. Consistent with minority stress theory and prior research, lesbian/gay and bisexual young adults reported higher levels of perceived social isolation than straight peers, even after adjustment for sociodemographic characteristics and psychological distress. Psychological distress was strongly and positively associated with perceived social isolation across models, underscoring the close linkage between mental health symptom burden and subjective social disconnection during young adulthood. In contrast, HED occasions were inversely associated with perceived social isolation overall, a finding that highlights the complex and potentially bidirectional relationship between alcohol use and social well-being in this life stage.

The primary contribution of this study is the identification of a significant interaction between sexual orientation and HED occasions. Among lesbian/gay young adults, reporting one or more HED occasions was associated with substantially lower perceived social isolation compared with lesbian/gay peers who reported no HED. This pattern was not observed among bisexual young adults, for whom perceived social isolation did not differ by HED status, nor was it as pronounced among straight respondents. These findings suggest that the social meaning and psychosocial correlates of HED may differ across sexual identity groups, with HED potentially reflecting social integration or community engagement rather than solely maladaptive coping for some sexual minoritized individuals.

The inverse association between HED and perceived social isolation among lesbian/gay young adults may be understood within the context of historically alcohol-centered LGBTQ+ social spaces. In the face of widespread structural stigma, exclusion from mainstream institutions, and limited access to affirming public spaces, bars, clubs, and nightlife venues have long served as critical sites of identity affirmation, community formation, and mutual support for lesbian and gay individuals (Beemyn, 2013; Wolfe, 1992). Sociological and historical scholarship highlights how queer nightlife environments function not merely as leisure spaces but as socially meaningful contexts where belonging, visibility, and collective identity are actively produced (Ghaziani, 2022, 2025; Hilderbrand et al., 2025). Within these environments, alcohol use is often normative and intertwined with social participation, such that engagement in drinking-related activities may coincide with greater perceived social integration and reduced feelings of isolation (Hunt et al., 2019). Empirical work further suggests that LGBTQ+ bars may operate as safe haven, particularly in socially conservative contexts, offering opportunities for connection and emotional support that are less readily available elsewhere (Croff et al., 2017). In this context, abstaining from or avoiding alcohol-centered social spaces may inadvertently constrain access to key avenues of social interaction and community belonging, potentially contributing to higher perceived social isolation among lesbian/gay individuals who do not engage in HED.

In contrast to lesbian/gay young adults, bisexual young adults may experience distinct social and structural vulnerabilities that limit the extent to which alcohol-related social participation translates into perceived connectedness. Bisexual individuals often face marginalization from both heterosexual and gay/lesbian communities, as well as identity invisibility and erasure, which can reduce access to affirming social spaces and community support (Alessi, 2014; McCormack, 2023; Pachankis et al., 2020). Consequently, participation in alcohol-centered social environments may not confer the same sense of belonging or reductions in perceived isolation observed among lesbian/gay peers. Instead, HED among bisexual young adults may be more strongly linked to coping with minority stress or normative young adult drinking behaviors rather than social integration, yielding no clear association with perceived social isolation (Kalb et al., 2018; Talley et al., 2016). This pattern demonstrates the heterogeneity within sexual minoritized populations and highlights the importance of considering identity-specific social and structural contexts when examining behavioral and psychosocial outcomes.

The differing patterns observed between lesbian/gay and bisexual adults warrant careful interpretation in light of theory and prior research. Minority stress frameworks suggest that bisexual individuals may experience unique stressors, such as identity invalidation and stigma from both heterosexual and lesbian/gay communities, that are distinct from those faced by lesbian and gay individuals (Feinstein & Dyar, 2023). These differences may contribute to heterogeneous health behavior patterns and could help explain why bisexual adults in our study did not significantly differ from heterosexual adults. Lesbian and gay populations may benefit from more cohesive community norms and greater exposure to targeted health interventions, which have been linked to differences in substance use behaviors and healthcare engagement (Lehavot et al., 2017). Intersectional stigma perspectives emphasize that overlapping social identities and structural factors shape health in complex and non-uniform ways across sexual minoritized subgroups (Lutete et al., 2022). Qualitative evidence further highlights substantial variability in substance use motivations and decision-making among bisexual and other bi+ individuals, underscoring within-group heterogeneity (Mata et al., 2023). Broader contextual factors, including socioeconomic status and psychological distress, may also differentially influence these associations (McGarrity, 2014; Singh et al., 2022). However, the lack of significant differences between bisexual and heterosexual adults in the present study may also reflect sampling variability or limited statistical power. These findings highlight the importance of disaggregating sexual minoritized subgroups and situating observed patterns within both minority stress theory and broader structural contexts.

HED was associated with lower perceived social isolation in the overall sample, even after adjustment for psychological distress. This finding aligns with a growing body of literature indicating that alcohol use during young adulthood cannot be understood solely as an indicator of psychosocial vulnerability (Quigley & Marlatt, 1996; Schulenberg & Maggs, 2002; White & Jackson, 2004). Emerging adults often engage in alcohol consumption within peer-oriented contexts where drinking serves both social and recreational functions, facilitating social bonding, identity exploration, and perceived connectedness (Brook & Willoughby, 2016; Mathes Winnicki et al., 2023; Villar et al., 2022). Consequently, alcohol use may simultaneously function as a risk behavior and a marker of social engagement, highlighting the dual roles of drinking in young adulthood. These complexities complicate public health narratives that frame HED exclusively as maladaptive and underscore the importance of considering social context and subgroup heterogeneity when evaluating alcohol-related outcomes (Brown et al., 2008; Gates et al., 2016).

The variation in perceived social isolation by racial/ethnic group relative to their NH White peers is noteworthy and warrants further consideration. Prior research suggests that resilience processes within minoritized communities, including stronger familial bonds, community cohesion, and culturally grounded norms around social connectedness, may serve as protective factors against loneliness and isolation (Marks et al., 2020; Rivas-Drake et al., 2022). Additionally, ethnic-racial identity and collective coping strategies have been identified as important sources of resilience in the face of structural adversity and discrimination, which may buffer against perceived isolation (Rivas-Drake et al., 2022). Empirical studies have also documented variation in loneliness and distress across racial and ethnic groups, with some evidence indicating lower reported loneliness among Black adults in certain contexts, potentially reflecting differences in social network structures and support systems (Nápoles et al., 2023). At the same time, research among sexual and gender minority youth of color highlights the complex interplay of risk and resilience, suggesting that despite exposure to multiple stressors, protective social and cultural resources may mitigate adverse psychosocial outcomes (Jowaheer, 2023; Vance et al., 2021). These findings underscore the importance of situating racial and ethnic differences in perceived isolation within broader frameworks of resilience and structural context, while recognizing the need for future research to more fully disentangle the social, cultural, and contextual mechanisms underlying these patterns, particularly in the context of HED.

Although we adjusted for available covariates where possible, unmeasured use of tobacco or other drugs may influence the observed associations. This is particularly relevant in the context of sexual minoritized populations, where practices such as chemsex, defined as the use of psychoactive substances in sexual contexts, have been documented and are shaped by complex social, cultural, and psychological drivers (García-Pérez et al., 2022; Jaspal, 2022; Mundy et al., 2025). However, it is important to situate these considerations within the broader public health landscape. Alcohol remains the most widely used psychoactive substance globally and contributes a substantially greater burden of disease and mortality than other forms of substance use (World Health Organization, 2024; Xie et al., 2025). Thus, while chemsex and other drug use may represent important contextual or subgroup-specific risk factors, alcohol use continues to represent a more pervasive driver of population-level health disparities. Future research would benefit from more comprehensive measurement of polysubstance use to better disentangle these overlapping influences.

4.1. Limitations and Strengths

Several limitations should be considered when interpreting these findings. First, the cross-sectional design precludes causal inference, and it is not possible to determine whether HED reduces perceived social isolation, whether lower isolation increases opportunities for HED, or whether both are shaped by unmeasured factors such as social network characteristics. Second, HED was assessed using a dichotomous measure of one or more occasions, which does not capture frequency, intensity, or drinking context. Third, sexual orientation categories were limited by sample size, precluding examination of additional identities or intersections with gender identity. Fourth, gender identity was not assessed restricting the ability to interpret the experiences of gender diverse populations. Fifth, all measures were self-reported and may be subject to recall or social desirability bias. Sixth, a key limitation of this study is that the measure of HED captures frequency but does not distinguish between social and solitary drinking contexts. While HED may co-occur with social engagement for some individuals, it may also occur in non-social settings. Future research should incorporate measures that explicitly assess the social context of drinking to better understand the relationship between alcohol use and perceived social isolation. Finally, while HINTS provides national representativeness, its primary focus is health information behaviors, which may limit the depth of alcohol-related measures.

Despite these limitations, this study has several important strengths, including the use of recent, nationally representative data; validated measures of perceived social isolation; and analytic approaches that explicitly test effect modification by sexual orientation. By focusing on perceived social isolation rather than objective social indicators alone, this study highlights a critical psychosocial dimension that may shape both mental health and substance use trajectories among young adults.

4.2. Intervention and Practice Implications

These findings have potential implications for intervention and practice, though they should be interpreted with caution. Efforts to reduce alcohol-related harm among sexual minoritized young adults may benefit from considering the social contexts in which alcohol use occurs, particularly for lesbian/gay individuals who may rely on alcohol-centered spaces for community and connection. While it is possible that interventions focused solely on reducing alcohol consumption could have unintended social consequences, this interpretation remains speculative and is not directly tested in the present study. Expanding access to alcohol-free, identity-affirming social spaces and strengthening community-based sources of connection may represent promising complementary strategies; however, additional research is needed to better understand how changes in alcohol use intersect with social connectedness and to inform evidence-based policy and intervention approaches.

4.3. Conclusions

This study demonstrates that the association between HED and perceived social isolation among young adults is not uniform but varies meaningfully by sexual orientation. Specifically, HED was associated with lower perceived social isolation among lesbian/gay young adults, whereas no such association was observed for bisexual peers. These findings highlight the importance of moving beyond one-size-fits-all models of risk and underscore the need to consider the broader social and community contexts in which alcohol use occurs. For lesbian/gay young adults, participation in alcohol-centered social spaces may provide opportunities for identity affirmation, peer connection, and community engagement, potentially mitigating feelings of isolation. In contrast, bisexual individuals often face unique structural and social vulnerabilities, including marginalization from both heterosexual and gay/lesbian communities and reduced access to affirming social spaces, which may limit the extent to which alcohol-related participation confers social benefits. These results emphasize that the psychosocial significance of alcohol use in young adulthood is shaped by intersecting factors of identity, social environment, and subgroup-specific experiences, rather than representing a universally maladaptive behavior.