Abstract
-
Purpose
Adolescents with same-sex or both-sex sexual experience face disproportionate health risks, yet research has predominantly examined single health problems rather than their multidimensional patterns. This study employed latent class analysis (LCA) to identify distinct health typological classes among Korean adolescents and characterize the distribution of sexual behavior patterns across these classes.
-
Methods
We analyzed 278,066 middle- and high-school students in the 2013–2016 Korea Youth Risk Behavior Survey. The LCA used 17 indicators encompassing sexual behavior patterns, health behaviors, health perceptions, and mental health. Complex-sample weighted analyses examined associations between latent classes and sociodemographic characteristics, sexual behaviors, and diseases.
-
Results
Four distinct classes were identified: mental health high-risk group (10.1%), substance use high-risk group (6.5%), average group (36.3%), and healthy group (47.1%). Adolescents with same-sex or both-sex sexual experience were disproportionately concentrated in the two high-risk classes, comprising 3.3% of the mental health high-risk group and 4.1% of the substance use high-risk group, compared with only 0.3% and 0.5% of the average group and the healthy group. The mental health high-risk group exhibited extremely elevated prevalence rates of suicidal ideation (96.2%), depressive symptoms (88.5%), and suicide attempts (27.5%), while the substance use high-risk group showed high rates of smoking (82.4%) and alcohol use (84.6%) but better mental health.
-
Conclusion
Adolescents with same-sex or both-sex sexual experience demonstrate health inequality through disproportionate classification in the mental health high-risk group and the substance use high-risk group. The findings underscore the need for integrated interventions combining mental health support and substance use prevention.
-
Key words: Adolescent; Health behavior: Latent class analysis; Mental health; Sexual behavior
INTRODUCTION
The proportion of sexual minorities in the global population has been steadily increasing [
1,
2]. A global survey of 18,515 individuals across 26 countries in 2024 found that approximately 11% of respondents identified as lesbian, gay, bisexual, transgender, or other sexual minorities, while the proportion was markedly lower in South Korea, at 5% [
2]. This large discrepancy may reflect underreporting, since South Korea lacks a nationwide survey system for sexual orientation and gender identity, and social prejudice makes such disclosures difficult, limiting accurate population estimation [
3].
Adolescence is a period of rapid physical, mental, and sexual changes, representing a critical developmental stage when sexual orientation emerges and identity formation begins [
4]. During this process, sexual-minority adolescents often navigate significant identity confusion and psychological uncertainty while exploring their sexual orientation [
5]. Coupled with these developmental vulnerabilities, this group encounters a dual burden of minority stress: (1) distal (external) stressors, such as bullying and family rejection, and (2) proximal (internalized) stressors, including identity concealment and internalized stigma [
6]. This situation results in sexual-minority adolescents experiencing substantial health inequalities. Research has consistently shown that the vulnerability across various health risk indicators is higher among sexual-minority adolescents than among their heterosexual peers. Specifically, sexual-minority adolescents demonstrate elevated risks of smoking [
7,
8], alcohol use [
7,
8], drug use [
7-
9], risky sexual behaviors [
10], sleep disorders [
11], depressive symptoms [
5,
7], and suicidal ideation and suicide attempts [
5,
7,
8]. Large-scale analyses of US adolescent survey data found that sexual-minority youth had significantly elevated odds of hopelessness (adjusted odds ratio [aOR], 4.6), suicidal ideation (aOR, 3.2), and suicide planning (aOR, 2.0) compared with their heterosexual peers, along with higher prevalence rates of cigarette use (15% vs. 7.8%), marijuana use (31.2% vs. 20.2%), and cocaine use (8.4% vs. 3.9%) [
8].
Recent research has also produced evidence of serious disparities in physical health between sexual-minority and heterosexual adolescents, with the former exhibiting greater physical health problems such as chronic pain [
12,
13] and elevated inflammatory markers [
14]. These patterns persist into adulthood, with sexual-minority adults showing higher prevalence rates of diabetes [
15], hypertension [
16], cardiovascular disease [
17], and cancer [
18,
19] than their heterosexual peers.
International research has increasingly employed latent class analysis (LCA) to examine multidimensional health patterns among sexual-minority adolescents. An LCA study of sexual-minority youth seeking crisis services identified four distinct classes of co-occurring behavioral health risks, with gender-nonconforming youth showing significantly higher odds of membership in high-risk classes [
20]. An analysis of over 119,000 US adolescents similarly found that gay, lesbian, and bisexual youth were disproportionately represented in polysubstance use classes compared with their heterosexual peers [
21]. These patterns extend to the co-occurrence of sexual violence, substance use, and mental health problems, with sexual-minority youth demonstrating elevated risk across multiple domains [
22]. In contrast, Korean research remains predominantly limited to single-variable approaches [
5,
7]. Accordingly, in this study we employed LCA as a person-centered method optimized for uncovering hidden group structures within categorical data. Unlike variable-centered approaches such as regression analysis that estimate relationships between variables across the entire population, LCA can capture within-group diversity, thereby providing essential information for identifying vulnerable groups where health risks are concentrated among adolescents and for facilitating the development of tailored intervention strategies [
23]. This approach is particularly relevant to nursing practice, since school and community health nurses—who maintain ongoing contact with adolescents—can utilize latent class profiles to identify high-risk individuals requiring integrated interventions rather than using fragmented approaches targeting isolated health problems.
The studies reviewed above assessed the sexual-minority status based on self-reported sexual identity or orientation, whereas South Korea lacks nationally representative surveys that have directly measured adolescent sexual orientation [
3]. Instead, the Korea Youth Risk Behavior Survey (KYRBS) conducted during 2013–2016 included questions about sexual-intercourse partners, and so enables the classification of sexual behavior patterns. Given that a same-sex or both-sex sexual experience does not necessarily equate to a sexual-minority identity [
24], the present study examined health patterns associated with sexual behavior patterns rather than sexual identity. However, it should be noted that the KYRBS partner-sex response options were modified after 2017, and so the 2013–2016 data remain the most-recent nationally representative data available for such an analysis.
This study aimed to systematically classify types of sexual behavior patterns among adolescents and their multidimensional health characteristics, and to identify the features and associated factors of each latent class. This information would support the development of practical policies and programs for improving the health status of adolescents.
METHODS
Ethical statements: This study obtained an exemption from the Institutional Review Board (IRB) at Daejeon University (IRB No. 1040647-202509-HR-003-03) as it involves the utilization of secondary data with anonymity.
1. Study Design
This study applied secondary data analyses to raw data obtained in the KYRBS performed from 2013 to 2016, which represented the 9th to 12th implementations of this survey. A cross-sectional descriptive design was used to systematically analyze the relationships between adolescents’ sexual behavior patterns and multidimensional health behaviors through LCA. The reporting of this study followed the guidelines outlined in the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines [
25].
The KYRBS is a nationwide anonymous self-administered online survey jointly conducted annually by the Ministry of Education and the Korea Disease Control and Prevention Agency that since 2005 has assessed adolescent health behaviors and provided foundational data for the development of relevant policies. The 2013–2016 surveys targeted middle- and high-school students nationwide, with approximately 800 schools and 68,000–75,000 students across 17 provinces and metropolitan cities selected through stratification, sample allocation, and sampling processes. The KYRBS data are publicly available for academic research purposes, with all individual identifiers removed prior to public release. The final analysis involved 278,066 survey participants.
3. Measures and Variable Definitions
This study analyzed the responses to questions regarding sexual behavior patterns, sociodemographic characteristics, health behaviors, health perceptions, mental health, sexual behaviors, and diseases. Variable values were defined based on the KYRBS indicator definitions and variable classifications in previous research [
5,
7].
1) Sexual behavior patterns
Sexual behavior patterns were assessed based on questions about sexual-intercourse partners among those with sexual experience. Participants were categorized into four groups: no sexual experience, opposite-sex experience only, same-sex experience only, and experience with both sexes.
2) Sociodemographic characteristics
The analyzed sociodemographic variables included sex (male or female), school grade (middle-school grades 1–3 or high-school grades 1–3), city size (metropolitan, small-to-medium city, or rural area), and living arrangements (with family, with relatives, boarding/self-boarding, dormitory, or welfare facility). Parental education levels were categorized into middle-school graduation or below, high-school graduation, college graduation or above, or unknown. A multicultural family status was determined by the parents’ birthplaces. Subjective academic performance and economic status were each classified into five levels: high, upper-middle, middle, lower-middle, and low. Weekly allowance, part-time employment, and hospital treatment due to violence victimization were also assessed.
3) Health behaviors
Current smoking was defined as smoking at least once during the past 30 days, and current alcohol use was defined as consuming at least one alcoholic drink during the past 30 days. Breakfast skipping was defined as being present when it occurred at least 5 days per week. The consumption of fast food and vegetables was assessed based on weekly frequency. Physical activity was defined as engaging in activity intense enough to increase the heart rate for at least 60 minutes daily on at least 5 days per week. The body-weight-control effort was categorized into no effort, weight loss, weight gain, or weight maintenance.
4) Health perceptions
Subjective health perception and body-image perception were assessed on 5-point scales and recategorized into three levels for the analysis: healthy, average, and unhealthy for health perception; and thin, average, and overweight for body image.
5) Mental health
Subjective happiness and perceived stress were assessed on 5-point scales and recategorized into three levels: happy, average, and unhappy for happiness; and high, moderate, and low for stress. Subjective sleep sufficiency was assessed according to whether the sleep duration was sufficient for fatigue recovery, and categorized into sufficient, average, or insufficient. Depressive symptoms were defined as feeling sad or hopeless enough to stop performing daily activities for at least 14 days consecutively during the past 12 months. Suicidal ideation, suicide planning, and suicide attempts during the past 12 months were classified into yes or no.
6) Sexual behaviors
Age at first sexual intercourse was categorized into elementary school or earlier, middle school, or high school. Sexual intercourse after consuming alcohol and the use of contraceptives were assessed among students with sexual experience, with contraceptive use categorized into always, sometimes, rarely, or never. Pregnancy experience was assessed among female students with sexual experience, and sexually transmitted infection (STI) experience was assessed among all sexually experienced students, with both classified into yes or no.
7) Diseases
Since asthma, allergic rhinitis, and atopic dermatitis were the only chronic conditions assessed in the survey, they were the only diseases included in this analysis. Participants were classified into yes or no based on the history of physician-confirmed diagnosis for these specific conditions.
4. Data Analyses
Data analyses were performed using R software ver. 4.3.0 (The R Foundation for Statistical Computing) considering the complex sample design characteristics of the KYRBS. LCA was conducted using the poLCA package. Because this package does not support sampling weights when estimating latent classes [
26], the class structure was derived from unweighted data. Although unweighted estimations may affect class solutions when the sampling design is informative [
27], the survey design features of stratification, clustering, and sampling weights were incorporated in all subsequent analyses of class prevalence, characteristics, and associations using the R survey package. In all analyses, statistical significance was considered to be present when
p<.05.
The data analyses involved four main steps. First, analyses of complex-sample descriptive statistics were conducted to determine the distributions of the participants’ characteristics. Second, LCA was conducted to derive latent patterns of adolescents’ sexual behavior patterns and health characteristics (health behaviors, health perceptions, and mental health variables) using 17 variables (
Supplement 1). The poLCA package was used to fit models with two to six latent classes, and the optimal model was selected by considering the Bayesian information criterion (BIC), Akaike information criterion (AIC), and entropy index [
23]. Third, complex-sample weighted analyses were used to characterize the derived latent classes. Class-specific distributions were calculated as weighted proportions. Fourth, Rao-Scott χ
2 tests were used to examine relationships between latent classes and exogenous variables including sociodemographic characteristics, sexual behaviors, and diseases.
RESULTS
1. Participant Characteristics
The sociodemographic characteristics, sexual behavior patterns, health behaviors, health perceptions, mental health, sexual behaviors, and diseases of the 278,066 study participants are presented in
Table 1. All percentages are weighted percentages that reflect the complex sample design.
The sample included 52.2% males. The distribution by school grade was relatively even across all grades. Most students (95.8%) lived with their families. Regarding sexual behavior patterns, 94.9% had no sexual experience, 4.1% had opposite-sex experience only, 0.6% had same-sex experience only, and 0.4% had experience with both sexes.
2. Selection of Optimal Latent Class Model
Models with two to six latent classes were fitted (
Table 2). Although both the AIC and BIC decreased as more classes were added, the four-class solution was selected based on this providing an adequate entropy index (0.855), conceptually distinct class profiles, and consistency with previous LCA studies that similarly identified four-class solutions for adolescent emotional and behavioral problems [
28-
30].
There were distinct differences in sexual behavior patterns, health behaviors, health perceptions, and mental health among the four derived latent classes (
Table 3).
1) Class 1: mental health high-risk group (10.1%)
Class 1 showed the most-vulnerable mental health indicators, with markedly higher rates of depressive symptoms (88.5%), suicidal ideation (96.2%), suicide planning (39.9%), and suicide attempts (27.5%) than for the other classes. High stress (85.0%), insufficient sleep (63.2%), and unhappiness (40.7%) were also prominent. Among those with sexual experience, 8.0% had opposite-sex experience only, 1.9% had same-sex experience only, and 1.4% had experience with both sexes. Health behaviors showed moderate risk levels, with smoking at 15.7% and alcohol use at 27.1%.
2) Class 2: substance use high-risk group (6.5%)
Class 2 exhibited the most-severe substance use behaviors, with markedly higher rates of current smoking (82.4%) and alcohol use (84.6%) than for the other classes. The sexual experience rate was 33.1%, with the rates of same-sex (2.3%) and both-sex (1.8%) experience being the highest among all classes. However, mental health indicators were more favorable for this class, with 57.0% feeling happy and low rates of suicide-related indicators: 6.0% for suicidal ideation, 1.4% for suicide planning, and 0.9% for suicide attempts. A positive subjective health perception was present in 72.4% of the participants.
3) Class 3: average group (36.3%)
Class 3 showed average characteristics across most indicators. This class had a very high rate of no sexual experience (98.4%), and the lowest rate of same-sex or both-sex experience (0.3%). Smoking (1.2%) and alcohol use (11.4%) rates were also low, but notable rates of high stress (65.3%) and insufficient sleep (57.8%) were present.
4) Class 4: healthy group (47.1%)
Class 4 was the largest group and showed the most-favorable health indicators. Feeling happy (93.4%) and good subjective health perception (88.9%) were prominent. Mental health indicators were the most favorable, with rates of depressive symptoms and suicidal ideation of 5.5% and 0.8%, respectively. This class also had a very high rate of no sexual experience (97.7%), and a rate of same-sex or both-sex experience (0.5%) that was similarly low to that for Class 3. Smoking (1.8%) and alcohol use (7.7%) rates were also low.
4. Associations between Latent Classes and Exogenous Variables
Significant differences were found across all sociodemographic variables, sexual behaviors, and diseases (
p<.001). The detailed distributions are presented in
Tables 4 and
5. Class 2 had markedly high proportions of males (81.5%) and high-school students (81.6%). Class 4 had the highest proportions of upper-level academic performance (42.5%) and economic status (41.5%). The rate of part-time employment was highest in Class 2 (49.5%) and lowest in Class 4 (8.3%). The rate of violence victimization requiring hospital treatment was highest in Class 1 (9.5%) and lowest in Class 4 (1.3%). Among sexually experienced adolescents, those in Class 2 had the highest proportion of first sexual intercourse during high school (58.4%), while Classes 1 and 4 had high proportions during elementary school (33.6% and 35.3%, respectively). The rate of STI experience was highest in Class 1 (16.4%); this class also showed the highest prevalence rates of asthma (12.1%), allergic rhinitis (37.2%), and atopic dermatitis (27.6%).
DISCUSSION
This study applied LCA to a large representative sample (N=278,066) from the 2013–2016 KYRBS to identify associations between sexual behavior patterns and multidimensional health characteristics. Four distinct classes were identified: mental health high-risk group (10.1%), substance use high-risk group (6.5%), average group (36.3%), and healthy group (47.1%). This four-class structure is broadly consistent with previous LCA research identifying four subgroups of adolescent emotional and behavioral problems, typically distinguishing between internalizing, externalizing, low-risk, and comorbid or maladjusted patterns [
28-
30]. The mental health high-risk group shares features with internalizing classes, the substance use high-risk group with externalizing classes, and the healthy group with low-problem classes. The average group, characterized by moderate stress and sleep insufficiency, did not correspond to the comorbid classes reported elsewhere but may represent a subclinical risk group specific to the indicators used in this study.
The rates of suicidal ideation (96.2%), depressive symptoms (88.5%), and suicide attempts (27.5%) in the mental health high-risk group were substantially higher than those in the overall sample. The extremely high prevalence rates of these indicators reflect the methodology of LCA clustering individuals with similar response patterns onto specific indicator variables. Similar methodological patterns have been reported elsewhere, with an LCA of undergraduate students finding that the high-risk class showed rates of suicidal ideation, prior suicide attempts, and depressive symptoms of 66.0%, 22.5%, and 97.6%, respectively [
31], illustrating that LCA typically yields extreme prevalence rates within identified risk classes. However, the possibility of overreporting in anonymous adolescent surveys cannot be excluded [
32]. These findings should be interpreted as identifying a subgroup where mental health vulnerabilities are concentrated, rather than as population-level prevalence estimates. Adolescents with same-sex or both-sex sexual experience comprised 3.3% of this group, which was markedly higher than their prevalence rates in the average group (0.3%) and the healthy group (0.5%). These findings are broadly consistent with international research indicating that adolescents with same-sex attraction are 5 to 7 times more likely to experience suicidal ideation and mental health problems than their heterosexual peers [
33], although that study assessed sexual orientation through self-reported attraction rather than sexual behavior patterns.
The mental health high-risk group demonstrated co-occurring vulnerabilities beyond mental health, including high rates of violence victimization (9.5%), STI experience (16.4%), and disease prevalence (12.1%, 37.2%, and 27.6% for asthma, allergic rhinitis, and atopic dermatitis, respectively). The elevated prevalence of these allergic conditions may be partly explained by the well-documented pathway through which chronic psychosocial stress dysregulates immune function, shifting the balance toward allergic inflammatory responses associated with asthma, allergic rhinitis, and atopic dermatitis [
34]. These findings underscore the need for integrated interventions that address mental health, physical health, and prevention of violence victimization simultaneously, particularly for those in this high-risk group who report same-sex or both-sex partners.
The substance use high-risk group was characterized by prominent smoking (82.4%) and alcohol use (84.6%) and had the highest proportion of those with same-sex or both-sex sexual experience (4.1%). The demographic profile of this group (81.5% males, 81.6% high-school students, and 49.5% in part-time employment) suggests greater developmental autonomy and economic resources that could facilitate substance access. The high rate of sexual intercourse after consuming alcohol (45.6%) indicates that substance use and risky sexual behaviors were tightly linked in the behavior patterns in this group. Despite these risk behaviors, this group exhibited relatively preserved mental health (57.0% feeling happy and 6.0% with suicidal ideation); however, this apparent well-being may be misleading since heavy substance use during adolescence is associated with increased risks of addiction, school dropout, chronic disease, and delayed-onset mental health problems in young adulthood. These findings suggest that substance use and mental health problems may represent temporally distinct pathways of health risk that do not necessarily co-occur during adolescence but may converge over time. Therefore, approaches to support early identification and interventions for this group should focus on substance use reduction and associated risky sexual behaviors, while recognizing that these adolescents might not present with the visible psychological distress typically targeted by mental health screening.
The average group (36.3%) showed moderate characteristics, but the elevated rates of stress (65.3%) and sleep insufficiency (57.8%) indicated latent risks. The very low proportion of adolescents with same-sex or both-sex sexual experience (0.3%) suggests that these individuals are concentrated in higher-risk classes.
The healthy group (47.1%) demonstrated the most-favorable indicators, with a high rate of feeling happy (93.4%), a low rate of depressive symptoms (5.5%), and minimal suicidal ideation (0.8%). However, adolescents with same-sex or both-sex sexual experience were markedly underrepresented, at 0.5%, compared with 5.1% of the overall sample. Notably, 35.3% of the sexually experienced adolescents in this class reported that their first sexual intercourse occurred during elementary school, which was the highest rate across all classes and is implausibly high for consensual activity. This pattern, shared with the mental health high-risk group (33.6%), raises concerns about either mischievous responding in anonymous surveys [
32] or early experience of sexual abuse. These findings illustrate that the “healthy” label reflects a favorable current health status based on the measured indicators rather than the absence of all developmental risks.
The disproportionate representation of adolescents with same-sex or both-sex sexual experience in the high-risk groups may be partly understood within the Korean sociocultural context. Conservative attitudes toward adolescent sexuality are reinforced by Confucian values emphasizing family conformity and traditional social norms [
35], and they may reduce access to comprehensive sexual health education and services for all sexually active adolescents, which would potentially have a greater impact on those whose sexual behaviors deviate from heteronormative expectations. In such an environment, adolescents with same-sex or both-sex sexual experience may face additional barriers to seeking health-related support, which could contribute to the clustering of health risks observed in this study.
Given this clustering of health risks, the clinical and nursing significance of these findings warrants consideration despite their modest absolute proportions (3.3% and 4.1% in the high-risk classes). These values represent overrepresentation rates of 6 to 8 times compared with the healthy group (0.5%), indicating that sexual behavior patterns serve as a meaningful marker for risk identification. These findings have several implications for nursing practice and policy. First, school and community health nurses—who maintain ongoing contact with adolescents—should include the making of sensitive inquiries about sexual experiences in routine health assessments and create safe environments for such disclosures. Second, given the co-occurrence of mental health vulnerabilities and substance use risks within distinct subgroups, integrated interventions combining mental health support and substance use prevention are necessary, rather than fragmented approaches targeting isolated health problems. Third, both schools and community organizations should establish support systems that enable adolescents with same-sex or both-sex sexual experience to access protective factors observed in the healthy group, such as supportive environments and mental health resources. Fourth, questions about sexual behavior patterns should be reintroduced into the KYRBS, since the modifications after 2017 eliminated the only nationally representative data source for monitoring health patterns according to sexual behaviors among Korean adolescents.
This study had several limitations. First, sexual behavior patterns were assessed only among the 5.1% of adolescents with sexual experience, thereby precluding classification of the 94.9% without sexual experience who may include adolescents with minority sexual orientations who are not yet sexually active. This would introduce selection bias, since our findings represent health patterns among sexually active adolescents categorized by the sex of their sexual partners, rather than all adolescents who may identify as sexual minorities. Therefore, the term “sexual behavior patterns” used throughout this study should be interpreted as referring specifically to the sex of sexual partners, and not as equivalent to a sexual-minority identity or sexual orientation. Second, the latent class structure was derived from unweighted data due to software constraints of the poLCA package. When sampling weights are informative, unweighted estimations may yield biased parameter estimates with smaller variances that do not fully reflect sampling variability [
27]. However, the excellent classification accuracy (entropy index=0.855) supports the reliability of the class solution. Nevertheless, future studies using software capable of integrating survey weights into LCA estimations (e.g., Mplus) could further validate these findings. Third, the use of 2013–2016 data meant that recent attitudinal changes were not included in the analyses. Although these data are the most-recent nationally representative data available, the behavioral and social environments of Korean adolescents have changed considerably since that period. For example, the increasing use of e-cigarettes, increased exposure to digital media, and deteriorations in mental health observed following the COVID-19 (coronavirus disease 2019) pandemic may have altered the composition and prevalence of the latent classes identified in this study. Accordingly, the present findings should be interpreted as providing a historical baseline rather than a direct representation of the current adolescent health landscape.
CONCLUSION
The application of LCA to a large representative sample (N=278,066) from the 2013–2016 KYRBS in this study has identified four distinct health classes. Adolescents with same-sex or both-sex sexual experience were disproportionately classified into the mental health and substance use high-risk groups, demonstrating their exposure to complex and severe health risks. These findings underscore the urgent need for integrated interventions and the reinstatement of assessments of sexual behaviors in national surveys. As the first large-scale multidimensional analysis of adolescent health patterns according to sexual behaviors in South Korea, this study has provided foundational data for developing evidence-based policies to address health inequalities.
ARTICLE INFORMATION
Supplementary material
Table 1.General characteristics of study participants (N=278,066)
|
Characteristic |
Category |
No. (%) |
Weighted % |
|
Sociodemographic characteristics |
|
|
|
|
Sex |
Male |
142,132 (51.1) |
52.2 |
|
Female |
135,934 (48.9) |
47.8 |
|
Grade |
Middle school 1st |
45,129 (16.2) |
14.9 |
|
Middle school 2nd |
46,347 (16.7) |
15.8 |
|
Middle school 3rd |
47,728 (17.2) |
16.8 |
|
High school 1st |
46,329 (16.7) |
17.5 |
|
High school 2nd |
46,200 (16.6) |
17.4 |
|
High school 3rd |
46,333 (16.7) |
17.6 |
|
City size |
Rural area |
25,247 (9.1) |
6.4 |
|
Metropolitan area |
123,617 (44.4) |
43.7 |
|
Small-to-medium city |
129,202 (46.5) |
49.9 |
|
Living arrangements |
With family |
264,906 (95.3) |
95.8 |
|
With relatives |
2,524 (0.9) |
0.8 |
|
Dormitory/boarding/self-boarding |
9,207 (3.3) |
2.9 |
|
Welfare facility |
1,429 (0.5) |
0.5 |
|
Father’s educationa)
|
≤Middle school |
8,060 (3.0) |
2.8 |
|
High school |
82,798 (31.0) |
30.3 |
|
≥College |
129,784 (48.6) |
50.7 |
|
Unknown |
45,599 (17.1) |
16.2 |
|
Missing |
11,825 |
- |
|
Mother’s educationa)
|
≤Middle school |
7,114 (2.7) |
2.5 |
|
High school |
103,754 (39.0) |
38.6 |
|
≥College |
112,087 (42.1) |
43.2 |
|
Unknown |
43,965 (16.5) |
15.6 |
|
Missing |
11,146 |
- |
|
Multicultural familya)
|
No |
255,495 (98.9) |
99.1 |
|
Yes |
2,873 (1.1) |
0.9 |
|
Missing |
19,698 |
- |
|
Subjective academic performance |
High |
34,095 (12.3) |
12.1 |
|
Upper-middle |
68,552 (24.7) |
24.6 |
|
Middle |
77,888 (28.0) |
28.1 |
|
Lower-middle |
66,481 (23.9) |
24.0 |
|
Low |
31,050 (11.2) |
11.1 |
|
Subjective economic status |
High |
23,295 (8.4) |
8.5 |
|
Upper-middle |
72,171 (25.9) |
26.3 |
|
Middle |
132,552 (47.7) |
47.4 |
|
Lower-middle |
39,813 (14.3) |
14.1 |
|
Low |
10,235 (3.7) |
3.6 |
|
Weekly allowance (KRW) |
0–9,999 |
89,494 (32.2) |
31.8 |
|
10,000–19,999 |
76,255 (27.4) |
27.2 |
|
20,000–29,999 |
36,844 (13.2) |
13.3 |
|
≥30,000 |
75,473 (27.1) |
27.7 |
|
Part-time work experience |
No |
240,347 (86.4) |
86.4 |
|
Yes |
37,719 (13.6) |
13.6 |
|
Hospital treatment due to violence |
No |
270,635 (97.3) |
97.3 |
|
Yes |
7,431 (2.7) |
2.7 |
|
Sexual behavior patterns |
No sexual experience |
264,439 (95.1) |
94.9 |
|
Opposite-sex experience only |
11,042 (4.0) |
4.1 |
|
Same-sex experience only |
1,470 (0.5) |
0.6 |
|
Experience with both sexes |
1,115 (0.4) |
0.4 |
|
Health behaviors |
|
|
|
|
Current smoking |
No |
255,530 (91.9) |
91.7 |
|
Yes |
22,536 (8.1) |
8.3 |
|
Current alcohol use |
No |
234,038 (84.2) |
83.8 |
|
Yes |
44,028 (15.8) |
16.2 |
|
Breakfast skipping (≥5 day/wk) |
No |
200,629 (72.2) |
72.3 |
|
Yes |
77,437 (27.8) |
27.7 |
|
Fast food consumption (≥3 times/wk) |
No |
237,032 (85.2) |
85.0 |
|
Yes |
41,034 (14.8) |
15.0 |
|
Vegetable consumption (≥3 times/day) |
No |
234,270 (84.3) |
84.5 |
|
Yes |
43,796 (15.7) |
15.5 |
|
Physical activity (60 min/day, ≥5 day/wk) |
No |
239,910 (86.3) |
86.6 |
|
Yes |
38,156 (13.7) |
13.4 |
|
Weight control efforts |
No effort |
138,845 (49.9) |
50.4 |
|
Weight loss effort |
91,621 (32.9) |
32.6 |
|
Weight gain effort |
16,902 (6.1) |
6.1 |
|
Weight maintenance effort |
30,698 (11.0) |
10.9 |
|
Health perceptions |
|
|
|
|
Subjective health perception |
Healthy |
198,094 (71.2) |
71.2 |
|
Average |
62,842 (22.6) |
22.6 |
|
Unhealthy |
17,130 (6.2) |
6.2 |
|
Subjective body image perception |
Thin |
73,626 (26.5) |
26.8 |
|
Average |
96,505 (34.7) |
34.6 |
|
Overweight |
107,935 (38.8) |
38.6 |
|
Mental health |
|
|
|
|
Subjective happiness |
Happy |
176,628 (63.5) |
63.2 |
|
Average |
76,516 (27.5) |
27.7 |
|
Unhappy |
24,922 (9.0) |
9.1 |
|
Stress perception |
High |
105,200 (37.8) |
37.9 |
|
Moderate |
118,692 (42.7) |
42.8 |
|
Low |
54,174 (19.5) |
19.3 |
|
Subjective sleep sufficiency |
Sufficient |
74,689 (26.9) |
26.4 |
|
Average |
89,959 (32.3) |
32.4 |
|
Insufficient |
113,418 (40.8) |
41.2 |
|
Depressive symptoms |
No |
204,033 (73.4) |
73.2 |
|
Yes |
74,033 (26.6) |
26.8 |
|
Suicidal ideation |
No |
240,851 (86.6) |
86.6 |
|
Yes |
37,215 (13.4) |
13.4 |
|
Suicide planning |
No |
265,466 (95.5) |
95.5 |
|
Yes |
12,600 (4.5) |
4.5 |
|
Suicide attempt |
No |
269,749 (97.0) |
97.0 |
|
Yes |
8,317 (3.0) |
3.0 |
|
Sexual behaviorsb)
|
|
|
|
|
Age at first sexual intercourse |
≤Elementary school |
4,042 (30.5) |
30.4 |
|
Middle school |
4,993 (37.7) |
37.4 |
|
High school |
4,208 (31.8) |
32.2 |
|
Missing |
384 |
- |
|
Sexual intercourse after drinking |
No |
9,152 (67.2) |
67.0 |
|
Yes |
4,475 (32.8) |
33.0 |
|
Contraception use |
Always |
4,081 (29.9) |
30.3 |
|
Sometimes |
1,619 (11.9) |
12.2 |
|
Rarely |
1,384 (10.2) |
10.3 |
|
Never |
6,543 (48.0) |
47.2 |
|
Pregnancy experience (female)c)
|
No |
3,468 (92.1) |
91.3 |
|
Yes |
298 (7.9) |
8.7 |
|
STI experience |
No |
12,381 (90.9) |
90.5 |
|
Yes |
1,246 (9.1) |
9.5 |
|
Diseases |
|
|
|
|
Asthma |
No |
252,771 (90.9) |
90.9 |
|
Yes |
25,295 (9.1) |
9.1 |
|
Allergic rhinitis |
No |
187,899 (67.6) |
66.8 |
|
Yes |
90,167 (32.4) |
33.2 |
|
Atopic dermatitis |
No |
211,321 (76.0) |
75.8 |
|
Yes |
66,745 (24.0) |
24.2 |
Table 2.Model fit indices for latent class analysis
|
No. of classes |
Log-likelihood |
AIC |
BIC |
Entropy |
|
2 |
–2,670,187 |
5,340,481 |
5,341,039 |
0.924 |
|
3 |
–2,641,249 |
5,282,658 |
5,283,501 |
0.865 |
|
4 |
–2,621,889 |
5,243,992 |
5,245,119 |
0.855 |
|
5 |
–2,605,025 |
5,210,319 |
5,211,730 |
0.806 |
|
6 |
–2,596,838 |
5,193,998 |
5,195,694 |
0.772 |
Table 3.Profiles of four latent classes (N=278,066)
|
Variable |
Class 1: mental health high-risk group |
Class 2: substance use high-risk group |
Class 3: average group |
Class 4: healthy group |
|
Sexual behavior patterns |
|
|
|
|
|
No sexual experience |
88.8 |
66.9 |
98.4 |
97.7 |
|
Opposite-sex experience only |
8.0 |
29.0 |
1.3 |
1.8 |
|
Same-sex experience only |
1.9 |
2.3 |
0.2 |
0.3 |
|
Experience with both sexes |
1.4 |
1.8 |
0.1 |
0.2 |
|
Health behaviors |
|
|
|
|
|
Current smoking (yes) |
15.7 |
82.4 |
1.2 |
1.8 |
|
Current drinking (yes) |
27.1 |
84.6 |
11.4 |
7.7 |
|
Breakfast skipping (≥5 day) |
34.5 |
39.9 |
30.6 |
22.7 |
|
Fast food (≥3 times/wk) |
22.0 |
30.6 |
14.6 |
11.2 |
|
Vegetables (≥3 times/day) |
16.6 |
13.0 |
12.0 |
18.8 |
|
Physical activity (60 min/day, ≥5 day/wk) |
15.2 |
20.6 |
8.4 |
16.6 |
|
Weight control efforts |
|
|
|
|
|
No effort |
39.9 |
48.7 |
49.1 |
52.9 |
|
Weight loss effort |
42.5 |
24.6 |
38.7 |
27.7 |
|
Weight gain effort |
6.5 |
18.3 |
3.1 |
6.7 |
|
Weight maintenance effort |
11.2 |
8.4 |
9.2 |
12.8 |
|
Health perceptions |
|
|
|
|
|
Subjective health perception |
|
|
|
|
|
Healthy |
49.8 |
72.4 |
53.8 |
88.9 |
|
Average |
31.9 |
21.1 |
36.6 |
10.2 |
|
Unhealthy |
18.3 |
6.5 |
9.6 |
0.9 |
|
Subjective body image perception |
|
|
|
|
|
Thin |
25.1 |
38.4 |
21.3 |
29.1 |
|
Average |
28.0 |
32.9 |
31.5 |
38.8 |
|
Overweight |
46.9 |
28.7 |
47.2 |
32.0 |
|
Mental health |
|
|
|
|
|
Subjective happiness |
|
|
|
|
|
Happy |
26.4 |
57.0 |
35.8 |
93.4 |
|
Average |
33.0 |
34.3 |
52.3 |
6.6 |
|
Unhappy |
40.7 |
8.7 |
11.9 |
0.0 |
|
Stress perception |
|
|
|
|
|
High |
85.0 |
40.9 |
65.3 |
6.5 |
|
Moderate |
11.3 |
43.2 |
34.5 |
55.5 |
|
Low |
3.7 |
15.9 |
0.2 |
38.0 |
|
Subjective sleep sufficiency |
|
|
|
|
|
Sufficient |
14.3 |
16.7 |
11.0 |
43.0 |
|
Average |
22.6 |
30.6 |
31.3 |
35.5 |
|
Insufficient |
63.2 |
52.6 |
57.8 |
21.5 |
|
Depression (yes) |
88.5 |
34.3 |
35.8 |
5.5 |
|
Suicidal ideation (yes) |
96.2 |
6.0 |
8.1 |
0.8 |
|
Suicide planning (yes) |
39.9 |
1.4 |
0.4 |
0.6 |
|
Suicide attempt (yes) |
27.5 |
0.9 |
0.3 |
0.1 |
Table 4.Sociodemographic characteristics by latent class (N=278,066)
|
Variable |
Class 1: mental health high-risk group |
Class 2: substance use high-risk group |
Class 3: average group |
Class 4: healthy group |
χ² |
p
|
|
Sex |
|
|
|
|
14,880.39 |
<.001 |
|
Male |
41.4 |
81.5 |
41.1 |
58.9 |
|
|
|
Female |
58.6 |
18.5 |
58.9 |
41.1 |
|
|
|
School level |
|
|
|
|
11,088.16 |
<.001 |
|
Middle school |
48.0 |
18.4 |
41.9 |
55.8 |
|
|
|
High school |
52.0 |
81.6 |
58.1 |
44.2 |
|
|
|
City size |
|
|
|
|
114.51 |
<.001 |
|
Rural |
6.5 |
7.5 |
6.2 |
6.4 |
|
|
|
Small-medium city |
49.7 |
51.6 |
50.5 |
49.3 |
|
|
|
Metropolitan |
43.8 |
40.9 |
43.3 |
44.4 |
|
|
|
Living arrangement |
|
|
|
|
1,019.98 |
<.001 |
|
With family |
93.5 |
92.8 |
95.9 |
96.7 |
|
|
|
Others |
6.5 |
7.2 |
4.1 |
3.3 |
|
|
|
Father’s educationa)
|
|
|
|
|
1,808.17 |
<.001 |
|
≤Middle school |
3.9 |
4.6 |
3.0 |
2.2 |
|
|
|
High school |
30.7 |
39.8 |
31.6 |
28.0 |
|
|
|
≥College |
50.1 |
41.7 |
49.5 |
52.9 |
|
|
|
Unknown |
15.3 |
13.8 |
15.9 |
16.9 |
|
|
|
Mother’s educationa)
|
|
|
|
|
1,884.58 |
<.001 |
|
≤Middle school |
3.7 |
4.0 |
2.7 |
1.9 |
|
|
|
High school |
39.0 |
47.4 |
40.7 |
35.9 |
|
|
|
≥College |
42.9 |
34.8 |
41.5 |
45.8 |
|
|
|
Unknown |
14.3 |
13.8 |
15.1 |
16.5 |
|
|
|
Subjective academic performance |
|
|
|
|
7,208.36 |
<.001 |
|
High (high & upper-middle) |
30.5 |
24.4 |
33.4 |
42.5 |
|
|
|
Middle |
23.7 |
23.8 |
28.7 |
29.2 |
|
|
|
Low (lower-middle & low) |
45.7 |
51.9 |
37.9 |
28.3 |
|
|
|
Subjective economic status |
|
|
|
|
9,994.49 |
<.001 |
|
High (high & upper-middle) |
30.3 |
30.6 |
28.1 |
41.5 |
|
|
|
Middle |
40.7 |
43.2 |
50.7 |
47.0 |
|
|
|
Low (lower-middle & low) |
29.0 |
26.2 |
21.2 |
11.6 |
|
|
|
Part-time job experience |
|
|
|
|
24,839.18 |
<.001 |
|
No |
79.0 |
50.5 |
88.2 |
91.7 |
|
|
|
Yes |
21.0 |
49.5 |
11.8 |
8.3 |
|
|
|
Hospital treatment due to violence |
|
|
|
|
7,630.46 |
<.001 |
|
No |
90.5 |
92.7 |
98.3 |
98.7 |
|
|
|
Yes |
9.5 |
7.3 |
1.7 |
1.3 |
|
|
Table 5.Sexual behavior characteristics and disease prevalence by latent class (N=278,066)
|
Variable |
Class 1: mental health high-risk group |
Class 2: substance use high-risk group |
Class 3: average group |
Class 4: healthy group |
χ2
|
p
|
|
Sexual behaviora)
|
|
|
|
|
|
|
|
Age at first sexual experience |
|
|
|
|
1,277.93 |
<.001 |
|
Elementary school |
33.6 |
16.6 |
27.4 |
35.3 |
|
|
|
Middle school |
28.3 |
25.0 |
30.7 |
33.4 |
|
|
|
High school |
38.1 |
58.4 |
41.9 |
31.3 |
|
|
|
Sexual intercourse after drinking |
|
|
|
|
1,305.48 |
<.001 |
|
No |
61.2 |
54.4 |
85.8 |
87.6 |
|
|
|
Yes |
38.8 |
45.6 |
14.2 |
12.4 |
|
|
|
Contraceptive use |
|
|
|
|
607.53 |
<.001 |
|
Always |
28.6 |
30.6 |
31.9 |
30.6 |
|
|
|
Sometimes |
13.8 |
15.6 |
7.3 |
6.4 |
|
|
|
Rarely |
10.9 |
14.4 |
5.8 |
4.1 |
|
|
|
Never |
46.6 |
39.4 |
55.1 |
58.9 |
|
|
|
Pregnancy experience (female)b)
|
|
|
|
|
84.44 |
<.001 |
|
No |
87.8 |
88.5 |
96.5 |
97.1 |
|
|
|
Yes |
12.2 |
11.5 |
3.5 |
2.9 |
|
|
|
STI experience |
|
|
|
|
307.38 |
<.001 |
|
No |
83.6 |
90.2 |
95.5 |
95.5 |
|
|
|
Yes |
16.4 |
9.8 |
4.5 |
4.5 |
|
|
|
Disease |
|
|
|
|
|
|
|
Asthma |
|
|
|
|
453.97 |
<.001 |
|
No |
87.9 |
89.9 |
90.6 |
91.8 |
|
|
|
Yes |
12.1 |
10.1 |
9.4 |
8.2 |
|
|
|
Allergic rhinitis |
|
|
|
|
899.94 |
<.001 |
|
No |
62.8 |
69.5 |
64.2 |
69.2 |
|
|
|
Yes |
37.2 |
30.5 |
35.8 |
30.8 |
|
|
|
Atopic dermatitis |
|
|
|
|
877.6 |
<.001 |
|
No |
72.4 |
79.7 |
73.5 |
77.7 |
|
|
|
Yes |
27.6 |
20.3 |
26.5 |
22.3 |
|
|
REFERENCES
- 1. Phillips GL, Curtis MG, Kelsey SW, Floresca YB, Davoudpour S, Quiballo K, et al. Changes to sexual identity response options in the Youth Risk Behavior Survey. JAMA Pediatr. 2024;178(5):506-508. https://doi.org/10.1001/jamapediatrics.2024.0024
- 2. LGBT+ Pride 2024 Global Survey: a 26-country Ipsos global advisor survey [Internet]. Ipsos; 2024 [cited 2025 Aug 1]. Available from: https://www.ipsos.com/sites/default/files/ct/news/documents/2024-05/Pride%20Report%20FINAL_0.pdf
- 3. Lee HR, Lee HM, Joo SS, Kim RY, Um YJ, Kim SS. Importance of collecting sexual orientation and gender identity information in nationally representative surveys of South Korea: a review of current status and measurement development. J Crit Soc Welf. 2022(74):175-208. https://doi.org/10.47042/ACSW.2022.2.74.175
- 4. Saewyc EM. Research on adolescent sexual orientation: development, health disparities, stigma and resilience. J Res Adolesc. 2011;21(1):256-272. https://doi.org/10.1111/j.1532-7795.2010.00727.x
- 5. Kwak Y, Kim JS. Associations between Korean adolescents’ sexual orientation and suicidal ideation, plans, attempts, and medically serious attempts. Iran J Public Health. 2017;46(4):475-484.
- 6. Frost DM, Meyer IH. Minority stress theory: application, critique, and continued relevance. Curr Opin Psychol. 2023;51:101579. https://doi.org/10.1016/j.copsyc.2023.101579
- 7. Lee DY, Kim SH, Woo SY, Yoon BK, Choi D. Associations of health-risk behaviors and health cognition with sexual orientation among adolescents in school: analysis of pooled data from Korean nationwide survey from 2008 to 2012. Medicine (Baltimore). 2016;95(21):e3746. https://doi.org/10.1097/MD.0000000000003746
- 8. Depa N, Desai S, Patel S, Silvi S, Hanif S, Rizvi S, et al. Mental health disparities amongst sexual-minority adolescents of the US: a national survey study of YRBSS-CDC. Psychiatry Res. 2022;314:114635. https://doi.org/10.1016/j.psychres.2022.114635
- 9. Duong J, Kelly D, Sciolla AF. Illicit drug use and associated health risk behaviors among sexual minority youth. Addict Behav. 2025;169:108397. https://doi.org/10.1016/j.addbeh.2025.108397
- 10. Rasberry CN, Lowry R, Johns M, Robin L, Dunville R, Pampati S, et al. Sexual risk behavior differences among sexual minority high school students: United States, 2015 and 2017. MMWR Morb Mortal Wkly Rep. 2018;67(36):1007-1011. https://doi.org/10.15585/mmwr.mm6736a3
- 11. Nagata JM, Lee CM, Yang JH, Kiss O, Ganson KT, Testa A, et al. Sexual orientation disparities in early adolescent sleep: findings from the Adolescent Brain Cognitive Development Study. LGBT Health. 2023;10(5):355-362. https://doi.org/10.1089/lgbt.2022.0268
- 12. Weiss KE, Li R, Chen D, Palermo TM, Scheurich JA, Groenewald CB. Sexual orientation/gender identity discrimination and chronic pain in children: a national study. Am J Prev Med. 2024;67(2):175-183. https://doi.org/10.1016/j.amepre.2024.03.010
- 13. Katz-Wise SL, Everett B, Scherer EA, Gooding H, Milliren CE, Austin SB. Factors associated with sexual orientation and gender disparities in chronic pain among U.S. adolescents and young adults. Prev Med Rep. 2015;2:765-772. https://doi.org/10.1016/j.pmedr.2015.09.011
- 14. McQuillan MT, Kuhns LM, Miller AA, McDade T, Garofalo R. Gender minority stress, support, and inflammation in transgender and gender-nonconforming youth. Transgend Health. 2021;6(2):91-100. https://doi.org/10.1089/trgh.2020.0019
- 15. Beach LB, Elasy TA, Gonzales G. Prevalence of self-reported diabetes by sexual orientation: results from the 2014 Behavioral Risk Factor Surveillance System. LGBT Health. 2018;5(2):121-130. https://doi.org/10.1089/lgbt.2017.0091
- 16. Caceres BA, Sharma Y, Doan D. Hypertension risk in sexual and gender minority individuals. Expert Rev Cardiovasc Ther. 2022;20(5):339-341. https://doi.org/10.1080/14779072.2022.2075345
- 17. Sherman J, Dyar C, McDaniel J, Funderburg NT, Rose KM, Gorr M, et al. Sexual minorities are at elevated risk of cardiovascular disease from a younger age than heterosexuals. J Behav Med. 2022;45(4):571-579. https://doi.org/10.1007/s10865-021-00269-z
- 18. Huang AK, Hoatson T, Chakraborty P, McKetta S, Soled KR, Reynolds CA, et al. Disparities in cancer incidence by sexual orientation. Cancer. 2024;130(24):4306-4314. https://doi.org/10.1002/cncr.35356
- 19. Rypka KJ, Wendland ZD, Steele MV, Wehner MR, Yeung H, Mansh MD. Sexual orientation and lifetime prevalence of skin cancer across racial and ethnic groups. JAMA Dermatol. 2024;160(9):977-983. https://doi.org/10.1001/jamadermatol.2024.2097
- 20. Srivastava A, Davis JP, Goldbach JT. Gender and sexual identities predicting patterns of co-occurring health risks among sexual minority youth: a latent class analysis approach. Prev Sci. 2020;21(7):908-916. https://doi.org/10.1007/s11121-020-01137-z
- 21. Coulter RW, Ware D, Fish JN, Plankey MW. Latent classes of polysubstance use among adolescents in the United States: intersections of sexual identity with sex, age, and race/ethnicity. LGBT Health. 2019;6(3):116-125. https://doi.org/10.1089/lgbt.2018.0149
- 22. Nelon JL, De Pedro KT, Gilreath TD, Patterson MS, Holden CB, Esquivel CH. A latent class analysis of the co-occurrence of sexual violence, substance use, and mental health in youth. Subst Use Misuse. 2019;54(12):1938-1944. https://doi.org/10.1080/10826084.2019.1618337
- 23. Lanza ST. Latent class analysis for developmental research. Child Dev Perspect. 2016;10(1):59-64. https://doi.org/10.1111/cdep.12163
- 24. Institute of Medicine (US) Committee on Lesbian, Gay, Bisexual, and Transgender Health Issues and Research Gaps and Opportunities. The health of lesbian, gay, bisexual, and transgender people: building a foundation for better understanding. National Academies Press (US); 2011. https://doi.org/10.17226/13128
- 25. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med. 2007;4(10):e296. https://doi.org/10.1371/journal.pmed.0040296
- 26. Linzer DA, Lewis JB. poLCA: an R package for polytomous variable latent class analysis. J Stat Softw. 2011;42(10):1-29. https://doi.org/10.18637/jss.v042.i10
- 27. Patterson BH, Dayton CM, Graubard BI. Latent class analysis of complex sample survey data: application to dietary data. J Am Stat Assoc. 2002;97(459):721-741. https://doi.org/10.1198/016214502388618465
- 28. Fine SL, Blum RW, Bass JK, Lulebo AM, Pinandari AW, Stones W, et al. A latent class approach to understanding patterns of emotional and behavioral problems among early adolescents across four low- and middle-income countries. Dev Psychopathol. 2023;35(4):1684-1700. https://doi.org/10.1017/S0954579422000384
- 29. Bianchi V, Brambilla P, Garzitto M, Colombo P, Fornasari L, Bellina M, et al. Latent classes of emotional and behavioural problems in epidemiological and referred samples and their relations to DSM-IV diagnoses. Eur Child Adolesc Psychiatry. 2017;26(5):549-557. https://doi.org/10.1007/s00787-016-0918-2
- 30. Kretschmer T, Barker ED, Dijkstra JK, Oldehinkel AJ, Veenstra R. Multifinality of peer victimization: maladjustment patterns and transitions from early to mid-adolescence. Eur Child Adolesc Psychiatry. 2015;24(10):1169-1179. https://doi.org/10.1007/s00787-014-0667-z
- 31. Bernanke J, Galfalvy HC, Mortali MG, Hoffman LA, Moutier C, Nemeroff CB, et al. Suicidal ideation and behavior in institutions of higher learning: a latent class analysis. J Psychiatr Res. 2017;95:253-259. https://doi.org/10.1016/j.jpsychires.2017.09.003
- 32. Robinson-Cimpian JP. Inaccurate estimation of disparities due to mischievous responders: several suggestions to assess conclusions. Educ Res. 2014;43(4):171-185. https://doi.org/10.3102/0013189X14534297
- 33. Kingsbury M, Hammond NG, Johnstone F, Colman I. Suicidality among sexual minority and transgender adolescents: a nationally representative population-based study of youth in Canada. CMAJ. 2022;194(22):E767-E774. https://doi.org/10.1503/cmaj.212054
- 34. Elenkov IJ, Chrousos GP. Stress hormones, Th1/Th2 patterns, pro/anti-inflammatory cytokines and susceptibility to disease. Trends Endocrinol Metab. 1999;10(9):359-368. https://doi.org/10.1016/s1043-2760(99)00188-5
- 35. Yang S. Young generation’s perceptions of same-sex sexuality and attitudes toward same-sex marriage in South Korea. SAGE Open. 2021;11(3):21582440211031886. https://doi.org/10.1177/21582440211031886
