Enquobahrie, Daniel AFreije, Sophia L2026-02-052026-02-052025Freije_washington_0250E_29046.pdfhttps://hdl.handle.net/1773/55226Thesis (Ph.D.)--University of Washington, 2025Background: Loneliness among adolescents has drawn public health concern due to its rising rates and established implications for mental and physical health. Global estimates indicate that the percentage of adolescents who reported feeling lonely “often” or “always” nearly doubled from 2012 (12%) to 2018 (23%). These widespread shifts may reflect societal influences and increased time spent online, warranting comprehensive investigation into contextual factors and digital technology interactions that may lead to loneliness. When persistent, the psychological and physiological strains of loneliness can lead to adverse health outcomes such as metabolic syndrome (MetS)—a cluster of factors that raise the risk of cardiometabolic diseases. However, reports on adolescent loneliness and MetS associations were inconsistent. This dissertation examined recent trends, correlates, and longitudinal metabolic risks of adolescent loneliness.Objectives: We aimed to: 1) characterize temporal trends in adolescent loneliness from 2000 to 2022 and associations with country-level factors (COVID-19 mortality, public health distancing laws, unemployment, income inequality, peacefulness, and internet use); 2) investigate associations of digital technology interactions with loneliness overall and among groups defined by risk for problematic internet use, gender identity, and race and ethnicity; and 3) investigate associations of adolescent persistent loneliness with MetS at ages 24-32 and 33-43, and examine if associations are mediated by depressive symptoms or moderated by sex and race/ethnicity. Methods: Aim 1 was addressed using data from the Programme for International Student Assessment (PISA), a school-based survey administered to 1,267,476 adolescents aged 15-16 across 38 Organization for Economic Cooperation and Development-member countries in 2000, 2003, 2012, 2015, 2018, and 2022. Aim 2 was addressed using data from the Youth Health and Social Media (YHSM) survey, a cross-sectional online survey administered to 12–17-year-old U.S. residents in 2019 (N=4,541). Aim 3 was addressed using data from the National Longitudinal Study of Adolescent to Adult Health (Add Health), a prospective cohort that followed participants from ages 11-21 in 1994 to ages 24-32 (n=10,069) and ages 33-43 (n=4,061). Loneliness was measured using a 6-item school loneliness questionnaire (PISA), the 3-item Comprehensive Inventory of Thriving Loneliness subscale (YHSM), and a single-item direct frequency measure (Add Health). To address Aim 1, multilevel piecewise regression was used to estimate trends in mean loneliness scores during three periods: 2000-2012, 2012-2015, and 2015-2022. Multilevel linear regression was used to estimate associations between country-level factors (COVID-19 mortality, public health distancing measures, internet use, unemployment rate, income inequality, and peacefulness) and loneliness scores. To address Aim 2, multivariable linear regression with robust standard errors was used to estimate the mean change in loneliness scores associated with quartiles of scores on the Adolescents’ Digital Technology Interactions and Importance (ADTI) scale and its three factors (Factor 1, bridging online and offline preferences; Factor 2, escaping offline environments; Factor 3, using technology for social connection). To address Aim 3, we used Poisson regression with robust standard errors to estimate associations between persistent loneliness in adolescence—defined as loneliness reported in two waves, one year apart—and MetS at Wave IV (ages 24-32) and Wave V (ages 33-43). We tested for mediation of the associations by depressive symptoms measured in Wave III (ages 18-26) using counterfactual approaches. Differences in associations according to risk for PIU (Aim 2), gender (Aim 2), sex (Aim 3), and race/ethnicity (Aim 2, 3) were evaluated using stratified analyses and models with interaction terms. Results. In analyses of PISA data, we observed no significant change in mean loneliness scores from 2000 to 2012 (95%CI: -0.022, 0.008), followed by yearly increases of 0.045 points from 2012-2015 (95%CI: 0.035, 0.054), and 0.017 points from 2015-2022 (95%CI: 0.008, 0.025). Within countries, a one-hour higher country-level weekly internet use and one percentage-point higher unemployment rate were associated with 0.012-point (95%CI: -0.021, -0.003) and 0.006-point (95%CI: -0.011, -0.001) lower loneliness scores per year, respectively. Other contextual measures were not associated with loneliness. In analyses of the YHSM survey, participants had mean ADTI and loneliness scores of 48.8 (SD 17.9) and 3.6 (SD 3.6), respectively. We observed a U-shaped relationship between quartiles of ADTI scores and loneliness scores, with participants with moderate scores (Q2, Q3) having similar or lower loneliness scores and participants with high scores (Q4) having similar or higher loneliness scores, compared to respective Q1 categories. For the total ADTI, participants in Q2 or Q3 had 0.24-point (95%CI:-0.45,-0.02) and 0.38-point (95%CI:-0.61,-0.15) lower loneliness scores, respectively, compared with Q1 participants. Participants in Q3 for Factor 1 (bridging online and offline preferences) had 0.31-point (95%CI:-0.52,-0.10) lower loneliness scores, while participants in Q4 for Factor 2 (escaping offline environments) had 0.30-point (95%CI:0.01, 0.59) higher mean loneliness scores, compared with participants in Q1. We observed significant multiplicative interactions between ADTI measures and PIU and gender identity (P-values <0.05) but not race and ethnicity. Among high-risk PIU participants and cisgender males, compared with their counterparts in Q1, moderate total ADTI and Factor 1 scores were associated with lower loneliness, whereas high total ADTI (males only), Factor 1 (males only), or Factor 3 scores were associated with higher loneliness (P-values<0.05). In the Add Health study, the prevalence of persistent loneliness at baseline was 19% across participants eligible for Wave IV or Wave V analyses. The prevalence of MetS was 20% in Wave IV and 27% in Wave V. Persistent loneliness was not associated with Wave IV MetS (aRR: 1.03, 95%CI: 0.89, 1.18) while it was associated with a 21% lower likelihood of Wave V MetS (95%CI: 0.64, 0.99). Loneliness-MetS associations were not modified by sex or race/ethnicity or mediated by depressive symptoms. Conclusions: Loneliness increased from 2012-2015 across 38 countries, followed by smaller annual increases until 2022. Greater increases in country-level unemployment rates and internet use over time were associated with lower loneliness. We also observed a U-shaped relationship between digital technology interactions and loneliness—a pattern observed and stronger among high PIU participants and cisgender men compared with low PIU participants, cisgender women, and TNG youth. Persistent loneliness during adolescence was associated with a lower risk of MetS at ages 33-43 but not at 24-32. Findings of loneliness trends underscore the urgency of promoting social connection through targeted public health strategies. Ongoing examinations of the trends and societal correlates of loneliness are warranted as updated data become available. Longitudinal studies are needed to clarify the temporality of digital technology interactions and loneliness, as well as comprehensive understanding of loneliness and MetS associations.application/pdfen-USCC BYAdolescenceDigital technologyLonelinessMetabolic SyndromePublic HealthSocial connectionEpidemiologyMental healthPublic healthEpidemiologyThesis