COVID-19 restrictions and age-specific mental health—U.S. probability-based panel evidence

Governments worldwide imposed social distancing, self-isolation, quarantining, and lockdowns to curb COVID-19 transmission. Interest in mental health support rose markedly early in the pandemic (e.g., Google searches for “mental health hotline” tripled in March 2020) and remained elevated. Prior evidence indicates increased psychological distress during COVID-19, with cohort studies showing higher rates of common mental disorders (CMDs), particularly among those without prior psychological problems. Many early studies relied on community-based surveys susceptible to response bias, underscoring the need for probability-based population evidence. This study examines whether COVID-19-related restrictions and lockdowns affected adults’ mental health similarly across age cohorts, gender, and comorbidities (including prior mental health diagnosis), using a U.S. probability-based survey (April 20–June 8, 2020). The main question is whether mental health impacts of restrictions are uniform across adults; based on prior indications that younger people are at greater risk during the pandemic, the authors hypothesized stronger effects of social restrictions on younger adults.

The paper situates its research within literature documenting psychological impacts of large-scale health shocks and disasters (e.g., SARS, MERS), including elevated risks of PTSD and psychological distress. Systematic reviews during COVID-19 show about one in three people experienced psychological distress. Longitudinal cohort studies report increased rates of common mental disorders during the pandemic, notably among individuals without prior psychological issues. However, many surveys are community-based and subject to response bias, highlighting the value of probability-based studies. Prior work during early U.S. pandemic phases showed increased acute stress and depressive symptoms in nationally representative samples. Evidence on age-specific effects is limited and mixed; this study adds age-stratified evidence on CMDs linked to specific COVID-19 restrictions.

Design and data source: The COVID-19 Household Impact Survey (NORC at the University of Chicago; funded by the Data Foundation) sampled a nationally representative U.S. adult population (≥18 years) from the AmeriSpeak Panel using 48 strata (age, race/ethnicity, education, gender). Coverage ~97% of U.S. households; recruitment rate 21.5% across 30,076 individuals. Three independent cross-sectional waves: Apr 20–26, 2020; May 4–10, 2020; May 30–Jun 8, 2020; total N=6,475. Measures: Mental distress was assessed via five items (past 7 days): (1) nervous/anxious/on edge; (2) depressed; (3) lonely; (4) hopeless about the future; (5) physical reactions (e.g., sweating, trouble breathing, nausea, pounding heart) when thinking about COVID-19. Responses coded 0–3 (0: not at all/ <1 day; 1: 1–2 days; 2: 3–4 days; 3: 5–7 days). A composite Psychological Distress Scale (T5) summed the five items (range 0–15); higher scores indicate greater distress. Internal consistency: Cronbach’s alpha ~0.84; individual items correlated 0.37–0.62 with each other and 0.57–0.82 with T5. Criterion validation used self-reported prior mental health diagnosis (PHYS8H). ROC analysis identified optimal threshold T5≥3 (sensitivity 0.70, specificity 0.73, total classification rate 0.70, AUC 0.77; alpha 0.84). Moderate distress was defined as T5≥3. Explanatory variables: Respondents reported whether their personal plans were affected (past 7 days; yes/no) by 19 restrictions/closures: K-12 school; Pre-K/childcare; college/training; bans on gatherings (>250, >50, >10 people); closure of places of worship; reduced public transportation; other reduced public services; closure of bars; restaurants; gyms/fitness; other businesses; canceled sports events; closure of work; work-from-home requirements; quarantine/stay-at-home orders; international travel restrictions/bans; domestic travel restrictions/bans. Additional variables: current employment (worked for someone else; self-employed; benchmark: did not work last week); job prospects in 90 days (extremely/very/moderately likely vs benchmark: not likely at all); self-rated physical health (excellent, very good, good, fair; benchmark: poor). Demographics: gender, race; household composition (alone; with other adult(s); with 1–2 children; more children as applicable). Statistical analysis: Multivariate logistic regressions modeled moderate mental distress (T5≥3) as the dependent variable. Independent variables: the 19 restriction indicators, controlling for gender, race, household composition, prior mental health diagnosis, physical health, employment status, 90-day job prospects, and survey wave fixed effects (April as benchmark). Analyses were stratified by age groups: 18–34, 35–54, and >55 years. Responses “Not sure” and “Skipped on the web” were excluded (treated as missing), focusing on yes/no responses; the final regression sample included 3,646 respondents. Robustness checks including models treating “Not sure/Skipped” as explicit categories in the full sample yielded qualitatively similar results. Analyses were conducted using SAS 9.4.

Prevalence and symptom severity: In the full sample (N=6,475), 36.79% reported moderate mental distress (T5≥3); mean T5=2.65 (95% CI: 2.57–2.73). Age-specific prevalence: 18–24: 52.42%; 25–34: 48.28%; 35–44: 42.85%; 45–54: 37.02%; 55–64: 25.95%; 65–74: 22.09%; 75+: 23.60%. Broad age groups: 18–34: 49.84% (T5=3.60); 35–54: 39.96% (T5=2.91); >55: 24.08% (T5=1.79). Confidence intervals across age groups did not overlap, motivating age-stratified analysis. Major correlates across models: - Prior mental health diagnosis strongly associated with higher odds of moderate distress: OR≈4.60 (18–34), 4.17 (35–54), 5.78 (>55); p<0.001. - Better physical health associated with lower odds of distress (vs poor health). For >55: excellent OR 0.24 (p<0.001), very good OR 0.37 (p<0.001), good OR 0.72 (ns). For 18–34: excellent OR 0.15 (p=0.03). For 35–54: excellent OR 0.34 (p=0.02). Overall, poor physical health was linked to substantially higher distress. - Gender: Males had lower odds vs females among 35–54 (OR 0.85; p=0.01) and >55 (OR 0.78; p<0.001); no significant gender effect in 18–34. - Household: Living alone was associated with higher odds among 18–34 (OR 1.58; p=0.02); not significant in older groups. - Employment and job prospects: Among 35–54, those “Very likely” to have a job in 90 days had higher odds (OR 2.40; p<0.001) vs “Not likely at all” (controlling for current employment). Among >55, “Extremely likely” (OR 1.68; p=0.05) and “Very likely” (OR 1.56; p=0.07) were associated with greater distress. Job prospects were not significantly related to distress among 18–34. Age-specific associations with restrictions (selected significant or near-significant, p≤0.10): - Young adults (18–34): Increased odds associated with reduced public transportation (OR 1.53; p=0.01), other reduced public services (OR 1.46; p=0.03), closure of restaurants (OR 1.50; p=0.02), quarantine/stay-at-home orders (OR 1.50; p=0.02), and domestic travel restrictions (OR 1.42; p=0.04). Living alone also elevated risk (above). - Middle-aged adults (35–54): Lower odds associated with reduced public transportation (OR 0.74; p=0.06). Higher odds associated with other reduced public services (OR 1.35; p=0.06), closure of other businesses (OR 1.35; p=0.06), and closure of places of worship (OR 1.32; p=0.08). - Older adults (>55): Higher odds associated with closure of work (OR 1.33; p=0.08) and bans on large gatherings (e.g., >250 people; OR 1.35; p=0.06). Overall pattern: Restrictions were most strongly and broadly associated with distress among young adults, less so among middle-aged, and least among older adults.

Findings indicate substantial age-specific differences in how COVID-19 restrictions relate to moderate psychological distress. Younger adults exhibited the highest prevalence and the broadest sensitivity to mobility- and social-activity-related restrictions (e.g., reduced transportation and services, restaurant closures, stay-at-home orders, travel restrictions), consistent with greater reliance on social interaction, mobility, travel, and leisure. Middle-aged adults showed associations with reductions in public services and closures of businesses and places of worship, and reduced distress with decreased public transport, suggesting differing stressors tied to community services and business operations. Older adults’ distress was comparatively less linked to restrictions but increased with bans on large gatherings and closure of work. Across all ages, prior mental health diagnosis and poorer physical health were robustly associated with higher distress; males in middle and older age groups reported lower distress than females. These results support targeted mental health responses that account for age-specific contexts and stressors rather than uniform interventions.

This probability-based U.S. study demonstrates that COVID-19 restrictions relate differentially to moderate psychological distress across age groups, with the strongest and most diverse associations among young adults and the weakest among older adults. Key correlates include prior mental health diagnosis and physical health status, as well as gender and living arrangements. The findings underscore the need for age-tailored mental health policies and interventions during pandemics, focusing on: mitigating the mental health impact of mobility/social restrictions for younger adults; addressing community service disruptions for middle-aged adults; and supporting older adults facing limitations on gatherings and work. Future research should examine longitudinal trajectories, pre-pandemic baselines, mechanisms underlying age-specific effects, and the impact of specific policy changes over time using designs that minimize selection and reporting biases.

- Mental health outcomes were derived from screening items; no clinical diagnostic evaluations for anxiety or depressive disorders were conducted. - Potential unmeasured confounders related to the pandemic may not have been included, despite multiple controls. - No pre-pandemic data for these age groups were available; time-series comparisons to pre-pandemic levels cannot be drawn. - Average survey response rate was ~20%; although panel-based sampling weights were used, response selection bias may remain. Additionally, exclusion of “Not sure/Skipped” responses may reduce power and risk Type II error, though robustness checks yielded similar qualitative results.