The "loneliness epidemic", intersecting risk factors and relations to mental health help-seeking: A population-based study during COVID-19 lockdown in Canada

Abstract

Background: Pandemic-induced social distancing and stay-at-home orders, while successful in decreasing the transmission of COVID-19, could exacerbate loneliness. Few studies have examined how pandemic-related social determinants intersect to shape pandemic loneliness and its relations to mental health care in Canada.

Methods: A population-representative sample of 3772 adults from the Canadian Perspective Survey Series (CPSS-6; January 25 to 31, 2021) was analyzed. Gender-specific logistic regression was employed to investigate the association between three-item loneliness scale (UCLA-3) with socio-demographics, job precarity, health behaviours, social isolation indicators, and mental health help-seeking. Classification and Regression Tree (CART) modelling was used to identify intersecting risk factors and the most important predictor of severe loneliness (UCLA-3 score ≥ 7).

Results: The estimated prevalence of severe loneliness was 34.7 % in Canada, with women significantly higher than men (38.1 % vs 31.3 %, p < 0.001). Pandemic loneliness were more prevalent in female (OR = 1.53, 99 % CI: 1.26-1.85), those who were younger (OR's range 1.42-3.00), women without college degree (OR = 1.44, 99 % CI: 1.01-2.04), those living alone (OR = 1.56, 99 % CI: 1.09-2.23), immigrant men (OR = 1.79, 99 % CI: 1.23-2.60), those with small network (OR's range: 1.73-3.26), those who were absent from work due to COVID-19 related reasons (OR = 2.11, 99 % CI: 1.04-4.28), past-month binge drinkers (OR's range: 1.39-1.70) and cannabis user (OR = 1.47, 99 % CI: 1.12-1.93). The CART algorithm identifies that immigrants who experienced pandemic-triggered job insecurity were the most-at-risk group of severely loneliness. Pandemic loneliness was positively associated with formal help-seeking from mental health professionals (OR = 1.71, 99 % CI: 1.21-2.41), informal help-seeking from social circle (OR = 1.51, 99 % CI: 1.17-1.95), and unmet mental health needs (OR = 1.78, 99 % CI: 1.29-2.49).

Limitations: Cross-sectional data prohibits causal inferences.

Conclusion: The COVID-19 pandemic converges with loneliness epidemic in Canada. Prevention and intervention programs should target upstream social determinants of mental health, especially the intersection of migration status and COVID-19-related job precarity, to eliminate loneliness during the pandemic.

Keywords: COVID-19; Loneliness; Machine learning; Psychiatric epidemiology; Social isolation.