Background Growing research and clinical attention to loneliness has led to a wide range of available measures; however, few studies directly compare different loneliness measures within the same population. Measures include single-item, direct questions and indirect, multi-item scales. This study compared responses and respondent characteristics between a direct single-item loneliness question and an indirect multi-item scale. Methods We conducted a cross-sectional analysis of population-based survey data from the Canadian Longitudinal Study on Aging (CLSA). The sample included baseline (2010−2015) and follow-up 1 respondents (2015−2018) from both the Tracking (computer-assisted telephone interviews) and Comprehensive (in-home assessment) cohorts (n = 43,235). Loneliness was assessed using a direct single-item question from the Center for Epidemiological Studies Depression Scale (CES-D-10) and the 3-item Loneliness Scale. We categorized loneliness using ordinal (not lonely, moderately lonely, severely lonely) and dichotomous (lonely/not lonely) classifications. Unweighted descriptive statistics were used to examine demographic, social, and health characteristics across measures and response categories. Results Among the 43,235 CLSA respondents, 14% (n = 5,848) were identified by both the single-item question and the scale as lonely, 67% (n = 28,998) were identified as not lonely by both, leaving 19% classified differently based on measure. Individuals identified as lonely respondents by both measures were more often older, women, less educated, and had lower income. Those classified as severely lonely on the single item but not lonely on the scale were older than other groups. In contrast, respondents who were moderately lonely on the single-item question tended to be married, report higher income, and have greater social contact. Conclusion The direct single-item question and the indirect multi-item scale provided comparable overall classifications of loneliness in our CLSA sample. However, respondent characteristics varied across the more granular response categories, highlighting the importance of measure selection depending on study aims.
Chamberlain et al. (Wed,) studied this question.