The Influence of Loneliness on Older Adults' Smoking and Physical Activity: A Lagged Dependent Approach

Background/Purpose: Social isolation is a dire social problem, and has been recognized as one of 12 Grand Challenges in social work. Loneliness, a subjective measure of social isolation, has been found to be associated with depressive symptoms, functional limitations, and increased mortality rates for older adults. However, the mechanism of loneliness’ influence on health outcomes for older adults is less examined, particularly through longitudinal analysis.

The Loneliness Model posits that perceived social isolation may lead to cognitive biases through fear and unconscious surveillance for social threat, thus result in the lack of self-control and maladaptive behaviors, which eventually lead to negative health outcomes. Existing studies on the association between older adults’ loneliness and behavioral outcomes such as smoking and physical activity are predominantly cross-sectional, with one exception of longitudinal study using a small local sample. The present study aims to fill the gap by analyzing longitudinal nationally representative data. We tested the following two hypotheses. Hypothesis 1: Loneliness is associated with the current smoking status. Hypothesis 2: Loneliness is associated with a decrease in physical activity. 

Methods: Data and Sample. This study used two waves (2010 and 2012) of data from the Health and Retirement Study (HRS), which collects biannual panel survey data from a representative sample over the age of 50 in the United States. After matching sample from two waves, we first excluded those who were either age-ineligible spouses, dead, or institutionalized. We also excluded those who had missing values for second wave variables due to attrition. The final sample size was 7,141.

Measures. Smoking was a dichotomous variable indicating whether respondents are currently smoking or not. Three kinds of physical activity (vigorous, moderate, and light physical activity) were measured based on its intensity. Each physical activity outcome had five response categories asking the frequency of physical activity ranging from every day (1) to never (5).  Loneliness was measured by an adapted UCLA Loneliness Scale. Control variables included self-reported health, difficulties in activities of daily living, and demographics.

Analysis Strategy. We adopted a lagged dependent approach for modeling, which requires controlling for baseline outcomes. Our independent variable and control variables were all baseline variables. We used logistic regression for smoking and heterogeneous choice models for ordinal physical activity outcomes. We used listwise deletion to handle missing values (2%).

Results: In line with our hypotheses, results showed that loneliness was associated with increased odds of smoking and decreased odds of engaging in all three levels of physical activities for older adults, after controlling for baseline outcomes and other covariates.

Conclusions and Implications: Using national representative longitudinal data, we found that loneliness may lead to smoking and reduced physical activity frequencies on all three levels: vigorous, moderate, and light. Our study contributes to the literature by establishing the temporal links between loneliness and behavioral health. The findings call for interventions targeting socially isolated older adults. Efforts to enhance social support will be crucial to eradicate the deleterious health impact of social isolation.