The data used in this cross-sectional study were collected between 2014 and 2016 as part of a larger project exploring healthy aging among older adults (35). Ethical approval was received from Curtin University's Human Research Ethics Committee. Eligibility criteria were being aged 60 years or older, living in a community setting, and being fully retired. Participants were recruited via a range of methods including notices in community newspapers, radio announcements, and flyers placed at seniors' events and retirement villages across the metropolitan area of Perth, Western Australia.

In total, 801 adults met the above eligibility criteria and provided written informed consent. Participation involved completing a self-administered survey that included psychological, social, and physical health measures validated for use in older adults. These measures are described in detail below. Unless otherwise stated, the items forming each of the scales included in the survey were summed for analysis purposes.

The final sample consisted of older adults ranging in age from 60 to 95 years (M = 71.93 years, SD = 6.68), 61% of whom were female. Characteristics of the present sample alongside those of the Australian older adult population are presented in Table 1. Pearson chi-square tests indicated the sample was representative in terms of gender, age, and education but not living arrangement: the present sample had a significantly higher proportion of older adults living alone compared to the Australian older adult population.

Depressive symptoms were assessed using the 20-item Center for Epidemiological Studies Depression Scale [CES-D: (37)]. Participants responded to each item (e.g., “I felt that everything I did was an effort”) on a 4-point scale that ranged from 0 (rarely or none of the time) to 3 (most or all of the time). Cronbach's alpha in the present study was 0.87, indicating good reliability.

The 14-item Personal Growth and Purpose in Life subscales of Ryff's (38) Psychological Well-Being Scale were used to measure personal growth and purpose in life, respectively. Responses to items (e.g., “I have the sense that I have developed a lot as a person over time”; “I have a sense of direction and purpose in life”) were made on a scale of 1 (strongly disagree) to 5 (strongly agree). Cronbach's alpha in the present study indicated good reliability for scores on both the Personal Growth (α = 0.86) and Purpose in Life (α = 0.88) subscales.

Self-esteem was measured using the 10-item Rosenberg Self-Esteem Scale (39). Respondents answered each item (e.g., “I feel that I'm a person of worth, at least on an equal plane with others”) on a 4-point scale that ranged from 0 (strongly disagree) to 3 (strongly agree). Cronbach's alpha for scores on this scale was 0.88, indicating good reliability. The 24-item Social Provision Scale (40) was used to assess social support. Each item (e.g., “There is someone I could talk to about important decisions in my life”) was measured on a 4-point scale that ranged from 1 (strongly disagree) to 4 (strongly agree). The scores on this scale were found to have excellent reliability (α = 0.92).

Self-efficacy was assessed using the 10-item General Self-Efficacy Scale (41). Participants responded to each item (e.g., “I can solve most problems if I invest the necessary effort”) on a 4-point scale that ranged from 1 (not at all true) to 4 (exactly true). Cronbach's alpha was 0.90, indicating excellent reliability. Life satisfaction was assessed by asking participants to rate how satisfied they are with their life on a scale of 1 (very satisfied) to 5 (very dissatisfied) [adapted from the World Values Survey: (42)]. For analysis purposes, this variable was reverse scored.

Level of physical activity was measured by asking participants: How many hours of moderate to vigorous activity (that is, physical activity that makes you breathe harder or puff and pant) would you do in an average week? The definition provided for moderate to vigorous activity was based on the Australian Department of Health's Physical Activity and Sedentary Behavior Guidelines (43). Response options were: 0 h, <1 h, between 1–2 h, between 2–3 h, between 3–4 h, between 4–5 h, and 5 or more h.

Consistent with previous research (44), self-rated health was assessed by asking participants to describe their physical health on a scale from 1 (very good) to 5 (very bad). For analysis purposes, this variable was reverse scored.

Sociodemographic variables included age (treated as continuous), gender (1 = male, 2 = female; treated as dichotomous), living arrangement (1 = lives alone, 2 = does not live alone; treated as dichotomous), and highest level of education attained (no formal school/primary school, high school, technical/trade certificate, undergraduate, postgraduate; treated as continuous).

Descriptive statistics for each modifiable protective factor are presented in Table 2. Univariate linear regression analyses were conducted to assess the relationships between each of the independent variables and the outcome variable of depressive symptoms (treated as continuous). Independent variables found to be significantly associated with depressive symptoms were then simultaneously entered into a linear multivariate regression model using SPSS 26. The Variance Inflation Factor (VIF) was used to assess for multicollinearity.

Moderation analyses were conducted using the PROCESS macro in SPSS to determine if the sociodemographic variables of gender, age, level of education, or living alone moderated the relationship between each of the significant independent variables and the outcome variable of depressive symptoms (45). Each of the independent variables found to be significant in univariate analyses and each of the sociodemographic variables were entered in analyses separately. Bootstrapping was performed (n = 5,000 samples), and a Bonferroni-adjusted alpha level of <0.005 was used to control for the family-wise error rate. Missing data were treated listwise.

Univariate regression analyses showed that life satisfaction, purpose in life, personal growth, self-esteem, social support, self-efficacy, physical activity, self-rated health, educational attainment, and living arrangement were all negatively associated with depressive symptoms (see Supplementary Table 1 for results of univariate regressions). A multivariate regression analysis combining these variables into a single model was used to explore the relative importance of these factors. VIF was <10 indicating that multicollinearity was not an issue. The model explained 55.6% of the variance in depressive symptoms (F_{(10, 734)} = 91.93, p < 0.001). The variables in the model that remained significantly and negatively associated with depressive symptoms at the Bonferroni-adjusted alpha level of <0.005 in descending order of effect were: life satisfaction, self-esteem, and purpose in life (see Table 3).

At the Bonferroni-adjusted alpha level of <0.005, a significant moderating effect of age was observed for life satisfaction (B = 0.13, SE = 0.04, p < 0.002, 95% CI for B [0.05, 0.21]). See Supplementary Table 2 for results of moderation analyses. Post-hoc investigation of this effect showed that life satisfaction was negatively associated with depressive symptoms for all age groups, but the strength of the association was stronger for those participants below the average age of this sample compared to those of mean age or older (see Table 4 and Supplementary Figure 1 for significant results). Age did not moderate the relationships between any of the other independent variables and depressive symptoms, nor were there significant moderating effects observed for gender, living arrangement, or educational attainment (see Supplementary Table 2).

The results from this study highlight the importance of life satisfaction, self-esteem, and purpose in life as focus areas for interventions aimed at preventing and ameliorating depressive symptoms among older people. While self-esteem exhibits trait-like stability (52), and is thus better suited to individual therapeutic intervention, life satisfaction and purpose in life have the potential to be modifiable through population-level interventions. As such, interventions that focus on ways of enhancing life satisfaction and purpose in life are likely to hold most potential for scalable prevention and amelioration strategies. Previous research suggests that encouraging people to participate in meaningful tasks such as hobbies, leisure activities, and/or volunteering can increase their life satisfaction and purpose in life (53–56). These activities have been suggested to provide older adults with opportunities that promote purpose in life and life satisfaction from (i) the relationships formed, (ii) the pursuit of goals, (iii) maintenance of independence, and (iv) engagement with the community (56, 57). These types of activities have also been found to be associated with social support, self-efficacy, and self-rated health (58–60). Further, participation in meaningful tasks appears to be beneficial in helping individuals adjust to age-related losses such as retirement (i.e., loss of work role) and bereavement (53, 55, 60). Encouraging older adults to engage in meaningful activities and facilitating relevant opportunities for them to do so may thus constitute means of preventing and ameliorating depressive symptoms in later life.

The main limitation of the present study was its cross-sectional design. Further research is needed to test the results longitudinally to assess whether the identified relationships hold over time. Another potential limitation was the use of convenience sampling, although the resulting sample was largely similar in profile to the population of older Australians, with the exception of living arrangement (see Table 1). Future studies should seek to access representative samples to test whether these results are generalizable. Similar research could also be conducted in other countries to assess the extent to which the identified relationships are relevant to other cultures. Given the inclusion of a broad range of psychological constructs in this study, a further potential limitation was the risk of social desirability bias in responses. To minimize this risk, an “arms-length” data collection method was used in the form of self-administered surveys. This approach has been found to result in more truthful responses to sensitive questions compared to when an interviewer is present (61).

The primary strength of this study was the large number of potential protective factors incorporated into a model that was tested on a substantial sample of community-dwelling older people. However, despite the wide range of factors included in this study, some potentially relevant variables were not assessed and could be incorporated into future research. In particular, including a measure of objective health could provide additional important data and overcome the limitations associated with relying solely on self-rated health as an indicator of physical well-being. Further, recent research suggests that sleep and diet quality may influence older people's experience of depressive symptoms (16), making these potentially important variables to include in future studies.