Measures

Past month probable depression was measured using the Patient Health Questionnaire (PHQ-9), a 9-item symptom severity rating scale for depression [21]. The PHQ-9 has excellent internal consistency, test–retest reliability, and construct validity [22]. The PHQ-9 has also produces similar results whether administered in person or on the telephone [23]. For analysis, baseline, wave 2, and wave 3 depression were dichotomised as probable depression (PHQ-9 score of 10 or higher) and no depression (PHQ-9 score of 9 and lower).

Physical activity was measured at baseline using the Physical Activity Scale for the Elderly (PASE). The PASE is a 10-item instrument designed to assess physical activity levels in older adults [24]. It covers physical activity domains including, leisure time activities, housework, sports and recreational activities. Specifically, these include: walking, light recreational activities, moderate recreational activities, strenuous recreational activities, exercise specifically to increase muscle strength and endurance, light housework, heavy housework, home repairs, lawn work or yard care, outdoor gardening, and caring for another person. The PASE also includes an item assessing physical activity from employment, but it was excluded in this mostly retired sample; the majority of the participants (88.3%) reported they were either retired, unable to work, unemployed or a student; only 11.7% of the survey participants were employed. The total PASE score is calculated by the average number of hours per day over a one-week period spent on each activity multiplied by a specific weight for each activity. Higher scores indicate higher levels of physical activity. The PASE has been shown to be a reliable and valid measure of physical activity among older adults [25].

Latent class analysis was used to identify groups of subjects with homogeneous patterns of activity: specific details on the methods used to create these classes have been described elsewhere [26]. Model Fit statistics are available in Supplementary data, Appendix Table S1, available in Age and Ageing online. Five classes of physical activity were examined: least active class (little to no activity), walker class (walking only and a little housework), domestic/gardening class (gardening, lawn/yard work, and home repairs), athletic class (mainly sports and/or recreational activities), and domestic/gardening athletic class (sports and recreational activities and home-related activities).

We also accounted for measures of demographic, psychosocial and physical health characteristics that are associated with physical activity and depression, including neuroticism, stressful life events, chronic health conditions, and disability [10, 11, 27]. Demographics variables included age, sex, race (Non-Hispanic White, Non-Hispanic Black, Hispanic, and Other) education (less than high school, high school/GED, some college, Bachelors or more) and household income (less than or equal to $20,000, more than $20,000 to $40,000, more than $40,000 to $80,000, and more than $80,000). Neuroticism was measured by the short neuroticism arm of the Eysenck Personality Questionnaire Brief Version, which has demonstrated high reliability (α: 78) and test–retest reliability (r: 0.92) [28]. Life stress events were measured with a seven-item survey asking respondents if they experienced various stressful events in the past 12 months (e.g. divorce/separation, serious illness in spouse or partner, major financial loss) [29]. Chronic conditions included the self-reported prevalence in the last 12 months (e.g. high blood pressure, stroke, heart problem, diabetes, cancer). Disability was measured using the Basic Activity of Living section of the Functional Status Questionnaire (FSQ); respondents were classified into no disability (score: 0–87) and warning zones of disability (score: 88–100) [30].