Abstract

OBJECTIVE: To assess the value of broadly based customary physical activity scores, derived from a questionnaire inventory, in predicting 10 year mortality among elderly people. DESIGN: A 10 year survival analysis of participants in the first wave of the Nottingham longitudinal study of activity and ageing who, in face to face interviews in 1985, provided detailed information on customary physical activity, health, and lifestyle. SETTING: Urban and suburban Nottingham PARTICIPANTS: A total of 1042 people aged 65 years and over randomly sampled from general practitioner records. MAIN RESULTS: On the basis of factor scores derived from the interview questionnaire, activity levels were graded as "high", "intermediate", or "low". In Cox regression models controlling for age, health status, and cigarette smoking at the time of the activity assessment, these gradings were significantly related to 10 year survival. Relative to the "high" activity groups, the risk of dying was significantly increased in both the "intermediate" (hazard ratio (HR) 1.53; 95% CI 1.12, 2.09) and "low" (HR 2.07; 95% CI 1.53, 2.79) groups for women, and in the "low" group (HR 1.59; 95% CI 1.12, 2.25) for men (p < 0.01 throughout). CONCLUSION: Since the survival model controlled for age, health status, and cigarette smoking, it is unlikely that the activity gradings used here are simple proxies for physical health. It is concluded, therefore, that within the elderly population, recall based survey assessments covering a wide range of customary or habitual physical activities, can provide indices showing both cross sectional utility and predictive validity.