Distant past exercise in women: measures may be reliable, but are they valid?

INTRODUCTION/PURPOSE: Methods to measure lifetime physical activity have been described together with evidence suggesting that they are reliable. We present study findings that challenge the validity of such measures. METHODS: One hundred twenty-seven women aged 50-70 interviewed (31 cases, 96 controls) during a pilot case-control study about risk factors for osteoarthritis of the knee. Interviews used a life course approach with visual cues. Information obtained on past physical activity: (i) Total METs for a large range of activities within each decade between age 10 and age 50, (ii) 10-point rating scales representing level of physical activity during each decade between age 10 and age 50, and (iii) lifetime hours' participation in competitive sports. RESULTS: Any statistical correspondence between METs and activity rating scales was best in relation to women's activity when in their teens (Spearman rank correlation r = 0.40). The highest correlation thereafter related to the most recent decade (r = 0.25), which was low, but there was little or no linear association between MET values and the activity ratings when women were in their 20s or 30s. Around one fifth (20.4%) of women who reported no competitive sports participation during their teens and 18.5% who reported none during their 20s also reported total METs corresponding to the top quartile of physical activity during those periods of their lives. CONCLUSION: Retrospective reports from women of distant past activity levels will likely yield different information depending on the form of questioning. It is unclear which method, if any, is accurate and unbiased. The validity of each method is therefore also questionable.