Recall of physical activity in the distant past: the 32-year follow-up of the Prospective Population Study of Women in Göteborg, Sweden.

It has been of interest to researchers in chronic disease epidemiology to determine whether elderly subjects can accurately recall their physical activity patterns in the distant past. As part of a 32-year follow-up of the Prospective Population Study of Women in Göteborg, Sweden, 433 participants currently aged 70-92 years recalled their leisure-time physical activity at the original examination in 1968, when they had been 38-60 years of age. Using a four-point scale, subjects were originally asked to describe leisure-time activity during the previous 12 months. Identical questions were asked in 2000 describing levels in 1968 and current levels. Subjects were more likely to overestimate their previous activity level than to underestimate it: 43.9% of women classified themselves consistently at both examinations; 48.7% overestimated and 7.4% underestimated their previous activity levels. Using current activity as a proxy for recalled levels did not improve the moderate agreement with activity as originally reported in 1968. The average magnitude of the recall bias (recalled minus original values) amounted to one half of a unit on the four-point scale and was not significantly associated with age or body mass index. In summary, systematic errors were observed in remote physical activity recalls by elderly women, who generally remembered being more active 32 years earlier than they had originally reported. Physical activity recall questionnaires for the distant past should be used with caution, particularly in older populations.