Occupational physical activity and risk for breast cancer in a nationwide cohort study in Sweden

Objective: Our purpose was to investigate effects of physical activity on r isk for breast cancer. Methods: From the Swedish nationwide censuses in 1960 and 1970 we defined t hree partly overlapping cohorts of women whose occupational titles allowed reproducible classification of physical demands at work in 1960 (n=704,904) , in 1970 (n=982,270), or with the same demands in both 1960 and 1970 (n=25 3,336). The incidence of breast cancer during 1971-89 was ascertained throu gh record linkage to the Swedish Cancer Register. We used Poisson regressio n to estimate relative risks (RR). Results: A total of 20,419, 22,840, and 8261 breast cancers, respectively, were detected in the three cohorts. In all three cohorts the risk for breas t cancer increased monotonically with decreasing level of occupational phys ical activity and with increasing socioeconomic status. Among women with th e same estimated physical activity level in 1960 and 1970 the RR was 1.3 fo r sedentary as compared with high/very high activity level (95% CI 1.2-1.4; p for trend < 0.001). Adjustment for socioeconomic status virtually elimin ated this association (RR 1.1; 95% CI 0.9-1.2; p for trend 0.12) leaving a statistically significant 30% gradient only among women aged 50-59 years at follow-up. The association between socioeconomic status and breast cancer risk was largely unchanged after adjustment for occupational physical activ ity. Conclusion: The protective effect of occupational physical activity on brea st cancer risk, if any, appears to be confined to certain age groups.