been shown to enhance risk. The suggested study in young women should distinguish between use before and after the prior benign condition.

Multiple studies of varying types have not consistently shown women who used oral contraceptives before the birth of their first child, or before the age of 25 years, to be at increased risk of breast cancer. Results among studies are more consistent for an increase in risk in women under the age of 45 who were long-term users of oral contraceptives (regardless of whether use occurred before or after a first birth or a particular age). Such users would, of course, tend to have been young at their initial use.

This observed increase in risk is not likely to be due to differential screening for breast cancer in oral contraceptive users. There is suggestive but inconsistent evidence that it could be a result of the pill stimulating growth in preexisting tumors. Alternatively, oral contraceptives, when used early in life, could be causing a small absolute increase in risk in young women. It is of extreme importance to determine which of these two possible biological mechanisms may be operating. If it is the former, then oral contraceptives are not causing any new breast cancers; if it is the latter, they are, and this could theoretically continue and become even more of a problem with the aging of the cohort of women who have been exposed early in life. Rigorous epidemiological studies should therefore be conducted to continue monitoring for any enhanced risk of breast cancer among women in birth cohorts exposed at a young age to prolonged use of oral contraceptives. Such studies should not be confined to women below a specific arbitrary age. They should include a pathological component to measure indices of proliferative activity and determine the likely cells of origin of the tumors (whether ductal or lobular). The analyses of both the histologic and epidemiological data should be planned to provide evidence for or against the two above-mentioned biological mechanisms.

Studies of survival in women with breast cancer would also be of value. If oral contraceptive users have more favorable survival than nonusers, this pattern would be evidence for a screening bias (or suggest that breast tumors that develop in response to oral contraceptives are less aggressive than other breast tumors). If survival is less favorable in users than in nonusers, this pattern would be evidence that oral contraceptives stimulate tumor growth (or suggest that breast tumors that develop in response to oral contraceptives are more aggressive than other breast tumors).

Limited evidence suggests that use of oral contraceptives near the time of menopause may increase a woman's risk of breast cancer. A