In relation to the question about whether breastfeeding practices (eg
supplementing with formula, feeding to schedule)
are taken into account in studies that have not found an association
between breastfeeding and incidence of breast cancer:

In the Michels et al study of the near 90,000 women in the US Nurses
Health Study the authors acknowledged in their discussion section that :
"The women in our cohort gave birth during the 1940s, 1950s and 1960s,
when breastfeeding declined to about 30% of parae.  We found that 64% of
women in our cohort had breastfed at some point in their life; however
generally for a shorter duration than earlier birth cohorts.  Changing
practices have affected not only the likelihood of breastfeeding but
also its duration and frequency.  Token breastfeeding - adhering to a
regular schedule and restricting frequency of breastfeeding - was popular
in the first half of this century, whereas feeding on demand became
popular only later.  London has suggested that even with the same total
duration women who breastfeed less frequently would have an earlier
return of ovulation and therefore less protection, provided lack of
ovulation is the mechansim by which breastfeeding lowers breast cancer
risk."

They comment that this may account for the differences between their
rersults and those of the case control study of Newcomb et al.(ref below)

The authors conclude with "We advocate further studies in younger cohorts
of premenopausal women." and do not rule out the possibility that
breastfeeding lowers the risk of breast cancer, although they found no
association.

Michels K, Willett W, Rosner B et al. Prospective assessment of
breastfeeding and breast canvcer incidence among 89,887 women.  Lancet
1996;347:431-36

Newcomb P, Storer B, Longnecker M et al.  Lactation and reduced risk of
premenopausal breast cancer.  N Engl J Med 1994;330:81-87w