272 THE CHINA STUDYHarvard School of Public Health, and the study he used is the famous
Nurses' Health Study.
Starting in 1976, researchers at the Harvard School of Public Health
had enrolled over 120,000 nurses from around the country for a study
that was intended to investigate the relationship between various dis-
eases and oral contraceptives, post-menopausal hormones, cigarettes
and other factors, such as hair dyes.^5 Beginning in 1980, Professor Wil-
lett added a dietary questionnaire to the study and four years later, in
1984, expanded the dietary questionnaire to include more food items.
This expanded dietary questionnaire was mailed to nurses again in 1986
and 1990.
Data now have been collected for over two decades. The Nurses'
Health Study is widely known as the longest-running, premier study
on women's health.^6 It has spawned three satellite studies, all together
costing $4-5 million per year.^6 When I give lectures to health conscious
audiences, upwards of 70% of the people have heard of the Nurses'
Health Study.
The scientific community has followed this study closely. The
researchers in charge of the study have produced hundreds of scien-
tific articles in the best peer-reviewed journals. The design of the study
makes it a prospective cohort study, which means it follows a group of
people, a cohort, and records information on diets before disease events
are diagnosed, making the study "prospective." Many regard a prospec-
tive cohort study as the best experimental design for human studies.
The question of whether diets high in fat are linked to breast cancer
was a natural outgrowth of the fierce discussion going on in the mid-
1970s and the early 1980s. High-fat diets not only were associated with
heart disease (the McGovern dietary goals), but also with cancer (the
Diet, Nutrition and Cancer report). What better study to answer this
question than the Nurses' Health Study? It has a good deSign, massive
numbers of women, top-flight researchers and a long follow-up period.
Sounds perfect, right? Wrong.
The Nurses' Health Study suffers from flaws that seriously doom its
results. It is the premier example of how reductionism in science can
create massive amounts of confusion and misinformation, even when
the scientists involved are honest, well intentioned and positioned at
the top institutions in the world. Hardly any study has done more dam-
age to the nutritional landscape than the Nurses' Health Study, and it
should serve as a warning for the rest of science for what not to do.