next up previous
Next: About this document

Breast Cancer Research

Topics Addressed

  1. Experimental Design
  2. Medicine

Case Study

Breast cancer is one of the most common malignancies among women in the U.S. If it is detected early enough-before the cancer spreads-chances of successful treatment are much better. The question is, do screening programs speed up detection enough to matter?

The first large-scale experiment to answer this question was run by a group called the Health Insurance Plan (HIP) of Greater New York, starting in 1963. The subjects (all members of this plan) were 62,000 women age 40 to 64, who were divided at random into two equal groups. In the treatment group, the women were encouraged to come in for annual screening, including a breast examination by a doctor and a mammogram (a kind of X-ray). About 20,200 women in the treatment group did come in for the screening, but 10,800 refused. The control group was offered standard health care that did not include the examination and mammogram. All the women were followed for many years; results for the first 5 years are shown below.

Deaths in the first five years of the HIP screening trial,
by cause. Rates are per 1,000 women.

Does screening cause a decrease in mortality? Which numbers in the table support your view? Explain briefly.
The death rate from all other (non-breast-cancer) causes among women who refused screening is about double the corresponding death rate for women who came in for the exam. Did screening cause this difference? Explain briefly. (Hint: Discuss this question briefly from the point of experimental design.)
In the first year of this trial, 67 breast cancers were detected in the examined group, 12 in the group that refused the examination, and 58 in the control group. Did screening cause an increase in breast cancer? Explain briefly.

next up previous
Next: About this document

Susan E. Stockdale
Fri Nov 4 11:58:56 PST 1994