Risk Factors and Breast Cancer Screening

Breast cancer is an important health concern for aging women. Though mammography can lead to early detection of breast cancer, it can also induce women without breast cancer to undergo the physical and emotional stress of cancer treatment. Current mammography screening guidelines, which are the subject of intense controversy, recommend mammography for women in their 50s but not for women in their 40s. These guidelines are based on large randomized controlled trials that occurred in the 1980s. I have obtained data from one of these trials, the Canadian National Breast Screening Study (CNBSS), through an agreement with the PI, Dr. Anthony Miller. The CNBSS is an ideal trial to study because it finds no impact of mammography on women in their 40s. Furthermore, the data contain extensive information on potential risk factors, and they follow women for 25 years.

I propose to re-analyze the CNBSS data using new econometric techniques that I have advanced to examine treatment effect heterogeneity in experiments. I have applied these methods to examine treatment effect heterogeneity in the Oregon Health Insurance Experiment in Kowalski (2016). I find that although compliers with the randomization increased their emergency room (ER) utilization when they gained health insurance through the lottery, always takers increased utilization even more, and never takers would have decreased their utilization. I also find that individuals with ER utilization in the pre-experimental period were most likely to be always takers and least likely to be never takers, thus identifying a “risk factor” for increased ER utilization. I propose to apply these methods to the CNBSS. I aim to use my findings to help to target mammography to the women most likely to benefit from it, ultimately improving women’s health.