CASE-CONTROL STUDY OF MAMMOGRAPHY

点击翻译按钮获取中文摘要

The effectiveness of screening mammography for women aged 40-49 and >70 has not been demonstrated in randomized controlled trials and remains a controversial issue despite its high degree of public health importance. No new RCTs, which is the best way to evaluate this question, are currently planned and case-control studies present an option for obtaining new information on the efficacy of screening mammography in a timely manner. We propose to perform such a study among women aged 40-87 in a large pre-paid health plan in Northern California. Far larger than any published to date, this study will include an estimated 1500 women with fatal breast cancer, diagnosed from 1980-1994, who were members of the health plan and potentially exposed to mammography for at least 3 years before their diagnosis. Exposure to screening mammography, as opposed to diagnostic mammography, will be determined by medical audits and compared between cases and controls for specified periods up to the diagnosis. Screening will be evaluated among women 40-49, 50-69, and 70-87 years of age taking into account the detectable preclinical phase of breast cancer. In addition we will evaluate the efficacy of mammography among approximately 300 black and 118 Asian women. The important problems of confounding and healthy screening biases will be addressed by comparing data on risk factors obtained from the medical records before diagnosis and by controlling for the number of previous health checkups and for the number of screening mammograms done prior to an interval being assessed. The estimated relative risk (OR) of breast cancer mortality will be determined by multiple logistic regression for matched sets and efficacy will equal 1 - OR. The research team is experienced in conducting these types of studies and the setting offers a defined and accessible population, a high level of exposure to mammography, an ability to link computer-stored datasets, and the centralized availability of medical records.