Article,
A population-based case-cohort evaluation of the efficacy of mammographic screening for breast cancer.
American journal of epidemiology, 140 (10): 889-901 (November 1994)2615<br/>Comment in: Am J Epidemiol 1995 Aug 15;142(4):448-50;<br/>Dissenys híbrids.
Abstract
Screening efficacy may be evaluated in population-based, observational studies, provided one addresses biases that arise in nonrandomized screening evaluations. The authors evaluated the association of mammographic screening with breast cancer mortality among women without prevalent breast cancer. The cohort consisted of 94,656 women, aged > or = 40 years, enrolled in a large health maintenance organization in Washington State during the period 1982-1988. A case-cohort design limited chart abstraction of risk factor and screening information to 4% of the cohort. There were 1,144 incident breast cancer cases, including 126 who died of breast cancer during the study interval. An age-stratified proportional hazards analysis compared the screening experience of the breast cancer deaths with that of 2,237 controls. The family history of breast cancer, previous biopsy, and parity were included in the model to control for potential confounding effects of selection bias. With a mean follow-up of only 3.5 years postscreening, there was a small but statistically nonsignificant reduction in the risk of breast cancer mortality among women screened 1 year prior to diagnosis compared with unscreened women. The relative risk was 0.80 (95% confidence interval 0.34-1.85) for women aged > or = 40 years and 0.61 (95% confidence interval 0.23-1.62) for women aged > or = 50 years. This paper illustrates how case-cohort methodology can be used to perform efficient assessment of screening efficacy in large cohorts, while eliminating or controlling for sources of bias.
