Résumé :
|
[BDSP. Notice produite par INIST IR0xA7XF. Diffusion soumise à autorisation]. Background So far, no randomized controlled trials with a mean mammographic screening interval of >=2 years has demonstrated statistically significant mortality reduction for women younger than age 50. The issue of screening frequency is vital in detection of primary breast cancer. Methods The study group consisted of cancers diagnosed in women who participated in a serial screening programme with a mean screening interval of 2 years. To study the effectiveness of the screening, a comparison is made between the distribution of age at which the tumour could be detected when biennial mammographic screening is the only detection method, and the distribution of age at which the tumour would be detected by either biennial mammographic screening or the development of symptoms. Some recently developed statistic methods, such as bootstrap, the maximum likelihood distribution estimator for doubly censored data and the EM algorithm, are used in estimation of these distributions. Results The hypothesis tests and confidence intervals show that the difference between the two distributions was statistically significant for women younger than 50 and 50-70 years old, but not for women over 70 years. Conclusions The statistical analysis indicates that for women younger than 50, and 50-70 years of age, a screening mammogram every other year is not frequent enough to detect primary breast cancer, but for women over 70 years, it might be sufficient.
|