[Oral contraceptives and breast cancer: analysis of the statistical power of the association].

The power of the association between oral contraceptives and breast cancer was analysed in all the papers published up to date. Seventy-seven publications (from 44 studies) were collected and graded as to quality using meta-analytical methods. Power achieved a figure of greater than or equal to 0.8 in a 10.8% of the associations studied. It showed a significant relationship with the existence of a significant relative risk of the oral contraceptives for breast cancer. The relationship with the sample size of a study was not linear. Power did not show any significant relationship to other variables related to the design of a study (apart from matching, being the power higher in unmatched studies), or to the biases detected, although studies considered as unbiased yielded a higher power. Logistic regression analysis included as predictors of a power greater than or equal to 0.80 the existence of a significant relative risk and the lack of biases in a research.