Analysis of pathological outcome of pregnancy.

A series of publications have demonstrated that women undergoing numerous pregnancies during their reproductive lives show considerably higher pregnancy order-specific abortion rates or perinatal mortality rates than women who have been pregnant less often. This gives rise to a statistical problem of control in connection with analytical epidemiology, especially since these differences are far greater than those expected for rates of abortion or perinatal mortality, for example between various occupational categories. The present article is concerned with statistical control problems of this nature. On the basis of computer simulations, the extent and direction of the bias which inadequate statistical control may bring about is calculated for situations in which the presumed deleterious effect is reflected in gravidity- and pregnancy order-specific relative risks.