Down Syndrome and Maternal Age
Down syndrome (trisomy 21) occurs when an individual is born with three copies of chromosome 21 instead of the normal two copies. The vast majority of these occur as a result of an error in meiosis during gamete formation, usually in oogenesis.
One can demonstrate that the frequency of Down syndrome increases with birth order, with a frequency of about 57 per 100,000 live births in 1st born children rising to about 164 per 100,000 in 5th born children.
[Data and graphs adapted from Rothman K: Epidemiology: An Introduction using data from Stark CR and Mantel N: Effects of maternal age and birth order on the risk of mongolism and leukemia. J. Natl. Cancer Inst. 37(5):687-98, 1966]
However, the frequency of Down syndrome also increases with maternal age, starting at about 40 per 100,000 live births in mothers under the age of 20 and rising slowly at first until age 35-39 when the frequency is about 270 per 100,000 live births, and then jumping to about 855 per 100,000 live births in mothers 40 years of age or older.
It is certainly not surprising that the increase with birth order correlates with the increase in maternal age. Mothers giving birth to their first born will have a younger age distribution than those giving birth to their 5th born. So, which matters? Birth order or maternal age? Is the association between birth order and Down syndrome confounded by maternal age? Or is the association between Down syndrome and maternal age confounded by birth order?
Stratifying the data by both birth order and maternal age, as in the three-dimensional bar graph below, clarifies this by showing the frequency of Down syndrome (on the vertical axis) stratified by both birth order (on the horizontal axis) and maternal age (on an axis projecting away from the reader). The figure shows that at any given maternal age, the birth order has little, if any, effect; the frequency of Down syndrome is low in young moms regardless of birth order, and the frequency of Down syndrome is high in older moms regardless of birth order.
In other words, if we control for maternal age, birth order is not associated with prevalence of Down syndrome; it is not an independent risk factor. However, within each stratum of birth order, prevalence increases with maternal age, meaning that, controlling for birth order, the strong association with maternal age persists. Given our definition of confounding, the effect of birth order was confounded by maternal age, since maternal age made it appear that there was an association with birth order. However, when stratified by both birth order and maternal age, we can see that birth order did not have an independent effect. Its apparent association was totally the result of confounding and overestimation caused by maternal age.