Non-response bias as a likely cause of the association between young maternal age at the time of delivery and the risk of cancer in the offspring.

Some epidemiological studies have shown an association between young maternal age at the time of delivery and risk of cancer in the offspring. In a recent German case-control study, there was a twofold increase in the leukaemia risk for children whose mothers were aged < 20 years at the time of delivery. As the prevalence of younger mothers among control families was particularly low, data on maternal age distributions for the general population of Germany were obtained in order to examine the representativeness of the control sample. Despite the excellent sampling frame based on data from complete and up-to-date population registries and a satisfactory response rate among controls ( approximately 71%), there seemed to be a material deficit of young mothers in the control sample. The prevalence of young mothers in the control group was 1.4%[95% CI 0.8%, 2.0%] compared with 2.9% in the general population and 3.3%[1.8%, 4.8%] among leukaemia cases. The most likely cause of this deficit was non-participation. Non-participation bias is hard to identify if no information is available on non-responders and if it results from only small subgroups among the total control sample. Maternal age and childhood leukaemia are an example where differential non-participation creates a spurious moderate association for a relatively rare exposure situation. But nowadays, many epidemiological studies target the detection of small to moderate risk increases for rare exposures. Therefore, the choice of a representative control group as well as the achievement of an excellent response rate are of prime importance.