The paper by Mortensen et al1 raises important questions in the study of time trends in fetal growth and other exposures and outcomes.

Fetal growth is usually measured by birthweight for gestational age, and small for gestational age (SGA), typically defined using a percentile cut-off point of the weight for gestational age distribution, is a commonly used outcome in perinatal epidemiology studies.2 SGA is also considered an important risk factor3 4 in studies of perinatal exposures on child and adult outcomes, and therefore implicitly an important intermediate between prenatal exposures and morbidity and mortality. Defining SGA requires reference to a population standard to identify the percentiles. Ignoring the potential for bias that arises using a live birth standard,5 there are several choices of population standards that may have an effect on the study. The primary choice is whether to select an absolute or a relative reference. An absolute reference, where all births are compared with the same reference population (either internal or external), implies that SGA is a …