The use of customised versus population-based birthweight standards in predicting perinatal mortality

Objective The objective of this study was to critically examine potential artifacts and biases underlying the use of 'customised' standards of birthweight for gestational age (GA). Design Population-based cohort study. Setting Sweden. Population A total of 782 303 singletons >= 28 weeks of gestation born in 1992-2001 to Nordic mothers with complete data on birthweight; GA; and maternal age, parity, height, and pre-pregnancy weight. Methods We compared perinatal mortality in four groups of infants based on the following classification of small for gestational age (SGA): non-SGA based on either population-based or customised standards (the reference group), SGA based on the population-based standard only, SGA based on the customised standard only, and SGA according to both standards. We used graphical methods to compare GA-specific birthweight cutoffs for SGA using the two standards and also used logistic regression to control for differences in GA and maternal pre-pregnancy body mass index (BMI) in the four groups. Main outcome measures Perinatal mortality, including stillbirth and neonatal death. Results Customisation led to a large artifactual increase in the proportion of SGA infants born preterm. Adjustment for differences in GA and maternal BMI markedly reduced the excess risk among infants classified as SGA by customised standards only. Conclusion The large increase in perinatal mortality risk among infants classified as SGA based on customised standards is largely an artifact due to inclusion of more preterm births.