The pitfalls of using birthweight centile charts to audit care

OBJECTIVES: Timely delivery of fetal growth restriction (FGR) is important in reducing stillbirth. However, targeted earlier delivery of FGR preferentially removes smaller babies from later gestations, thereby right-shifting the distribution of birthweights at term. This artificially increases the birthweight cutoffs defining the lower centiles and redefines normally grown babies as small by population-based birthweight centiles. Our objective was to compare updated Australian national population-based birthweight centile charts over time with the prescriptive INTERGROWTH-21st standard. METHODS: A retrospective descriptive study of all singleton births ≥34 weeks' gestation in Victoria, Australia in five two-year epochs: 1983-84, 1993-94, 2003-04, 2013-14, and 2016-17. The birthweight cutoffs defining the 3rd and 10th centile from three Australian national population-based birthweight centile charts, for births in 1991-1994, in 1998-2007, and 2004-2013 respectively, were applied to each epoch to calculate the proportion of babies with birthweight <3rd and <10th centile. The same analysis was done using the INTERGROWTH-21st birthweight standard. To assess change over gestation, proportions were also calculated at preterm, early term and late term gestations. RESULTS: From 1983-84 to 2016-17, the proportion of babies with birthweight <3rd fell across all birthweight centile charts, from 3.1% to 1.7% using the oldest Australian chart, from 3.9% to 1.9% using the second oldest Australian chart, from 4.3% to 2.2% using the most recent Australian chart, and from 2.0% to 0.9% using the INTERGROWTH-21st standard. A similar effect was evident for the <10th centile. The effect was most obvious at term gestations. Updating the Australian population birthweight chart progressively right-shifted the birthweight distribution, changing the definition of small over time. The birthweight distribution of INTERGROWTH-21st was left-shifted compared to the Australian charts. CONCLUSIONS: Locally-derived population-based birthweight centiles are better for clinical audit of care but should not be updated. Prescriptive birthweight standards are less useful in defining 'small' due to their significant left-shift.