I had the pleasure of working with midwife Sarah Grundy last year as she completed research for her Master of Public Health degree. Sarah’s paper has recently been published (Grundy et al., 2021). In it she examined many factors that were associated with babies being born small for gestational age (SGA) and demonstrated that being a migrant woman was an independent risk factor. The full text version of the paper is free to access, and I recommend that you read it.
One of the joys of supervising research students is the new knowledge that you are exposed to along the way. What I learned from Sarah’s work was the extent to which I took growth charts for granted . I had not previously questioned whether assigning a percentile to a particular birth weight was a stable concept that didn’t change over time. It turns out that the situation is not that straight forward.
Growth charts assume that in a healthy population, most biological parameters like birth weight are normally distributed. That is, when graphed they form a bell-shaped curve. Most individuals fit somewhere towards the middle of the curve, with small numbers at the outside ends – the very heavy or the very light babies. Given that babies grow during pregnancy, we don’t expect a baby born at 28 weeks to weigh the same as one born at 41 weeks. Growth charts plot the mean weight (the fiftieth percentile) for each gestational age and typically include an upper and lower boundary line considered to represent the boundaries of “normal”. This is often the 10th and 90th, or the 5th and 95th percentiles. See for example the chart offered by the Fetal Medicine Foundation.
In maternity care we use percentile charts widely. Until I read Sarah’s thesis-in-progress exploring the construction of such charts, I had not really given much thought to how such charts were constructed. I was surprised to learn that growth charts are not stable over time.
Our clinical practice is focussed on attempting to eliminate, or at least manage, the tails at either end of the growth chart. We aim to detect the SGA baby so we can electively shorten the duration of pregnancy with the aim of preventing stillbirth. In recent years we have also changed the parameters used to diagnose gestational diabetes and we manage blood glucose levels more aggressively to prevent babies being large for gestational age (over the ninetieth percentile). As these changes impact on the population from which the percentiles are determined, the birth weights we use to define the percentiles change.
As Sarah wrote in her thesis:
Analysis of births in Victoria identified that planned birth of the fetus at risk of late onset fetal growth restriction in order to prevent stillbirth removes smaller babies from being born at later gestations, resulting in a right-shift in the population distribution of term birthweights (Selvaratnam et al., 2020). This clinical management had artificially increased the birthweight thresholds that defined population centiles (Selvaratnam et al., 2020). As a result 53.8% of the 2,748 term neonates that were over the 3rdcentile in 1983-84 would be classified as under the 3rd centile in 2016-17 (Selvaratnam et al., 2020).
As we modify growth charts, we may end up assigning incorrect significance to the percentiles. For example, if weighing 2500 g and being on the fifth centile in 1983 were associated with a 4-fold increase in risk of perinatal mortality, is the risk still the same for a baby on the fifth centile in 2016, but who weighs 2800 g? For percentile charts to remain useful predictors of outcomes, each modification needs to be accompanied by a recalculation of the actual rate of outcomes occurring with the new percentile boundaries.
Discovering the instability of percentile charts gave me quite a headache! It was a good reminder that there are dangers in taking for granted the “rules of thumb” we use in clinical practice. Nothing is ever quite as simple as it first seems. We need maternity clinicians to be experts in critical thinking rather than simple followers of rules.
Grundy, S., Lee, P., Small, K., & Ahmed, F. (2021). Maternal region of origin and small for gestational age: a cross-sectional analysis of Victorian perinatal data. BMC Pregnancy and Childbirth, 21(1). https://doi.org/10.1186/s12884-021-03864-9
Selvaratnam, R. J., Davey, M. A., Mol, B. W., & Wallace, E. M. (2020). Increasing obstetric intervention for fetal growth restriction is shifting birthweight centiles: a retrospective cohort study. British Journal of Obstetrics and Gynaecology. https://doi.org/10.1111/1471-0528.16215
Categories: New research