• Measurement
• Neonatology
• Obesity

Babies of 4 kg or more at birth—‘big fat babies’—tend to get a bad press. There are several reasons for this. That they are more likely to have a difficult time squeezing themselves down the birth canal is well known and unsurprising, hence their increased rates of shoulder dystocia and clavicle fracture. But they also get caught in the wake of infants of diabetic mothers, which are also big fat babies, and which historically have had a variety of problems attributed to them: polycythaemia, hypocalcaemia, hypoglycaemia, and a propensity for respiratory distress, among others. Babies who either by virtue of their genetic endowment or for no particular reason are merely large can thus suffer investigations and potentially unnecessary treatments, because ‘everyone knows’ that these are dodgy babies who should be treated as if they were infants of diabetic mothers.

Therefore, it is important that Linder et al,1 in their paper, have tackled this belief system. Their data are important on several fronts. Most strikingly, the numbers are very large: from a base population of over 60 000 they had nearly 2766 ‘macrosomic’ infants of 4 kg or more, who were not infants of diabetic mothers, and compared these with an equal number of controls; all collected over 10 years. They present anthropometric data on their subjects, which allows them to develop the notion of ‘symmetric’ and ‘asymmetric’ largeness for dates, and thereby potentially to separate those that were fat from those that were just big. And although they found that rates of problems of all kinds were increased compared to the controls, they were actually very uncommon in absolute terms.

So what can we take away from this work, and what might be its problems? One obvious issue is inherent in the methodology: the big babies were the subject of protocol-driven investigations, while the control group only got tested if they were ‘symptomatic’. In an observational study using routinely collected data this is inevitable, and the main effect will be to inflate the relative risk of any adverse measurement with respect to the controls.

The timing of first blood glucose measurement also creates a concern, in that the first measurement in the macrosomic babies was made at a time when many babies have not recovered from the normal physiological dip in blood glucose that occurs after birth. Had the control babies received a measurement at this time, it is quite possible that very similar rates of ‘hypoglycaemia’ might have been uncovered, all in babies who remained asymptomatic.

The second problem is that just as not all big babies are fat, so not all fat babies are big. By taking an arbitrary cutoff of 4000 g the authors can only show the effect of ‘fatness’ at the upper end of the scale of macrosomia, yet, any effects of fatness may operate at much lower birth weights as well. This paper can shed no light on that issue, but it raises a challenge that others might wish to take up.

The anthropometry is also interesting. One would like to know the properties of the length measurement: ‘routinely measured by trained nurses’, we are told, but anyone who has tried doing this knows that it can be done well or badly, and the authors present no data on the accuracy (validity) or precision (reliability) of the length measurements. For this reason, as the authors state, the weight to length ratio is much the most robust derived value in this study; calculation of Body Mass Index (BMI) requires the length to be squared, and Ponderal Index (PI) requires it to be cubed, both of which greatly magnify any errors in the measurement. We should remember that PI was developed almost a century ago2 on the basis that its cubic transformation of length made it particularly suitable for the measurement of fatness in infants and children, so it is curious that it performed so poorly. Perhaps either systematic underestimation, or systematic overestimation of length may be responsible for the apparently low utility of BMI and PI in this study. Furthermore, while BMI and PI both correlate with adiposity in populations of babies, neither are very good predictors of relative fat mass in individual subjects.

Those familiar with life in the resource-poor world will be disappointed that there were no measurements of mid-arm circumference. The mid-arm circumference to head circumference ratio is widely used globally, and both measurements (arm and head) can be made much more accurately and reproducibly than a baby’s length. If it is proposed that some linear measurement of all babies is valuable for discriminating those at metabolic risk from those who can be left alone, mid-arm circumference has the great advantage that it only takes one person to do it accurately, and there is no reason to square or cube it. It deserves more investigation.

The authors went to some lengths to ensure that none of the mothers of the macrosomic babies were diabetic, but as we know, diabetes is not an all-or-nothing concept, and thresholds for its diagnosis (especially gestational diabetes) are arbitrary. That the mothers of the subjects had not formally been diagnosed as diabetic does not exclude the possibility that they might be genuinely diabetic in a subsequent pregnancy, as mothers of macrosomic babies often progress either to gestational or type 2 diabetes. Therefore, some of the babies in this study will undoubtedly have been infants-of-not-quite-diabetic mothers; but this is a real-world situation and does not detract from the robustness of the data.

None of the above caveats should be taken to detract from the central findings of the paper. The message is this: rates of hypocalcaemia and polycythaemia in babies of 4 kg or more are very low, so there is simply no justification for their routine measurement in those who are asymptomatic. A similar message pertains for blood glucose, for which the overall rate was 1.2%, and even in babies over 4.5 kg at birth was less than 6%. If you want to use the weight to length ratio to further refine targeted measurements, ‘symmetric’ babies, by this index, had a rate of hypoglycaemia of 0.3%.

I would argue that these data support the view that provided that there are no other factors in play, big fat babies (even those over 4 kg at birth) can be left to establish feeds just like any other baby. Unless they show relevant symptoms, they do not need any ‘routine’ or ‘screening’ measurements.