Re: Letter by Mongelli and Gardosi

Sir,

We thank Mongelli and Gardosi for their interest in our paper (Narchi and Skinner 2009). They correctly noted, as we did, that the observed prevalence of complications in neonates diagnosed to be macrosomic by the customised growth method, was higher than in the group diagnosed by the standard population method. Unfortunately, that difference did not reach statistical significance. The sample size of over 6,000 neonates might not have been large enough to have enough power to detect such a difference. Assuming that hypoglycaemia can occur in 5–10% of macrosomic infants of diabetic mothers, two groups, each with 100 infants of diabetic mothers, one diagnosed by the standard method and the other by the customised method, would be needed to have enough power to detect a significant difference in the rate of that complication. This would translate into a sample size of 60,000 deliveries. We would therefore strongly recommend a repeat of our study on a much larger scale.

We could not understand the odds ratio of 136 which Mongelli and Gardosi calculate for hypoglycaemia, as this ratio cannot be calculated when one (or more) of the outcomes is nil (i.e. no hypoglycaemia in one of the groups). Adding the total number of complications in each group before comparing the odds ratio of the resulting total in each group sounds like an attractive option. However, this is not statistically sound, as the number of complications in one group exceeds the number of infants in that group, as some infants have more than one complication (illustrated in the customised group). Establishing a morbidity score for each infant, as they suggested, would be more appropriate but we are not aware of such a validated score with regard to neonatal complications specific to maternal diabetes. Using composite morbidity scores validated for other conditions such as prematurity or growth retardation might not necessarily be appropriate for infants of diabetic mothers. Had we suggested developing such a new and unvalidated score in our study to try to prove the point they are making, might have appeared as an attempt to prove an association, despite statistical evidence to the contrary.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.