We appreciate Dr Pharoah's comments on the definitions of incidence and prevalence.[1] However, his interpretation of the data appears erroneous.[2]

Dr Pharoah suggests that perhaps the dexamethasone-treated infants were sicker than the control infants and were saved by the intervention only to go on to suffer from cerebral palsy. This interpretation is incompatible with the data. There were no differences in the prenatal characteristics, delivery type or severity of disease after birth between the groups. There was no evidence to suggest that dexamethasone improves survival as has been borne out by other studies.

In addition, Dr Pharoah points to the infant who had cerebral palsy with normal ultrasound scans. The most likely explanation of these findings is that ultrasound has inadequate sensitivity in order to detect subtle fidings which may be picked up by other methods such as MRI.

Thus, our study does not address issues of prenatal timing of the pathogenesis of cerebral palsy, but rather points to a serious and potentially preventable postnatal cause.

References
(1) Pharoah POD. Dexamethasone treatment and cerebral palsy [Rapid Response]. http://adc.bmjjournals.com/cgi/eletters/fetalneonatal;83/3/F177#EL1 (8 November 2000)

(2) Shinwell ES, Karplus M, Reich D, et al. Early postnatal dexamethasone treatment and increased incidence of cerebral palsy. Arch Dis Child Fetal Neonatal Ed 2000;83:F177-81.

Both the title of this paper and the first paragraph of the discussion imply that the use of postnatal dexamethasone may lead to cerebral palsy. However, it is the misuse of the term "incidence" that gives rise to this interpretation. The authors did not, neither could they, provide incidence data. What they presented was cerebral palsy prevalence data.

If it is accepted that prevalence and not incidence data are provided, then a very different interpretation of the findings can be made. Supposing the cerebral impairment of cerebral palsy occurred prepartum, then the use of dexamethasone may have allowed these children to survive whereas, had they received the placebo, they would have died before a diagnosis of cerebral palsy could be made. This would account for the higher prevalence of cerebral palsy in the dexamethasone compared with the placebo group. Support for this interpretation comes from the author's statement:
Eleven (22%) of the 51 children with cerebral palsy had normal neonatal ultrasound scans. All 11 of these infants were treated with dexamethasone.

This suggests that the cerebral impairment was prenatal in timing. Further support comes from the observation that there were fewer IVH in the dexamethasone group, although the difference did not quite attain the conventional level of statistical significance.

It needs to be appreciated that:
Prevalence = Incidence x Duration of disease.

The duration of the disease is affected by how long the child (or fetus) survives. Unless it is known what happens to the fetus from the time of conception, it is not possible to determine incidence in those diseases that have their origin during uterine development.

Professor Emeritus
POD Pharoah