Prediction of survival for preterm births

Editor—We would like to clarify one or two issues raised in the correspondence about our recent paper in the BMJ.^1–4

The data from Trent were presented not as being good or bad but simply to reflect what was actually happening. They will become less useful with time, but, having established the methodology, we hope to offer biennial updates. It would be possible to provide a more complex model of predicted outcome using additional factors, but we were aware that this would add little to the accuracy of prediction.

We do not agree with Ferriman et al that hospital based data are an acceptable alternative.³ The small numbers make the predictions far less accurate, and the inevitable referral bias also has a marked effect on the results of each unit.⁵ We are currently looking at the quality of the survival of preterm infants at discharge from neonatal care⁴ in terms of respiratory and neurological morbidity. Although this may be of interest, what parents really want to know is the probable long term health status of their infant. Population based outcomes of this type for large numbers of preterm infants are, however, not currently available.

All three letters report survival rates higher than those from Trent. None provides data relating to the outcome of all babies, of the relevant gestation, alive at the onset of labour. This is essential if any comparison is going to compare like with like. Doyle et al comment: “The improving survival rates are reported to obstetricians, who decide whether a preterm infant will be born alive and whether the paediatric team will participate.”⁴ We accept that determining viability is a difficult area for parents and clinicians and that practice varies between centres in the United Kingdom with regard to the most preterm infants. Variation in how these infants are defined and treated will, however, affect survival rates for “liveborn infants.” In units where all liveborn infants are not necessarily admitted to neonatal units or seen by a paediatrician, the sickest infants may not be classified as liveborn, and survival rates will seem more favourable. We have recently reported data supporting this concept.⁵ This study showed that babies aged 28 weeks or less who had been transferred postnatally for intensive care had significantly better survival rates than predicted from scores for disease severity and better than infants whose whole course was in a tertiary centre. These seemed to be simply a selected group.