Intrapartum risk factors are important in developing world

EDITOR—Badawi et al's case-control study of neonatal encephalopathy in Western Australia shows that clinical evidence of dysfunction of the central nervous system in newborn infants is associated with a wide range of disorders.1 For most of their study population these disorders had origins before the onset of labour. We have two concerns about generalising their findings to other populations: their definition of neonatal encephalopathy, and the greater importance of intrapartum risk factors for neonatal encephalopathy in poorer populations in the developing world.

The omission of intrapartum criteria from the case definition of neonatal encephalopathy has been advocated previously and removes an important bias affecting other studies.2 The investigators' broad clinical definition of neonatal encephalopathy, however, makes comparison with prevalence studies in other settings problematic. For instance, isolated neonatal seizures are difficult to ascertain clinically.3 What proportion of the 109 infants reported to have seizures had interictal evidence of neurological dysfunction? Other investigators have chosen to exclude established causes of encephalopathy such as overt congenital infection and hypoglycaemia from prevalence studies, so it would be helpful for comparative purposes to know what proportion of the authors' cases had evidence of these.

Finally, the inclusion of 37 infants with birth defects (23% of the cases) clearly has implications for the subsequent analysis of the likely time of the insult. In a high income setting with almost universal antenatal care and a relatively low stillbirth rate, the live birth of neurologically impaired fetuses will probably be maximised. In many low income countries mothers are stunted, do not access antenatal care, have high stillbirth rates, and receive poor obstetric care. Under these conditions intrapartum factors probably remain more important in the causation of neonatal encephalopathy.

Inverse association of risk may be due to easier delivery with elective caesarean section

EDITOR—Badawi et al highlight the complex interaction between antenatal and intrapartum risk factors in the aetiology of newborn encephalopathy among term infants. 1 2 They have suggested that elective caesarean section is a protective factor; this requires some clarification.

The number of cases managed by elective caesarean section is small, and I suggest that the inverse association of risk of encephalopathy is in fact due purely to the absence of difficulty of delivery in this group. This would make more sense both statistically and on the grounds of biological plausibility. If one compares all babies in each group exposed to an apparently difficult delivery then the risk of encephalopathy is greatly increased (table). The corollary is that all babies in each group exposed to an apparently uncomplicated delivery will have a reduced risk of encephalopathy. This does not necessarily mean that the delivery mode is protective.

This report of an apparently protective effect of elective caesarean section on the risk of newborn encephalopathy might be used by those who wish to strengthen their argument in favour of routine elective section. The figures show that 32% of the study population had a complicated delivery. Altogether, 174 additional elective sections would therefore be required to prevent one case of newborn encephalopathy; this is based on the assumption that one can predict antenatally which women will have difficulty in labour and that the encephalopathy is due to a difficult delivery alone. Badawi et al have shown that this is unlikely to be the case.2 The authors have also omitted to report the apparently protective effect of spontaneous vaginal delivery on the risk of newborn encephalopathy (49 (30%) v 161 (40%); unadjusted odds ratio 0.63 (95% confidence interval 0.43 to 0.93)).

Badawi et al have tried to unravel the complexity of the interaction of risk factors in the aetiology of newborn encephalopathy, but it would be a shame if results based on small numbers were taken out of context.

Authors' reply

EDITOR—Ellis and Costello raise two crucial issues. Firstly, they point out that the risk factor profile and the relative importance of individual risk factors in developing nations may differ from those in the developed world. This is important, because the global burden of perinatal morbidity and mortality falls mainly on the developing nations.

Secondly, they emphasise the need for a standard definition of newborn encephalopathy that does not assume an intrapartum aetiology and explicitly states exclusion criteria. Such criteria may include some birth defects, but this requires debate. In our study, all infants meeting entry criteria were included. This maximises generalisability. Only 11 case infants had birth defects sufficient, on their own, to explain the encephalopathy. Exclusion of these infants made no material difference to our conclusions. Of the 109 case infants with seizures, 104 had evidence of interictal neurological dysfunction.

Murphy raises the role of elective caesarean section. As she points out, when uncomplicated deliveries are compared with complicated deliveries, uncomplicated deliveries are associated with a lower risk (crude odds ratio 0.36 (95% confidence interval 0.25 to 0.53)). We would not choose this as a primary analysis because the complexity of delivery is in part an outcome of the causal sequence leading to the newborn encephalopathy and should not therefore be treated as a simple explanatory variable. Nevertheless, the stated association clearly exists. But, crucially, it strengthens our belief about the appropriate way to interpret our findings.

We believe four points to be true.

(1) There is a strong negative association between uncomplicated delivery and newborn encephalopathy.

(2) There is a strong negative association between elective caesarean section and newborn encephalopathy.

(3) If analysis is restricted to women who (when conventional criteria are used) might be expected to benefit from an elective section, a strong negative association between elective section and newborn encephalopathy remains.

(4) A relatively large proportion of case mothers who might have been expected to benefit from an elective section did not receive one.

These observations may be interpreted in two ways. Firstly, our study is observational, and the causal pathways leading to newborn encephalopathy are complex. Consequently, our findings may reasonably be interpreted as showing no more than an interesting association. This conclusion would reflect the fact that we have provided absolutely no definitive evidence of a (negative) causal link between elective section and newborn encephalopathy. Conversely, although the study is observational and the causal pathways are complex, it may be argued that one simple hypothesis that would be entirely consistent with the observed associations is that an elective section can, in some at risk pregnancies, mitigate the risk of newborn encephalopathy.

In conclusion, our study does not prove that an elective section protects against newborn encephalopathy. Nevertheless, our data are consistent with such a protective effect. Furthermore, it is difficult to come up with an alternative explanation that is as coherent. In this situation we believe that one must take a decision-theoretical approach to inference by considering the costs and benefits of accepting one interpretation over the other. The former interpretation implies that nothing needs to be done. The latter suggests that further research needs to be undertaken, with study designs that will allow more to be said about the role of elective section in the aetiology of newborn encephalopathy. Given that elective section may genuinely be protective, we believe that to assume the former interpretation is unacceptable.