The analysis by Yates and Newell points out the need for new trials to determine if there is a safe and effective way to give postnatal steroids for preterm infants. A conclusion that I can only agree with. However their article is limited by one major error, and by the failure to describe one of the serious limitations of the available evidence.
The Major error is contained in the section "CLD severity and CP risk". The authors state that the meta-analysis of Doyle et al "found in babies at high risk of CLD that steroid treatment reduces the risk of CP". This is untrue. It would indeed have been a remarkable result as there is not one single individual trial in their review which has shown a reduction in CP with postnatal steroids.
The meta-regression which these authors are referring to was a comparison of control group CLD frequency to the COMBINED OUTCOME OF DEATH OR CP. This has clearly very different implications. If we examine the data used by Doyle et al for that meta-analysis, the studies with a greater than 50% rate of CLD among controls, CP frequency INCREASED among the steroid treated babies by 36%, while mortality decreased. The combined outcome of death or CP among this subgroup of trials is in fact 65/206 steroid treated, and 69/197 controls.
The serious limitation of all the current evidence that was not discussed is the very high frequency of contamination (treatment of control babies with steroids) in the majority of the studies. Many of the studies having over 50% (and up to 75%)rates of contamination. Such an enormous exposure of controls to steroids, has been one of the major hindrances to understanding the effects on long term outcomes. This rate of contamination also affects meta-regression discussed above, as the studies with higher rates of control group CLD also had high rates of contamination.
The multiple systematic reviews of steroid therapy do show certain circumstances where mortality appears to be reduced, and under certain circumstances the reduction in mortality may be greater than the increase in CP. Future studies of postnatal steroids should target such circumstances and should severely limit the use of steroids among controls, as far as ethically appropriate.
Conflict of Interest:
None declared
Conflict of Interest:
None declared