Reopening the Debate on Corticosteroids: In Reply

Abstract

In Reply.—We thank Drs Fernandes and Johnson for their letter about the DART (Dexamethasone: A Randomized Trial) study. We wholeheartedly agree with them that clinicians should consider both risks and benefits of any therapy, including postnatal corticosteroids. Risks might include outcomes such as death or cerebral palsy; both are possible consequences when clinicians consider postnatal corticosteroid treatment for a chronically ventilator-dependent infant. On the other hand, benefits might include not only extubation but also, perhaps paradoxically, avoidance of death or cerebral palsy. In a systematic review of all randomized trials of postnatal corticosteroids, we have shown that the risk of the combined outcome of death or cerebral palsy varies with risk of chronic lung disease (CLD) in the control group; infants at low risk might experience net harm from corticosteroids, whereas infants at high risk of CLD might experience net benefit.1 The infants in the DART study2 were clearly at very high risk of CLD: the rate of CLD in the control group was 91%, higher than in any of the randomized, controlled trials in the systematic review.1The other issue to consider in the balance of risks versus benefits is the distinction between a prevention trial (such as that by Watterberg et al3) compared with a treatment trial (such as DART). Thus, unless prevention trials are based on highly accurate predictors of CLD, a larger proportion of infants involved will be exposed to the risks with no expectation of benefit; as such, prevention trials perhaps require a different “safety margin” compared with treatment trials such as the DART study, in which the risk of CLD was so much higher.2Fernandes and Johnson suggest a search for subgroups that might be expected to benefit more from postnatal corticosteroids, as well as considering corticosteroids other than dexamethasone. The DART study2 focused on 1 subgroup: infants who are ventilator dependent after the first week of life; most were much older than 1 week when entered into the study, with a median age at entry of 23 days (interquartile range: 17–32 days). Fernandes and Johnson focus on a possible role for early low-dose hydrocortisone in infants exposed to histologic chorioamnionitis. We encourage them to test their hypothesis in a large randomized, controlled trial that encompasses not only short-term morbidity, such as rates of CLD, but also long-term survival and neurologic outcome so that they will be able to establish all the benefits and risks of early hydrocortisone. However, if they intend, as did Watterberg et al,3 to commence hydrocortisone soon after birth, they may have difficulty in establishing who has histologic chorioamnionitis soon enough and with enough certainty.