Mike: Hello, I'm Mike Clarke, podcast editor for the Cochrane Library. The question of whether and how to treat preterm infants with postnatal corticosteroids for the management of bronchopulmonary dysplasia, which is a chronic lung problem due to prematurity, has been a neonatologist's quandary for decades. In August 2023, Cochrane Neonatal published a network meta-analysis on the use of postnatal corticosteroids for its prevention. In this podcast, Roger Soll of Cochrane Neonatal, talks with the lead author Susanne Hay, an attending neonatologist at Beth Israel Deaconess Medical Center in the US, about her team's work.

Roger: Hello, Susanne. First off, could you tell us why it's important to have reliable evidence about the effects of postnatal corticosteroid treatment for these patients?

Susanne: Hello, Roger. Bronchopulmonary dysplasia, or BPD for short, is a major cause of illness in preterm infants, and corticosteroids are one of the few treatments that effectively prevent it. However, these drugs have their own potential risks, and their use in this population dropped dramatically in the late 90s and early 2000s with reports of concerning side effects, including serious neurosensory problems such as cerebral palsy. Corticosteroid treatment courses have since been modified to mitigate these risks, but the optimal treatment regimen to balance these benefits and harms is unknown.

Roger: So, we have the individual trials but how might a network meta-analysis help us to navigate this?

Susanne: Although some trials have shown that corticosteroids are effective, there are few studies that directly compare different systemic corticosteroid regimens, such as dexamethasone and hydrocortisone. Therefore, the technique of network meta-analysis, or NMA, allows us to use existing direct comparisons to build indirect comparisons and, fortunately, the patients in the various placebo-controlled trials are likely similar enough to allow us to combine their results using this technique, and to build the necessary indirect comparisons.

Roger: That's a lovely idea. Has it been done before?

Susanne: Yes, two other groups have applied network meta-analysis to this question but what distinguishes our NMA is that we built two separate networks, to distinguish between evidence for early treatment, which is in the 7 days after birth, and for late treatment, 7 days or more after birth. We felt that this was important to ensure sufficient similarity between the included patients, because early- and late-treatment groups likely represent different patient populations, with those requiring ventilator support beyond the first week of life likely to be distinctly sicker. We hoped that this would allow us to provide more reliable results in our analysis.

Roger: Thanks. Did you find the evidence you needed and what does it say about the effects of dexamethasone and hydrocortisone for preventing BPD?

Susanne: Our network meta-analysis was only able to add a small amount of information to the existing data. In the absence of many direct comparisons, our resulting network comparisons relied heavily on indirect evidence, and there was notable heterogeneity and imprecision across the networks. Our results suggest that late high-dose dexamethasone or early moderate-dose dexamethasone may have the best effects for survival without BPD, but the certainty of the evidence is low. There was not enough evidence to make a judgement on a preferred treatment with regard to neurodevelopmental outcomes.

Roger: Overall, what's your take-home message about the evidence on systemic corticosteroids for BPD?

Susanne: While our inability to show a clear winner for a corticosteroid treatment regimen using the NMA technique is disappointing, it's also an important lesson in understanding the limitations of this analysis tool and the implications of network design and interpretation. In our case, overall, there are insufficient data to guide treatment with regard to the optimal corticosteroid regimen for BPD. More trials directly comparing treatment regimens are needed, especially with trials designed to evaluate survival without major neurosensory disability.

Roger: Thanks, Susanne. If people want to read your team's network meta-analysis, where can they find it?

Susanne: It's available online at Cochrane Library dot com. If they search for "corticosteroids and BPD", they'll see a link to our network meta-analysis near the top of the list.