DISCUSSION
In this study we investigated which clinical characteristics may be associated with successful extubation following the commencement of dexamethasone in less than 30 week GA neonates with BPD. Our analysis show that after we adjusted for the multiple factors measured, three factors remained as likely to be prognostically informative. These included less mature at birth, higher MAP, and greater FiO2 requirement.
There are very few similar studies: most involving older patient cohorts and focus on long-term outcomes. Our study is the first looking specifically at extubation success and the occurrence of both modifiable and unmodifiable individual patient variables. Cuna17evaluated the outcomes of neonates with evolving BPD who were ventilator dependent and treated with systemic dexamethasone at either 14-28 days old vs 29-42 days old. They found that delaying dexamethasone had worse outcomes; it is difficult to compare Cuna’s study to ours as both the patient cohort and the primary outcomes differ, and their data do not allow predicting successful extubation at the time of starting steroids.
A likely explanation as to why the individual variables affect extubation success is that they either increase the risk of BPD (younger GA) or are markers of the severity of BPD (MAP and FiO2requirement). Higher MAP and greater FiO2 requirement will decrease the chance of successful extubation as it is these neonates that have more severe pulmonary inflammation and require higher tidal volumes and oxygen delivery for adequate ventilation and oxygenation. The trends in data (both Table 2 and Table 3 - unadjusted ORs) support this, as those not successfully extubated by day 14 had higher MAPs (13, IQR 11-14 cm H2O) and higher average FiO2 requirements (0.40, IQR 0.33-0.48) compared to those successfully extubated MAP (11, range 10-12 cm H2O) and FiO2 (0.35, range 0.30-0.39). These neonates were also less mature at birth putting them at greater risk of BPD. They also had a higher proportion of grade 3 or 4 CXR changes (66% vs 32%) and higher proportion on HFOV use (41% vs 21%). The higher proportion of neonates requiring HFOV in this study cohort is noteworthy, given that its use is predominantly as a rescue strategy once conventional ventilation was no longer adequate. Whilst we have identified variables associated with failing extubation, it is important to acknowledge which are potentially modifiable, and therefore which factors we as clinicians, can act on to maximise the chances of successful extubation in this high-risk population. Given the large variability of MAP and FiO2 requirement we cannot provide cut-off values at which they are no longer negatively predictive of successful extubation, so clinical acumen is required. It is a balance between the risk and benefit of staying on the ventilator to achieve lung growth and potentially decrease FiO2 and MAP requirements versus risking extubation failure and increasing the number of repeated ventilation courses which we know is detrimental.1 It would be require a randomised controlled trial to investigate such strategies.
Another counter-intuitive finding is that we could not demonstrate the starting dose of dexamethasone to be associated with successful extubation. Having said this, the retrospective nature of this observational study means we were unable to systematically vary neither dose or duration independent of other clinical variables. Different clinicians made these decisions based on a variety of factors, several of which, were possibly not recorded in our dataset. To clarify this issue we would suggest performing a prospective study.
There were some limitations to this study. Its retrospective nature means that the reasoning for the timing of dexamethasone treatment is unknown. There may also have been a confounding effect in our study related to practice variation (across the ten-year period) in practices not related to dexamethasone prescription including but not limited to feeding methodology, medication types, parenteral nutrition usage, and some subtle changes to ventilation practices. It was not possible within the constraints of this dataset to control for these variables. The results may also not be generalizable to other neonatal units with different extubation practices and, given the complexity, a larger sample collected from across multiple neonatal units may be useful.
The strengths of this study include a comparatively large sample size and that as far as we know prescribing practices for dexamethasone remained consistent across the duration of the study. All neonates receiving systemic dexamethasone older than 7 days of age.9 A potential decrease in confounding could be inferred from the reasonable sample size of our study allowing multivariable analysis to adjust for multiple confounding factors including GA, weight and postnatal age.
This study has established a number of factors present at the onset of the first course of dexamethasone are predictive of extubation success. However further work could be done: a more comprehensive multi-centre prospective study which would allow first, a better understanding of the suitability and utility in deferring steroid treatment, and second, the development of prognostic scoring models.