Discussion
In this study, we used electronic patient medical records to determine if ICS use and ICS initiation correlated with patient characteristics and respiratory outcomes in infants and children with BPD in the outpatient setting. We found that by completion of the first pulmonary visit, 29% of children with BPD were on an ICS while 71% were not. The infants and children prescribed an ICS in the outpatient setting were found to have patient characteristics different from those who were not prescribed an ICS at all or those prescribed at NICU discharge. These findings suggest that there exist several distinct outpatient BPD phenotypes based on ICS use and initiation. For instance, the children prescribed an ICS as outpatients had a greater likelihood of acute care usage and were more likely to be diagnosed with asthma compared to those not on an ICS. Interestingly we found that ICS initiation in the outpatient setting was driven primarily by healthcare providers that were not pulmonologists. Another interesting observation was that we found no associations between BPD severity scores in the NICU and outpatient ICS phenotypes. In brief, our data indicates a correlation between a decision to initiate and the timing of ICS initiation (i.e., NICU discharge versus outpatient management) with patient demographics and outpatient respiratory outcomes; suggesting an emergence of several different outpatient BPD phenotypes defined in part, by ICS use.
Our study revealed that factors related to the prescription of ICS therapy at the time of NICU discharge were related to the degree of prematurity and respiratory disease. Notably, gestational age, birth weight, need for supplemental oxygen at discharge, length of NICU stay, and BPD severity, (which trended towards significance), were associated with ICS therapy prescribed at discharge. In contrast, children initiated on ICS therapy as outpatients were less likely to be on home supplemental oxygen and more likely to have shorter NICU lengths of stay, implying that a different set of factors may be contributing to decision-making for prescribing an ICS in the outpatient setting. Our outcome data, suggests that the more frequent ED visits and rehospitalizations observed between NICU discharge and the first outpatient pulmonary clinic visit, for a select group of infants and children, may be driving non-pulmonary providers to initiate outpatient ICS therapy.
After the first pulmonary visit, we found a decrease in the magnitude of the adjusted odds ratios for ED visits and rehospitalizations, in subjects prescribed outpatient ICS therapy prior to their first visit. Nevertheless, it should be noted, that the odds ratio for these outcomes still remained significantly above 1, when compared to subjects not on an ICS. This association implies that ICS therapy may only partially ameliorate the likelihood of adverse outpatient outcomes. Other factors, such as social determinants of health, neutrophilic-driven small airway inflammation or small airway dysanapsis may be greater drivers of outpatient respiratory outcomes once children are discharged from the NICU. Further studies are needed to examine these relationships.
In the longer term, we observed striking differences in the development of asthma associated with the timing of prescription of ICS therapy. While the frequency of a diagnosis of asthma is not associated with ICS therapy at the time of NICU discharge (p =0.45), it is associated with being on ICS therapy at the conclusion of the first pulmonary clinic visit (p <0.001). Moreover, over 70% of those prescribed ICS therapy after NICU discharge and before the first pulmonary clinic visit had a diagnosis of asthma, which may suggest the emergence of a new respiratory phenotype, that presents in the outpatient setting. Although our study is limited in identifying potential causes for the emergence of this phenotype, we would speculate that both genetic and environmental factors may be involved, given the associations we observed with a family history of asthma and public insurance coverage for the group of subjects placed on ICS therapy after NICU discharge.
This study did not address whether ICS use was beneficial in modifying respiratory outcomes in the BPD patient population prescribed an ICS. Indeed, a beneficial response to ICS in the BPD population has been shown to be variable.(11, 12) ICS has been shown to improve small airflow obstruction caused by eosinophilic inflammation, however their effectiveness in neutrophilic airway inflammation is less clear. (14) Several studies have suggested that small airway obstruction in children with BPD, is likely driven by a non-eosinophilic inflammatory process. (15-18). One study of ICS use in formerly preterm children between 6 and 12 years of age found that by 12 weeks, ICS use resulted in modest lung function improvement. (19) Although a modest improvement in lung function was found in this study, results may have been different if participants had been enriched for a family history of asthma and asthma diagnosis. Another smaller study in children born <34 weeks-gestation found that the combination of an ICS coupled with a long-acting β agonist was superior to an ICS alone.(20) Larger studies will be needed to address which phenotypes of BPD children will potentially benefit from ICS use to mitigate respiratory morbidities and symptoms.
We also did not address why an ICS was prescribed (at NICU discharge or in the outpatient setting). Nevertheless, three different phenotypes emerged that were associated with distinct demographic features and outpatient respiratory outcomes, independent of BPD severity in the NICU. These different outpatient BPD phenotypes (as described by ICS use and initiation of ICS use) may be driven by genetic influences. (21) however, we cannot rule out a component of caregiver preference in deciding to initiate an ICS or not. In a single center study, we previously reported that children with BPD and a family history of asthma had a greater likelihood of ED visits, and systemic steroid use in the outpatient setting, suggesting that there may be a genetic component influencing ICS use. (22) A more precise understanding of outpatient BPD phenotypes may help with prognosticating respiratory outcomes and modifying long-term outcomes in children at risk for lifetime respiratory morbidities.
Limitations of this study include its retrospective nature and the use of electronic patient records which may contain inaccuracies based on ICD codes and encounter information. Furthermore, our results reflect a large urban center which may not be generalizable to other geographic areas. Furthermore, practice variations in use of ICS therapy in infants and children with BPD likely exist. Nevertheless, different demographic characteristics were found among the three groups based on ICS use and timing of initiation.
In summary, insights gained from this study may improve our understanding of outpatient clinical phenotypes in preterm children with BPD, further improving our understanding of their long-term respiratory outcomes. Furthermore, identifying specific cohorts of children with BPD who stand to benefit from ICS therapy will allow for more robust, evidence-based guidelines for outpatient ICS use in this population moving forward.