Methods
This was an observational cohort study with two groups to the cohort. For the first “NICU group”, we included infants that were part of a larger prospective study of infants admitted to our single-center level IV NICU from 2015-2017. Eligible infants were born <32 weeks’ gestation, with BPD defined as respiratory support at 36 weeks post-menstrual age, and discharged with home oxygen10,11. We excluded infants with major surgical non-respiratory comorbidities and tracheostomies, and we included only one member of a multiple gestation. In our NICU, infants anticipating discharge with home oxygen receive a pediatric pulmonary consultation in the NICU, pass an overnight pulse oximetry test on the prescribed liter flow of oxygen, and generally have a capillary blood gas with a PCO2 ≤ 60 mm Hg prior to discharge. Because of this clinical practice, we thought few infants in our center would have high pre-discharge pCO2s, so we also reviewed records from our pediatric pulmonary BPD clinic to include a “referral group” of infants discharged from other local level III NICUs and referred to our clinic for home oxygen management. Infants in the referral group had the same inclusion and exclusion criteria as the prospective study cohort but did not have the same centralized inpatient discharge practice guidelines. Caregivers were consented for the larger prospective observational cohort study in our NICU; the referral cohort was considered exempt as a retrospective chart review. Only records from our institution, either our NICU or outpatient clinic, were used for the study.
In our BPD clinic, both the NICU and referral groups are managed beginning around 4-6 weeks after NICU discharge by a single pediatric pulmonary care team, using a home oxygen weaning algorithm (Figure 1). Infants who pass a room air trial in clinic are followed every 4-6 weeks until home oxygen is discontinued. Infants who do not pass a room air trial in clinic are followed every 2-3 months; after a room air trial is passed, they are followed every 4-6 weeks until they are off oxygen. Successful weaning from oxygen is confirmed by a home oximetry trial, performed with the Respironics 920M Plus or Nonin 2500 PalmSTAT pulse oximetry monitors, and interpreted by our team.
Our primary exposure variable of interest was the closest capillary blood gas pCO2 prior to NICU discharge. This information was available both on infants discharged from our NICU and those referred from other hospitals because we record pre-discharge blood gas data as part of our clinic intake. For infants in our NICU group, we evaluated a secondary exposure variable of pCO2 at 36 weeks corrected gestational age. Our hospital practice is to obtain capillary blood gas testing at 36 weeks corrected age for all infants still receiving respiratory support, as part of pulmonary hypertension screening; this information was not available for infants in the referral group. We included this secondary exposure variable because we thought pCO2 obtained at a standardized time point may be more reflective of NICU illness severity3.
Our primary outcome was readmission by one year after NICU discharge for respiratory reasons, defined as overnight admission with increased respiratory symptoms as identified by manual chart review. Secondary outcomes included other respiratory hospital encounters, medical management related to BPD, and home oxygen weaning. Hospital encounter outcomes included intubation, readmission to an intensive care unit, visits to the emergency room, and total duration of inpatient admissions; these encounters were rare outside of our health system but if they occurred, we abstracted details from the clinic record. Medical respiratory management outcomes included receiving systemic steroids or increased diuretics, bronchodilators, or inhaled corticosteroids above baseline; these were recorded from all clinic, inpatient or emergency department encounters. Home oxygen outcomes included the total duration of home oxygen in weeks post-NICU discharge, any increases in home oxygen flow above baseline, and the number of failed room air trials in clinic, home oximetry trials and sleep studies. We also recorded the number of BPD clinic visits in the first year following discharge, number of missed clinic visits and the number of times oxygen was weaned unsupervised.
We reviewed patients’ NICU charts for demographic data and neonatal variables including sex, multiple gestation, gestational age at birth, birth weight, antenatal steroid use, surfactant use, patent ductus arteriosus ligation, BPD severity in grades as proposed by Jensen et al12; the number of days of ventilation, non-invasive ventilation, and supplemental oxygen. At discharge, we recorded the corrected gestational age at discharge, amount of home oxygen, and receipt of diuretics, bronchodilators, and inhaled corticosteroids. Infants referred to our institution from other NICUs were limited in the availability of some neonatal variables by what was abstracted from the referring NICU discharge summary as part of our clinic intake.