Results:
In the original cohort of 204 infants, 41 needed GT placement before discharge. In the validation cohort of 187 infants, 37 needed GT placement. Respiratory parameters were highly associated with need for GT at discharge in the original test cohort (Table 1) and the validation cohort (Table 2).
We examined both a broader categorization of respiratory support defined as high support (CPAP, NIPPV, CV, HFV) vs. low support (HFNC, LFNC, no support) (Figure 3), as well as the more detailed distribution of respiratory mode (Figure 4). When examined in more detail, the difference in respiratory support in the two groups (GT vs. non-GT) is quite striking. Infants who failed to achieve full oral feeding were on significantly more respiratory support, as well as higher oxygen (Figure3), throughout the NICU course. Both FiO2 and the level of respiratory support do improve over time regardless of GT outcome (Figure 1). Every variable examining duration of continuous respiratory support, including total ventilator days, total ventilator + CPAP days, PMA at last ventilator day, and PMA at last CPAP day (Figure 4), were significantly different between the GT and non-GT infants.
We also looked specifically at need for HFV and GT status. At MUSC, HFV is used as a rescue mode after failure to achieve adequate ventilation and/or oxygenation on conventional ventilator support; hence, infants exposed to HFV at some point during their NICU course would have had significantly higher respiratory support needs at that time. In both the original and validation cohorts, infants who later received a GT were significantly more likely to have some exposure to HFV (P<0.0001).
Given the large differences between respiratory variables in GT and non-GT infants, we also examined the association between respiratory support and PMAff (Figure 2). In the original test cohort for the predictive model, infants who received a GT had their first oral feeding attempt at 40.3w PMA while the non-GT infants’ PMAff on average was 33.6 weeks.6 All variables measuring duration of respiratory support were also highly correlated with PMAff with r values of 0.642- 0.884 and with p values all < 0.001.
To further investigate an additional pulmonary complication seen in the NICU, we examined the incidence of bronchopulmonary dysplasia (BPD) between GT and non-GT infants in the original test cohort (Table 1).