74 - Association of Unit-level First and 36-week Oxygen Saturation Targets with Bronchopulmonary Dysplasia

Publication Number: 3070.74

Birte Staude, University Medical Center Ulm, Ulm, Baden-Wurttemberg, Germany; Héloïse Torchin, Université Paris Cité, PARIS, Ile-de-France, France; Laëtitia Marchand, INSERM, Paris, Ile-de-France, France; Harald Ehrhardt, University Medical Center Ulm, 89075, Baden-Wurttemberg, Germany; Jennifer Zeitlin, Inserm, Paris, Ile-de-France, France

Background: Bronchopulmonary dysplasia (BPD) prevalence varies widely between neonatal units, suggesting potential for improvement through better care practices. We previously found that part of this variance can be attributed to differences in first-week oxygen saturation targets, but the impact of these targets at different timepoints has not been evaluated. While first-week targets are a surrogate of harmful oxygen exposure to the immature lung, targets at 36 weeks post menstrual age (PMA) may influence BPD diagnosis, introducing artefactual measurement differences.

Objective: We aimed to evaluate the associations of unit-level first-week and 36-week oxygen saturation targets with BPD in infants born very preterm (VPT, < 32 weeks' gestation).

Design/Methods: Data come from two prospective population-based cohorts (EPICE and Epipage-2) in 11 European countries in 2011-2012. We included VPT infants admitted to units with ≥40 VPT admissions during the study period (6809 infants, 111 units). Minimum oxygen saturation targets in the first week and at 36 weeks PMA were available from unit questionnaires. We modelled the impact of these targets on BPD (receiving oxygen or ventilatory support at 36 weeks PMA) adjusting for patient characteristics using a multiple-membership model which allows consideration of unit transfers between timepoints. We ran three models: (1) using the new unit's 36-week targets for all transferred infants (2) using the first unit's 36-week targets for infants transferred without oxygen, as their BPD risk is unlikely to be affected by the new unit's targets (3) excluding transferred infants.

Results: BPD prevalence varied between neonatal units from 6 to 42% after adjustment for patient risk factors (Figure 1). The range of minimum targets was 80-92% in the first week and 80-95% at 36 weeks. Targets were correlated within units (Spearman ρ:0.43, Figure 2a) and infants (ρ:0.40, Figure 2b). 32% of infants changed units. Five (0.4%) infants transferred from the first unit without oxygen had BPD. First week targets were similarly associated with BPD in all models (OR[95%-CI] 1.13[1.04,1.22]; 1.12[1.04,1.21]; 1.13[1.03,1.23], Table 1) while the association of 36-week targets was weaker, with no effect in model 1 (1.00[0.94,1.07]) and a less pronounced effect in models 2 and 3 (1.06[0.99,1.13], 1.05[0.97, 1.14]).

Conclusion(s): These results confirm that higher unit first-week minimum oxygen saturation targets are associated with a higher risk of BPD. However, they also suggest that units' targets at 36 weeks may affect assessments of the BPD burden, supporting growing calls to standardize BPD assessment.

Figure 1. Unit-prevalence of Bronchopulmonary Dysplasia.
Prevalence of bronchopulmonary dysplasia (BPD) in infants born very preterm is shown by unit as unadjusted (red) and adjusted for perinatal factors and stabilized using standardized morbidity rates as suggested by the Centers of Medicare and Medicaid Services (blue) prevalence.

Figure 2. Correlation of first- and 36-week minimum oxygen saturation targets.
Minimum oxygen saturation targets in the first week of life and at 36 weeks by neonatal unit (a) and infant (b). Correlation was assessed by Spearman rank correlation.

Table 1. Association of unit-level first week and 36-week minimum saturation targets on BPD risk using multiple membership models.
Note: Missing data on the individual level was imputed by multiple imputation by chained equations. Unit level data was not imputed. Presented are odds ratios with 95%-credible intervals (OR [95%-CI]). n = number of patients, k = number of units. Adjustment included: gestational age, birth weight z-score, sex, 5-minute Apgar score <7, maternal age, mother born outside of Europe, multiparous, admission for preterm labor.