689 - Association of the five-minute Apgar score with neonatal outcomes after very preterm birth in an international context

Poster Number: 689
Publication Number: 689.444

Harald Ehrhardt, University Medical Center Ulm, Division of Neonatology and Pediatric Intensive Care Medicine, Ulm, Baden-Wurttemberg, Germany; Rolf F.. Maier, Philipps University of Marburg, Marburg, Hessen, Germany; Ulrika Ådén, Karolinska Institutet and Linköping University, Stockholm, Stockholms Lan, Sweden; Adrien M. Aubert, Université Paris Cité, Inserm, INRAE, CRESS, Paris, Ile-de-France, France; Elizabeth S. Draper, University of Leicester, Leicester, England, United Kingdom; Anna Gudmundsdottir, Karolinska Universitetssjukhuset, Solna, Stockholms Lan, Sweden; Heili Varendi, University of Tartu, Tartu, Tartumaa, Estonia; Tom Weber, University of Copenhagen, Værløse, Hovedstaden, Denmark; Michael Zemlin, Saarland University, Homburg, Saarland, Germany; Jennifer Zeitlin, Inserm, Paris, Ile-de-France, France

Background: The Apgar score was developed for clinical triage, but is used by researchers to identify newborns at risk of poor outcome. Its prognostic validity may be undermined by varying practices for assigning scores, especially in an international context and in very preterm infants (VPTI).

Objective: To assess associations between the Apgar score at 5 minutes of life and mortality and severe neonatal morbidity in an international cohort of VPTI and interactions with country of birth.

Design/Methods: Data of the population-based EPICE cohort of VPTI < 32 weeks from 19 regions in 11 European countries (N=7900 live births) were abstracted from medical records. We compared Apgar score distributions by country of birth. We used a modified Poisson regression to estimate relative risks and confidence intervals describing the association between a score < 7 and adverse outcomes: in-hospital mortality, intraventricular hemorrhage (IVH) grades 3 and 4, late onset sepsis (LOS) among infants surviving to at least 72 hours, bronchopulmonary dysplasia (BPD) among survivors to 36 weeks and retinopathy of prematurity (ROP) grade III or more and surgical necrotizing enterocolitis (NEC) among survivors to discharge. Model 1 included country only; Model 2 added perinatal characteristics (gestational age (GA), multiples, sex and small for GA (< 3rd and 10th percentiles); Model 3 added management factors (antenatal steroids, cesarean, inborn). Using Model 3, we investigated interactions with country.

Results: The mean Apgar score at 5 minutes was 7.7 (SD 2.2) and the median 8, ranging from 6.5 (2.5) in Poland to 8.6 (2.3) in Denmark and 7 to 10 respectively; 20.2% of VPTI had an Apgar score < 7, but variation between countries was high (Figure 1). A score < 7 increased risks of mortality, IVH, BPD and ROP, but not NEC or LOS (Figure 2). No significant interactions with country were detected for mortality (P=.23) or BPD (P=.20), but interactions were significant for IVH (P=.016), with more marked associations in Belgium, Denmark and Poland, and for LOS (P=.009), with stronger associations in Poland, Portugal and Italy. Interactions were significant for NEC (P< .001) and ROP (P< .001), but a smaller number of events complicated the interpretation of individual country effects.

Conclusion(s): In our European sample, Apgar score was associated with most adverse neonatal outcomes.  While associations were consistent across countries for mortality and BPD, significant interactions were observed for the other outcomes. These results imply that caution is needed when using Apgar scores in prognostic models without local validation.