97 - Outcomes in Small for Gestational Age infants categorized by Intergrowth-21st and Fenton's standards. A multilevel analysis of the Very Low Birthweight cohort from South American NEOCOSUR network.

Poster Number: 97
Publication Number: 97.438

Cecilia Cocucci, Hospital Universitario Austral, Pilar, Buenos Aires, Argentina; Ivonne L. D'Apremont, Pontificia Universidad Católica de Chile, SANTIAGO, Region Metropolitana, Chile; Maria B. ITALIA CENERE, Austral Universitary Hospital, Buenos Aires, Ciudad Autonoma de Buenos Aires, Argentina; Daniela Masoli, Pontificia Universidad Católica de Chile, Santiago, Region Metropolitana, Chile; Gabriel Musante, Hospital Universitario Austral, Pilar, Buenos Aires, Argentina

Background: Infants with birth weight (BW) in centile < 10th for gestational age (GA) are considered Small for Gestational Age (SGA). Although extremely sensitive, this wide cutoff point lacks specificity for adverse outcomes. In addition, Intergrowth-21st and Fenton's standards provide different centile cutoffs that could introduce classification bias on grey zones.

Objective: To assess the effect of BW for GA condition in VLBW on adverse outcomes considering individual and centers' effect.

Design/Methods: We analyzed a cohort of VLBWs with GA 24 to 31wk born at NEOCOSUR centers between 2013-2020. Infants with major congenital malformations, limitation of therapeutic effort at delivery room and multiple births were excluded. BW was categorized in 3 groups by Intergrowth21st and Fenton's centiles as Adequate for GA (AGA) (≥10th) SGA ( < 10th ≥3rd) and Very SGA (VSGA) ( < 3rd). This BW condition was our main exposure variable. Main outcome was Death anytime at NICU. Secondary outcomes were Perforated Necrotizing Enterocolitis and combined outcomes of Death or Bronchopulmonary Dysplasia, Severe Intracranial Hemorrhage and Periventricular Leukomalacia. To account for centers' cluster effect, we built two multilevel logistic models, one for Intergrowth21st and one for Fenton's. After exploring classical adjusting variables, GA, BW, gender and prenatal steroids were added to the model. Interaction effect of GA on BW condition was explored. Intraclass Correlation Coefficient (ICC) was estimated to explain centers' variance in the model.

Results: We analyzed 6763 infants from 23 centers (Table 1). With both standards, VSGA condition was associated with significantly poorer outcomes in mortality and selected morbidity than AGA infants. However, in contrast to Fenton’s, no significant differences were found in infants classified as SGA by Intergrowth-21st. There was no significant interaction effect of GA on BW condition. Based on estimated ICC 8% of total variance was attributable to centers (Tables 2&3)Conclusion(s): SGA infants as below 10th centile constitute a heterogeneous group where infants below 3rd centile may have significantly higher mortality and morbidities than those over. Using centile 10th as cutoff point may lead to important classification bias that underestimates adverse outcomes in VSGA infants. In addition, using Intergrowth BW standards could increase this bias when compared to Fenton’s standards. High ICC found in our model suggest that centers' clustering effect in large networks should be considered and modelled properly to better explain associations between risk factors and outcomes.
Table 1. Demographic characteristics by standard and birthweight for gestational age condition* Chi Square Test for categorical data. One Way ANOVA test for continuous data.
Table 2. Crude and adjusted outcomes for infants categorized by Intergrowth-21st standardsMultilevel Logistic Regression. ICC=Intraclass Correlation Coefficient