372 - End organ involvement predicts mortality in infants with HIE

Publication Number: 372.4980

Ulrike Mietzsch, University of Washington School of Medicine, Seattle, WA, United States; Niranjana Natarajan, University of Washington School of Medicine, Seattle, WA, United States; Kendell German, University of Washington School of Medicine, Seattle, WA, United States; Sandra E. Juul, University of Washington, Seattle, WA, United States; Tai-Wei Wu, University of Southern California, Keck School of Medicine, Los Angeles, CA, United States; Hannah C. Glass, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Dennis E. Mayock, UWMC, Lake Forest Park, WA, United States; Yvonne Wu, UCSF, San Francisco, CA, United States; Patrick Heagerty, University of Washington, Seattle, WA, United States; Thomas R. Wood, University of Washington School of Medicine, Seattle, WA, United States

Background: Accounting for the independent contribution of end-organ injury to mortality and long-term outcome in neonates affected by HIE is challenging. Several scoring systems that include detailed estimations of end-organ involvement have been proposed. However, these scoring systems can be labor intensive, costly, and difficult to implement widely as the parameters routinely measured in infants with HIE differ across countries.

Objective: We hypothesized that a simplified, universally applicable method categorizing end-organ involvement in infants with HIE would predict long-term neurodevelopmental outcome and mortality independent of HIE severity and perform similarly to a previously published and more detailed end-organ assessment score.

Design/Methods: We used a published weighted score (WS) and generated a simplified categorical score (CS) classifying each system as affected or not (Table 1). Both scoring systems were compared using data from infants with moderate or severe HIE enrolled in the multi-center HEAL Trial. Infants treated with ECMO were excluded. Death and moderate-severe neurodevelopmental impairment (NDI) in survivors were predicted using the components of the WS and CS, as well as HIE severity, using Poisson regression with log link and robust standard errors. Model performance was compared using the Akaike Information Criterion (AIC) and area under the receiver operator curve (AUROC).

Results: A total of 461 eligible infants with documented outcome data were included and of those, 398 survived to discharge. For both scoring tools, organ score was non-linearly associated with higher mortality and moderate-severe NDI (Fig 1). When predicting NDI in survivors, AIC was lower for WS, but AUROC was similarly poor for both (Fig 2A). No individual component of either score was significantly associated with NDI. When predicting death, the AUROC was similar (0.90 for the CS and 0.91 for the WS), with a lower AIC for the CS suggesting slightly better performance (Fig 2A). Of the individual components, respiratory and renal components of the CS were associated with a significantly higher risk of death (RR 7.55 and 2.06; Fig 2B), whereas in the WS, respiratory and endocrinology involvement were significant (RR 1.47 and 1.36; Fig 2C).

Conclusion(s): Both a weighted (WS) and a categorical (CS) assessment of end-organ injury show similar association with mortality and moderate-severe NDI. The simplified CS can therefore be used to assess the degree of end-organ involvement and its impact on outcome, especially in relation to mortality risk, in neonates affected by moderate or severe HIE.

Table 1
Composition of organ scores and definition for cut-offs. Respiratory: Continuous positive airway pressure (CPAP) or mechanical ventilation (MV) outside the delivery room; Cardiovascular: Hypotension (defined as mean arterial blood pressure < gestational age at birth) and/or left ventricular dysfunction (diagnosed on echocardiography) requiring treatment with inotropic medications and/or hydrocortisone during the first 3 days after birth. Renal: acute kidney injury defined as creatinine > 1.5 mg/dL within the first seven days after birth; Liver: Alanine aminotransferase (ALT) > 100 U/dL within five days of birth; Hematology: Thrombocytopenia defined as platelet count < 100 K and/or disseminated intravascular coagulopathy defined as International Normalized Ratio (INR) > 2.0 within 3 days of birth; Endocrinology: Hypoglycemia defined as blood glucose < 50 mg/dL or Hyperglycemia defined as blood glucose > 200 mg/dL within 24 hours of birth.

Figure 1
Shown are the unadjusted association between weighted score (WS, A) and categorical score (CS, B) and proportion of infants experiencing mortality (red) and moderate-severe neurodevelopmental impairment (NDI) in survivors (blue). Higher scores are associated with increased mortality and moderate-severe NDI. Both, the WS and CS show a similar association with death and NDI. Lines are from a locally estimated scatterplot smoothing (loess) model with standard error in grey. Moderate-severe NDI was defined as a Gross Motor Function Classification System (GMFCS) level of 1 and cerebral palsy (CP), a GMFCS level of ≥2, quadriplegic CVP, or a Bayley Scales of Infant Development, 3rd edition, cognitive score of <85.

Figure 2
Performance of both the weighted score (WS) and categorical score (CS), and association between the subcategories of each scoring tool and mortality. (A) Model performance was similar for the WS and CS predicting death and moderate-severe neurodevelopmental-impairment (NDI) in survivors using 1) Akaike Information Criterion (AIC), where a lower value indicates a better balance of model complexity and accuracy, and 2) area under the receiver operator curve (AUROC), with bootstrapped confidence intervals. (B) Relative risk of death for CS subcomponents after adjusting for all other components and hypoxic-ischemic encephalopathy (HIE) severity. The respiratory and renal categories were associated a greater relative risk (RR) of death. (C) Relative risk of death for WS subcomponents after adjusting for all other components and HIE severity. For the WS, the endocrinology and respiratory categories were associated with a higher risk of death. RR is per unit increase in subscore. RRs in B and C are the outputs from a single Poisson regression model with log link and robust standard errors, with 95% CI. Moderate-severe NDI was defined as a Gross Motor Function Classification System (GMFCS) level of 1 and cerebral palsy (CP), a GMFCS level of ≥2, quadriplegic CP, or a Bayley Scales of Infant Development, 3rd edition, cognitive score of <85.