Outcomes of Preterm Infants following Discussions about Withdrawal or Withholding of Life Support

Abstract

Objectives: To describe the frequency of postnatal discussions about withdrawal or withholding of life-sustaining therapy (WWLST), ensuing WWLST, and outcomes of infants surviving such discussions. We hypothesized that such survivors have poor outcomes.

Study design: This retrospective review included registry data from 18 centers of the National Institute of Child Health and Human Development Neonatal Research Network. Infants born at 22-28 weeks of gestation who survived >12 hours during 2011-2013 were included. Regression analysis identified maternal and infant factors associated with WWLST discussions and factors predicting ensuing WWLST. In-hospital and 18- to 26-month outcomes were evaluated.

Results: WWLST discussions occurred in 529 (15.4%) of 3434 infants. These were more frequent at 22-24 weeks (27.0%) compared with 27-28 weeks of gestation (5.6%). Factors associated with WWLST discussion were male sex, gestational age (GA) of ≤24 weeks, birth weight small for GA, congenital malformations or syndromes, early onset sepsis, severe brain injury, and necrotizing enterocolitis. Rates of WWLST discussion varied by center (6.4%-29.9%) as did WWLST (5.2%-20.7%). Ensuing WWLST occurred in 406 patients; of these, 5 survived to discharge. Of the 123 infants for whom intensive care was continued, 58 (47%) survived to discharge. Survival after WWLST discussion was associated with higher rates of neonatal morbidities and neurodevelopmental impairment compared with babies for whom WWLST discussions did not occur. Significant predictors of ensuing WWLST were maternal age >25 years, necrotizing enterocolitis, and days on a ventilator.

Conclusions: Wide center variations in WWLST discussions occur, especially at ≤24 weeks GA. Outcomes of infants surviving after WWLST discussions are poor.

Trial registration: ClinicalTrials.gov: NCT00063063.

Keywords: disability; ethics; newborn; palliative care; prognosis.

Figure 1

Flow diagram for all eligible infants up until time of discharge. References to discussions and WWLST here refer to documented discussions and documented WWLST. For the 397 infants who died before 12 hours, the majority (61.2%) did not have any intensive care initiated.

Figure 2

Variation in mortality by GA at each center. The x-axis shows 18 centers labeled A–R. The y-axis shows mortality rate as a percentage. Centers ranged from 9% to 60% mortality at 22–24 weeks, from 3% to 21% mortality at 25–26 weeks, and from 0%–14% mortality at 27–28 weeks gestation.

Figure 3

Rates of discussion and WWLST for each center. The x-axis shows 18 centers labeled A–R. The y-axis represents rates as percentage of total infants. Rates of discussion varied (range 6.4%–29.9%), as did rates of ensuing WWLST (range 5.2%–20.7%).

Figure 4

The y-axis displays aORs of factors assessed by logistic regression, for discussion about WWLST. The OR is represented by the mean and the 95% CI. Factors with statistically significantly increased OR are above the line and exclude 1 (dotted line). These include male sex (95% CI 1.11–1.68), white race (95% CI 1.16–2.47), GA ≤ 24 weeks (95% CI 3.14–4.86), SGA (95% CI 1.97–3.33), congenital syndromes or malformations (95% CI 1.73–4.17), early-onset sepsis (95% CI 1.57–4.78), proven NEC (95% CI 2.12–3.90), brain injury (95% CI 2.14–3.37), and pulmonary hemorrhage (95% CI 1.55–2.99). In contrast, values below the line and excluding 1 are factors with a significantly lower OR. Only 1 factor analyzed is in this range, and this is any surgery (95% CI 0.51–0.83). Additional factors included in the logistic regression model that were not significant and are not shown are maternal age >25 years (OR 0.98, 95% CI 0.78–1.22); maternal education ≥college (OR 0.95, 95% CI 0.71–1.26); and delivery room epinephrine (OR 1.52, 95% CI 0.97–2.40).

Figure 5

The y-axis displays aORs of factors assessed by logistic regression, for WWLST. The OR is represented by the mean and the 95% CI. Factors with statistically significantly increased OR are above the line and exclude 1 (dotted line). These include maternal age >25 years (range 95% CI 1.02–2.75), proven NEC (95% CI 2.11–9.80), and proportional of days on a ventilator (95% CI 1.10–1.52). In contrast, values below the line and excluding 1 are factors with a significantly lower OR. These include delivery room epinephrine (95% CI 0.14–0.90), steroids for BPD (95% CI 0.26–0.84), and surgery (95% CI 0.19–0.62). Additional factors included in the logistic regression model that were not significant and are not shown are maternal education ≥college (OR 1.52 95% CI 0.75–3.07); GA ≤24 weeks (OR 1.04, 95% CI 0.63–1.74); SGA (OR 1.21, 95% CI 0.67–2.18); syndrome or malformations (OR 1.06, 95% CI 0.45–2.52); early onset sepsis (OR 1.4, 95% CI 0.4–4.6); late onset sepsis (OR 0.79, 95% CI 0.46–1.32); brain injury (OR 0.95, 95% CI 0.59–1.54); pulmonary hemorrhage (OR 0.87, 95% CI 0.45–1.70); and maximum fraction of inspired oxygen (FiO₂) (OR 1.02, 95% CI 0.79–1.31).