Neurodevelopmental and Growth Impairment Among Extremely Low-Birth-Weight Infants With Neonatal Infection FREE

The increased survival of extremely low-birth-weight (ELBW) infants (ie, weighing 401-1000 g at birth) has heightened awareness of the importance of assessing and improving long-term outcomes associated with prematurity.^{1 - 4} It is estimated that as many as 15% of the most immature infants develop cerebral palsy (CP) and approximately half develop cognitive and behavioral deficits.^{1 - 5} Cerebral white matter damage, identified by cranial ultrasound or magnetic resonance imaging, is a powerful predictor of CP in low-birth-weight preterm infants.^{5 - 6} Investigators have reported associations among antenatal infection, a fetal inflammatory response, vasculitis, white matter damage, and long-term disability.^{7 - 10}

Infections are a frequent complication among ELBW preterm infants^{11 - 12} and are associated with short-term sequelae and an increased risk of death. This study was undertaken to determine if neonatal infections are associated with adverse neurodevelopmental and growth sequelae in early childhood.

The National Institute of Child Health and Human Development (NICHD) Neonatal Research Network maintains a registry of very low-birth-weight (VLBW) infants (weighing 401-1500 g at birth).¹³ Trained personnel collect maternal and delivery data soon after birth and infant data for 120 days (with end points of discharge or death). Surviving infants who weighed 1000 g or less at birth are asked to return for a comprehensive visit at 18 to 22 months of corrected gestational age. Outcomes of ELBW infants born between January 1, 1993, and August 31, 2001, who participated in the VLBW registry and follow-up studies were assessed. Infants with major congenital malformations/syndromes and those with ventricular shunts were excluded because of possible adverse effects on outcome. The institutional review boards at each center approved participation in the registry and the follow-up studies. Written informed consent was obtained from parents or legal guardians for follow-up.

Maternal and neonatal data included maternal age, antenatal antibiotic and steroid use, timing of rupture of membranes, mode of delivery, birth weight, gestational age, sex, race/ethnicity, postnatal surfactant, antibiotic, and steroid use, and diagnoses of respiratory distress syndrome (RDS), patent ductus arteriosus (PDA), intraventricular hemorrhage (IVH), periventricular leukomalacia (PVL), bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), and infection. Maternal race, determined by chart abstraction, was recorded for the infant. Bronchopulmonary dysplasia was defined by use of supplemental oxygen at 36 weeks of postconceptional age (PCA), NEC was classified according to the system of Bell et al,¹⁴ and IVH was classified according to the method of Papile et al.¹⁵

Growth charts developed by Alexander et al¹⁶ were used to classify infants as small for gestational age at birth, defined by a birth weight lower than the 10th percentile for sex and gestational age. Infants were classified at less than the 10th percentile for head circumference at birth using intrauterine growth data reported by Thomas et al.¹⁷ These data were further used to classify study infants at less than the 10th percentile for weight and head circumference at 36 weeks of PCA, compared with infants born at 36 weeks of gestation (data used to create curves in article by Thomas et al,¹⁷ Reese Clark, Pediatrix Medical Group, written communication, April 1, 2004). When the 36-week PCA measurements were unavailable, the closest measurement between 34 and 38 weeks was used.

Early-onset (≤72 hours of birth) and late-onset (>72 hours) sepsis (EOS and LOS, respectively) were defined by a positive blood culture and antibiotic therapy for 5 or more days. Meningitis was defined by a positive cerebrospinal fluid culture and antibiotic therapy for 5 or more days. Cultures positive for organisms generally considered to be contaminants (ie, corynebacterium, propionibacterium, diphtheroids) were excluded. Infants were classified by type of infection as follows: uninfected (no EOS, LOS, late-onset culture-negative clinical infection, NEC, or meningitis); clinical infection alone (late-onset cultures negative but antibiotic treatment for ≥5 days); sepsis alone (EOS/LOS); sepsis and NEC (not always concurrent); or meningitis with or without sepsis. Patients with NEC were included because of the strong association of NEC with infection.¹⁸

The follow-up visit included an interview with the infant’s mother or other primary caregiver, an assessment of cognitive and neuromotor development using the Bayley Scales of Infant Development II¹⁹ and a neurologic examination (both by certified examiners), ascertainment of vision and hearing by caregiver report, and measurement of weight, length, and head circumference. Bayley scale scores provide mental (MDI) and psychomotor (PDI) developmental indexes. The mean score is 100; a score of less than 70 (>2 SDs below the mean) indicates significant delay. Infants judged to be so severely delayed that they were untestable were assigned MDI and PDI scores of 49. Cerebral palsy was defined as a nonprogressive disorder characterized by abnormal tone in at least 1 extremity and abnormal control of movement and posture. Vision impairment was defined as blindness in one or both eyes or need for corrective lenses. Hearing impairment was defined as hearing aids in one or both ears. A composite outcome, neurodevelopmental impairment, was defined as MDI score of less than 70, PDI score of less than 70, CP, bilateral blindness, or bilateral hearing impairment. Centers for Disease Control and Prevention growth charts were used to determine growth status at follow-up (weight, length, and head circumference <10th percentile for sex and age).²⁰

In a secondary analysis, infants in the sepsis alone group were classified by the following pathogen types: coagulase-negative staphylococci; other gram-positive; gram-negative; fungal; and combinations (>1 episode of infection, each with a different pathogen, or polymicrobial bacteremia); and compared with uninfected infants.

Statistical significance for unadjusted comparisons was determined by χ² tests. Logistic regression models, adjusting for potential confounding variables, were used to compare infected and uninfected infants. Weight and head circumference at 36 weeks of PCA and outcomes evaluated at follow-up (MDI <70; PDI <70; CP; vision impairment; hearing impairment; neurodevelopmental impairment; and weight, length, and head circumference <10th percentile) were compared. Models included infection group (or pathogen type), study center, gestational age, birth weight, sex, race/ethnicity, rupture of membranes more than 24 hours before delivery, mode of delivery, multiple birth, antenatal antibiotic and steroid use, postnatal surfactant and steroid use, RDS, BPD, PDA, IVH grade 3 or 4, PVL, and maternal age at the time of delivery. Covariates were entered as categorical variables except maternal age at the time of delivery. Models fitted to the follow-up outcomes also included caregiver’s education at the follow-up visit (high school degree or not). Because not all infants were evaluated for PVL, a 3-level variable (yes, no, or missing) was included in all models. Each model was also run adding head circumference of less than the 10th percentile at 36 weeks of PCA to explore whether head circumference was a mediating variable between infection and outcome. Head circumference was entered as a 3-level variable because of missing values. Odds ratios (ORs) and 95% confidence intervals (CIs) are reported. Wald χ² tests were used to determine significance for comparisons between infection or pathogen groups, with P<.05 considered statistically significant. Analyses were performed using SAS statistical software.²¹

Between January 1, 1993, and August 31, 2001, 13 013 ELBW infants were entered into the network’s prospective registry; 4160 infants died before discharge and 226 died after discharge. Infants with major congenital malformations (n = 337), those with ventricular shunts (n = 159), and those with positive blood cultures but receiving antibiotic therapy for less than 5 days (and therefore considered probable contaminants; n = 239) were excluded from this analysis. Thus, 7892 infants were eligible for follow-up and 6314 (80%) completed the follow-up visit.

Of the 6314 infants who completed follow-up, 6093 were analyzed and categorized as follows: 2161 (35%) were uninfected; 1538 (25%) had clinical infection alone; 1922 (32%) had sepsis alone (EOS, n = 52; LOS, n = 1844; both, n = 26); 279 (5%) had sepsis/NEC (EOS, n = 5; LOS, n = 266; both, n = 8); and 193 (3%) had meningitis (with sepsis, n = 136 [EOS, n = 2; LOS, n = 128; both, n = 6], or without sepsis, n = 57). Thirteen infants could not be classified because of missing information and 208 infants with various combinations of sepsis, meningitis, and NEC were excluded because of small numbers.

Infection status of the infants who died (and were therefore unavailable for follow-up) and of all survivors was assessed. Because limited information was available for those who died in the first 12 hours after delivery, we evaluated infants who died more than 12 hours after delivery. Infection was included in the coded cause of death for 27% of these infants. Because most deaths among ELBW infants occurred in the first week of life, many did not survive long enough to develop LOS. Thus, infants who died more than 12 hours after delivery were more likely than survivors to be uninfected (46% vs 35%) and less likely to have clinical infection (11% vs 24%) or sepsis alone (23% vs 30%). Overall, they were more likely to have had NEC with or without sepsis (17% vs 7%). Approximately 4% in each group had meningitis. Compared with the survivors seen at the follow-up visit, infants who were alive but did not complete their follow-up visit were more likely to be uninfected (39% vs 35%), and the percentages in each infection group were 1% to 2% lower than for study infants (P = .001).

Maternal and neonatal characteristics of the study population are presented in Table 1. Compared with the uninfected infants, those with infection were more immature, had lower birth weights, were more frequently male, and had several complications that have been linked to adverse long-term sequelae, including severe IVH, PVL, BPD, postnatal steroid use, and poor growth (Table 2). After adjusting for study center and maternal and neonatal variables, infants in each infection group were significantly more likely than the uninfected infants to have a head circumference of less than the 10th percentile at 36 weeks of PCA (adjusted OR vs uninfected for clinical infection alone, 1.8; 95% CI, 1.5-2.2; for sepsis alone, 2.5; 95% CI, 2.1-3.0; for sepsis/NEC, 4.0; 95% CI, 2.9-5.6; and for meningitis, 3.0; 95% CI, 2.1-4.4) and to be at less than the 10th percentile for weight at 36 weeks of PCA (adjusted OR vs uninfected for clinical infection alone, 1.8; 95% CI, 1.5-2.2; for sepsis alone, 2.2; 95% CI, 1.9-2.7; for sepsis/NEC, 3.1; 95% CI, 2.1-4.5; and for meningitis, 2.2; 95% CI, 1.4-3.3).

Overall, 41% of children assessed at 18 to 22 months of corrected gestational age had at least 1 adverse neurodevelopmental outcome (data not shown). In unadjusted comparisons, children with infection had significant increases in most adverse outcomes (Table 3). In general, infants without infection were least likely to have adverse outcomes, while those with sepsis/NEC were most likely. Overall 62% of children in the cohort had weight, length, or head circumferences that were less than the 10th percentile at follow-up (data not shown). By univariate analyses, infected children were more likely than those who were uninfected to have growth failure (<10th percentile) at follow-up (Table 4).

After adjusting for study center and maternal and neonatal variables, statistically significant differences for all neurodevelopmental outcomes except hearing impairment were found between children in most infection groups and the uninfected group (Table 5). Only children who had sepsis or sepsis/NEC were at an increased risk of hearing impairment. When head circumference that was less than the 10th percentile at 36 weeks was added to each model, differences between infected and uninfected groups were reduced somewhat; however, a significant association remained for all outcomes except CP (without head circumference, P = .009; with head circumference, P = .07).

Children in each infection group were significantly more likely to have head circumferences that were less than the 10th percentile at follow-up than uninfected children (Table 5). In contrast, only children in the sepsis/NEC and the meningitis groups were more likely than those who were uninfected to be at less than the 10th percentile for length and only those in the sepsis/NEC group were more likely to be at less than the 10th percentile for weight.

Children in the sepsis alone group were evaluated by infecting organisms (coagulase-negative staphylococci, n = 925; other gram-positive, n = 278; gram-negative, n = 197; fungal, n = 105; and combinations, n = 399) and compared with uninfected children. Eighteen children with sepsis were excluded because pathogen information was unavailable. The majority of children had 1 episode of sepsis. By definition, patients in the combination group had more than 1 episode of sepsis, each with a different organism, or had 1 episode of polymicrobial bacteremia. Unadjusted comparisons between pathogen group and outcomes are shown in Table 6. All adverse neurodevelopmental outcomes except hearing impairment were higher among infected children regardless of pathogen type. Hearing impairment was significantly higher for those with gram-negative or fungal infections and those in the combination group. Compared with uninfected children, those in each pathogen group were more likely to have head circumferences that were less than the 10th percentile at follow-up. Fewer differences were found between children in the pathogen groups and uninfected children for weight and length.

After adjusting for study center and maternal and neonatal variables, statistically significant differences were found for children in some pathogen groups compared with those not infected (Table 7). However, for all outcomes except hearing impairment, we detected no differences in risk by type of infecting pathogen. Compared with infants with coagulase-negative staphylococcus infection, those with gram-negative infections (OR, 2.5; 95% CI, 1.0-6.3) and those in the combination group (OR, 2.7; 95% CI, 1.3-5.5) were significantly more likely to have hearing impairment.

Infection is an important problem among VLBW preterm infants. Previous studies from the NICHD Neonatal Research Network have demonstrated that as many as 25% of these infants have 1 or more positive blood cultures and about 5% have a positive cerebrospinal fluid culture over the course of their neonatal hospitalization.^{11 - 12 ,22} Rates of infection increase with decreasing birth weight and gestational age. Moreover, postnatal infection is associated with an increased risk of neonatal complications, prolonged hospitalization, and death.^{11 - 12}

This study was undertaken to determine if neonatal infections are associated with adverse sequelae in early childhood. Although few studies have directly addressed infection and outcome, neonatal sepsis was first linked to cerebral white matter damage in term infants more than 30 years ago.²³ Follow-up studies of preterm infants have suggested an association between sepsis and cerebral white matter damage and CP.^{24 - 27} A recent report from the United Kingdom noted a 4-fold increase in CP among VLBW infants with a history of neonatal sepsis compared with infants with no history of neonatal infection.²⁷

To our knowledge, this is the largest study to date to evaluate the impact of neonatal infection on adverse outcomes in early childhood. More than 6000 ELBW infants were classified by infection during initial hospitalization and were seen in follow-up at 18 to 22 months of corrected age. Culture-confirmed neonatal infection increased the rate of several adverse neurodevelopmental outcomes that contribute to long-term disability, including significantly delayed mental and psychomotor Bayley Scales of Infant Development II scores, vision and hearing impairment, and CP. Of interest, clinical infection alone was also associated with increased risk for adverse outcomes and infants with sepsis/NEC were at highest risk. Although we suspected that infants with gram-negative or fungal infections might have worse neurodevelopmental outcomes than those infected with other pathogens, the type of infecting pathogen had little effect on outcome among survivors. This may be explained in part by the high mortality rate among those infected with gram-negative and fungal organisms.^{11 - 12}

Intrauterine infection, as evidenced by clinical and/or histologic chorioamnionitis, is associated with cerebral white matter injury and subsequent neurodevelopmental impairment.^{28 - 32} A key role for an inflammatory response by the fetus as well as the mother in the pathogenesis of brain injury has been postulated.^{33 - 35} Proinflammatory cytokines in amniotic fluid and in fetal or neonatal blood appear to increase the risk for neonatal brain injury and adverse long-term outcome.

Studies to elucidate the pathophysiology of brain injury in infants with neonatal infection are warranted. Infecting organisms and/or their microbial products can stimulate the production of proinflammatory cytokines. Experimental data indicate that inflammatory cytokines may be neurotoxic in vitro and in vivo and may increase the permeability of the preterm blood-brain barrier.^{36 - 40} Although it appears that the inflammatory cytokine response precedes and contributes to brain injury, a cytokine response may also be the result and/or marker of damaged white matter. Efforts to reduce the inflammatory responses of the neonate with infection might reduce the risk of brain injury associated with infection, but must be approached with caution.^{41 - 42}

Neonates with infection are also at risk for circulatory and/or respiratory insufficiency with decreased systemic blood pressure, hypoxemia, and pathologic alterations in cerebral blood flow. A maturation-dependent impairment in regulation of cerebral blood flow in ELBW⁴³ infants increases the risk of ischemic injury. The long-term impact of cerebral ischemia-reperfusion precipitated by sepsis-related cardiovascular instability is unclear. Volpe^{44 - 46} and others have shown that oligodendroglial precursor cells, the major cellular target in the pathogenesis of white matter injury/PVL, are particularly vulnerable to free radicals that are generated in response to ischemia-reperfusion. The role of infection and cytokines in the pathogenesis of PVL might be related to effects on cerebral hemodynamics, to the generation of reactive oxygen species/free radicals, or to direct toxic effects on vulnerable oligodendroglial precursors.^{44 ,46}

Hearing impairment was more frequent among children who survived neonatal sepsis or sepsis/NEC, especially if they were infected with gram-negative agents or had polymicrobial bacteremia or multiple infections. This finding likely reflects that aminoglycosides are often used to treat these infections. At first glance, it is curious that infants with meningitis were not more likely to have hearing impairment. However, because gram-positive organisms have been more commonly associated with neonatal meningitis among network infants,²² prolonged therapy with aminoglycosides may be less likely. Aminoglycoside ototoxicity is a known complication of prolonged high drug levels; at lower levels there appears to be genetic susceptibility to ototoxicity.^{47 - 48} It is unclear whether infants with hearing loss who received aminoglycosides in the neonatal period should be screened for the presence of the mitochondrial mutations known to increase risk.

Many VLBW infants fail to achieve their ultimate growth potential as a result of intrauterine and postnatal growth failure.⁴⁹ Impaired head growth and subnormal head size are associated with poor cognitive function and academic achievement at 8 years of age,⁵⁰ suggesting a link between postnatal growth and neurodevelopmental potential. Recently, the NICHD Neonatal Research Network reported that while 16% of ELBW infants were small for gestational age at birth, by 36 weeks of PCA, 89% had growth failure.⁵¹ Furthermore, by 18 to 22 months of corrected age, 40% still were at less than the 10th percentile for weight, length, and head circumference.⁵² Our study demonstrates that infection affects weight and head circumference at both 36 weeks of PCA and 18 to 22 months of corrected gestational age. Of greatest concern was our finding that infants with neonatal infections were significantly more likely to have poor head growth. The long-term impact of impaired growth during and following neonatal infections deserves further study.

The major limitation of our study is that neurodevelopmental and growth status were assessed by a single follow-up visit. Outcomes may change as children get older; ie, some may improve, some remain unchanged, or some may worsen. Although patients who returned for follow-up had slightly higher rates of infection, it is unlikely that these differences between survivors who did and did not return for follow-up were clinically significant or affected the study’s robust findings. It also is difficult to speculate on the possible bias introduced by infants who died before follow-up. Because we studied the most immature ELBW infants, many of whom had very complicated neonatal courses, we do not know whether our findings are applicable to larger preterm or term infants.

In summary, this large cohort study demonstrates an association between neonatal infection in ELBW infants and increased risk of poor neurodevelopmental and growth outcomes. Possible interventions to reduce brain injury associated with infection might include earlier diagnosis and improved therapies, including efforts to stabilize blood pressure and maintain adequate oxygenation, reduction of systemic inflammation and generation of proinflammatory cytokines, and pharmacologic interventions to reduce the impact of reactive oxygen species on vulnerable oligodendroglial precursors. Ultimately, efforts to reduce the high rates of infection in ELBW infants are the most important interventions.