Abstract
Objective
To determine the relative contribution of clinical data versus head ultrasound (HUS) in predicting neurodevelopmental impairment (NDI) in extremely low birth weight (ELBW) infants.
Study design
2103 ELBW infants (<1000g) admitted to a National Institute of Child Health and Human Development Neonatal Research Network center who had a HUS within the first 28 days, a repeat one around 36 weeks’ post-menstrual age, and neurodevelopmental assessment at 18–22 months corrected age were selected. Multivariate logistic regression models were developed using clinical and/or HUS variables. The primary outcome was the predictive abilities of the HUS done before 28 days after birth and closer to 36 weeks post-menstrual age, either alone or in combination with “Early” and “Late” clinical variables.
Results
Models using clinical variables alone predicted NDI better than models with only HUS variables at both 28 days and 36 weeks (both p < 0.001), and addition of the HUS data did not improve prediction. NDI was absent in 30% and 28% of the infants with grade IV intracranial hemorrhage or periventricular leukomalacia, respectively, but was present in 39% of the infants with a normal head ultrasound.
Conclusions
Clinical models were better than head ultrasound models in predicting neurodevelopment.
Keywords: Logistic models, Predictive value of tests, ROC curve, Infant, premature, Intracerebral hemorrhage, Leukomalacia, periventricular
Advances in perinatal care have increased survival of extremely low birth weight (ELBW) infants (< 1000 grams).1,2 However, improvements in survival have not led to better neurodevelopmental outcomes.3,4 Many ELBW infants who survive develop major physical or behavioral disabilities such as severe neurological impairment, cerebral palsy, deafness, or blindness.4–6 In addition to these neurodevelopmental disabilities, survivors are at high risk for learning difficulties leading to school failure, stressed families, and increased health care costs.7–11
Clinicians have attempted to predict neurodevelopmental impairment using many perinatal risk factors such as gestational age, race, maternal socioeconomic status, antenatal steroid use, mode of delivery, low Apgar scores, birth weight, severe intracranial hemorrhage, periventricular leukomalacia, bronchopulmonary dysplasia, nosocomial sepsis, and necrotizing enterocolitis. 4,12–15 Decisions to limit therapeutic interventions to avoid futile therapy and postponement of death are often based on a prediction of mortality and neurodevelopmental impairment. Identification of abnormal head ultrasound findings such as severe intracranial hemorrhage (grades III or IV) and periventricular leukomalacia are generally considered strong predictors of handicap 4,14,16 and mortality.17,18 However, the variance in neurodevelopmental impairment explained by severe intracranial hemorrhage/periventricular leukomalacia has been generally low in these studies − 8% for a low mental developmental index (<70 or below 2 Standard Deviations of the mean) at 2 years of age 14 and 5–7% for cognitive, language, and achievement performance at 8 years of age.19 The existing literature on the predictive ability of head ultrasound for neurodevelopmental outcome consists mostly of studies from single centers with small numbers of patients. Identification of severe intracranial hemorrhage (grades III or IV) and periventricular leukomalacia may occur early in the first few hours after birth or later in the clinical course,20 with progression or resolution commonly observed. 16,21–23
Despite the uncertainty regarding the prediction of neurodevelopmental outcome, decisions to withdraw or withhold life-saving support are frequently based, at least in large part, on neurosonographic findings.17,18,24–26 Therefore, this study was designed to compare clinical variables and head ultrasound data in the prediction of neurodevelopmental impairment at 18–22 months corrected age.
METHODS
Population
This study analyzed data from a retrospective cohort of all ELBWinfants (401–1000 g) who were admitted (both inborn and outborn) to any of the 19 participating centers of the National Institute of Child Health and Human Development Neonatal Research Network from January 1, 1998 to June 30, 2001. Infants were included if they had at least two head ultrasounds, one within 28 days after birth and a later ultrasound close to 36 weeks postmenstrual age, as well as follow-up assessment at 18–22 months corrected age. Exclusion criteria were lethal congenital malformations and chromosomal abnormalities. Data were routinely collected by trained research personnel and entered into a database as previously described.4 Data collection was approved by the institutional review board at each of the participating institutions.
Procedures
A comprehensive assessment was performed at 18–22 months corrected age. This assessment consisted of medical history, physical and neurologic examinations, and developmental assessment. The neurologic assessment (including muscle tone, strength, reflexes, angle, and posture) was performed using the Amiel-Tison method.27 Cerebral palsy was defined as a non-progressive central nervous system disorder characterized by abnormal muscle tone in at least one extremity and abnormal control of movement and posture. The developmental assessment consisted of the Bayley Scales of Infant Development II, 28 which included the mental and psychomotor developmental indices. Hearing impairment was defined as the use of hearing aids. Visual impairment was defined as blind with some functional vision or no useful vision. Neurodevelopmental impairment was defined as the infant having one or more of the following: psychomotor developmental index <70, mental developmental index <70, cerebral palsy, and hearing or visual impairment. Functional status was assessed by the presence or absence of independent walking and independent feeding.
Assessment of Head Ultrasounds
According to the standardized Neonatal Research Network data collection procedures, early head ultrasound (HUS-28) was the most abnormal head ultrasound obtained within the first 28 days after birth, and head ultrasound closer to 36 weeks (HUS-36) was the head ultrasound obtained after day 28 and closest to 36 weeks post-menstrual age. HUS-28 data were classified as: normal, blood/echodensity in germinal matrix/subependymal area (grade I), blood/echodensity in the ventricles (grade II), ventricular size enlarged (grade III), blood/echodensity in the parenchyma (grade IV), cystic area in the parenchyma (if a previous grade IV hemorrhage was documented previously), and/or periventricular leukomalacia (cystic area in the parenchyma was also considered periventricular leukomalacia if no grade IV was documented previously) by trained research nurses based on local radiology reports. HUS-36 data were classified as: normal, ventricular size enlarged, cystic area in the parenchyma (if a grade IV was documented previously), cystic periventricular leukomalacia, porencephalic cyst, and/or shunt. Data on size or location of the lesions were not collected. Central readers were not used. The specific frequency and timing of the head ultrasounds was based on clinical status and local protocols at each Network center.
Variables Analyzed
The variables analyzed were those that have been reported as associated with poor neurodevelopmental outcome by other investigators. Clinical variables were grouped as “Early” if they could be assessed by postnatal day 28. These included maternal education, antenatal steroids, antenatal antibiotics, mode of delivery, outborn, male, race, gestational age, 1 minute Apgar score <3, 5 minute Apgar score <3, 5 minute Apgar score 3–6, intubation at delivery, medications for resuscitation, birth weight (in 100 gram increments), intrauterine infection, breech presentation, surfactant, indomethacin <24 hours, indomethacin for patent ductus arteriosus treatment, early sepsis (culture positive), late onset sepsis (culture positive), proven necrotizing enterocolitis (≥ Bell Stage II), seizures, pneumothorax, conventional ventilation, and high frequency ventilation. Clinical variables were grouped as “All” if they could be assessed before discharge home, death, and at 120 days, and included the “Early” clinical variables as well as supplemental O2 at 36 weeks postmenstrual age, steroids for bronchopulmonary dysplasia, and threshold retinopathy of prematurity. Head ultrasound variables were grouped as early (HUS-28) or late (HUS-36) depending on ultrasound timing.
Statistical Analyses
The data were randomly divided into a development set of 70% of the population (n=1472) and a validation set of 30% (n=631). Stepwise variable selection was performed on the development set to obtain clinical and/or sonographic variables significantly associated with neurodevelopmental impairment, at p = 0.1 for the entry and exit criteria. For the combined head ultrasound/clinical models, the head ultrasound variables were forced to remain in the model. Using the selected clinical and head ultrasound variables, multiple logistic regression analysis was performed to develop a regression equation on the development set. This equation was applied to the validation set to predict neurodevelopmental impairment and its components. Neurodevelopmental impairment was defined as mental developmental index<70, psychomotor developmental index<70, cerebral palsy, deafness and/or blindness. The process was repeated for HUS-28 and “Early” clinical variables, as well as for HUS-36 and “All” clinical variables alone and in combination. Predictive abilities of the different models were compared using the area under the curve of the receiver operating characteristic curves.
RESULTS
Demographic and Clinical Characteristics
Of 5867 ELBWinfants admitted to any of the participating centers, 3096 infants had both HUS-28 and HUS-36 (Figure). HUS-28 was done before 7 days in 50%, 39% between 7 and 14 days, and 12% between 15 and 28 days of age. The median age at HUS-36 was day 57 (25th–75th centile of 37–74 days).
Figure.
Flow chart showing numbers of infants available for analysis in the study.
Of the 3009 infants without major malformations/syndromes, death before and after discharge occurred in 6.6% of the patients with normal head ultrasound or grade I intracranial hemorrhage, 11% of those with grade II intracranial hemorrhage, and 14% and 18% of those with grade III and IV intracranial hemorrhage, respectively. Cystic periventricular leukomalacia was noted in 5% of patients who died. 2,271 infants survived to follow-up at 18 to 22 month corrected age. Some or all neurodevelopmental impairment data were missing in 168 infants. Thus, a total of 2,103 infants were included in the analysis of neurodevelopmental impairment. A comparison of the infants lost to follow-up to those followed up shows that infants were comparable for birth weight and grade of intracranial hemorrhage, but there was a lower incidence of periventricular leukomalacia in infants lost to follow-up (Table I).
Table 1.
Comparison of infants lost to follow-up to those followed up. The lost to follow-up includes the infants that had follow-up but data on NDI was missing.
| Lost to Follow-up | Followed up | p-Value | |
|---|---|---|---|
| N | 647 | 2103 | |
| Birth Weight | 777 ± 137 | 773 ± 136 | 0.53 |
| Gestational Age | 26.0 ± 2.0 | 25.8 ± 1.8 | 0.04 |
| Male | 310 (48.0%) | 988 (47.0%) | 0.69 |
| IVH Grade 0 | 425 (66.4%) | 1308 (63.4%) | 0.22 |
| IVH Grade 1 | 71 (11.1%) | 244 (11.8%) | |
| IVH Grade 2 | 52 (8.1%) | 151 (7.3%) | |
| IVH Grade 3 | 62 (9.7%) | 215 (10.4%) | |
| IVH Grade 4 | 30 (4.7%) | 145 (7.0%) | |
| PVL | 24 (3.7%) | 134 (6.4%) | 0.01 |
| Cystic PVL | 7 (1.1%) | 50 (2.4%) | 0.06 |
The mean gestational age for the study population was 26 weeks (± 1.8 weeks SD) and the mean birth weight was 773 grams (± 136 g SD). Eighty percent of the infants were exposed to at least one dose of antenatal steroids, 22% were multiple births, 53% were females, and 47% were black/non-Hispanic.
The overall rate of the combined outcome of death or neurodevelopmental impairment was 50%, with 45% (638/1430) infants with a normal HUS-28, 46% (121/266) of those with grade I intracranial hemorrhage, 58% (102/176) with grade II, 63% (163/259) with grade III, and 76% (140/184) with grade IV having neurodevelopmental impairment or death (p<0.0001 for death or neurodevelopmental impairment by grade of intracranial hemorrhage). 77% (127/164) of those with periventricular leukomalacia and 81% (50/62) of those with cystic periventricular leukomalacia also either died or developed neurodevelopmental impairment. Overall, 2.3% (50/2153 infants) died after discharge, and the post-discharge death rate was comparable among those with normal HUS-28 (2.2%), grade 4 intracranial hemorrhage (2.7%) and those with periventricular leukomalacia (2.2%).
Prediction of Neurodevelopmental Impairment
Neurodevelopment impairment increased with worse head ultrasound findings (Table II). However, neurodevelopmental impairment was not present in 30% and 24% of the infants with grade IV intracranial hemorrhage and cystic periventricular leukomalacia, respectively, but was present in 39% of infants with normal head ultrasounds.
Table 2.
Percentage of infants with each neurological outcome at 18–22 months corrected age by head ultrasound findings.
| Head ultrasound variable | NDI n=929 | MDI<70 n=174 | PDI<70 n=478 | Cerebral palsy n=347 | Blindness n=66 | Deafness n=42 | Non-independent walking n=260 | Non-independent feeding n=318 |
|---|---|---|---|---|---|---|---|---|
| Normal (n=1308) Intracranial hemorrhage | 39.4 | 31.9 | 18.8 | 10.1 | 1.6 | 1.5 | 7.7 | 12.8 |
| grade 1 (n=244) | 40.6 | 31.5 | 18.0 | 17.2 | 2.9 | 1.2 | 10.7 | 13.9 |
| grade 2 (n=151) | 51.0 | 36.9 | 22.3 | 17.2 | 4.0 | 3.3 | 9.3 | 13.9 |
| grade 3 (n=215) | 55.4 | 43.3 | 36.7 | 31.3 | 7.0 | 2.8 | 25.1 | 23.4 |
| grade 4 (n=145) | 69.7 | 52.6 | 55.5 | 51.4 | 11.2 | 4.9 | 42.4 | 28.5 |
| Periventricular leukomalacia (n=134) | 72.4 | 60.3 | 52.8 | 50.0 | 10.5 | 3.7 | 44.0 | 29.1 |
| Cystic periventricular leukomalacia (n=50) | 76.0 | 60.4 | 64.6 | 64.0 | 18.0 | 6.3 | 50.0 | 32.0 |
All infants were counted only once and were assigned the highest grade of intracranial hemorrhage/leukomalacia from either head ultrasound.
Missing values in either the row or column variable were excluded from the analysis:
The “Early” and “All” clinical models were better predictors of neurodevelopmental impairment (areas under the curve = 0.68 for both), compared with the corresponding HUS-28 (areas under the curve = 0.58) and HUS-36 models (areas under the curve = 0.57) (p<0.001 vs. the clinical models) (Figure 2[L1]). The HUS-36 model did not improve prediction of neurodevelopmental impairment when compared to the HUS-28 model (p=0.4). The combined head ultrasound and clinical models were comparable to isolated clinical models for neurodevelopmental impairment prediction for both “Early” (0.68 vs 0.68, p=0.5) and “All” models (0.68 vs. 0.68, p=0.9).
Predictions of the major neurological outcomes were comparable for the HUS-36 model and HUS-28 model except for non-independent walking for which the HUS-36 model was superior (0.71 vs. 0.65, p<0.01). There was an improvement in the predictive ability for mental developmental index<70 (0.72 vs. 0.69, p<0.05), cerebral palsy (0.78 vs. 0.72, p<0.01), and non-independent walking (0.79 vs. 0.74, p<0.01) for the HUS-36/“All” clinical model as compared to the HUS-28/“Early” clinical model.
Multiple logistic regression analyses controlling for clinical variables and timing of head ultrasound revealed that only periventricular leukomalacia and shunt placement in HUS-36 were significantly associated with neurodevelopmental impairment (Table III; available at www.jpeds.com). Head ultrasound and clinical variables findings were significantly associated with various components of neurodevelopmental impairment.
Table 3.
Odds ratios and 95% confidence intervals of clinical and head ultrasound predictors (at p<0.05) on neurodevelopmental outcomes.
| Outcome | “Early” Predictors | OR (CI 95%) | “All” Predictors | OR (CI 95%) |
|---|---|---|---|---|
| Neurodevelopmental impairment | ||||
| Clinical | ||||
| Male | 2.0 (1.6–2.5) | Male | 1.9 (1.5–2.4) | |
| Race+ | 1.3 (1.0–1.6) | Race | 1.3 (1.0–1.7) | |
| Surfactant | 1.7 (1.2–2.3) | Surfactant | 1.5 (1.0–2.1) | |
| High frequency ventilation | 1.5 (1.2–2.0) | Chronic lung disease | 1.4 (1.1–1.8) | |
| Seizure | 2.2 (1.3–3.7) | Late onset sepsis | 1.6 (1.2–2.0) | |
| Late onset sepsis | 1.7 (1.4–2.2) | Retinopathy of prematurity | 1.9 (1.4–2.6) | |
| Birth weight§ | 0.8 (0.7–0.9) | Birth weight | 0.8 (0.8–0.9) | |
| Head ultrasound | ||||
| None statistically significant at p<0.05 | Periventricular leukomalacia | 3.3 (1.7–6.4) | ||
| Shunt | 3.6 (1.6–8.2) | |||
| Mental Developmental Index <70 | ||||
| Clinical | ||||
| Male | 2.4 (1.9–3.0) | Male | 2.3 (1.8–2.9) | |
| Race | 1.5 (1.2–2.0) | Race | 1.6 (1.2–2.1) | |
| Surfactant | 1.6 (1.1–2.3) | Surfactant | 1.5 (1.0–2.3) | |
| High frequency ventilation | 1.5 (1.1–2.0) | Chronic lung disease | 1.4 (1.1–1.9) | |
| Seizure | 2.2 (1.4–3.6) | Late onset sepsis | 1.5 (1.2–2.0) | |
| Late onset sepsis | 1.6 (1.2–2.0) | Retinopathy of prematurity | 1.6 (1.1–2.2) | |
| Birth weight | 0.8 (0.7–0,9) | Birth weight | 0.8 (0.7–0.9) | |
| Head ultrasound | ||||
| Periventricular leukomalacia | 2.4 (1.1–5.2) | Periventricular leukomalacia | 2.3 (1.3–4.4) | |
| Shunt | 2.2 (1.1–4.7) | |||
| Psychomotor Developmental Index <70 | ||||
| Clinical | ||||
| Male | 1.5 (1.1–1.9) | Male | 1.4 (1.0–1.8) | |
| High frequency ventilation | 1.7 (1.2–2.2) | Steroids for bronchopulmonary dysplasia | 1.5 (1.1–2.1) | |
| Seizure | 2.8 (1.7–4.6) | Seizure | 2.3 (1.4–3.9) | |
| Late onset sepsis | 1.8 (1.4–2.4) | Late onset sepsis | 1.8 (1.3–2.4) | |
| Birth weight | 0.9 (0.8–1.0) | Retinopathy of prematurity | 1.8 (1.3–2.6) | |
| Birth weight | 0.9 (0.8–1.0) | |||
| Gestational age | 1.2 (1.0–1.2) | |||
| Head ultrasound | ||||
| Parenchymal blood | 1.9 (1.1–3.3) | Periventricular leukomalacia | 3.2 (1.7–6.0) | |
| Ventricular enlargement | 2.3 (1.4–3.7) | Ventricular enlargement | 1.7 (1.0–2.7) | |
| Shunt | 3.0 (1.5–6.4) | |||
| Cerebral Palsy | ||||
| Clinical | ||||
| Male | 1.5 (1.1–2.1) | Male | 1.4 (1.0–2.0) | |
| Pneumothorax | 1.8 (1.1–2.9) | Pneumothorax | 1.7 (1.1–2.8) | |
| Seizure | 3.4 (2.1–5.5) | Steroids for bronchopulmonary dysplasia | 1.5 (1.0–2.1) | |
| Birth weight | 0.9 (0.8–1.0) | Seizure | 2.8 (1.7–4.6) | |
| Indomethacin | 1.4 (1.0–1.9) | |||
| Birth weight | 0.9 (0.8–1.0) | |||
| Head ultrasound | ||||
| Normal | 0.5 (0.3–0.8) | Periventricular leukomalacia | 5.2 (2.8–9.6) | |
| Parenchymal blood | 2.3 (1.3–3.9) | Ventricular enlargement | 1.9 (1.2–3.1) | |
| Ventricular enlargement | 2.0 (1.3–3.2) | Shunt | 3.7 (1.8–7.8) | |
| Non-independent feeding | ||||
| Clinical | ||||
| Male | 1.8 (1.3–2.4) | Male | 1.8 (1.3–2.5) | |
| Pneumothorax | 1.8 (1.1–2.9) | Pneumothorax | 2.0 (1.2–3.2) | |
| High frequency ventilation | 1.6 (1.2–2.3) | Chronic lung disease | 1.5 (1.1–2.1) | |
| Maternal education* | 1.9 (1.2–3.1) | High frequency ventilation | 1.7 (1.2–2.4) | |
| Birth weight | 0.8 (0.7–0.9) | Gestational age | 1.1 (1.0–1.2) | |
| Birth weight | 0.7 (0.7–0.9) | |||
| Maternal education | 1.9 (1.2–3.1) | |||
| Head ultrasound | ||||
| Ventricular enlargement | 2.3 (1.3–3.9) | Ventricular enlargement | 2.2 (1.3–3.8) | |
| Non-independent walking | ||||
| Clinical | ||||
| Steroids | 0.6 (0.4–0.9) | Steroids | 0.6 (0.4–0.9) | |
| Male | 1.4 (1.0–2.1) | Chronic lung disease | 1.7 (1.2–2.6) | |
| Seizure | 4.5 (2.7–7.3) | Seizure | 3.8 (2.2–6.3) | |
| High frequency ventilation | 1.6 (1.1–2.4) | High frequency ventilation | 1.5 (1.0–2.2) | |
| Late onset sepsis | 1.8 (1.3–2.6) | Late onset sepsis | 1.8 (1.2–2.6) | |
| Birth weight | 0.9 (0.7–1.0) | Birth weight | 0.8 (0.7–1.0) | |
| Head ultrasound | ||||
| Periventricular leukomalacia | 2.4 (1.0–5.5) | Periventricular leukomalacia | 4.4 (2.4–8.3) | |
| Parenchymal blood | 2.2 (1.2–4.0) | Ventricular enlargement | 2,1 (1.2–3.5) | |
| Ventricular enlargement | 3.0 (1.7–5.3) | Shunt | 2.4 (1.2–5.0) | |
Race compares Black and non-Hispanic versus others
Odds ratios for birth weight are given for every 100 grams increment in birth weight
Comparison of maternal education up to 12th grade versus college/graduate degree
DISCUSSION
This study addresses the ability of clinical models, head ultrasound models, and their combination to predict neurodevelopmental impairment at 18–22 months corrected age. Many ELBW infants with a normal head ultrasound later develop neurodevelopmental impairment, although some infants with abnormal head ultrasounds do not have significant impairment. Clinical models were stronger predictors for neurodevelopmental impairment and its components than head ultrasound models. Isolated head ultrasound findings had a poor predictive ability for neurodevelopmental outcome when compared to clinical data and generally demonstrated no improvement in predictive ability over time. Only when combined with clinical data was a head ultrasound closer to 36 weeks post-menstrual age better than an ultrasound in the first 28 days for the prediction of low mental developmental index, cerebral palsy, and non-independent walking. Overall, these models are not optimal for use in individual neonates due to lack of sufficient accuracy, but the models are suitable for evaluation of the relative importance of the predictors.
Periventricular leukomalacia was an important risk factor for neurodevelopmental impairment and its components but only if considered after the first 28 days. Echodensities diagnosed early in life may partially or completely resolve and consequently may not predict neurodevelopmental impairment.22,29,30 Independent of the time of the exam, enlarged ventricles was the only head ultrasound finding that predicted poor neurodevelopmental outcome. It is possible that enlarged ventricles may be a marker of abnormal white matter development. 23,31
Infants with normal head ultrasound and grades I and II intracranial hemorrhage had an incidence of neurodevelopmental impairment ranging from 39% to 51%. When corrected for other variables by regression analysis, a normal head ultrasound was not associated with normal outcome, which is consistent with the recent work by Laptook et al,32 who demonstrated that nearly 30% of ELBW infants with a normal head ultrasound had either cerebral palsy or a low MDI. It is likely that neurodevelopmental impairment will not always be preceded by abnormal ultrasound findings. Detection of subtle white matter damage may be better with magnetic resonance imaging.33,34
Severe intracranial hemorrhage and periventricular leukomalacia are believed to be strong predictors of poor neurodevelopmental outcome.4,14–19 Abnormal ultrasonographic findings are considered often in decisions to withdraw or withhold therapy. Clinicians may consider the cranial ultrasound to provide additional prognostic data to the “pre-test probability” estimated based upon multiple clinical variables (birth weight, antenatal variables, need for resuscitation, severity of illness). The current study shows that head ultrasounds do not reliably predict neurodevelopmental impairment in survivors. When controlled for clinical variables and timing of the exam, only periventricular leukomalacia diagnosed closer to 36 weeks and shunt placement were significantly associated with subsequent neurodevelopmental impairment. The high prevalence of neurodevelopmental impairment in infants with normal head ultrasound or minor grades of intracranial hemorrhage, and the frequent absence of severe neurodevelopmental impairment despite grade IV intracranial hemorrhage and/or periventricular leukomalacia indicate that the association between head ultrasound findings and neurodevelopmental outcome is not as strong as previously believed.14,19
A limitation of this study is that the analysis was constrained to already existing data; however, the database was comprehensive and collected by trained personnel. Although head ultrasound interpretations are subject to interobserver variability35,36 and central readers were not used, the effect of inter-reader variability was minimized by using a predefined classification. Our study may also more closely approximate routine clinical practice as central readers are not used for clinical decision making. Another limitation is that the timing of the head ultrasound (both early and late) were variable, and this variability in timing may influence the observations. Again, however, this may more closely approximate routine clinical practice in which timing is variable. Only infants who survived to follow-up were evaluated. Therefore, differential bias is possible as infants in whom support was withdrawn as a result of severe intracranial hemorrhage would not have been included. Differential bias is also possible in this study as examiners during follow-up were not masked to the infants’ clinical course, head ultrasound findings, or clinical findings on prior evaluations.
Important strengths of the current study include the analysis of a large multicenter cohort of ELBWinfants with a standardized and comprehensive evaluation of neurodevelopment at 18–22 months corrected age. Furthermore, the analyses performed on head ultrasound findings were controlled not only for clinical variables but also for the time of the exam.
Clinical variables were stronger predictors than head ultrasound findings as the addition of head ultrasound data did not improve the predictive abilities of models with only clinical variables. The poor predictive ability for head ultrasound may be partly explained by the use of a classification that considers increasing grades of IVH as a progression of a single disease with cumulative effect.37 Ventriculomegaly is considered a consequence of intraventricular hemorrhage although it may represent atrophy secondary to white matter damage from other causes.31 In addition to the anatomical site and extension of the hemorrhage,38,39 a classification that considers the degree of white matter damage,22 its location,40 and whether these findings are persistent or transient41 may improve the prediction of neurologic outcome. Additional limitations of the current methods of ultrasound analysis include the lack of standardization in determination of ventricular size and of reporting of cerebellar lesions. It is possible that a classification of head ultrasound that takes into account the site, extension, and persistence of the hemorrhage as well as the degree of white matter damage may correlate better with longer-term outcome.
Clinicians often overestimate the incidence of major disability or death in these extremely sick babies, and this may lead to restriction of life support therapies.42,43 The current study documents that head ultrasound findings are poor predictors of outcome and indicates that decisions on withdrawal or withholding support in preterm infants should not be made based solely on head ultrasound findings. A large study with more accurate imaging techniques and/or classifications is required to identify specific characteristics that may improve the predictive ability of imaging studies for neurodevelopmental outcomes.
Acknowledgments
Funding Sources: Supported by cooperative agreements with the National Institute of Child Health and Human Development: U10 HD34216 (Dr. Carlo), U10HD27853 (Dr. Donovan), U10HD40461 (Dr. Finer), U10HD21364 (Dr. Fanaroff), U10HD40461 (Dr. Goldberg), U10HD27851 (Dr. Stoll), U10HD27856 (Dr. Lemons), U10HD21397 (Dr. Duara), U10DH27881 (Dr. Papile), U10HD40521 (Dr. Phelps), U10HD27880 (Dr. Stevenson), U10HD21415 (Dr. Korones), U10HD40689 (Dr. Laptook), U10HD21373 (Dr. Tyson), U10HD40498 (Dr. O’Shea), U10HD21385 (Dr. Shankaran), U10HD27904 (Dr. Oh), U10HD27871 (Dr. Ehrenkranz)
Appendix
List of Participating NICHD Neonatal Research Network Centers during the period of the study
| Center | Principal Investigator (PI) | Follow up Principal Investigator (FPI) | Network Coordinator (NC) | Follow-up Coordinator (FC) | |
|---|---|---|---|---|---|
| 1 | Brown University | William Oh, MD | Betty Vohr, MD | Angelita Hensman, RNC | Lucy Noel, RNC |
| 2 | Case Western Reserve University | Avroy A. Fanaroff, MB BCh | Dee Wilson, MD | Nancy Newman, RN | Bonnie Siner, RN |
| 3 | Duke University | Ronald N Goldberg, MD | Ricki Goldstein, MD | Kathy Auten, RN, BS | Melody Lohmeyer, RN |
| 4 | Emory University | Barbara J. Stoll, MD | Barbara J. Stoll, MD | Ellen Hale, RNC, BS | Ellen Hale, RNC, BS |
| 5 | Harvard University | Ann R. Stark, MD | Ann R. Stark, MD | Kerri Fournier, RN | |
| 6 | Indiana University | James A. Lemons, MD | Anna Dusick, MD | DeeDee Appel, RN | Leslie Richards, RN |
| 7 | Stanford University | David K. Stevenson, MD | Susan Hintz, MD | Bethany Ball, RN, BS | Bethany Ball, RN, BS |
| 8 | University of Alabama | Waldemar A. Carlo, MD | Kathy Nelson, MD | Monica Collins, RN | Vivien Phillips, RN |
| 9 | University of California, San Diego | Neil N. Finer, MD | Yvonne Vaucher, MD | Chris Henderson, RN | Martha Fuller, RN |
| 10 | University of Cincinnati | Edward F. Donovan, MD | Jean Steichen, MD | Cathy Grisby, RN | Tari Gratton, RN |
| 11 | University of Miami | Shahnaz Duara, MD | Charles Bauer, MD | Ruth Everett, RN | Mary Allison, RN |
| 12 | University of New Mexico | Lu-Ann Papile, MD | Lu-Ann Papile, MD | Conra Backstrom, RN | |
| 13 | University of Rochester | Dale L. Phelps, MD | Gary Myers, MD | Linda Reubens, RN | Diane Hust, RN |
| 14 | University of Texas. Houston | Jon E. Tyson, MD, MPH | Brenda Morris, MD | Georgia McDavid, RN | Shannon Rossi |
| 15 | Wake Forest University | T. Michael O’Shea, MD | Robert Dillard, MD | Nancy Peters, RN | Barbara Jackson, RN |
| 16 | Wayne State University | Seetha Shankaran, MD | Yvette Johnson, MD | Gerry Muran, BSN | Debbie Kennedy, RN |
| 17 | Yale University | Richard A. Ehrenkranz, MD | Linda Mayes, MD | Pat Gettner, RN | Elaine Romano, MSN |
| 18 | NICHD | Linda L. Wright, MD, Rosemary D. Higgins, MD | Beth B. McClure, MS | ||
| 19 | Research Triangle Institute | W. Kenneth Poole, PhD | W. Kenneth Poole, PhD | Betty Hastings, Carolyn Petrie, MS | Beth B. McClure, MS |
Footnotes
Presented in part at the Pediatric Academic Societies’ Meeting, San Francisco, CA, 2004
Reprints: None
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Horbar JD, Badger GJ, Carpenter JH, Fanaroff AA, Kilpatrick S, LaCorte M, et al. Trends in mortality and morbidity for very low birth weight infants, 1991–1999. Pediatrics. 2002;110:143–51. doi: 10.1542/peds.110.1.143. [DOI] [PubMed] [Google Scholar]
- 2.Arias E, MacDorman MF, Strobino DM, Guyer B. Annual summary of vital statistics—2002. Pediatrics. 2003;112:1215–30. doi: 10.1542/peds.112.6.1215. [DOI] [PubMed] [Google Scholar]
- 3.Lorenz JM, Wooliever DE, Jetton JR. A quantitative review of mortality and development disability in ELBW. Arch Pediatr Adolesc Med. 1998;152:425–35. doi: 10.1001/archpedi.152.5.425. [DOI] [PubMed] [Google Scholar]
- 4.Vohr BR, Wright LL, Dusick AM, Mele L, Verter J, Steichen JJ, et al. Neurodevelopment and functional outcomes of extremely low birth weight infants in the national institute of child health and human development neonatal research network, 1993–1994. Pediatrics. 2000;105:1216–26. doi: 10.1542/peds.105.6.1216. [DOI] [PubMed] [Google Scholar]
- 5.Wood N, Marlow N, Costeloe K. Neurologic disability after extremely preterm birth. N Engl J Med. 2000;343:378–84. doi: 10.1056/NEJM200008103430601. [DOI] [PubMed] [Google Scholar]
- 6.Doyle LW, Casalaz D Victorian Infant Collaborative Study Group. Outcome at 14 years of extremely low birth weight infants: a regional study. Arch Dis Child Fetal Neonatal Ed. 2001;85:F159–64. doi: 10.1136/fn.85.3.F159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Singer LT, Salvator A, Guo S, Collin M, Lilien L, Baley J. Maternal psychological distress and parenting stress after the birth of a very low birth weight infant. JAMA. 1999;281:799–805. doi: 10.1001/jama.281.9.799. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bhutta AT, Cleves MA, Casey PH, Cradock MM, Anand KJ. Cognitive and behavioral outcomes of school-aged children who were born preterm: a meta-analysis. JAMA. 2002;288:728–37. doi: 10.1001/jama.288.6.728. [DOI] [PubMed] [Google Scholar]
- 9.Petrou S, Sach T, Davidson L. The long-term costs of preterm birth and low birth weight: results of a systematic review. Child Care Health Dev. 2001;27:97–115. doi: 10.1046/j.1365-2214.2001.00203.x. [DOI] [PubMed] [Google Scholar]
- 10.Stevenson RC, Pharoah PO, Stevenson CJ, McCabe CJ, Cooke RW. Cost of care for a geographically determined population of low birth weight infants to age 8–9 years. II Children with disability. Arch Dis Child Fetal Neonatal Ed. 1996;74:F118–21. doi: 10.1136/fn.74.2.f118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Saigal S, den Ouden L, Wolke D, Hoult L, Paneth N, Streiner DL, et al. School-age outcomes in children who were extremely low birth weight from four international population-based cohorts. Pediatrics. 2003;112:943–50. doi: 10.1542/peds.112.4.943. [DOI] [PubMed] [Google Scholar]
- 12.Zernikow B, Holtmannspoetter K, Michel E, Pielemeier W, Hornschuh F, Westermann A, et al. Artificial neural network for risk assessment in preterm neonates. Arch Dis Child Fetal Neonatal Ed. 1998;79:F129–34. doi: 10.1136/fn.79.2.f129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Stoll BJ, Hansen NI, Adams-Chapman I, Fanaroff AA, Hintz SR, Vohr B, et al. Neurodevelopmental and growth impairment among extremely low-birth-weight infants with neonatal infection. JAMA. 2004;292:2357–65. doi: 10.1001/jama.292.19.2357. [DOI] [PubMed] [Google Scholar]
- 14.Ambalavanan N, Nelson KG, Alexander G, Johnson SE, Biasini F, Carlo W. Prediction of neurologic morbidity in extremely low birth weight infants. J Perinatol. 2000;20:496–503. doi: 10.1038/sj.jp.7200419. [DOI] [PubMed] [Google Scholar]
- 15.Hack M, Wilson-Costello D, Friedman H, Taylor GH, Schluchter M, Fanaroff AA. Neurodevelopment and predictors of outcomes of children with weight less than 1000 g. Arch Pediatr Adolesc Med. 2000;154:725–31. doi: 10.1001/archpedi.154.7.725. [DOI] [PubMed] [Google Scholar]
- 16.Pierrat V, Duquennoy C, van Haastert IC, Ernst M, Guilley N, de Vries LS. Ultrasound diagnosis and neurodevelopmental outcome of localized and extensive cystic periventricular leukomalacia. Arch Dis Child Fetal Neonatal Ed. 2001;84:F151–6. doi: 10.1136/fn.84.3.F151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.McMenamin JB, Shackelford GD, Volpe JJ. Outcome of neonatal intraventricular hemorrhage with periventricular echodense lesions. Ann Neurol. 1984;15:285–90. doi: 10.1002/ana.410150315. [DOI] [PubMed] [Google Scholar]
- 18.Volpe JJ. Neurology of the newborn. 4. 2001. Intracranial hemorrhage: germinal matrix-intraventricular hemorrhage of the premature infant; pp. 428–93. [Google Scholar]
- 19.Vohr BR, Allan WC, Westerveld M, Schneider KC, Katz KH, Makuch RW, et al. School-age outcomes of very low birth weight infants in the indomethacin intraventricular hemorrhage prevention trial. Pediatrics. 2003;111:340–6. doi: 10.1542/peds.111.4.e340. [DOI] [PubMed] [Google Scholar]
- 20.Perlman JM, Rollins N. Surveillance protocol for the detection of intracranial abnormalities in premature neonates. Arch Pediatr Adolesc Med. 2000;154:822–6. doi: 10.1001/archpedi.154.8.822. [DOI] [PubMed] [Google Scholar]
- 21.Goetz MC, Gretebeck RJ, Oh KS, Shaffer D, Hermansen MC. Incidence, timing, and follow-up of periventricular leukomalacia. Am J Perinatol. 1995;12:325–7. doi: 10.1055/s-2007-994486. [DOI] [PubMed] [Google Scholar]
- 22.Lai FF, Tsou KY. Transient periventricular echodensities and developmental outcome in preterm infants. Pediatr Neurol. 1999;21:797–801. doi: 10.1016/s0887-8994(99)00092-2. [DOI] [PubMed] [Google Scholar]
- 23.Kuban K, Sanocka U, Levinton A, Allred EN, Pagano M, Dammann O, et al. White matter disorders of prematurity: association with intraventricular hemorrhage and ventriculomegaly. J Pediatr. 1999;134:539–46. doi: 10.1016/s0022-3476(99)70237-4. [DOI] [PubMed] [Google Scholar]
- 24.Cook LA, Watchko JF. Decision making for the critically ill neonate near the end of life. J Perinatol. 1996;16:133–6. [PubMed] [Google Scholar]
- 25.Wall SN, Partridge JC. Death in the intensive care nursery: physician practice of withdrawing and withholding life support. Pediatrics. 1997;99:64–70. doi: 10.1542/peds.99.1.64. [DOI] [PubMed] [Google Scholar]
- 26.Cuttini M, Nadai M, Kaminski M, Hansen G, de Leeuw R, Lenoir S, et al. End-of-life decisions in neonatal intensive care: physicians’ self-reported practices in seven European countries. EURONIC Study Group. Lancet. 2000;355:2112–8. doi: 10.1016/s0140-6736(00)02378-3. [DOI] [PubMed] [Google Scholar]
- 27.Amiel-Tison C. Neuromotor status. In: Taeusch HW, Yogman MW, editors. Follow-up Management of the High-Risk Infant. Boston, MA: Little, Brown Company; 1987. pp. 115–26. [Google Scholar]
- 28.Bayley N. Bayley Scales of Infant Development-II. San Antonio, TX: The Psychological Corporation; 1993. [Google Scholar]
- 29.Ringelberg J, van de Bor M. Outcome of transient periventricular echodensities in preterm infants. Neuropediatrics. 1993;24:269–73. doi: 10.1055/s-2008-1071555. [DOI] [PubMed] [Google Scholar]
- 30.Fawer CL, Diebold P, Calame A. Periventricular leukomalacia and neurodevelopmental outcome in preterm infants. Arch Dis Child. 1987;62:30–6. doi: 10.1136/adc.62.1.30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Leviton A, Gilles F. Ventriculomegaly, delayed myelination, white matter hypoplasia, and “periventricular” leukomalacia: how are they related? Pediatr Neurol. 1996;15:127–36. doi: 10.1016/0887-8994(96)00157-9. [DOI] [PubMed] [Google Scholar]
- 32.Laptook AR, O’ Shea TM, Shankaran S, Bhaskar B the NICHD Neonatal Network. Adverse neurodevelopmental outcomes among extremely low birth weight infants with a normal head ultrasound: prevalence and antecedents. Pediatrics. 2005;115:1–8. doi: 10.1542/peds.2004-0667. [DOI] [PubMed] [Google Scholar]
- 33.Maalouf EF, Duggan PJ, Counsell SJ, Rutherford MA, Cowan F, Azzopardi D, et al. Comparison of findings on cranial ultrasound and magnetic resonance imaging in preterm infants. Pediatrics. 2001;107:719–27. doi: 10.1542/peds.107.4.719. [DOI] [PubMed] [Google Scholar]
- 34.Barnes PD, Keller K, Constantinou JC, Fleisher BE, Hintz SR, Ariagno RL. Neonatal brain magnetic resonance imaging before discharge is better than serial cranial ultrasound in predicting cerebral palsy in very low birth weight preterm infants. Pediatrics. 2004;114:992–8. doi: 10.1542/peds.2003-0772-L. [DOI] [PubMed] [Google Scholar]
- 35.Pinto J, Paneth N, Kazam E, Kairam R, Wallenstein S, Rose W, et al. Interobserver variability in neonatal cranial ultrasonography. Paediatr Perinat Epidemiol. 1988;2:43–58. doi: 10.1111/j.1365-3016.1988.tb00179.x. [DOI] [PubMed] [Google Scholar]
- 36.O’Shea TM, Volberg F, Dillard RG. Reliability of interpretation of cranial ultrasound examinations of very low-birthweight neonates. Dev Med Child Neurol. 1993;35:97–101. doi: 10.1111/j.1469-8749.1993.tb11611.x. [DOI] [PubMed] [Google Scholar]
- 37.Papile LA, Burstein J, Burstein R, Koffler H. Incidence and evolution of subependymal and intraventricular hemorrhage: a study of infants with birthweights less than 1500g. J Pediatr. 1978;92:529–534. doi: 10.1016/s0022-3476(78)80282-0. [DOI] [PubMed] [Google Scholar]
- 38.Paneth N. Classifying brain damage in preterm infants. J Pediatr. 1999;134:527–9. doi: 10.1016/s0022-3476(99)70231-3. [DOI] [PubMed] [Google Scholar]
- 39.de Vries LS, Van Haastert ILC, Rademaker KJ, Koopman C, Groenedaal F. Ultrasound abnormalities preceding cerebral palsy in high-risk preterm infants. J Pediatr. 2004;144:815–20. doi: 10.1016/j.jpeds.2004.03.034. [DOI] [PubMed] [Google Scholar]
- 40.Holling EE, Leviton A. Characteristics of cranial ultrasound white-matter echolucencies that predict disability: a review. Dev Med Child Neurol. 1999;41:136–9. doi: 10.1017/s0012162299000262. [DOI] [PubMed] [Google Scholar]
- 41.Dammann O, Leviton A. Duration of transient hyperechoic images of white matter in very-low-birthweight infants: a proposed classification. Dev Med Child Neurol. 1997;39:2–5. [PubMed] [Google Scholar]
- 42.Meadow W, Frain L, Ren Y, Lee G, Soneji S, Lantos J. Serial assessment of mortality in the neonatal intensive care unit by algorithm and intuition: certainty, uncertainty, and informed consent. Pediatrics. 2002;109:878–86. doi: 10.1542/peds.109.5.878. [DOI] [PubMed] [Google Scholar]
- 43.Morse SB, Haywood JL, Goldenberg RL, Bronstein J, Nelson KG, Carlo WA. Estimation of neonatal outcome and perinatal therapy use. Pediatrics. 2000;105:1046–50. doi: 10.1542/peds.105.5.1046. [DOI] [PubMed] [Google Scholar]