INTRODUCTION

Intraventricular hemorrhage (IVH), periventricular leukomalacia (PVL), and ventriculomegaly (VM) are major neuropathological complications occurring commonly in premature infants, and are associated with high mortality and adverse neurodevelopmental outcome.^{1,2,3,4,5,6,7,8,9} The incidence of IVH has decreased over the past 10 years, with an overall occurrence rate of approximately 15%.¹ Most IVH occurs in the first 3 to 4 postnatal days, with more than 95% of hemorrhages being first detected by bedside head ultrasound (HUS) studies before the fifth day of life.^9,10 Infants with the earliest onset of hemorrhage may have extension subsequently, and may later develop posthemorrhagic hydrocephalus, VM, and/or PVL.^9,10,11 The reported incidence of PVL varies from 3 to 15%.^12,13 On HUS, PVL may initially appear as increased periventricular white matter echogenicity that subsequently evolves into leukomalacia and/or cyst formation 1 to 3 weeks later.¹³ Thus cystic PVL is usually detected on HUS after the first 2 to 3 weeks of life.^14,15 Nonobstructive VM detected by a HUS study may be a manifestation of early diffuse white matter injury not associated with IVH or initial periventricular echogenicity (PVE).¹⁶

Previous studies have demonstrated that severe IVH, VM and PVL can be clinically occult.^17,18,19 Therefore, serial screening HUS studies are necessary to identify those infants having these intracranial abnormalities and at risk for subsequent neurodevelopmental deficits. When the initial screening HUS and a follow-up study fail to detect neuropathology, the timing and frequency of subsequent HUS studies becomes more difficult to determine, and is often a matter of clinical judgment. This study was designed to test the hypothesis that premature neonates who have had two normal screening HUS studies 7 or more days apart are likely to have subsequently normal HUS studies. The objective of this retrospective analysis was to determine the positive predictive value of having a normal repeat HUS after having had two previous normal studies.

METHODS

All preterm infants 32 weeks gestational age (GA) admitted to the neonatal intensive care unit (NICU) at University Hospital and Medical Center, Stony Brook, between January 1998 and July 2000 were reviewed for this study. Computerized radiology records and case notes were extracted for data collection. Demographic data and medical information collected included: date of birth, birth weight (BW), GA, multiple births, delivery route (vaginal or cesarean section), inborn vs out-born birth and transported, Apgar scores, hospital length of stay (LOS), dates and reports of HUS studies, and the occurrences of medical complications for those infants with abnormal repeat scans. Obstetrical dating and the newborn examination were used to estimate GA.

The ultrasound screening practice in our institution for examining preterm infants of these GAs includes the performance of an initial HUS between days of life (DOL) one and four. If this initial study is normal, a repeat follow-up HUS is considered at 1 week after the initial screening study and at 1 month of age. In the more premature infants, an additional HUS is often obtained before hospital discharge, a practice similar to that supported by other authors.¹²

Standard coronal and parasagittal images were obtained with a portable, mechanical-sector, real-time unit using a 7.5 MHz transducer directed through the anterior fontanel. The HUS scans were interpreted by one of six radiologists on a rotating basis as part of our standard clinical practice. Although inter-reader agreement was not assessed directly because of the retrospective nature of this report, interpretation for IVH grading is reasonably reliable for readers differentiating normal or grade I hemorrhages from grades II to IV.^20,21 There may be a tendency for over-reading (false positive) PVL abnormalities; however, that might bias interpretations away from normal.²¹ The total number of HUS studies performed, the DOL for each HUS study, and the results reported from each study were recorded for each infant. The classification and definition of HUS abnormalities are as outlined in Table 1. IVH grading was reported using the Papile classification.¹⁸

Table 1 - Classification and Definition of HUS Abnormalities.

Full table

Since neurodevelopmental outcomes for infants with minor abnormalities (e.g. grade I IVH) are similar to those with normal HUS studies,²² minor abnormalities in the present study were grouped with normal scans. This grouping was made only after infants with minor abnormalities and those with normal studies demonstrated similar BW, GA and hospital LOS (p>0.2). Studies with major abnormalities were grouped separately and included IVH grade II, PVL, VM and PVE. Although reports of outcome for infants with grade II IVH is similar to those with normal HUS,²² this finding in the present study was placed into the major abnormalities group because of common inconsistencies in interpretation between grades II and III,²⁰ and a greater likelihood for grade II IVH to extend to a higher grade, and/or the later development of posthemorrhagic hydrocephalus.¹

A total of 520 neonates 32 weeks in GA were admitted over the 2½ years study period (Figure 1) — overall, 41 had no HUS (16 died before a scan was done), 379 had normal (or only minor abnormality) studies, and 100 (21% of all those scanned) had major HUS abnormalities detected (Table 1), comparable to previous reports.^{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15} We excluded from further analysis the 41 babies having no HUS study, the 198 having only one study, and 40 infants who had two studies, but less than 7 days apart (the 20 normal infants from this group had no repeat scans). Of the remaining 241 infants who had two or more scans, with the first two 7 days apart, 43 were abnormal on the first or second study, with six of these reverting to normal on a third scan. Of the remaining 198 infants, having had at least two normal initial HUS studies 7 days apart, 100 had no repeat scans and were also excluded. Discrimination analysis was performed then, on the 141 subjects highlighted at the bottom of Figure 1. The predictive value of having a normal third HUS study after having had two normal HUS studies 7 days apart was tested using a [2 2] discrimination table.²³

Figure 1.

Population of neonates.

Full figure and legend (32K)

All clinical data were summarized as meanSD unless otherwise specified. Group comparisons were made using an independent Student's t-test, ², and the Fisher's exact test where appropriate. A probability value of <0.05 was considered significant to detect group differences.

RESULTS

Of the 198 infants who had two normal HUS studies 7 days apart, 100 had no subsequent studies, and 98 had additional follow-up evaluations. The 100 infants who had no subsequent HUS studies were more mature, had a larger BW, and had shorter LOS when compared to the 98 who had subsequent scans (Table 2). The DOL for obtaining the second normal HUS was also significantly later in the group with no subsequent studies.

Table 2 - Infants with Two Normal (or Mildly Abnormal) HUS Studies 7 Days Apart.

Full table

The 98 neonates who had subsequent HUS studies had a mean gestation of 272 weeks, BW 967313 g, and LOS 7830 days (Table 2). The majority of this group was inborn, and 64% were delivered by cesarean section. The mean DOL for obtaining the first normal scan was 21 and 137 days for the second normal scan. Each infant in this group had a total of 42 HUS studies performed.

Of the 98 infants with repeat scans, 92 (94%) had normal studies, and six became abnormal (Table 3). The six infants who had a repeat abnormal scan were significantly less mature (GA 252 vs 272 weeks, p<0.05), had a lower BW (727239 vs 983313 g, p<0.05), and had longer hospitalizations (LOS 10727 vs 7730 days, p<0.02) than the normal group. Of normal infants, 11 (12%) were <25 weeks gestation. Repeat abnormal HUS studies were observed in 27% of infants <25 weeks of GA, compared to 2% of infants 25 weeks of GA. The timing of the first two normal HUS studies was similar in infants with normal vs abnormal follow-up scans. The mean DOL for repeat normal vs abnormal scans was 3318 vs 258 days, p<0.10.

Table 3 - Characteristics of Infants Who Had Subsequent Repeat HUS Studies After Having Had Two Normal HUS 7 Days Apart.

Full table

The case histories for the six infants with an abnormal repeat scan are summarized in Table 4. All had respiratory distress syndrome (RDS) and received treatment with surfactant. All received antibiotics for 7 to 10 days for suspected sepsis. Two infants had culture-positive bacterial sepsis after the normal HUS studies were obtained, one of whom also had bacterial meningitis. Three of the four infants who were <25 weeks GA (cases 1, 2, and 3) were of multiple births, two of whom had one sibling death within the first two days of life, and the third had a sibling that died in utero. Three infants required high-frequency oscillatory ventilation (HFOV) rescue, one of whom had multiple bilateral pneumothoraces requiring repeated chest tube insertions (five procedures) during the first 3 weeks of life. Three infants had hypotension requiring repeated fluid bolus volume infusions and/or vasopressor support. Two infants (cases 2 and 3) died before discharge, and the four survivors were all discharged on supplemental oxygen for bronchopulmonary dysplasia (BPD). Two infants had necrotizing enterocolitis (NEC) with bowel perforation requiring surgery after the second normal HUS.

Table 4 - Case Histories of Infants with Repeat Abnormal HUS Studies.

Full table

The initial abnormalities detected in these six infants with repeat abnormal scans included three infants with IVH grade II (cases 1, 3, and 4), and two infants with moderate- to-severe VM who both later developed IVH grade III and PVL (cases 2 and 5, Table 4). One additional infant (case 6) whose abnormal scan showed "small PVE" seen only in the coronal view in a subsequent scan later had repeat normal scans prior to discharge.

The discrimination analysis demonstrated that two normal HUS studies 7 days apart predicted subsequently normal studies with a sensitivity of 94%, a specificity of 86%, and a positive predictive value of 94% (Table 5, p<0.0001).

Table 5 - Discrimination Analysis for Two Normal HUS Studies 7 Days Apart Predicting a Normal Third HUS Subsequently (See Last Line of Figure 1).

Full table

COMMENTS

This report demonstrates that the vast majority (94%) of premature infants who have had two normal or slightly abnormal (less than grade II IVH) HUSs 7 days apart continue to have normal repeat studies. Our results are similar to those reported by Perlman and Rollins¹² in which 91% of infants <1500 g at birth who had two normal or slightly abnormal HUS studies at 3 to 5 days of life and again at 10 to 14 days, had subsequently normal studies. Moreover, 95% of infants in their study with normal or slightly abnormal studies at 1 week, and a third normal study at 4 weeks, had normal repeat studies, a proportion similar to our results. In an earlier report by Perlman et al.,²⁴ 99% of infants who had a normal or mildly abnormal HUS study at 10 to 14 days of life continued to have normal studies subsequently. However, in that study, infants who had IVH grade II or transient PVE that did not progress to cystic PVL were not included in the abnormal group, thus increasing the percentage of normal follow-up studies.

In our study, those few infants whose HUS results became abnormal after two earlier normal HUS studies 7 days apart were the most unstable and vulnerable extremely low-BW infants, who often required surgery, and who experienced a high mortality. Additionally, all our infants with repeat abnormal HUS studies had significant morbidity. All had BPD, a risk factor for the development of VM,²⁵ and five of the six infants also had one or more other complications that have also been associated with HUS abnormalities. These associations included pneumothorax, NEC, sepsis, surgical intervention for any reason, multiple gestation births, intrauterine fetal death in a sibling, acidosis, hypotension, PDA, and prolonged rupture of membranes.^8,24,26 In addition, the use of HFOV rescue in our preterm infants <30 weeks in gestation may have contributed to the appearance of HUS abnormalities, as reported by Moriette et al.²⁷

The incidence of significant intracranial abnormalities varies with GA and BW, occurring in the smallest and most immature preterm neonates.^1,12,28,29 Similar to our study, Batton et al.²⁸ found that major IVH occurred in 16% of infants 25 weeks gestation, compared to only 2.1% in those >25 weeks. Also, Perlman and Rollins¹² reported a 12.6% incidence of major intracranial abnormalities (severe IVH, VM, or PVL) in a cohort of 317 infants 1500 g BW, with 21% in those neonates <1000 g, and 7.4% in those with BWs between 1000 and 1500 g. In our study, repeat abnormal HUS examinations were also observed more commonly in infants <25 weeks gestation, suggesting a unique vulnerability of the most immature infants to develop serious intracranial abnormalities, even after having had a second normal study at about 2 to 3 weeks of age.

Owing to its prognostic usefulness, HUS has become the standard of care for detecting and following severe intracranial pathology in preterm infants.^2,3,4,25 The significance of mild and transient HUS abnormalities is less clear, however, and more subtle neurodevelopmental sequelae may only be recognized after long-term follow-up studies have been evaluated.³⁰ Indeed, extremely low-BW infants <25 weeks gestation are at the highest risk for developmental delay, even in the absence of abnormal intracranial pathology.^29,31,32,33 Therefore, long-term neurodevelopmental follow-up programs are necessary to detect and treat delays in this population regardless of HUS screening in the early neonatal period.

Our study was a retrospective review of HUS studies performed according to our clinical practice, and at the discretion of the attending staff. As a result, a large number of infants were excluded without a sufficient number of scans to determine the value of a third or later scan, a weakness of this report. Owing to clinical bias, greater numbers of HUS studies were performed in the sickest and most premature neonates. Most infants 27 weeks GA had four or more HUS studies performed, while the majority of more mature babies had only one HUS study, and a minority had none. It is reassuring that our more mature preterm infants who had only two normal HUS studies (7 days apart) were generally more than 2 weeks of age at the time of their second scan, a time when most abnormalities should have been detected.

This study may also raise concerns about our underdetection of late-onset PVL if only two HUS studies 7 days apart are recommended during the first 2 weeks of hospitalization. We potentially could miss significant pathology. A number of investigators have suggested more intensive schedules for performing HUS evaluations.^12,34,35,36 In the report by Pierrat et al.³⁴ infants with extensive PVL (grade III PVL) developed cystic changes within the first 2 to 3 weeks of life; and infants with localized PVL (grade II PVL) were noted to develop cysts only after the first month of life. However, 97% of all their infants with grade II PVL had echodensities on their first screening study, a finding that would have placed those infants into the major abnormality group in our study. They recommended obtaining a weekly HUS in all infants at risk until discharge from the hospital, a costly strategy. Townsend et al.³⁵ found the incidence of major HUS abnormalities (cystic PVL and persistent ventricular enlargement) after 28 days in 28% of high-risk infants tested. In contrast to our study, 36% of their infants with cystic PVL and 21% of those with VM were greater than 30 weeks GA. They recommended obtaining a HUS study in the second week of life and a later study after 28 days in all infants 30 weeks GA. However, few infants were screened in the first week of life, and some may have had early PVE, which would have been classified as abnormal in our study, requiring follow-up.

Andre et al.¹⁴ reported on the occasional occurrence of late-onset PVL in premature infants. These authors noted that while cystic PVL usually appeared during the first 2 weeks after birth, up to 9% of infants had normal early screens, with late-onset cysts first appearing after 1 month of life. Similar to our study, 80% of their infants developing late-onset PVL were critically ill having intercurrent events including symptomatic bacterial sepsis, NEC, and volvulus. As in our study, such critically ill babies clearly deserve repeat scans. Goetz et al.,³⁶ similar to the study of Andre et al.¹⁴ and to our study, noted that infants who had initial normal HUS and developed late cystic PVL were of shortest GA and lowest BW. Goetz et al. recommended different schedules of HUS screening based on BW. All infants 1100 g were scanned in the first week of life, and repeated in the second week. If either was abnormal, a third study was indicated. If both were normal, a repeat scan was suggested for 45 to 60 days of life, and a fourth at 100 days. Babies >1100 g were scanned in the first week, and the second week study was optional. A third scan was suggested at 60 days or at hospital discharge.

Our study is unique in classifying infants with IVH grade II, transient PVE and/or a question of these findings in the abnormal group. Many of these infants would have been classified as normal using the definitions of other authors. Despite the many recommendations for screening, our cohort of infants with two normal scans 7 days apart would likely yield identical results, since our definition for a normal HUS was more stringent.

CONCLUSION

Using our classification of normal HUS findings, our data suggest that stable premature infants 25 weeks gestation without intervening deterioration are unlikely to have an abnormal HUS study after having had two previously normal studies 7 days apart. Conversely, unstable infants including any infant with late intercurrent events or extremely premature infants <25 weeks gestation are subject to late severe IVH, VM, and PVL, and would therefore need repeat HUS studies, even when initial studies are normal. In most of the reports we have cited, a first screening HUS study was obtained at 24 to 72 hours of life. In the present report we suggest the second screening HUS may be performed as early as 7 days after the first to rule out significant pathology in the majority of infants. A third screening HUS may not be necessary in uncomplicated cases, and we raise the question of the cost-effectiveness of this practice for future investigation. Clearly, all high-risk babies require long-term neurodevelopmental follow-up regardless of HUS findings.

References

1. Volpe JJ. Neurology of the Newborn, 4th ed. Philadelphia, PA: WB Saunders Co.; 2001. p. 428–493.
2. Van der Bor M, Ouden LD, Guit GL. Value of cranial ultrasound and magnetic resonance imaging in predicting neurodevelopmental outcome in preterm infants. Pediatrics 1992;90:196–199. | PubMed | ChemPort |
3. Whitaker AG, Feldman JF, Rossem RV, et al. Neonatal cranial ultrasound abnormalities in low birth weight infants: Relation to cognitive outcomes at six years of age. Pediatrics 1996;98:719–729. | PubMed | ChemPort |
4. Stewart AL, Renolds EOR, Hope PL, et al. Probability of neurodevelopmental disorders estimated from ultrasound appearance of brains of very preterm infants. Dev Med Child Neurol 1987;29:3–11. | PubMed |
5. Vohr B, Garcia-Coll C, Flanagan P, Oh W. Effects of intraventricular hemorrhage and socioeconomic status on perceptual, cognitive, and neurologic status of low birth weight infants at 5 years of age. J Pediatr 1992;121:280–285.
6. Pidcock FS, Graziani LJ, Stanley C, Mitchell DG, Merton D. Neurosonographic features of periventricular echodensities associated with cerebral palsy in preterm infants. J Pediatr 1990;116:417–422. | PubMed | ChemPort |
7. Paneth N, Rudelli R, Monte W, et al. White matter necrosis in very low birth weight infants: Neuropathologic and ultrasonographic findings in infants surviving six days or longer. J Pediatr 1990;116:975–984.
8. Leviton A, Paneth N. White matter damages in preterm newborns — An epidemiological perspective. Early Hum Dev 1990;24:1–22. | PubMed |
9. Ment LR, Schneider KC, Ainley MA, Allan WC. Adaptive mechanisms of developing brain. The neuroradiologic assessment of the preterm infant. Clin Perinatol 2000;27:303–323.
10. Vohr B, Ment LR. Intraventricular hemorrhage in the preterm infant. Early Hum Dev 1996;44:1–16.
11. Kuban K, Sanocka U, Leviton A et al. White matter disorders of prematurity: Association with intraventricular hemorrhage and ventriculomegaly. J Pediatr 1999;134:539–546. | PubMed |
12. Perlman JM, Rollins N. Surveillance protocols for the detection of intracranial abnormalities in premature neonates. Arch Pediatr Adolesc Med 2000;154:822–826. | PubMed | ISI | ChemPort |
13. Volpe JJ. Brain injury in the premature infant — from pathogenesis to prevention. Brain Dev 1997;19:519–534.
14. Andre P, Thebaud B, Delavaucoupet J et al. Late-onset cystic periventricular leukomalacia in premature infants: A threat until term. Am J Perinatol 2001;18:79–86.
15. Hayakawa F, Okumura A, Kato T, Kuno K, Watanabe K. Determination of timing of brain injury in preterm infants with periventricular leukomalacia with serial neonatal electroencephalography. Pediatrics 1999;104:1077–1081.
16. Leviton A, Gilles F. Ventriculomegaly, delayed myelination, white matter hypoplasia, and "periventricular" leukomalacia: How are they related? Pediatr Neurol 1996;15:127–136. | Article | PubMed |
17. Kirks DR, Bowie JD. Cranial ultrasonography of neonatal periventricular/intraventricular hemorrhage: Who, how, why and when? Pediatr Radiol 1986;16:114–119.
18. Papile LA, Burstein J, Burstein R, Koffler H. Incidence and evolution of subependymal and intraventricular hemorrhage: A study of infants with birth weight less than 1500 grams. J Pediatr 1978;92:529–534. | Article | PubMed | ISI | ChemPort |
19. Paul DA, Pearlman SA, Finkelstein MS, Stefano JL. Cranial sonography in very-low-birth-weight infants: Do all infants need to be screened? Clin Pediatr 1999;38:503–509.
20. Babcock DS, Han BK. The accuracy of high resolution, real-time ultrasonography of the head in infancy. Radiology 1981;139:665–676.
21. Baarsma R, Laurini RN, Baerts W, Okken A. Reliability of sonography in non-hemorrhagic periventricular leucomalacia. Pediatr Radiol 1987;17:189–191.
22. Papile L, Munsick-Bruno G, Schaefer A. Relationship of cerebral intraventricular hemorrhage and early childhood neurologic handicaps. J Pediatr 1983;103:273–277. | PubMed | ChemPort |
23. Feinstein XXXI AR. On the sensitivity, specificity, and discrimination of diagnostic tests. Clin Pharmacol Ther 1975;17:104–116. | PubMed | ISI | ChemPort |
24. Perlman JM, Risser R, Broyles RS. Bilateral cystic periventricular leukomalacia in the premature infant: Associated risk factors. Pediatrics 1996;97:822–827. | PubMed | ISI |
25. Ment LR, Vohr B, Allan W, et al. The etiology and outcome of cerebral ventriculomegaly at term in very low birth weight preterm infants. Pediatrics 1999;104:243–248. | PubMed |
26. Kuban KCK, Leviton A. Cerebral palsy. N Engl J Med 1994;330:188–195.
27. Moriette G, Paris-Llado J, Walti H, et al. Prospective randomized multicenter comparison of high-frequency oscillatory ventilation and conventional ventilation in preterm infants of less than 30 weeks with respiratory distress syndrome. Pediatrics 2001;107:363–372. | Article | PubMed | ISI | ChemPort |
28. Batton DG, Holtrop P, DeWitte D, Pryce C, Roberts C. Current gestational age-related incidence of major intraventricular hemorrhage. J Pediatr 1994;125:623–625.
29. Allen MC, Donohue PK, Dusman AE. The limit of viability — Neonatal outcome of infants born at 22 to 25 weeks' gestation. N Engl J Med 1993;329:1597–1601. | Article | PubMed | ISI | ChemPort |
30. Jongmans M, Mercuri E, De Vries L, Dubowitz L, Henserson SE. Minor neurological signs and perceptual-motor difficulties in prematurely born children. Arch Dis Child 1997;76:F9–14.
31. Vohr BR, Wright LL, Dusick AM, et al. Neurodevelopmental 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;105:1216–1226. | Article | PubMed | ISI | ChemPort |
32. Hack M, Wilson-Costello D, Friedman H, Taylor GH, Schluchter M, Fanaroff AA. Neurodevelopment and predictors of outcomes of children with birth weights of less than 1000 g, 1992–5. Arch Pediatr Adolesc Med 2000;154:7255–7731.
33. Wood NS, Marlow N, Costeloe K, Gibson AT, Wilkinson AR. Neurologic and developmental disability after extremely preterm birth. N Engl J Med 2000;343:378–384. | Article | PubMed | ISI | ChemPort |
34. Pierrat V, Duquennoy C, van Haastert IC, Guilley N, de Vries LS. Ultrasound diagnosis and neurodevelopmental outcome of localized and extensive cystic periventricular leucomalacia. Arch Dis Child Fetal Neonatal Ed 2001;84:F151–F156.
35. Townsend SF, Rumack CM, Thilo EH, Merenstein GB, Rosenberg AA. Late neurosonographic screening is important to the diagnosis of periventricular leukomalacia and ventricular enlargement in preterm infants. Pediatr Radiol 1999;29:347–352.
36. Goetz MC, Gretebeck RJ, OH KS, Shaffer D, Hermansen M. Incidence, timing, and follow-up of periventricular leukomalacia. Am J Perinatol 1995;12:325–327.