117
Early Humon Development, 23 (1990) 117-128 Elsevier Scientific Publishers Ireland Ltd. EHD 01082
Neonatal intracranial hemorrhage: I. Changing pattern in inborn low-birth-weight infants Cynthia Strand, Abbot R. Laptook, Sharon Dowling, Nancy Campbell, Robert E. Lasky, Lawrence A. Wallin, Ann M. Maravilla and Charles R. Rosenfeld Deportment of Pediatrics, University of Texas Southwestern Medical Center at Dallas, 5323 Hurry Hines Blvd., DoNus, TX 75235 (U.S.A.) (Received 3 January 1990; revision received 15 April 1990; accepted 6 June 1990)
Summary Many reports of the occurrence of periventricular-intraventricular hemorrhage (PVH-IVH) are biased by the inclusion of both inborn and outborn infants. To obviate this selection bias we examined a large inborn population of low-birthweight infants to determine if the incidence of PVH-IVH changed over a 3-year interval from March, 1982 through February, 1985. Serial cranial ultrasonography was performed in 463 consecutive infants of birth weight less than or equal to 1500 g who survived for more than 8 h. The incidence of PVH-IVH decreased from 31.5% and 29.3% in years 1 and 2, respectively, to 23.7% in year 3 (P< 0.05). The latter reflected a fall in the incidence of grades III and IV PVH-IVH, but no change in the incidence of grades I and II. This observation was not attributable to changes in mortality, the distribution of infants by birth weight and estimated gestational age in each year of the study, or infants excluded from the analysis. Contrary to most reports, 21.9% of all PVH-IVH during the 3 years were first diagnosed after 14 days postnatal age and were predominantly grade I. These results document not only a change in the epidemiology of PVH-IVH in an inborn population, but also the importance of serial cranial ultrasonography beyond the first week of life. low-birth-weight; inborn.
periventricular-intraventricular
hemorrhage;
Correspondence to: Abbot R. Laptook. 037%3782/90/$03.50 0 1990 Elsevier Scientific Publishers Ireland Ltd. Published and Printed in Ireland
sonography;
118
Introduction
There have been numerous reports that the incidence of periventricular-intraventricular hemorrhage (PVH-IVH) diagnosed by either computerized tomography (CT) or ultrasonography (US) in low-birth-weight infants ranges between 30% and 50% [l-5]. However, critical evaluation of many of these reports reveals heterogeneity concerning important issues potentially influencing the incidence of PVH-IVH. Amongst these there are four of primary importance. (1) The incidence of PVH-IVH over time in regional referral centers may be difficult to interpret since the referral population may change and a difference has been demonstrated in the occurrence of PVH-IVH for inborn compared to outborn patients [6,7]. (2) In a number of reports the incidence of PVH-IVH, using either CT [1,6] or US [5,7] was based upon the presence or absence of PVH-IVH in the first 7 postnatal days, thus assuming the absence of PVH-IVH after this age. More recent investigations using serial US during the neonatal period have demonstrated the onset of PVH-IVH after 7 days postnatal age [8,9]. (3) In several investigations, determination of the incidence of PVH-IVH has been limited to infants who survive at least 24 h [lo, 1I] or until the time of the first diagnostic examination [12]. (4) Finally, although most investigators have defined their study population as infants whose birth weight is less than 1500 g, others have used a different birth weight [4,6,12] or gestational age
L&%101. There has been a change in the outcome of low-birth-weight infants who have sustained PVH-IVH [ 121 that may be associated with the evolution of obstetric and neonatal care and more comprehensive surveillance for PVH-IVH. Batton et al. [ 131 have retrospectively investigated whether a change in the incidence of PVH-IVH had occurred between 1981 and 1984. No alteration in the overall incidence of PVH-IVH was detected although the incidence of grade III IVH decreased. This investigation, however, was from a regional tertiary care center and a selection bias of infants transported for intensive care may have affected these results. Similar preliminary observations have been reported from another regional referral center [14]. In contrast, a recent report from a regional referral center retrospectively analyzed their experience from 1980 to 1987 and demonstrated a decline in the incidence of PVHIVH [ 151. The present investigation prospectively examined a totally inborn population at Parkland Memorial Hospital in order to determine the incidence of PVHIVH in infants whose birth weight was less than or equal to 1500 g and to assess if the incidence changed over a 3-year interval from 1982 to 1985. A change in the incidence of PVH-IVH in this population has been documented. Methods
Parkland Memorial Hospital serves a large indigent urban population and is the sole county facility in Dallas County. Between March 1982 and February 1985 all infants born at Parkland Memorial Hospital whose birth weight was 6 1500 g were prospectively evaluated as part of an investigation of PVH-IVH in low-birthweight infants. Maternal and neonatal data were collected on all patients using a database
119
previously validated in this institution for the accuracy of the collected information (M4NDATE [16]). Gestational age was determined by one investigator (SD.) over the 3-year interval using a modified Dubowitz examination (pediatric estimate) or calculated from either the last menstrual period when well documented, or ultrasound if obtained at less than 14 weeks gestation (obstetric estimate). All patients underwent serial cranial US examinations which were scheduled at postnatal ages of l-3 days, 4-7 days, day 21 and at 36 weeks post-conceptual age. Technical and scheduling difficulties precluded rigid compliance. Examinations were done using a Diasonics ultrasound machine with a ~-MHZ neonatal probe. Scans were obtained in three coronal planes (at the level of the frontal horn, third ventricle and trigone) and three sagittal planes (midline and right and left parasagittal views through the germinal matrix). All US were interpreted by one pediatric radiologist (A.M.) who was blinded to the clinical course. Interpretations were done without knowledge of an individual patient’s previous or subsequent scans. PVHIVH was classified according to the grading system of Papile et al. [l]. Specifically, a grade I hemorrhage was localized to the caudothalamic junction of the subependyma1 matrix, a grade II hemorrhage extended into the ventricle without ventricular enlargement, a grade III hemorrhage extended into the ventricle with ventricular enlargement and a grade IV hemorrhage extended into the brain parencyhma. During the 3-year interval 33,672 infants were born and 575 had a birth weight 8 h, results of the 41 infants with missing data were accounted for by incorporating results from the 19 available autopsies and assuming the presence of PVH-IVH in the remaining 22 of 41 infants. Statistical significance was considered present at a level of P < 0.05. Results For the 3-year interval, infants with a birth weight 8 h
120 15
5
?
2 AZ .t: 3 E $ $ a
0 m L3 @j
10
Grade Grade Grade Grade
I II Ill IV
p 8 h between 1982 and 1985. The percent of infants with PVH-IVH is plotted by severity of the hemorrhage and the year of the study. As indicated, the incidence of grades III and IV PVH-IVH was significantly reduced during the third year.
had an incidence of PVH-IVH of 27.6% (1281463). Of the 128 infants sustaining PVH-IVH, 48.4% were grade I, 7.4% were grade II, 30.4% were grade III and 14.1% were grade IV. When analyzed by year of study, the incidence of PVH-IVH decreased from 31.5% (41/130) and 29.3% (431147) in years 1 and 2, respectively, to 23.7% (44/186) in year 3 (P < 0.05). As illustrated in Fig. 1, the reduction of occurrence of PVH-IVH during year 3 of the study reflects a significant decrease in the incidence of grades III plus IV PVH-IVH while the occurrence of grades I plus II PVH-IVH remained constant. To ensure that the above observations did not reflect a change in the study population, the distribution of estimated gestational age and birth weight for each year of the investigation is listed in Table I. There was no change in these characteristics amongst infants with a birth weight G1500 g and surviving > 8 h between 1982 and 1985. An increase in mortality for all infants < 1500 g could account for the observed incidence of PVH-IVH between 1982 and 1985. However, mortality decreased from 35.7qo and 29.7% in year 1 and 2 respectively, to 24.2% in year 3 (P < 0.05). No relationship was present between the total incidence of PVH-IVH and either estimated gestational age or birth weight for the 3-year interval. However, when the results were separated by severity of the hemorrhage (i.e., grades I and II versus grades III and IV) a relationship became apparent; there was a decreasing incidence of grades III and IV PVH-IVH with increasing gestational age (P < 0.05) and birth weight (P < 0.05), while grade I and II PVH-IVH remained constant (Fig. 2). Although cranial US was scheduled at specific postnatal ages, multiple reasons prevented strict adherence to this protocol (e.g., intravenous catheter in the scalp,
20.8’ 15.6 17.2
23.8 23.8 24.7
28-29 32-33
19.2 19.7 19.4
30-31
21.6 29.9 32.8
“Results are a % of the total number of infants in each year.
3/82-2/83 (n = 130) = 147) 3/83-2/84(n 3/84-2/85 (n = 186)
24-2-l
Gestational age (weeks)
6.2 8.2 4.3
34-35
751-1000
23.8 19.0 19.4
500-750
3.1’ 8.2 4.8
Birth weight (g)
28.5 29.9 30.1
1001-1250
44.6 42.9 45.7
1251-1500
Estimated gestational age and birth weight distribution amongst infants with birth weight < 1500 g and surviving > 8 h between 1982 and 1985.
TABLE I
122
60 r 0
I
I
II III IV
Grade Grade 6l Grade @?I Grade
0
60
0
SOO750
r
751iOO0
10011250
1251i500
Birth Weight (grams)
24-25
26-27
28-29
30-31
32-33 34-35
~36
Estimated Gestational Age (weeks) Fig. 2. The incidence of PVH-IVH by birth weight and estimated gestational age amongst 463 infants. A higher incidence of grades 111 and IV PVH-IVH was present for infants with a birth weight of 500-1000 g compared to infants with a birth weight of lOOI--1500 g (P < 0.05). Infants with an estimated gestational age of G29 weeks had a higher incidence of PVH-IVH than infants 232 weeks for grades 111 and IV only (P< 0.05).
unstable medical condition, when the sonographer was available, etc.). Given these limitations, 89% of the 413 infants studied had the first cranial US examination by 7 days postnatal age and 96% had the first US by 9 days postnatal age (Fig. 3A). Interestingly, only 66% of PVH-IVH were first diagnosed by 7 days postnatal age and 21.9% [28] of all PVH-IVH were first diagnosed after 14 days postnatal age (Fig. 3B). Since 11% of the study population had their first cranial US examination after 7 days postnatal age, it was possible that the high incidence of ‘late’ PVH-IVH (i.e., PVH-IVH diagnosed after 14 days postnatal age) reflected the postnatal day that the initial cranial US was obtained for this sub-population. Of those infants with ‘late’ PVH-IVH, all had at least one normal cranial US prior to the diagnosis of PVHIVH. In 8 of the infants with late PVH-IVH, a normal cranial US was present only in the first 7 postnatal days. In the remaining infants normal cranial ultrasound examinations were found in 71% (20/28) after 7 days postnatal age, 53% (15/28)
123
5
6
7
0
9
10
8. 20-
$ >I iI0 3 79%
t I 74%
o-
I
I
10
ll44
21
20
35
Aw(day6)
Fig. 3. (A) Postnatal age at which first cranial ultrasound study was performed amongst 463 infants 2 h) Acidemia (070)(pHa 47.20)
(7 Days (n = 85)
>14Days(n
1085 + 251” 28.7 f 2.2 83.7 21.7 72.8
1215 f 234 30.3 + 2.2 57.1 3.6 39.3
= 28)
P-value < 0.017 < 0.010 < 0.008 < 0.025 < 0.003
“Values are means + S.D.
after 10 days postnatal age and 46% (13/28) after 14 days postnatal age. Thus, the phenomenon of ‘late’ PVH-IVH is real and not a reflection of limitations of our data collection. In Table II, characteristics of infants with PVH-IVH diagnosed on or before 7 days postnatal age (n = 85) and after 14 days postnatal age (n = 28) are listed. Infants with ‘late’ PVH-IVH were of greater birth weight and maturity and had a lower incidence of ventilator support, hypoxemia and acidemia than infants sustaining PVH-IVH in the first week of life. There were no differences between the two groups in the incidence of hyaline membrane disease, pneumothorax, delivery by cesarean section, patent ductus arteriosus, hypercapnia and Apgar scores at 1 and 5 min. The distribution of the worst grade of PVH-IVH for the ‘late’ group was 82% grade 1, 10.7% grade II, and 7% grade III; grade IV did not occur. Discussion
In this investigation we have demonstrated a significant change in the incidence of PVH-IVH over a 3-year interval amongst a large inborn population of infants whose birth weight was < 1500 g and who survived > 8 h. In most previous investigations the change in incidence of PVH-IVH over time either was not reported, despite durations of 4-4.5 years [8,17], or the study interval was too short (2-12 months) to permit adequate evaluation [l-4]. Exceptions to this were the studies of Batton et al. [13], who reported an unchanged incidence of PVH-IVH over a four-year interval and Philips et al. [15], who documented a decline in the incidence of PVHIVH over 7 years. A number of important differences distinguish these investigations from our experience at Parkland Memorial Hospital. First and most importantly, the experience of Batton et al. and Philips et al. reflect that of regional referral centers where as many as 36% of deliveries occurred in outlying hospitals. Thus, the determination of which infant is transferred for tertiary care may influence the incidence of PVH-IVH in the referral center. Secondly, the investigations of Batton et al. and Philips et al. were retrospective and thus cranial sonography was performed routinely only in the first week of life and follow-up scans were per-
125
formed only ‘as indicated.’ The latter presumably means that infants with evidence of PVH-IVH during the first week or the presence of specific clinical indications after the first week resulted in further cranial sonograms. Thus, PVH-IVH occurring after the first week may have been missed in a substantial number of infants. In contrast to the conclusion of Batton et al. [13] and in agreement with that of Philips et al. [15], we have demonstrated that it is possible for the incidence of PVH-IVH to change from year to year. Moreover, our findings are not confounded by the influence of random transfer of outborn infants and the well-known hazards of retrospective surveys. Whether the incidence of PVH-IVH remains lower is yet to be determined. Even if the incidence of PVH-IVH does not remain low, this does not detract from important implications with regard to clinical investigation of PVHIVH. Specifically, surveillance for PVH-IVH limited to short intervals of time (for example, 1 year) may lead to inappropriate interpretation of incidence figures for nurseries which implement routine use of medication to prevent PVH-IVH. The latter point has been emphasized by Philips et al. [15] and Szymonowicz et al. [ 181. Szymonowicz et al. [18] documented a fall in incidence of PVH-IVH during two consecutive time intervals (1982 and 1983-84) implementing specific care guidelines to minimize cardiorespiratory instability during the second time interval. In contrast, no specific guidelines or change in approach to management of low-birthweight infants occurred in our nursery and a fall in incidence of PVH-IVH was documented. Thus, our data also underscores the potential problem introduced into investigation of PVH-IVH when historical control groups are used. A number of variables needed to be considered to ensure that a selection bias for our study population did not contribute to the above conclusion. If neonatal mortality increased over the 3-year interval, a reduced incidence of PVH-IVH may reflect deaths of infants who otherwise may have survived and developed PVH-IVH. During the interval of this investigation the number of infants with a birth weight Q 1500 g delivered at Parkland Memorial Hospital was higher (P < 0.05) in year 3 (38.1% of 575) than in years 1 (29.7%) and year 2 (32.2%), but as noted in the Results section, mortality was lower (P < 0.05) in year 3 compared to years 1 and 2. Thus, the change in infant mortality, i.e., decrease, would not account for the observed reduction in incidence of PVH-IVH. A second consideration would be a change in the distribution of birth weight and estimated gestational age of the study population over the 3-year interval. If year 3 was characterized by the survival of larger, more mature babies, a change in the incidence of PVH-IVH would be anticipated. Characterization of the study population for each year by gestational age assessment and birth weight indicates a similar distribution of infants surviving > 8 h in each year of the study. Although infants who expired at 8 h is similar to that reported for other inborn populations [17,19]. The large number of infants included in this study reduces the risk of a non-representative sample error. Investigation of a large inborn population offers the distinct advantages of a relatively consistent obstetric approach for delivery of low-birth-weight infants, uniform delivery room stabilization and somewhat greater uniformity in neonatal intensive care. Review of obstetric and neonatal practices during the interval of this investigation has not allowed determination of a readily identifiable etiology for the change in incidence of PVH-IVH. Specifically, our nursery has not adopted routine postnatal administration of phenobarbital, indomethacin, vitamin E, ethamsylate or pancuronium, all of which may influence the incidence of PVH-NH. Our general approach to ventilator management of low-birth-weight infants, fluid management, use of heparin in intravenous fluids, and judicious use of vasoactive drugs has not changed. In addition our obstetric colleagues have not changed the antepartum evaluation and treatment of high risk, preterm mothers (steroids, tocolytic agents and vitamin K are not routinely used), nor has the management during labor or delivery been altered. Attempts at timing the occurrence of PVH-IVH initially relied upon clot analysis for hemoglobin A and chromium-labelled red blood cells from post-mortem tissue and blood samples [20-221. With the advent of cranial US, more precise timing of PVH-IVH became possible and infants have been studied as often as every 8 h through the first 3 days of life [2]. Although there is variation in the exact time of occurrence of PVH-IVH, the consensus of most studies is that the majority (75100%) occur by 72 h postnatal age [2-4,7,19,23,24]. In the present investigation we made no attempt to time accurately the occurrence of PVH-IVH. However, the design of our investigation allowed confirmation’of the phenomenon of ‘late’ PVHIVH, i.e., occurrence after 14 days postnatal age. In our study 13 infants had a normal cranial US after 14 days postnatal age and subsequently were diagnosed to have PVH-IVH. The maximum age at which a normal ultrasound occurred prior to diagnosis of a ‘late’ PVH-IVH was 33 days. As discussed in the Results section, the timing of cranial US examinations precludes an exact incidence of ‘late’ PVH-IVH. While the occurrence of PVH-IVH in infants more than 7 days postnatal age has been previously documented [9,25], in only three other reports has the occurrence of PVH-IVH after 14 days postnatal age been noted [8,26,27]; an additional report mentions the occurrence of this diagnosis in a single infant [23]. The scant number of reports of this observation probably reflects limitations of previous investigations where surveillance for PVH-IVH was limited to the first 7-10 days postnatal age. Consistent with our observations, ‘late’ PVH-IVH has been characterized as usually being grade I in severity [8,26,27]. In this investigation we have demonstrated a reduction in the incidence of PVHIVH over a 3-year interval in a large inborn population of low-birth weight infants.
127
Although an identifiable etiology for the change in the epidemiology of PVH-IVH was not apparent, our observation was not attributable to changes in mortality, the distribution of infants by birth weight and gestational age in each year of the study, or infants excluded from the analysis. The timing of occurrence of PVH-IVH in this study has important implication for surveillance of the presence of PVH-IVH in neonatal nurseries. Previous recommendations for optimal timing of neonatal cranial US examinations have been an initial US between 4 and 7 days and follow-up examinations on day 14 and at 3 months of age [28]. However, a follow-up US at 3 months may miss ‘late’ grade I PVH-IVH due to prior resolution; furthermore, many infants have already been discharged from the hospital by this age. Thus, based upon the results of the present investigation, we would suggest altering the time of final US examination from 3 months to 1 month postnatal age. Acknowledgent
The authors wish to thank Ms. Marilyn Dixon for secretarial support and the staff in the Special Care Nursery for facilitating this investigation. This work was supported by funds from the Crystal Charity Ball of Dallas. References 1
2 3 4 5 6 7 8 9 10 11 12 13 14
Papile, L.A., Burstein, J., Burstein, R. et al. (1978): Incidence and evolution of subependymal intraventricular hemorrhage: A study of infants with birth weights less than 1500 g. J. Pediatr.,
and 92,
529-534. Van De Bor, M., Van Bel, F., Lineman, R. et al. (1986): Perinatal factors and periventricularintraventricular hemorrhage in preterm infants. Am. J. Dis. Child., 140, 1125-l 130. Thorburn, R.J., Lipscomb, A.P., Stewart, A.L. et al. (1982): Timing and antecedents of periventricular hemorrhage and of cerebral atrophy in very preterm infants. Early Hum. Dev., 7,221-238. Szymonowicz, W. and Vyh, Yu. (1984): Timing and evolution of periventricular haemorrhage in infants weighing 1250 g or less at birth. Arch. Dis. Child., 59, 7-12. Cooke, R.W.I. (1981): Factors associated with periventricular haemorrhage in very low-birthweight infants. Arch. Dis. Child., 56,425-431. Clark, C.E., Clyman, R.I., Roth, R.S. et al. (1981): Risk factor analysis of intraventricular hemorrhage in low-birth-weight infants. J. Pediatr., 99,625-628. De Crespigny, L.C., Mackay, R., Murton, L.J. et al. (1982): Timing of neonatal cerebroventricular haemorrhage with ultrasound. Arch. Dis. Child., 57,231-233. Enzmann, J.D., Murphy-Irwin, K., Stevenson, D. et al. (1985): The natural history of subependyma1 germinal matrix hemorrhage. Am. J. Perinatol., 2, 123-133. Perlman, J.M. and Volpe, J.J. (1986): Intraventricular hemorrhage in extremely small premature infants. Am. J. Dis. Child., 140, 1122-I 124. Dykes, F.D., Lazzara, J.A., Ahmann, P. et al. (1980): Intraventricular hemorrhage: A prospective evaluation of etiopathogenesis. Pediatrics, 66,42-49. Levene, M.I., Fawer, Cl. and Lamont, R.F. (1982): Risk factors in the development of intraventricular haemorrhage in the preterm neonate. Arch. Dis. Child., 57,410-417. Shinnar, S., Molteni, R.A., Gammon, K. et al. (1982): Intraventricular hemorrhage in the premature infant: A changing outlook. N. Engl. J. Med., 306, 1464-1468. Batton, D.G., DeWitte, D.B., Boal, D.K. et al. (1986): Incidence and severity of intraventricular hemorrhage: 1981-1984. Am. J. Perinatol., 3,353-356. Hitti, J., Gulati, R.K., Soll, R.F. et al. (1987): No trend in incidence of intraventricular hemorrhage among very low-birth-weight infants. Pediatr. Res., 21,363 (abstract).
128 15
Philip, A.G.S., Allan, W.C. and Tito, A.M. (1989): lntraventricular hemorrhage in preterm infants: Declining incfdence in the 1980s. Pediatrics, 84,797-801. 16 Sanchez, A.M., Mize, S.G., Jimenez, J.M. et al. (1975): Systems approach to the evaluation of maternal and neonatal care. Proceedings, Twelfth Hawaii International Conference of Systems, J. Sciences, 3,140-151. 17 Morales, W.J. and Koerten, J. (1986): Obstetric management and intraventricular hemorrhage in very-low-birth-weight infants. Obstet. Gynecol., 68,35-39. 18 Szymonowicz. W., Yu, V.Y.H., Walker, A. and Wilson, F. (1986): Reduction in periventricular haemorrhage in preterm infants. Arch. Dis. Child., 61,661-665. 19 Dolfin, T., Skidmore, M.B., Fong, K.W. et al. (1983): Incidence, severity and timing of subependymal and intraventricular hemorrhages in preterm infants born in a perinatal unit as detected by serial real time ultrasound. Pediatrics, 71,541-546. 20 Emerson, P., Fujimura, M., Howat, P. et al. (1977): Timing of intraventricular haemorrhage. Arch. Dis. Child., 52, 183-187. 21 Dyer, N.C., Brill, A.B., Tsiantos, A.K. et al. (1973): Timing of intracranial bleeding in newborn infants. J. Nucl. Med., 14.807-811. 22 Tsiantos, A.K., Victorin, L., Relier, J.P. et al. (1974): Intracranial hemorrhage in the prematurely born infant: Timing of clots and evaluation of clinical signs and symptoms. J. Pediatr., 85, 854859. 23 Ment, L.R., Duncan, CC., Ehrenkranz, P.A. et al. (1984): lntraventricular hemorrhage in the preterm neonate: Timing and cerebral blood flow changes. J. Pediatr., 104,419-425. 24 Beverley, D.W., Chance, G.W. and Coates, C.F. (1984): lntraventricular haemorrhage-timing of occurrence and relationship to perinatal events. Br. J. Obstet. Gynaecol., 91, 1007-1013. 25 Levene, M.I., Wigglesworth, J.S. and Dubowitz, V. (1981): Cerebral structure and intraventricular haemorrhage in the neonate: a real-time ultrasound study. Arch. Dis. Child., 56,416-424. 26 Hecht, S.T., Filly, R.A., Callen, P.N. et al. (1983): Intracranial hemorrhage: Late onset in the preterm neonate. Radiology, 149,679-699. 27 Trounce, J.Q., Rutter, N. and Levene, M.I. (1986): Periventricular leukomalacia and intraventricular haemorrhage in the preterm neonate. Arch. Dis. Child, 61, 1196-1202. 28 Partridge, J.C., Babcock, D.S., Steichen, J.L. et al. (1983): Optimal timing for diagnostic cranial ultrasound in low-birth-weight infants: Detection of intracranial hemorrhage and ventricular dilation. J. Pediatr., 102,281-287.
Early Humon Development, 23 (1990) 117-128 Elsevier Scientific Publishers Ireland Ltd. EHD 01082
Neonatal intracranial hemorrhage: I. Changing pattern in inborn low-birth-weight infants Cynthia Strand, Abbot R. Laptook, Sharon Dowling, Nancy Campbell, Robert E. Lasky, Lawrence A. Wallin, Ann M. Maravilla and Charles R. Rosenfeld Deportment of Pediatrics, University of Texas Southwestern Medical Center at Dallas, 5323 Hurry Hines Blvd., DoNus, TX 75235 (U.S.A.) (Received 3 January 1990; revision received 15 April 1990; accepted 6 June 1990)
Summary Many reports of the occurrence of periventricular-intraventricular hemorrhage (PVH-IVH) are biased by the inclusion of both inborn and outborn infants. To obviate this selection bias we examined a large inborn population of low-birthweight infants to determine if the incidence of PVH-IVH changed over a 3-year interval from March, 1982 through February, 1985. Serial cranial ultrasonography was performed in 463 consecutive infants of birth weight less than or equal to 1500 g who survived for more than 8 h. The incidence of PVH-IVH decreased from 31.5% and 29.3% in years 1 and 2, respectively, to 23.7% in year 3 (P< 0.05). The latter reflected a fall in the incidence of grades III and IV PVH-IVH, but no change in the incidence of grades I and II. This observation was not attributable to changes in mortality, the distribution of infants by birth weight and estimated gestational age in each year of the study, or infants excluded from the analysis. Contrary to most reports, 21.9% of all PVH-IVH during the 3 years were first diagnosed after 14 days postnatal age and were predominantly grade I. These results document not only a change in the epidemiology of PVH-IVH in an inborn population, but also the importance of serial cranial ultrasonography beyond the first week of life. low-birth-weight; inborn.
periventricular-intraventricular
hemorrhage;
Correspondence to: Abbot R. Laptook. 037%3782/90/$03.50 0 1990 Elsevier Scientific Publishers Ireland Ltd. Published and Printed in Ireland
sonography;
118
Introduction
There have been numerous reports that the incidence of periventricular-intraventricular hemorrhage (PVH-IVH) diagnosed by either computerized tomography (CT) or ultrasonography (US) in low-birth-weight infants ranges between 30% and 50% [l-5]. However, critical evaluation of many of these reports reveals heterogeneity concerning important issues potentially influencing the incidence of PVH-IVH. Amongst these there are four of primary importance. (1) The incidence of PVH-IVH over time in regional referral centers may be difficult to interpret since the referral population may change and a difference has been demonstrated in the occurrence of PVH-IVH for inborn compared to outborn patients [6,7]. (2) In a number of reports the incidence of PVH-IVH, using either CT [1,6] or US [5,7] was based upon the presence or absence of PVH-IVH in the first 7 postnatal days, thus assuming the absence of PVH-IVH after this age. More recent investigations using serial US during the neonatal period have demonstrated the onset of PVH-IVH after 7 days postnatal age [8,9]. (3) In several investigations, determination of the incidence of PVH-IVH has been limited to infants who survive at least 24 h [lo, 1I] or until the time of the first diagnostic examination [12]. (4) Finally, although most investigators have defined their study population as infants whose birth weight is less than 1500 g, others have used a different birth weight [4,6,12] or gestational age
L&%101. There has been a change in the outcome of low-birth-weight infants who have sustained PVH-IVH [ 121 that may be associated with the evolution of obstetric and neonatal care and more comprehensive surveillance for PVH-IVH. Batton et al. [ 131 have retrospectively investigated whether a change in the incidence of PVH-IVH had occurred between 1981 and 1984. No alteration in the overall incidence of PVH-IVH was detected although the incidence of grade III IVH decreased. This investigation, however, was from a regional tertiary care center and a selection bias of infants transported for intensive care may have affected these results. Similar preliminary observations have been reported from another regional referral center [14]. In contrast, a recent report from a regional referral center retrospectively analyzed their experience from 1980 to 1987 and demonstrated a decline in the incidence of PVHIVH [ 151. The present investigation prospectively examined a totally inborn population at Parkland Memorial Hospital in order to determine the incidence of PVHIVH in infants whose birth weight was less than or equal to 1500 g and to assess if the incidence changed over a 3-year interval from 1982 to 1985. A change in the incidence of PVH-IVH in this population has been documented. Methods
Parkland Memorial Hospital serves a large indigent urban population and is the sole county facility in Dallas County. Between March 1982 and February 1985 all infants born at Parkland Memorial Hospital whose birth weight was 6 1500 g were prospectively evaluated as part of an investigation of PVH-IVH in low-birthweight infants. Maternal and neonatal data were collected on all patients using a database
119
previously validated in this institution for the accuracy of the collected information (M4NDATE [16]). Gestational age was determined by one investigator (SD.) over the 3-year interval using a modified Dubowitz examination (pediatric estimate) or calculated from either the last menstrual period when well documented, or ultrasound if obtained at less than 14 weeks gestation (obstetric estimate). All patients underwent serial cranial US examinations which were scheduled at postnatal ages of l-3 days, 4-7 days, day 21 and at 36 weeks post-conceptual age. Technical and scheduling difficulties precluded rigid compliance. Examinations were done using a Diasonics ultrasound machine with a ~-MHZ neonatal probe. Scans were obtained in three coronal planes (at the level of the frontal horn, third ventricle and trigone) and three sagittal planes (midline and right and left parasagittal views through the germinal matrix). All US were interpreted by one pediatric radiologist (A.M.) who was blinded to the clinical course. Interpretations were done without knowledge of an individual patient’s previous or subsequent scans. PVHIVH was classified according to the grading system of Papile et al. [l]. Specifically, a grade I hemorrhage was localized to the caudothalamic junction of the subependyma1 matrix, a grade II hemorrhage extended into the ventricle without ventricular enlargement, a grade III hemorrhage extended into the ventricle with ventricular enlargement and a grade IV hemorrhage extended into the brain parencyhma. During the 3-year interval 33,672 infants were born and 575 had a birth weight 8 h, results of the 41 infants with missing data were accounted for by incorporating results from the 19 available autopsies and assuming the presence of PVH-IVH in the remaining 22 of 41 infants. Statistical significance was considered present at a level of P < 0.05. Results For the 3-year interval, infants with a birth weight 8 h
120 15
5
?
2 AZ .t: 3 E $ $ a
0 m L3 @j
10
Grade Grade Grade Grade
I II Ill IV
p 8 h between 1982 and 1985. The percent of infants with PVH-IVH is plotted by severity of the hemorrhage and the year of the study. As indicated, the incidence of grades III and IV PVH-IVH was significantly reduced during the third year.
had an incidence of PVH-IVH of 27.6% (1281463). Of the 128 infants sustaining PVH-IVH, 48.4% were grade I, 7.4% were grade II, 30.4% were grade III and 14.1% were grade IV. When analyzed by year of study, the incidence of PVH-IVH decreased from 31.5% (41/130) and 29.3% (431147) in years 1 and 2, respectively, to 23.7% (44/186) in year 3 (P < 0.05). As illustrated in Fig. 1, the reduction of occurrence of PVH-IVH during year 3 of the study reflects a significant decrease in the incidence of grades III plus IV PVH-IVH while the occurrence of grades I plus II PVH-IVH remained constant. To ensure that the above observations did not reflect a change in the study population, the distribution of estimated gestational age and birth weight for each year of the investigation is listed in Table I. There was no change in these characteristics amongst infants with a birth weight G1500 g and surviving > 8 h between 1982 and 1985. An increase in mortality for all infants < 1500 g could account for the observed incidence of PVH-IVH between 1982 and 1985. However, mortality decreased from 35.7qo and 29.7% in year 1 and 2 respectively, to 24.2% in year 3 (P < 0.05). No relationship was present between the total incidence of PVH-IVH and either estimated gestational age or birth weight for the 3-year interval. However, when the results were separated by severity of the hemorrhage (i.e., grades I and II versus grades III and IV) a relationship became apparent; there was a decreasing incidence of grades III and IV PVH-IVH with increasing gestational age (P < 0.05) and birth weight (P < 0.05), while grade I and II PVH-IVH remained constant (Fig. 2). Although cranial US was scheduled at specific postnatal ages, multiple reasons prevented strict adherence to this protocol (e.g., intravenous catheter in the scalp,
20.8’ 15.6 17.2
23.8 23.8 24.7
28-29 32-33
19.2 19.7 19.4
30-31
21.6 29.9 32.8
“Results are a % of the total number of infants in each year.
3/82-2/83 (n = 130) = 147) 3/83-2/84(n 3/84-2/85 (n = 186)
24-2-l
Gestational age (weeks)
6.2 8.2 4.3
34-35
751-1000
23.8 19.0 19.4
500-750
3.1’ 8.2 4.8
Birth weight (g)
28.5 29.9 30.1
1001-1250
44.6 42.9 45.7
1251-1500
Estimated gestational age and birth weight distribution amongst infants with birth weight < 1500 g and surviving > 8 h between 1982 and 1985.
TABLE I
122
60 r 0
I
I
II III IV
Grade Grade 6l Grade @?I Grade
0
60
0
SOO750
r
751iOO0
10011250
1251i500
Birth Weight (grams)
24-25
26-27
28-29
30-31
32-33 34-35
~36
Estimated Gestational Age (weeks) Fig. 2. The incidence of PVH-IVH by birth weight and estimated gestational age amongst 463 infants. A higher incidence of grades 111 and IV PVH-IVH was present for infants with a birth weight of 500-1000 g compared to infants with a birth weight of lOOI--1500 g (P < 0.05). Infants with an estimated gestational age of G29 weeks had a higher incidence of PVH-IVH than infants 232 weeks for grades 111 and IV only (P< 0.05).
unstable medical condition, when the sonographer was available, etc.). Given these limitations, 89% of the 413 infants studied had the first cranial US examination by 7 days postnatal age and 96% had the first US by 9 days postnatal age (Fig. 3A). Interestingly, only 66% of PVH-IVH were first diagnosed by 7 days postnatal age and 21.9% [28] of all PVH-IVH were first diagnosed after 14 days postnatal age (Fig. 3B). Since 11% of the study population had their first cranial US examination after 7 days postnatal age, it was possible that the high incidence of ‘late’ PVH-IVH (i.e., PVH-IVH diagnosed after 14 days postnatal age) reflected the postnatal day that the initial cranial US was obtained for this sub-population. Of those infants with ‘late’ PVH-IVH, all had at least one normal cranial US prior to the diagnosis of PVHIVH. In 8 of the infants with late PVH-IVH, a normal cranial US was present only in the first 7 postnatal days. In the remaining infants normal cranial ultrasound examinations were found in 71% (20/28) after 7 days postnatal age, 53% (15/28)
123
5
6
7
0
9
10
8. 20-
$ >I iI0 3 79%
t I 74%
o-
I
I
10
ll44
21
20
35
Aw(day6)
Fig. 3. (A) Postnatal age at which first cranial ultrasound study was performed amongst 463 infants 2 h) Acidemia (070)(pHa 47.20)
(7 Days (n = 85)
>14Days(n
1085 + 251” 28.7 f 2.2 83.7 21.7 72.8
1215 f 234 30.3 + 2.2 57.1 3.6 39.3
= 28)
P-value < 0.017 < 0.010 < 0.008 < 0.025 < 0.003
“Values are means + S.D.
after 10 days postnatal age and 46% (13/28) after 14 days postnatal age. Thus, the phenomenon of ‘late’ PVH-IVH is real and not a reflection of limitations of our data collection. In Table II, characteristics of infants with PVH-IVH diagnosed on or before 7 days postnatal age (n = 85) and after 14 days postnatal age (n = 28) are listed. Infants with ‘late’ PVH-IVH were of greater birth weight and maturity and had a lower incidence of ventilator support, hypoxemia and acidemia than infants sustaining PVH-IVH in the first week of life. There were no differences between the two groups in the incidence of hyaline membrane disease, pneumothorax, delivery by cesarean section, patent ductus arteriosus, hypercapnia and Apgar scores at 1 and 5 min. The distribution of the worst grade of PVH-IVH for the ‘late’ group was 82% grade 1, 10.7% grade II, and 7% grade III; grade IV did not occur. Discussion
In this investigation we have demonstrated a significant change in the incidence of PVH-IVH over a 3-year interval amongst a large inborn population of infants whose birth weight was < 1500 g and who survived > 8 h. In most previous investigations the change in incidence of PVH-IVH over time either was not reported, despite durations of 4-4.5 years [8,17], or the study interval was too short (2-12 months) to permit adequate evaluation [l-4]. Exceptions to this were the studies of Batton et al. [13], who reported an unchanged incidence of PVH-IVH over a four-year interval and Philips et al. [15], who documented a decline in the incidence of PVHIVH over 7 years. A number of important differences distinguish these investigations from our experience at Parkland Memorial Hospital. First and most importantly, the experience of Batton et al. and Philips et al. reflect that of regional referral centers where as many as 36% of deliveries occurred in outlying hospitals. Thus, the determination of which infant is transferred for tertiary care may influence the incidence of PVH-IVH in the referral center. Secondly, the investigations of Batton et al. and Philips et al. were retrospective and thus cranial sonography was performed routinely only in the first week of life and follow-up scans were per-
125
formed only ‘as indicated.’ The latter presumably means that infants with evidence of PVH-IVH during the first week or the presence of specific clinical indications after the first week resulted in further cranial sonograms. Thus, PVH-IVH occurring after the first week may have been missed in a substantial number of infants. In contrast to the conclusion of Batton et al. [13] and in agreement with that of Philips et al. [15], we have demonstrated that it is possible for the incidence of PVH-IVH to change from year to year. Moreover, our findings are not confounded by the influence of random transfer of outborn infants and the well-known hazards of retrospective surveys. Whether the incidence of PVH-IVH remains lower is yet to be determined. Even if the incidence of PVH-IVH does not remain low, this does not detract from important implications with regard to clinical investigation of PVHIVH. Specifically, surveillance for PVH-IVH limited to short intervals of time (for example, 1 year) may lead to inappropriate interpretation of incidence figures for nurseries which implement routine use of medication to prevent PVH-IVH. The latter point has been emphasized by Philips et al. [15] and Szymonowicz et al. [ 181. Szymonowicz et al. [18] documented a fall in incidence of PVH-IVH during two consecutive time intervals (1982 and 1983-84) implementing specific care guidelines to minimize cardiorespiratory instability during the second time interval. In contrast, no specific guidelines or change in approach to management of low-birthweight infants occurred in our nursery and a fall in incidence of PVH-IVH was documented. Thus, our data also underscores the potential problem introduced into investigation of PVH-IVH when historical control groups are used. A number of variables needed to be considered to ensure that a selection bias for our study population did not contribute to the above conclusion. If neonatal mortality increased over the 3-year interval, a reduced incidence of PVH-IVH may reflect deaths of infants who otherwise may have survived and developed PVH-IVH. During the interval of this investigation the number of infants with a birth weight Q 1500 g delivered at Parkland Memorial Hospital was higher (P < 0.05) in year 3 (38.1% of 575) than in years 1 (29.7%) and year 2 (32.2%), but as noted in the Results section, mortality was lower (P < 0.05) in year 3 compared to years 1 and 2. Thus, the change in infant mortality, i.e., decrease, would not account for the observed reduction in incidence of PVH-IVH. A second consideration would be a change in the distribution of birth weight and estimated gestational age of the study population over the 3-year interval. If year 3 was characterized by the survival of larger, more mature babies, a change in the incidence of PVH-IVH would be anticipated. Characterization of the study population for each year by gestational age assessment and birth weight indicates a similar distribution of infants surviving > 8 h in each year of the study. Although infants who expired at 8 h is similar to that reported for other inborn populations [17,19]. The large number of infants included in this study reduces the risk of a non-representative sample error. Investigation of a large inborn population offers the distinct advantages of a relatively consistent obstetric approach for delivery of low-birth-weight infants, uniform delivery room stabilization and somewhat greater uniformity in neonatal intensive care. Review of obstetric and neonatal practices during the interval of this investigation has not allowed determination of a readily identifiable etiology for the change in incidence of PVH-IVH. Specifically, our nursery has not adopted routine postnatal administration of phenobarbital, indomethacin, vitamin E, ethamsylate or pancuronium, all of which may influence the incidence of PVH-NH. Our general approach to ventilator management of low-birth-weight infants, fluid management, use of heparin in intravenous fluids, and judicious use of vasoactive drugs has not changed. In addition our obstetric colleagues have not changed the antepartum evaluation and treatment of high risk, preterm mothers (steroids, tocolytic agents and vitamin K are not routinely used), nor has the management during labor or delivery been altered. Attempts at timing the occurrence of PVH-IVH initially relied upon clot analysis for hemoglobin A and chromium-labelled red blood cells from post-mortem tissue and blood samples [20-221. With the advent of cranial US, more precise timing of PVH-IVH became possible and infants have been studied as often as every 8 h through the first 3 days of life [2]. Although there is variation in the exact time of occurrence of PVH-IVH, the consensus of most studies is that the majority (75100%) occur by 72 h postnatal age [2-4,7,19,23,24]. In the present investigation we made no attempt to time accurately the occurrence of PVH-IVH. However, the design of our investigation allowed confirmation’of the phenomenon of ‘late’ PVHIVH, i.e., occurrence after 14 days postnatal age. In our study 13 infants had a normal cranial US after 14 days postnatal age and subsequently were diagnosed to have PVH-IVH. The maximum age at which a normal ultrasound occurred prior to diagnosis of a ‘late’ PVH-IVH was 33 days. As discussed in the Results section, the timing of cranial US examinations precludes an exact incidence of ‘late’ PVH-IVH. While the occurrence of PVH-IVH in infants more than 7 days postnatal age has been previously documented [9,25], in only three other reports has the occurrence of PVH-IVH after 14 days postnatal age been noted [8,26,27]; an additional report mentions the occurrence of this diagnosis in a single infant [23]. The scant number of reports of this observation probably reflects limitations of previous investigations where surveillance for PVH-IVH was limited to the first 7-10 days postnatal age. Consistent with our observations, ‘late’ PVH-IVH has been characterized as usually being grade I in severity [8,26,27]. In this investigation we have demonstrated a reduction in the incidence of PVHIVH over a 3-year interval in a large inborn population of low-birth weight infants.
127
Although an identifiable etiology for the change in the epidemiology of PVH-IVH was not apparent, our observation was not attributable to changes in mortality, the distribution of infants by birth weight and gestational age in each year of the study, or infants excluded from the analysis. The timing of occurrence of PVH-IVH in this study has important implication for surveillance of the presence of PVH-IVH in neonatal nurseries. Previous recommendations for optimal timing of neonatal cranial US examinations have been an initial US between 4 and 7 days and follow-up examinations on day 14 and at 3 months of age [28]. However, a follow-up US at 3 months may miss ‘late’ grade I PVH-IVH due to prior resolution; furthermore, many infants have already been discharged from the hospital by this age. Thus, based upon the results of the present investigation, we would suggest altering the time of final US examination from 3 months to 1 month postnatal age. Acknowledgent
The authors wish to thank Ms. Marilyn Dixon for secretarial support and the staff in the Special Care Nursery for facilitating this investigation. This work was supported by funds from the Crystal Charity Ball of Dallas. References 1
2 3 4 5 6 7 8 9 10 11 12 13 14
Papile, L.A., Burstein, J., Burstein, R. et al. (1978): Incidence and evolution of subependymal intraventricular hemorrhage: A study of infants with birth weights less than 1500 g. J. Pediatr.,
and 92,
529-534. Van De Bor, M., Van Bel, F., Lineman, R. et al. (1986): Perinatal factors and periventricularintraventricular hemorrhage in preterm infants. Am. J. Dis. Child., 140, 1125-l 130. Thorburn, R.J., Lipscomb, A.P., Stewart, A.L. et al. (1982): Timing and antecedents of periventricular hemorrhage and of cerebral atrophy in very preterm infants. Early Hum. Dev., 7,221-238. Szymonowicz, W. and Vyh, Yu. (1984): Timing and evolution of periventricular haemorrhage in infants weighing 1250 g or less at birth. Arch. Dis. Child., 59, 7-12. Cooke, R.W.I. (1981): Factors associated with periventricular haemorrhage in very low-birthweight infants. Arch. Dis. Child., 56,425-431. Clark, C.E., Clyman, R.I., Roth, R.S. et al. (1981): Risk factor analysis of intraventricular hemorrhage in low-birth-weight infants. J. Pediatr., 99,625-628. De Crespigny, L.C., Mackay, R., Murton, L.J. et al. (1982): Timing of neonatal cerebroventricular haemorrhage with ultrasound. Arch. Dis. Child., 57,231-233. Enzmann, J.D., Murphy-Irwin, K., Stevenson, D. et al. (1985): The natural history of subependyma1 germinal matrix hemorrhage. Am. J. Perinatol., 2, 123-133. Perlman, J.M. and Volpe, J.J. (1986): Intraventricular hemorrhage in extremely small premature infants. Am. J. Dis. Child., 140, 1122-I 124. Dykes, F.D., Lazzara, J.A., Ahmann, P. et al. (1980): Intraventricular hemorrhage: A prospective evaluation of etiopathogenesis. Pediatrics, 66,42-49. Levene, M.I., Fawer, Cl. and Lamont, R.F. (1982): Risk factors in the development of intraventricular haemorrhage in the preterm neonate. Arch. Dis. Child., 57,410-417. Shinnar, S., Molteni, R.A., Gammon, K. et al. (1982): Intraventricular hemorrhage in the premature infant: A changing outlook. N. Engl. J. Med., 306, 1464-1468. Batton, D.G., DeWitte, D.B., Boal, D.K. et al. (1986): Incidence and severity of intraventricular hemorrhage: 1981-1984. Am. J. Perinatol., 3,353-356. Hitti, J., Gulati, R.K., Soll, R.F. et al. (1987): No trend in incidence of intraventricular hemorrhage among very low-birth-weight infants. Pediatr. Res., 21,363 (abstract).
128 15
Philip, A.G.S., Allan, W.C. and Tito, A.M. (1989): lntraventricular hemorrhage in preterm infants: Declining incfdence in the 1980s. Pediatrics, 84,797-801. 16 Sanchez, A.M., Mize, S.G., Jimenez, J.M. et al. (1975): Systems approach to the evaluation of maternal and neonatal care. Proceedings, Twelfth Hawaii International Conference of Systems, J. Sciences, 3,140-151. 17 Morales, W.J. and Koerten, J. (1986): Obstetric management and intraventricular hemorrhage in very-low-birth-weight infants. Obstet. Gynecol., 68,35-39. 18 Szymonowicz. W., Yu, V.Y.H., Walker, A. and Wilson, F. (1986): Reduction in periventricular haemorrhage in preterm infants. Arch. Dis. Child., 61,661-665. 19 Dolfin, T., Skidmore, M.B., Fong, K.W. et al. (1983): Incidence, severity and timing of subependymal and intraventricular hemorrhages in preterm infants born in a perinatal unit as detected by serial real time ultrasound. Pediatrics, 71,541-546. 20 Emerson, P., Fujimura, M., Howat, P. et al. (1977): Timing of intraventricular haemorrhage. Arch. Dis. Child., 52, 183-187. 21 Dyer, N.C., Brill, A.B., Tsiantos, A.K. et al. (1973): Timing of intracranial bleeding in newborn infants. J. Nucl. Med., 14.807-811. 22 Tsiantos, A.K., Victorin, L., Relier, J.P. et al. (1974): Intracranial hemorrhage in the prematurely born infant: Timing of clots and evaluation of clinical signs and symptoms. J. Pediatr., 85, 854859. 23 Ment, L.R., Duncan, CC., Ehrenkranz, P.A. et al. (1984): lntraventricular hemorrhage in the preterm neonate: Timing and cerebral blood flow changes. J. Pediatr., 104,419-425. 24 Beverley, D.W., Chance, G.W. and Coates, C.F. (1984): lntraventricular haemorrhage-timing of occurrence and relationship to perinatal events. Br. J. Obstet. Gynaecol., 91, 1007-1013. 25 Levene, M.I., Wigglesworth, J.S. and Dubowitz, V. (1981): Cerebral structure and intraventricular haemorrhage in the neonate: a real-time ultrasound study. Arch. Dis. Child., 56,416-424. 26 Hecht, S.T., Filly, R.A., Callen, P.N. et al. (1983): Intracranial hemorrhage: Late onset in the preterm neonate. Radiology, 149,679-699. 27 Trounce, J.Q., Rutter, N. and Levene, M.I. (1986): Periventricular leukomalacia and intraventricular haemorrhage in the preterm neonate. Arch. Dis. Child, 61, 1196-1202. 28 Partridge, J.C., Babcock, D.S., Steichen, J.L. et al. (1983): Optimal timing for diagnostic cranial ultrasound in low-birth-weight infants: Detection of intracranial hemorrhage and ventricular dilation. J. Pediatr., 102,281-287.
