Acta Pñ diatr 91: 212± 217.

2002

Early versus late treatment of posthaemorrhagic ventricular dilatation:

results of a retrospective study from Ž ve neonatal intensive care units
in The Netherlands
LS de Vries1, KD Liem2, K van Dijk2, BJ Smit3, L Sie4, KJ Rademaker1 and AWD Gavilanes5; on behalf of
the Dutch Working Group of Neonatal Neurology
Departments of Neonatology , Wilhelmina Children’s Hospital1, UMC Utrecht; University Medical Centre Nijmegen2; Academic Medical
Centre Amsterdam (AMC)3; Academic Hospital Free University Amsterdam (AZVU)4; Academic Hospital Maastricht (AZM)5, The
Netherlands

De Vries LS, Liem KD, van Dijk K, Smit BJ, Sie L, Rademaker KJ, Gavilanes AWD; on behalf
of the Dutch Working Group of Neonatal Neurology. Early versus late treatment of posthaemor-
rhagic ventricular dilatation: results of a retrospective study from Ž ve neonatal intensive care units
in The Netherlands. Acta Pædiatr 2002; 91: 212–217. Stockholm. ISSN 0803-5253
Posthaemorrhagic ventricular dilatation (PHVD) in very preterm infants carries a poor prognosis.
As earlier studies have failed to show a beneŽ t of early intervention, it is recommended that
PHVD be Ž rst treated when head circumference is rapidly increasing and/or when symptoms of
raised intracranial pressure develop. Infants with PHVD, admitted to 5 of the 10 Dutch neonatal
intensive care units were studied retrospectively, to investigate whether there was a difference in
the time of onset of treatment of PHVD and, if so, whether this was associated with a difference in
the requirement of a ventriculo-peritoneal (VP) shunt and/or neurodevelopmental outcome. The
surviving infants with a gestational age µ34 wk, born between 1992 and 1996, diagnosed as
having a grade III haemorrhage according to Papile on cranial ultrasound and who developed
PHVD were included in the study. PHVD was deŽ ned as a ventricular index (VI) exceeding the
97th percentile according to Levene (1981), and severe PHVD as a VI crossing the p 97 ‡ 4 mm
line. Ninety-Ž ve infants met the entry criteria. Intervention was not deemed necessary in 22
infants, because of lack of progression. In 31 infants lumbar punctures (LP) were done before the
p 97 ‡ 4 mm line was crossed (early intervention). In 20/31 infants, stabilization occurred. In 9 a
subcutaneous reservoir was placed, with subsequent stabilization in 6. In 5/31 infants a VP shunt
was eventually inserted. In 42 infants treatment was started once the p 97 ‡ 4 mm line was crossed
(late intervention). In 30 infants LPs were performed and in 17 of these a VP shunt was eventually
inserted. In 11 infants a subcutaneous reservoir was immediately inserted and in 8 of these infants
a VP shunt was needed. In one infant a VP shunt was immediately inserted, without any other
form of treatment. Infants with late intervention crossed the p 97 ‡ 4 mm earlier (p 0.03) and
needed a shunt (26/42; 62%) more often than those with early intervention (5/31; 16%). Early LP
was associated with a strongly reduced risk of VP-shunting (odds ratio = 0.22, 95% conŽ dence
interval: 0.08–0.62). The number of infants who developed a moderate or severe handicap was
also higher (11/42; 26%) in the late intervention group, compared with those not requiring any
intervention (3/22; 14%) or treated early (5/31; 16%).
Conclusion: In this retrospective study, infants receiving late intervention required shunt insertion
signiŽ cantly more often than those treated early. A randomized prospective intervention study,
comparing early and late drainage, is required to further assess the role of earlier intervention.
Key words: Hydrocephalus, intraventricular haemorrhage, posthaemorrhagic ventricular dilata-
tion; preterm infants
LS de Vries, KE 04 123.1, Department of Neonatology, Wilhelmina Children’s Hospital, PO Box
85090, NL-3508AB Utrecht, The Netherlands (Fax. ‡31 30 2505 320, e-mail. l.devries@wkz.a-
zu.nl)

The prevalence of periventricular haemorrhage-intra-
ventricular haemorrhage (PVH-IVH) in very low birth-
weight (VLBW) infants has decreased considerably
during the past decade (1). This can be attributed
especially to the increased antenatal use of corticoster-
oids and to overall improvement in neonatal intensive
care (2). Until recently this decrease did not coincide
with a decrease in the prevalence of infantile hydro-
cephalus. The latest study from Fernell and Hagberg
(3), however, did show a decline in the prevalence of

Ó 2002 Taylor & Francis. ISSN 0803-525 3

ACTA PÆDIATR 91 (2002)
infantile hydrocephalus from 17 to 14 infants per 1000
born at less than 32 wk gestation. In the majority of
cases in this age group, infantile hydrocephalus could be
attributed to posthaemorrhagic ventricular dilatation
(PHVD).
In the absence of associated parenchymal injury,
PHVD is the most serious complication of an IVH and is
known to occur in about 35% of the infants with an
IVH; the larger the initial haemorrhage, the greater the
risk of developing PHVD. Among those who develop
PHVD, about 35% require some form of intervention
and a further 15% of those receiving intervention go on
to need insertion of a ventriculo-peritoneal (VP) shunt.
Early insertion of a VP shunt is not without complica-
tions, with infection and dysfunction being especially
common (4–6). The children usually remain shunt
dependent and will require several shunt revisions
during early childhood. Neurodevelopmental outcome
in infants with infantile hydrocephalus is reported as
being very poor, with about 75% of the children
developing cerebral palsy (CP) and 50% learning
disabilities and a smaller percentage affected by
epilepsy and/or cerebral visual impairment (7, 8).
Although the best way to prevent infantile hydro-
cephalus is prevention of PVH-IVH, optimal manage-
ment of PHVD is also important and this has been
attempted many times over the years using different
methods of intervention. As the mechanism of PHVD is
related to blockage of cerebrospinal  uid (CSF)
channels by small blood clots, early removal of bloody
CSF by lumbar or even ventricular taps has been
performed in small studies and a large multicentre
study, but without any beneŽ cial effect (8–10). Intra-
ventricular administration of Ž brinolytic agents has also
been used, but a recent review of 62 cases reported in
the literature was unable to show a positive effect, with
a need for shunt insertion varying from 11 to 100% (11).
In the majority of the studies mentioned above, the
infants were enrolled once the ventricular width
measurement was more than 4 mm above the 97th
percentile for age (12).
The aim of the present retrospective multicentre
study was to investigate whether there was a difference
in the timing and mode of treatment of PHVD in The
Netherlands and, if so, whether this was associated with
a difference in the percentage of children requiring
insertion of a VP shunt and whether this had an effect on
neurodevelopmental outcome.
Patients and methods
The study, initiated by the Dutch Working Group of
Neonatal Neurology, included neonatologists and pae-
diatric neurologists with a special interest in neonatal
neurology and working in any of the 10 Dutch regional
intensive care units. In 5 of the 10 units, the data of
infants born between January 1992 and December 1996
Late treatment of posthaemorrhagi c ventricula r dilatation 213
were studied. Only those preterm infants with a
gestational age of 34 wk or below who had a large
IVH, Ž lling the lateral ventricle by > 50% causing acute
ventricular dilatation, without any evidence of paren-
chymal involvement (grade III, according to Papile et
al. (13)) were eligible for this study. Infants with
evidence of a unilateral parenchymal haemorrhage or
cystic periventricular leukomalacia (PVL), diagnosed
using cranial ultrasound, were excluded.
Cranial ultrasound was performed by the paediatric
radiologist and/or the attending neonatologist . Exam-
inations were carried out regularly to diagnose the
maximum extent of the haemorrhage, to establish the
development of the PHVD and to follow the course of
the PHVD and the effect of any intervention. The
ventricular width was measured in the coronal view at
the level of the foramen of Monro according to Levene
(12). This measurement was available in all centres and,
depending on the policy of the unit, was prospectively
or retrospectively plotted in the graph. Special attention
was paid to crossing the p 97 ‡ 4 mm line. This line
represents ‡2SD above the p 97 for the ventricular
width according to Levene (12). Furthermore, it was
recorded whether intervention was initiated before or
after crossing the p 97 ‡ 4 mm line. It was also recorded
whether the p 97 ‡ 4 mm line was still crossed in
infants whose intervention was started before the p
97 ‡ 4 mm line had been exceeded.
The mode of intervention was recorded and varied
from lumbar punctures, ventricular taps (uncommon),
placement of a subcutaneous reservoir and insertion of a
VP shunt. In most centres a volume of 10 ml/kg was
allowed to drain during a lumbar or ventricular tap, but
the volume aimed for was not always achieved. The
need for a repeat LP was based on measurement of the
VI and/or the shape of the ventricles. Owing to the
retrospective nature of the study, the exact amount of
 uid drained in each intervention, as well as the
pressure measured, was not always available.
Outcome measures
The primary endpoints included (i) shunt placement
before discharge home and before the corrected age of
1 y and (ii) neurodevelopmental outcome at 2 y of age.
Owing to the limitations of a retrospective study, only
the presence or absence of a moderate disability
(including spastic diplegia, hemiplegia, moderate learn-
ing disabilities (DQ 50–69)) or severe disability
(including quadriplegia, blindness, deafness, uncon-
trolled epilepsy, severe learning disability (DQ <50
and multiple disabilities) could be established.
Statistical analysis
Statistical analysis was done using w2, Mann-Whitney
and Kruskal Wallis tests, using SPSS for Windows 7.5.
A p-value below 0.05 was considered signiŽ cant.
Logistic regression analysis was used to relate timing
214 LS de Vries et al. ACTA PÆDIATR 91 (2002)

of LPs (early, late) to requirement of insertion of a VP

shunt (dependent variable). In order to assess whether
other variables (GA, BW, centre) were confounders in
this relation these variables were subsequently entered
in the model.

Results
Ninety-Ž ve infants with grade III IVH and a VI >p 97
were eligible for the study. Thirty-Ž ve of the infants had
a gestational age (GA) of 27 wk or below and 60 a GA
of 28–34 wk.
Echolucencies in the brain parenchyma were not
detected by cranial ultrasonography in any of the Fig. 1. Flow chart for the 31 children who had “early intervention ”.
infants.

Treatment and 8 of these infants subsequently required shunt

Three groups could be identiŽ ed. The Ž rst group
comprised 22 infants who exceeded the p 97 but did
not require intervention, as there was no progression in
the ventricular dilatation. The mean GA was 28.2 wk
(range 26–32 wk). The second group comprised 31
infants in whom LPs were started before the p
97 ‡ 4 mm line had been crossed. The mean GA was
28.3 wk (range 25–33). The third group comprised of 42
infants. Treatment was started once the p 97 ‡ 4 mm
line had been crossed. The mean GA was 28.4 wk (25–
34 wk). There was no signiŽ cant difference between the
three groups for either GA or birthweight. None of the
22 infants in group 1 crossed the p 97 ‡ 4 mm line.
insertion. In one infant a VP shunt was inserted but was
not preceded by any other form of treatment (Fig. 2).
Infants who were Ž rst treated when the p 97 ‡ 4 mm
line had been crossed needed shunt insertion (26/42;
62%) signiŽ cantly more often than those who were
treated before crossing the ‡4 mm line (5/31; 16%).
Early LP was associated with a strongly reduced risk
of VP-shunting (odds ratio = 0.22, 95% conŽ dence
interval: 0.08–0.62). This association was not explained
by other variables. Entering GA, BW (birthweight) and
a variable indicating centre simultaneously into the
model did not change the estimates (OR = 0.22, 95%
CI: 0.07–0.67).

Early intervention group

In 31 infants LPs were started before the p 97 ‡ 4 mm Comparison of early and late intervention group
line had been crossed. In 13 out of 31 infants (42%) the As mentioned above, there was no statistically sig-
p 97 ‡ 4 mm line was crossed after initiation of niŽ cant difference in GA or birthweight between the
intervention and 3 of the infants required shunt insertion two groups. The median day when the p 97 mm line was
compared to 2 out of 18 infants whose VI remained crossed was day 7 for both the early intervention group
below the p 97 ‡ 4 mm. and the late intervention group (NS). The median day
In 20/31 infants who had LPs, stabilization subse- when the p 97 ‡ 4 mm was crossed was day 14 in the
quently occurred; in 9 infants a subcutaneous reservoir early intervention group compared with day 10 in the
was placed because the PHVD persisted or was rapidly late intervention group (p = 0.03).
progressive, requiring ongoing withdrawal of CSF. In 6
of these 9 infants, stabilization subsequently occurred,
while the other 3 did not stabilize and eventually
required shunt insertion. In another two infants a VP
shunt was placed following a period of apparent
stabilization and discharge home, but with subsequent
recurrence of progressive ventricular dilatation requir-
ing late shunt insertion (Fig. 1).

Late intervention group

In 42 infants, treatment was started once the p
97 ‡ 4 mm line had been crossed. LPs were done in
30 infants and 17 of these subsequently required shunt
insertion. In 11 infants a subcutaneous reservoir was
inserted as soon as the ‡4 mm line had been crossed Fig. 2. Flow chart for the 42 children who had “late intervention ”.
ACTA PÆDIATR 91 (2002)
Table 1. Neurodevelopmenta l outcome at 24 mo of the 95 infants with posthaemorrhagi c ventricula r dilatation (PHVD).
No intervention (n = 22)
Normal 14
Mild disability 5 9
Moderate 2 4
Severe 1 1
Neurodevelopmental outcome
All the children were seen for follow-up and were at
least 24 months of age when last seen (Table 1).
There was no signiŽ cant difference in outcome
between those treated “early” and those treated “late”
Drain revisions and infection
Seven of the 29 (24%) infants with a reservoir
developed an infection. This was due to a coagulase
negative staphylococci in all but 2 of the infants.
Thirteen of the 31 (42%) infants with a VP shunt
required at least one shunt revision during the Ž rst year
and in 5 (16%) cases this was associated with a drain
infection.
Discussion
Despite the retrospective nature of this study and the
lack of randomization among the centres, the data show
that the need to insert a VP shunt was signiŽ cantly
higher among infants in whom treatment was initiated
once the p 97 ‡ 4 mm line had been crossed. The
percentage of infants, requiring a shunt in the late
intervention group (62%) was similar to that found in
the ventriculomegaly trial, where the p 97 ‡ 4 mm line
was used as the starting-point for randomization of
lumbar taps versus conservative treatment (8). It was
also noted that those treated late were more likely to
develop moderate or severe disabilities, but this did not
reach statistical signiŽ cance.
Our data came from a very large group of infants with
the same grade of haemorrhage. Neonatal magnetic
resonance imaging was not available in the majority of
the infants, but regular cranial ultrasound scans were
performed and children who had developed cysts in the
brain parenchyma were excluded. Thus, an attempt was
made to select a group of infants with a similar type of
IVH and to study the effect of treatment of PHVD on the
need of a VP shunt and on subsequent neurodevelop-
mental outcome.
Treatment of PHVD varied considerably among the
Ž ve participating neonatal intensive care units. In one of
the units the routine was always to wait for the p
97 ‡ 4 mm line to be crossed before immediately
inserting a subcutaneous reservoir. In another unit,
treatment was almost always initiated before the
‡4 mm line was exceeded and the decision to intervene
was made on the basis of progression of the PHVD and
Late treatment of posthaemorrhagi c ventricula r dilatation 215
Early intervention (n = 31) Late intervention (n = 42)
17 23
8
8
3
especially on the changing shape of the ventricle, i.e. a
more “ballooning” shape in the midcoronal view.
Treatment was more variable in the other three units.
Pressure measurements were recorded but as these data
were not available for all taps and not always
considered to be reliable, they were not included in
the analysis. Although the infants were not randomized
for early or late intervention, a comparison was made
between those receiving early versus those receiving
late treatment (12). One could of course argue and say
that those infants who were treated early would not have
exceeded the p 97 ‡ 4 mm line if left alone. As the
infants were not randomized, we cannot be certain
about this, but in fact almost half (42%) of those treated
early still crossed the p 97 ‡ 4 mm line during the
course of therapy, despite early intervention. It was
interesting to see that those who were treated early were
observed to cross the p 97 ‡ 4 mm signiŽ cantly later
than those treated later. This suggests that LPs were able
to delay the rate of progression of increase in ventricular
size.
While there is still insufŽ cient data about the role of
early versus late intervention on subsequent neuro-
developmental outcome, it is well known that paren-
chymal injury, either haemorrhagic parenchymal
infarction or cystic PVL, is the main determinant of
neurodevelopmental outcome (14). Previous multicen-
tre studies enrolled infants with large haemorrhages,
grade III and IV according to Papile et al. (13). In these
multicentre studies it was difŽ cult to judge the effect of
PHVD per se on subsequent neurodevelopmental out-
come, as all infants with large haemorrhages were
included and the data were not analysed separately for
those with and without parenchymal injury. Of the
infants without evidence of parenchymal injury, 46%
enrolled in the ventriculomegaly trial (8) developed a
moderate or severe disability. As the ventriculomegaly
study was performed in the early 1980s, it is likely that
cystic PVL remained undiagnosed in a number of cases
and, in fact, in 15% of the infants parenchymal lesions
were Ž rst detected after trial entry (8). In our study, only
22% of children with a grade III haemorrhage devel-
oped a moderate or severe disability. It cannot be
excluded that withdrawal of intensive care treatment
was more common in The Netherlands than in other
countries involved in the ventriculomegaly trial. As the
number of infants who died before trial entry is not
reported in the multicentre study, this point cannot be
resolved. In the more recent international controlled
216 LS de Vries et al.
trial for acetazolamide and furosemide, almost half of
the infants had parenchymal lesions before trial entry,
but no separate outcome data are given for the children
without parenchymal injury (15).
The best timing for intervention of PHVD is still a
matter of debate (16). In the recent multicentre
randomized acetazolamide and furosemide trial (15),
it was deemed advisable to delay the removal of CSF
until either head growth exceeded twice the normal rate
for 2 weeks or the infant showed clinical symptoms or
signs of raised intracranial pressure. The advice stems
from the lack of positive effect of early taps, as shown in
the ventriculomegaly trial (8). In both the ventriculo-
megaly trial and the acetazolamide trial, the p
97 ‡ 4 mm line has been taken as the point of entry
for the study. One cannot, however, exclude that
intervention based on the value of the VI and started
after the p 97 ‡ 4 mm has been crossed is too late, i.e.
that one has already reached the “point of no return”.
Little data is available on earlier intervention, based
on the rate of progression of the ventricular width and
the change in shape of the lateral ventricle. The fact that
the change in shape, “the depth” of the lateral ventricle,
as measured in the midcoronal view, changes before the
VI increases was already pointed out by Sauerbrei et al.
in 1981 (17). They identiŽ ed a normal depth as 1–3 mm
and a depth ¶5mm as being increased. Recently, Davies
et al. reported similar measurements (18). Even when
treatment is started early, lumbar punctures are often
ineffective, as the amount of CSF that can be drained
per tap does not often reach the desired 10 ml/kg.
Ventricular taps are an alternative, but needle tracks do
occur and can lead to porencephalic cysts when multiple
tracks coalesce. Several recent publications have
advocated the use of an external drainage system or a
subcutaneous access device (19–21). The external drain
has a poor reputation with regard to infection, but this is
no longer the case with careful tunnelling and im-
plementation under strict sterile conditions provided
that the system is changed every 2–3 wk (19).
Berger et al. (19) studied 37 infants, 11 of whom did
not require permanent shunts. Two infants developed
ventriculitis (5.4%). The mean duration of drainage was
23 days.
A subcutaneous reservoir was used in a number of
our infants, usually following the performance of
several lumbar taps. The risk of infection was consider-
able (24%), with coagulase-negative staphylococci
being the organism in all but two cases. The risk of
infection in our own unit (WKZ) has been reduced to
12%, including the years 1996–1999, and was 15% in
one of the other participating units (6). The largest study
using subcutaneous reservoirs showed good results as
far as wound problems, blockage and infection were
concerned, but a high percentage of infants (88%) in
this study eventually required a shunt (21). Direct
placement of a reservoir without performing LPs was
recently advocated to give more effective intervention,
ACTA PÆDIATR 91 (2002)
especially in those without apparent parenchymal injury
(22).
Adverse effects of PHVD, without the symptoms
mentioned above, have been shown using various
techniques in clinical studies. A delay in latency of
somatosensory and visual evoked potentials was ob-
served during progressive PHVD and was found to
normalize following shunt insertion (23). A change in
the peak systolic velocity and even an absent or
reversed diastolic velocity have been shown during
progressive PHVD (24) as was cerebral blood  ow
measured using PET following a LP (25). An improve-
ment in oxidation of cytochrome aa3 was shown using
near infrared spectroscopy (NIRS), re ecting improve-
ment in cellular oxygenation of brain tissue (26), and,
Ž nally, raised CSF hypoxanthine values have also been
reported (27). In animal studies, a decrease in white
mater perfusion was especially noted in long-standing
hydrocephalus (28, 29). A recent study, using magnetic
resonance spectroscopy, showed progressive tissue
injury in rats that were only treated at a late stage of
their infantile hydrocephalus (30).
In view of the data obtained in this retrospective
study, a prospective randomized multicentre study is
presently being designed to compare early (<p
97 ‡ 4 mm) and late (¶p 97 ‡ 4 mm) intervention of
PHVD. Once progression in ventricular size is noted,
only a few lumbar taps will be allowed in both groups,
before inserting a subcutaneous reservoir, to enable the
caretakers to drain the amount of CSF that was aimed
for. We hope that this study will resolve the issue of the
best time for intervention with the ultimate aim of
reducing the number of shunts and the number of infants
surviving with disabilities.
Acknowledgements.—We thank M. Jonkers and P. Rosias for their help in
extracting the data from the notes and also the paediatric neurosurgeon s in
the Ž ve centres for performing the surgical procedures . We also thank C.
Uiterwaal for advice regarding the statistics and R. Gooskens, F. van Bel
and J. S. H. Vles for their fruitful discussions.
References
1. Philip AGS, Allan WC, Tito AM, Wheeler LR. Intraventricula r
haemorrhage in preterm infants: declining incidenc e in the
1980s. Pediatrics 1989; 84: 797–801
2. Shankaran S, Bauer CR, Bain R, Wright LL, Zachary J. Prenatal
and perinatal risk and protectiv e factors for neonatal intracrania l
haemorrhage . National Institute of Child Health and Human
Development . Arch Paediatr Adolesc Med 1996; 150: 491–7
3. Fernell E, Hagberg G. Infantile hydrocephalus : declining
prevalenc e in preterm infants. Acta Paediatr 1998; 87: 392–6
4. Dykes FD, Dunbar B, Lazarra A, Ahmann PA. Posthaemorrhagi c
hydrocephalu s in high risk infants: natural history, management
and long term outcome. J Pediatr 1989; 114: 611–8
5. Lin J-P, Goh W, Brown J, Steers AJ. Neurologica l outcome
following neonatal post-haemorrhagi c hydrocephalus : the effect
of maximum raised intracrania l pressure and ventriculoperito -
neal shunting. Child Nerv Syst 1992; 8: 190–7
6. Bruinsma N, Stobberingh EE, Herpers MJ, Vles JS, Weber BJ,
Gavilanes DA. Subcutaneous ventricula r catheter reservoir and
ACTA PÆDIATR 91 (2002)
ventriculoperitonea l drain-relate d infection s in preterm infants
and young children. Clin Microbiol Infect 2000; 6: 202–6
7. Fernell E, Hagberg G, Hagberg B. Infantile hydrocephalus : an
indicator of enhance d survival. Arch Dis Child 1994; 70: F123–8
8. Ventriculomegal y Trial Group 1990. Randomised trial of early
tapping in neonatal posthaemorrhagi c ventricular dilatation.
Arch Dis Child 1990; 65: 3–10
9. Anwar M, Kadam S, Hiatt IM, Hegyi T. Serial lumbar punctures
in preventio n of posthaemorrhagi c hydrocephalu s in preterm
infants. J Pediatr 1985; 107: 446–9
10. Mantovani JF, Pasternak JF, Mathew OP, Allan WC, Mills MT,
Casper J, et al. Failure of daily lumbar puncture s to prevent
neonatal post haemorrhagi c hydrocephalus . Child Nerv Syst
1997; 13: 73–6
11. Haines SJ, Lapointe M. Fibrinolytic agents in the management of
posthaemorrhagi c hydrocephalu s in preterm infants: the evi-
dence. Childs Nerv Syst 1999; 15: 226–34
12. Levene MI. Measurement of the growth of the lateral ventricle in
preterm infants with real-time ultrasound . Arch Dis Child 1981;
56: 905–10
13. Papile L, Burstein J, Burstein R, Kof er H. Incidenc e and
evaluatio n of subependyma l haemorrhage : a study of children
with a birthweight less than 1500 g. J Pediatr 1978; 92: 529–34
14. Cooke RWI. Determinants of major handicap in post-haemor -
rhagic hydrocephalus . Arch Dis Child 1987; 62: 504–17
15. Internationa l PHVD drug trial group. Internationa l randomised
controlle d trial of acetazolamid e and furosemide in posthaemor -
rhagic ventricular dilatation in infancy. Lancet 1998; 352: 433–
40
16. Du Plessis AJ. Posthemorrhagi c hydrocephalu s and brain injury
in the preterm infant: dilemmas in diagnosi s and management.
Semin Pediatr Neurol 1998; 5: 161–79
17. Sauerbrei EE, Digney M, Harrison PB, Cooperberg PL. Ultra-
sonic evaluatio n of neonatal intracrania l hemorrhag e and its
complications . Radiology 1981; 139: 677–85
18. Davies MW, Swaminathan M, Chuang SL, Betheras FR.
Reference ranges for the linear dimensions of the intracrania l
ventricle s in preterm neonates. Arch Dis Child FetalNeonatal Ed
2000; 82: F218–23
19. Berger A, Weninger M, Reinprecht A, Haschke N, Kohlhauser C,
Pollak A. Long-term experience with subcutaneousl y tunneled
external ventricula r drainage in preterm infants. Child’s Nerv
Syst 2000; 16: 103–9
Late treatment of posthaemorrhagi c ventricula r dilatation 217
20. Gaskill SJ, Marlin AE, Rivera S. The subcutaneou s ventricula r
reservoir : an effective treatment for posthaemorrhagi c hydro-
cephalus. Child’s Nerv Syst 1988; 4: 291–5
21. Hudgins RJ, Boydston WR, Gilreath CL. Treatment of post-
hemorrhagic hydrocephalu s in the preterm infant with a
ventricula r access device. Pediatr Neurosurg 1998; 29: 309–13
22. Shankaran S, Baker JD, Becker C, Canady A, Delaney-Black V.
Aggressive management of posthaemorrhagi c ventriculomegal y
may limit morbidity in VLBW infants with grade III ICH. Ped
Res 2000; 47: 323A
23. de Vries LS, Pierrat V, Minami T, Smet M, Casaer P. The role of
somatosensor y evoked response s in infants with rapidly pro-
gressive ventricular dilatation . Neuropediatric s 1990; 21: 136–9
24. Kempley ST, Gamsu HR. Changes in cerebral artery blood  ow
velocity after intermitten t cerebrospina l  uid drainage . Arch Dis
Child 1993; 69: 74–6
25. Perlman J, Herscovitc h P, Tarby T, Kreusser KL, Raichle ME,
Volpe JJ. The effect of lumbar puncture on regional cerebral
blood  ow as determined by positron emission tomography in
neonatal posthemorrhagi c hydrocephalus . Ann Neurol 1985; 18:
379–80
26. Du Plessis A, Miles K, Tsuji HN, Volpe JJ. Near infrared
spectroscop y (NIRS) shows pronounce d effects of CSF removal
on cerebral haemodynamic s in infantile hydrocephalus . Pediatr
Res 1995; 37: 377A
27. Bejar R, Saugstad OD, James H, Gluck L Increase d hypox-
anthine levels in cerebrospina l  uid of infants with hydrocepha -
lus. J Pediatr 1983; 103: 44–8
28. White RP, Kef er CW, Bada HS. Eicosanoid levels in CSF of
premature infants with posthemorrhagi c hydrocephalus . Am J
Med Sci 1990; 299: 230
29. Weller RO, Schulman K. Infantile hydrocephalus : clinical,
histologica l and ultrastructura l study of brain damage. J
Neurosurg 1972; 36: 255–65
30. Jones HC, Harris NG, Rocca JR, Andersohn RW. Progressive
tissue injury in infantile hydrocephalu s and prevention /reversal
with shunt treatment. Neurol Res 2000; 22: 89–96
Received March 26, 2001; revision received Oct. 12, 2001; accepted
Oct. 22, 2001