Professional Documents
Culture Documents
©AANS, 2014
Object. Intraventricular hemorrhage (IVH) is the most common cause of hydrocephalus in the pediatric popula-
tion and is particularly common in preterm infants. The decision to place a ventriculoperitoneal shunt or ventricular
access device is based on physical examination findings and radiographic imaging. The authors undertook this study
to determine if head circumference (HC) measurements correlated with the Evans ratio (ER) and if changes in ven-
tricular size could be detected by HC measurements.
Methods. All cranial ultrasound (CUS) reports at the authors’ institution between 2008 and 2011 were queried
for terms related to hydrocephalus and IVH, from which a patient cohort was determined. A review of radiology
reports, HC measurements, operative interventions, and significant clinical events was performed for each patient in
the study. Additional radiographic measurements, such as an ER, were calculated by the authors. Significance was set
at a statistical threshold of p < 0.05 for this study.
Results. One hundred forty-four patients were studied, of which 45 (31%) underwent CSF diversion. The mean
gestational age and birth weight did not differ between patients who did and those who did not undergo CSF diver-
sion. The CSF diversion procedures were reserved almost entirely for patients with IVH categorized as Grade III or
IV. Both initial ER and HC were significantly larger for patients who underwent CSF diversion. The average ER and
HC at presentation were 0.59 and 28.2 cm, respectively, for patients undergoing CSF diversion, and 0.34 and 25.2
cm for those who did not undergo CSF diversion. There was poor correlation between ER and HC measurements
regardless of gestational age (r = 0.13). Additionally, increasing HC was not found to correlate with increasing ERs
on consecutive CUSs (φ = -0.01, p = 0.90). Patients who underwent CSF diversion after being followed with multiple
CUSs (10 of 45 patients) presented with smaller ERs and HC than those who underwent CSF diversion after a single
CUS. Just prior to CSF diversion surgery, the patients who received multiple CUSs had ERs, but not HC measure-
ments, that were similar to those in patients who underwent CSF diversion after a single CUS.
Conclusions. The HC measurement does not correlate with the ER or with changes in ER and therefore does not
appear to be an adequate surrogate for serial CUSs. In patients who are followed for longer periods of time before
CSF shunting procedures, the ER may play a larger role in the decision to proceed with surgery. Clinicians should
be aware that the ER and HC are not surrogates for one another and may reflect different pathological processes.
Future studies that take into account other physical examination findings and long-term clinical outcomes will aid in
developing standardized protocols for evaluating preterm infants for ventriculoperitoneal shunt or ventricular access
device placement.
(http://thejns.org/doi/abs/10.3171/2014.5.PEDS13602)
I
ntraventricular hemorrhage (IVH) is the most com- < 25 weeks) will develop severe hemorrhage classified as
mon cause of hydrocephalus in the pediatric popula- Grade III or IV.6,13 The long-term sequelae of IVH include
tion. It is estimated that approximately 30%–59% of parenchymal injury, hydrocephalus, cerebral palsy, and
all live-born, preterm infants will develop IVH and that cognitive delay.1,16,17 Currently, there is no consensus re-
36% of all extremely premature infants (gestational age garding the indications for, or the timing of, the treatment
of hydrocephalus following IVH in premature infants.
Abbreviations used in this paper: CUS = cranial ultrasound; ER Clinical suspicion for IVH can be raised through as-
= Evans ratio; HC = head circumference; ICP = intracranial pres- sessment of several factors including apneic or bradycard-
sure; IVH = intraventricular hemorrhage; VAD = ventricular access ic episodes, a bulging anterior fontanelle, and increased
device; VPS = ventriculoperitoneal shunt. head circumference (HC), and is confirmed by findings
on cranial ultrasound (CUS). The presence of an apneic or of hemorrhage and documented posthemorrhagic hy-
bradycardic episode suggests that hemorrhage is already drocephalus, but in whom no IVH was present on CUS
severe and strongly predicts placement of a temporizing imaging obtained at our institution, were labeled as hav-
device or shunt.13 Studies have demonstrated a positive ing “resolved hemorrhage” but were still included in the
correlation between findings on palpation of the anterior study. If periventricular leukomalacia and cystic brain
fontanelle and intracranial pressure (ICP).9,15 The reliabil- parenchymal changes followed a diagnosis of Grade IV
ity of this physical examination finding has been verified, hemorrhage, the patient was considered to continue car-
with an acceptable interrater reliability in evaluating the rying a Grade IV classification for analysis. The label of
anterior fontanelle.18 Nevertheless, guidelines for how hydrocephalus was applied if the radiographic dictation
these assessments can be used in clinical practice have not included the terms “hydrocephalus,” “ventriculomegaly,”
yet been established. “ventricular dilatation,” or “enlarged ventricles.” We mea-
Measuring HC at regular intervals in this population sured the ER, the ratio of the bifrontal horn diameter to
has long been thought to be an objective way to assess the biparietal bone diameter, for each CUS scan.7,10
progression of hydrocephalus. However, the criterion of a
2 cm/week increase in HC as an indicator of posthemor- Statistical Analysis
rhagic hydrocephalus has been shown to be unreliable.13 Potential predictors of CSF diversion were analyzed,
Using serial CUSs has become standard practice for many including initial birth weight, gestational age, presence
pediatric neurosurgeons to evaluate the need for shunt of hydrocephalus on the initial CUS, interval worsening
placement, although there is no evidence-based rationale of hydrocephalus between CUSs, and ER on each CUS.
for this practice.4,14 Data were analyzed using the Student t-test of means. For
Current practice is for physicians to use physical ex- the analysis of differences in ER and HC by IVH grade,
amination and radiographic findings without evidence ANOVA, Student t-test, and correlation studies were used
of which factors are most predictive of a need for shunt as indicated in the text. Interscan changes in ER and HC
placement. There is also no clear understanding of how were calculated and compared using the Pearson chi-
well clinical and radiographic findings correlate in pre- square test for association. Significance was set at p < 0.05
term infants. This study was undertaken to establish the for all parts of this study.
correlation, or lack thereof, between HC and CUS find-
ings, and to determine which of these clinical factors are
most predictive for CSF diversion. Results
Initial ER and HC as Predictors of CSF Diversion
Methods
Patient Selection Presenting patients’ demographic data, initial HC
measurements, and radiographic findings are shown in
Institutional review board permission was obtained Table 1, allowing comparison between patients who pro-
for this study. We queried our electronic medical record gressed to surgical CSF diversion (n = 45) and those who
for CUS reports entered between 2008 and 2011 and did not (n = 99). The vast majority of patients in the study
containing the phrases “intracranial hemorrhage,” “intra- who underwent CSF diversion did so after receiving only
ventricular hemorrhage,” “germinal matrix,” “bleed,” or 1 CUS at our institution (35 of 45 patients, 78%). Birth
“IVH.” Patients with concurrent diagnosis of congenital weight and gestational age at birth did not differ between
malformations or hydrocephalus due to causes other than patients who underwent CSF diversion and those who did
IVH were excluded. Patients born of term pregnancy (> 38 not.
weeks) were excluded. As shown in Table 1, there was a significant differ-
A retrospective chart review was then conducted to ence in initial ER for children who did and did not un-
collect data on the 144 patients identified. The following dergo CSF diversion. Overall, the average ER in the group
information was collected for each patient: gestational age undergoing diversion was 0.59, compared with 0.34 in
at birth; birth weight; HC measurements; characteristics the group not progressing to diversion (p < 0.001). When
of each ultrasound performed, including the time interval patients were stratified by hemorrhage severity, a larger
between the scans; grade of IVH; presence and progres- initial ER was observed in patients needing CSF diversion
sion of associated hydrocephalus; and Evans ratio (ER). for resolved, mild-moderate, and severe hemorrhage. As
Other forms of imaging, including CT scans and MRI, with the ER, the initial HC was significantly different be-
were not included for analysis because the number of pa- tween children who did and did not progress to CSF diver-
tients receiving imaging by these modalities was small. sion (p < 0.001), but this difference became less reliable
Additional clinical information was analyzed, including as patients were stratified by hemorrhage severity. For pa-
the date of placement of a VAD or shunt, the age at which tients with Grade I or II hemorrhage, initial HC was not a
diversion occurred, any revisions of shunts or shunt infec- significant predictor of diversion (p = 0.087); patients with
tions, and the date and cause of death of the patient where CSF diversion who had resolved or severe (Grade III or
available. IV) hemorrhage had initial HC measurements that were
significantly larger, but by a relatively narrow margin (p =
Ultrasound Imaging 0.029 and 0.041, respectively).
The authors evaluated all CUS images and reports.
Correlation Between ER and HC
Grading of IVH followed the scale defined by Papile et
al.12 Children who presented to our facility with a history We used a correlation analysis to determine if HC
* Key demographic data obtained for all patients, stratifying many categories by grade of hemorrhage (resolved; mild-moderate
[Grade I or II]; severe [Grade III or IV]) to demonstrate factors significantly different between patients who later received surgical
CSF diversion and those who did not. Unless otherwise indicated, the numbers in parentheses represent the SD. GA = gestational
age.
† The p values were calculated with either a chi-square or Student t-test. Significant differences (p < 0.05) are in boldface.
and ER correlated on initial scans for patients overall or CSF diversion in patients who are followed for longer pe-
if stratified by gestational age (≥ 30 or < 30 weeks) at the riods before surgery, we compared patients who received
time of imaging (Fig. 1). Initial HC measurements were multiple CUS images before CSF diversion to those who
not available for 14 patients. Contrary to our hypothesis, received only 1 CUS prior to CSF diversion (Table 4). In
there was a significant correlation between initial ER and 10 (22%) of 45 patients, > 1 scan was obtained before un-
HC only in patients who underwent the first CUS at less dergoing CSF diversion. Patients in this late CSF diver-
than 30 weeks gestational age, and no significant correla- sion group received an average of 4 CUSs before under-
tion between initial CUS and ER overall (Table 2). going CSF diversion. Gestational age at birth and birth
weight were not significantly different between patients
Correlation Between Changes in ER and HC who underwent early versus late CSF diversion, nor was
A contingency analysis was used to test if increases in the presence of high-grade IVH. Additionally, gestation-
ER were accompanied by changes in HC. We examined al age at the time of CSF diversion was not significantly
all patients who received > 1 scan without undergoing different between the 2 groups. Initial HC and ER were
CSF diversion, analyzing all intervals between scans. A significantly smaller in patients who underwent delayed
significant change in ER was set at 0.2, because this was CSF diversion (p < 0.01 for each); however, ER did not
similar to the difference between initial ERs in patients differ significantly between groups just prior to diversion
who did and did not undergo CSF diversion (Table 1). We (p = 0.22). Head circumference remained slightly smaller
chose 1.5 cm as a meaningful change in HC, taken from in patients who underwent CSF diversion at a later time
the study by Müller and Urlesberger, which showed 1.5 (p = 0.034).
cm to be a sensitive indicator of increasing hydrocepha- Discussion
lus.11 The contingency association was analyzed with the
Pearson chi-square test, revealing no significant associa- Predictors of CSF Diversion: ER Versus HC
tion between changes in ER and changes in HC (Table 3).
Management of infants with hydrocephalus from IVH
has been the subject of controversy, because many infants
Characteristics of Patients Undergoing Delayed Versus
with IVH may have spontaneous resolution of symptoms
Immediate CSF Diversion without shunt insertion.5 However, early interventions
To better understand the decision to proceed with such as lumbar puncture have been associated with lower
* Correlation between initial HC and ER in all patients prior to diversion, stratified by age as shown in Fig. 1. Correlation was
significant (p < 0.05; boldface) only for the subgroup of patients < 30 weeks old at initial scan, but not for patients ≥ 30 weeks old
or overall.
† Fourteen patients did not have an HC measured at initial scan.
TABLE 4: Comparison between patients with early surgical intervention after 1 CUS image and those who underwent
intervention after ≥ 2 CUS images*
CSF Diversion
Characteristic Early Late† p Value
no. of patients 35 10
GA in wks at birth 27.5 (3.1) 26.1 (2.6) 0.18
GA in wks at time of diversion 31.1 (3.3) 31.4 (3.2) 0.80
birth weight in g 1114.6 (469.0) 860.8 (298.8) 0.11
initial hemorrhage category
resolved 5 (14.3%) 0 (0%) 0.57
Grade I or II 3 (8.6%) 1 (10%) 1.00
Grade III or IV 27 (77.1%) 9 (90%) 0.66
ER
initial 0.61 (0.11) 0.50 (0.13) <0.01
before CSF diversion 0.61 (0.11) 0.57 (0.11) 0.22
HC in cm
initial 29.0 (3.0) 25.8 (4.0) <0.01
before CSF diversion 29.0 (3.0) 26.7 (1.8) 0.034
no. of scans before CSF diversion 1 (0) 4.0 (2.4) <0.001
* Demographic and measured data for patients who received early surgical intervention, defined as CSF diversion after 1 CUS,
compared to patients with later surgical intervention. Both HC and ER were significantly smaller in patients who underwent delayed
CSF diversion (p < 0.001 for both measures). Although initial ER measurements differed significantly between the 2 groups, the
ER on the last scan before diversion did not differ significantly between the early diversion and delayed surgery groups. The HC
measurements remained significantly different prior to diversion, regardless of whether patients underwent diversion early or late
(p = 0.034). Unless otherwise indicated, the numbers in parentheses represent the SD. Significant differences (p < 0.05) are in
boldface.
† At least 2 CUSs were done before VAD placement.
VPS or VAD at a later time. Future work on identifying ed them from developing HCs as large as the early CSF
baseline characteristics of patients with IVH who develop diversion group. It is important to note that an enlarging
progressive hydrocephalus would allow this subgroup to ER in the absence of increased HC or other signs of hydro-
be identified and treated earlier. cephalus could be due to parenchymal loss rather than hy-
drocephalus. This highlights the fact that a single clinical
Weaknesses of This Study and Future Directions measurement should not be used in isolation to determine
A weakness of our study is that other clinical indica- if patients require CSF diversion and that relying on ven-
tors, such as fontanelle tension or apneic/bradycardic epi- tricular size alone could lead to overtreatment.
sodes, were not evaluated. Neurosurgeons do take these Additional work is needed to develop standardized
into account when determining the need for CSF diversion clinical practices for CSF diversion in infants. Although
in infants, although these episodes are not always pres- our study did not show a correlation between HC mea-
ent, even with severe hydrocephalus. Although fontanelle surements and ER, it is possible that our retrospective de-
tension has been shown to have good interrater reliabil- sign was not optimally suited for this, because CUSs were
ity, there is no common method of measuring this tension performed at nonstandardized intervals. Prospective trials
quantitatively.8,9 Correlation of our findings to measured with clinical assessments of the anterior fontanelle timed
fontanelle tension or actual ICP measurement would help closely with CUS studies could better determine if clini-
determine whether ER or HC relates most closely to the cal assessments of the fontanelle can be sensitive enough
elevated ICP. to predict development of hydrocephalus. Additionally,
Patients who underwent late CSF diversion presented emerging research has emphasized the potential for using
with significantly smaller ER and HC than those who un- MRI flow monitoring to predict relative responses to CSF
derwent early CSF diversion. When these patients with so- diversion.2 Continued investigation into these diagnostic
called late CSF diversion were examined just prior to sur- assessments may lead to improved evaluation of IVH and
gery, their ER, but not their HC, no longer differed from allow for decreased incidence and morbidity of associated
those in patients with so-called early CSF diversion. This hydrocephalus.
suggests that clinicians may place more weight on ER than
HC measurements when following patients over time. Al- Conclusions
ternatively, these patients’ hydrocephalus could have been
slower progressing, and CSF diversion may have prevent- Despite the realization that hydrocephalus leads to
poor clinical outcomes and impaired cognitive function, 7. Evans WA Jr: An encephalographic ratio for estimating ven-
there is no standardized way to evaluate the efficacy of tricular enlargement and cerebral atrophy. Arch Neurol Psy-
shunt placement by an institution.3 Ultimately, the stan- chiatry 47:931–937, 1942
8. Gross SJ, Oehler JM, Eckerman CO: Head growth and devel-
dardization of criteria for which patients undergo CSF opmental outcome in very low-birth-weight infants. Pediat-
diversion or other treatments for hydrocephalus, and the rics 71:70–75, 1983
confirmation that these standards are efficacious, could 9. Kaiser AM, Whitelaw AG: Intracranial pressure estimation by
lead to an efficient protocol for placing VPSs in neonates. palpation of the anterior fontanelle. Arch Dis Child 62:516–
Such improvements in treatment algorithms could reduce 517, 1987
unnecessary testing and subjective test interpretations, 10. Kulkarni AV, Drake JM, Armstrong DC, Dirks PB: Measure-
and could potentially improve overall outcomes for in- ment of ventricular size: reliability of the frontal and occipital
fants with IVH. horn ratio compared to subjective assessment. Pediatr Neuro-
surg 31:65–70, 1999
Disclosure 11. Müller WD, Urlesberger B: Correlation of ventricular size and
head circumference after severe intra-periventricular haemor-
The authors report no conflict of interest concerning the mate- rhage in preterm infants. Childs Nerv Syst 8:33–35, 1992
rials or methods used in this study or the findings specified in this 12. Papile LA, Burstein J, Burstein R, Koffler H: Incidence and
paper. evolution of subependymal and intraventricular hemorrhage: a
Author contributions to the study and manuscript preparation study of infants with birth weights less than 1,500 gm. J Pedi-
include the following. Conception and design: Miller. Acquisition of atr 92:529–534, 1978
data: Ingram, Huguenard. Analysis and interpretation of data: Miller, 13. Riva-Cambrin J, Shannon CN, Holubkov R, Whitehead WE,
Ingram, Huguenard. Drafting the article: Ingram. Critically revising Kulkarni AV, Drake J, et al: Center effect and other factors
the article: all authors. Reviewed submitted version of manuscript: influencing temporization and shunting of cerebrospinal fluid
all authors. Approved the final version of the manuscript on behalf of in preterm infants with intraventricular hemorrhage. Clinical
all authors: Miller. Statistical analysis: Ingram, Huguenard. Admini article. J Neurosurg Pediatr 9:473–481, 2012
strative/technical/material support: Chern. Study supervision: Miller, 14. Stonestreet B: Intraventricular hemorrhage in premature in-
Chern. fants. Cerebrospinal Fluid Res 6:S3, 2009
15. Taylor GA, Madsen JR: Neonatal hydrocephalus: hemody-
namic response to fontanelle compression— correlation with
References intracranial pressure and need for shunt placement. Radiology
1. Aquilina K: Intraventricular haemorrhage in the newborn. Adv 201:685–689, 1996
Clin Neurosci Rehabil 11:22–24, 2011 16. Volpe JJ: Brain injury in the premature infant. Neuropathology,
2. Bradley WG Jr, Whittemore AR, Kortman KE, Watanabe AS, clinical aspects, pathogenesis, and prevention. Clin Perinatol
Homyak M, Teresi LM, et al: Marked cerebrospinal fluid void: 24:567–587, 1997
indicator of successful shunt in patients with suspected nor- 17. Volpe JJ: Neurology of the Newborn, ed 5. Philadelphia:
mal-pressure hydrocephalus. Radiology 178:459–466, 1991 Saunders/Elsevier, 2008
3. Brouwer A, Groenendaal F, van Haastert IL, Rademaker K, 18. Wellons JC III, Holubkov R, Browd SR, Riva-Cambrin J,
Hanlo P, de Vries L: Neurodevelopmental outcome of preterm Whitehead W, Kestle J, et al: The assessment of bulging fonta-
infants with severe intraventricular hemorrhage and therapy nel and splitting of sutures in premature infants: an interrater
for post-hemorrhagic ventricular dilatation. J Pediatr 152: reliability study by the Hydrocephalus Clinical Research Net-
648–654, 2008 work. Clinical article. J Neurosurg Pediatr 11:12–14, 2013
4. Brouwer M: Treatment and Outcome of Neonatal Haem-
orrhagic Brain Injury [dissertation]. Utrecht, The Nether-
lands: Utrecht University, 2011
5. de Vries LS, Liem KD, van Dijk K, Smit BJ, Sie L, Rademaker
KJ, et al: Early versus late treatment of posthaemorrhagic ven- Manuscript submitted November 26, 2013.
tricular dilatation: results of a retrospective study from five Accepted May 8, 2014.
neonatal intensive care units in The Netherlands. Acta Paedi- Please include this information when citing this paper: pub-
atr 91:212–217, 2002 lished online June 20, 2014; DOI: 10.3171/2014.5.PEDS13602.
6. Dykes FD, Dunbar B, Lazarra A, Ahmann PA: Posthemor- Address correspondence to: Brandon A. Miller, M.D., Ph.D.,
rhagic hydrocephalus in high-risk preterm infants: natural his- Department of Neurosurgery, Emory University, 1365 Clifton Rd.
tory, management, and long-term outcome. J Pediatr 114:611– NE, Bldg. B, Ste. 2200, Atlanta, GA 30322. email: brandon.miller@
618, 1989 emory.edu.