£1.35. .I-ir. ...,

 

. yflnri .

:3: :a.

 

 

I
x 3
an}...
a 5:.
15;.

. 1 ~‘ r on:
a gay...
eh:

\5. l
5.2...
t

:33..qu

3m

I: .
a... Ends
at. 3&2»... .5“
u. ‘ . 1"! .
(Eryn... «if

V t ‘

"El-itvlts,

.2:

. 4m, _
.. .3 335mm“
#3:»... .

is

It. V .1
.‘li‘fi . .. ,. {J
.4. v . A .
:33. A am?
Em??? meswy u.
. .
k5,. .

.
V’s.
t

.1... 5 {'55.
.3 I}!!!
f).

x. u: .r
9.72.. a
‘1

E3 .32...

—\I «lung
.11
I.

i.‘ I”), 35:: ..
..«u‘g!3l9t a
, .. \lzx

mhtunulflaiu 31:19:”

.55. .31. a
’ §’--!, 2‘
tuhbif 53

Ml _ luff}.
it..»...fiu2...§$§. 5",

fit I: (DA—305

5
$1.“;

1‘ .Et‘s.rlnn

.1. 7171.
:3 2:;

. .. in Ev
73:: 2‘55
. Z . . ,

9.. firs?

r:
:-

virilzfih I.

.uu.
.rvu. Vin" ”gnu-f

. z... 4‘

. 32.03..
n. .:i.. w

:55.
It.

$3....Et V

a to z . 1h
1:3,...3p4um 33.94%
. 1 Mann

 

a 3.. I..." . .
59.12:)!!! at!

:3. a: .3
.3. S... » . L
line-Qil‘r‘ftlv‘l‘ d \
‘Iw «tin/Iv ‘5‘ (J 3L
£.§\?r.(.l.v
alatnouv .13 .
a 1.11.. 5...: (of

 

 
 

[3:25.-

:2 . ' LIBRARY
Michigan State
9 0 0 7 University

This is to certify that the
thesis entitled

A Systematic Review of Neurodevelopmental and Cognitive
Outcomes in Premature Children Born with Germinal Matrix
and lntraventricular Hemorrhage (GM/IVH)

presented by

Jun-tsui Fan

has been accepted towards fulfillment
of the requirements for the

MS. degree in Epidemiology

 

i. ~

' \ l “‘ t“ 7F N
\ - '.—-—-~'-"~-~- , ._,.*'
\-’\._. {1&5 J B“

 

Major Professor’s Signature
3.1/1 s /0 5,

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE
JUN 2 2 200

 

V

 

r» u

I.’ A; 1‘ ; OI:
«,V '5 i '-‘ o I,

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/C|RC/Date0ue.indd-p.1

A SYSTEMATIC REVIEW OF NEURODEVELOPMENTAL AND
COGNITIVE OUTCOMES IN PREMATURE CHILDREN WITH
GERMIN AL MATRIX AND INTRAVENTRICULAR HEMORRHAGE

By
J un-tsui Fan

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Epidemiology

2006

ABSTRACT

A SYSTEMATIC REVIEW OF NEURODEVELOPMENTAL AND
COGNITIVE OUTCOMES IN PREMATURE CHILDREN WITH
GERMINAL MATRIX AND INTRAVENTRICULAR HEMORRHAGE
By

J un-tsui Fan

To determine whether isolated GM/IVH is associated with neurodevelopmental and
cognitive abnormalities, I adopt a systematic review approach that combines qualitative and
quantitative methods in this thesis. MEDLINE and 181 Web of Knowledge are the main
sources for literature searches. A ranking of importance for the selected studies is provided
after combining the evaluation with causal criteria and study quality criteria. The studies of
van de Bor et al. (1988, 1993, and 2004) are regarded as important studies according to the
ranking. Weighted OR shows that children with GM/IVH have a 2.5 times higher risk of
abnormal neurodevelopment and abnormal general cognition than those without. The
differences in weighted means of neurodevelopmental and cognitive tests range from 1.2 to
3.3. The results suggest that GM/IVH may slightly impair children’s special cognitive
outcomes, while its effects on children’s abnormal global cognition is not significant. In
addition, GM/IVH in preterm infants leads to an increased risk of adversely affected
neurodevelopmental performance. The effect of GM/IVH on school performances of

children or adolescents is unfavorable.

This thesis is dedicated with love to
Moi-kien Yi, Li-Tai-moi Y1, Ngiuk-Siuk Y1,
Yang-tse Fan, J un-huei Fan,
Kai-hui Chang, and Yen-yi Ho.

III

ACKNOWLEDGEMENTS

This thesis originates from a lot of people’s wisdom, enthusiasm, and aspiration. I
wrote it down with a great hope of contributing the organized knowledge and interesting
discovery to people who want to do further relevant explorations. I feel very glad for the
accomplishment of this thesis and have an immense gratitude to people who have helped,
supported, and took care of this work.

From the bottom of my heart, I thank Dr. Paneth, my advisor, for your enlightened
teaching, generous assistance, and admirable patience. In the past two years, my abilities,
confidence, and enthusiasm for researches were all elevated miraculously because of
working with you. I appreciate Dr. Kama and Dr. Breslau, my committee, for your
valuable suggestions and nice comments. Thank you so much.

Thank Mom for always being my strongest supporter either for my spirit or my daily
life. \Vrthout you, I cannot achieve this challenging mission. Thank grandfather and
grandmother for your encouragement and love. Thank Jun-huei, my twin sister, for always
making me happy and cheering me up. Thank Yang-tse, my brother, for taking care of our
family and allowing me not to won'y about anything. Thank Yen-yi for answering all my
research questions. Finally, I would like to thank Kai-hui for editing and formatting my
thesis, helping me prepare the defense, taking care of my daily life, and giving me perpetual
love and happiness. I am sincerely grateful for all that you have done. Thank you very

much.

IV

TABLE OF CONTENTS

LIST OF TABLES .................................................................................................................. VI
Chapter One Introduction .......................................................................................................... 1
I. Research Purpose ........................................................................................................ 1

11. Background ............................................................................................................... 1

111. Definition of GMH/IVH .......................................................................................... 2

IV. Classification of IVH in the Premature Infant ......................................................... 4

V. Diagnosis of IV H ....................................................................................................... 7
Chapter Two Methods of this Research .................................................................................. 10
I. Hypothesis ................................................................................................................ 10

11. Literature Search ..................................................................................................... 10

1. Sources for the Literature Search ........................................................................ 10

2. Determining Key Concepts and Generating Search Terms ................................ l4

3. Strategies for Performing MEDLINE! PubMed ................................................. 15

III. Literature Selection ................................................................................................ 16

1. Inclusion and Exclusion Criteria ......................................................................... 16

2. Selection with Titles, Abstracts, and Texts of Literature .................................... 17

IV. Literature Analysis ................................................................................................. 18

1. Determining the Type of Literature Review ....................................................... 18

2. Methods for Summarizing Evidence .................................................................. 21

Chapter Three Results ............................................................................................................. 26
1. Studies Identified ...................................................................................................... 26

II. Introduction of Birth Cohorts .................................................................................. 26

III. Measures and Classification of IVH ...................................................................... 28

IV. Follow-up and Assessment Rates in Studies .......................................................... 29

V. Study Quality Evaluation ......................................................................................... 31

VI. Ranking of Importance for Selected Studies ......................................................... 35

VII. Frequencies and Odds Ratios of Developmental Abnormalities ......................... 36
VIII. Averages and Standard Deviations of Developmental Tests .............................. 38

IX. Overall Evaluation and Interpretation ................................................................... 39
Chapter Four Discussion ......................................................................................................... 42
1. Evaluation of Causal Criteria .................................................................................... 42

11. Limitations ............................................................................................................... 44

III. Strengths ................................................................................................................. 46

IV. Conclusions ............................................................................................................ 47

V. Further Discussion ................................................................................................... 47
TABLES .................................................................................................................................. 50
BIBLIOGRAPHY ................................................................................................................... 90

LIST OF TABLES

Table l-A (l of 2): The Numbers, Gestational Ages, and Birth Weights of Baseline Cohorts

in Studies without a Full-Term Control Group ....................................................................... 50
Table l-A (2 of 2): The Numbers, Gestational Ages, and Birth Weights of Baseline Cohorts
in Studies without a Full-Term Control Group ....................................................................... 51
Table l-B (l of 2): The Numbers, Gestational Ages, and Birth Weights of Baseline Cohorts
in Studies with a Full-Term Control Group ............................................................................ 52
Table l-B (2 of 2): The Numbers, Gestational Ages, and Birth Weights of Baseline Cohorts
in Studies with a Full-Term Control Group ............................................................................ 53
Table 2-A (1 of 2): Follow-up Rates, Assessment Rates, and Number of Assessment in
Studies without a Full-Term Control Group ........................................................................... 54
Table 2-A (2 of 2): Follow-up Rates, Assessment Rates, and Number of Assessment in
Studies without a Full-Term Control Group ........................................................................... 55
Table 2-B (1 of 2): Follow-up Rates, Assessment Rates, and Number of Assessment in
Studies with a Full-Term Control Group ................................................................................ 56
Table 2-B (2 of 2): Follow-up Rates, Assessment Rates, and Number of Assessment in
Studies with a Full-Term Control Group ................................................................................ 57
Table 3-A (1 of 2): Study Quality Evaluation ...................................................................... 58
Table 3-A (2 of 2): Study Quality Evaluation ...................................................................... 59
Table 3-B (l of 2): Study Quality Evaluation ...................................................................... 60
Table 3-B (2 of 2): Study Quality Evaluation ...................................................................... 61
Table 4 ( l of 2): Ranking of Importance for Selected Studies ............................................ 62
Table 4 (2 of 2): Ranking of Importance for Selected Studies ............................................ 63
Table S-A (1 of 2): Frequencies and Odds Ratios of Abnormal Outcomes in Studies without
a Full-Term Control Group ..................................................................................................... 64
Table S-A (2 of 2): Frequencies and Odds Ratios of Abnormal Outcomes in Studies without
a Full-Term Control Group ..................................................................................................... 66
Table 5-B (l of 2): Frequencies and Odds Ratios of Special Cognitive Outcomes in a Study
with a Full-Term Control Group ............................................................................................. 68
Table 5-B (2 of 2): Frequencies and Odds Ratios of Special Cognitive Outcomes in a Study
with a Full-Term Control Group ............................................................................................. 69
Table 6—A (1 of 2): Means and Standard Deviations of Developmental Tests in Studies
without a Full-Term Control Group ........................................................................................ 70
Table 6-A (2 of 2): Means and Standard Deviations of Developmental Tests in Studies
without a Full-Term Control Group ........................................................................................ 72
Table 6-B (l of 2): Means and Standard Deviations of Cognitive Tests in Studies with a
Full-Term Control Group ........................................................................................................ 74

Table 6-B (2 of 2): Means and Standard Deviations of Cognitive Tests in Studies with a

VI

Full-Term Control Group ........................................................................................................ 75
Table 7-A Weighted OR for Abnormal Neurodevelopment ................................................ 76
Table 7-B Weighted OR for General Cognition .................................................................. 76
Table 7-C Weighted Means for Neurodevelopment: Motor Abilities ................................. 77
Table 7-D Weighted Means for General Cognition ............................................................. 77
Table 7-E Weighted Means for Special Cognition - Language ........................................... 78
Table 7-F Weighted Means for Special Cognition— Reasoning & Memory ....................... 78
Table 8 ( l of 2): Causal Criteria for Selected Studies ......................................................... 79
Table 8 (2 of 2): Causal Criteria for Selected Studies ......................................................... 80
Appendix Table l (l of 2): Measurements and Classification of IVH ................................ 81
Appendix Table 1 (2 of 2): Measurements and Classification of IVH ................................ 83
Appendix Table 2 (l of 2): Measurements of Outcomes ..................................................... 85
Appendix Table 2 (2 of 2): Measurements of Outcomes ..................................................... 87

VII

Chapter One
Introduction

I. Research Purpose

Many studies have discussed the relationship between severe brain damage (germinal
matrix and/or intravenuicular hemorrhage (GM/IVH) with parenchymal lesions/ ventricular
enlargement (PUVE)) and developmental abnormalities in premature children. However,
the influence of GM/IVH alone on the outcomes of premature children has been
insufficiently studied. The purpose of this research is to compare the outcomes of
premature surviving children with and without GM/IVH in order to determine whether
isolated GM/IVH is associated with neurodevelopmental and cognitive abnormalities.

11. Background

Intracranial hemorrhage (ICH) may affect newborns of all gestational ages and is often
clinically silent. Germinal matrix hemorrhage and intraventricular hemorrhage (GMH/IV H)
are most common in the premature population. Estimates of GMH/IVH frequency in
preterm infants have declined from 30 to 40% in the early 19805 to less than 20% in the
19908 [1]. GM/IVH was once found in about 40% of very low birth weight (V LBW;
birthweight< 1500 grams) infants [2]-[7]. However, a recent study suggests that 25-35% of
extremely low birth weight (ELBW; birthweight< 1000 grams) infants will have some
degrees of IVH identified in the early newbom period [8], and the prevalence in infants

whose gestational age are less than 32 weeks is about 15% [6].

1

Many studies have reported the relationship between premature infants and disabilities.
Nearly 8% of VLBW infants have disabling cerebral palsy (CP) at age 2 [9] [10]. 5% of
VLBW infants have severe (moderate to profound) mental retardation at school age [11]
[12]. The percentage of VLBW infants -— who have learning difficulties, attention deficit
disorder and strabismus - has been estimated to be as high as 40%, 30% and 15%,
respectively [13].

Several studies have reported that 30% to 50 % of prematurely born children who
display impaired academic achievements and/or behavioral disorders require additional
educational resources [14]-[ 19]. However, the association between GM/IV H and
developmental outcome in premature born children is still uncertain and controversial. It is
unclear whether GM/IVH leads to children’s developmental disabilities. Therefore, the

topic is worthy of further investigation.

III. Definition of GMH/IVH
Germinal Matrix Hemorrhage (GMH):

Understanding the anatomy and neuropathology of germinal matrix is essential,
because majority of the intraventricular hemorrhage originates from the germinal matrix [20].
Gerrninal matrix tissue is found mean the caudate nucleus [13] [20] directly ventrolateral to
the lateral ventricle [20]. The germinal matrix is a temporary embryonic structure of the
telencephalon existing before 36 weeks of gestation. It not only serves as the source of

cerebral neuronal precursors from 10 to 20 weeks of gestation but also plays a role in the

2

formation of cerebral oligodendroglia and astrocytes. Cells for the pulvinar (dorsal and
posterior portions) of the thalamus originate from the caudothalamic germinal matrix [21]
that provides the neuronal cell clusters for cortical association regions to connect with the
thalamic nuclei [22].

The germinal matrix from 28 to 32 weeks is noticeable on ultrasound (US) in the
thalamostriate groove at the level of the head of the caudate nucleus at the site, of or lightly
posterior to, the foramen of Monro [23]-[28]. That is the most common site for germinal
matrix hemorrhage [20]. The germinal matrix is surrounded by, and is divided from, the
cerebrospinal fluid (CSF) of the lateral ventricles by a single layer of ependyma. Once the
ependymal lining breaks, intraventricular hemorrhage (IVH) occurs through extension of the
bleeding into the ventricles.

The definition of germinal matrix hemorrhage is based on US diagnosis, and it is
described as follows: on at least one ultrasound scan, there is a focal echodensity in the
thalamocaudate groove that is just lateral to the frontal horns of the lateral ventricles. It
sometimes extends to the head of the caudate nucleus [12].

The germinal matrix generally has involuted by the 32nd to 34th week of gestation [13]
and is essentially exhausted by term [20]. That is the reason why the incidence of GMH
among term neonates is rare but this type of intracranial hemorrhage is commonly seen in
preterm infants [13].

lntraventricular hemorrhage (IVH)

3

In accordance with the above statement, intraventricular hemorrhage is a further
pathological progress usually originating from germinal matrix hemorrhage. The most
general source of intraventricular hemorrhage is rupture of the ependymal layer and
extension of bleeding into the lateral ventricles [13]. It is difficult to ascertain the origin of
bleeding in terms of the specific germinal matrix location [l3].The following is a brief
definition of IVH based on US diagnosis: on at least one ultrasound scan, there is an
echodense focus or foci within the lateral, third or fourth ventricles separate from the choroid
plexus, and it must be at least as echodense as the choroid plexus [12]. It can also be
diagnosed when inequality of the choroid plexus margin reveals adherent intraventricular
blood [12].

IV. Classification of IVH in the Premature Infant

Papile, Burstein, Burstein, and Koffler advanced the first classification strategy for
intraventricular hemorrhage in 1978 [2]. Although Papile et al. used computed
tomography (CT) scan to describe the results of brain imaging [2], their classification system
provides the most common and widely used set of terms to describe ultrasound images of
brain damage in preterm infants. lntraventricular hemorrhages have been classified into
four separate grades in a hierarchical scheme numbered from Ito IV based on the CT
abnormalities to describe the “varied natural history of IVH” [2].

Grade I: Subependymal hemorrhage

Grade 1]: lntraventricular hemorrhage without ventricular dilatation

4

Grade HI: lntraventricular hemorrhage with ventricular dilatation

Grade IV: lntraventricular hemorrhage with parenchymal hemorrhage

The classification of Papile and her coworkers is based on the location and amount of
bleeding in an infant’s brain. In addition, whether the hemorrhage enlarges or not is also
important for the grades of severity of intraventricular hemorrhage.

Neonatal cranial ultrasound (US) scanning was introduced in the early 19803, which
enabled people to obtain more knowledge about the anatomy and neuropathology of an
infant’s brain. Volpe in 1987 provided a classification of IVH modified from that of Papile
et al. In accordance with neuropathologic findings, Volpe suggested that the presence of
parenchymal lesions (IV H grade IV in Papile’s category) should be noted separately. That
is because periventricular hemorrhagic infarction or other parenchymal lesions usually do
not only result from extension of GM/ IVH into normal brain parenchyma. Therefore,
there are three grades for IVH and a separate notation for periventricular echodensity in
Volpe’s classification [20]. The basis of Volpe’s classification is the occurrence and
proportion of blood in the germinal matrix and lateral ventricles [20].

Grade I: Gerrninal matrix hemorrhage with no or minimal lntraventricular hemorrhage
(10% of ventricular area on parasagittal view)

Grade II: lntraventricular hemorrhage (10%-50% of ventricular area on parasagittal
view)

Grade 111: lntraventricular hemorrhage (>50% of ventricular area on parasagittal view;

5

usually distends lateral ventricle)

Separate notation: Periventricular echodensity (location and extent)

Although Papile’s system is widely used, Paneth in 1999 indicated it has some
weaknesses. First, it is cumulative. As a result, entities such as isolated ventricular
enlargement or isolated parenchymal hemorrhage fall nowhere in Papile’s system, because
each grade that is above grade II includes the entities below it in the hierarchy [29].
Second, it lacks strong pathologic proof to indicate that grade IV hemorrhage is usually an
extension of subependymal hemon'hage [29]. Preterm infants’ parenchymal hemorrhages
tend to be a component of white matter damage (WMD). For instance, bleeding into a
pre-existing infarction, or microscopic hemorrhage presents together with periventricular
leukomalacia (PVL) [29]. In addition, so-called “grade IV hemorrhage”, which on US is
an extensive echodense region in white matter surrounding the lateral ventricle, is sometimes
not bleeding at all [29]. Many interpretations about “grade IV hemorrhage” based on
ultrasound images are not appropriate for natural history of IVH [29]. Therefore, Paneth
addressed the following three general categories involving not only GM/IV H but also the
common brain lesions in premature infants. These categories are more concordant with
neuropathologic discoveries.

1. WMD

2. Hemorrhages in non-parenchymal areas of the brain

3. Lesions in other brain locations: cerebellum, basal ganglia, brain stem, etc [29].

6

Whitaker (1996, 1997) et al. and Pinto-Martin (1999) et al. published studies that did
not grade the severity of IVH in a hierarchical scheme numbered from I to IV. Instead,
there were only two groups in their studies: (1) Isolated germinal matrix hemorrhage and/ or
intraventricular hemorrhage (GM/IVH). (2) Parenchymal lesion and/ or ventricular
enlargement (PUVE) with or without GM/IVH [lO]-[12]. The two groups correspond to
grades I/ II IVH and grades HI/ IV IVH respectively.

This thesis discusses the outcomes in premature children with germinal matrix and
intraventricular hemorrhage (GM/IVH). “Grade HI/ IV IVH” or “PUVE” are not studied
here. Therefore, the term “GM/IVH” will be utilized consistently in the thesis, which
would not conflict with the different terms such as “grade III] IVH” but specifically excludes
“grade III/ IV IVH” and “PUVE” as described above.

V. Diagnosis of IVH

Ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI)
are the primary methods currently used in the evaluation of premature infants’ brains [30].
CT has been available for the past three decades [30]. US scan has been widely used since
its introduction in the early 19805 [10][14]. Although MRI has been shown to provide
superior images of IVH, particularly after the first few days of brain hemorrhage [31]-[34], it
currently cannot supplant US in the evaluation of IV H [20]. This is because MRI has some
disadvantages: (1) it requires transporting the infant to the scanner, (2) it has a relatively long

data acquisition time, (3) it prevents use of metallic materials found on standard neonatal

7

monitoring and support equipment, and (4) it is expensive [20]. Therefore, the following
discussions exclude MRI and only focus on US and CT whose principles, strengths, and
weaknesses are described.

Computed Tomography (CT)

Computed tomography (CT) is sometimes called CAT scan. Special x-ray equipment
is used to obtain many images from different angles and then these images are combined to
show a cross-section of body tissues and organs [30]. CT scan is able to demonstrate the
site and extent of intraventricular hemorrhage effectively [2] [4] [20] [35]-[42].

CT scan provides a poorer image of the precise details of soft tissues (particularly the
brain) compared with MRI [30]. On the other hand, it is more sensitive than
ultrasonography in cases where extra axial (subdural, subarachnoid) or posterior fossa
haemorrhage is suspected [20]. However, CT scan has major disadvantages: (l) the infant
is required to be removed from the intensive care unit (ICU), and (2) the brain and eyes of
the infant will be exposed to ionizing radiation [20] [30]. In addition, it is not a portable
technology and is thus gradually replaced by US scan [20].

Ultrasound (US) Brain Scan

Ultrasound scan has high reliability and versatility in identifying the degree of severity
of IV H, which ranges from isolated germinal matrix hemorrhage to major degrees, with or
without periventricular hemorrhagic infarction [20]. Most cranial ultrasound studies in

neonates utilized a sector or linear-array transducer with a frequency of 7 MHz or higher

8

[43]. Images of ultrasound scans are collected through the anterior fontanelle, in both the
coronal and sagittal planes [13] [20]. The posterior fossa and midbrain can be imaged well
through the posterior fontanelle [13] [20].

Performing three US scan at 4 hours, 24 hours, and 7 days of life can improve the
sensitivity of US detection to more than 75% for GM/IVH [13]. The screening
examination at 7-14 days of life can detect more than 80% of GM/IVH which may result in
post-hemonhagic hydrocephalus [20] [30] [43]. Furthermore, the sensitivity of ultrasound
for detecting white matter abnormalities can be increased further by performing weekly
scans after the first month of life [43]. Later scans at 36 to 40 weeks postrnenstrual age are
recommended by the Committee of the American Academy of Neurology and the Child
Neurology Society to evaluate infant prognosis [30].

The major advantage of US is its portability [30]. This is because US can be
performed without moving infants from the intensive care unit (ICU) [30]. In addition, it is

safer and cheaper than CT and MRI [20].

Chapter Two

Methods of this Research
I. Hypothesis

This study has a null hypothesis: there is no significant difference for premature
children’s neurodevelopmental and cognitive outcomes between the group of infants who
have GM/IVH and those who with normal results of ultrasound scanning. On the other
hand, an alternative hypothesis of this study is that there are significant differences for those
outcomes between GMH/IVH group and normal ultrasound scanning group. I here extract
evidence from the literature to evaluate and ascertain this hypothesis by systematic literature
review in this thesis.

11. Literature Search
1. Sources for the Literature Search
( i ) Bibliographic Databases on the Internet

Search engines or bibliographic databases on the Internet are the major sources of
literature that will be analyzed in this thesis. These sources store most of the current
research and are convenient for literature search. These sources include MEDLINE/
PubMed, ISI Web of KnowledgeSM, electronic journals, and a powerful searching engine
“Scholar Google”.

MEDLINE] PubMed— MEDLINE is a comprehensiveand cross-referenced database

of citations provided by the National Library of Medicine (NLM) [44]. It includes medical

lO

literature from 1966 to the present. Searching MEDLINE through PubMed provides
researchers several advantages such as a user-friendly interface, cosmopolitan search
resources, and access to full-text articles [45]. The following components are included in
the MEDLINE database: Title, Author(s), Affiliation, Abstract, Language, Publication Date,
Journal Title, and Medical Subject Heading (MeSH) terms [46]. NLM defines the MeSH
terms chosen from 22,568 descriptors and continually updates them [46].

The PubMed tt ://www.ncbi.nlm.nih. ov/entrez/ ue .fc i?db=PubMed provides a

 

typical single search window (Figure 1). By entering keywords in that window, a link to
the full-text of more than 4,400 journals will be provided by the PubMed searches [47].
Results of the literature searches from PubMed include literature ( l) with free full texts, (2)
with abstracts but without free full texts, or (3) without abstracts and free full texts [44].
For literature with abstracts but without free full texts, usually only subscribers of those
journals can access the full text. However, non-subscribers may still access those articles
by paying a fee [44]. Once a useful literature is found but cannot be downloaded from
PubMed, other sources and search methods will need to be used.

ISI Web of KnowledgeSM— ISI Web of KnowledgeSM is another electronic academic
literature source, which is as important as MEDLINE] PubMed for this thesis. It enhances
the function of ISI Web of Science and provides three major databases: “Science Citation
Index Expanded”, “Social Sciences Citation Index” and “Arts & Humanities Citation Index”.

For this thesis, we only use Science Citation Index Expanded for literature search in ISI Web

11

 

of Knowledge“, because it provides articles that belong to various science fields.
Through ISI Web of Knowledge“, it is possible to search for articles from more than 8,500
journals in different fields of science. In addition, it has a unique function, called “Cited
Reference Searching”, which provides an easy pathway of cross citation.

However, it is not convenient for individuals because the use of “ISI Web of

SM”

Knowledge needs to be licensed through academic registration. Therefore, people with

no “ID” and “Password” from an academic register cannot access “ISI Web of

SM”. For this thesis, literature searches from ISI Web of KnowledgeSM are

Knowledge
administered by registering via the Proxy Server of Michigan State University Libraries.

Electronic Joumals— It is possible that people obtain only abstracts without full-text
articles while searching for articles on MEDLINE/ PubMed or ISI Web of Knowledge“.
Under this circumstance, searching for literature from electronic journals is an appropriate
and practical solution. The results of literature searches fiom MEDLINE/ PubMed or ISI
Web of KnowledgeSM do not always provide the address of a joumal’s website. Therefore,
the best way to search a joumal’s website is through the libraries’ networks of academic
institutions, such as universities, colleges, or research centers. In this thesis, I access
individual joumal’s website through the homepage of Michigan State University Libraries
(http://www.lib.m_su.edu/).

Scholar Google— Scholar Google (http://scholangooglecofl) is a very powerful and

useful search engine similar to Google (http://wwwgooglecofl). The difference between

12

Scholar Google and Google is that Scholar Google focuses only on academic papers and
books but not on other irrelevant links. Obtaining the knack of how to effectively and
efficiently utilize this public search engine is crucial for literature retrieval [48]. The main
advantage of using Scholar Google for literature search is that it not only combines several
academic databases but also provides links of citations. Scholar Google is a valuable
source for literature search in this thesis because it can find literature that cannot be found
from PubMed or 131 Web of Knowledge“.
( ii ) Medical or Epidemiological Journals in Libraries

The medical or epidemiological journals stored in libraries are other important literature
sources for this thesis. Databases on the Internet are convenient sources for searching
recent publications. For epidemiological observational studies, however, only some of the
relevant articles can be found by a computer-aided literature search [49]. The reason is that
academic websites do not always provide the complete databases, and the searches may not
return all relevant literature. In addition, some of the full-text literature published a long
time ago may not be in the databases. Therefore, printed journals in libraries still have a
great value and cannot be replaced or ignored. In this thesis, some articles are from printed
journals such as Joumal of Developmental and Behavioral Pediatrics, Pediatrics, Archives
of General Psychiatry, Seminars in Perinatology, Developmental Medicine and Child
Neurology, Journal of Pediatric Ophthalmology and Strabismus, the Journal of Pediatrics

and Early Human Development. Most of these journals were found in libraries of

13

Michigan State University (MSU) and University of Michigan at Ann Arbor. If a journal
could not be found in these two libraries, I asked the librarian to recall the journal from other
libraries, or I recalled the journal on the website of MSU library by myself.

2. Determining Key Concepts and Generating Search Terms

The research question of this thesis is: “Is GM/IVH causally associated with
neurodevelopmental and cognitive abnormalities in premature children?” Breaking down
the research question into some key concepts and then generating useful search terms is the
first step to perfonning a literature search [46] [50].

(1) Concepts about exposure include the following terms: germinal matrix hemorrhage,
subependymal hemonhage, intraventricular hemorrhage, periventricular hemorrhage,
periventricular-intraventricular hemorrhage, mild brain damage, and intracranial
hemorrhage.

(2) Concepts about other risk factors include the following terms: low birth weight,
low—birth-weight, low birthweight, preterm, preterm infant, premature and, prematurity.

(3) Concepts about outcomes include the following terms: neurodevelopment,
neurodevelopmental, neurodevelopmental outcome, neurodevelopmental performance,
cognition, cognitive, cognitive outcome, cognitive performance, learning, behavior, outcome,
and school performance.

To perform the literature search effectively, I chose two or more terms from different

concept groups and then typed them in the search box. The results are shown after clicking

14

the “Go” button on the PubMed web page. MeSH terms lead retrievals into limited records
on that specific topic [51] and may improve the specificity of literature searches [52]. On
the other hand, a free-text search, which is searching through generic words, retrieves
records including the given word anywhere in the title or abstract even if these records are
not essentially related to the subject [53]. To obtain optimal results, therefore, free-text
search needs to be performed with a MeSH search [54].

Finally, it is very important to note that English is the sole language adopted by
MEDLINE field contents [46]. Hence, all keywords have to be in English for any literature
search through MEDLINE.

3. Strategies for Performing MEDLINE/ PubMed

MEDLINE allows Boolean operators [55] to combine keywords for the searches [46].
AND, OR, and NOT are the connectors which may improve search specificity [53] [55].

All the connectors have to be capital letters in MEDLINE [55] [56]. To retrieve a set of
records simultaneously covering all given search terms, the AND operator should be utilized
[46]. On the other hand, records contain any of the given terms can be retrieved by the OR
operator [46]. By using NOT operator, only records that involve the first key word will be
returned, and those related to the second will be excluded [46]. In addition, utilizing the
truncation function (*) can increase the search recall due to synonyms (e.g., hemorrhage and
bleeding) and variants (e.g., behaviour and behavior) of a term [46]. For a multiword

keyword, typing the keyword in quotation marks (e. g., “low birth weight”) is appropriate

15

and beneficial [46]. Following these strategies may help researchers to conduct efficient
and effective literature search.

111. Literature Selection

1. Inclusion and Exclusion Criteria

Any review paper should have explicit inclusion and exclusion criteria in order to
appropriately select all relevant studies [56] [57]. The criteria of this review are listed as
follows.

(1) Only published papers written in English are included. Published papers written in
other languages, such as Chinese or Spanish, are excluded.

(2) This thesis includes only relevant papers published after 1980 and excludes papers
published before 1980.

(3) The included studies should involve GM/IVH (IVH Grade VII) and have a clear
classification of IV H. Studies are excluded if the authors only address PUVE (IV H Grade
III/IV) or combine GM/IVH with other types of brain damage (e.g. PINE).

(4) Included studies must involve preterm (gestational age g 37 weeks) or low birth
weight (birthweight< 2.5 kilograms) subjects.

(5) Studies included in this thesis must have a premature normal-ultrasound group to
compare with the premature GM/IV H group.

(6) All the included studies should be related to neurodevelopment, mental

development, school performance, behavior, or cognition.

16

(7) Sample size is an important criterion for selecting literature. The sample size of
each study group in the included papers has to be larger than 10.

(8) The age of assessment has to be equal to or older than 10 months and equal to or
younger than 18 years old.

(9) All the included studies should provide the number of surviving infants.
2. Selection with Titles, Abstracts, and Texts of Literature

The next step for literature selection is scanning titles and abstracts once an acceptable
number of abstracts have been found [52]. During the literature selection of observational
studies, investigators may find that the titles of the publications usually have clear
descriptions of the subjects, exposure, outcome, or research method. The information
provided by those titles should be examined through the inclusion and exclusion criteria. It
is really a quick and initial measure for literature selection. However, the major misgiving
during the scan of study titles is that authors and selectors may differ in how they define
concepts [58]. These differences in defining concepts are important causes for failed
database searches and selection [59]. A literature selection relying on keywords in the titles
of studies will result in the neglect of many relevant publications [58]. For instance, some
publications related to the topic of this thesis have different keywords for the ultrasound scan
results in their titles, such as “intraventricular haemorrhage”,
“periventricular-intraventricular haemorrhage”, or “cranial ultrasound abnormalities”. If a

study selector has different definitions of those key terms, many useful relevant publications

17

may be missed. Therefore, good literatures may not be selected if only the titles of the
papers are scanned. Scans of abstracts are also important and necessary.

Structured abstracts usually consist of one or more of the following components to
assist in the selection of relevant publication: background, objective, design, methods, results,
and conclusion. Nevertheless, few abstracts address all the essential information, and the
study method or design is sometimes missing [58]. Therefore, a researcher should select
literature not only through scanning titles and abstracts but also through carefully reading the
texts of publications.

IV. Literature Analysis
1. Determining the Type of Literature Review

Literature reviews can be classified into at least four types: (1) qualitative and narrative
reviews, (2) quantitative reviews of published data (usually called meta-analysis), (3)
meta-analyses with individual data (usually called pooled analysis), and (4) prospectively
planned pooled analyses [57]. Each type of literature review has its own strengths and
limitations. Clearly evaluating those strengths and limitations is an inevitable stage in
determining the type of literature review that is appropriate for the research in this thesis.

Traditional narrative reviews only present qualitative results [57]. Publication bias
[60][6l], often called “file-drawer” problems [62], is the main limitation of traditional
narrative reviews. The “file-drawer” problems arise because the results of unpublished

literature are imagined to be guzzled in investigators’ file cabinets [62] [95]. To avoid

l8

subjective judgments of the selected studies, an a-priori strict protocol for the review is
absolutely necessary. Following the indication of the explicit protocol, an extensive
overview of the included studies can be done within a short period of time and at a low cost
[57].

Meta-analysis from published data has several main limitations. First, publication bias
cannot be avoided because some epidemiological researches may be adventurously done,
and the positive results are more likely to be published. Therefore it is possible that
meta-analysis provides an overestimation of the risk estimate [57]. Second, the included
studies may differ in their design, assessment of exposure or outcome, and definition of
confounder variables. Adjusting for different confounders in separate studies is very
difficult for meta-analysis [57]. In addition, reliable results cannot be derived from
meta-analysis if the heterogeneity among included studies is high [57].

Meta-analysis with individual data can avoid some of the problems that arise with
meta-analysis from published data. This type of analysis may eliminate publication bias
because it includes unpublished data. With individual data, investigators can conduct a
statistical re-analysis and reduce some errors in the analysis [57]. Furthermore, it is
feasible for investigators to examine the effect of rare exposures with a large sample size
from several separate studies [57]. However, this type of literature review has its own
limitations. It takes a lot of time and money to do the analysis [631-[65]. In addition, the

direct cooperation between study coordinators is necessary [63]-[65].

l9

Prospectively planned pooled analysis requires a joint core protocol for data collection
and analysis between separate studies to eliminate their large heterogeneity [57]. However,
it is very time-consuming and difficult to perform. In addition, prospectively planned
pooled analysis may multiply errors in the design of single studies [57].

The literature review method used in this thesis does not belong to either meta-analysis
review or prospectively planned pooled analysis review. The main reason is that the
heterogeneity among the included studies is very high. Although most of the included
publications use cohort study design and adopt ultrasound scan to measure the exposure, the
assessed age and the measurement of outcomes among those studies are highly different.
Therefore, it is inappropriate to classify this literature review method as meta-analysis. In
addition, this thesis presents only the published data, and it excludes unpublished data or
primary data from all included studies. For this thesis, it is impractical to obtain the
original data from the coordinators of the included studies due to the lack of financial
support, scanty manpower, and limited time.

In the thesis, published data from the included studies were carefully organized and
synthesized. The relevant components of each study, including authors, publication year,
study design, sample size and assessment procedures, are carefully tabulated. Procedures
for summarizing the Evidence among studies include: (1) systematic organization and clear
presentation for the statistical data, and (2) objective judgment and logical explanation for

the results. Methodologic guidelines were adopted for this literature review [56][57].

20

By generalizing the above statements, as well as considering the materials,
characteristics, and sources of included literature in this thesis, I decided to use a review
strategy that combines qualitative and quantitative methods and called this a “systematic”
review.

2. Methods for Summarizing Evidence

A review published by Breslow et al. indicated that having clear methods for
summarizing evidence among studies is essential [66]. Pooling of odds ratios, relative risks,
and methodologic quality ranks are some of the quantitative methods adopted by most of the
review papers examined by Breslow et al. [66]. I observe that the validity of individual
studies can be appropriately assessed through the study-quality evaluation process.
Therefore, I apply two methods in this thesis to obtain high-quality literature summaries,
including evaluations using causal criteria and study-quality criteria.

(i) Evaluation with Causal Criteria

Two series of review papers, which were reviewed by Weed et al. [67], evaluated the
study of causation using causal criteria [68]. Most of those review papers used Hill’s
causal criteria [69]. Those criteria provided a practical method to judge whether causation,
which is based on the evidence that appeared in research, exists [70]- [72]. Therefore in
this thesis we adopt and administer an evaluation using Hill’s causal criteria for literature
analysis.

Hill’s nine causal criteria are strength, consistency, specificity, temporality, biological

21

plausibility, biological gradient (dose-response), experiment, coherence and analogy [69].
Parts of the nine criteria are established through the review process and are used to evaluate
the 16 selected studies in order to approach the possible existent causation.

Strength— All studies selected in the thesis adopt a cohort study design. For cohort
studies, strength is a useful criterion to evaluate the extent of association between exposure
and outcome. The association is achieved by comparing the outcome of an exposure group
with that of a non-exposure group. Strength can also be used to examine the association in
case-control studies by comparing the exposure of a disease group with that of a non-disease
group.

Causal inferences are more likely to be derived from strong associations than from
weak ones [73] [74]. However, it does not mean that weak associations cannot support the
establishment of causal association. Sometimes the strength of association is small because
of the effects from other risk factors or bias. Therefore, third variables and bias have to be
seriously estimated when researchers find weak associations in their studies.

Consistency— This criterion is used to evaluate whether the association has been
repeatedly observed by different persons, in different places, at different times, and under
different circumstances [69] [73].

Specificity— This criterion means that a specific exposure results in a specific outcome.
If specificity is established, stronger case for causation may be made [73]. In this thesis,

however, specificity cannot be established because the included studies have various

22

outcomes, even if they have the same exposure (GM/IVH).

Temporality— This criterion is established in all selected studies because they all
adopted prospective cohort study design. In addition, GM/IV H always occurs before
children’s developmental conditions can be examined. Since temporality is established in
every selected study, this criterion is not useful for judging the causation in this thesis.

Biological plausibility— This criterion means that there is a coincidence between the
observed associations and current biological knowledge [69] [73] [74]. It is useful to
consider biological plausibility because biases or confounding factors may be easily
identified by this criterion [74]. Many clinical or bio-medical studies use biological
markers to establish biological plausibility. In this thesis, however, no study adopted
biological markers or similar relevant methods to establish the biological plausibility.

Biological gradient (dose-response)— This thesis discusses the causal association
between GM/IVH and children’s neurodevelopmental and cognitive outcomes. There is no
biological gradient established in this thesis since the results of more severe brain
hemorrhage are not included and discussed.

Experiment— It is impossible to split subjects into different groups and then expose
them to a risk (GM/IVH) or non-risk (no hemorrhage) condition. This criterion is not used,
because all included studies use a non-experimental design.

Coherence— Coherence means there are no irreconcilable conflicts between causal

associations derived from study data and general epidemiological, medical, or biological

23

knowledge [69] [74]. Most included studies comply with this criterion.

Analogy— The criterion is not used in this review because of limited evidence with
other kinds of subjects (e.g. assessing the outcomes of adults’, elders’, or rats’ brain
bleeding).

In this thesis, a criterion that is established in all or none of the studies is not useful for
judging causation. As a result, it is not useful to adopt a criterion established in all selected
studies such as “temporality”; similarly, it is not useful to adopt criteria that are not
established in any of the selected studies, such as “specificity”, “biological plausibility”,
“biological gradient”, “experiment”, and “analogy”. Since only the criteria that are
established in part of the selected studies can show the difference of causation, I use strength,
consistency and coherence to evaluate the causation in this research.

(ii) Criteria for Evaluating the Quality of Each Study

Setting criteria for study-quality evaluation [56] is essential for systematic literature
reviews. In this thesis, we first consider several aspects of a study and then provide a study
quality checklist [52].

The criteria used in this thesis to evaluate study quality include: (1) whether the
follow-up rate is equal to or larger than 90%, (2) whether the assessment rate is equal to or
larger than 85%, (3) whether the mean of gestational age between the study groups is not
significantly different, (4) whether the mean of birth weight between the study groups is not

significantly different, (5) whether US brain scan was used for GM/IVH diagnosis, (6)

24

whether the study has a full-term control group, and whether the full-term control group is
well-matched, (7) whether the sample size is equal to or larger than 30, (8) whether it has
careful or standard measurements of outcomes, (9) whether key confounders are mentioned
and adjusted, (10) whether the other risk factors are mentioned and controlled, and (11)
whether it has a clear conclusion about GM/IV H.

Based on the above criteria, a study will be given one point if its answer for a criterion
is “yes”. On the other hand, it will be given zero point if its answer for a criterion is “no”.
The more points a study receives, the higher study quality it has.

In addition, the age of assessment is the twelfth criterion. A study with an older age of
assessment is regarded as a higher quality study. The age of assessment is divided into five
ranges, and the score for each range is determined as follows: (1) _2_ 10 months & g 1.5
years: 1 point; (2) >15 years & 3.3 years: 2 points; (3) >3 years & $5 years: 3 points; (4)

>5 years & g 10 years: 4 points; (5) >10 years & —_<— 18 years: 5 points.

25

Chapter Three
Results
1. Studies Identified

Two hundred to three hundred publications were reviewed. By verifying these
publications against the inclusion and exclusion criteria, almost 90% of them were excluded.
Initially, twenty-three publications were identified to be useful. However, some of them
did not meet all the inclusion criteria: three of these publications combined GM/IVH with
other types of brain damage and had no clear IVH classification; two of these publications
assessed the subjects at the age of three months; one publication had no premature normal
ultrasound group; one publication did not provide the size of the birth cohort. Therefore,
seven of the twenty-three publications were excluded. Finally, sixteen journal articles were
selected for this thesis.

11. Introduction of Birth Cohorts

All included studies adopted the cohort study design. Tables l-A and l-B present the
essential information about the birth cohort of each included study. The information
includes: study location, number of birth cohort members, admission criterion (i.e.
gestational age and birth weight), average of gestational age and average of birth weight in
each group. Table 1-B provides the information on the birth cohorts with both a premature
group and a full-term control group, while Table l-A includes studies with only premature

groups. Three studies conducted by van de Bor et al. (1988, 1993, and 2004) were from

26

the same birth cohort (Table 1-A). Two studies conducted by Ross et al. (1992; 1996) were
from the same birth cohort (Table l-B). In addition, studies conducted by Pinto-Martin
(1999) are of the same birth cohort as the study conducted by Whitaker et al. (1996; 1997)
(Table l-A). Therefore, eleven birth cohorts are represented in the sixteen selected studies.

Ten of the studies were conducted in six different cities or counties in the United States
[10] [12] [14] [75]-[81]. Of the other studies, three were conducted in the Netherlands [1]
[82] [83], two in Australia [84] [85], and one in the United Kingdom [86].

The gestational age of the premature birth cohort is not uniform but is generally less
than 34 weeks for all included studies. However, nine studies did not provide the range of
gestational age for recruiting the cohort [10] [12] [14] [75] [76] [78] [81] [85] [86]. Table
l-A shows that gestational age is significantly different between the GM/IVH group and the
premature normal ultrasound group in eight studies [1] [10] [12] [14] [76] [78] [82] [83].
Overall, the mean gestational age of the GM/IVH groups of all studies is 29.6 weeks (SD:
1.9). The mean gestational age of the premature normal-ultrasound groups of all studies is
30.7 weeks (SD: 1.9). The mean gestational age of subjects with GM/IVH is significantly
different than those with normal ultrasound (P: 0.01). This means that gestational age may
be a confounding factor for the relationship between GM/IVH and children’s developmental
outcomes. In addition, the mean gestational age of the full-term control group of all studies
is 39.4 weeks (SD: 1.1).

Birth weight is classified into three ranges: low birth weight (LBW, (2500 grams), very

27

low birth weight (VLBW, <1500 grams) and extremely very low birth weight (EVLBW,
<1000 grams). Four studies involved the LBW cohort [10] [12] [14] [77], nine studies the
VLBW cohort [l] [75] [76] [78] [81]-[83] [85] [86] and one study the EVLBW cohort [84].
Two studies did not provide the enrolling criteria of birth weight for their cohort [79][80].
From Table l-A, I observe that in four studies [10][12][l4][78], the birth weight is
significantly different between the GM/IVH group and the premature normal ultrasound
group. Overall, the mean birth weight of the GM/IVH groups of all studies is 1300.5
grams (SD: 274.2). The mean birth weight of the premature normal ultrasound groups of
all studies is 1393.] grams (SD: 282.5). The mean birth weight of subjects with GM/IVH
is not significantly different than those with normal ultrasound (P: 0.08). Furthermore, the
mean birth weight of the full-term control group of all studies is 3719.3 (SD: 349.7).
111. Measures and Classification of IVH

Appendix Table 1 summarizes the measures used and classification system of IVH in
the studies. From this table, I observe that the measurement of IV H tends to be consistent
among the included studies: fifteen studies utilized ultrasound scans to examine the infant’s
brain pathological status. But, two studies published by Papile et al. (1983) and Ment et al.
(1985) adopted both ultrasound scans and computed tomography brain scans to examine
their subjects, while one study, reported by Lowe et al. ( 1990), utilized only computed
tomography brain scans to measure the subjects’ IVH status.

Times and timeframe of IVH measurements among the included studies are not

28

uniform but are very similar. IVH status was exarrrined at least two times and at most four
times. The initial examination was conducted right after birth or right after the subjects
were admitted into the intensive care unit of hospitals. The timeframe of the last
examination was slightly different among studies; however, it was never longer than the first
two months of life.

Various terms and abbreviations were adopted by the included studies for the
classification of NH. Eight of the studies directly used Papile’s system for classifying the
IVH [1] [75] [76] [78] [81]-[84]. The other studies utilized a modified classification to
assign the subjects into the study groups. The IV H classifications are all compared in
Appendix Table 1. In general, “Grade I/II IVH”, “Grade I/II CIVH”, “Grade I/II PIV H”,
“Low-grade hemorrhages”, and “S/IVH” are uniformly called “GM/IVH” in this thesis.
IV. Follow-up and Assessment Rates in Studies

Tables 2-A and 2-B provide important statistical information about assessed children:
the age of assessment, the follow-up rate, the assessment rate, and the number of children
who successfully completed the developmental tests (i.e. number of assessment). Table
2-B summarizes information about studies that assessed children with a premature groups
and a full-term control group. In Table 2-A, studies include only premature groups.

The neurodevelopmental and cognitive outcomes were assessed at different ages in the
selected studies. Five studies had one or more ages of assessment [10] [75]-[78]. The

ages of assessments were classified into five ranges in this thesis to analyze the potential

29

heterogeneity: (1) 10 months to 18 months, (2) 18 months to 3 years, (3) 3 years to 5 years,
(4)5 years to 10 years, and (5) 10 years to 18 years. Five studies were in the range from 10
months to 18 months [75]-[78] [80]. Seven studies were in the range from 18 months to 3
years [10] [75]-[79] [82]. The range from 3 years to 5 years included three studies [83] [85]
[86]. The range from 5 years to 10 years included five studies [10] [12] [14] [81] [84].
There was one study in the range from 10 to 18 years [1]. Overall, the mean age of
assessment of all studies is 4.5 years which is larger than 3 years and is sufficient for high
study quality.

Follow-up rate and assessment rate of the selected studies were extremely different,
ranging froml9% [81] to 100% [80]. This thesis emphasizes the follow-up rates of the
premature groups, and does not focus on those of full-term control groups. This is because
most studies with a full-term control group did not recruit the full-term subjects at birth,
except the study published by Sherlock et al (2005).

Follow-up rate represents the percentage of children who were still alive in the age of
assessment and were contacted successfully. Assessment rate means the percentage of
follow-up children who accepted the given developmental tests and could complete the tests
successfully. Therefore, the assessment rate is usually lower than the follow-up rate.
Higher follow-up and assessment rates are better. In this thesis, if a selected study has a
follow-up rate higher than 90% and has an assessment rate higher than 85%, it is regarded as

a study with high quality. Based on Tables 2-A and 2-B, eight studies reported respectively

30

by Kitchen et al. (1990), van de Bor et al. (1988, 1993 , and 2004), Levene et al. (1992),
Sherlock et al. (2005), and Ross et al. (1992; 1996) met this criterion.

Overall, the mean follow-up rate and the mean assessment rate among studies are 83%
and 75%, respectively. Although the mean follow-up rate and the mean assessment rate do
not meet the criteria of high study quality, they are still acceptable and can provide some
evidence for this research.

In addition, “number of assessments” provides a clear picture of the composition of
study groups that indicate the number of children who successfully completed the various
development tests. Table 2-B shows that Lowe et al. (1990) and Ross et al. (1992; 1996)
conducted studies focused on GM/IVH but did not focus on PUVE.

V. Study Quality Evaluation

In this review, the main quality variation of the included publications is caused by
difference in follow-up rates, assessment rates, sample sizes of study groups, age of
assessment, and adjustments for key confounders and other risk factors. Any variation in
one or more factors affects the quality of the included publications.

A study whose follow-up rate and assessment rate are equal to or higher than 90% and
85% respectively is recognized as a high quality study. Table 3-A shows that half of the 16
included studies meet the former criterion. On the other hand, 10 studies meet the later
criterion; in other words, that the assessment rate is equal to or higher than 85%.

Birth weight and gestational age, combined with cranial ultrasound findings, are the

31

 

most sensitive predictors of sensory, cognition, language, psychomotor and academic
performance in early school-aged children as well as in younger children [87]. As a result,
if there is a significant difierence in mean birth weight and gestational age between
premature children with GM/IVH and those without GM/IV H, the birth weight and the
gestational age may confound children’s developmental outcomes. Therefore, a study is
regarded as high quality if its enrolled infants, with and without GM/IVH, had no significant
differences in mean birth weight and mean gestational age. Based on Table 3-A, I observe
that seven studies had no significant difference in mean gestational age and twelve studies
had no significant difference in mean birth weight.

Ultrasound brain scan is a reliable and standard technology for examining the presence
of brain hemorrhage. The study reported by Lowe et al. did not utilize ultrasound for brain
screening and thus does not meet the criterion of high-quality study (Table 3-A).

Five studies reported by Lowe et al. (1990), Levene et al. (1992), Sherlock et al. (2005),
and Ross et al. (1992; 1996) had matched full-term control groups (Table 3—A); therefore,
they are regarded as high-quality studies. Those full-term control groups matched the
premature groups in different dimensions such as race, age, gender, and parents’
socio-economic status (T able l-B). Through a careful match, the effects of those variables
can be controlled. Although premature normal ultrasound group already plays the role of
control group, a study with a matched full-term control group still provides additional

valuable advantages. With a full-term control group, the effects of birth weight and

32

 

gestational age on outcomes can be investigated. On the other hand, studies without a
full-term control group can only investigate the effects of GM/IVH on the outcomes. In
addition, the external validity of a research can be increased by the comparison between the
full-term control group and the premature group.

A larger sample size can lead to a more robust result. Based on the Central Limit
Theorem [88], a distribution with a larger sample size is closer to a normal distribution.
Usually, a sample size larger than 30 is assumed to result in an approximately normal
sampling distribution. Table 3-A shows that three studies reported by Sostek et al. [77],
Lowe et al. [81] and Ross et al. [79] do not meet the criterion of high study quality because
some of the groups in those studies have sample sizes smaller than 30 (Tables 2-A and 2-B).
If a sample size is very small, confidence interval will become so large that the Central Limit
Theorem cannot be applied. In this case, no difference will be found between the GM/IVH
group and the normal ultrasound group. Ross at al. obtained some significant findings
when they administered the special cognitive tests (i.e. invisible displacement tasks and
object discrimination reversal tasks), but Sostek et al. and Lowe et al. did not find any
significant results. This rrright be due to their different sample sizes of groups: the group
sizes in studies of Ross et al. are close to 30 (N=27 and 28), but those in studies of Sostek et
al. (N=23 and 20) and Lowe et al. (N=22 and 11) are much smaller than 30. Obviously,

further evaluation of the results from these studies is necessary because of their small sample

sizes.

33

Several different measures were administered in the sixteen included studies, reporting
a variety of developmental outcomes (Appendix Table 2). Attempting to design related
studies with a uniform measurement is often challenging because human development and
cognition are complex and comprehensive processes. As a result, they cannot be fully
studied by evaluating any single outcome. Therefore, in this review, any study I
administering careful or standard measurements of outcomes is recognized as a hi gh-quality

study. All selected studies met this criterion (Table 3-B).

 

To explore the long-term neurodevelopmental and cognitive outcomes, a study with the
age of assessment larger than 3 years is regarded as a high-quality study in this review.
Three points were given if a study had the age of assessment between 3 and 5 years. Four
points were given if a study had the age of assessment between 5 and 10 years. In addition,
five points were given if a study had the age of assessment between 5 and 18 years. Table
3-B shows that 9 studies receive at least three points due to the criterion of the age of
assessment.

Other risk factors and key confounders have to be either controlled while the
researchers design a study or adjusted while the researchers do the statistical analysis [89].
Confounding may cause errors when estimating the association between a typical exposure
(e.g. GM/IVH) and its outcome (e.g. cerebral palsy) [89]. Other risk factors, which may be
precursors or mediators, also need to be recognized in order to estimate their influence on

the outcomes. Six studies in Table 3-B identified and adjusted for the key confounders

34

such as gestational age, birth weight, Apgar score at 5 minutes < 7, small for gestational age,
respiratory distress syndrome, seizures of all origins, apnea (more than 15 sec. and/ or
bradycardia), hyperbilirubinemia (maximum neonatal serum total bilirubin > 200 a mol/l)
and bronchopulmonary dysplasia. On the other hand, nine studies mentioned and adjusted
other risk factors such as social disadvantage, gender, outbom, assisted ventilation and
multiple gestations (Table 3-B). These studies meet the study quality criteria.

Table 3-B reveals that four selected studies emphasized the outcomes caused by severe
brain damage (e. g. ventricular enlargement and/or parenchymal lesion) and did not make a
clear conclusion for the effects of GM/IVH. Therefore, these studies did not meet the
criterion of high study quality.

VI. Ranking of Importance for Selected Studies

Every study was evaluated by the study quality criteria (Tables 3-A, 3-B). When a
study met any of the first eleven criteria, it would obtain one point. In addition, a study
with an older age of assessment would receive more points. After the evaluation, a study
receives a total score that represents its quality (Table 3-B). Table 4 is created to provide
the ranking of importance for selected studies by combining the evaluation of causal
inference with that of study quality. The study of van de Bor et al. (2004), which is on the
first place, is the most important study. On the other hand, the study of Sostek et al. (1987),
is on the last place, showing that it is the least important study (Table 4). Studies of

Kitchen et al. (1990), Levene et al. (1992), Sherlock et al. (2005), and Ross et al. (1992;

35

 

FF? {7" 1t 1 I P. ‘_ '

1996) tied as fourth place winners (1’ able 4). Eight studies, listed from the first to the

fourth places, were regarded as important and reliable studies.

VII. Frequencies and Odds Ratios of Developmental
Abnormalities

In Tables 5-A and 5-B, the data used to calculate the frequencies of abnormal or special
cognitive outcomes were abstracted from twelve of the included studies, and the percentages
of outcomes were calculated manually. In Table 5-A, the frequencies of abnormal

outcomes in the GM/IVH groups and the normal ultrasound groups are compared. Table

 

5-B compares the GM/IVH groups with not only the normal ultrasound groups but also the
full-term control groups.

In Table 5-A, eight studies [1] [12] [78] [81]-[85] had one or more abnormal outcomes
with higher frequencies in the GM/IVH groups than in the normal ultrasound groups. The
frequencies of abnormal outcomes in the GM/IVH groups were at least twice as high than in
the normal ultrasound groups in the eight studies. Take the study reported by Kitchen
(1990) for instance, the prevalence rate of cerebral palsy (CP) in the GM/IVH group is
18.18%, and in the normal ultrasound group is 3.13% (Table 8-A). As a result, the CP rate
in the GM/IVH group is six times higher than in the normal ultrasound group. Performing
the same comparison for the study published by Ross et al. in Table 5-B, I observe that the
abnormal rate of children who habituated in the GM/IVH group (46.67%) is significantly

lower than the normal ultrasound group (96.67%), as well as the full-term control group

36

(90.00%). In addition, the frequency of failed reversal trials in the GM/IVH group
(73.33%) is significantly higher than not only the normal ultrasound group (56.67%) but also
the full-term control group ( 16.67%). Based on the comparison of frequencies, I may infer
that there might be significant differences in the outcomes between the GM/IVH groups and
the normal ultrasound groups.

Odds ratios (OR), 95% confidence intervals (CI) and p-values were provided by the
authors in five studies [1] [12] [14] [82] [83]. In the rest of the studies, those statistic
results were calculated using the SAS program. Significant differences between the
GM/IVH groups and the normal ultrasound groups existed in six studies [1] [12] [78] [82]
[83] [85] for twelve abnormal outcomes in Table 5-A. Children with GM/IVH had more
disability than those with normal ultrasound in two studies [83] [85]. In 1988 and 2004,
two studies reported by Van de Bor et al. showed that children with GM/IVH had more
handicap than those without GM/IVH at the age of 2 (OR=2.10; 95%CI=1.30-3.30; P<0.01)
and 5 (OR=2.70; 95%CI=1.21-6.08; P=0.01), respectively. Based on the odds ratio and
95% CI in the study of Kitchen et al., cerebral palsy prevalence (OR=6.89;
95%CI=1.62-29.39; P=0.003) was regarded as significantly different between the GM/IVH
group and the normal ultrasound group. Mental retardation (OR=5.00; 95%CI=1.50-16.80;
P<0.01) was significantly elevated by GM/IVH in the study of Whitaker et al. (1996), and
abnormal neurodevelopment (OR=4.90; 95%CI=1.96-12.27; P=0.0003) was significantly

elevated by GM/IVH in the study of Hanigan et al. (1991). In the study of van de Bor et al.,

37

which was reported in 2004, the risk that children needed special education at 5 (OR=3.04;
95%CI=1.24-7.43; P=0.01), 9 (OR=2.72; 95%CI=1.26-5.88; P=0.009), and 14 (OR=2.10;
95%CI=1.01—4.35; P<0.05) years old were all significantly high.

In Table 5-B, the study reported by Ross et al. showed that there were significant
differences for children who habituated between the GM/IVH group and the full-term
control group (OR=0.1; 95%CI=0.02-0.39; P=0.0003). This means the GM/IVH group
(N=14) had significantly smaller number of children who habituated than the full-term
control group (N=27).

The significantly different results in Tables 5-A and SB were consistent with the above
comparison of frequencies. On the other hand, studies reported by Papile et al. (1983),
Lowe et al. (1990) and Sherlock et al. (2005) showed no significant difference in the

abnormal outcomes between the GM/IVH groups and the normal ultrasound groups.

VIII. Averages and Standard Deviations of Developmental

Tests

Tables 4—A and 4-B listed the means and standard deviations (SD) of various
neurodevelopmental or cognitive tests. In Table 3-A, Sherlock et al. (2005) found that the
normal ultrasound group had significantly lower scores than the GM/IVH group in Wide
Range Achievements Test (WRAT3). The p—values for reading, spelling, and arithmetic
were 0.004, 0.048, and 0.026, respectively. The results derived from WRAT3, which were

reported by Sherlock et al., do not conform to common anticipation. The rest of the studies

38

in Tables 4-A showed that there were no significant differences between the GM/IVH groups
and the normal ultrasound groups.

Table 3-B lists the means and standard deviations (SD) of various cognitive tests in
studies with a full-term control group. Ross et al. (1992) reported that the GM/IVH group
had significantly lower scores on Bayley Mental Developmental Index than both the
premature normal ultrasound group and the full-term control group. In addition, Ross
reported that there were significant differences between not only the GM/IVH group and the
premature normal ultrasound group (P<0.03) but also the GM/IVH group and the full-term
control group (P<0.05) on the habituation test. Ross et al. (1996) found that the GM/IV H
group had significantly lower scores than both the premature normal ultrasound group
(P<0.01) and the full-term control group (P<0.01) for the invisible displacement trials score
of invisible displacement tasks. On the other hand, they reported that the GM/IVH group
had significantly lower scores only than the full-term control group for the systematic search
trials score of invisible displacement tasks (P<0.05). For the object discrimination reversal
task in the same study, the CW group had significantly lower scores than both the
premature normal ultrasound group (P<0.05) and the full-term control group (P<0.05). The
study published by Lowe et al. (1990) did not show any significance for all of the cognitive
tests between the GM/IVH group and the premature normal ultrasound group, as well as the

GM/IVH group and the full-term control group.

IX. Overall Evaluation and Interpretation

39

In order to overcome the limitation due to lack of specific outcomes, I aggregated
various outcome variables that appeared in the selected studies into three categories:
neurodevelopment, general cognition, and special cognition. Because of the high
heterogeneity among the included studies, it is infeasible to utilize meta-analysis.

Therefore I calculated the weighted odds ratios (Tables 7-A and 7-B) to find a trend of the
effect of GM/IVH on children’s development and cognition as follows. Firstly, a
sum-of-product is generated by adding each odds ratio of the outcome variable in a specific
category multiplied by its sample size. Secondly, the sum-of-product is divided by the total
sample sizes of the outcome variables, which produces the weighted odds ratio. In addition,
the median and the range of the ORs were given, because the median of ORs can reduce the
effect of extreme values of ORs, and the range of the ORs can be used to estimate the
difference among the ORs. Furthermore, the weighted means of developmental tests were
provided through similar methods (Tables 7-C, D, E, and F).

Eleven outcome variables (e. g. handicap, disability, and cereme palsy etc.) that
belonged to neurodevelopment were aggregated (Table 7-A). The median of ORs is 2.1
and the range of ORs is 5.96. In addition, weighted odds ratio is 2.53. I observed that the
weighted OR slightly differs from the median of ORs. This means the weighted OR is
slightly influenced by the extreme value of ORs. That the weighted odds ratio is 2.53
means children with GM/IVH have a risk of abnormal neurodevelopment 2.53 times higher

than those without.

40

To evaluate the effect of GM/IVH on children’s general cognitive outcomes, Table 7-B
shows that the median of ORs is 1.4 and the range of ORs is 4.23. In addition, the
weighted odds ratio is 2.49. I observe that the weighted OR differs from the median ORs,
suggesting that the weighted OR is influenced by the extreme value of ORs. Based on the
weighted OR, children with GM/IVH have a risk of abnormal general cognition 2.49 times
higher than those without.

Tables 7-C, D, E, and F present un-weighted means, weighted means, and differences
in weighted means. With the differences in weighted means, the effect size of GM/IV H
can be inferred. In Table 7-C, the weighted means of the GM/IVH group and the normal
ultrasound group in children’s neurodevelopment are 97.4 and 100.7 respectively. Table
7-D shows that the effect size of GM/IVH on children’s general cognition and the difference
in means is between 1.2 and 2.0. In addition, Table 7-F shows that the effect size of
GM/IVH on children’s reasoning and memory and its difference in means is 1.98. The
results show that it is possible that GM/IVH has some unfavorable influence on children’s
neurodevelopment, general cognition, and reasoning and memory. However, Table 7-E
provides an opposite result that children with GM/IVH may obtain higher scores than those

without. The difference in weighted means of beneficial effect size of GM/IVH is 2.42.

41

Chapter Four
Discussion

I. Evaluation of Causal Criteria

By applying Hill’s causal criteria to evaluate the selected studies, I found that five of
the criteria (i.e. specificity, biological plausibility, biological gradient, experiment, and
analogy) were not established and one (i.e. temporality) was established in all studies. The
other three criteria, strength, consistency and coherence, were thus adopted to evaluate the
causal nature of the association between GM/IVH and developmental outcomes in the
selected studies. Table 8 shows the evaluation of causal criteria for each included study.

Strength— “Strength” is established by eight of the selected studies [1] [ l2] [78]-[80]
[82] [83] [85], and it means that children with GM/IVH have higher risk than those with
normal ultrasound for the abnormal outcomes. The results abstracted from the studies of
van de Bor et al in 1988, 1993 and 2004 reveal that the odds ratio from handicap in children
with GM/IVH is twofold (OR=2.1) to threefold (OR=2.7) the odds ratio in children with
normal ultrasound (Table 8-A). In Kitchen’s study, the odds ratio from cerebral palsy in
children with GM/IVH is over six times (OR: 6.89) the odds ratio in children with normal
ultrasound (Table 8-A), while that from disability is fourfold (OR: 4.04). From the
comparison between the GM/IVH group and the premature normal ultrasound group (Table
8-A), I found that the odds ratios from mental retardation (OR=5.0) and abnormal

neurodevelopment (OR=4.9) are both fivefold the normal group, as reported in studies of

42

Whitaker [l2] and Hanigan et a1 [78]. In 1992 and 1996, Ross et al. conducted studies to
evaluate special cognition (i.e. habituation test, AB task, and invisible displacement task) for
infants, and they found that the “strength” existed when the GM/IVH group was compared
with the normal ultrasound group (Tables 3-B and 4-B).

Based on Cohen's scale [94] and Hopkins’s scale [96], an OR that ranges from 1.5 to
3.5 is regarded as having a “small” effect size, 3.5 to 9.0 has a “moderate” effect size, and
9.0 to 32 has a “large” effect size. Therefore, the included studies, whose strength criterion
is established, have the effect size between small and moderate levels.

Strength was established among half of the 16 included studies, and it may support the
causation between GM/IVH and developmental outcomes. Because these selected studies
lack sufficient evaluations for specific outcomes, their various outcomes were aggregated
into three categories: neurodevelopment, general cognition and special cognition. By
evaluating the pooled outcome variables in each category, the causal criterion of strength is
moderately established, and it is necessary to obtain more evidence from related studies to
further strengthen causation.

Consistency— Table 8 shows that children’s disability reported by Kitchen et al. [85]
was consistent with that reported by van de Bor et al. [83] (Table 8-A). In addition, studies
performed by van de Bor et al. (1988; 2004) also show consistent results (Tables 3-A and 5).
Furthermore, the results of children’s neurodevelopment reported in the study of Hanigan et

al. [78] agreed with that of van de Bor et al. [82] (Tables 3-A and 5). This observation

43

shows that the outcomes of the GM/IVH groups are significantly different from those of the
normal ultrasound groups. Due to the lack of sufficient evaluations for specific outcomes,
the causal criterion of consistency was built in the selected studies only to a certain extent.

Cohemnw— While most included studies revealed that either GM/IVH is not
associated with children’s developmental outcomes or the GM/IVH group has a higher risk
for developmental outcomes than the normal ultrasound group (Table 8), Sherlock et al.
(2005) found that the normal ultrasound group had significantly lower scores than the
GM/IVH group in Wide Range Achievements Test (WRAT3) (Table 3-A). This result did
not conform to the common anticipation. However, fifteen of the sixteen selected studies
produced results which conformed to the general epidemiological and medical knowledge.
Therefore, the causation of coherence was strongly established among these fifteen studies.

Overall, the studies conducted by Kitchen [85], van de Bor [82] [83] and Hanigan [78]
et al. included the most evidence to support the causation because “strength”, “consistency”
and “coherence” were all established (Table 8). Therefore, the causation between GM/IVH
and developmental outcomes in these four studies is stronger and more credible than the
other studies.
II. Limitations

That it is difficult to obtain a larger number of publications for the topic of this thesis is
the first limitation. Previous researchers did not give enough attention to the effect of

GM/IVH on children’s development and cognition. Therefore, the lack of abundant

44

 

literature undermines the establishment of stronger inference for the association between
GM/IV H and neurodevelopment, as well as the association between GM/IVH and cognition.

Secondly, publication bias, one type of selection bias, is a major limitation for
systematic reviews [90]. This is because studies with statistically significant results are
more likely to be published than those without significant results [90]. This review
included relevant published studies but could not include unpublished studies; therefore, it is
difficult to minimize the publication bias. In addition, language bias, another type of
selection bias, is manifested because all included studies in this research were published in
English [91]. Selection bias, including publication bias and language bias, may threat the
internal validity of this research.

Neurodevelopmental and cognitive outcomes, which were investigated in this research,
exhibit various and comprehensive conditions. Based on previous studies, GM/IV H did
not have significant influences on any typical outcome. As a result, most researchers
selected some of the outcomes that they were interested in. Therefore, it is hard to pool a
large number of publications for a typical outcome, such as intelligence deficiency or
cerebral palsy.

A similar limitation for this review is that it is not easy to collect studies approaching
long-term developments such as school performances or learning abilities. In this review,
only one selected study assessed children at the age of 14. Therefore, evidence for the need

of special education is very limited because of insufficient relevant studies.

45

The heterogeneity among the selected studies is a generic problem for this review.
The differences in age of assessment, follow-up rate, assessment rate, and various
developmental outcomes are the main causes of heterogeneity. In addition, due to various
measurements used in the selected studies, this review lacks efficient and practicable
methods to evaluate the validity and reliability of their outcomes. Similarly, various
methods of statistic analyses in the selected studies lead to the heterogeneity that threatened
the comparability. Because of heterogeneity, I tend to use descriptive statistic methods to
analyze the results of the included studies in this thesis.

III. Strengths

This research is the first systematic review to study the effects of GM/IV H on
neurodevelopmental and cognitive outcomes. It is not only a beginning for studying
related issues but also a valuable guide for further investigations. \Vrth this systematic
review, the neglected links between GM/IVH and children’s development were explored.
Consequently, more reliable evidence of association was established in this thesis.

Because this is a systematic review, it provides an improved reflection of reality [92].
Applying explicit scientific principles such as quantitative methods, selection criteria, causal
criteria and criteria of study quality, the random and systematic errors of bias is likely to be
reduced [93]. Therefore, this thesis not only avoids the disadvantages of traditional
reviews but also provides more accurate inferences. Several high quality studies were

pooled in this thesis, and highly credible interpretations were presented.

46

In order to overcome the limitation that no specific outcome was evaluated by sufficient
studies, I tried to pool the various outcomes into larger categories. With this technique, a
trend was found and possible inferences were derived.

IV. Conclusions

After evaluating the selected studies with quantitative methods, causal criteria and
criteria of study quality, 1 summarize the results as follows.

Firstly, GM/IV H does seem to slightly impair children’s special cognitive outcomes,
especially special mental performances, such as infants’ attention for an object, infants’
development of memory for the location of an object, and children’s reasoning and memory.

Secondly, general cognitive tests (e. g. the Full-Scale IQ from the Wechsler Intelligence
Scales for Children- Third Edition and the Wechsler Preschool and Primary Scales of
Intelligence) show that there is no significant difference between children who were with
GM/IVH and those without it in global cognitive performances. However, GM/IVH seems
to have certain effects on children’s general cognition. It still needs more evidence to
provide stronger support.

Thirdly, GM/IVH in preterm infants leads to an increased risk of adversely affected
neurodevelopmental performance, including CP, disability, and handicap. In addition, the
effect of GM/IVH on school performances of children or adolescents is unfavorable.

Finally, a conclusion is made that the older the age of assessment, the better the evaluation.

V. Further Discussion

47

I observed that the rates of handicap and/or abnormal neurodevelopment differ between
the GM/IV H group and the normal ultrasound group, but there is usually no difference in
general cognitive scores, such as the mean IQ. In order to explain this phenomenon, I list
some possible reasons as follows:

12 studies examined both outcomes except the studies of Papile et al. (1983), van de
Bor et al. (1988; 1993), and Ment et al. (1985). Therefore, this phenomenon is not due to
that the outcomes were examined by different studies.

Ross et al., in 1992 and 1996, conducted the studies that excluded “infants with
congenital malformations, moderate to severe neurosensory deficits, or possible exposure to
substance abuse in utem”. They found the effect of GM/IVH on children’s cognition is
significantly unfavorable. The study of Pinto-Martin et al. (1999) excluded children with
major motor or cognitive disability, and they did not find any cognitive outcomes that differ
between children with GM/IVH and those without. Other studies that examined both
outcomes did not provide the exclusion criteria for enrolling children. Therefore, it is hard
to know whether the following two explanations caused this phenomenon: (1) children with
developmental abnormalities were excluded in cognitive scores, and (2) the number of
children with abnormal development was too small to affect the overall cognitive mean.

A possible supposition is that GM/IV H may influence various neurodevelopmental
functions. In addition, the assessment rate may affect the reliability of a study: it is possible

that the loss-of-follow-up children had poor cognitive performance or abnormal

48

neurodevelopment. To find the correct answer to this question, more relevant researches

need to be conducted.

49

TABLES

Table 1-A(1 of 2): The Numbers, Gestational Ages, and Birth Weights of
Baseline Cohorts in Studies without a Full-Term Control Group

 

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Year of Study Location Number of Number of who Admitted
Publication Being Admitted survived and can Criterion:
Cohort be contacted. GA (wks)
Papile! 1983 Albuquerque, 260 232 N/A
N.M., U.S.A.
Ment/ 1985 New Haven, CT, 218 164 N/A
U.S.A.
Kitchen/ 1990 Melbourne, 227 154 N/A
Parkville,
Australia.
Sostek/ 1987 Washington, DC, N/A 113 § 34
USA.
van de Bor/ 1988 Netherlands 484 294 < 32
van de Bor/ 1993 Netherlands 484 304 < 32
van de Bor/ 2004 Netherlands 484 304 < 32
Whitaker/ 1996 New Jersey, 1105 898 N/A
U.S.A.
Whitaker/ 1997 New Jersey, 1105 898 N/A
U.S.A.
Pinto-Martin! New Jersey, 1105 Age 2: 887 N/A
1999 USA. Age 6: 873
Age 9: 868
Hanigan/ 1991 USA. 459 336 N/A

 

 

 

 

 

GA: Gestational Age; BW: Birth Weight; SD: Standard Deviation; wks: weeks; g: grams;
N/A: Not Available; *Normal: premature normal ultrasound group; GM/IVH: Germinal
Matrix and/or lntraventricular Hemorrhage; 1‘: Significant; 1: Slightly significant.

50

 

Table LA (2 of 2): The Numbers, Gestational Ages, and Birth Weights of
Baseline Cohorts in Studies without a Full-Term Control Grorm

 

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Year Admitted GA of Groups: BW of Groups:
Criterion: MeaniSD (wks) MeaniSD (g)
BW (g)
Papile! 1983 < 1501 *Normal: 30.2 *Normal: 1180
GM/ IVH: 29.3 CM] IVH: 1086
Ment/ 1985 g 1250 *Normal: 29.7 i 2.3 *Normal: 1020 i 165
All IVH: 28.5 i 1.7 “t AllIVH: 1007 i 154
Kitchen/ 1990 500 to 1500 All preterm: 29.3 i 1.9 All preterm: 1179 i 214
Sostek/1987 <1750 *Normal: 29.9 i 2.14 *Normal: 1262 i 291
GM/ IVH:29.2 i 2.21 3'; GM/IVH: 1193 i‘ 306
van de Bor/ 1988 <1500 *Normal: 29.7 i 1.5 *Normal: 1298 i 319
GM/IVH: 29.2 i 1.5 i GM/IVH: 1261 i 278
van de Bor/ 1993 < 1500 *Normal: 29.7 i’ 1.5 *Normal: 1298 i 319
GM/IVH: 29.2 i 1.5 'i GM/IVH: 1261 i 278
van de Bor/ 2004 <1500 *Normal: 29.4 i 1.5 *Normal: 1297 i 329
GM/IVH: 28.8 i 1.5 i‘ GM/IVH: 1258 i 291
Whitaker/ 1996 501 to 2000 *Normal: 32.1 i 3.0 *Norrnal: 1529.5 i 342.9
GM/IVH: 30.1 i 3.0 i GM/IVH: 1339.2 ’1 367.0 f
Whitaker/ 1997 501 to 2000 *Normal: 32.1 i 3.0 *Normal: 1529.5 i 342.9
GM/IVH: 30.1 i 3.0 i GM/IVH: 1339.2 i 367.0 T
Pinto-Martin/ 501 to 2000 *Normal: 32.1 i 3.0 *Normal: 1529.5 i 342.9
1999 GM/IVH: 30.1 i 3.0 t GM/IVH: 1339.2 i 367.0 ”t
Hanigan! 1991 < 1500 *Normal: 30.5 i 0.17 *Normal: 1208.95 i 16.1

 

 

GM/IVH: 29.2 :1: 0.34 '3‘

 

GM/IVH: 1150.72 i 34.8 T

 

GA: Gestational Age; BW: Birth Weight; SD: Standard Deviation; wks: weeks; g: grams;
N/A: Not Available; *Normal: premature normal ultrasound group; GM/IVH: Germinal
Matrix and/or lntraventricular Hemorrhage; T: Significant; 1:: Slightly significant.

51

 

Table l-B (l of 2): The Numbers, Gestational Ages, and Birth Weights of
Baseline Cohorts in Studies with a Full-Term Control Group

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Study Location Number of Being Number of who Admitted Admitted
Year Admitted Cohort survived and can Criterion: Criterion:
be contacted. GA (wks) BW (g)
Lowe/ Albuquerque, FTControls: N/A FT Controls:N/A FT Controls:N/A FT Cont: N/A
1990 N.M., U.S.A. Preterm: 260 Preterm: 198 PretermzN/A Pretermz<1501
Levene! U.K. FT Controls: N/A FT Controls:N/A FTControlszN/A FTCont: N/A
1992 Preterm: 200 Preterm: 155 PretermzN/A Preterm:<1501
Sherlock] Victoria, FTControls: 265 FT Controls: 262 FT Controls:N/A FT Contz>2499
2005 Australia. Preterm: 568 Preterm: 298 Pretermz<28 Pretermz<1000
Ross/ New York, FT Controls: 30 FT Controls: 30 FT Controls:N/A FT Cont: N/A
1992 U.S.A. Preterm: 60 Preterm: 60 Pretermz28~32 Preterm: N/A
Ross/ New York, FT Controls: 30 FT Controls: 30 FT Controls:N/A FT Cont: N/A
1996 U.S.A. Preterm: 60 Preterm: 60 Pretermz28~32 PretermzN/A

 

GA: Gestational Age; BW: Birth Weight; wks: weeks; g: grams; SD: Standard Deviation;
N/A: Not Available; SES: Socio-Economic Status; FT: Full Term; Cont: Controls; *Normal:
premature normal-ultrasound group; GM/IVH: Germinal Matrix and/or lntraventricular
Hemorrhage.

52

 

Table LB (2 of 2): The Numbers, Gestational Ages, and Birth Weights of
Baseline Cohorts in Studies with a Full-Term Control Group

 

 

 

 

 

 

 

Author/ GA of Groups: BW of Groups: Match
Year MeaniSD (wks) MeaniSD (g)
Lowe! 1990 FT Controls: N/A FTControls: N/A Race; age; gender; SES.
*Nomral: 30.4 *Normal: 1157
GM/ IVH: 30.5 GM/ IVH: 1137
Levene! 1992 FT Controls: N/A FT Controls: N/A Age; class; school.
*Norrnal: N/A *Normal: N/A
GM/ IV H: N/A GM/ IVH: N/A
Sherlock! 2005 FT Controls: 39.3 $1.4 FTControls: 3407 i 443 Gender; the mother’s
All preterm: 26.7 i 2 All preterm: 883 i 162 country of birth; health
insurance status.
Ross/ 1992 FT Controls: 39.4 i0.9 FT Controls: 3875.2 i 321 GA; birthweight (within
*Normal: 30.5 i 1.5 *Normal: 1494.7 i 256 100 g); days of assisted
GM! IVH: 30.2 i 1.4 GM/ IVH: 1431.1 1’ 226 ventilation (within 1
week); SES; gender“, race.
Ross/ 1996 FT Controls: 39.4 10.9 FTControls: 3875.8 i 285 GA; birthweight (within

 

*Norrnal: 30.5 i 1.5
GM/ IVH: 30.2 i 1.4

 

*Normal: 1483.6 i 266
GM/ IVH: 1432.8 i 227

 

100 g); days of assisted
ventilation (within 1

week); SES; gender", race.

 

GA: Gestational Age; BW: Birth Weight; wks: weeks; g: grams; SD: Standard Deviation;
N/A: Not Available; SES: Socio-Economic Status; FT: Full Term; Cont: Controls; *Normal:
premature normal-ultrasound group; GM/IVH: Germinal Matrix and/or lntraventricular

Hemorrhage.

53

 

Table 2-A (1 of 2): Follow-up Rates, Assessment Rates, and Number of
Assessment in Studies without a Full-Term Control Group

 

 

 

 

Author/ Age of Assessment Follow-up Rate (%) Assessment Rate (%)
Year
Papile/ 1983 12 or 24 months 198/ 232: 85.34% 197/ 232: 84.91%
Ment/ 1985 12, 18, and 30 142/ 164: 86.59% 142/ 164: 86.59%
months
Kitchen/ 1990 5 years 139/ 154: 90.26% 135/ 154: 87.66%

 

Sostek/ 1987

Mean age at 12.6
months and 21.5

At 12.6 months
89/ 113: 78.76%

At 12.6 months
89/ 113: 78.76%

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

months
At 21.5 months At 21.5 months
86l113=76.11% 86/113=76.11%
van de Bor/ 1988 2 years 294/ 294: 100.00% 294/ 294: 100.00%
van de Bor/ 1993 5 years 304/ 304: 100.00% 301/ 304: 99.01%
van de Bor/ 2004 14 years 278/ 304: 91.45% 278/ 304: 91 .45%
Whitaker/ 1996 6 years 685/ 898: 76.28% 597/ 898: 66.48%
Whitaker/ 1997 6 years 685/ 898: 76.28% 564/ 898: 62.81%
Pinto-Martin/ 1999 Motor Ability Motor Ability
2 years 777/ 887: 87.59% 462/ 887: 52.09%
6 years 685/ 873: 78.47% 538/ 873: 61.63%
9 years 658/ 868: 75.81% 487/ 868: 56.11%
General Cognition General Cognition
2 years 777/ 887: 87.59% 611/ 887: 68.88%
6 years 685/ 873: 78.47% 538/ 873: 61.63%
9 years 658/ 868: 75.81 % 488/ 868: 56.22%
Hanigan/ 1991 18 months and 219/ 336: 65.18% 216/ 336: 64.29%
3 years

 

N: Number; *Normal: premature normal ultrasound group; GM/IVH: Genninal Matrix
and/or lntraventricular Hemorrhage; VIE/PL: Ventricular Enlargement and/or Parenchymal
Lesion; N/A: Not Available.

54

 

Table 2-A (2 of 2): Follow-up Rates, Assessment Rates, and Number of
Assessment in Studies without a Full-Term Control Group

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Number of Assessment
Year Total N *Nonnal GM/ IVH VE/ PL

Papile! 1983 197 115 51 31

Ment/ 1985 142 94 62 8

Kitchen/ 1990 135 96 33 6
Sostek/ 1987

89 36 23 30

86 36 20 30

van de Bor/ 1988 294 225 52 17

van de Bor/ 1993 301 234 50 17

van de Bor/ 2004 278 216 45 17

Whitaker/ 1996 597 468 83 46

Whitaker/ 1997 564 454 78 32

Pinto-Martin/ 1999

462 382 61 19

538 449 70 19

487 402 65 20

611 504 83 24

538 449 70 19

488 403 65 20

Hanigan/ 1991 216 153 47 16

 

 

 

 

 

 

N: Number; *Normal: premature normal ultrasound group; GM/IVH: Germinal Matrix

and/or lntraventricular Hemorrhage; VIE/PL: Ventricular Enlargement and/or Parenchymal
Lesion; N/A: Not Available.

55

Table 2-B (1 of 2): Follow-up Rates, Assessment Rates, and Number of
Assessment in Studies with a Full-Term Control Group

 

 

 

 

 

 

 

 

 

 

Author/ Year Age of Follow-up Rate (%) Assessment Rate (%)
Assessment
Lowe/ 1990 From 5 to 6 FT Controls: N/A FT Controls: N/A
years Preterm: 38/ 198: 19.19% Preterm: 38/ 198: 19.19%
Levene/ 1992 5 years FTControls: N/A FT Controls: N/A
Preterm: 152! 155: 98.06% Preterm: 140/ 155: 90.32%
Sherlock/ 8 years FT Controls: 221/262=84.35% FT Controls: 221/262: 84.35%
2005 Preterm: 275/ 298: 92.28% Preterm: 270/ 298: 90.60%
Ross/ 1992 10 months FTControls: 30/ 30: 100% FT Controls: 30/ 30: 100%
Preterm: 60/ 60: 100% Preterm: 60/ 60: 100%
Ross/ 1996 2 years FT Controls: 27/ 30: 90% FT Controls: 27/ 30: 90%
Preterm: 55/ 60: 91.67% Preterm: 55/ 60: 91.67%

 

FT: Full Term; N/A: Not Available; N: Number; *Normal: premature normal ultrasound
group; GM/IVH: Gernrinal Matrix and/or lntraventricular Hemorrhage; VIE/PL: Ventricular
Enlargement and/or Parenchymal Lesion; T: the sum of prolonged flare (PF) group and
GM/IVH with PF group.

56

 

Table 2-B (2 of 2): Follow-up Rates, Assessment Rates, and Number of
Assessment in Studies with a Full-Term Control Group

 

 

 

 

 

 

 

 

 

Author/ Year Number of Assessment
Total N FT Controls *Normal GM/ IVH VE/ PL
Lowe/ 1990 60 22 27 11 N/A
Levene/ 1992 284 144 64 54 22T
Sherlock/ 2005 491 221 180 72 18
Ross/ 1992 90 30 30 30 N/A
Ross/ 1996 82 27 28 27 N/A

 

 

 

 

 

 

FT: Full Term; N/A: Not Available; N: Number; *Normal: premature normal ultrasound
group; GM/IVH: Germinal Matrix and/or lntraventricular Hemorrhage; VFJPL: Ventricular
Enlargement and/or Parenchymal Lesion; T: the sum of prolonged flare (PF) group and
GM/IVH with PF group.

57

Table 3-A (1 of 2): Study Quality Evaluation

 

Author/ Year Follow-up rate Assessment rate Ms of GA are Ms of BW are
; 90% g 85% NS. NS.

 

Papile/ 1983

O

1

 

Ment/ 1985

 

Kitchen/ 1990

 

Sostek/ 1987

 

van de Bor/ 1988

 

van de Bor/ 1993

 

van de Bor/ 2004

 

Whitaker/ 1996

 

Whitaker/ 1997

 

ooo—-—-—-o-—o
ooo——~—o-——

cooooco—o—
ooo——-—-—-—-—-—-

Pinto-Martin/
1 999

 

Hanigan] 1991

 

Lowe/ 1990

 

Levene! 1992

 

Sherlock/ 2005

 

0 0
0 l
1 l
1 l
1 1

Ross/ 1992

 

 

0
I
1
l
1
1

Ross/ 1996 1 1 1

 

 

 

 

 

1: Its answer for a criterion is “yes”; 0: Its answer for a criterion is “no”; Ms: Means; GA:
Gestational Age; NS: Not Significant; US: Ultrasound; FT: Full-Term.

58

 

Table 3-A (2 of 2): Study Quality Evaluation
Author/ Year Utilizing US scan Has well matched Sample size of each
FT controls study group (case)
g 30

1

 

 

Papile! 1983
Ment/ 1985
Kitchen] 1990
Sostek/ 1987
van de Bor/ 1988
van de Bor/ 1993
van de Bor/ 2004
Whitaker/ 1996
Whitaker/ 1997
Pinto—Martin/
1999
Hanigan/ 1991 1 0
Lowe! 1990 0 1
1 1
1 1
1 1

 

 

 

 

 

 

 

 

 

OOOOOOOOOO

l
1
0
l
l
1
l
l
l

 

 

 

Levene! 1992
Sherlock/ 2005
Ross/ 1992

Ross/ 1996 l l

1: Its answer for a criterion is “yes”; 0: Its answer for a criterion is “no”; Ms: Means; GA:
Gestational Age; NS: Not Significant; US: Ultrasound; FT: Full-Term.

 

 

 

 

 

 

 

 

Or—nr—tr—no—

 

59

Table 3-B (1 of 2): Study Quality Evaluation

 

Author/ Year Has careful or standard Age of Assessment Key confounders are
measurements of mentioned and
outcomes adjusted.

 

Papile! 1983 0

 

Ment/ 1985

 

Kitchen/ 1990

 

Sostek/ 1987

 

van de Bor/ 1988

 

van de Bor/ 1993

 

van de Bor/ 2004

 

Whitaker/ 1996

 

Whitaker/ 1997

 

—l———I—n—————
#A-RMUJN—WNH

Pinto-Martin/
1999

 

Hanigan/ 1991

 

Lowe/ 1990

 

Levene! 1992

 

Sherlock/ 2005

 

Ross/ 1992

 

 

 

 

 

 

N—‘AUJ-bN
OOOOOO

Ross/ 1996

 

1: Its answer for a criterion is “yes”; 0: Its answer for a criterion is “no”; GM/IVH: Germinal
Matrix and/or lntraventricular Hemorrhage. Age of assessment: (1) g 10 months & $1.5
years: 1 point; (2) >15 years & $3 years: 2 points; (3) >3 years & g5 years: 3 points; (4)
>5 years & :10 years: 4 points; (5) >10 years & g 18 years: 5 points.

60

Table 3-B (2 of 2): Study QualitLEvaluation

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Year The other risk factors are Has a clear conclusion Total scores
mentioned and adjusted about GM/IV H of study quality

Papile! 1983 0 1 8
Ment/ 1985 0 1 8
Kitchen/ 1990 0 0 10
Sostek/ 1987 0 0 4
van de Bor/ 1988 1 1 11
van de Bor/ 1993 1 1 12
van de Bor/ 2004 l 1 14
Whitaker/ 1996 1 1 10
Whitaker/ 1997 1 1 10
Pinto-Martin/ 1999 1 0 9
Hanigan/ 1991 0 0 5
Lowe/ 1990 1 1 10
Levene! 1992 0 1 12
Sherlock/ 2005 0 1 13
Ross/ 1992 1 1 11
Ross/ 1996 1 1 ll

 

 

 

 

 

1: Its answer for a criterion is “yes”; 0: Its answer for a criterion is “no”; GM/IVH: Germinal
Matrix and/or lntraventricular Hemorrhage.

61

Table 4 (1 of 2): Ranking of Importance for Selected Studies

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Year Ranking of Importance Total scores
of study quality

van de Bor/ 2004 1 14
van de Bor/ 1993 2 12
van de Bor/ 1988 3 11
Sherlock/ 2005 4 13
Levene/ 1992 4 12
Ross/ 1992 4 11
Ross/ 1996 4 11
Kitchen! 1990 4 10
Whitaker/ 1996 5 10
Whitaker/ 1997 6 10
Lowe/ 1990 6 10
Pinto-Martin/ 1999 7 9
Papile! 1983 8 8
Ment/ 1985 8 8
Hanigan/ 1991 9 5
Sostek/ 1987 10 4

 

 

 

 

 

Ranking of importance from 1 to 10 (1 means the most important study; 10 means the last
important study). Plus mark (+) means the criterion is established. Minus mark (—)

means the criterion is not established.

62

Table 4 (2 of 9: Ranking of Importance for Selected Studies

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Year Ranking of Causal Criteria
Importance Strength Consistency Coherence

van de Bor/ 2004 1 + — +
van de Bor/ 1993 2 + + +
van de Bor/ 1988 3 + + +
Sherlock/ 2005 4 — - —
Levene/ 1992 4 — —- +
Ross/ 1992 4 + — +
Ross/ 1996 4 + — +
Kitchen/ 1990 4 + + +
Whitaker/ 1996 5 + — +
Whitaker/ 1997 6 — — +
Lowe! 1990 6 — — +
Pinto-Martin/ 1999 7 — — +
Papile/ 1983 8 — — +
Ment/ 1985 8 - — +
Hanigan/ 1991 9 + + +
Sostek/ 1987 10 — — +

 

 

 

 

 

 

Ranking of importance from 1 to 10 (1 means the most important study; 10 means the last
important study). Plus mark (+) means the criterion is established. Minus mark (—)
means the criterion is not established.

63

Table 5-A (1 of 2): Frequencies and Odds Ratios of Abnormal Outcomes

in Studies without a Full-Term Control Group

 

 

 

 

 

Author/ Year Abnormal Outcomes GM/ IVH (%)
Papile! 1983 Bayley Scales of Infant Development
Abnormal outcome 5/51= 9.80%
Neurodevelopmental Examination
Minor handicap 21/51= 41.18%
Major handicap 5/51= 9.8%
Kitchen/ 1990 CP 6/33= 18.18%
Disability 6/33= 18.18%
Sostek/ 1987 At 12.6 months
Mental Delay 8.6%
Motor Delay 30.5%
van de Bor/ 1988 TNeurodevelopmental outcome (Handicap) 19/52= 36.5%

 

 

 

 

 

 

 

 

 

 

van de Bor/ 1993 :EzDisability (including Handicap) 21/50=42%
iTotal Handicap 13/50= 26%
van de Bor/ 2004 Disability or Handicaps at 5 Years Overa_ll
None 29/45= 64.44%
Disability 16/45= 35.56%
Handicap 11/45= 24.44%
School Performance at 5 Yefi
Mainstream 31/40= 77.5%
Special Education 9/40= 22.5%
School Performance at 9 Yea_rs
Normal 22/40= 55.0%
Slow Leamer 5/40= 12.5%
Special Education 13/40= 32.5%
School Performance at 14 Yea_rs
Normal 25/45= 55.6%
Slow Leamer 10/45= 22.2%
:EISpecial Education 10/45= 22.2%
Whitaker/ 1996 iBorderline Intelligence 5/83= 6.0%
iMR 5/83= 6.0%
Whitaker/ 1997 TAny disorder in DISC 2.1 P 17/7 8: 21.8%
Hanigan/ 1991 Abnormal Neurodevelopment 12/47= 25.53%

 

 

 

64

 

 

 

 

 

 

Lowe/ 1990 Abnormality on McCSCA 4l11=36.36%
Abnormality on TERA 5/11=45.45%
Abnormality on DT-VMI m1 8. r 8%

Sherlock/ 2005 Abnormal Movement ABC 17/64: 26.56%
CP 9/72= 12.50%
Major neurosensory disability 10/72= 13.89%

 

 

GM/ IVH: Germinal Matrix and/or lntraventricular Hemorrhage; *Normal: premature
normal ultrasound group; CP: Cerebral Palsy; MR: Mental Retardation; DISC 2.1 P: the
Diagnostic Interview Schedule for Children-Parent version 2.1P; McCSCA: the McCarthy
Scales of Children’s Ability; TERA: the Test of Early Reading Ability; DT-VMI:
Developmental Test of Visual-Motor Integration; N/A: Not Available; Movement ABC: the
Movement Assessment Battery for Children; OR: Odds Ratio; CI: Confidence Interval; NS:
Not Significant; T: significant; The frame mark ( [:1 ) means it contributed to the cell of
two-by-two table has a expected count less than 5; T: the odds ratio, 95% Confidence
Interval, and p-value were provided by the publication itself.

65

Table S-A (2 of 2): Frequencies and Odds Ratios of Abnormal Outcomes

in Studies without a Full-Term Control Group

 

 

 

 

 

 

 

 

 

 

 

Author/ Year *Norrnal (%) OR; (95% CI); P-value
Papile/ 1983
1 1/1 15: 9.57% OR=1.03 (0.34-3.13) NS
46/115= 40% OR=1.05 (0.54-2.05) NS
12/115= 10.43% OR=0.93 (0.31-2.80) NS
Kitchen/ 1990 13%|: 3.13 % 0R=6.89 (162-2939) P=0.003T
5/96= 5.21 % OR=4.04 (1.15-14.29) P=0.02T
Sostek/ 1987
16.7% N/A
22.2% N/A
van de Bor/ 1988 40/225= 17.8% OR=2.10 (1.30-3.30) P<0.01T
van de Bor/ 1993 57/234: 24% OR=2.20 (1.12-4.40) P<0.05T
31/234= 13% OR=1.90 (0.81-4.34) NS
van de Bor/ 2004
164/216: 75.92%
49/216= 22.68% OR=1.88 (0.95-3.74) NS
23/216= 10.65% OR=2.70 (1.21-6.08) P=0.01T
178/195: 91.3%
17/195= 8.7% OR=3.04 (1.24-7.43) P=0.01T
107/ 193: 55.4%
57/ 193: 29.5% OR=0.34 (0.13-0.91) P=0.02T
29/193= 15.0% OR=2.72 (1.26-5.88) P=0.009T
97/2 16: 44.9%
93/216: 43.1% OR=0.38 (0.18-0.80) P=0.009T
26l216= 12.0% OR=2.10 (1.01 -4.35) P<0.OSrL
Whitaker/ 1996 26/468= 5.6% OR=1.10 (0.40-3.10) NS
6/468= 1 .3% OR=5.00 (1.50-16.80) P<0.01T
Whitaker/ 1997 93/454: 20.3% OR=1.40 (0.70-2.70) NS

 

 

 

66

 

 

 

 

 

Hanigan/ 1991 10/153= 6.54% OR=4.90 (1.96-12.27) P=0.0003T

Lowe/ 1990 5/27=18.52% OR=2.51 (0.52-12.04) NS
14/27=51 .85% OR=0.77 (0.19-3.16) NS
8/27=29.63% OR=O.53 (0.09-3.01 ) NS

Sherlock/ 2005 39/173= 22.54% OR=1.24 (0.64-2.40) NS
12/180= 6.67% OR=2.00 (0.80-4.98) NS
28/180= 15.56% OR=0.88 (0.40-l.91) NS

 

 

 

 

GM/ IVH: Germinal Matrix and/or lntraventricular Hemorrhage; *Normal: premature
normal ultrasound group; CP: Cerebral Palsy; MR: Mental Retardation; DISC 2.1 P: the
Diagnostic Interview Schedule for Children-Parent version 2.1P; McCSCA: the McCarthy
Scales of Children’s Ability; TERA: the Test of Early Reading Ability; DT-VMI:
Developmental Test of Visual-Motor Integration; N/A: Not Available; Movement ABC: the
Movement Assessment Battery for Children; OR: Odds Ratio; CI: Confidence Interval; NS:
Not Significant; T: significant; The frame mark ( I: ) means it contributed to the cell of
two-by-two table has a expected count less than 5; i: the odds ratio, 95% Confidence
Interval, and p-value were provided by the publication itself.

67

Table 5-B (l of 2): Frequencies and Odds Ratios of Special Cognitive

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Outcomes in a Study with a Full-Term Control Group
Author/ Measures GM/ IV H (%) *Nonnal (%) FT Controls (%)
Year
Ross/ Children who habituated l4/30= 46.67% 29/30= 96.67% 27/30= 90.00%
1992
Response on AB tails
(a) Failed reversal trials 22/30= 73.33% 17/30= 56.67% 5/30= 16.67%
(b) AB error without delay 6/30= 20.00% 8/30= 26.67% [1733: 13.33%
(c) 1-3 seconds delay ”Q 6.67% @ 13.33% 17/30= 56.67%
(d) g 5 seconds delay IW 0.00% "17371 3.33% [W]: 13.33%

 

GM/IVH: Gemrinal Matrix and/or lntraventricular Hemorrhage; *Normal: premature
normal ultrasound group; FT: Full-Term; OR: Odds Ratio; CI: Confidence Interval; T:
significant; T: slightly significant; NS: Not Significant; The frame mark ( E] ) means it
contributed to the cell of two-by-two table has a expected count less than 5.

68

 

Table 5-B (2 of 2): Frequencies and Odds Ratios of Special Cognitive
Outcomes in a Study with a Full-Term Control Group
Author/ OR; (95% CI); P-value OR; (95% CI); P-value
Year (Compared with *Normal) (Compared with FT Controls)
Ross/ OR=0.03 (0.004-0.25) P<.0001T OR=0.10 (0.02-0.39) P=.0003T
1992

 

 

 

 

 

 

OR=2. 10 (0.71 -6.22) NS OR=13.75 (3.92-48.27) P<.0001T
OR=0.69 (0.21-2.30) NS OR=1 .63 (0.41 -6.47) NS
OR=0.46 (0.08275) NS OR=0.06 (0.01-0.27) P<.0001 T
OR=0.32 (0.01-8.24) NS OR=O.10 (0.01-1.88) P=0.04T.

 

GM/IVH: Gemrinal Matrix and/or lntraventricular Hemorrhage; *Normal: premature
normal ultrasound group; FT: Full-Term; OR: Odds Ratio; CI: Confidence Interval; T:
significant; it: slightly significant; NS: Not Significant; The frame mark ( [:1 ) means it
contributed to the cell of two-by-two table has a expected count less than 5.

69

Table 6-A (1 of 2): Means and Standard Deviations of Developmental Tests

in Studies without a Full-Term Control Group

 

 

 

 

 

 

 

 

 

Author/ Year Measures and/or outcomes GM/IVH
Mean (SD); N
Kitchen/ 1990 WPPSI Full Scale 107.50 (11.85) N: 33
Ment/ 1985 Bayley 12 Mental 93.05 (15.23) N: 49
Bayley 18 Mental 88.93 (17.37) N: 42
SBIS 30 Month CA 87.02 (14.85) N: 40
PPVT 30 Month CA 88.58 (17.86) N: 35
Sostek! 1987 At 12.6 months
Bayley Scales of Infant Development
(a) Mental 105.80 (16.10) N: 23
(b) Motor 93.60 (22.60) N: 23
At 21.5 months
Bayley Scales of Infant Development
(a) Mental 105.20 (16.90) N: 20
(b) Motor 96.40 (20.50) N: 20
Whitaker/ 1996 language
(a) Overall (TOLD SLQ) 94.60 (18.40) N: 73
(b) Receptive (FOLD LIQ) 95.50 (15.20) N: 73
(c) Expressive (TOLD SPQ) 94.50 (19.60) N: 73
(d) Verbal reasoning (SB area score) 102.90 (10.50) N: 73
Short-term memory (SB area score) 99.20 (11.00) N: 73
Quantitative reasoning (SB area score) 107.20 (9.10) N: 73
Visual perceptual organization
(a) Abstract visual reasoning (SB area 101.20 (10.40) N: 73
score) 92.20 (10.70) N: 73
(b) Vrsual-motor integration 105.70 (20.60) N: 71
(c) TVPS perceptual quotient
Pinto-Martin/ Motor Ability
1999 Age_2 98.90 (15.80) N: 61
Agfl 7.00 (2.90) N: 70
Age_9 7.90 (2.80) N: 65
General Cognitive Ability
Age_2 105.20 (18.80) N: 83

 

70

 

 

Age_6 102.90 (10.40) N: 70
Age_9 99.40 (12.80) N: 65
Levene/ 1992 TOMI- Overall: Median (Ranges) 3.00 (0-13) N: 54
WPPSI (vocabulary): Mean (SD) 19.20 (6.64) N: 54
Sherlock/ 2005 Cognitive and educational outcomes

 

 

 

 

 

 

(a) For IQ Score 104.17 (15.32) N: 72
(b) For WRAT3

Reading 100.85 (14.80) N: 69
Spelling 96.85 (11.55) N: 68
Arithmetic 93.10 (12.75) N: 68

 

GM/IV H: Germinal Matrix and/or lntraventricular Hemorrhage; SD: Standard Deviation;
*Normal: premature normal ultrasound group; N: Number; WPPSI: the Wechsler Preschool
and Primary Scales of Intelligence; SBIS: Stanford-Binet Intelligence Test; PPVT: Peabody
Picture Vocabulary Test; CA: Corrected Age; TOLD : Test of Language Development; SLQ:
speaking quotient; LIQ: listening quotient; SB: the Composite Index of the Stanford Binet;
TVPS: the Test of Visual Perceptual Skills; IQ: Intelligence Quotient; SAS: Standard Age
Score; ECA: Expressive Communication Age; RCA: Receptive Communication Age; TOMI:
the Test of Motor Impairment; WRAT3: Wide Range Achievements Test; NS: Not
Significant; T: premature normal ultrasound group had significantly lower score than
GM/IVH group.

71

Table 6-A (2 of 2): Means and Standard Deviations of Developmental Tests

in Studies without a Full-Term Control Group

 

 

 

 

 

 

 

 

 

 

Author/ Year *Normal Significance
Mean (SD); N

Kitchen/ 1990 106.30 (14.20) N: 96 NS

Ment/ 1985 95.5 (13.1) N: 65 NS
90.9 (16.0) N: 5 NS
88.2 ( 17.2) N: 47 NS
88.2 (17.0) N: 36 NS

Sostek/ 1987
105.20 (23.50) N: 36 NS
96.40 (25.00) N: 36 NS
103.10 (27.50) N: 36 NS
93.70 (24.80) N: 36 NS

Whitaker/ 1996
95.60 (16.10) N: 429 NS
97.20 (13.70) N: 428 NS
94.70 (17.10) N: 432 NS
104.30 (10.90) N: 436 NS
101.40 (1 1.40) N: 436 NS
108.80 (10.00) N: 436 NS
102.10 (13.00) N: 436 NS
93.10 (10.70) N: 431 NS
106.90 (19.70) N: 421 NS

Pinto-Martin/ 1999
101.70 (14.80) N: 382 NS
7.00 (3.00) N: 449 NS
7.80 (3.00) N: 402 NS
106.20 (19.80) N: 504 NS
103.70 (12.10) N: 449 NS

 

72

 

 

 

 

 

 

 

99.10 (14.50) N: 403 NS
Levene/ 1992 2.25 (0-16) N: 64 NS

19.30 (5.82) N: 64 NS
Sherlock/ 2005

104.19 (15.35) N=180 NS

95.20 (15.70) N: 172 P=0.004T

93.60 (12.40) N: 172 P=0.048T

88.30 (14.30) N: 170 P=0.026T

 

GM/IVH: Germinal Matrix and/or lntraventricular Hemorrhage; SD: Standard Deviation;
*Normal: premature normal ultrasound group; N: Number“, WPPSI: the Wechsler Preschool
and Primary Scales of Intelligence; SBIS: Stanford-Binet Intelligence Test; PPVT: Peabody
Picture Vocabulary Test; CA: Corrected Age; TOLD : Test of language Development; SLQ:
speaking quotient; LIQ: listening quotient; SB: the Composite Index of the Stanford Binet;
TVPS: the Test of Vlsual Perceptual Skills; IQ: Intelligence Quotient; SAS: Standard Age
Score; ECA: Expressive Communication Age; RCA: Receptive Communication Age; TOMI:
the Test of Motor Impairment; WRAT3: Wide Range Achievements Test; NS: Not
Significant; T: premature normal ultrasound group had significantly lower score than

GM/IVH group.

73

 

Table 6-B (l of 2): Means and Standard Deviations of Cognitive Tests in
Studies with a Full-Term Control Group

Author/ Year Measures

Lowe/ 1990 (a) McCSCA Mean (Range)

 

 

 

(b) TERA Mean (Range)

(c) DT—VMI Mean (Range)

((1) BFL

R Mean (Range)

L Mean (Range)
Ross/ 1992 (a) Bayley MDI

(b) Bayley PDI

(c) Habituation- Trials to habituation

(d) Habituation- Increase in % time looking at novel face
Ross/ 1996 Bayley MDI

Bayley PDI

Habituation- Trials to habituation

Habituation- Increase in % time looking at novel face
Invisible displacements task- ID trials score (maximum of 8.0)
Invisible displacements task- SS trials score (maximum of 4.0)
Object discrimination reversal task (Number of reversals)

GM/IVH: Germinal Matrix and/or lntraventricular Hemorrhage; SD: Standard Deviation;
*Normal: premature normal ultrasound group; FT: Full-Term; McCSCA: the McCarthy
Scales of Children’s Ability; TERA: the Test of Early Reading Ability; DT-VMI:
Developmental Test of Vlsual-Motor Integration; BFL: Benton Finger Localization; R: Right;
L: Left; Bayley MDI: Bayley Mental Developmental Index; Bayley PDI: Bayley
Psychomotor Development Index; ID: Invisible Displacement; SS: Systematic Search; NS“:
No significant difference on the test between GM/IVH and *Normal groups; NSb: No
significant difference on the test between GM/IVH and FT Control groups; T: there is a
significant difference on the test between GM/IVH and *Normal groups; thhere is a
significant difference on the test between GM/IVI-I and FT Control groups.

 

 

 

 

 

74

 

 

 

 

Table 6-B (2 of 2): Means and Standard Deviations of Cognitive Tests in
Studies with a Full-Term Control Grou
Author/ GM/IVH *Nonnal FT Controls Significance
Year Mean (SD) Mean (SD) Mean (SD)
Lowe/ 95(71-120) 97(75-123) 117(89-141) NS3
1990 90.8(66-126) 82.6(55-110) 104.1(79-139) NS“
10.8(5-15) 9.4(4-15) 11.9(9-16) Nsa
7.3(3-10) 7.6(5-10) 9.2(8-10) NS“
7.1(3-10) 7.0(3-10) 9.1(7-10) Nsa
Ross/ 104.4 (18) 113.4 (11) 110.0 (12) P< 0.057;:
1992 94.3 (14) 99.7 (16) 102.8 (13) NS“; Nsb
25.2 (6) 20.3 (6) 21.2 (7) P<0.03T; P< 0.053,:
6.2 (4) 6.8 (2) 6.3 (3) NS“; Nsb
Ross/ 101.8 (16.2) 105.2 (19.1) 109.4 (20.0) Ns“; NSb
1996 98.2 (19.0) 97.1 (15.5) 106.3 (16.3) NS“; NSb
21.6 (5.7) 20.6 (6.4) 20.6 (6.9) NS“; Nsb
6.4 (10.2) 12.2 (12.4) 13.4 (15.1) NS“; Nsb
3.9 (1.7) 5.2 (1.3) 5.3 (1.6) P<0.01"; P<0.0l:]:
1.5 (1.2) 1.6 (1 .2) 2.4 (1 .2) NS“; P<0.055r
1.52 (0.5) 2.0 (0.6) 2.1 (0.5) P<0.05“; P< 0.05::

 

 

 

 

 

 

 

GM/IVH: Germinal Matrix and/or lntraventricular Hemorrhage; SD: Standard Deviation;
*Normal: premature normal ultrasound group; FT: Full-Term; McCSCA: the McCarthy
Scales of Children’s Ability; TERA: the Test of Early Reading Ability; DT-VMI:
Developmental Test of Visual-Motor Integration; BFL: Benton Finger Localization; R: Right;
L: Left; Bayley MDI: Bayley Mental Developmental Index; Bayley PDI: Bayley
Psychomotor Development Index; ID: Invisible Displacement; SS: Systematic Search; NS“:
No significant difference on the test between GM/IVH and *Normal groups; NS": No
significant difference on the test between GM/IVH and FT Control groups; T: there is a
significant difference on the test between GM/IVH and *Normal groups; thhere is a
significant difference on the test between GM/IVH and FT Control groups.

75

Table 7 -A

Weighted OR for Abnormal Neurodevelopment

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Year Measures/ Outcomes OR
Papile! 1983 Major handicap 0.93
Kitchen/ 1990 CP 6.89
Kitchen/ 1990 Disability 4.04
van de Bor/ 1988 Handicap 2.10
van de Bor/ 1993 Disability (including handicap) 2.20
van de Bor/ 1993 Handicap 1.90
van de Bor/ 2004 Disability 1.88
van de Bor/ 2004 Handicap 2.70
Hanigan/ 1991 Abnormal neurodevelopment 4.90
Sherlock/ 2005 Abnormal movement ABC 1.24
Sherlock/ 2005 CP 2.00
Median of 0R8 2.1
Range of ORs 5.96
Un-weighted OR 2.80
Weighted OR 2.53

 

 

 

Table 7 -B Weighted OR for General Cognition

 

 

 

 

 

 

 

 

 

 

 

Author/ Year Measures/ Outcomes OR
Whitaker/ 1996 Borderline intelligence 1.]
Whitaker/ 1996 Mental retardation 5
Whitaker/ 1997 Any disorder in DISC 2.1 P 1.4

Lowe/ 1990 Abnormality on McCSCA 2.51

Lowe/ 1990 Abnormality on TERA 0.77
Median of 0R8 1.4

Range of ORs 4.23
Un-weighted OR 2.20
Weighted OR 2.49

 

 

 

76

 

 

Table 7 -C Weighted Means for Neurodevelopment: Motor Abilities

 

 

 

 

 

 

 

 

 

Author/ Year Measures/ Mean of the Sample size of the Mean of the Sample size of
Outcomes GM/IVH group GM/IV H group *Normal group the *Normal
group
Sostek! 1987 Bayley PDI 96.4 20 93.7 36
Pinto-Martin/ Bayley PDI 98.9 61 101.7 382
1999
Ross/ 1992 Bayley PDI 94.3 30 99.7 30
Ross/ 1996 Bayley PDI 98.2 27 97.1 28
Un-w Means 96.95 98.05
W Means 97.40 100.70
D in W Means 3.3

 

 

 

 

 

 

Un-w Means: Un—weighted Means; W Means: Weighted Means; D in W Means: Difference in

Weighted Means.

Table 7 -D Weighted Means for General Cognition

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Author/ Year Measures/ Mean of the Sample size of Mean of the Sample size of
Outcomes GM/IVH the GM/IV H *Normal the *Normal
group group amup group
Kitchen/ 1990 WPPSI Full 107.5 33 106.3 96
Scale
Ment/ 1985 SBIS 30 87.02 40 88.2 47
Month CA
Ment/ 1985 PPVT 30 88.58 35 88.2 36
Month CA
Sostek/ 1987 Bayley MDI 105.2 20 103.1 36
Pinto-Martini WISC-HI 99.4 65 99. 1 403
1999
Sherlock/ 2005 IQ Score 104.17 72 104.19 180
Lowe! 1990 McCSCA 95 11 97 27
Ross/ 1992 Bayley MDI 104.4 30 113.4 30
Ross/ 1996 Bayley MDI 101.8 27 105.2 28
Un-w Means 99.23 100.52
W Means 99.46 100.67
D in W Means 1.21

 

 

 

 

 

 

Un-w Means: Un-weighted Means; W Means: Weighted Means; D in W Means: Difference in

Weighted Means.

77

 

 

 

 

 

 

 

 

 

 

Table 7-E Weighted Means for Special Cognition - Language
Author! Year Measures! Mean of the Sample size of Mean of the Sample size of
Outcomes GM/IVH the GM/IVH *Normal the *Normal
group group group group
Whitaker/ Overall TOLD 94.6 73 95.6 429
1996 SLQ
Sherlock/ WRAT3 100.85 69 95.2 172
2005 Reading
Sherlock! WRAT3 96.85 68 93.6 172
2005 Spelling
Lowe! 1990 TERA 90.80 11 82 27
Un-w Means 95.77 91.6
W Means 97.05 94.63
D in W Means 2.42

 

 

 

 

 

 

 

 

Un-w Means: Un-weighted Means; W Means: Weighted Means; D in W Means: Difference in
Weighted Means.

Table 7 -F Weighted Means for Special Cognition— Reasonin & Memory

 

 

 

 

 

 

 

 

 

 

 

 

 

Author! Year Measures/ Mean of the Sample size of Mean of the Sample size of
Outcomes GM/IVH the GM/IVH *Normal the *Normal
group group group group
Whitaker! Short-term 99.2 73 101.4 436
1996 memory
Whitaker/ Quantitative 107.2 73 108.8 436
1996 reasoning
Whitaker! Abstract visual 101.2 73 102. 1 436
1996 reasoning
Sherlock/ WRAT3 93. 1 68 88.3 170
2005 (Arithmetic)
Un-w Means 100.18 100.15
W Means 100.30 102.28
D in W Means 1.98

 

 

 

Un-w Means: Un-weighted Means; W Means: Weighted Means; D in W Means: Difference in
Weighted Means.

78

Table 8 (l of 2):

Causal Criteria for Selected Studies

 

Author] year

Strength

Consistency

Coherence

Specificity

Temporality

 

Papile! 1983

 

Ment/ 1985

 

Kitchen! 1990

 

Sostek] 1987

 

van de Bor/ 1988

 

van de Bor/ 1993

 

van de Bor/ 2004

 

Whitaker] 1996

 

Whitaker] 1997

 

Pinto-Martin] 1999

 

Hanigan] 1991

 

Lowe! 1990

 

Levene! 1992

+++++++++++++

 

Sherlock] 2005

 

Ross/ 1992

+

+++++++++++++++

 

 

Ross] 1996

 

 

 

+

 

 

+

 

Plus mark (+) means the criterion is established. Minus mark (—) means the criterion is

not established.

79

 

 

Table 8 (2 of 2): Causal Criteria for Selected Studies
Author] year Biological Biological Experiment Analogy

Plausibility Gradient

Papile! 1983 - — — —
Ment/ 1985 — — — —
Kitchen] 1990 — - — -
Sostek] 1987 — — -— —
van de Bor/ 1988 —- — — —
van de Bor/ 1993 — — — —
van de Bor/ 2004 — — — -
Whitaker] 1996 — — — -
Whitaker] 1997 — — — —
Pinto-Martin] 1999 — — — —
Hanigan] 1991 — -- - -
Lowe] 1990 — — — —
Levene] 1992 - - — -
Sherlock] 2005 — — — —
Ross! 1992 — — - —
Ross] 1996 — - — —
Plus mark (+) means the criterion is established. Minus mark (—) means the criterion is
not established.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

80

Appendix Table 1 (1 of 2):

Measurements and Classification of IVH

 

 

 

 

Author] year Times and Timeframe of Exposure Measurement
Papile! 1983 Examined twice.
I“: performed CT between 3 and 10 days of age.
2'“: performed CT or US one week after the initial scan.
Ment/ 1985 Examined at least 3 times.
Performed CT scans within the first three weeks of life.
Performed US scans ; twice in the first eight weeks of life.
Kitchen] 1990 Performed US several times between birth and after discharge until any

ventricular dilation had either resolved or stabilized.

 

Sostek] 1987

Examined four times: performed US at the time of birth, on days 1 and 7,
and at discharge.

 

van de Bor/ 1988

Performed US at least four times.
18‘: as soon as after admission.
2"d-3'd: within the first week of life.
4‘“: after the first week of life until discharge.

 

van de Bor/ 1993

Performed US at least four times.
18‘: as soon as after admission.
2"“-3“: within the first week of life.
4'“: after the first week of life until discharge.

 

van de Bor/ 2004

Performed US at least four times.
1‘”: as soon as after admission.
2““3'“: within the first week of life.
4"“: after the first week of life until discharge.

 

Whitaker] 1996

Performed US 3-4 times.

1“- 3'“: 98% of the cohort was examined at least one time at 4 hours,
24 hours, and in 7 days of life.

4'“: 47% of the cohort was examined between the third and fifth
hospital week and] or before discharge.

 

Whitaker] 1997

Performed US 3-4 times.

1"- 3'“: 98% of the cohort was examined at least one time at 4 hours,
24 hours, and in 7 days of life.

4‘“: 47% of the cohort was examined between the third and fifth
hospital week and] or before discharge.

 

 

Pinto-Martin!
l 999

 

Performed US 3—4 times.

1‘“- 3rd: 98% of the cohort was examined at least once at 4 hours, 24
hours, and in 7 days of life.

4‘“: 47% of the cohort was examined between the third and fifth
hospital week and] or before discharge.

 

81

 

 

Hanigan] 1991

Performed US 4 times.
I“: at 12 hours of age.
2%“: at 2, 3, 7, and 14 days after birth.

 

Lowe! 1990

Performed a CT brain scan at 5 to 10 days of age. Given infants with
PIVH weekly CT brain scan until the hemorrhage resolved.

 

Levene] 1992

Examined at least twice: weekly performed US for the first month and
then every week until discharge.

 

Sherlock] 2005

Performed at least one US in the first week of life, at 28 days, and prior
to discharge.

 

Ross] 1992

Performed US three times.
1": at two weeks from birth.
2"“: at before discharge.
3'“: at 1 month after birth.

 

 

Ross/ 1996

 

Performed US three times.
1": at two weeks from birth.
2'”: at before discharge.
3'“: at 1 month after birth.

 

IVH: lntraventricular Hemorrhage; CT: Computed Tomographic; US: Ultrasound; CIVH:
Cereme lntraventricular Hemorrhage; GMH: Germinal Matrix Hemorrhage; GMH/IVH:
Germinal Matrix Hemorrhage and/or lntraventricular Hemorrhage; CVH:
Cerebroventricular Hemorrhage; ICH: Intracerebral Hemorrhage; VD: Ventricular Dilation;
PIVH: Periventricular-lntraventricular Hemorrhage; PUVE: Parenchymal Lesion and/or
Ventricular Enlargement; PVH: Periventricular hemorrhage; FT: Full Term.

82

 

Appendix Table 1 (2 of 2):

Measurements and Classification of IVH

 

Author] year

Classification of IVH

 

Papile! 1983

CIVH: grade from I to IV.
Grade I: isolated GMH.
Grade H: IVH with normal ventricular size.
Grade 111: IVH with ventricular dilation.
Grade IV: IVH with parenchymal hemorrhage.

 

Ment] 1985

GMH/IVH; grade from I to IV; according to the Papile classification.

 

Kitchen] 1990

CVH and ventricular dilation were defined as follows:

(1) GMH: “hemorrhage, usually less than 1 cm diameter, confined
to the germinal matrix.”

(2) IVH: “all hemorrhages within the lateral ventricles, regardless of
size, but not extending into brain parenchyma.”

(3) ICH: “any hemorrhage extending into or originating in brain
parenchymalateral or superior to the lateral ventricles.”

(4) VD: “width of the lateral ventricles exceeding 3 mm in the
coronal view at the level of the foramen of Monro.”

 

Sostek] 1987

IVH; grade from I to IV; according to a modified Papile classification.
Grade I: GMH without ventricular blood.
Grade H: IVH in the lateral ventricles without ventricular distention.
Grade III: IVH with ventricular distention.
Grade IV: IVH extending into the brain parenchyma.

 

van de Bor/ 1988

PIVH; grade from I to IV; according to the Papile classification.

 

van de Bor/ 1993

PIVH; grade from I to IV: according to the Papile classification.

 

van de Bor/ 2004

PIVH; grade from I to IV; according to the Papile classification.

 

Whitaker] 1996

Three study groups:
(1) No abnormality (NA)
(2) GMH] IVH
(3) PUVE

 

Whitaker] 1997

Three study groups:
(1) No abnormality (NA)
(2) GMH] IVH
(3) PUVE

 

Pinto-Martin]
1 999

Three study groups:
(1) No abnormality (NA)
(2) GMH] IVH
(3) PUVE

 

 

Hanigan! 1991

 

PVH; grade from I to IV; according to the Papile classification.
(1) Low-grade hemorrhages: Papile’s classification Grades I & II.

 

83

 

 

(2) High-grade hemorrhages: Papile’s classification Grades III & IV.

 

Lowe! 1990 PIVH; grade from I to H; according to the Papile classification.

 

Levene] 1992 Five study groups:

(1) Premature normal ultrasound group

(2) Prolonged flare (PF)

(3) GM/IV H, without parenchymal hemorrhage, but no evidence of
PF.

(4) Both GM/IVH and PF

(5) FT Controls group

 

Sherlock] 2005 IV H; grade from I to IV; according to the Papile’s classification.

 

Ross] 1992 Three study groups:

(1) Premature normal ultrasound group

(2) Subependymal and lntraventricular Hemorrhage (S/IV H)
(3) FT Controls group

 

 

Ross] 1996 Three study groups:

(1) Premature normal ultrasound group

(2) Subependymal and lntraventricular Hemorrhage (S/IV H)
(3) FT Controls group

 

 

 

 

IVH: lntraventricular Hemorrhage; CT: Computed Tomographic; US: Ultrasound; CIVH:
Cerebral lntraventricular Hemorrhage; GMH: Germinal Matrix Hemorrhage; GMH/IVH:
Germinal Matrix Hemorrhage and/or lntraventricular Hemorrhage; CVH:
Cerebroventricular Hemorrhage; ICH: Intracerebral Hemorrhage; VD: Ventricular Dilation;
PIVH: Periventricular-lntraventricular Hemorrhage; PUVE: Parenchymal Lesion and/or
Ventricular Enlargement; PVH: Periventricular hemorrhage; FT: Full Term.

84

Appendix Table 2 (1 of 2):

Measurements of Outcomes

 

Author] Year

Outcome

 

Papile! 1983

1. Overall development.
2. Neuromotor.
3. Vision and hearing.

 

 

Ment/ 1985 1. Development (IQ)
2. Development (IQ)
Kitchen] 1990 1. Vision, hearing, and the central nervous system (i.e.: Spastic CP; Ataxic

CP)
2. Psychologic assessment

 

Sostek] 1987

1. Mental and motor development
2. Muscle tone, primitive reflexes, deep tendon reflexes, protective and
equilibrium reactions, and range of motion.

 

 

 

van de Bor/ 1. Neurodevelopmental outcome
1988
van de Bor] 1. Mental development
1993 2. Abnormalities of gross and fine motor function
3. Neurological status
4. Minor neurological dysfunction
5. Central motor deficit
6. Vision and visual functions.
7. Hearing
8. Speech and language development
van de Bor/ 1. Neurodevelopmental outcome was assessed during a home visit at 5
2004 years (including neurological status, vision and visual function, hearing,

and language and speech). & Mental development.
2. School performance at 9 years of age.
3. School performance at 14 years of age.

 

Whitaker] 1996

Age 6 years
1. Global cognitive outcomes

2. Specific cognitive outcomes
(a) Language; (b) Short-term memory; (c) Quantitative reasoning; (d)
Visual perceptual organization.

 

Whitaker] 1997

Age 6 years
1. DSM-HI-R psychiatric disorders; Behavior problems (i.e.: children with

ADHD)
2. Intelligence
3. Motor problems

 

 

Pinto-Martin]

 

Age 2 years

 

85

 

 

 

1999

1. Neurodevelopmental status
2. “Standardized assessments of infant development”; “Examinations of
vision, hearing, and CP”

,Age6 years

1. Psychopathology
2. “Standardized assessments of general intellectual and motor
functioning”

Age 9 veala

 

1. School performance
2. “Standardized assessments of general intellectual and motor
functioning”

 

Hanigan] 1991

1. Neurodevelopment_al outcomg

Ex: “Abnormalities in muscle tone, reflexes, postural reactions and motor
patterns”

2. Cogai'tive status

3. Developmen_tal quotieng:

Ex: Visual motor] problem-solving skills, receptive language, and
expressive language

 

Lowe! 1990

1. Cognitive performance
2. Early readying skills

3. Visuomotor integration
4. Reading potential

 

Levene] 1992

1. Motor functions
2. Vocabulary

 

Sherlock] 2005

1. CP; blindness; deafness
2. Intellectual impairments (IQ score< -1 SD.)

 

Ross/ 1996

1. Motor & Mental.
2. Three special cognitive abilities.

 

 

Ross] 1992

 

1. “Infants’ information processing”
2. “The development of memory for location of an object”
3.Motor & Mental.

 

86

 

 

Appendix Table 2 (2 of 2):

Measurements of Outcomes

 

Author] Year

Measurements of Outcome

 

Papile! 1983

1. Bayley Scales of Infant Development (including mental (MDI) and
motor (PDI) scales).

2. Neuromotor examination

3. Ophthalmologic and audiologic examinations

 

Mentl 1985

1. BSID at 3, 6, 12, and 18 months.
2. SBIS and PPVT-R at 30 months.

 

Kitchen] 1990

l. Diagnosed by the developmental pediatrician.
2. WPPSI (a test in English)

 

Sostek] 1987

1. The Bayley Scales of Infant Development (Bayley, 1969)
2. An extensive neurologic examination based on Anriel-Tison and
Grenier (1980)

 

van de Bor/
1988

1. Gesell test

 

van de Bor/
1993

1. A language and speech test

2. DDST

3. Part of Touwen’s standardized and age-specific neurological
examination

4. Observation of its presence

5. SRCMDASID

6. Physical, motor, and functional examination of vision

7. Puretone audiometry with a hand-held audiometer (Hortrnann DA 323)
8. Screening test developed by Gerritsen (1989)

 

van de Bor/
2004

l. Validated standardized tests (was assessed at 5 years).

Registration of school attendance at 5 years of age.

2. Using questionnaires to be filled by parents (outcomes: (1) normal, (2)
slow learners, and (3) special education).

3. Using questionnaires to be filled by adolescents (outcomes: (1) normal,
(2) slow learners, and (3) special education).

 

 

Whitaker] 1996

 

Age 6 years
1. SB fourth edition composite score (for general intellectual functioning);

VABS composite score (for overall adaptive functioning).

2. (a) TOLD (for overall, receptive, expressive language); SB verbal
reasoning area score (for verbal reasoning); (b) SB short-term memory
area score (for short-term memory); (c) SB quantitative reasoning area
score (for quantitative reasoning); ((1) SB abstract visual reasoning area
score (for abstract visual reasoning); VMI (for visual motor integration;
TVPS (for visual perceptual skills).

 

87

 

 

 

Whitaker] 1997

Age 6 years
1. DISC 2.1 P (the children’s version was not used).

2. SB
3. RMPI

 

Pinto—Martin]
1 999

AgeL'eara

 

1. Motor outcome measure: Bayley PDI(Bay1ey 1969)
2. General cognitive outcome measure: Bayley MDI (Bayley 1969) and
SB3 (Tennan and Merrill 1960).

Age 6 years
1. Motor outcome measure: RMPI (Riley 1976) (assess fine, gross, and

oral).
2. General cognitive outcome measure: SBzFE (Thorndike et al. 1986)

Age 9 years
1. Motor outcome measure: RMPI (assess fine, gross, and oral).

2. General cognitive outcome measure: WISC—III (Wechsler 1991)

 

Hanigan] 1991

l. The Peabody DMQs

2. A nonstandardized battery of items taken from Gesell, Cattell, and
Bayley.

BSID (for the 18-month visit).

SBIS (for the 3-year visit).

 

Lowe! 1990

1. McCSCA
2. TERA

3. BDTVMI
4. BFL

 

Levene] 1992

1. T‘OMI (Stott et al., Henderson)

2. WPPSI

3. A number of measures of basic literacy, language, representational and
mathematical concepts and skills (Tizard et al.).

4. A checklist of behavioral statements.

 

 

 

Sherlock] 2005 l. Diagnosed by pediatricians; Movement ABC.
2. WISC-III; Several special cognitive tests (including VCI, POI, FDI,
PSI, TOL, RCF, WRAT3)

Ross] 1996 1. The Bayley Scales of Infant Development (Bayley, 1969)

 

2. A habituation/novelty preference task (Janowsky, 1985); an object
displacement task; an object discrimination reversal task

 

88

 

 

 

 

Ross] 1992 1. A habituation/novelty preference task (J anowsky, 1985)

2. The AB object permanence task (Diamond, 1985)

3. The Bayley Scales of Infant Development (Bayley, 1969): Bayley MDI,
PDI.

 

 

 

N: Number; DDST: Denver Developmental Screening Test; SRCMDASID: the Standard
Recording of Central Motor Deficit and Associated Sensory and Intellectual Deficit;
Movement ABC: the Movement Assessment Battery for Children; WISC-III: the Wechsler
Intelligence Scale for Children-Third Edition; VCI: Verbal Comprehension Index; POI:
Perceptual Organization Index; FDI: Freedom from Distractibility Index; PSI: Processing
Speed Index; TOL: Tower of London; RCF: Rey Complex Figure; WRAT3: Wide Range
Achievements Test; CP: Cerebral Palsy; Bayley PDI: Bayley Psychomotor Development
Index; Bayley MDI: Bayley Mental Developmental Index; SB3: the Stanford Binet Third
Edition, Form L-M; RMPI: the Riley Motor Problems Inventory; SBzFE: the Composite
Index of the Stanford Binet: Fourth Edition; WISC-III: the Full-Scale IQ from the Wechsler
Intelligence Scales for Children- Third Edition; DISC 2.1 P: the Diagnostic Interview
Schedule for Children-Parent version 2.1P; SB: the Composite Index of the Stanford Binet;
VABS: Vineland Adaptive Behavior Scale; TOLD: Test of Language Development; VMI:
the Developmental Test of Visual Motor Integration; TVPS: the Test of Visual Perceptual
Skills; The Peabody DMQs: the Peabody Developmental Motor Quotients; BSID: the
Bayley Scales of Infant Development; SBIS: the Stanford-Binet Intelligence Scale; WPPSI:
the Wechsler Preschool and Primary Scales of Intelligence; K-ABC: the Gestalt Closure
subtest from the Kaufman Assessment Battery for Children; HVOT: the Hooper Visual
Organization Test; GP test: the Grooved Pegboard test; ROCF: the Rey-Osterreith Complex
Figure; McCSCA: the McCarthy Scales of Children’s Ability; TERA: the Test of Early
Reading Ability; BDTVMI: the Beery Developmental Test of Visual-Motor Integration; BFL:
Benton Finger Localization; SBISFE: the Stanford-Binet Intelligence Scale, Fourth Edition;
VR: Verbal Reasoning; AN R: Abstract] Visual Reasoning; STMSSAS: Short Term Memory
Subscale Standard Age Scores; SICD: the Sequenced Inventory of Communication
Development; ECA: Expressive Communication Age; RCA: Receptive Communication Age;
TOMI: the Test of Motor Impairment; WPPSI: the vocabulary subscale of the Wechsler
Preschool and Primary Scale of Intelligence; SBIS: the Stanford-Binet Intelligence Scale;
PPVT-R: the Peabody Picture Vocabulary Test-Revised.

89

 

 

I?
l

BIBLIOGRAPHY

1. van de Bor M and Ouden L. School performance in adolescents with and without
periventricular-intraventricular hemorrhage in the neonatal period. Sem. Perin. Vol 28, No
4, 2004

2. Papile LA, Burstein J, Burstein R, Koffler H. Incidence and evolution of

subependymal and intraventricular hemorrhage: a study of infants with birth weights less
than 1,500 gm. J Pediatr. 1978 Apr;92(4):529-34.

 

p-
3. Dykes FD, Lazzara A, Ahmann P, Blumenstein B, Schwartz J, Brann AW. 9
lntraventricular hemorrhage: a prospective evaluation of etiopathogenesis. Pediatrics. 1980 5;;
Jul;66( 1 ):42-9.
i

4. Ahmann PA, Lazzara A, Dykes FD, Brann AW Jr, Schwartz JF. lntraventricular
hemorrhage in the high—risk preterm infant: incidence and outcome. Ann Neurol. 1980 LL

Feb;7(2): 118-24.

5. Fitzhardinge PM, Flodmark O, Fitz CR, Ashby S. The prognostic value of computed
tomography of the brain in asphyxiated premature infants. J Pediatr. 1982
Mar;100(3):476-81.

6. Shankaran S, Slovis TL, Bedard MP, Poland RL. Sonographic classification of
intracranial hemonhage. A prognostic indicator of mortality, morbidity, and short-term
neurologic outcome. J Pediatr. 1982 Mar;100(3):469-75.

7. Merit LR, Scott DT, Ehrenkranz RA, Rothman SG, Duncan CC, Warshaw J B.
Neonates of less than or equal to 1,250 grams birth weight: prospective neurodevelopmental
evaluation during the first year post-term. Pediatrics. 1982 Aug;70(2):292-6.

8. Rennie JM. The immature brain. In: Rennie JM, editor. Neonatal cerebral ultrasound.
Cambridge University Press; 1997. 124.

9. Escobar GJ, Littenberg B, Petitti DB. Outcome among surviving very low birthweight
infants: a meta-analysis. Arch Dis Child. 1991 Feb;66(2):204-11.

10. Pinto-Martin J A, Whitaker AH, Feldman JF, Van Rossem R, Paneth N. Relation of

cranial ultrasound abnormalities in low-birthweight infants to motor or cognitive
performance at ages 2, 6, and 9 years. Dev Med Child Neurol. 1999 Dec;41(12):826-33.

90

11. Aylward GP, Pfeiffer SI, Wright A, Verhulst SJ. Outcome studies of low birth weight
infants published in the last decade: a meta analysis. J Pediatr. 1989 Oct;115(4):515-20.

l2. Whitaker A. H., Feldman J. F., J. A. Pinto-Martin, N. Paneth. Neonatal Cranial
Ultrasound Abnormalities in LBW infants: relation to cognitive outcomes at six years of age.
Pediatrics, vol 98, No.4. 1996

13. Paneth N., Rudelli R., Kazam E., Monte W. Brain damage in the preterm infant.
1994 Mac Keith Press

14. Whitaker A.H., Rossem R. V., J. A. Pinto-Martin, N. Paneth. Psychiatric Outcomes
in low-birth-weight children at age 6 years: relation to neonatal cranial ultrasound
abnormalities Arch G Psy vol 54, sep 1997

15. Hack M, Flannery DJ, Schluchter M, Cartar L, Borawski E, Klein N. Outcomes in
young adulthood for very-low-birth-weight infants. N Engl J Med. 2002 Jan
l7;346(3): 149-57.

16. Hack M, Taylor HG, Klein N, Mercuri-Minich N. Functional limitations and special
health care needs of 10- to 14-year-old children weighing less than 750 grams at birth.
Pediatrics. 2000 Sep;106(3):554-60.

17. Hack M, Fanaroff AA. Outcomes of children of extremely low birthweight and
gestational age in the 1990's. Early Hum Dev. 1999 Jan;53(3): 193-218.

18. Sai gal S, Szatmari P, Rosenbaum P, Campbell D, King S. Cognitive abilities and
school performance of extremely low birth weight children and matched term control
children at age 8 years: a regional study. J Pediatr. 1991 May;118(5):751-60.

19. Whitfield MF, Grunau RV, Holsti L. Extremely premature (< or = 800 g)
schoolchildren: multiple areas of hidden disability. Arch Dis Child Fetal Neonatal Ed.
1997 Sep;77(2):F85-90.

20. Volpe JJ. Neurology of the newborn Fourth Edition. 2001

21. Rakic PT. Embryonic development of the pulvinar-LP complex in man. In: Cooper
IS, Riklan M, Rakic PT (Eds) The pulvinar-LP complex. 1974, Springfield, IL: Thomas
CC: 3-35

91

 

 

22. Rakic PT, Sidman RL. Telencephalic origin of pulvinar neurons in the fetal human
brain. Zeitschrift fur Anatomic and Entwicklungsgeschichte. 1969; 129: 53-82

23. Hambleton G, Wigglesworth J S. Origin of intraventricular haemorrhage in the
preterm infant. Arch Dis Child. 1976 Sep;51(9):651-9.

24. Larroche JC. lntraventricular hemorrhage in the premature neonate. In Korobkin R,

Guilleminault C, editors: Advances in perinatal neurology, vol 1, New York, 1979, SP
Medical and Scientific Books.

25. Yakovlev PI, Rosales RK. Distribution of the terminal hemorrhages in the brain wall
in stillborn premature and nonviable neonates. In Angle CR, Bering EA Jr, editors:
Physical trauna as an etiologic agent in mental retardation, Washington DC, 1970, US
Government Printing Office.

26. Gmenwald P. Subependymal cerebral hemorrhage in premature infants, and its
relation to various injurious influences at birth. Am J Obstet Gynecol. 1951
Jun;61(6):1285-92.

27. Leech RW, Kohnen P. Subependymal and intraventricular hemorrhages in the
newborn. Am J Pathol. 1974 Dec;77(3):465-75.

28. Rorke LB. Pathology of perinatal brain injury, New York, 1982, Raven Press.

29. Paneth N. Classifying brain damage in preterm infants. J Pediatr. 1999
May; 134(5):527-9.

30. Jeffrey JN, Terrie El. Imaging perinatal brain injury in premature infants. Semin
Perinatol. 2004;28:433-443

31. McArdle CB, Richardson CJ, Hayden CK, Nicholas DA, Crofford MJ, Amparo EG
Abnormalities of the neonatal brain: MR imaging. Part I. Intracranial hemorrhage.
Radiology. 1987 May;l63(2):387-94.

32. Haddad J, Constantinesco A, Brunot B. Single photon emission computed
tomography of the brain perfusion in neonates. In Haddad J, Christmann D, Messer J,
editors: Imaging techniques of the CNS of the neonates, New York, 1991, Springer-Verlag.

92

33. Zuerrer M, Martin E, Boltshauser E. MR imaging of intracranial hemorrhage in
neonates and infants at 2.35 Tesla. Neuroradiology. 1991 ;33(3):223-9.

34. Barkovich AJ. Pediatric neuroimaging, ed 2, New York, 1995, Raven Press.
35. Krishnamoortlry KS, Fernandez RA, Momose KJ, DeLong GR, Moylan FM, Todres
ID, Shannon DC. Evaluation of neonatal intracranial hemorrhage by computerized

tomography. Pediatrics. 1977 Feb;59(2):165-72.

36. Scott WR, New PF, Davis KR, Schnur JA. Computerized axial tomography of
intracerebral and intraventricular hemorrhage. Radiology. 1974 Jul;112(1):73-80.

37. Pevsner PH, Garcia-Bunuel R, Leeds N, Finkelstein M. Subependymal and
intraventricular hemorrhage in neonates. Early diagnosis by computed tomography.
Radiology. 1976 Apr;1 19(1): 1 1 1-4.

38. Volpe JJ. Neonatal intracranial hemorrhage. Pathophysiology, neuropathology, and
clinical features. Clin Perinatol. 1977 Mar;4(1):77-102.

39. Rumack CM, McDonald MM, O'Meara OP, Sanders BB, Rudikoff JC. CT detection
and course of intracranial hemorrhage in premature infants. AJR Am J Roentgenol. 1978
Sep;13l(3):493-7.

40. Lee BC, Grassi AE, Schechner S, Auld PA. Neonatal intraventricular hemorrhage: a
serial computed tomography study. J Comput Assist Tomogr. 1979 Aug;3(4):483-90.

41. Albright L, Fellows R. Sequential CT scanning after neonatal intracerebral
hemorrhage. AJR Am J Roentgenol. 1981 May;l36(5):949-53.

42. Siegel MJ, Patel J, Gado MH, Shackelford GD. Cranial computed tomography and
real-time sonography in full-term neonates and infants. Radiology. 1983 Oct;149(1):111-6.

43. O'Shea TM, Counsel] SJ, Bartels DB, Dammann 0. Magnetic resonance and
ultrasound brain imaging in preterm infants. Early Hum Dev. 2005 Mar;81(3):263-71.

44. Tonukari NJ. Searching the online biomedical literature from developing countries.
African J of Biotechnology. 2005 Aug; 4 (8): 758-762

45. Detmer WM. Medline on the web: ten questions to ask when evaluating a web based

93

service. Internet Working Group Newsletter 1997; 3: 11-13.

46. Vincent B, Vincent M, Ferreira CG. Making PubMed searching simple: learning to
retrieve medical literature through interactive problem solving. Oncologist. 2006
Mar;11(3):243-51.

47. Wheeler DL, Barrett T, Benson DA, Bryant SH, Canese K, Church DM, DiCuccio M,
Edgar R, Federhen S, Helmberg W, Kenton DL, Khovayko O, Lipman DJ, Madden TL,
Maglott DR, Ostell J, Pontius JU, Pruitt KD, Schuler GD, Schliml LM, Sequeira E, Sherry
ST, Sirotkin K, Starchenko G, Suzek TO, Tatusov R, Tatusova TA, Wagner L, Yaschenko E.
Database resources of the National Center for Biotechnology Information. Nucleic Acids
Res. 2005 Jan 1;33(Database issue):D39-45.

48. Day I. The quest for information: a guide to searching the Internet. J Contemp
Dent Pract. 2001 Nov 15;2(4):33-43.

49. Dickersin K, Scherer R, Lefebvre C. Identifying relevant studies for systematic
reviews. BMJ. 1994 Nov 12;309(6964):1286-91

50. Murphy LS, Reinsch S, Najm WI, Dickerson VM, Seffinger MA, Adams A, Mishra SI.
Searching biomedical databases on complementary medicine: the use of controlled

vocabulary among authors, indexers and investigators. BMC Complement Altem Med.
2003 Jul 7;3:3. Epub 2003 Jul 7.

51. Coletti MH, Bleich HL. Medical subject headings used to search the biomedical
literature. J Am Med Inform Assoc. 2001 Jul-Aug;8(4):3l7-23.

52. Dunning J, Prendergast B, Mackway-Jones K. Towards evidence-based medicine in
cardiothoracic surgery: best BETS. Interactive Cardiovascular and Thoracic Surgery 2003;
2: 405-409

53. Pappas DE, Owen Hendley J. Otitis media. A scholarly review of the evidence.
Minerva Pediatr. 2003 Oct; 55(5): 407-14.

54. Harrison J. Designing a search strategy to identify and retrieve articles on
evidence-based health care using MEDLINE. Health Libr Rev. 1997 Mar;14(1):33-42.

55. Stewart MG, Kuppersmith RB, Moore AS. Searching the medical literature on the
Internet. Otolaryngol Clin North Am. 2002 Dec; 35(6): 1163-74, v-vi.

94

56. Weed DL. Methodologic guidelines for review papers. J Nation. Canc. Instit. 1997
Jan; 89(1): 6-7

57. Blettner M, Sauerbrei W, Schlehofer B, Scheuchenpflug T, Friedenreich C.
Traditional reviews, meta-analyses and pooled analyses in epidemiology. Int J Epidemiol.
1999 Feb;28(1):1-9.

58. Evans D. Database searches for qualitative research. J Med Libr Assoc. 2002
Jul;90(3):290-3.

59. Lowe HJ, Barnett GO. Understanding and using the medical subject headings
(MeSH) vocabulary to perform literature searches. JAMA. 1994 Apr 13;271(14):1103-8.

60. Dickersin K. The existence of publication bias and risk factors for its occurrence.
JAMA. 1990 Mar 9;263(10): 1385—9.

61. Dickersin K. How important is publication bias? A synthesis of available data.
AIDS Educ Prev. 1997 Feb;9(1 Suppl):15-21.

62. Rosenthal R. The “file-drawer problem” and tolerance for null results. Psychol Bull.
1979;86:638-41

63. Stewart LA, Parrnar MK. Meta-analysis of the literature or of individual patient data:
is there a difference? Lancet. 1993 Feb 13;341(8842):418-22.

64. Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and
hormonal contraceptives: collaborative reanalysis of individual data on 53 297 women with
breast cancer and 100 239 women without breast cancer from 54 epidemiological studies.
Lancet. 1996 Jun;347(9017): 1713-27

65. Stewart LA, Clarke MJ. Practical methodology of meta-analyses (overviews) using
updated individual patient data. Cochrane Working Group. Stat Med. 1995 Oct
15;14(19):2057-79.

66. Breslow RA, Ross SA, Weed DL. Quality of reviews in epidemiology. Am J Public
Health. 1998 Mal",88(3):475-7.

67. Weed DL, Gorelic LS. The practice of causal inference in cancer epidemiology.

95

 

Cancer Epidemiol Biomarkers Prev. 1996 Apr;5(4):303-11.
68. Weed DL. On the use of causal criteria. Int J Epidemiol. 1997 Dec;26(6):1137-41.

69. Hill AB. The environment and disease: association or causality? Proc R Soc Med.
1965;58:295-300.

70. Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical epidemiology--a basic
science for clinical medicine. Boston: Little, Brown, 1985.

71. Holland PW, Statistics and causal inference. J Am Statist Assoc. 1986;81:945-60.

72. Peterson HB, Kleinbaum DGt Interpreting the literature in obstetrics and gynecology.
1. Key concepts in epidemiology and biostatistics. Obstet Gynecol 1991 ;78:710-7.

73. Nicol-Smith L. Causality, menopause, and depression: a critical review of the
literature. BMJ. 1996 Nov 16;313(7067):1229-32.

74. Elwood J M. Causal relationships in medicine. A practical system for critical
appraisal. Oxford medical publications. 1988

75. Papile LA, Munsick-Bruno G, Schaefer A. Relationship of cerebral intraventricular
hemorrhage and early childhood neurologic handicaps. J Pediatr. 1983 Aug;103(2):273-7.

76. Merit LR, Scott DT, Ehrenkranz RA, Duncan CC. Neurodevelopmental assessment
of very low birth weight neonates: effect of germinal matrix and intraventricular hemorrhage.
Pediatr Neurol. 1985 May-Jun;1(3): 164-8.

77. Sostek AM, Smith YF, Katz KS, Grant EGt Developmental outcome of preterm
infants with intraventricular hemorrhage at one and two years of age. Child Dev. 1987
Jun;58(3):779-86.

78. Hanigan WC, Morgan AM, Anderson R], Bladle P, Cohen HS, Cusack TJ,
Thomas-McCauley T, Miller TC. Incidence and neurodevelopmental outcome of
periventricular hemorrhage and hydrocephalus in a regional population of very low birth
weight infants. Neurosurgery. 1991 Nov; 29(5):701-6.

79. Ross G, Boatright S, Auld PA, Nass R. Specific cognitive abilities in 2-year-old
children with subependymal and mild intraventricular hemorrhage. Brain Cogn. 1996

96

Oct;32(l):l-l3.

80. Boss G, Tesman J., Auld PAM, Nass R. Effects of Subependymal and Mild
lntraventricular Lesions on Visual Attention and Memory in Premature Infants.
Developmental Psychology 1992. V0128. No. 6. 1067-1074

81. Lowe J., Papile L.A. Neuridevelopmental Performance of Very-Low-Birth-Weight
Infants With Mild Periventricular, lntraventricular Hemorrhage. Outcome at 5 to 6 Years of
Age. AJDC Vol 144, Nov. 1990

82. van de Bor M, Verloove-Vanhorick SP, Baerts W, Brand R, Ruys J H. Outcome of
periventricular-intraventricular hemorrhage at 2 years of age in 484 very preterm infants
admitted to 6 neonatal intensive care units in The Netherlands. Neuropediatrics. 1988
Nov; 19(4): 1 83-5.

83. van de Bor M, Ens-Dokkum M, Schreuder AM, Veen S, Brand R, Verloove-Vanhorick
SP. Outcome of periventricular-innaventricular haemorrhage at five years of age. Dev
Med Child Neurol. 1993 Jan;35(1):33-41.

 

84. Sherlock R. L., Anderson P. J ., Doyle L. W. Neurodevelopmental sequelae of
intraventricular haemorrhage at 8 years of age in a regional cohort of ELBW/ very preterm
infants. E.H.D. (2005)

85. Kitchen WH, Ford GW, Rickards AL, Doyle LW, Kelly E, Murton LJ. Five-year
outcome of infants of birthweight 500 to 1500 grams: relationship with neonatal ultrasound
data. Am J Perinatol. 1990 Jan;7(l):60-5.

86. Levene M, Dowling 8, Graham M, Fogelrnan K, Galton M, Phillips M. Impaired
motor function (clumsiness) in 5 year old children: correlation with neonatal ultrasound
scans. Arch Dis Child. 1992 Jun;67(6):687-90

87. Hack M, Taylor HG, Klein N, Eiben R, Schatschneider C, Mercuri-Minich N.
School-age outcomes in children with birth weights under 750 g. N Engl J Med. 1994 Sep
22;331(12):753-9.

88. Pagano M., Gauvreau K. Principles of biostatistics. Duxbury press, 1993.

89. Bhopal R. Concepts of epidemiology- an integrated introduction to the ideas,
theories, principles and methods of epidemiology. Oxford University Press. 2002.

97

90. Rychetnik L, Hawe P, Waters E, Barratt A, Frommer M. A glossary for evidence
based public health. J Epidemiol Community Health. 2004 Jul;58(7):538-45.

91. Dickersin K. Systematic reviews in epidemiology: why are we so far behind? Int J
Epidemiol. 2002 Feb;3](1):6-12.

92. Mulrow CD. Rationale for systematic reviews. BMJ. 1994 Sep 3;309(6954):597-9.

93. Oxman AD, Guyatt GH. Guidelines for reading literature reviews. Can Med Assoc J
1988;138:697-703.

94. Cohen, J. Statistical power analysis for the behavioral sciences (2nd ed.). New
Jersey: Lawrence Erlbaum. 1988

95. Scargle JD. Publication bias: the “file-drawer” problem in scientific inference. J. of
Scientific Exploration. 2000; 14(1): 91-106.

 

96. http://sportsci.org/resourcdstats/index.html

98