5.8.? r...
k . 3
“fig:
fining“
I O b I Ii t“

A. a I! {mmmwfi

.1 ..w» . Quasi... ..ew
bifmmés ..a............¢.mn.
.3. a in...
.1... .9 .

 

 

   

 

 

 

  

                

  

 

l.’o

. v ...... .5.
.. $3.541.

      

I
-n a

I"

        

”L.

 

I
I

 

           

r
a

 

 

h§h.5....}hz.e.
.2... .22}... a?
. ...: . -. V
.....r....n..........v.......u.,..

 

 

 

      

 

 

 

                       

.3. «5.1!
1.5))»:

 

 

33:33:

                   

1 .1... ‘1. .
4r!‘~9$.:.t
i ,9..~A.u;aԤ\

     

   

.. .... m...
..fln... ....1..-¢.F..
1.; .5...
..
...... I... 1
Pl. Eight h
.l< ¢~ .vlnu‘lflni- .i‘Q-K3I1.
an... . 2.... Fit... "Nun,- .3."
01,3 1.. . 2.. ....
z and {5...}... _. ...».WC...
...... -..
p».

 

 

   
    

 
  

 

   

 

   

u...
Ruizntilxst ..I.
b.4484 .2...“ .0.
All) {gun ...-1

av Oil vrut

rl’l ..Iw. :31‘5 1:0": (0
x .3329, 519%”..o‘3litrl outfit-..."... uh"
4 .1332...n..g..l~.......r..wo1.riat..2 .3.
“.rigvaaeétslssg. p!.153.!. W
42.33.93... iii:
.‘A!!.‘ t».

 

 

     

          

      

  

 

      

 

. It. filal.§

.612): K E’s .- § ..l 3!“.- {Run

$41!“. 911.1,... I3I\’|3$;V.. gift. 1‘;

4n .5 ...: i 3.: .2
y . ... . .F. ...

 

{1.31.1 Id .Utl L

. 3.3.... l....x..ll..l.. vi?

13.79.23":le It)
.c$.u..lt..5.

i .)

 

 

       

..
9"“,114; ..
2......” 5......4... Lfidnmmfim

c 3‘! it...“ {rt-iii \: fur).-

      

                              

 

 

.r$. 99...}, .ihuwflflwnwfi

50:1»: 1:, 0
Runs. ” .... .71.»! (a
if ‘1. I. ..nLithflfi (...!le .16.

 

  

...}? Cl
-3 ‘ . Ivllrtiq 1-6.: ». ...:
..n‘v ian ...... .. ...I\ CIA: 3.4.1:.
$..J.r..l|ovs.l$.l.:é.x
...:

x

        

 

     

      

     

 

.: S ...
c. :6
A. 3. .

{gamut ..
.v . ...:nuu

 

      

..finkfi.§

...\n73.\l\v|l.¢lfi'.1.l {1‘ X\ . -

 

gfikmw

II

 

THESb

 

2am

 

Illllllllllllllllllllllllllllllllllllllllllllllllllllilllll

LIBRARY 3 1293 02048 8965
Michigan State
University

 

 

This is to certify that the

thesis entitled

Prevalence and Risk Factors of Sensori-
neural Hearing Loss in Low Birth Weight
Children

presented by

Maria Lourdes Coronado

has been accepted towards fulfillment
of the requirements for

Masimis— degree in _Epidem_i_ology

Major professor

Date z/{q/OO

0-7639 MS U is an Aflirmative 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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11100 momma-p.14

 

PREVALENCE AND RISK FACTORS OF SENSORINEURAL HEARING LOSS IN
LOW BIRTH WEIGHT CHILDREN

By

Maria Lourdes Coronado

A THESIS

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

MASTER OF SCIENCE

Department of Epidemiology

2000

ABSTRACT

PREVALENCE AND RISK FACTORS OF SENSORINEURAL HEARING LOSS IN
LOW BIRTH WEIGHT CHILDREN

By

Maria Lourdes Coronado

Background--The prevalence rates of sensorineural hearing loss reported in the literature
among low birth weight populations have been found to be higher than in the normal
birth weight population, but have varied substantially from study to study. In infants
<2,500g, prevalence rates of sensorineural loss ranged from 15 to 190 per 1000
(Campenelli et al, 1958; Clarke et al, 1986; Abramovich et al, 1979; Salamy et al, 1989;
Veen et al, 1993; Wright et al, 1972; Swigonski et al, 1987; Doyle et al, 1992; Leslie et
al, 1995; McClelland et al, 1992; Winkel et al, 1978; Bergman et a1, 1985; Agnostakis et
al, 1982). Low birth weight infants represent an important population in which to study
the risk factors for hearing impairment.

Methods--The main objective of this thesis was to report the prevalence rate of hearing
loss in a low birth weight population. A literature review was first completed in order to
summarize the prevalence rates and different definitions of hearing loss utilized in other
studies. All available hearing variables in the NBH study were then assessed in terms of
their validity and reliability. While this is primarily a descriptive analysis, the author
hopes to suggest risk factors in need of further evaluation and improvements in methods

for assessing sensorineural hearing loss.

Results--The principal findings of this study are 1) hearing loss is more prevalent in the
low birth weight population, 2) maternal report of hearing loss is a poor predictor of 2
year olds failing to respond to l or more VRA stimuli, 3) 32 children (4.6%) failed to
respond to l or more VRA stimuli, 9 (1.2%) children needed or were prescribed hearing
aids by age 9, and 9 (1.4%) children were defined by 1 audiologist as having
sensorineural hearing loss, (4) children with germinal matrix hemorrhaging were 4.9 (1 .3-
184) times more likely to have severe hearing loss requiring a hearing aid than those
without germinal matrix hemorrhaging, (5) children with white matter damage were 6.0
(LS-24.7) times more likely to need hearing aids and were 7.0 (2.0-24.2) times more
likely to have sensorineural or untyped hearing loss defined by l audiologist than those

without white matter damage.

Copyright by
Maria Lourdes Coronado
2000

To all the dedicated professors of epidemiology whose wisdom is passed on to the next
generation of epidemiologists

ACKNOWLEDGMENTS

A publication such as this would not be possible without the guidance of faculty
members in the departments of Epidemiology and Audiology at Michigan State
University. Their dedication to the sciences and efforts in review of this thesis have
improved my understanding of the discipline, dedication, and techniques required in this
exciting and at times frustrating process of manuscript submission. Acknowledgment
must also be given to the secretaries of the department of Epidemiology whose reminders
of deadlines expedited my defense, to those colleagues who supported me in continuing
this endeavor, to my husband for his constant support, to my family, and to the families
participating in the NBH study.

Finally, I would like to give a special thanks to those specific individuals who
shared an enormous amount of effort with editions and collection of data. This thesis
could not be completed without their help. I will always be indebted to Dr. Paneth, my
thesis advisor, whose patience, support, and wisdom allowed me to grow not only as an
epidemiologist but a scientific writer. To Dr. Holzman, another member of my thesis
committee, I am grateful for giving me the understanding of causation and the ability to
orally present results. I thank Dr. Pathak, also a thesis committee member, for teaching
me the necessary statistical tools to analyze scientific data. Dr. Elfenbein, the fourth
member of my thesis committee, and Dr. Gravel I owe my appreciation of hearing.
Lastly, my gratitude goes to Dr. Whitaker, Dr. Pinto-Martin, and again Dr. Paneth, who

have ascertained 10 years of information, for allowing me to learn from the NBH data.

vi

TABLE OF CONTENTS

LIST OF TABLES ................................................................................. viii
LIST OF FIGURES .................................................................................. x
LIST OF ABBREVIATIONS ...................................................................... xi
INTRODUCTION .................................................................................... 1
THE LITERATURE ON HEARING LOSS IN LBW INFANTS ............................. 3

NEONATAL BRAIN HEMORRAGE STUDY: VALID HEARING MEASURES. . ....30

HEARING LOSS IN THE NBH STUDY ....................................................... 45
VALIDITY OF MATERNAL REPORT OF HEARING PROBLEMS ...................... 61
FUTURE RESEARCH IN NBH HEARING ................................................... 73
APPENDICES ....................................................................................... 94
REFERENCES ..................................................................................... 104

vii

LIST OF TABLES

1.] Hearing Loss Studies in Low Birth Weight Infants Using Single Source

1.2 Hearing Loss Studies in Low Birth Weight Infants Using Conditional Multiple
Sources

1.3 Hearing Loss Studies in Low Birth Weight Infants Using Unconditional Multiple
Sources

1.4 Hearing Loss in Low Birth Weight Infants Using Studies Which Could Not Be
Included in The Above Categories Or Were Obtained from Secondary Sources

2.1 Descriptive Analysis of Overall Population, Deaths, and Survivors

2.2 Descriptive Analysis of Survivors

2.3 Descriptive Analysis of Survivors and Children with Hearing Screen

2.4 Descriptive Analysis of Survivors and Children with Medical Abstraction Forms
2.5 Descriptive Analysis of Survivors and Children with Age 2 Maternal Interviews
2.6 Descriptive Analysis of Survivors and Children with Age 6 Maternal Interviews

2.7 Descriptive Analysis of Survivors and Children with Hearing Aid Information at Age
2 or 9 Maternal Interviews

3.1 Frequency of Hearing and Middle Ear Problems by Source @Age 2

3.2 Frequency of Cases with the Following Middle Ear Dysfunction or Hearing Problems
at Age 6

3.3 Frequency of Hearing and Middle Ear Problems by Source @Age 9
3.4 Assessing Incidence of Hearing Problems Reported by Mothers
3.5 Prevalence Rates of 4 Possible Case Definitions of Sensorineural Hearing Loss

4.1 The Association between Maternal Report of Hearing Problems with Maternal Report
of Ear Infections and Ear Surgeries at Age 2

viii

4.2 The Association between Maternal Report of Hearing Problems with Any Decibel

Loss in the Audiometry Exam at Age 2, Stratified by Ear Infections

4.3 The Association between Maternal Report of Hearing Problems with Concerns on

Medical Chart and Medical Chart Abstractor-Judged Hearing Loss

5.1 Hemorrhage in Relation to Hearing Loss

5.2 Respiration in Relation to Hearing Loss

5.3 Bilirubin in Relation to Hearing Loss

5.4 Infections in Relation to Hearing Loss

5.5 Homeostatic Imbalances in Relation to Hearing Loss

A.

B.

C.

Descriptive Analysis of Survivors with and without Prenatal Health Forms
Excluding Children Tested at Higher Screening Levels

Sensitivity and Specificity Values of Needing or Wearing Hearing Aids Substituted
for Sensorineural Hearing Loss

Coronado Program for SPSS to Calculate Multiple Odds Ratios with 95% Confidence
Intervals from Literature Reviews Using the Woolf Method

Variables of Risk not Biologically Plausible or Consistently Insignificant in Relation
to Hearing Loss

Estimated Cases Using Sample Size Calculations

LIST OF FIGURES

1.1 Factors Indicating High Risk for Hearing Loss for Infants 0-29 Days Old (from the
Joint Committee of Infant Hearing)

1.2 Assessment of Hearing in Children

1.3 Prevalence Rates by Source Groups

2.] Exact Sources of Information about Hearing Problems

4.1 Odds Ratios for the Association of Report of Hearing Problems with Ear Infections
5.1a Blood Upon the Tip of the Cupula of the Canals

5.1b Blood on the Tip and Sides of the Cupula

5.2 Blood around the Semicircular Canal

5.3 Massive Hemorrhage Especially in the Perilymphatic Spaces

LIST OF ABBREVIATIONS

ABR - Auditory Brainstem Response Audiometry
AC - Air Conduction (Pure Tone Audiometry)
ACR - Acoustic Reflex

B - Bilateral

BC - Bone Conduction (Pure Tone Audiometry)
BOA - Behavioral Observation Audiometry
CHL - Conductive Hearing Loss

CPA - Conditioned Play Audiometry

CPAP - Continuous Positive Airway Pessure
CPPV - Continuous Positive Pressure Ventilation
CSF - Cerebrospinal Fluid

CI - Confidence Interval

dB - Decibel

dB SPL - Decibel Sound Pressure Level
ECOG - Electrocochleography

EEG - Electroencephalogram

ENT - Ear, Nose, Throat

EPA - Conventional Earphone Audiometry

F BW - Full Birth Weight

GM - Germinal Matrix

HI - Hearing Impairment or Loss

HL - Hearing Level

Hz - Hertz

ICN - Intensive Care Neonatal

ICU - Intensive Care Unit

IPPV - Intermittent Positive Pressure Ventilation
IRDS - Infant Respiratory Distress Syndrome
IVH - Intraventricular Hemorrhage

LBW - Low Birth Weight

MRL - Minimum Response Level

N - Population

11 - Sample

NBH - Neonatal Brain Hemorrhage

NICU - Neonatal Intensive Care Unit

OAE - Otoacoustic Emissions Test

OR - Odds Ratio

PAM - Post Auricular Myogenic

PE - Parenchymal Echodensities

PEL - Parenchymal Lesions

RD - Respiratory Distress

Resp - Respiration

xi

sd - Standard Deviation

SL - Sensation Level

SNHL - Sensorineural Hearing Loss

TORCH - Toxoplasmosis, Rubella, Cytomegalovirus, Herpes
TYMP - Tympanogram

VE - Ventricular Enlargement

VLBW - Very Low Birth Weight

VRA - Visual Reinforcement Audiometry

WHO - World Health Organization

xii

INTRODUCTION

The prevalence rates of sensorineural hearing loss reported in the literature among
low birth weight populations have been found to be higher than in the normal birth
weight population, but have varied substantially from study to study. In infants <2,500g,
prevalence rates of sensorineural loss ranged from 36 to 190 per 1000 (Campenelli et al,
1958; Clarke et al, 1986; Abramovich et al, 1979; Salamy et al, 1989; Veen et al, 1993;
Wright et al, 1972; Swigonski et a1, 1987; Doyle et al, 1992; Leslie et al, 1995;
McClelland et al, 1992; Winkel et al, 1978; Bergman et al, 1985; Agnostakis et al,
1982). Two studies by Campenelli et a1 (1958) and Clarke et al (1986) found none of the
children had sensorineural hearing loss who were born weighing >2,500g. Low birth
weight infants represent an important population to study the risk factors for hearing
impairment.

Thirty to 60 percent of the causes of hearing loss remain unknown (McCormick,
1988) and the number of research studies on hearing in the low birth weight population
remains low. Although over 65 causal variables in the low birth weight population have
been evaluated through univariate analysis by researchers, very few are measured
consistently in each study (Veen et al, 1993; Doyle et al, 1992; Campenelli et al, 1958;
Agnostikas et al, 1982; Salamy et al, 1989; McDonald, 1964; Abramovich et al, 1979;
Leslie et al, 1995; Halpem et al, 1987; Bergman et al, 1985). Additionally, only 5
studies had enough power to control for confounding using multivariate analysis.

The aims of this thesis are to create case definitions of sensorineural hearing loss,

report the prevalence rate of hearing loss in a low birth weight population, and suggest

I

improvements of measure. A literature review was first completed in order to compare
the prevalence rates along with the different definitions utilized in other studies. All
available hearing variables in the NBH study were then assessed in terms of validity and
reliability. Prevalence of the best case definitions were then analyzed. While this is
primarily a descriptive analysis, the author hopes to suggest risk factors in need of further

evaluation and improvements in order to assess sensorineural hearing loss.

Chapter 1

THE LITERATURE ON HEARING LOSS IN LBW INFANTS

C 0MPARABILIT Y
Incorporated within hearing research are differences in methodology. In order to
compare the results in the literature, 4 methodological issues must first be addressed:

population selection, ascertainment of cases, study design, and analysis.

POPULATION SELECTION

Population selection has varied widely in hearing studies using low birth weight
infants. Inclusion criteria have been based not only on weight, but also on gestational age
and on risk factors associated with hearing impairment. Although most studies use
conventional weight categories such as <2,500g for low birth weight, <1,500g for very
low birth weight, and <1 ,000g for extremely low birth, other categories such as <2,000g,
<1 ,800g, and <1 ,250g have also been used. Gestational age inclusion criteria have ranged

from <28 weeks to <38 weeks. Some risk factors used for population selection have been

 

FIGURE 1.1: Factors Indicating High Risk for Hearing Loss for Infants 0-29 Days
Old (from the Joint Committee on Infant Hearing)

 

Family History of Hereditary, Childhood Hyperbilirubinemia Requiring
Sensorineural Hearing Loss Transfirsions
Congenital Perinatal Infection Bacterial Meningitis
(cytomegalovirus, rubella, herpes, Apgar<4 @ 1min or <6 @ 5 min
toxoplasmosis, syphilis, etc.) Ototoxic Medications
Craniofacial Anomalies Mechanical Ventilation >5 days
<1 ,500g Birth Weight Stigmata or Syndromes Associated with

Hearing Loss

 

American Speech-Language-Hearing Association, 1994

 

3

ototoxic drugs, seizures, and elements of the High Risk Registry created by the Joint

Committee on Infant Hearing (Figure 1.1).

CASE ASCERTAINMENT AND HEARING MEASUREMENT

Adding to difficulties of comparability of studies are the diverse methods of
ascertaining caseness, including factors such as age at assessment, measurement of
hearing loss, and case definitions. Age of hearing assessment in the low birth weight
population has been as early as 6 months (Bergman et al, 1985; Halpem et a1, 1987) and
as late as 12 years (Abramovich et al, 1979). A majority of researchers begin assessing at
<9 months (Salamy et al, 1989; Clarke et al, 1986; Agnostakis et al, 1982; Veen et al,
1993; Wright et al, 1972;.Swigonski et al, 1987; Doyle et al, 1992; Leslie et al, 1995;
McClelland et a1, 1992). Others use 1 to 3 years (Campenelli et al, 1958; Winkel et al,
1978; Halpem et al, 1987) and some over 4 years (Abramovich et al, 1979; Bergman et
al, 1985).

TeChniques for measurement of hearing loss depend on the developmental skills
of the child, and therefore, on age. Audiometric exams include conventional earphone
audiometry (EPA) with voluntary response such as hand raise, conditioned play
audiometry (CPA), visual reinforcement audiometry (VRA), and behavioral observation
audiometry (BOA). The first 2 audiology exams require cooperation of the child.
Children who are under 3 years of age or developmentally delayed may be assessed with
the audiometer using the latter 2 options. The BOA, however, is no longer recommended

by the ASHA (1997).

Recently advances have been made in creating objective tools to measure hearing
in the newborn (Buttross et al, 1995). Two widely used screening devices are the
otoacoustic emissions test (OAE) and auditory brainstem response audiometry (ABR),
(Arehart et al, 1998). For very young children, ABR is the diagnostic tool of choice.
Figure 1.2 lists some hearing tests with brief descriptions, minimum assessment age,

developmental skills required, and limitations of each.

 

FIGURE 1.2: Assessment of Hearing in Children

 

Test Description Age Skill Limitation
EPA Voluntary Response-Hand 35 yr. High Need Compliance
Raise
CPA Reinforced Response-Game 3-5 yr. Moderate Need Compliance
VRA Reinforced Response-Head .5-3 yr. Low Need Compliance
Turn
*BOA Reflexive Response-Startle 0-3 yr. Low High False Positives
OAE Cochlear Otoacoustic 0-5 yr. None Some False Positives
Emission
ABR Neural Response 0-.5 yr. None Some False Positives

 

Buttross, et al, 1995.
* No longer recommended by the ASHA, 1997.

 

Focusing attention mainly on studies using audiometric tests did not eliminate all
issues of comparability. Research studies differed in their definitions of a case of hearing
loss, and also in categorizing the severity and the type of loss. Some studies used a mild
loss or greater category such as >20 dB HL, while others used a moderate loss or greater
category of >40 dB HL. Very few studies used the WHO classification of mild (26-40
dB HL), moderate (41-60 dB HL), severe (61-80 dB HL), and profound (_>_81 dB HL), and
only 1 study classified hearing loss using the WHO definitions of impairment, disability,

and handicap (Gell, 1992).

STUDY DESIGN AND ANALYSIS

In comparing studies of risk factors, the study design and level of analysis must be
considered. Although the majority of hearing loss studies of the low birth weight
population had cohort designs with prospective follow-up, some retrospectively
ascertained risk factors were obtained years later from available records, rather than by
collecting data from the entire population before hearing was tested. Because of the low
prevalence of hearing loss even in this high risk population, most research was
constrained to providing univariate or descriptive analysis. With these challenges in

mind, prevalence rates will be compared in the next section.

LITERATURE REVIEW: PREVALENCE OF HEARING LOSS IN LOW BIRTH WEIGHT
0R INTENSIVE CARE UNIT INFANTS

This part of the chapter will focus on a literature review of prevalence rates of
hearing loss in low birth weight or intensive care unit populations. The prevalence rates
of sensorineural hearing loss among extremely (<1,000g), very (<1,800g included with
<1 ,500g in this review), and low birth weight (<2,000 included with <2,500g in this
review) populations are high, but vary across studies.

Studies of prevalence rates were summarized if the populations being followed
contained low birth weight or intensive care unit infants from developed countries. To
compare the following studies, 4 categories were created. Research about hearing loss
was categorized, according to type of measures used, as single source, conditional
multiple source, unconditional multiple sources, and other. 1) Studies in the single

source group performed the same hearing screen or evaluation by audiometer or ABR on

6

all survivors with no other follow-up assessments. 2) The conditional multiple source
group includes studies that used an audiometer or ABR screen and then assessed children
during follow-up only if they failed the screening exam. The initial screen results should
be comparable to the single source studies. 3) Those in the unconditional multiple source
group defined the hearing of survivors with more than 1 type of measure either in a single
examination or in several follow-up examinations. Studies will be ordered by birth
weight within the groups. Studies using the same birth weight criterion will be ordered
by screening levels used in hearing assessments. 4) Another section, called other, will
report results from studies which could not be included in the above categories, or were
not obtained directly but from a review in another paper.

Using only studies with sample selection exclusively by low birth weight groups
or by receipt of care in newborn intensive care units, sensorineural hearing loss rates were
reported to be 95 per 1000 extremely low birth weight infants (Doyle et al, 1992), 36 to
97 per 1000 very low birth weight infants (Bergman et al, 1985; Abramovich et al, 1979;
Salamy et al, 1989; Cox et al, 1992), 59 to 190 per 1000 low birth weight infants (Winkel
et al, 1978; Campenelli et al, 1958; Clarke et al, 1986; Agnostakis et al, 1982), and 15 to
1 18 per 1000 very low birth weight and/or very pre-term infants (Veen et al, 1993; Leslie

et al, 1995).

SINGLE SOURCE
Three measures used by studies in the single source group were auditory
brainstem response (ABR), conventional earphone audiometry (EPA) with voluntary

response, and behavioral observation audiometry (BOA). Auditory brainstem response

7

screening in newborns has been shown to be highly specific (88%) and sensitive (100%)

to sensorineural hearing loss (McClelland et al, 1992). The strongest asset of the ABR is
its ability to assess hearing in the newborn. This helps to prevent loss to follow-up being
an issue when examining the hearing of the low birth weight population. The first study

reported here utilized the ABR.

Galambos & Despland (1980) assessed infants between the ages of 26 and 42
weeks in their isolettes or open bassinets. The thresholds of adults with normal hearing
were used to establish a 0 dB reference level in the room. They found that 14 (14.0%) of
100 unselected infants in the intensive care nursery had abnormal auditory brainstem
evoked response in at least 1 ear (threshold ABR appears at an intensity of 340 dB).
Three infants produced no ABR to clicks beyond the limit of the testing apparatus (90 dB
threshold). Nine (9.0%) had SNHL with responses at 45 - 75 dB threshold and 2 (2.0%)
had CHL approaching 40 dB threshold (Table 1.1). Blinding and training of examiners
were not reported in the methods section. Because the 100 infants represented 97% of
those admitted to the ICN unit during the 5 month period, selection bias was unlikely.

Conventional earphone audiometry (EPA), using pure-tone audiometry with
earphones and voluntary response such as hand raising or button pressing, is the “gold
standar ” measure for children around the age of 5. This approach was used by most of
the studies in the single source group. The use of earphones and voluntary responses
requires a higher developmental level and more cooperation of the child than other
assessments, such as the ABR and BOA, which require only involuntary responses.
Because children of low birth weight incur more neurological damage than do children of

normal weight, a higher proportion of the low birth weight population may have to be

8

excluded, or have less reliable evaluations. It should also be noted that with any follow-
up assessment in the low birth weight population, loss of subjects may become
disproportionately large over time. As the period of assessment becomes farther from the
initial time of assembly of the cohort, those who are most impaired may be lost. In the
opposing direction, an overestimation of those with hearing loss may occur if a higher
proportion of mothers who suspect hearing impairment volunteer to participate in the
study.

Clarke et al (1986) studied children of an average age of 6.5 years. A sound-proof
booth was available for the hearing evaluation, which used an audiometer to test air-
conduction and, when necessary, bone-conduction in each ear. Of 205 children weighing
$2,500g, 34 (16.6%) had moderate to severe hearing loss (>45 dB HL at any frequency
from .5 - 4000Hz), only 12 (5.9%) children had sensorineural hearing loss, bilateral in 8
(3.9%). Sensorineural hearing loss was defined as bone-conduction (BC) hearing levels
resembling air-conduction (AC) while conductive hearing loss had normal BC hearing
level. Thirteen (6.3%) low birth weight children had conductive hearing loss. Minimal
high-frequency hearing loss, defined as >30 dB HL at 4000Hz or >35 dB HL at 6000
and 8000Hz was reported in 9. The prevalence of hearing impairment in a control group
of 123 children who weighed more than 2,500g at birth was much lower. Children of
higher birth weight experienced no sensorineural loss. Only 3 (2.4 %) children had
conductive hearing loss and 5 (4.0%) had minimal high-frequency loss. Because the
author who performed the hearing tests could not be blind to all variables, a reduction in

internal validity may have been produced. The present author believes that this study has

the best methodological description of sensorineural typing and has results which can be
generalized both to infants of normal weight and low birth weight.

Campenelli et a1 (1958) assessed children between the ages of 6 and 7 years using
a clinically calibrated audiometer by obtaining AC HL from 125 Hz to 8000 Hz and BC
HL from 250 Hz to 4000 Hz in a sound-treated testing room. Children with otologic
disease (defined by the author as pathology present or suspected during a routine
otolaryngological inspection) were excluded from the study. Mean air, bone, and speech
thresholds were used to categorize loss by severity. The mean threshold measurements
for AC and BC were plotted for each group (control versus cases) rather than each
individual. They found 7 (15.9%) of 44 children weighing 52,500g at birth possessed
moderate to severe loss, presumably sensorineural in type. No sensorineural losses were
discovered in a control group of 44 infants weighing >2,500g. Differences between
Clarke et al (1986) and this study may be due to the different time period of the study,
typing of sensorineural hearing loss, and different calibration methods used on
audiometers in the 1950’s. Description of blinding of examiners was not reported.
Generalizability was strong since all infants identified were included in the study.

Winkel et a1 (1978) found 17 (19%) of 91 infants weighing 52,000g (excluding
infants given ototoxic drugs other than kanamycin) had sensorineural hearing loss
(defined according to ISO R389 (1967)) >20 dB HL with one or more stimulus
frequencies from 125 to 8000 Hz using EPA when children were assessed between 4 and
6 years of age. The air/bone audiometry evaluation was performed by 2 experienced
audiometry assistants in a sound proof chamber. Severe hearing loss was defined as

hearing loss affecting social functioning, where hearing aid is required and bilateral

lO

threshold is >30 dB HL in the conversation area. Children with conductive hearing loss
were excluded from the study. Five (out of 10) cases in the kanamycin group were
classified as moderate to severe hearing impairment while all cases in the group without
kanamycin were classified as slight hearing loss. The authors did not describe the
definition based on ISO R389.

Behavioral observation audiometry (BOA) is the easiest play audiometric test for
younger children. A stimulus is presented through speakers. A pass and fail is dependent
on the observation of an unconditioned response. Unlike the EPA where a threshold can
be measured the BOA can only obtain the minimum response level (MRL). Because it
takes a louder stimulus to elicit involuntary detectable changes, the MRL is normally
higher than the threshold. This exam is based on the reflexive response (Flexer, 1994).
The BOA is no longer recommended by the American Speech-Language-Hearing
Association (ASHA, 1997) as a screening tool but was used by the following study.

Johnson & Ashurst (1990) found 132 (32%) of 41 16 infants, who weighed
<2,000g or >2,000 g but admitted to a special care unit, failed the distraction test at age 1.
Failure was defined as no response by the infant to a sound made at lor both ears. At age
3, audiometric testing was given to each child and sensorineural hearing loss was defined
as loss >30 dB MRL at any frequency from .5 - 4000Hz in at least 1 ear or children
requiring a hearing aid with SNHL. Seven (.6%) of the 1133 children born <2,000g had
sensorineural hearing loss according to audiometric testing, while 11 (.3%) of 3,394 who
weighed 22,000g had sensorineural hearing loss. As part of their routine exam, health

visitors performed the distraction tests but the methods were not carefully described by

the authors. Blinding of the examiners, typing of loss, and environment of testing site
were not reported.

Agnostakis et al (1982) used a screening level of 30 decibels, with assessment
made at around 6.5 years, and birth weight below 1,800g used as the sample selection
criterion. In 98 children, the frequencies of any (unilateral or bilateral) sensorineural
hearing loss of >30 dB HL at 500 - 8000Hz were 9 (9.2%) and 5 (5.1%) respectively.
Pure-tone audiometry and impedance (an indication of middle ear dysfirnction)
audiometry were performed on 75 (77%) of the survivors. Only impedance audiometry
results were available for uncooperative children. Information on examiner blinding,
specific audiometric procedure used, typing of loss, and environment of test site were not
available. Internal validity is difficult to assess.

A Dutch study by Veen et al (1993), utilizing <32 weeks gestational age or
<1 ,500g as the inclusion criteria found, of 890 (94% of all possible) survivors at age 5,
136 (15.3%) had hearing loss. Only 13 (1.5%) had sensorineural hearing loss, of which 8
(.9%) were bilateral. One hundred and twenty-three (13.8%) children had conductive
hearing loss, of which 43 (4.8%) were bilateral. Hearing loss was defined as average
thresholds of >26 dB HL at 500, 1000, 2000, or 4000Hz. The assessments were
conducted by 3 trained pediatricians in the homes of the children and without a sound
proof booth. Air/bone-conduction was used to type hearing loss as sensorineural,
conductive, or mixed in each ear. Blinding was not well defined. Hearing loss was lower
in prevalence than in previous studies. Because this cohort is nationwide, the study is
very representative of the Dutch infants with <1 ,500g weight or <32 weeks gestational

age.

 

TABLE 1.1: Hearing Loss Studies in Low Birth Weight Infants Using Single

 

 

 

 

 

 

 

 

 

 

Source
Study Age N Selection Prevalence per Test dB
Criteria 1000 Survivors
Children’s 26-42 100 intensive care H1 140 ABR >30
Hospital wks unit SNHL 90 HL
Galambos & CHL 20
Despland, 1980
Vancouver 6.5 205 <2500g SNHL 59 EPA, >45
Clarke et al, years CHL 63 BC HL
1986 123 >2500g SNHL 0
CHL 24
Washington DC 72-84 44 <2500g excluding SNHL 159 EPA,
Campenelli mos otologic disease BC
et al, 1958 44 >2,500g SNHL 0
Rhigshospital 4-6 91 <2000g SNHL 190 EPA, >20
Winkel et al, years excluding drugs BC HL
1978 except kanamycin
Oxford Region 1 year 4116 All 32 HI DT
Johnson & 3
Ashurst, 1990 years 1133 <2,000g SNHL 6 BOA >30
MRL
3394 32,000g in SNHL 3 BOA
special care
nursery
Athens 6.5 98 <1800g SNHL 92 EPA, >30
University years BSNHL 51 IMP HL
Agnostakis
et al, 1982
POPS 5 890 <1500g or H1 153 EPA, >25
Veen et al, 1993 years <32weeks SNHL 15 BC HL
CHL 138
Stanford 6mos 820 <1500g, with risk Cochlear BOA
University -3 factors Loss 61
Halpem et al, years
1987
Royal Women 8 42 <1000g SNHL 95 EPA >25
Doyle et al, years CHL 119
1992

 

 

 

 

 

 

 

 

B=bilateral or better ear, SNHL=sensorineural hearing loss, CHL=conductive hearing
loss, HI=hearing impairment or loss, EPA=conventional earphone audiometry,
BC=bone conduction, DT=distraction test, BOA= behavioral observation audiometry,
ABR=auditory brainstem response, IMP=Impedance, italics: sound-proof chamber.

 

 

Halpem et al (1987) used at least 1 of the High Risk Registry factors (Gerkin,
1984) as the inclusion criteria for their study. Of 820 survivors from the intensive care
nursery and successfully followed, 50 (6.1%) possessed cochlear hearing loss, based on
behavioral sound field testing (specific type of test was not defined) done between the
ages of 6 months to 3 years. Screening level, differentiation of cochlear loss, blinding of
examiners, instrumentation, and use of sound-proof room was not recorded. Internal
validity is difficult to assess. This sample size used, however, is one of the largest to
record hearing loss in VLBW infants.

Doyle et a1 (1992), in a sample of 42 survivors weighing <1 ,000g who had pure-
tone audiometry and impedance assessments at 8 years of age, found 4 (9.5%) children
had sensorineural loss and 5 (11.9%) had conductive loss. Only 1 (2.3%) child had a
bilateral loss of 25 decibel or more in 2 or more frequencies. Audiology exams were
performed in a sound-proof booth by blinded audiologists. Hearing impairment in at least
1 ear was defined as >25 decibel loss. Typing of sensorineural and conductive hearing
loss, stimulus frequencies used, and dB referents (e.g., HL, SPL) were not reported.

External validity may be weakened by the small sample size.

CONDITIONAL MULTIPLE SOURCES

McClelland et a1 (1992), found 85 (21.0%) of 405 neonates in an intensive care
unit failed hearing screens. The hearing screening consisted of ABR, where abnormality
was defined as >30 dB normal hearing level (nHL). The infants were tested in their
bassinets. All failures were referred to a regional audiology center. Follow-up

information was gathered for more than 5 years. Definitive audiological assessments

l4

were performed on 62 of the survivors who failed the screen. With a sample of 299
intensive care unit survivors at age 3, 5 (1.8%) had severe bilateral sensorineural hearing
loss and 12 (4.0%) had conductive loss requiring surgery. Seventy-three percent of
children who passed were examined at 9 months, 3 years, and first year of school through
the clinical medical officer service and confirmed to have no hearing impairment.
Otoscopic examination, pure-tone audiometry, acoustic impedance exam, or the ABR
were performed during the definitive audiology assessment. Typing of loss was not
clearly described, test environment, and blinding of examiners were not mentioned. This
cohort was one of the larger studies to assess hearing (Table 1.2).

Watkin et al (1991) found 12 (3.7%) children with 2 moderate hearing loss among
322 screened with ABR as neonates with at least 1 of the risk factors on the risk registry
of the American Joint Committee on Infant Hearing (Gerber, 1990) in addition to
gestational age of <32 weeks, needing more than 4 hours of ventilation, and with cerebral
illness symptoms. Hearing loss was based on follow-up testing. Infants who failed were
given a diagnostic audiometric assessment 4 to 6 weeks later which consisted of the BOA
and impedance audiometry testing. Children who passed were reassessed by an
experienced audiologist at 7 months using the distraction test and impedance audiometry.
Children not at risk at 7 months were first screened by a parental questionnaire of age
expected normal auditory behavior and by the infant distraction test performed by the
health visitor. None of the 596 children without risk factors, who were also assessed, had
bilateral hearing loss of moderate severity or worse. Distraction test, blinding of

examiners, and testing site were not described by the authors.

l5

Roberts et al (1982) found 3 (2.3%) of 128 infants in the intensive care unit with
confirmed moderate to severe sensorineural impairment. Neonates were tested 2 or more
times in their bassinets using ABR. Both ears were tested, and failure was defined as no
response at 40 dB nHL or 70 dB nHL. In approximately 70% of infants who failed,
definitive audiometric testing 6 months later was obtained. Thirteen mail or telephone
surveys of report of non-response to noise were substituted for missing follow-up data.
All mothers reported that their babies responded well to noise. Typing of sensorineural
hearing loss, blinding of examiners, and environment of testing site were not reported.

Stein et a1 (1983) found 20 (20%) of 100 infants from the neonatal intensive care
unit failed to show a response at 40 dB nHL or 60 dB nHL to the ABR screen. Infants
were tested in a quiet room next to the NICU area in their bassinets or incubators.

During follow-up of those with no response to 60 dB nHL, 2 (2.0%) were confirmed to
have severe to profound sensorineural hearing loss. Definitive exams consisted of ABR,
impedance, and or behavioral testing and were completed about 7 months after discharge.
This population was comprised of infants weighing >2000g (40%), from 1500 to 2000g
(31%), from 1000 to 1500g (21%), and <1000g (8%). Typing of sensorineural loss and
blinding of examiners were not described.

Duara et a1 (1986) found 35 (12.6%) of 278 infants in a tertiary care unit failed the
ABR screening. A specially designated hearing laboratory adjacent to the ICN was
constructed in order to decrease the effects of noise on the ABR testing. Abnormality
was defined as an absence in response to >35 dB nHL in both ears. At 4 months
corrected age, failures were referred to further investigation such as otorhinolaryngolgic

evaluation, tympanometry, and behavioral observation audiometry in a quiet room by a

16

trained audiologist. Definitive hearing impairment was defined as >25 dB nHL in serial
retesting which occurred in 2 to 3 month intervals. Twenty-seven (77%) of those failing
the screen were available for further investigation. After subsequent follow—up 4 (1.5%)
children were confirmed to have permanent hearing loss. Although blinding of
volunteers and examiner was not reported, this study contains one of the best described
methods in measuring hearing loss and one of the larger samples recruited.

Cox et al (1992) found that 13 (17.1%) of 76 infants failed the ABR screen. At
age 8, 2 (3.6%) children continued to have sensorineural hearing loss and 6 (10.7%) to
have conductive hearing loss. The neonates assessed were part of a larger cohort (N =
202). Time constraints limited the number of children screened, and no description of the
selection of those to be assessed were reported. At 4 months of age, ABR and acoustic
immitance were performed on the infants at the outpatient clinic for high risk infants.
Failures were defined as no response at 30 dB nHL in at least 1 ear. During the follow-up
at age 2 and 8, BOA was primarily utilized. F ollow-up fails were defined as loss at >25
dB HL at 1000 - 4000Hz. Typing of loss, blinding, or the presence of a sound proof
booth were not described in the article.

In a study of 70 infants weighing 51,500g, Wright et al (1972) found 7 (10.0%)
with hearing loss. This study did not describe the methods of measuring hearing, only
reporting that examiners performed an audiological screen, and that failures were further
evaluated by 1 of the authors.

Salamy et al (1989) used ABR as an infant screen along with follow-up pure-tone
audiometry and tympanometry assessment on suspected impaired children at age 3.

Incorporating data on past audiometric results, they found that 11 (7.5%) of 199 survivors

l7

who weighed <1,500g had sensorineural hearing loss of whom 9 (4.5%) required hearing
aids. Infants were examined in the nursery. Hearing loss was defined as no wave V
response at 30 dB nHL for a sleeping child and no wave V response at 40 dB nHL for a
child who was awake. ABR was available in 70% of the cohort at about 4 months, 70%
at 1 year, and 50% at around 3 years. The Hear Kit behavioral audiometry was performed
when ABR failure was found. If hearing loss was suspected, air-conduction was
performed at 250 - 8000Hz to obtain speech thresholds. Follow-up fail was defined as
>40 dB SL Hear Kit tests, typing of loss, blinding. and testing environment were not
well described by the authors.

Swigonski et al (1987) also used multiple inclusion criteria on infants weighing
<1 ,250g or <1 ,500g having 1 or more risk factors (ventilation, low apgar sore, sepsis, or
seizures). Of 137 neonates tested with the ABR, 34 (24.8%) had no response in either ear
with 80 dB HL. Of 83 survivors who received final hearing assessments of behavioral
observation audiometry along with typanometry, 4 (3.3%) possessed moderate to severe
sensorineural hearing loss. ABR assessment was performed in the bassinets at the
intensive care units. Infants who failed or had no response at 60 dB HL, were given a
behavioral follow-up hearing and tynpanometry tests. Hearing loss was defined as >25
dB HL at 6 months and >15 dB HL at 9 months. Abnormal findings were referred to an
ENT clinic for final testing. Blinding of examiners, typing of hearing loss, and testing
environment were not recorded. The generalizability of the results is diminished due to
the additional risk factors in the inclusion criteria.

Visual reinforcement audiometry (VRA) is one of the most popularly used hearing

tests performed on young children. A sound stimulus is presented through speakers or

18

 

TABLE 1.2: Hearing Loss Studies in Low Birth Weight Infants Using Conditional

 

 

 

 

 

 

 

 

 

 

 

Multiple Sources
Study Age N Selection Prevalence Test dB
Criteria per 1000
Survivors
Belfast Hospital new- 405 special care H] 210 ABR >30
McClelland born unit infants (fails) >65
et al, 1992 3 years 382 SNHL 18 BOA,ABR, nHL
TYMP
Watkin et al, new- 918 special care BHI 51 ABR
1991 born unit infants (fails) >40
7 mos 322 with risks H1 37 BOA, nHL
596 w/out risks HI 0 TYMP
Roberts et al, new- intensive ABR >40
1982 born care unit (fails) nHL
6 mos 128 SNHL 23 ABR
Michael Reese new- 100 intensive BHI 200 ABR >40
Hospital born care unit (fails) nHL
Stein et al, 1983 7 mos 99 SNHL 20 ABR, BOA
University of new— 278 tertiary care BHI 126 ABR >35
Maryland born 270 center BHI 37 (fails) >25
Duara et al, 1986 16 mos PH] 15 ABR, BOA nHL
Rainbow Babies new- 76 very low BHI 171 ABR >30
Cox et al, 1992 born birth weight nHL
8 years 56 SNHL 36 (fails) >25
CHL 107 BOA HL
Chicago 10 70 51500g H1 100 screen
Wright et al,1972 years (fails)exam
UCSF new- 199 <1500g SNHL 75 ABR,BOA >30
Salamy et al, born (suspected) nHL
1989 4 SNHL 55 EPA, >40
years TYMP SL
Whitcomb Riley new- 137 <1250g, H1 248 ABR >40
Swigonski et al, born <1500g with (fails) HL
1987 6-9 122 risks SNHL 33 BOA &
mos TYMP
Royal N Shore 8-10 102 <28weeks BSNHL 68.6 VRA & >70
Leslie et al, 1995 mos or <1000g BSNHL 118 ABR >40
nHL

 

 

 

 

 

 

 

B=bilateral or better ear, SNHL=sensorineural hearing loss, CHL=conductive hearing

 

loss, HI=hearing impairment or loss, TYMP=tympanometry (an impedance exam),
P=permanent, BOA= behavioral observation audiometry, ABR=auditory brainstem
response, EPA=conventional earphone audiometry, VRA=visual response audiometry,
italics =sound-proof chamber.

 

 

earphones. The result of the exam is based on a conditioned response such as head
turning towards a toy or light. VRA can therefore measure hearing threshold, unlike the
BOA which again only measures MRL. To complete this test children require a higher
mental development than is needed for the BOA. But ability to comply is still lower than
that needed of the EPA. This type of hearing screen is excellent for children who are
capable of being conditioned, following basic instructions (F lexer, 1994). The next study
uses this type of screen.

Leslie et al (1995), in a sample of 102 survivors weighing <1,000g or <28weeks
with formal audiological evaluations using visual reinforcement audiometry and auditory
brainstem response audiometry, found that 12 (11.8%) had consistent bilateral
sensorineural hearing loss >40 dB nHL requiring hearing aids. All infants at the age of 8
to 10 months were brought to recognized hearing centers. Those with abnormal results
were reexamined over a period of time. Blinding was not reported, but internal validity

seems strong due to formal audiological assessments.

UNCONDITIONAL MULTIPLE SOURCES

Bjeere (1975) found that 2 (1.4%) of 139 infants born in Malmo during 1966, of
whom 90% weighed <2,500g, had hearing impairment. At age 5, non-equivalent exams
were used for testing. Fifty percent of the survivors were tested using play audiometry in
the children’s welfare center, while the other half was tested only using the whisper

method at 4m during the neurological exam. The loss was not typed as sensorineural or

20

conductive. Screening level used in the audiometry assessment and blinding of
examiners were not reported (Table 1.3).

Abramovich et al (1979) used a sample with an average age of 6.5 years and who
weighed <1,500g. Of 111 children assessed, 10 (9%) had sensorineural hearing loss, of
which 8 (7.2%) were bilateral. Pure-tone audiometry was utilized to obtain threshold,
and abnormal hearing was defined as >20 dB HL (250 - 8000Hz) for most of the children
between the ages of 4 and 12. Children unable to cooperate were examined using sound
field audiometry, the posterior auricular muscle response method (PAM), and tympanic
electrocochleography (ECOG). The approach used to type loss as conductive or
sensorineural hearing loss was not clear, nor was blinding or control for sound level of
the testing area reported. Again, internal validity is difficult to assess. This study also
used more than 1 source to prevent losing data on uncooperative children.

Bergman et al (1985) utilized multiple measures such as the auditory brainstem
response along with behavioral observation, conditioned play, and conventional earphone
audiometry, to assess hearing between the ages of 6 months and 5.5 years. Twenty-three
(9.7%) children of 237 survivors weighing 51,500g at birth with comprehensive follow-
up assessments had bilateral sensorineural hearing loss of >55 dB HL at 500 - 4000Hz.
Typing of hearing loss, blinding, and test site environment was not mentioned.

Bradford et a1 (1985) found that 10 (8.5%) of 117 newborns with gestational age
of <33 weeks failed the ABR at >100 dB SPL (or 80 dB HL). At 1 year of age 9 (7.7%)
of 108 survivors were established to have sensorineural hearing loss according to the
ABR, PAM, or ECOG with >76 dB SPL (or 60 dB HL). Abnormal ABR was defined as

absence of wave 1, III, V, and N. F ollow-up testing was repeated after 3 months. At 6

2|

 

TABLE 1.3: Hearing Loss Studies in Low Birth Weight Infants Using
Unconditional Multiple Sources

 

 

 

 

 

 

 

 

 

 

Study Age N Selection Prevalence Test dB
Criteria per 1000
Survivors
Malmo General 5 139 52,500g H1 14 CPA or
Hospital years whisper
Bjeere, 1975
University 4-12 111 <1500g SNHL 90 EPA >20
College years BSNHL 72 PAM, HL
Abramovich ECOG, or
et al, 1979 sound field
McGee 6mos 422 <1500g BSNHL 97 BOA, CPA, >55
Bergman et al, -5.5 152 seizures BSNHL 167 or EPA HL
1985 years 42 <1500g & BSNHL 286
seizures

University new- 117 <33 weeks HI 85 ABR >80
College born (all) HL
Hospital 1 year 108 SNHL 77 ABR or >60
Bradford et al, PAM or HL
1985 ECOG
GottingenGerm neo- preterm SNHL 124 ABR &
any nate Audiometry
Schulte &
Stennert, 1978

 

 

 

 

B=bilateral or better ear, SNHL=sensorineural hearing loss, CHL=conductive hearing
loss, HI=hearing impairment or loss, CPA=conditioned play audiometry,
EPA=conventional earphone audiometry, BOA= behavioral observation audiometry,
ABR=auditory brainstem response, PAM = posterior auricular muscle response method,
ECOG= tympanic electrocochleography

 

22

 

months, 107 (91%) infants of the original cohort were examined with the PAM with
retesting 3 months later being done on those who failed. Griffith’s baby test was used on
108 (92%) of the surviving infants. Any failure of the above tests, or suspicion of hearing
loss during developmental observations, received a referral to a neurologist (rather than
an audiologist or otolaryngologist) for otological investigation, acoustic impedance
testing, behavioral distraction tests, and electrocochleography. Blinding was only
reported during the Griffith’s baby test, typing, and testing sites were not described.
lntemal validity seems strong, and results may be generalized to the young gestational age
children.

An annotation by Schulte & Stennert (1978) briefly described the results of their
research on preterm infants. The infants were tested in their incubators using ABR.
During follow-up, audiometric exams were performed. Although the total sample was
not reported, 12.4% were described to have sensorineural hearing loss. Blinding of

examiners and approach to defining type of loss were not reported.

OTHER STUDIES

McDonald (1964) found 19 (1.8%) of 1081 children born <1,800g traced at ages
6 to 8 had moderate to severe perceptive deafness. Deafness was reported by health
visitors and, when possible, copies of audiograms were obtained. The specific definition
of deafness and blinding of health visitors were not reported. This study contains one of
the largest cohort of LBW infants with hearing information.

Drillien et al (1980) found that 2 (.8%) of 266 children weighing $1,500g were

deaf. Children were assessed at around 7 years of age. Assessment of deafness was not

23

described and the definition of deafness was not reported. The results were part of a large
scale psychological battery of tests. Though the psychologist was not blind to birth
weight, blinding to factors causing impairment did occur. Testing was normally done at
the primary schools.

Saigal et al (1982) found that only 1 (.6%) of 179 survivors weighing between 501
- 5,000g at birth was deaf. Follow-up visits were made 3 times in the first year, twice in
the second year, and once at ages 3 and 5. If medical personnel suspected hearing loss, or
delayed acquisition of speech was detected, the children were referred for audiometric
testing. In a later paper (Saigal et al, 1990), 73 children had their hearing assessed by
audiometry. Nine (12.3%) were reported to have untyped hearing loss. The specific
assessment and screening level used were not described, nor was blinding of the
examiners or clinical staff performing the assessment reported.

The rest of the results were obtained from a secondary source such as a literature
review article. Efforts were made to obtain the original articles, but due to constraints of
time and resources, these efforts were not successful. Three articles (Clarke et al, 1986;
Stein et al, 1983; Veen et al, 1993) made reference to 16 other studies pertaining to low
birth weight or intensive care unit survivors which assessed some kind of hearing
impairment. Robinson & Robinson (Clarke et al, 1986) were reported to have found .7%
of their low birth weight cohort to be deaf (Table 1.4).

Jacobson et a1 (Stein et al, 1983), Galambos et al (Stein et al, 1983), Cox et al
(Stein et al, 1983), and the Siegel Institute (Stein et al, 1983) were described as having
assessed severe to profound bilateral hearing loss, and to have obtained the following

results: 3.6%, 1.8 to 4.0%, 2.0 to 3.0%, and 2.0% respectively in neonatal intensive care

24

 

TABLE 1.4: Hearing Loss in Low Birth Weight Infants Using Studies Which
Could Not Be Included in The Above Categories Or Were Obtained from

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Secondary Sources
Authors Case Selection Age Prevalenc Secondary
Criteria e Per 1000 Source
McDonald, 1964 deaf <1,800g 6-8 years 18 N/A
41
& 532 weeks 6
>32 weeks
Drillien, 1980 deaf _<_1,500 6.5-7 years 8 N/A
Saigal et al, 1982 deaf 501-1,500g 1 to 5 years 6 N/A
Robinson & deaf LBW Child 7 Clarke
Robinson, 1965
Jacobson et al, SEV NICU Child 36 Stein
1981 BHI
Galambos et al, SEV NICU Child 18 to 40 Stein
1980 BHI
Cox et al, 1981 SEV NICU Child 20 to 30 Stein
BHI
Siegel Institute, SEV NICU Child 20 Stein
unpublished BHI
Harper & Wiener , SNHL LBW Child 36 Clarke
1965 FBW 10
Lubchenco et al, SNHL _<_1,500g 20-25 years 105 Clarke
1963
Lloyd, 1984 SNHL <1,501g 2-7 years 60 Veen
Johnson et al, 1987 SNHL 500-1,499g 1.5 - 3 years 0 Veen
Dann et al, 1964 SNHL 51,000g Infants 20 Clarke
Davey, 1962 SNHL 51,360g Child 22 Clarke
CHL 269
Steiner et al, 1980 SNHL 501-1,500g 6-16 years 15 Veen
CHL 8
Comer, 1960 HI <1,500g Infants 46 Clarke
Drillien, 1972 HI <1.360g Child 120 Clarke
Mercer et al, 1978 HI VLBW 4 jears ll Clarke

 

 

 

 

 

 

 

B=bilateral or better ear, SNHL=sensorineural hearing loss, CHL=conductive hearing
loss, HI=hearing impairment or loss, CPA=conditioned play audiometry,
EPA=conventional earphone audiometry, BOA= behavioral observation audiometry,
ABR=auditory brainstem response, PAM = posterior auricular muscle response method,
ECOG= tympanic electrocochleography, SEV=severe

 

25

 

units. Harper & Wiener (Clarke et al, 1986) found 3.6% of the low birth weight
survivors, while only 1.0% of the normal birth weight children, had sensorineural hearing
loss. Lubchenco et a1 (Clarke et al, 1986) assessed survivors weighing <1,500g at birth
between 20 to 25 years of age and found 10.5% to have sensorineural hearing loss. Lloyd
et a1 (Veen et a1, 1993) found that 6.0% of survivors weighing < 1,501 g at ages between 2
and 7 were sensorineurally impaired. In contrast, Johnson et a1 (Veen et al, 1993) found
that none of 143 children between 500 to 1,499g at birth had sensorineural hearing loss
when they were evaluated at 1.5 to 3 years of age. Dann et al (Clarke et al, 1986)
reported that 2.0% of 100 infants weighing 51,000g had sensorineural deafness.
Similarly, Davey et al (Clarke et al, 1986) found that 2.2% of survivors weighing 52,360g
had sensorineural deafness.

In addition, Davey (Clarke et al, 1986) also found that 26.9% of children weighing
_<_1,360g had conductive hearing loss. Steiner et al (Veen et al, 1993), who evaluated '
survivors weighing 501 to 1,500g between the ages of 6 and 16, found only .8% to have
conductive loss, while 1.5% had sensorineural hearing loss. Corner (Clarke et al, 1986)
found 4.6% of infants weighing <1 ,500g to have hearing loss. Mercer et al (Clarke et al,
1986), assessing very low birth weight children at 4 years of age, found that 1.1% had

hearing loss.

COMPARING PRE VA LENC E RA TES
By categorizing the hearing impairment studies by common methodological
practices the variability in the wide range of prevalence rates often reported begins to be

explained. When researchers used screening methods without follow-up assessments, or

26

when multiple non-related sources were used to define sensorineural hearing loss the
prevalence rates were high for the low birth weight population. In studies based on a
single detailed assessment, prevalence tended to lie between 5.9 and 19.0% with the
exception of Veen et a1 (1993) and Johnson & Ashurst (1990), who reported rates of
1.5% and .6%, respectively. Similarly, in the multiple unconditional source group, the
range was between 7.7 and 12.4%. In contrast, the studies defining sensorineural hearing
loss by retesting screened fails, or that used formal audiology assessments, obtained lower

prevalence rates. The range for those assessing confirmed sensorineural hearing loss was

1.8 to 11.8% (Figure 1.3).

FIGURE 1.3: Prevalence Rates by Source Groups

  
 
  
  

.. w e-l

180 1 ,, ,
g 1.0 ‘ 1 IMultiple
g 1.0 _ 1 ISingle
8 120 3 1 DConditional
= 100 - .,‘
a .. , , l
a .. «1
I- 40 i
3" 20 Multiple

0 Single
Conditional

Studies 2 1

The high prevalence of hearing impairment in low birth weight infants can be

demonstrated in 2 ways: by comparisons to national estimates or by comparisons to the

27

prevalence in control groups. The best approach is found in studies which obtained
hearing impairment in a control population using the same methodological practices, and
matching for measures such as socioeconomic status, sex, race, etc. (studies in Table 1.1
and Table 1.4 with control groups). On the other hand, difficulties are found when
hearing impairment in low birth weight populations is compared to the findings of
national estimates. Definitions of hearing loss differ, including screening levels used,
typing of sensorineural loss or deafness, and whether findings are for the better or either
ear. Additionally, although low birth weight populations are normally studied
prospectively using hearing evaluations, no study that present national rates of hearing
loss in the overall population here used prospective audiometric assessments. National
estimates of hearing loss are usually cross-sectional studies based on surveys. Surveys at
best only capture disabling hearing loss, or deafness, which require hearing aids (Durkin
et al, 1995). The few cross-sectional studies in which audiometric assessments were
completed on a representative or regional sample will be used to compare hearing loss in
low birth weight children.

The following are national/regional estimates from Europe, Australia, and United
States. As can be seen, low birth weight children have much higher prevalence rates than
the overall population. Martin et al (1991) found in 9 European countries that the
prevalence of deafness (not specifically defined in literature) was .8 to .9 per 1000 (.08 to
.09%) in 8 year old children. McDonald (1964) reported an estimated prevalence of
deafness of .7 to l per 1000 (.O7%) in school age children from the Ministry of
Education, and a rate of .7 per 1000 (.07%) from a survey in Northumberland and

Durham. In Australia, the general population was found to have sensorineural hearing

28

loss of 1 per 1000 (.10 %) (Doyle et al, 1992). This is much lower than any of the
prevalence rates reported in low birth weight children from European countries.

The following are national/regional estimates from the United States. Two
studies, The National Census of the Deaf Population/Birth Defects and The National
Foundation, found a prevalence of deafness of 2 per 1000 (.20%) based on self reports.
The Bureau of Education for the Handicapped of the US office of Education reported a
slightly lower rate of deafness of .75 per 1000 (.08%) school age children to be deaf.
Yeargin-Allsopp et al (1992) found 1 per 1000 (.10%) 10 year old children in
Metropolitan Atlanta to have a hearing impairment of 240 dB HL at 500-2000Hz in the
better ear, based on data obtained from multiple agencies. All of the national surveys find
much lower prevalence rates for deafness than is reported in low birth weight children in

the United States.

29

Chapter 2

NEONATAL BRAIN HEMORRHAGE STUDY: VALID HEARING MEASURES

The Central New Jersey Neonatal Brain Hemorrhage (NBH) Cohort, now
prospectively followed into preadolescence, originated as a cohort of 1,105 infants
weighing 501 to 2,000g at birth who were delivered in, or transferred to 10f 3 intensive
care units (Jersey Shore, Monmouth, and St. Peters) in central NJ between September 1,
1984 and June 30, 1987. In the counties of Ocean, Monmouth, and Middlesex, the above
ICUs delivered and/or cared for 84.8% (384/453) of all neonates weighing <2,000g and
91.5% (214/234) of infants weighing <1,500g born in the counties in the first year of the
study (Pinto-Martin et al, 1992; Paneth et al, 1994; Pinto et al, 1988). Of all tri-county
neonates <2,000g during the entire study period, 83.2% (1097/1318) were enrolled
(Holzman et a1, 1995; Pinto-Martin et al, 1992; Hegyi et al, 1994; Hegyi et al, 1996).

Hearing problems in this study were recorded and tested in many ways. Exact
sources of information are listed in Figure 2.1. Of 901 survivors at age 2, follow-up
information was available in 777 (86%), including physical examinations in 725 (80%),
hearing forms in 686 (76%), maternal interviews about the child’s health in 718 (80%),
and abstraction of medical records in 729 (81%). At age 6, follow-up assessment was
available in 685 (76.0%). At age 9, maternal interviews were available in 657 (72.9%),
physical assessment was available in 542 (60.1%), the morbidity questionnaire was

available in 537 (59.6%), and the teacher report (still being gathered at this time) was

30

FIGURE 2.1: Exact Sources of Information about Hearing Problems

 

 

 

Hearing Form at Age 2
TYPE OF PROBLEM TEST CRITERIA
Fail with Visual Reinforcement Fail at 1KHz@40 dB SL, Fail at lKHz@20 dB SL
Audiometry (VRA) Fail at 4KHz@40 dB SL, Fail at 4KHz@20 dB SL
Any Fail with VRA Failure on any of the above screening levels

 

Bilateral F ail w/ Acoustic Reflex

Failure at both ears in acoustic reflex exam

 

Bilateral Fail w/ Tympanometry

Failure at both ears tympanometry exam

 

Medical Record Abstraction at Age 2

 

 

 

TYPE OF PROBLEM WORDING OF QUESTION

Hearing Concern on Chart Concern noted in chart about hearing?
Fail on ABR, Audiometry, Normal versus Abnormal Exam?
Tympanometry

 

Hearing Loss Impression
Judged by Medical Abstractor

Categorize this child as to whether he or she has or
does not have hearing loss : a) definitely has, b)
probably has, c) cannot tell, d) probably does not
have, e) definitelydoes not have

 

 

 

Hearing Aid Prescribed Hearing Aid Prescribed?
Maternal Interview at Age 2
TYPE OF PROBLEM WORDING OF QUESTION

 

Hearing Problem Suspected by
Mom

Do you think your child has a problem with
hearing?

 

Medical Suggestion of Hearing

Has a doctor or a nurse ever told you your child

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Problem Reported by Mom has a problem with his hearing?
Hearing Aid Prescribed Has your child ever had a hearing aid prescribed?
Maternal Report at Age 6
TYPE OF PROBLEM WORDING OF QUESTION
Hearing Problem Reported by Aside from ear infections, any trouble with ears or
Mom hearing problems? What kind?
Maternal Report at Age 9
TYPE OF PROBLEM WORDING OF QUESTION
Hearing Problems Does your child presently have hearing problems?
Hearing Aid Does your child need or use a hearing aid?
Morbidity at Age 9
TYPE OF PROBLEM WORDING OF QUESTION
Ability to See, Hear & Speak Ability to see, hear, and speak normally for age
Needs Equipment Requires equipment to see or hear or speak
Limited Sight, Hearing, and Sees, hears, or speaks with limitations even with
Speaking equ_ipment
Blind, Deaf, or Mute Blind, deaf, or mute
Teacher Report at Age 9
TYPE OF PROBLEM WORDING OF QUESTION
Deaf How is the student classified: Deaf?

 

 

Hearing Impaired

How is the student classified: Hearing Impaired?

 

3|

 

 

 

 

available in 110 (12.2%). The age 9 data are incomplete and have not yet been analyzed
carefully for completeness and accuracy as of this writing. The main forms used in this
thesis to measure hearing problems were Hearing, Medical Abstraction, Review by
Audiologist, and Maternal Interview about the child at ages 2, 6, and 9.

The NBH study has strength not only in its extensive collection of
exposure/outcome variables but also in its prospective design, large sample size, and
consistent tracking methods. One possible weakness which would distort the results of
any cohort study is disproportional loss to follow-up among participants. According to
Kelsey et al (1986), effects of non-participation can effect validity, both internal
(association of measures) and external (generalizability). In order to assess equal
representation throughout time, a descriptive analysis was performed comparing all
survivors at age 2 to the initial cohort, to survivors after 1 month, and to various subsets
of the cohort with hearing information. For continuous variables, means and standard
deviations were calculated, and for categorical variables, frequencies and percents were
calculated. When differences occurred in the continuous variables, F-test was used to
compare variances and t-test to compare means. Odds ratios and Chi-square were used
for categorical variables.

Parental host factors originated from the initial maternal interview. Infant host
factors originated from 1) the initial physical or 2) initial resuscitation forms. Table 2.1
lists descriptive variables from the above 3 forms that were recorded shortly after birth
(Reuss et al, 1995). The survivors at age 2 were compared to non-survivors As expected,

the survivors resemble the overall group since they represent 85% of the initial population

32

 

TABLE 2.1: Descriptive Analysis of Overall Population, Deaths, and Survivors

 

 

 

 

 

 

 

 

 

 

 

 

 

Variables Overall Deaths@2 Years Survivors@2 Years
N=1,105 N=204 N=90l
Gestational Age
Weeks mean (sd) 30.9 (3.6) 28.1 (3.7) 31.6 (3.2)
Weight mean (sd) 1396.4 (406.2) 1013.3 (390.3) 1479.8 (359.1)
Maternal Age
mean (sd) 26.7 (5.8) 26.2 (6.1) 26.8 (5.8)
Paternal Age
mean (sd) 30.0 (6.9) 29.9 (7.4) 30.1 (6.8)
Race n (%)
Caucasian 736 (66.6%) 134 (65.7%) 602 (66.8%)
African American 287 (26.0%) 54 (26.5%) 233 (25.9%)
Hispanic 38 (3.4%) 7 (3.4%) 31 (3.4%)
Asian 15 (1.4%) 3 (1.5%) 12 (1.3%)
Gender 11 (%)
male 571 (51.7%) 120 (58.8%) 451 (50.1%)
Welfare n (%) 160 (14.5%) 23 (11.3%) 137 (15.2%)
Unmarried n (%) 231 (20.9%) 38 (18.6%) 193 (21.4%)
College Father
n (%) 223 (20.0%) 18 (8.8%) 203 (22.5%)
Father Unemployed
n (%) 60 (5.4%) 11 (5.4%) 49 (5.4%)
College Mother
n (%) 313 (28.3%) 32 (15.7%) 281 (31.2%)
Mother
Unemployed n (%) 149 (13.5%) 19 (9.3%) 130 (14.4%)

 

 

Bolded = significant (p<.05)

 

 

SOURCE: NBH Cohort (1,105 Newbornsg2,000g; 901 Survivors@Age2) 1984-1986
& 1986-1988 Columbia University, NY

 

33

 

but differ from the non-survivors. Those who died differed in host factors compared to
the overall sample. The children who died by age 2 tended to be lower in birth weight,
younger in gestational age, more likely to be male, and had less educated parents
compared to the overall cohort.

A more relevant assessment of disproportionate loss to follow-up than a
comparison of deaths and survivors is a comparison of infants who survived to 1 month
with infants who survived to age 2. Of the 204 children who died by age 2, 159 (78%)
died during the first month. As can be seen in Table 2.2, excluding the infants who died
during the first month eliminates the differences between survivors. The 2 groups of
survivors do not differ in any of the assessed variables. Thus, the population of survivors
from whom information have been obtained in the first month is quite comparable to the
population of survivors who were possible participants in follow-up at age 2. To show
that our loss to follow-up was proportional at all ages and between exams, we now
compare all survivors at age 2 to those who actually participated in the study at ages 2, 6,
and 9.

The hearing screening was performed by a nurse or a nurse practitioner
(examiners), blinded to the children’s newborn clinical status, but not blinded to mother’s
concerns about hearing. Both were trained by a certified and licensed audiologist in the
office. Reliability testing using the audiologist as a gold standard occurred only once in
the field. The examiners had practice sessions with adults in order to feel comfortable
with the operation of both pieces of equipment. In addition, principles of visual
reinforcement audiometry were reviewed and the examiners were instructed with regard

to the protocol which involved the use of visual reinforcement delivery contingent upon a

34

 

TABLE 2.2: Descriptive Analysis of Survivors

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variables Survivors@l Month Survivors@2 Years
N=940 N=901
Gestational Age Weeks
mean (sd) 31.5(3.3) 31.6 (3.2)
Weight mean (sd) 1462.6 (368.4) 1479.8 (359.1)
Maternal Age mean (sd) 27.7 (5.8) 26.8 (5.8)
Paternal Age mean (sd) 30.1 (6.7) 30.1 (6.8)
Race n (%)
Caucasian 627 (66.7%) 602 (66.8%)
African American 247 (26.3%) 233 (25.9%)
Hispanic 31 (3.3%) 31 (3.4%)
Asian 12 (1.3%) 12 (1.3%)
Gender n (%)
male 480 (51.1%) 451 (50.1%)
Welfare n (%) 145 (15.4%) 137 (15.2%L
Unmarried n (%) 205 (21.8%) 193 (21 .4%)
College Father n (%) 209 (22.2%) 203 (22.5%)
Father Unemployed n (%) 53 (5.6%) 49 (5.4%)
College Mother n (%) 292 (31.1%) 281 (31.2%)
Mother Unemployed n (%) 138 (14.7%) 130 (14.4%)

 

 

SOURCE: NBH Cohort (1,105 Newbomsg2,000g; 901 Survivors@Age2) 1984-1986

& 1986-1988 Columbia University, NY

 

35

 

correct response (head turn towards the reinforcer during or immediately after stimulus
presentation). Children were trained on the procedure at a supra-threshold level until
contingent responding was demonstrated. The presentation level was lowered to the
screening level and the presence or absence of the head turn response was noted (personal
communication with Pinto-Martin, 1997).

Hearing was assessed by visual reinforcement audiometry and middle ear function
by tympanometry. The instruments used were a Beltone portable audiometer, a
loudspeaker, a visual reinforcer (an animated toy housed in dark smoked plexiglas that
was illuminated and activated during reinforcement periods), and a Grason-Stadler
screening tympanometer. The sound field test arrangement was a custom built
arrangement, supplied by Guinta Associates. The child was seated on his/her parent's lap
facing the examiner. The visual reinforcer was located approximately 90-degrees to the
child's left side. The examiner served to distract the child, and through a hand-held
switch turned on the reinforcer when appropriate. The pass-refer criterion was a reliable
response to a 20 dB sensation level signal at 1000 and 4000 Hz. While 40 dB sensation
level was included on the response sheet, responses only at that level would result in
referral for formal audiology evaluation.

The screening stimulus was set at 20 dB above the examiner’s own threshold
(each had normal hearing sensitivity). Because of varying background noise levels and
some variability introduced by changes in the loud speaker volume control, the examiner
performed a biological check of her own thresholds at 1000 and 4000 Hz at the beginning
of each test day and in each new test environment. This was the only practical way to

screen in a sound field (as opposed to under earphones which would have been very

36

difficult in 2 year old children) especially when the screening sites were in the home. The
tympanometry pass was a type A tympanogram with peak pressure between +100 and
-250 mmHZO. Suggested screening practices recommended by the American Speech-
Language-Hearing Association were followed (personal communication with Gravel,
1997).

Children who failed the screen were referred to 1 of 2 audiology centers for
definitive hearing assessments. The office visits were offered free of charge, and
sponsored by the grant. Survivors and children with Hearing Forms were similar in
means of gestational age, birth weight, maternal and paternal age, and distributions of
race, gender, public assistance, marital status, paternal and maternal education and
employment (Table 2.3).

At age 2, Medical Abstraction was completed by 1 nurse not involved in the
hearing screening. She was asked to note any record of concerns about hearing in
medical records, primarily from the child’s pediatrician. Prescription of hearing aids and
any results of auditory brainstem response, audiometry, or typanometry exams were
recorded. On the basis of the information in the medical records the abstractor
categorized the child as definitely has, probably has, probably does not have, definitely
does not have hearing loss or cannot tell. Number of Otitis Media (OM) episodes found
in the chart, and any language exams, were also recorded. Survivors and children with
Medical Abstraction Forms were similar in means of gestational age, birth weight,
maternal and paternal age, and distribution of race, gender, public assistance, marital

status, paternal and maternal education and employment (Table 2.4).

37

 

TABLE 2.3: Descriptive Analysis of Survivors and Children with Hearing Forms

 

 

 

 

 

 

 

 

 

 

 

 

 

 

at Age 2
Variables Survivors Hearing Forms
N=901 n = 686
Gestational Age Weeks
mean (sd) 31.6132) 31.6 (3.2)
Weight mean (sd) 1479.8 (359.1) 1478.9 (358.9)
Maternal Age mean (sd) 26.8 (5.8) 27.2 (5.8)
Paternal Age mean (sd) 30.1 (6.8) 30.2 (6.8)
Race 11 (%)
Caucasian 602 (66.8%) 474 (69.0%)
African American 233 (25.9%) 166 (24.2%)
Hispanic 31 (3.4%) 20 (2.9%)
Asian 12 (1.3%) 7 (1.0‘79
Gender n (%)
male 451 (50.1%) 337 (49.1%)
Welfare n (%) 137 (15.2%) 104 (15.1%)
Unmarried n (%) 193 (21.4%) 140 (20.3%)
College Father n (%) 203(22.5%) 170 (24.8%)
Father Unemployed n (%) 49 (5.4%) 40 (5.8%)
College Mother n (%) 281 (31.2%) 227 (33.3%)
Mother Unemployed n (%) 130 (14.4%) 103 (15.0%)

 

 

SOURCE: NBH Cohort (1,105 Newboms:2,000g; 901 Survivors@Age2; 686
VRA@Age2; 669 Visual Reinforcement Audiometry Exams@Age2) 1984-1986 &

1986-1988 Columbia University, NY

 

38

 

 

TABLE 2.4: Descriptive Analysis of Survivors and Children with Medical
Abstraction Forms

 

 

 

 

 

 

 

 

 

 

 

 

 

Variables Survivors Medical Abstraction
=901 N=729
Gestational Age Weeks
mean (sd) 31.6 (3.2) 31.6 (3.2)
Weight mean (sd) 1479.8 (359.1) 1481.0 (360.2)
Maternal Age mean (sd) 26.8 (5.8) 27.2 (5.8)
Paternal Age mean (sd) 30.1 (6.8) 30.3 (6.8)
Race n (%)
Caucasian 602 (66.8%) 507 (70.0%)
African American 233 (25.9%) 172 (23.6%)
Hispanic 31 (3.4%) 19 (2.6%)
Asian 12 (1.3%) 9 (1.2%L
Gender n (%)
male 451 (50.1%) 354 (48.6%)
Welfare n (%) 137 (15.2%) 107 (14.7%)
Unmarried n (%) 193 (21.4%L 150 (20.6%)
College Father n (%) 203 (22.5%) 179 (24.6%)
Father Unemployed n (%) 49 (5.4%) 38 (5.2%)
College Mother n (%) 281 (31.2%) 251 (34.4%)
Mother Unemployed n (%) 130 (14.4%) 113 (15.5%)

 

 

 

SOURCE: NBH Cohort (1,105 Newbomsg2,000g; 901 Survivors@Age2; 729
Medical Abstractions@Age2) 1984-1986 & 1986-1988 Columbia University, NY

 

39

 

At age 2, mothers also provided information in the maternal interview (separate
from the hearing screen) on hearing aids prescribed, their own suspicion of hearing
problems in the child, and their report of any medical personnel suggestion of hearing
problems. Also reported were number of ear infections and any tympanostomy tube
insertions. Survivors and children with maternal interviews were similar in means of
gestational age, birth weight, maternal and paternal age, and distribution of race, gender,
public assistance, marital status, paternal and maternal education and employment (Table
2.5).

At age 6, the only available hearing information was in the maternal interview.
Mothers were asked about any trouble with ears or hearing problems and asked to
describe the trouble in an open ended question. Mothers were also queried about special
school services and language tests performed. Similarly, there were no differences
between survivors and children with maternal interviews in gestational age, birth weight,
maternal and paternal age, and distribution of race, gender, public assistance, marital
status, paternal and maternal education and employment (Table 2.6).

At age 9, mothers were asked to report about hearing problems and need for
hearing aids. The morbidity questionnaire asks the mother about the ability of the child
to hear, speak, see normally or equipment required for the child to hear, speak, or see;
whether there are any limitations of hearing, speaking, or seeing; and whether the child is
mute, deaf, or blind. Teachers were asked about sensory impairments and special
classroom interventions. The teacher forms, however, were not yet all collected at the

time of this writing. As with the age 2 and age 6 maternal interviews, similar results were

40

 

TABLE 2.5: Descriptive Analysis of Survivors and Children with Age 2 Maternal

 

 

 

 

 

 

 

 

 

 

 

 

 

Interviews
Variables Survivors Age 2
N=901 N=718
Gestational Age Weeks
mean (sd) 31.6 (3.2) 31.6 (3.2)
Weight mean (sd) 1479.8 (359.1) 1478.9 (358.9)
Maternal Age mean (sd) 26.8 (5.8) 27.2 (5.7)
Paternal Age mean (sd) 30.1 (6.8) 30.4 (6.9L
Race n (%)
Caucasian 602 (66.8%) 505 (69.9%)
Afiican American 233 (25.9%) 166 (23.0%)
Hispanic 31 (3.4%) 20 (2.8%)
Asian 12 (1.3%) 9 (1 .3%)
Gender n (%)
male 451 (50.1%) 358 (49.4%)
Welfare n (%) 137 (15.2%) 103 (14.3%)
Unmarried n (%) 193 (21.4%) 142 (19.6%)
College Father n (%) 203 (22.5%) 176 (24.5%)
Father Unemployed n (%) 49 (5.4%) 40 (5.6%)
College Mother n (%) 281 (31 .2%) 242 (33.7%)
Mother Unemployed n (%) 130 (14.4%) 106 (14.6%)

 

 

 

SOURCE: NBH Cohort (1,105 Newborns:2,000g; 901 Survivors@Age2; 718
Maternal Reports@Age2L1 984-1986 & 1986-1988 Columbia University, NY

 

41

 

 

TABLE 2.6: Descriptive Analysis of Survivors and Children with Age 6 Maternal

 

 

 

 

 

 

 

 

 

 

 

 

 

Interviews
Variables Survivors Age 6
N=901 N=685
Gestational Age Weeks 31.6 (3.2) 31.6 (3.2)
mean (sd)

Weight mean (sd) 1479.8 (359.1) 1479.1 (358.3)
Maternal Age mean (sd) 26.8 (5.8) 27.5 (5.5)
Paternal Age mean (sd) 30.1 (6.8) 30.6 (6.6)

Race n (%)
Caucasian 602 (66.8%) 496 (72.4%)

African American 233 (25.9%) 151 (22.0%)

Hispanic 31 (3.4%) 16 (2.3%)

Asian 12 (1.3%) 10 Q.5%)

Gender n (%)

Male 451 (50.1%) 340 (49.6%)

Welfare n (%) 137 (15.2%) 90 (13.1%)
Unmarried n (%) 193 (21.4%) 124 (18.1%)
College Father n(%) 203 (22.5%) 174 (25.4%)

Father Unemployed n (%9 49 (5.4%) 34 (4.9%)
College Mother n (%) 281 (31.2%) 253 (36.9%)
Mother Unemployed n Q/oL 130 (14.4%) 102 (14.9%)

 

 

 

SOURCE: NBH Cohort (1 ,105 Newborns:2,000g; 901 Survivors@AgeZ, 685
Maternal Report @Age6) 1984-1986 & 1986-1988 Columbia University, NY

 

42

 

found between all survivors and children with maternal interviews in means of gestational
age, birth weight, maternal and paternal age, and distribution of race, gender, public
assistance, marital status, paternal and maternal education and employment (Table 2.7).
Disproportionate loss to follow-up was not observed across the ages or the variety
of assessments utilized in relation to hearing information. Additional analysis showed
that no differences occurred in the distributions of host factors (ie. BW, GA, race, etc.)
among infants with and without Labor and Delivery, Neonatal 1, or Neonatal 2 forms. A
higher birth weight was found in children with Prenatal Forms than in infants without
these forms (mean of 1,497.6 and 5d. of 355, Appendix A). Assessment of bias in
relation to information from the Initial Physical, Initial Resuscitation, and Ultrasound
forms have not been completed as of this writing. Further investigation of proportional
representation of forms containing independent variables is warranted before performing

causal analysis about the relationships of exposures to hearing loss.

43

 

TABLE 2.7: Descriptive Analysis of Survivors and Children with Hearing Aid
Information at Age 2 or 9 Maternal Interviews

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variables Survivors Report of Hearing Aid
=90] at Age 2 or 9
N=782
Gestational Age Weeks
mean (sd) 31.6 (3.2) 31.6 (3.2)
Weight mean (sd) 1479.8 (359.1) 1475.6 (359.7)
Maternal Age mean (sd) 26.8 (5.8) 27.1 (5.7)
Paternal Age mean (sd) 30.1 (6.8) 30.3 (6.9)
Race n (%)
Caucasian 602 (66.8%) 542 (68.7%)
Afi'ican American 233 (25.9%) 193 (24.7%)
Hispanic 31 (3.4%) 21 (2.7%)
Asian 12 (1.3%) 9(1.2%)
Gender n (%)
male 451 (50.1%) 384 (48.5%)
Welfare n (%) 137 1L5.2%L 116 (14.7%)
Unmarried n (%) 193 (21.4%) 160 (20.2%)
College Father n (%) 203 (22.5%) 187 (23.9%)
Father Unemployed n (%) 49 (5.4%) 42 (5.2%)
College Mother n (%) 281(31.2%) 264 (33.8%)
Mother Unemployed n (%) 130 (14.4%) 117 (15.0%)

 

 

SOURCE: NBH Cohort (1,105 Newborns:2,000g; 901 Survivors@Age2; 718
Maternal Reports@Age2; 657 Maternal Reports@Age9) 1984-1986 & 1986-1988

Columbia University, NY

 

44

 

Chapter 3

HEARING LOSS IN THE NBH STUDY

Much of the variability in the prevalence of hearing loss reported in literature is
due to the variety of measures used and the different definitions applied in research. This
thesis attempted to organize such differences in methodology into 3 major groups (single
source, conditional multiple sources, and unconditional multiple sources). Because the
NBH study utilized several different methods of ascertaining hearing problems, different
types of case definitions can be produced in this study as well. Prevalence rates will first
be reported according to method of ascertainment, taking account of validity and
reliability.

The least subjective hearing loss measure originated from the age 2 hearing data.
At age 6 and age 9 no audiometry hearing screen was performed during follow-up. For
this reason, special attention was paid to this measure. To ensure the quality of the data,
all age 2 follow-up charts were abstracted for hearing screen information. The hearing
forms were compared to the computer data for coding errors. Approximately 900 cases
were inspected.

Ninety-three percent of the charts were free of coding errors. Twenty new forms
were found which were never entered to computer storage. Twenty-one of the screening
forms were missing from the follow-up charts, and therefore could not be checked for
errors. The new forms were compared to the computer entries of the missing forms to

establish whether accidental switching could have occurred. Four of the forms were

45

confirmed to have been switched. Twenty-eight were discovered to have minor coding
errors. The errors were reported to the data manager for further investigation.
Verification of changes and corrections were given by the Principal Investigator and/or
data manager. This resulted in 9 additional children with hearing screen information. An
additional 58 children were later found in comment fields of the hearing form as having
been referred to the audiology centers because of difficulty in testing, delay of speech and
language, or hearing problems suspected by the mother or the examiner. The records of
these 58 children have not yet been carefully abstracted.

At age 2, maternal interviews were available in 718, hearing forms in 695
(initially, 686), and medical abstraction in 729 children (Table 3.1). Five children
requiring hearing aids were reported, 2 in the maternal interview, 2 in the medical
abstract, and 1 child in both. Twenty-seven (3.8%) children were suspected by their
mothers of having hearing problems and 35 (4.9%) had hearing problems their mothers
reported as having been suggested by medical personnel. Thirty-two (4.6%) toddlers
failed the visual reinforcement audiometry screen at either 20 or 40 dB SL for stimuli at 1
or 4KHz. Medical record review judged 26 (3.6%) children as having hearing
impairment. In 10 the judgment was based on ABR data and in 18 the judgment was
based on other audiometry exams. Tympanometric data indicated bilateral middle ear
dysfunction in 118 (19.1%) children.

In the NBH study, 293 (42%) hearing forms had the numbers 20 and 40 crossed
out and replaced with higher numbers by one of the examiners. The examiner may have
been recording the HL audiometer dial readings 20 or 40 dB above his/her threshold,

though this was not part of the protocol. The second alternative may be that the examiner

46

 

TABLE 3.1: Frequency of Hearing and Middle Ear Problems by Source at Age 2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Frequencies
Maternal Hearing Medical
Type of Problem Interviews Forms Abstraction
N=718 N=695 N=729
Hearing Problem Suspected 27 (3.8%)
by Mother
Medical Suggestion of 35 (4.9%)
Hearing Problem Reported
by Mother
Hearing Aid Prescribed in 3 (.4%)
Maternal Report
Fail to respond 1 or more 32 (4.6%)
screening levels with Visual
Reinforcement Audiometry
(VRA)
Bilateral Dysfunction in the 118 (19.1%)
Tympanometry Exam
Hearing Concern on Chart 85 (12.1%)
Past Fail on BSA 10 (1.4%)
Past Fail on Audiometry 18 (2.5%)
Past Fail on Tympanometry 35 (4.8%)
HearingLoss Impression 26 (3.6%)
Hearing Aid Prescribed 3 (.4%)

 

 

SOURCE: NBH Cohort (1,105 Newbomsg2,000g; 901 Survivors@Age2; 718
Maternal Reports@AgeZ; 695 Hearing Forrns@Age2; 669 Visual Reinforcement
Audiometry Exams@Age2) 1984-1986 & 1986-1988 Columbia University, NY

 

47

 

(for unknown reasons) decided to present the stimuli above the background noise, at
higher dB SL levels than requested. The first hypothesis is the more likely of the two, but
dealing with this situation is difficult and cannot be accurately ascertained for the purpose
of this masters thesis.

Our solution was first to see who these 293 children were and find out if they
differed from the rest of the population. They did not differ in means of gestational age,
birth weight, maternal and paternal age, and distribution of race, gender, public
assistance, marital status, paternal and maternal education and employment. If all hearing
data are valid, thirty-two (4.6%) toddlers failed the visual reinforcement audiometry
screen at either 20 or 40 dB SL for stimuli at l or 4KHz. If the data represent true dB SL
stimulus presentation level, then prevalence rates must be calculated omitting data from
the 283 children who passed screens at high levels. The prevalence was elevated to 7.9%
if all possible questionable forms are deleted (Appendix B). If the changes on the forms
reflect the dB HL values used to achieve the desired dB SL, then they provide evidence of
high background noise levels. If we omit the 28 children whose records indicate that they
might have been tested at 60 dB HL or greater, we reduce the risk of including children
who had severe to profound hearing losses among the number who passed the screen.
Doing this produced a prevalence rate of 4.8% (32/660). It is important to note that
examiners were not required to record dB HL values, thus there is no way to know
whether other examiners were screening at high dB HL.

The true prevalence rate is expected to be between 4.8% to 7.9%. This is the best
range we can obtain from the study at this point. Since it is more likely that the examiner

was reporting the dB HL dial readings, it is more likely that the prevalence rate is closer

48

to 4.8%. Because the 28 children in question is only 4% of the population, the results
without the 28 children are expected to be similar to the results had all data been valid.
Next the maternal report analysis (Chapter 4) was rerun with and without the 293 children
to see if any statistical dilution occurred. The results were essentially the same. No
dilution was produced. Even if all 293 were omitted from the study, similar results are
still expected. Further work may elucidate what these questionable hearing tests mean.
At age 6, follow-up assessment was available on 685 children, among whom
information was available on any ear/hearing problems from 561. Eighty-two percent of
mothers with maternal interviews at age 6 described the hearing problem, if present, in an
open ended question. The maternal descriptions were grouped into temporary hearing
loss versus possible permanent hearing loss (Table 3.2). Seventy-two children (12.8%)
were identified by their mothers as having trouble with ears or hearing. Of those with

information, 41 (7.3%) were reported to have hearing problem. In 27 (4.8%) hearing loss

 

 

 

 

 

 

 

 

 

 

was reported as temporary.
TABLE 3.2: Frequencies of Cases with the Following Middle
Ear Dysfunction or Hearing Problems at Age 6
Problem Chapter Frequency
Type of Question Abbreviations N = 561
Any other trouble with ears, Any Trouble 72 (12.8%)
hearing?
OPEN ENDED (TYPE OF
PROBLEM) WERE
CATEGORIZED INTO THE
FOLLOWING:
Temporary Loss or Hearing Any Hearing Loss 41 (7.3%)
Loss not noted as disappearing
Temporary Hearing Loss Temporary Loss 27 (4.8%)
SOURCE: NBH Cohort (1,105 Newboms:2,000g; 901 Survivors@Age2; 685w/
Follow -up Info@Age6; 561 w/ Ear & Hearing Problem Info@Age6) 1984-1986 &
1986-1988 Columbia University, NY; 1990-1992 NY State Psychiatric Institute, NY

 

 

49

At age 9, 4 forms were available that contained hearing information: Health
Maternal Report, Morbidity Questionnaire, Teacher Report, and Physical Assessment
(Table 3.3). Thirty-seven (5.6%) children were reported by mothers to have hearing
problems; 7 (1.1%) were reported to need hearing aids; 159 (29.6%) required equipment
to hear, see, or speak; 17 (3.2%) were limited in hearing, seeing, and speaking even with
equipment; 6 (1.1%) were deaf, blind, or mute. As of the time of this writing only 110
Teacher Reports were available. One child (1.0%) was reported to be deaf and 7 (6.4%)

to be hearing impaired.

 

TABLE 3.3: Frequency of Hearing and Middle Ear Problems by Source at Age 9

 

Frequencies

 

Type of Problem

Maternal Report
N=657

Morbidity
N=537

Teacher Report
N=l 10

 

Problems with
Hearing

37 (5.6%)

 

Hearing Aids
Needed

7(1.1%)

 

Equipment
Needed to Hear,
See, Speak

159 (29.6%)

 

Continued
Limitation in
Hearing, Seeing,
Speaking

17 (3.2%)

 

Deaf, Blind, Mute

6(1.1%)

 

Deaf

1 (1.0%)

 

Hearinngpaired

 

7 (6.4%)

 

 

SOURCE: NBH Cohort (1,105 Newboms:2,000g; 901 Survivors@Age2; 657
Maternal Reports@Age9; 537 Morbidity@Age9; 110 Teacher Report @Age9; 542
Physical Assessments @Age9) 1984-1986 & 1986-1988 Columbia University, NY

 

Mothers at age 2 follow-up reported the lowest percent of hearing problems

(3.8%), the highest proportion was reported at age 6 (7.3%) and an intermediate

proportion was reported at age 9 (5.6%). Four hundred and eighty-five children had

50

 

information on maternal report of hearing problems in all 3 time periods. Sixty-four
(13%) had a hearing problem reported in at least 1 period. Only 4 children were reported
to have problems in all 3 periods. Reports of hearing recommendations or fittings were
present for 3 of these children.

Seven children were reported to have hearing problems at ages 6 and 9, but not
at age 2. Only one of these was reported to need a hearing aid. Sixteen children were first
reported to have hearing problems at age 9. Of these, only 2 were reported to need
hearing aids. Seven children were reported to have hearing problems at age 2 and 6
which were not reported at age 9. Twelve children were reported to have hearing
problems at age 2, but not reported at the other ages. Twenty-four children were first
reported to have hearing problems at age 6, but were not reported at age 2 or 9. At age 2,
1 child was reported to have a problem which was not reported at age 6 but reported at
age 9 (Table 3.4). At age 2, there were 5 children reported to have been prescribed
hearing aids with 1 child being possibly misclassified, where 1 mother answered “yes” to
needing a hearing aid but “no” to having hearing problems. Ten children were reported to
have hearing aids at age 2 or age 9 including the possible misclassification. Of the 283
children who passed the questionable screens, 1 (.4%) child was reported as needing a
hearing aid at age 9.

At age 6, 27 (9.5%) of the 283 children were reported by the mothers as having
a hearing problem. Twenty-four of these children, however, had temporary written in the
comment fields. Only 3 (1.1%) of these children had possible permanent hearing
problems. At age 9, 12 (4.2%) children were reported by the mothers as having a hearing

problem. Whether the hearing problem was temporary or permanent was not

51

distinguished at this age. Three children (1.1%) were reported by the mothers to have

hearing problems at age 6 and age 9.

 

 

 

 

 

 

TABLE 3.4: Assessing Incidence of Hearing Problems Reported by Mothers
Age 2 Age 6 Age 9 N = 485
no No no 421 (86.8%)
yes Yes yes 4 (.8%)
no Yes yes 7 (1.4%)
no No yes 16 (3.3%)
yes Yes no 7 (1.4%)
yes No no 12 (2.5%)
no Yes no 24 (4.9%)
yes No yes 1 (.2%)
SOURCE: NBH Cohort (1,105 Newboms:2,000g; 901 Survivors@Age2; 718
Maternal Reports@Age2; 561 Maternal Report @Age6; 657 Maternal Reports@Age9)
1984-1986 & 1986-1988 Columbia University, NY

 

 

After examining all these variables and making necessary corrections the 4
following case definitions were produced. Only the hearing screen results at age 2 were
utilized for the single source definition. The results from the referred audiology centers
were used for the conditional multiple source definition. For the unconditional multiple
sources all hearing information was utilized, but in addition 2 audiologists reviewed all
available audiology evaluations. Reports of hearing aids were used for a possible
preliminary definition of sensorineural hearing loss. The prevalence rates of each
definition of hearing loss will be compared to those of the past studies after addressing

the strengths and weaknesses of each definition.

SINGLE SOURCE
Failure to detect any of the test stimuli presented as part of the VRA measure

should produce the most valid definition of hearing loss from the data available at age 2.

52

To review (please see chapter 2 for details), the children were screened with 1000 and
4000 Hz stimuli at 20 and 40 dB sensation level which was completed in the field.
Earphones were attempted with this cohort, but only 10 children complied. The results
produced are thus of the better ear. At age 2, 32 (4.6%) of the toddlers failed to respond
to at least 1 stimulus level. This prevalence of 46 per 1000 was lower than all of the other
single source studies or the conditional multiple source studies before follow-up
assessments were performed.

The actual prevalence rate of the NBH cohort is suspected to be higher. Since the
children were tested at an early age, the conventional earphone audiometry (pure-tone
audiometry with earphones and voluntary response) could not be applied. Because better
ear data were collected, unilateral losses could not be identified and type of loss could not
be determined. Sound proof rooms were not available. False negatives, children who
passed at high screening levels, but might have failed at lower screening levels, might
have occurred. This reduced sensitivity might decrease the prevalence rate and dilute
statistical results of association or cause. The impact of false negatives is poSsible in each
of the following calculations that include these data.

Another method of obtaining an estimate of sensorineural hearing loss was
through consultations with audiologists utilizing the tympanogram information. In this
process, the following was assumed. Children who experienced hearing loss with middle
ear dysfunction (abnormal tympanograms) had hearing loss most probably of a
conductive or mixed nature. Therefore the children who failed to respond to at least 1
stimulus level without bilateral tympanogram fails most probably experienced

sensorineural hearing loss. Of the 32 who failed to respond to at least 1 stimulus level, 14

53

(2.0%) did not have abnormal bilateral tympanograms. This prevalence of sensorineural
hearing loss was lower than found in most of the single source studies that typed the
hearing loss as sensorineural, with the exception of Veen et a1 (1993) who obtained a

similar prevalence rate of 15 per 1000.

CONDITIONAL MULTIPLE SOURCE

Children who failed to respond to at least 1 stimulus level presented during the
VRA at age 2 were referred to audiology centers at Robert Wood Johnson and Monmouth
Medical Center, for definitive audiology evaluations. Of the 32 children in our study who
failed to respond to at least 1 stimulus level, only 15 (46%) have formal evaluations in the
follow-up charts. A very similar type of definitive follow-up was performed by
McClelland et al (1992), Roberts et a1 (1982) and Duara et al (1986). Of those who failed
their screen, 72%, 70%, and 77% were available for further testing in these studies,
respectively. Comparing the NBH children who participated in referrals to those who did
not, showed that the 2 groups were similar in demographic information (such as maternal
and paternal age, employment, education, race, gender, welfare, marital status). Twelve
(1.7%) children were found to have hearing loss, in 5 (.8%) of whom it was sensorineural
in type. This prevalence rate is lower than any of the previous studies except Duara et a1

(1986).

UNCONDITIONAL MULTIPLE SOURCES
Along with the visual reinforcement audiometry screen, past audiometric and past

ABR results were available through medical abstraction during the age 2 follow-up.

54

Many of these children were part of low birth weight intervention and follow-up
programs or general medical care which performed routine exams independently of the
NBH study. Using any fails (excluding those who could not be assessed) on the above
exams will be the case definition in the unconditional multiple sources category. Fifty-
five children (approximately 7.9%) by age 2 had experienced a fail of 1 of the above
hearing screens. This finding is similar to the results of Bradford et al (1985).

To achieve a more accurate picture, 2 certified audiologists were asked to review
all the hearing information available to age 2 in children suspected of having hearing loss
based on the above failures. These children were reported to have hearing problems by
the mother, had abnormal ear exam at birth, failed the VRA screen at age 2, or were
judged by a nurse abstractor to have possible hearing loss.

One hundred and twenty children were suspected of hearing loss according to the
above criteria. Ninety-seven cases had some information in the follow-up charts (which
included the VRA, formal audiology evaluations, maternal report of hearing problems
and medical abstraction of any past audiology screens or exams) and were sent
to be reviewed. Any recorded risk factor for hearing loss, or outcome variable related to
hearing loss, were deleted from the forms to insure unbiased review by the audiologists.
The audiologists were asked to judge definite hearing loss and type these as sensorineural
versus conductive based on all hearing information found in the charts. Many of these
cases did not contain enough information for such a distinction. Eighty had the original
VRA screen but only 37 had some other type of audiology exam in the charts. No data

beyond the age 2 assessments were used in this exercise.

55

At the onset of this manuscript 1 audiologist had completely reviewed follow-up
information sent on all 97 children. Six hundred and forty-six children had information
on the VRA screen, were reported as having or not having hearing problems by the
mother, or were judged by the abstractor to have normal hearing or possible hearing loss.
At age 2, 15 (2.3%) were reported by the audiologist as having hearing loss. Nine (1.4%)
were categorized as sensorineural hearing loss, 3 (.5%) as conductive hearing loss, and 3
(.5%) as hearing loss which could not be typed. This prevalence rate of sensorineural
hearing loss was lower than the prevalence rate found in all of the unconditional multiple

source studies that typed the hearing loss as sensorineural.

HEARING AID

Hearing aid in this study was defined as requiring hearing aids as reported at
either age 2 or 9. At age 2, 5 children were prescribed hearing aids (with 1 possible
misclassification where this mother also answered that this child did not have a hearing
problem). By age 9, 9 (1 .2%) children needed or were prescribed hearing aids of 782
with hearing aid information at either age. The child with the inconsistent maternal
report was confirmed as not needing a hearing aid at age 9 while the other 4 were
confirmed as needing hearing aids at both time periods. This 1.2% prevalence rate of
hearing aid requirement is similar to the results (1.0% and 1.8%, respectively) found in
studies by Veen et al (1993) and McClelland et a1 (1992). Abramovich et a1 (1979),
Bradford et a1 (1985), and Doyle et a1 (1992) found higher prevalence rates (7.7% to

9.5%) of hearing aid requirement.

56

Table 3.5 summarizes the prevalence rates using the above case definitions. 1)
Of 782 children with hearing aid information at age 2 (maternal report and medical
abstract) or age 9 (maternal report), 9 (1.2%) children required hearing aids. Two
children required aids at both ages. Five were new cases at age 9 while 2 required
hearing aids at age 2 but had no information at age 9. 1a) Four of 761 (.5%) children,
with hearing aid information in maternal report or medical abstract, needed hearing aids
at age 2. 1b) Of 657 with maternal report of hearing aids at age 9, 7 (1.1%) children
required hearing aids. Obtaining the prevalence of the multiple source definition by the
audiologist was very difficult for the following reason: The denominator ( from N= 646)
can only be an estimate originating from the children who had any hearing information in
the maternal report, medical abstract, or hearing form since not all cases were reviewed
by the audiologist. 2) Nine (1.4%) were reported as having sensorineural hearing loss by
the audiologist. 3) Of 669 VRA exams, 5 children (.7%) were confirmed after
subsequent audiological evaluation to have sensorineural hearing loss. 4) Fourteen

children (2.0% of 699) had possible sensorineural hearing losses based on the VRA fail at

 

TABLE 3.5: Prevalence Rates of 4 Possible Case Definitions of Sensorineural

 

 

 

Hearing Loss
Case Definitions Prevalence Rates per 1000
1) Hearing Aids from Age 2 or Age 9 12
a) at Age 2 5
b) at Age 9 11
2) Multiple Source Defined by 1 Audiologist 14
3) Referral Center Findings 7
4) Age 2 VRA Fail at Any 1 Level, Excluding 20
Any Bilateral Tympanometry Fail

 

SOURCE: NBH Cohort (1,105 Newborns:2,000g; 718 Maternal Reports@Age2; 686
Hearing Forrns@Age2; 669 Visual Reinforcement Audiometry Exams@Age2, 657
Maternal Reports@Age9) 1984-1986 & 1986-1988 Columbia University, NY

 

 

 

57

any 1 level and bilateral tympanogram information. The true prevalence rate of
sensorineural hearing loss in this population including children who may have been

missed by the first screen is suspected to be between 1.2 and 2.0%.

USABLE CASE DEFINITIONS

Four possible definitions of sensorineural hearing loss have already been
discussed. Although a few children wearing hearing aids could have conductive hearing
loss, this is unlikely, since the overwhelming proportion of children needing hearing aids
in low birth weight studies have bilateral sensorineural hearing loss and mixed hearing
loss. The following sensitivity and specificity values were calculated by the present
author (Appendix C).

In the Veen et a1 (1993) study, all 9 children who wore hearing aids had bilateral
sensorineural hearing loss. Thirteen children had sensorineural hearing loss, 11 were
bilateral. Wearing hearing aids produced a sensitivity of 82% for bilateral sensorineural
hearing loss, a reduced sensitivity of 69% when including unilateral sensorineural hearing
loss, and a specificity of 100% for both. Similar results were found with McClelland et al
(1992). All children requiring hearing aids had sensorineural hearing losses. Seven
children had sensorineural hearing loss, 6 were bilateral. Requiring hearing aids
produced a sensitivity of 83% for bilateral sensorineural hearing loss, a reduced
sensitivity of 71% when including unilateral hearing loss, and a specificity of 100% for
both. Improved values were seen in the Abramovich et al (1979) study. All children

wearing hearing aids and needing support services or special education had sensorineural

58

hearing losses. Ten children had sensorineural hearing loss, 8 were bilateral. Sensitivity
including unilateral hearing loss was 80%, but 100% for bilateral hearing loss.
Specificity values were 100% for both. In the Bradford et a1 (1985) study, all 9 children
who required hearing aids had sensorineural hearing loss. Nine children had
sensorineural hearing loss, 8 were bilateral. Sensitivity was 100% for both unilateral and
bilateral sensorineural hearing loss. Specificity was also 100% for unilateral and 99% for
bilateral sensorineural hearing loss. Doyle et a1 (1992) had the lowest sensitivity value.
All children requiring hearing aids had sensorineural hearing losses. Four children had
sensorineural hearing loss, 1 child had bilateral loss. This produced a 100% sensitivity
for bilateral sensorineural hearing loss but only a 25% sensitivity including unilateral
hearing loss. Both specificity values were 100%.

Wearing hearing aids as an indicator of sensorineural hearing loss may not work
unless all children have had access to audiologic services, insurance coverage for hearing
aids, and good follow through from the parent. Needing or requiring hearing aids
(regardless of whether they were bought or worn), however, seems to be a good indicator
of bilateral sensorineural hearing loss and could be used as a cost effective preliminary
variable until funds are available for a valid and reliable hearing screen for sensorineural
hearing loss or formal audiological exam.

The other method to define sensorineural hearing loss was to abstract as much of
the hearing information as possible with the hopes of producing more cases during the
audiology review. The complete review of all cases sent to the second audiologist was
not yet finished at the onset of this thesis. Reliability of this measure can not yet be

accurately performed. These 2 definitions remain our only usable definition of

59

sensorineural hearing loss possible for causal analysis at this time. The hearing screen
alone did not permit valid typing of sensorineural hearing loss. Although referrals to
audiology centers were used to provide a definitive exam, the participant rate was lower
(45.5%) than other studies (72-77%) and the number of cases was too low to enable
causal research. Using the conditional method of excluding bilateral tympanogram fails

created the most cases but also the most misclassification errors.

60

Chapter 4

VALIDITY OF MATERNAL REPORT OF HEARING PROBLEMS

Maternal report of hearing problems has been utilized by clinicians and
researchers as an indicator of hearing loss in children. Proponents of the use of such
reports, such as Barry McCormick (1988) and Gell et a1 (1992), conclude that parental
suspicions are reliable and valid “indicators of the presence of hearing disorders”. But
Marie McCormick (1997) argues from personal experience that hearing difficulties
suspected by parents are influenced by a history of otitis media episodes. Although many
studies have shown that reliance on parental observation can delay detection of auditory
impairment (Harrison & Roush, 1996; Stein et al, 1983; Ellsman et al, 1987; Stein et al,
1990; Mauk et al, 1991; Thompson & Thompson, 1991) and some studies have partially
assessed either the false positive rate or the sensitivity of maternal report of hearing loss
(Parving & Christensen, 1992; Lee et al, 1996; Watkin, 1990), no study known to the
author has reported sensitivity, specificity, and predictive values (which will vary based
on the prevalence or hearing loss in the population) of maternal report of hearing loss in
relation to audiometry results.

This chapter was designed to assess the correlation among a variety of measures
of hearing problems in low birth weight infants, and especially the predictive value of
maternal report in relation to audiometry results.

61

The initial null hypothesis was that maternal report does not predict failure to
respond to any screening level of the visual reinforcement audiometry screen. A second
null hypothesis was that ear infections and/or ear surgeries do not predict maternal report
of hearing problems. The 2 types of maternal reports of hearing problems (self & medical
personnel suggested) were assessed in terms of the strength of their association with the
VRA findings, the significance of the associations, and their level of prediction. Odds
Ratios and Confidence Intervals (95%) were used to assess the strength and significance
of the associations. Sensitivity, Specificity, and Positive Predictive Values were
calculated to assess the level of prediction. The number of ear infections in children with
and without (both) maternal reports of hearing problems were compared using the Mann-
Whitney test. A predictive model was built using Logistic Regression, including ear
infections, ear surgery (tympanostomy), concerns on the medical chart, and interaction

terms to assess prediction of self maternal report of hearing problems.

MATERNAL REPORT IN RELATION TO VRA

To compare (both) maternal reports to the VRA screen, hearing loss was defined
as failure at level 1 or level 2 on any frequency. Children with any failure were more
likely to have self maternal report of hearing problems than children with no failure
{OR=5.6 (1 .8-17.9)}, but the positive predictive value was only 17% and sensitivity only
15%. Children with any failure were not significantly more likely to have MD maternal
report of hearing problems, but was approaching significance {OR=3.1 (.9-10.9)}.

62

MATERNAL REPORTS OF INFECTION AND EAR SURGERY

Initially, ear infections (none versus any) were not found to be significantly
related to self maternal report of hearing problems. Categorizing any ear infections,
however, yielded different results. Children with >9 ear infections were 6.3 (1.9-20.6)
times more likely to have self maternal report of hearing problems than were children
without any ear infections. and 8.7 times more likely (2.8-26.5) to have MD maternal
report of hearing problems (Figure 4.1). Similar results were obtained when substituting
episodes of OM documented on the medical charts for ear infections.

The distribution of ear infections in children with (both) maternal reports of
hearing problems differed from that found in children without such maternal report. The
median number of infections was 3 in children with self maternal report of hearing
problem and only 1 in children without. A similar pattern was seen for MD maternal
report. Because the distribution of ear infections was positively skewed (N=.21,
p<.0001), the Mann-Whitney nonparametric test was performed. On this test the
difference in ear infections reported was significant (p<.01) for MD maternal report and
approached significance for maternal self report.

Children who had ear surgery were 4.7 (1.3-17.2) times more likely to have self
maternal report of hearing problems than children without ear surgery. Similarly,
children who had ear surgery were 12.8 (4.7-34.8) times more likely to have MD maternal
report of hearing problems than children without ear surgery. Both maternal reports were
more predictive of history of ear infections (positive predictive value 74% to 82%) than

63

Figure 4.1

: Odds Ratios for the Association of

Maternal Report of Hearing Problems with

Number of Ear Infections

NBH Cohort (1.105 newborn 5 2,000g; 901 survivors@age2; 725 examined@age2)

100

Self Maternal Report MD Maternal Report

U1 0
E e
0
sh
E
3 10 ~ .
5
a ' °
=
c e
U 0 e
3
In
°‘ -
a .
a
a u
8 I I
0*: 1 n - l V‘ ~
g C N o O\ o N as as
m a a A .9 .9 A
g ... M .-— m
O

0.1 .

Number of Ear Infections

64

of a history of ear surgery (positive predictive value of 11% to 21%), reflecting the much

higher incidence of ear infections (Table 4.1).

 

TABLE 4.1: The Association between Maternal Report of Hearing Problems with
Maternal Report of Ear Infections and Ear Srfleries at Age 2

 

 

 

 

 

 

Maternal Odds Ratio Sensitivity Specificity Positive
Report (95 %CI) Predictive
Values
Ear Self 4.7 (1.3-17.2) 15% 96% 11%
Surgeries MD 12.8 (4.7-34.8) 35% 96% 21%
Ear Self 1.6 (.7-3.8) 4% 97% 74%
Infections MD 2.6 (1.1-6.4) 6% 98% 82%

 

 

SOURCE: NBH Cohort (1,105 Newboms:2,000g; 901 Survivors@Age2; 777 w/
F ollow-up Information@Age2; 725 examined@Age2; 718 Maternal Reports@Age2;
686 Hearing Forrns@Age2; 669 VRA@ageZ)

 

 

INTERACTION BETWEEN EAR INFECTIONS AND SURGERY AND PREDICTION
OF VRA BY MATERNAL REPORT

For ear infections and ear surgery to act as confounders of the relationship
between (both) maternal reported hearing problems and VRA screen results, these
exposures would have to be related to VRA fails. No relationship of any ear infection,
>9 ear infections, or ear surgery to any fail on the VRA screen was found. We explored
the possibility, however, that these factors moderated the association between (both)
maternal report and VRA.

Stratifying by ear surgery or >9 ear infections was not possible because no child
with surgery failed the VRA screen, and no child with >9 ear infections had failed the

VRA screen corresponding to (both) maternal report of a hearing problem. These empty

65

 

cells can partially be explained by the very high correlation between surgery and >9 ear
infections. Children with >9 ear infections were 11 (4.1-30.4) times more likely to have
had ear surgery than were children with _<_9 ear infections and 45 (5.4-3 75.9) times more
likely to have had ear surgery than children without any ear infections.

The crude odds ratio for the association of self maternal report with VRA screen

is 15.7 (2.5-100.6) in children without infection. Improvement is also seen in the positive

 

predictive value (20%) and sensitivity (33%) in this group. In the stratum with infection,
however, the corresponding odds ratio is only 3.2 (.7-15.4). Thus a strong interaction of

the self maternal report/V RA association is found with infections (Table 4.2).

 

TABLE 4.2: The Association between Maternal Report of Hearing Problems with
Any Fail in the VRA Screen at Age 2, Stratified by Ear Infections

 

 

 

 

 

 

 

 

 

Selection Maternal Odds Ratio Sensitivity Specificity Positive
Report (95 %CI) Predictive

Values
All Self 5.6418479) 15% 97 % 17 %
Subjects MD 3.1 (.9-10.9) 12% 96% 10%
with Ear Self 3.2 (745.4) 11 % 96% 12%
Infections MD 2.4 (.5-11.4) 12% 95% 9%
without Ear Self 15.7 (2.5-100.6) 22% 98% 33%
Infections MD 6.9 (.7-69.0) 1 1% 98% 20%

 

 

SOURCE: NBH Cohort (1,105 Newborns_<_2,000g; 901 Survivors@Age2; 777 w/
F ollow-up Information@Age2; 725 examined@Age2; 718 Maternal Reports@Age2;
686 Hearing Forrns@Age2; 669 Visual Reinforcement Audiometry Exams@Age2)

 

 

MATERNAL REPORT AND MEDICAL ABSTRACTION
Children for whom the nurse abstractor indicated that there was concern about

hearing on their charts (N =85) were significantly more likely to have self maternal report

66

 

of hearing problems {OR=10.2 (4.4-23.7)} and significantly more likely to have MD
maternal report {OR=12.1 (5.6-26.0)} than children whose medical records did not raise
concern (Table 4.3). Fewer children were judged by the abstractor to have hearing
impairment (N =25) than to have records indicating concern about hearing (n=85), but

children identified as having hearing loss were more likely to have self or MD maternal

report of hearing problems {OR=20.8 (8.0-53.7) and OR=14.1 (5.7-35.0), respectively}.

 

TABLE 4.3: The Association between Maternal Report of Hearing Problems with
Concerns on Medical Chart and Medical Chart Abstractor-Judged Hearing

 

 

 

 

 

 

Impairment
Medical Maternal Odds Ratio Sensitivity Specificity Positive Predictive
Chart Report (95%CI) Values
Concern Self 10.2 (4.4-23.7) 17 % 98% 54%
MD 12.1 (5.6-26.0) 22% 98% 57%
Judged Self 20.8 (8.0-53.7) 36% 97% 35%
HI MD 14.1 (5.7-35.0) 36% 96% 27%

 

 

SOURCE: NBH Cohort (1,105 Newbornsi2,000g; 901 Survivors@Age2; 777 w/
F ollow-up Information@Age2; 725 examined@Age2; 718 Maternal Reports@Age2;
686 Hearing Forms@Age2; 729 Medical Record Abstraction@Age2)

 

 

 

INTERACTION OF OTITIS/SURGERY AND ABSTRACTOR JUDGED HEARING
IMPAIRMENT

An approaching significant relationship was found between medical chart
abstraction and VRA results {OR=3.5 (1.0, 12.6)}. The relationships between (both)
maternal reports of hearing problems, abstractor-judged HI, and ear infections were
investigated. A higher proportion of children with (both) maternal reports of hearing

problems tend to have hearing impairment judged by the abstractor and also cluster in the

67

group with >9 ear infections. Both maternal reports of hearing problems interact with the

presence of frequent ear infections and abstractor judged hearing impairment.

MULTIVARIATE MODELS OF DETERMINANTS OF MATERNAL REPORT

Because self maternal report of hearing problems were not strongly predictive of
our objective measure of hearing loss, the VRA screen, we were interested in finding
exactly what they were reporting. A predictive model for self maternal report was
developed. Greater than 9 ear infections, ear surgery, concern on medical charts, and
possible interaction terms were included. No interaction term improved the model.
Greater than 9 ear infections {OR=5.3 (2.0-13.9)}, ear surgery {OR=4.7 (1.3-17.3)}, and
concern on medical charts {OR=10.3 (4.4-24.0)} taken 1 at a time significantly improved
the model for self maternal report of hearing problems {Chi—square (p-value) = 8.56
(<.05), 4.08(<.05), 26.70(<.001), respectively}. Ear surgery and ear infections, however,
dropped out of the model when concern on the medical charts was added to it. Not only
did children with surgery have more ear infections, but they were more likely to have
medical charts indicating concern about hearing. Therefore, the full model consisted of
only concern on the medical chart. In addition, concern on medical chart was more
strongly associated with ear surgery {OR=7.0 (2.6-18.7)} and >9 ear infections {OR=4.7
(2.3-9.7)} than it was with VRA fail {OR=2.9 (1.3-6.9)}.

The 3 principal findings of this chapter are 1) (both) maternal reports of hearing
problems are poor predictors of VRA screen in 2 year olds, 2) (both) maternal reports of
hearing problems are more highly concordant with concern about hearing noted in

medical charts, and 3) (both) maternal reports of hearing problems and concern about
68

hearing in medical record are strongly influenced by >9 ear infections and by ear surgery
in the past, independent of confirmed hearing loss.

Although (both) maternal reports of hearing problems are strongly associated with
evidence of concern about hearing on medical charts and abstractor-j udged hearing
impairment, none of these 4 variables proved to be strong predictors of any fail in the
VRA screen. Both types of maternal report were influenced by a history of frequent ear
infections and of ear surgery. Ear infections and ear surgery, however, were not
themselves associated with any fail. Because children with >9 ear infections are much
more likely to have ear surgery and because children with ear surgery are more likely to
have medical charts with concerns about hearing, ear surgery does not explain self
maternal report of hearing problems when either ear infections or concern on the medical
charts are included in the model. At age 2, (both) maternal reports of hearing difficulties
are thus poor indicators of any fail on the VRA screen and is a better indicator of ear
infections, as suggested by Marie McCormick (1997).

In 3 population based studies (Durkin et al, 1995; Durkin et al, 1994; Zaman et al,
1990) the following validity results for maternal report of hearing problems, unadjusted
for other disabilities, were: Very low positive predictive value (2.3% to 3.3%), low to
moderate specificity (29.7% to 54.0%), and high sensitivity (89% to 100%). The positive
predictive value for the Zaman et a1 (1990) study for maternal report of any disability
was 22%, but was lower for sensory disorders. In the Jellinek et al (1988) study the
sensitivity for maternal report for any psychosocial functioning was high (95.0%) and
specificity was moderate (68.0%). Durkin et al (1995) obtained sensitivity values for

maternal report of sensory disabilities ranging from 80% to as low as 4% for milder

69

forms, also found poor positive predictive results, which resulted in overestimation of
serious disability by more than 300%.

Studies (Parving & Christensen, 1992; Lee et al, 1996; Watkin, 1990) partially
assessing the predictive performance of maternal report for audiometry results show
similar findings to the present study. Parving & Christensen (1992), utilizing a
population referred to an audiology center, showed a high false positive rate (44%) and
moderate sensitivity (60%) for parental suspicion about hearing in relation to behavioral
play audiometry (obtained in 65% of the population). Another study, using national data
and contrasting maternal self report of hearing problems for ages 6-11 and teenager self
report of hearing problems for ages 12-17 to audiometry results, found a sensitivity of
only 25%. Watkin (1990) showed, in a hearing impaired population, low sensitivity of
parental concern about hearing to permanent hearing loss. Many studies (Harrison &
Roush, 1996; Stein et al, 1983; Ellsman et al, 1987; Stein et al, 1990; Mauk et al, 1991;
Thompson & Thompson, 1991) have shown that relying on parents’ suspicion of hearing
loss delays detection and intervention to remedy hearing loss. Yet, maternal report of
hearing problems continues to be utilized by clinicians and researchers as a primary
indicator of hearing impairment in children.

Barry McCormick (1988) argues that parental suspicions are “reliable indicators
of the presence of hearing disorders”, citing 3 studies reporting an increase in referrals
about hearing to professionals due to active surveillance of hearing problems by
questioning mothers. Gell et al (1992) expressed the view that parental suspicions are

“valid indicators of the presence of hearing disorders”.

70

At age 2, visual reinforcement audiometry is the hearing test of choice. In the
normal, low-functioning, and hearing impaired infant and toddler population, the
reliability (Rudmin, 1984; Thompson et al, 1989; Gravel & Traquina, 1992; Greenberg et
al, 1978; Liden & Harford, 1985; Ruth et al, 1983; Lansioni et al, 1989; Primus, 1988) of
the VRA is well established. One study (Moore et a1, 1992) investigated premature
infants and found VRA to be reliable as long as the child was 8 months corrected age or
6 months mental age or older. One follow-up study (Talbot, 1987) using a hearing
impaired population tested the validity of the VRA in relation to conditioned play
audiometry, which is a more sophisticated hearing test, and found no significant
difference between the 2 tests. If the NBH VRA results are a good indication of hearing
loss at age 2, then it can be concluded that maternal report of hearing problems are not. It
is important to note, however, that the NBH VRA was a screening procedure not a
diagnostic procedure and that it was conducted in the field not in a test booth. As a result,
some children who failed the screening likely had hearing within normal limits and some
children with hearing losses, particularly those with mild and moderate hearing losses,
might have passed the screening.

These findings have important implications for both clinical and research settings.
The first implication is that reliance on parental suspicion as the primary screening tool
impacts the psycho-social development of the child through delayed detection. A second
implication of our findings is that we need to improve survey techniques for ascertaining
hearing problems in children. In other surveys some questions typically asked were: 1)
Does the child appear to have difficulty hearing?, 2) Aside from ear infections is there

any trouble with ears or hearing problems?, 3) Does the child have the ability to hear

71

normally?, 4) Does the child require equipment to hear?, 5) Does your child hear with
limitation?, 6) Is the child deaf?, 7) Does the child have hearing problems?, 8) Does your
child need a hearing aid?, 9) Is the child classified as deafl, 10) Is the child hearing
impaired (N BH surveys, unpublished; POPS surveys, unpublished; McMaster surveys,
unpublished)? The present study asks 3 questions: 1) Has your child ever been prescribed
a hearing aid, 2) Do you think your child has a problem with hearing or 3) Has a doctor or
nurse ever told you that your child has a problem with his hearing? Bias might be
reduced with the addition of “which is permanent” at the end of the hearing problem
questions. Another improvement might be to separate questions about whether the child
had any ear infections from questions about hearing problems, which was not done in our
questionnaire.

In studies published by the National Center for Health Statistics (Roberts &
Ahuja, 1975; Roberts & Federico, 1975) on hearing sensitivity in children, mother's
suspicion over-reported hearing loss (compared to audiometry results) at all ages.
Hearing loss perceived by mothers may also miss mild to moderate hearing loss.
Hopefully, parental suspicion of hearing impairment may be more valid at older ages.
We are beginning to investigate this subject in our own cohort at age 6 and age 9.
Although parental suspicion should not be disregarded in clinical settings, our study
supports a strong recommendation for utilizing more accurate measures of hearing loss in

toddlers in the research setting.

72

Chapter 5

FUTURE RESEARCH IN NBH HEARING

Providing a valid and reliable outcome measure is only 1 initial step in paving the
way to causal research. This chapter addresses the subsequent stages needed to ensure
accurate analysis of the causes of sensorineural hearing loss in the NBH data. First, use of
hearing aids and audiologist’s review as case definitions for sensorineural hearing loss
will be assessed. A preliminary assessment of 1 causal factor previously associated with
sensorineural hearing loss will be made. Findings similar to past studies should be found
if prescription of hearing aids is a good case-definition for sensorineural hearing loss.
Other causal factors shown to be strongly and or consistently associated with
sensorineural hearing loss will be emphasized. Next, improvements to the NBH data are
suggested in a proposal to obtain a more specific case definition. Lastly, suggestions are
made for improving all future hearing research.

Irn‘ants of low birth weight and young gestational age possess immature cerebral
vascular beds which are at risk of intracranial hemorrhage (Volpe, 1989; Paneth & Pinto-
Martin, 1990; Holzman et al, 1995). This risk is elevated when other complications are
present such as asphyxia and hypertension. Most sensorineural damage originates in the
_ cochlea (Clarke & Conry, 1979; Kelemen, 1963; Bosher, 1972; Fisch, 1954). Some
forms of hearing loss may also depend on damage to the eighth nerve, the ventral
cochlear nucleus, dorsal cochlear nucleus (Hall, 1961; Clarke & Conry, 1979; McDonald,

1964; Spector et al, 1978), the inferior colliculi (Clarke & Conry, 1979), and the

73

geniculate bodies of the thalamus (Clarke & Conry, 1979). Destruction of auditory
organs via hemorrhaging is produced by the absence of blood, leading to lack of
necessary nutrients and other vital substances needed for cell survival, in areas near the
site of lesion (Butler, 1994). Alternatively, blood itself can act as an ototoxin (Kelemen,
1963; Buch, 1966; Spector et al, 1978). Depending on the portion of the brain destroyed,
comorbid conditions with sensorineural hearing loss can include cerebral palsy, mental
retardation, and visual impairment.

Some studies without cranial ultrasounds have used these comorbid conditions as
evidence for or against intracranial damage in sensorineural hearing loss. Bergman et a1
(1985, pg. 99) writes “the best argument against the etiologic role of hypoxic-ischemic
injury or intracranial hemorrhage in producing hearing loss is the lack of motor or
cognitive deficits in 61% of our hearing-impaired children”. But even in a population
weighing 51,500g, finding that 39% of hearing impaired children have motor or cognitive
deficits implies excess risk of brain damage as one would expect about 20% to have such
deficits. Roberts et a1 (1982) also found a relationship approaching significance between
IVH and ABR fail at >70 dB HL but not between IVH and ABR fail at >40 dB HL.

Clarke et a1 (1986) stated that Voss (1923) and Albrecht (1930) hypothesized
hemorrhage in the inner ear causes sensorineural hearing loss. Kelemen (1963)
reports that Voss (1923) actually credits Toynbee (1860) to have been the first to show
extravasates (a discharge or escape of blood from a vessel into the tissue) in the labyrinth.
Pathology studies were the first to discover the effects of hemorrhaging on the middle and

inner ear. Kelemen (1963) was one of the first to show the ototoxicity of blood to the

74

organs of hearing. This author reported 8 patients in whom hemorrhages reached the

endolymphatic space of the semicircular canals (Figure 5.1a and 5.1b).

      

’ - J :_:.‘ (Ir; a? ’1’.» :
FIGURE 5.1a: Blood upon the Tip of FIGURE 5.1b: Blood on the Tip
the Cupula of the Canals. and Sides of the Cupula.
Kelemen, 1963.

   

 

 

Buch (1966) collected and examined 135 temporal bones from 73 newborn
infants, who were primarily premature. Hemorrhaging was most commonly found in the
internal acoustic meatus (internal auditory canal meatus) in the bony labyrinth. The
membranous labyrinth, perilymphathic spaces, especially in the cochlea and to a lesser
extent the vestibular portions, were the regions most commonly destroyed by
hemorrhages. In the cochlea, the scala tympani contained the most extravasated blood
followed by the scali vestibuli. The most extensive damage occurred in the scala tympani
(Figure 5.2). In 52% of the cases, bleeding was found in both the middle and inner ears,

in 40% bleeding was only in the inner ear, and in 4% bleeding was only in the middle ear.

75

 

FIGURE 5.2: Bloodaround the Semicircular Canal.
Buch, NH, 1966.

76

A retrospective comparison was made to controls without hemorrhages. Use of forceps,
difficult delivery, twin births, and primary asphyxia were suggested causes.

Spector et a1 (1978) found findings similar to those of Buch (1966). Autopsies
and temporal bone assessments were completed on 28 infants who experienced
respiratory distress (RD) before dying and on 24 infants who died of other natural causes.
A double blind retrospective design was used to compare the inner ears of the 2 groups.
In most of the respiratory distress (RD) cases, the bleeding was located in the
periventricular matrix with possible extension to the intraventricular CSF system. The
origin of hemorrhaging was suggested to be intracerebral, extending into the
periventricular subependymal cortex, breaking through ventricles and then into the
subarachnoid space. The most common location for extravasates was in the inner ear’s
perilymphatic space. A large proportion of the cases had hemorrhaging in the internal
acoustic meatus. Spector et a1 (1978) suggested 5 portals of entry of blood into the inner
ear: 1)VIII nerve along with nerve cell bodies, 2) cochlear aqueduct, 3) cochlear vein, 4)
inner ear spontaneously, and 5) otic capsule, the first being the most common.

The following hypothesis was formed by Spector et a1 (1978) about the pathway
to auditory destruction: Respiratory distress -) Intracranial hemorrhage -) Inner ear
hemorrhage with the interaction of ventilatory support (IPPV or CPAP). Hemorrhaging
in the periventricular matrix (Figure 5.3) was found in 24/28 infants who had had
respiratory distress, but only in 2/24 infants who had not had respiratory distress, neither
of whom had inner ear bleeding. Infants with respiratory distress were 66 (110-3954)

times more likely to have hemorrhaging in the inner ear then children dying of other

77

1‘ . 1 3 w' ...n 44'3”".5‘: 1 . r . a - , ~ .
,‘i‘." )‘1- ‘ ‘% .mfgq' {t:_() ”RAINE’Z‘ ,. It 7' s
. 3‘ . \ .. _. ‘ .1. .' . I-_ 3 ‘. I, ',- .

    

Figure 5.3: Massive Hemorrhage Especially in the Perilymphatic Spaces.
SV=scala vestibuli, SM=scala media, ST=scala tympani, Hem=hemorrhage.

 

Spector et al, 1978.

 

78

causes (OR and 95% CI calculated by present author). Infants with the most severe
bleeding were the most premature and of the lowest birth weight, or were very ill and
needed ventilatory support. Infants with less bleeding died in utero or were term babies
who needed very little ventilatory assistance.

This relationship has not been greatly investigated “in the living” because few
studies possess a valid and reliable measure of both sensorineural hearing loss and
cerebral hemorrhaging on surviving infants. In the literature review, only half of the
studies contained univariate analysis, only 5 studies completed multivariate analysis, and
only 5 published studies (Doyle et al, 1992; Clark et al, 1986; Campenelli et al, 1958;
Winkel et a1, 1978; and Veen et al, 1993) implemented pure-tone, air-conduction
thresholds as the sole measurement for hearing loss. Of these, 4 reported using standard
techniques (bone conduction) to distinguish sensorineural from conductive loss. But only
2 used large low birth weight samples, 890 in the Netherlands and 205 in Canada. Most
past research on hearing in the low birth weight population has had a sample size of
approximately 100 and only 9 other studies in the literature review reported sample sizes
of over 200.

Each study assessed about 1 dozen causal or predictive variables. Of the 6
variables which were similar in both studies, the only one found to be significantly
different by the 2 research teams was cerebral damage. Using Goodman and Kruskal’s
Gamma correlation coefficients, a significant relationship (p = .013) was found by Clarke
et al (1986) between sensorineural hearing loss and cerebral dysfunction which was
measured by EEG. Using Fisher’s exact t and Student-t test, Veen et a1 (1993) showed a

very significant association (p<.01) between intracranial hemorrhage (which was

79

measured by ultrasound or computed tomography, rapid or saltatory deterioration, and/or
fall in hematocrit level) and sensorineural hearing loss. The OR and 95% CI for the latter
study was calculated with the Woolf method by the present author (Table 5.1). The
relationship between cerebral hemorrhage and sensorineural hearing loss seems to also be

strong, 4.4 (1.5-13.3).

 

TABLE 5.]: Hemorrhage in Relation to Hearing Loss

 

 

First Author Definition Variable Odds Ratio
(95% CI)
Veen sensorineural intracranial hemorrhage diagnosis 4.4 (1.5-13.3)
Roberts ABR @ 70 IVH 3.3 (.9-12.0)
Roberts ABR @40 IVH .9 (.3-3.0)

 

The NBH study has one of the most valid and reliable measures of cerebral
hemorrhaging in the newborn (Paneth et al, 1994; Pinto et al, 1988; Holzman et al, 1995).
The cohort was designed to assess the causes and consequences of brain injury, and
cranial ultrasounds were used to image brain damage in the newborn period. Three
lesions in the brain were of interest: Germinal matrix/intraventricular hemorrhage
(GM/IVH), parenchymal echodensities and echolucencies (PE) (which marked white
matter damage) and ventricular enlargement (VE). Pinto et al (1988) tested the reliability
of the ultrasound readings using a sample of the population. The findings were as
follows: moderate kappa of .53 for germinal matrix, good kappa of .67 for ventricles,
strong kappa of .73 for parenchymal abnormalities. Reliability was later tested by Pinto-
Martin et al (1992) using the whole cohort with an overall agreement of 89%, a GM/IVH
kappa of .69 and ventricular hemorrhage kappa of .77. Another strength of the NBH
ultrasound interpretation is describing the specific location of lesions. Sites closest to the

auditory organs should have a stronger relationships with sensorineural hearing loss.
80

If Spector et al (1978) was correct, a significant and strong relationship must first
be found with hemorrhage and sensorineural hearing loss. A stronger relationship should
be found with parenchymal (white matter) lesions than with germinal matrix lesions since
auditory axons are contained in white matter. Using hearing aid and the audiologist’s
review as our best measures to date, OR and 95% CI for the relationship of sensorineural
hearing loss and cerebral damage were calculated with SPSS. Similar findings to the
results found in Netherlands with the Veen et al (1993) cohort were found. NBH results
showed that children with any GM/IVH (not isolated) were 4.9 (1.3-18.4) times more
likely to need hearing aids than those without GM/IVH. A stronger relationship occurred
with PELVE. Children with white matter damage were 6.0 (1.5-24.7) times more likely
to need hearing aids than those without white matter damage. Children with GM/IVH did
not produce significant results {2.1 (.5-8.5)} in children reported as having sensorineural
hearing loss by the audiologist. Children with PELVE, however, approached significance
{4.0 (.8-19.8)}. Analysis was also completed including the 2 children with untyped
hearing loss, since they can not be ruled out at this time as not having sensorineural
hearing loss. Children with GM/IVH again produced insignificant results {2.1 (.6-7.1)}.
Children with PELVE, however, were 7.0 (2.0-24.2) times more likely to have
sensorineural or untyped hearing loss reported by the audiologist. Past studies have found
a strong and significant relationship of intracranial hemorrhage to sensorineural hearing
loss. Because causal research with this hypothesis is lacking, the present author suggests
that investigations with our cohort begin here.

To further investigate the Spector et al (1978) hypothesis, the effects of

Respiratory Distress (RD) and mechanical ventilation on sensorineural hearing loss

81

should be investigated. Veen et al (1993) found IPPV and/or CPAP to be significantly
associated with sensorineural hearing loss with a strong OR of 6.0 (1.3-27.2), which was
calculated by the present author. Many studies (Table 5.2), have found the relationship
between RD and hearing loss to be strong, with ORs of 4.0 to 46.8 (Agnostakis et al,
1982; Winkel et al, 1978; Abramovich et al, 1979; Leslie et al, 1995; McDonald, 1964).
Studies using continuous variables have also found significant findings (Bergman et al,
1985; Doyle et al, 1992; Salamy et al, 1989). After multivariate analysis, Leslie et a1
(1995) and Bergman et a1 (1985) continued to find respiratory variables to be strongly
associated with sensorineural hearing loss. Both researchers along with Doyle et al
(1992) believe that respiratory therapy or problems may just be a sign of sicker infants.
Leslie et a1 (1995) also postulates that hypoxia may damage the brain stem auditory
nuclei and that noise via oxygen humidifier, which is 24 dB higher than normal NICU
background noise, may be a possible confounder. Some researchers, however, have
failed to show significant findings between the 2 variables (Veen et al, 1993; Agnostakis
et al, 1982; Winkel et al, 1978; McDonald, 1964; Halpem et al, 1987; Cox et al, 1984;
Cox et al, 1992).

Other biologically plausible factors worth studying which have also so far
produced inconsistent results are the effects of hyperbilirubinemia, infections, and the
interactions homeostatic imbalances. The relationship between hyperbilirubinemia and
sensorineural hearing loss (Table 5.3) has been found to be strong with OR of 3.7 to 17.3
by some researchers (Agnostakis et al, 1982; Winkel et al, 1978; Abramovich et al,
1979). Significance also existed using continuous variables (Doyle et al, 1992; Bergman

et al, 1985; Clarke et al, 1986, Salamy et al, 1989). After multivariate analysis by

82

 

TABLE 5.2: Respiration in Relation to Hearing Loss

 

 

First Author Definition Variable Odds Ratio
(95% CI)
Agnostakis sensorineural apnea w/ bradychardia & cyanosis 46.8 (5.4-405.8)
Cox2 ABR fail apgar <6 @ 5min 20.0 (1.8-225.0)
Winkel sensorineural late asphyxia: cyanosis-apnea 8.1 (2.1-30.8)
Abramovich sensorineural apnea needing with intubation, 5.4 (1.4-20.7)
without ventilation
Veen sensorineural IPPV and/or CPAP 6.0 (1.3-27.2)
Leslie sensorineural max FiO2 >90 5.6 (1.2-26.9)
Leslie sensorineural oxygen >90 4.0 (1.1-15.6)
McDonald deaf cyanotic attacks 12.1 (3.1- 47.8)
McDonald deaf any respiratory difficulties 6.0 (1.7-21.1)
McDonald deaf oxygen therapy 4.3 (1.5-12.5)
Halpem impaired CPPV 1.8 (1 .0-3.4)
Halpem impaired perinatal asphyxia 2.0 (l .0-3.9)
Cox ABR fail apgar <6 @ 5min 4.3 (.9-20.7)
Veen sensorineural apgar @5 min <7 2.5 (.7-9.2)
Veen sensorineural IRDS clinical diagnosis 1.9 (.6-5.6)
Veen sensorineural apnea(315 s) or 1.4 (.4-4.2)
w/bradychardia<1 OO/min
Agnostakis sensorineural CPPV 3.6 (.3-38.6)
Winkel sensorineural birth asphyxia: apgarg7, l-5min 1.1 (.3-3.4)
McDonald deaf white asphyxia 3.3 (.7-15.8)
McDonald deaf rib recession or grunting 4.0 (.7-24.3)
respiration
McDonald deaf resuscitation 4.6 (.6-37.0)
McDonald deaf blue asphyxia 1.3 (.4-3.8)
Halpem impaired apgar _<_3 1.5 (.6-3.6)
Halpem impaired respiratory distress .6 (.3-1.1)
Cox2 ABR fail respiratory distress syndrome 3.4 (.8-15.9)
severe
Cox ABR fail apnea of prematurity 2.2 (.6-8.0)
Cox ABR fail respiratory distress syndrome 2.7 (.5-13.6)
Cox ABR fail apgar <6 @ 1min 1.7 (.5-6.0)
Cox2 » ABR fail respiratory distress syndrome mild .3 (.0-3.l)
Cox2 ABR fail gpgar <6 @ 1min .3 Q27)

 

83

Bergman et al (1985) and Clarke et al (1986), hyperbilirubinemia remained significant. A
level above 20mg over 100ml was suggested to cause damage to the auditory pathway,
particularly the cochlear nucleus (Fisch, 1961; Clarke et al, 1986; Bergman et al, 1985;
Leslie et al, 1995). Some researchers, however, have not found significant results (Veen

et al, 1993; McDonald, 1964; Halpem et al, 1987).

 

TABLE 5.3: Bilirubin in Relation to Hearing Loss

 

 

First Author Definition Variable Odds Ratio
(95% CI)
Agnostakis sensorineural hyperbilirubin >14mg/dl 17.3 (3.3-91.4)
Agnostakis sensorineural hyperbilirubin: hemolytic process 6.9 (1.4-34.8)
Winkel sensorineural hyperbilirubinemia 4.8 (1.0-22.7)
Abramovich sensorineural bilirubin >170micromole/l 3.7 (1.0-14.2)
Veen sensorineural bilirubin >200micromol/l 1.8 (.6-5.4)
McDonald deaf jaundice 2.0 (.8-5.5)
Halpem impaired bilirubinemia >12 1.0 (.4234

 

Some researchers have found a strong relationship between infections and
sensorineural hearing loss (Table 5.4) with an OR of 4.3 to 8.2 (Veen et al, 1993;
Halpem et al, 1987) even after multivariate analysis (Halpem et al, 1987). Others have
failed to find significant findings (Abramovich et al, 1979; Agnostakis et al, 1982;
Halpem et al, 1987; Cox et al, 1992). Studies that group the infections rather than
analyze the infections separately have a higher probability of finding significant results
since the exposure rate is relatively small in developed countries. Some of the congenital
perinatal infections known to cause sensorineural hearing loss include toxoplasmosis,
rubella, cytomegalovirus, herpes, and syphilis (Halpem et al, 1987; American Speech-
Language Hearing Association, 1994). Because the relationship between TORCH

infections and sensorineural hearing loss has been well documented and understood,

84

additional research may not be helpful since they account for a small percentage of

hearing impairment in this country (Halpern et al, 1987).

 

TABLE 5.4 Infections in Relation to Hearing Loss

 

 

First Author Definition Variable Odds Ratio
(95% CI)
Veen sensorineural sepsiszwhite blood count/+ blood 4.7 (1.4-15.3)
culture

Halpem impaired TORCH infection 8.2 (2.7-25.0)
Halpem impaired meconium aspiration 4.3 (2.0-9.2)

Abramovich sensorineural meningitis 11.1 (.6-193.0)
Agnostakis sensorineural sepsis 3.0 (.6-13.7)
Agnostakis sensorineural positive culture 2.5 (.5-14.1)
Abramovich sensorineural rubella 5.5 (.5-66.7)
Veen sensorineural congenital infection 2.4 (.3-18.7)
Agnostakis sensorineural meningitis 2.7 (.3-26.7)
Agnostakis sensorineural clinical diagnosis: infection 1.7 (.2-16.2)
Cox ABR fail infection 2.9 (.6—15.2)

 

Homeostatic imbalances have been found by some researchers to be strongly

related to sensorineural hearing loss (Table 5.5) with an OR of 5.6 to 9.9 (Agnostakis et

al, 1982; Leslie et al, 1995). Bergman et al (1985) found that low serum sodium

remained significant after multivariate analysis. Some researchers postulate that

homeostatic imbalance (such as acidosis, hypothermia, hyperosmolarity, or

hypoalbumineria) can serve as effect modifiers to sensorineural hearing losses interacting

with bilirubin, infection, or aminoglycosides (Bergman et al, 1985; Leslie et al, 1995).

Leslie et a1 (1995) hypothesizes that hyponatraemia may damage the auditory brainstem

nuclei by causing demyelination and alkalosis could damage the auditory pathway

through decreased blood flow. Others believe that this may just be another representation

of severely ill infants (Bergman et al, 1985). Some studies have not found significant

results (Veen et al, 1993; Abramovich et al, 1979).

85

 

TABLE 5.5: Homeostatic Imbalances in Relation to Hearing Loss

 

 

First Author Definition Variable Odds Ratio
(95% CI)
Agnostakis sensorineural Hypothermia 9.9 (2.3-43.0)
Leslie sensorineural max pH > 7.6 (1.1-89.0)
Leslie sensorineural plasma Na <125 5.6 (1.1-27.8)
McDonald deaf Oedema 3.3 (1.2-8.9)
Agnostakis sensorineural hypoglycemia: <25mg/dl 3.2 (.8-13.0)
Veen sensorineural pH <7.1 within 30 min 3.1 (615.2)
Veen sensorineural hypothermia first day <35.5 1.4 (.5-4.4)
degrees Celsius
Abramovich sensorineural pH <7.2 w/in 2hrs 2.5 (.4-15.7)

 

Whenever possible OR and CI were calculated from past studies for these
variables with the exception of Leslie et a1 (1995) who reported the odds ratios. Other
variables which were not biologically plausible or produced consistently insignificant
relationships such as comorbid conditions, aminoglycosides, some host and delivery
factors can be seen in the appendix.

Although the present author has shown that use of hearing aids and audiologist’s
review (including untyped) are related to possible causal factors for sensorineural hearing
loss, they are only the most usable measures available thus far in the NBH data. Results
using hearing aids and audiologist’s review can be reported but should remain
preliminary in this cohort. Inter-rater reliability testing on the cases sent to both
audiologists is not yet completed. The audiologist’s review which was completed must
go through further data cleaning. To produce a more sensitive and specific measure,
sensorineural hearing loss must be assessed using pure-tone audiometry using both air
and bone-conduction during a subsequent follow-up period, possibly at age 14. A
daughter grant using the “Small Grant Program For The National Institute On Deafness

And Other Communication Disorders” may be applicable. Two other possible grants can
86

be obtained through “Grants For Health Services Dissertation Research” and “Child And
Adolescent Development And Psychopathology Research Centers”.

Of the 28 carefully reviewed studies, 11 performed univariate analysis specific to
sensorineural hearing loss, 3 performed unspecific univariate analysis, and only 5
performed multivariate analysis. The studies in the single source category performed the
most inferential analysis. Seven (78%) performed univariate analysis specific to
sensorineural hearing loss and 3 (3 3%) multivariate analysis. Two (20%) of the
conditional multiple source studies performed univariate analysis specific to sensorineural
hearing loss, 2 (20%) performed unspecific (confirmed hearing loss and ABR fail)
univariate analysis, and only 1 (10%) multivariate analysis. Only 1 (25%) of the studies
from the other group performed unspecific (deafness) univariate analysis specific while
none performed multivariate analysis. Two (40%) of the unconditional multiple source
group studies performed univariate analysis specific to sensorineural hearing loss while
only 1 (25%) performed multivariate analysis. If causal analysis of sensorineural hearing
loss is the main objective of the low birth weight researcher, designing a study using a
very sensitive single source measure for sensorineural hearing loss may offer a higher
probability of successfully obtaining enough power (more cases).

The measures of central tendency (mean, median, and mode) from the causal
(performed analysis) and the non-causal (did not perform analysis) studies which
obtained specific sensorineural hearing loss variables were compared. The causal studies
had a mean of 15.3 cases with a median of 12.0 and a mode of 12.0. The non-causal
studies had a mean of 5.1 cases with a median of 4.5 and a mode of 2. To calculate the

approximate number of cases needed to explore causal analysis such as germinal matrix

87

 

intraventricular hemorrhage and sensorineural hearing loss in the NBH study, a sample
size calculation for cohort studies was performed (Appendix F). Using 20.7% as the
estimated exposure rate, 4.9 as the estimated risk, an alpha of .05, and power of 60, the
approximate number was 18 cases. If a higher power of 80 is desired, approximately 36
cases would be needed. In order to perform inferential analysis a minimum number of 12
to 18 sensorineural hearing loss cases should be obtained. A single source design study
may be the most successful. The single source studies produced the highest mean of
14.2, the unconditional multiple source studies produced a mean of 14.0, and the
conditional multiple source studies produced a low mean of 5.6.

Hearing losses defined through surveys and multiple sources are not suggested
and should only be used as last resorts, if time, money, and staff are lacking. Surveys are
the cheapest and easiest measure to obtain, but the least objective technique. Multiple
sources could be a reliable way of assessing sensorineural hearing loss only if all exams
are given to and completed by a large proportion of the children. A study that uses both
CPA and whisper on all children would be more reliable than a study that uses CPA for
50% of the children and whisper test for the other half. Most of the studies in the
multiple source group used many different assessments to patch and paste information
where data would otherwise be missing. Although this technique provides enough cases,
it may not be the most reliable design.

Another past alternative was to perform a traditional screen such as the ABR,
VRA, or BOA (to capture but not type possible hearing impairment) then referring the
children who failed the screen to receive free formal audiology assessments at

participating audiology centers. In studies with a nonexistent loss to follow-up rate, this

88

is a viable and very cost effective option. Unfortunately, past low birth weight studies
along with our own have shown loss to follow-up (an important epidemiological issue)
rates to formal audiology assessments after screening of 23% to 54%. If the Veen et a1
(1993) study would have relied on a conditional multiple source design, an estimated
23% (3 cases) to 54% (7 cases) could have been lost. Ten to 6 sensorineural hearing loss
cases may not have been enough to permit inferential analysis. For some children in low
birth weight populations, the screen may be the only chance to assess hearing. Since
traditional screens do not contain enough information to type the hearing loss, the
researcher must rely heavily on the formal audiology follow-up assessment. A highly
prevalent disease may be more forgiving of moderate follow-up rates. Since the
prevalence rate of sensorineural hearing loss is very low, every single case lost to follow-
up dramatically decreases the power to perform inferential analysis. Further investigation
should explore the reasons for these losses to follow-up. Implementing interventions,
which prevent these losses, should be provided before attempting this conditional
multiple source design.

Although nothing can replace a single source design using a formal audiology
exam performed by certified or licensed audiologists at audiology offices, financial and
loss to follow-up constraints may make it impossible to produce a representative sample
in the low birth weight populations. In an ideal situation, all children should have a
formal audiology assessment performed at an audiology office which contains a sound
proof room by a certified or licensed audiologist, who is trained in pediatric audiology.
This, however, would take an enormous amount of funding. In the east coast, a formal

audiology assessment (with pure-tone audiometry testing air and when necessary bone

89

conduction to define sensorineural versus conductive hearing loss) costs $80.00-$100.00
per person. For the NBH study, this would mean a total of $56,000.00-$70,000.00, an
estimate for 700 children. Some of the audiology centers contacted in the area were
willing to give 50% discounts for academic research, decreasing the price to $28,000.00
to $35,000.00. This may only be an option for a study with a small sample size.

Even if a gracious sponsor was found to fund all 700 assessments, another
epidemiological problem would be disproportionate loss to follow-up. In the low birth
weight population a large proportion of the sample report low socioeconomic conditions.
Let’s compare, for example, the possible hurdles overcome by a mother with a high
socioeconomic status versus the hurdles overcome by a mother with a low socioeconomic
status to transport a child to the audiology center for a free hearing assessment. On
average a mother with high socioeconomic status will have less children, own an
automobile, and can afford to take half a day off of work. On average a mother with low
socioeconomic status will have more children (needing a possible baby-sitter), does not
own an automobile (may need to take a bus or the subway), and may not be able to afford
time off of work. The hurdles facing the low socioeconomic families are not new
problems in research and have been shown to cause disproportionate loss to follow-up. A
solution has been to offer evaluations in the homes if a family can not make it to the site.
Assimilating formal audiological assessments in the field assists in eliminating
disproportionate loss to follow-up. It is also a cost effective, yet, can be a sensitive
enough tool to type sensorineural hearing loss and perform inferential analysis.

Successful single source protocols, like the Veen et al (1993) and Clarke et a1

(1986) studies, need to be followed for the sake of achieving multivariate analysis. Both

90

studies utilized audiometers (Hortman DA 323 and Maico Model 24, respectively) with
earphones to test pure-tone air-conduction sensitivity as well as a vibrator to test pure-
tone bone-conduction sensitivity when necessary. Both studies tested hearing in each ear
using 500, 1000, 2000, and 4000Hz. Clarke et a1 (1986) additionally tested at 250, 3000,
6000, and 8000Hz. Threshold levels were recorded by both. Both studies applied
standard techniques to type sensorineural hearing loss (where bone-conduction hearing
threshold level patterns resembled air-conduction threshold level) and conductive hearing
loss (where bone-conduction hearing threshold levels were normal). Typing of unilateral
losses were made possible through masking in both studies. Both hearing evaluations
were performed by the authors {pediatricians in Veen et a1 (1993) and psychologists in
Clarke et al (1986)} who were trained to use the testing equipment. Clarke et a1 (1986)
also had a sound-treated booth which met ANSI standard custom-built for the study. The
above simulations to formal audiology assessments created sensitive enough screens to
permit multivariate analysis.

Without a sound proof room, <25 dB HL may not be detected. Because of this,
all past publications without a sound chamber produced an elevated population curve,
which consists of false negatives (children who may have failed with lower stimulus
levels). Misclassification errors can dilute analysis, producing weaker or no associations
when they really exist. Since sensorineural hearing loss has a low prevalence rate, the
power to accomplish causal research is limited. Producing a lower screening level could
increase power. Those who used EPA with sound proof rooms had higher rates (5.9 to
19.0%) than the one which did not (1.5%). It is estimated that an evaluation like Veen et

a1 (1993) could produce 11 sensorineural hearing loss cases and an evaluation like Clarke

91

et a1 (1986) could produce 41 cases of sensorineural hearing loss cases in approximately
700 NBH children. The higher prevalence rate may be attributed to the more thorough
evaluation used by Clarke et al (1986). The additional frequencies and the sound proof
booth probably produced higher sensitivity to formal mild and moderate hearing losses.

In conclusion to this chapter, if the main objective of a researcher is to perform
inferential analysis on sensorineural hearing loss, then a single source design (using an
age appropriate audiometric assessment to type the cases) may be the most successful in
producing enough cases in the low birth weight population. If money is no object and
losses to follow-up are not expected, then a formal audiology assessment performed at an
audiology center in a sound proof room and by a certified or licensed audiologist trained
in pediatric audiology would be the best option. If money and losses to follow-up are an
issue, then an evaluation like Clarke et a1 (1986) using additional frequencies and a sound
proof booth would be the next best option. More cases can be obtained by lowering the
threshold used in the case definition. If in addition to money and losses to follow-up,
sound proof booth and time are problems then an evaluation such as Veen et al (1993)
would be suggested. All those who fail the screen should still be referred to a
participating audiology center for intervention. But, the data will not be hindered by
relying only on the formal audiology assessment for typing. An improvement to loss to
follow-up may be to have the formal assessments performed by a certified or licensed
audiologist trained in pediatric audiology at the sites and homes of all the children or, if
not affordable, only of those who failed. Dealing with the problem of background noise
will be a key element in the success of any protocol. Background noise can make it

impossible to conduct a valid audiologic evaluation. After these suggestions of design

92

 

have been considered can the present author support that the NBH study has the best
possible definition for sensorineural hearing loss and therefore generate the best quality
causal research.

Lastly, causal (inferential) analysis of sensorineural hearing loss requires a hybrid
approach of epidemiology and audiology. Issues in both fields must be recognized to
develop solutions which will be successful and acceptable by epidemiologists and
audiologists. This thesis attempted to examine possible methodological limitations
present in sensorineural hearing loss field research and inferential analysis. The
suggested solutions and information in this work have taken half a decade of research,
thought, and edits. But the present author, still a neophyte in both the world of
epidemiology and audiology, invites the experts to tackle these inferential analysis issues
in sensorineural hearing loss field research and to design better solutions to the problems
introduced. The author hopes, that this work, helps to inspire and instigate seasoned
researchers of sensorineural hearing loss to improve the methods applied in field research,

where inferential analysis can be achieved with minimal error.

93

APPENDICES

94

APPENDIX A

 

TABLE A: Descriptive Analysis of Survivors with and without Prenatal Health

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Forms
Prenatal Health Forms
Variables without with
Gestational Age Weeks mean
(sd) 29.8(6.9L 30.4(6.3)
Weight mean (sd) 1451.8(363.3) 1497.6(355.7)
Maternal Age mean (sd)
26.5(6.0) 27.0(5.7)
Paternal Age
mean (sd) 29.7(7.1) 30.2(6.7)
Race 11 (%)
Caucasian 229(66.6%) 373(68.8%)
African American 98(28.5%) 135(24.9%)
Hispanic 10(2.9%) 21 (3.9%)
Asian 3(.9%) 9(1 .7%)
Gender n (%)
Male 185(53 .0%) 266(48.7%)
Welfare n (%) 41(23.7%) 96(21.8%)
Unmarried n (%) 57(31.1%) 136(29.1%)
College Father
n (%) 52(34.0%) 152(39.6%)
Father Unemployed n (%)
12(6.4%) 37(7.8%)
College Mother
n (%) 73(3_9.0%) 209(44.4%)
Mother Unemployed n (%)
40(21 .3%) 90(19.0%)

 

 

95

 

APPENDIX B

EXCLUDING CHILDREN TESTED AT HIGHER SCREENING LEVELS

No significant differences in host factors (GA, weight, race, gender, marital status,
welfare, and maternal and paternal education and employment) were found with children
tested at 20 and 40 dB SL and children tested at higher screening stimuli.

Excluding children tested at higher screening levels did not significantly change the
results using all children tested in Chapter 4.

Maternal report and fail to respond to at least 1 screening level
OR = 4.3 (.87, 21.4)
Specificity = 96.8%
Sensitivity =12.5%
Positive Predictive Value = 15.4%

MD suspicion and fail to respond to at least 1 screening level
OR = 4.0 (.83, 20.0)
Specificity = 96.8%
Sensitivity = 11.8%
Positive Predictive Value = 15.4%

Abstractor judged and fail to respond to at least 1 screening level
OR = 3.7 (.75, 18.0)

Concern on medical chart and fail to respond to at least 1 screening level
OR = 1.6 (.44, 5.8)

Excluding children tested at higher screening levels may change the prevalence of
sensorineural and overall hearing loss in Chapter 3 .

293 children tested at higher screening level (283 passed and 10 failed)

385 children tested at normal screening level using the VRA

10 children tested using earphones

7 children missing report of screening level

385+10+10=405

Possible Estimates Had All Children Been Tested at Normal Screening Levels

32 failed VRA 14 possible sensorineural hearing losses
32/405 = .079 based on the VRA fail at any 1 level and
283*.079 = 22 bilateral tympanogram information
22 + 32 = 54 14/405 = .035
54/688 = .078 283*.035 = 10
.078 compared to .046 10 + 14 = 24

24/688 = .035

.035 compared to .020

96

APPENDIX C

SENSITIVITY AND SPECIFICITY VALUES OF NEEDING OR WEARING
HEARING AIDS SUBSTITUTED FOR SENSORINEURAL HEARING LOSS

Sensitivity = Number with disease who have agnositive test
Number with disease
Specificity = Number without disease who have a negative test
Number without disease Lilienfeld & Stolley (1994)

UNILATERAL SENSORINEURAL HEARING LOSS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Veen et al (1993) EPA
+ -
Hearing + 9 0 9
Aid - 4 877 881
13 877 890
Sensitivity = 9/13 = 69%
Specificity = 877/877 = 100%
McClelland et al (1992) EPA
+ -
Hearing + 5 0 5
Aid - 2 55 57
7 55 62
Sensitivity = 5/7 = 71%
Specificity = 55/55 = 100%
Abramovich et al (1979) EPA
+ -
Hearing + 8 0 8
Aid - 2 101 103
10 101 111
Sensitivity = 8/10 = 80%
Specificity =101/101 =100%
Bradford et al (1985) EPA
+ -
Hearing + 9 0 9
Aid - O 112 112
9 112 121

 

 

 

 

Sensitivity = 9/9 = 100%
Specificity = 112/112 = 100%

97

 

 

 

 

APPENDIX C

SENSITIVITY AND SPECIFICITY VALUES OF NEEDING OR WEARING
HEARING AIDS SUBSTITUTED FOR SENSORINEURAL HEARING LOSS

 

 

 

 

 

 

 

 

 

 

Watkin et al (1991) EPA
+ -
Hearing + 12 0 12
Aid - 0 310 310
12 310 322
Sensitivity = 12/ 12 = 100%
Specificity = 310/310 = 100%
Doyle et al (1992) EPA
+ ..
Hearing + 1 0 l
Aid - 3 38 41
4 38 42

 

 

 

 

Sensitivity = '/4 = 25%
Specificity = 38/38 = 100%

BILATERAL SENSORINEURAL HEARING LOSS

 

 

 

 

 

 

 

 

 

 

Veen et al (1993) EPA
+ -
Hearing + 9 0 9
Aid - 2 879 881
11 879 890
Sensitivity = 9/11 = 82%
Specificity = 879/879 = 100%
McClelland et al (1992) EPA
+ -
Hearing + 5 0 5
Aid - 1 56 57
6 56 62

 

 

 

 

Sensitivity = 5/6 = 83%
Specificity = 56/56 = 100%

98

 

 

 

 

APPENDIX C

SENSITIVITY AND SPECIFICITY VALUES OF NEEDING OR WEARING
HEARING AIDS SUBSTITUTED FOR SENSORINEURAL HEARING LOSS

Abramovich et al (1979) EPA

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

+ -
Hearing + 8 0 8
Aid - 0 103 103
8 103 111
Sensitivity = 8/8 = 100%
Specificity = 103/103 = 100%
Bradford et al (1985) EPA
+ -
Hearing + 8 1 9
Aid - 0 108 108
8 109 117
Sensitivity = 8/8 = 100%
Specificity =108/109 = 99%
Watkin et al (1991) EPA
+ -
Hearing + 10 0 10
Aid - 2 310 312
12 310 322
Sensitivity = 10/ 12 = 83%
Specificity = 310/310 = 100%
Doyle et al (1992) EPA
+ -
Hearing + l 0 1
Aid - O 41 41
1 41 42

 

 

 

 

Sensitivity = 1/1 = 100%

Specificity =41/41 = 100%

99

 

 

 

 

APPENDIX D

CORONADO PROGRAM FOR SPSS TO CALCULATE MULTIPLE ODDS RATIONS
WITH 95% CONFIDENCE INTERVALS FROM LITERATURE REVIEWS USING
THE WOOLF METHOD

. Name header variables study, outcome, cause, a, b, c, d

. Enter information in appropriate header column for study, outcome variable, causal
variable, and cell frequency from outcome/causal 2X2 table
Outcome Variable
+ -
Causal + a b
Variable - c d

 

 

 

 

 

 

. Calculating Odds Ratios

Transform 9 Compute 9 Name Target Variable “oddratio” 9 Enter numeric
expression as (a*d)/(b*c) 9 OK ‘

. Calculating natural log Odds Ratios
Transform 9 Compute 9 Name Target Variable “lnor” 9 Enter numeric
expression as LN function (oddratio) 9 OK

. Calculating variance
Transform 9 Compute 9 Name Target Variable “varlnor” 9 Enter numeric
expression SQRT function (l/a + Nb + l/c + l/d) 9 OK

. Calculating Z times variance
Transform 9 Compute 9 Name Target Variable “strlnor” 9 Enter numeric
expression as 1.96*varlnor 9 OK

. Calculating lower limits
Transform 9 Compute 9 Name Target Variable “lowlim” 9 Enter numeric
expression EXP function (lnor - strlnor) 9 OK

. Calculating upper limits
Transform 9 Compute 9 Name Target Variable “upperlim” 9 Enter numeric
expression EXP function (lnor + strlnor) 9 OK

Odds Ratios and Confidence Intervals will be directly outputted onto the data window.

100

APPENDIX E

 

TABLE E: Variables of Risk not Biologically Plausible or Consistently

Insignificant in Relation to Hearing Loss

 

 

 

 

 

 

 

First Author Definition Variable Odds Ratio
(95% CI)
Halpem impaired craniofacial anomalies 8.1 (2.3-27.7)
Winkel sensorineural exchange transfusion 1.4 (.5-4.2)
Campenelli sensorineural race: CA 13.1 (1.7-103.7)
Abramovich sensorineural singleton 3.6 (1.0-13.5)
Abramovich sensorineural born in UCH 2.1 (.6-7.6)
Veen sensorineural male 1.2 (.4-3.4)
Abramovich sensorineural male 1.4 (.4-5.0)
Campenelli sensorineural male 1.6 (.3-8.1)
Abramovich sensorineural small for GA .3 (.0-2.8)
Halpem impaired birth weight <1500 .5 (.3-.9)
Halpem impaired gestational age <38 .3 (.l-.5)
Cox ABR male 2.6 (.6-8.1)
Cox ABR intrauterine grth failure 1.0 (.3-3.9)
Cox ABR race: CA .6 (.2-2.3)
Cox ABR maternal education: high school .8 (.2-4.3)
Cox ABR social class: class 1-3 .6 (.1-2.4)
Cox ABR multiple birth .3 (.1-1 .7)
Cox ABR social class: class 4-5 .2 (.l-.8)
Abramovich sensorineural hyaline membrane disease 1.3 (.3-4.8)
Halpem impaired retrolental fibroplasia .6 (.2-1.7)
Cox ABR neurological abnormality @8 1.8 (.3-11.0)
years
Cox ABR neurological abnormality @20 1.7 (.1- 20.5)
mos
Bergman deaf kanamycin 3.7 (.9-14.9)
Bergman deaf ampicillin 2.3 (.9-6.2)
Winkel sensorineural kanamycin 1.0 (.3-2.8)
Agnostakis sensorineural gentamicin 1.4 (.3-6.2)
Abramovich sensorineural streptomycin 2.5 (.3-19.7)
Agnostakis sensorineural kanamycin .9 (.2-3.5)
Abramovich sensorineural kanamycin 1.4 (.2-9.3)
Abramovich sensorineural dextrostix <25 .7 (.1-5.9)
Halpem impaired aminoglycosides 1.1 (.6-2.0)
Veen sensorineural bradycardia <100/min w/out apnea 2.0 (.7-5.9)
Agnostakis sensorineural congenital heart disease 2.7 (.3-26.7)
Abramovich sensorineural heart rate <100 @birth .7 (.1-3.4)

 

101

APPENDIX E

 

TABLE E: Variables of Risk not Biologically Plausible or Consistently

Insignificant in Relation to Hearing Loss

 

 

 

First Author Definition Variable Odds Ratio
(95% CI)

Winkel sensorineural incubator 1.5 (.4-5.8)
Halpem impaired intubated 1.5 (.6-3.7)
Abramovich sensorineural forcep 1.7 (.4-7.3)
Abramovich sensorineural vertex 1.4 (.4-5.2)
Abramovich sensorineural breech 1.1 (.2-5.5)
McDonald deaf bleeding 1.5 (.5-4.4)
McDonald deaf toxemia(>150/100mmHg) .5 (.l-2.3)
McDonald deaf sugically induced labor .3 (.0-2.0)

 

102

APPENDIX F

ESTIMATED CASES USING SAMPLE SIZE CALCULATIONS
n = [1.96V—2p—q + .27\/—pEJ;-p;1;]2 / (pl- p0)2 Schlesselman (1984)
p0 = exposure rate among non-diseased

P1 = P0 R/l1+ Po(R-1)l

p = '/2(p1- po)

q = 1 - p

(11 = 1' PI

QO = 1- P0

01 == .05 Z = 1.96

l-B = .60 Z = .27

p0 = .207 exposure rate using hearing aid at age 2 or 9 data

p1 = .5612

p = .3841

q = .6159

ql = .4388

qo = .793

on = .05 Z = 1.96

l-B=.60 Z=.27 ifl-B=.80 Z=.84

= 18 cases 11 = 36 cases

103

REFERENCES

104

REFERENCES

Abramovich, S.J., Gregory, S., Slemick, M., Stewart, A. Hearing Loss in Very Low
Birthweight Infants Treated with Neonatal Intensive Care. Archives of Disease in
Childhood. 54: 421-426 (1979).

Agnostakis, D., Petmezakis, J ., Papzissis, G., Messaritakis, J ., Matsaniotis, N. Hearing
Loss in Low-Birth-Weight Infants. American Journal of Diseases in Childhood. 136:
602-604 (1982).

Albrecht, W. Z. Hals, Nase, und Ohrenheilk. 27: 495 (1930).

American Speech-Language-Hearing Association. Joint Committee on Infant Hearing
1994 Position Statement. 36 (12): 38-41 (1994).

American Speech-Language-Hearing Association. Guidelines for Audiologic Screening.
(1997)

Arehart, K. H., Yoshinaga-Itano, C., Thompson, V., Gabbard, S.A., Brown, A.S., State of
the States: The Statur of Universal Newborn Hearing Screening, Assessment, and
Intervention Systems in 16 States. American Journal of Audiology. 7 (2). [Online]
Available http://journals.asha.org/aj a, (1998).

Bergman, I., Hirsch, R.P., Fria, T.J., Shapiro, S.M., Holzman, I., Painter, M.J., Cause of
Hearing Loss in the High Risk Premature Infant. The Journal of Pediatrics. 106: 95-100
(1985).

Bjeere, I. Neurological Investigation of 5-Year-Old Children with Low Birthweight.
Acta Paediatr Scand. 64: 859-864 (1975).

Bosher, S.K. Neonatal Changes in the Endolymph and their Possible Relationship to
Peri-Natal Deafiress. Acta Otolaryngology. 73: 203-211 (1972).

Bradford, R.C., Baudin, J ., Conway, M.J., Hazel], J .W.P., Stewart, A.L., Reynolds,
E.O.R. Identification of Sensory Neural Hearing Loss in Very Preterm Infants by
Brainstem Auditory Evoked Potentials. Archives of Disease in Childhood. 60: 105-109
(1985)

Buch, NH. The Inner Ear of Newborn Infants. The Journal of Laryngology and Otology.
16: 765-777 (1966).

Butler, K.G. Hearing Impairment and Language Disorders: Assessment and
Intervention. Gaithersburg, An Aspen Publication (1994).

105

Buttross, S.L., Gore Gearhart, J ., Peck, J .E. Early Identification and Management of
Hearing Impairment. American Family Physician. 51 (6): 1437-1446 (1995).

Campenelli, P.A., Pollock, F.J., Henner, R. In Oto-Audiological Evaluation of Forty-
Four Premature Children. Archives of Otolaryngology. 67: 609-615 (1958).

Clarke, B. R., Conry, R.F. Hearing Impairment in Children of Low Birth Weight. The
Journal of Auditory Research. 19: 277-291 (1979).

Clarke, B.R., Conry, R.F., Dunn, H.G. Hearing Impairment and Associated Factors.
Sequelae of low birthweight: The Vancouver Study/ Edited by Henry G. Dunn.
Philadelphia, J. B. Lippincott Co. 147-165 (1986).

Corner, B. Prematurity. London, Cassell (1960).

Cox, C., Hack, M., Aram, D., Borawski, E. Neonatal Auditory Brainstem Response.
Pediatric Research. Pediatric Research. 31 (1): 68-72 (1992).

Cox, L.C., Hack, M., Metz, D.A. Auditory Brain Stem Response Abnormalities in the
Very Low Birthweight Infant: Incidence and Risk Factors. Ear and Hearing. 5 (1): 47-
51 (1984).

Cox, L.C., Hack, M., Metz, D.A. Brainstem Evoked Response Audiometry in the
Premature Infant Population. Journal of Pediatric Otorhinolaryngology. 3: 213 (1981).

Dann, M., Levine, S.Z., New, E.V. A Long-tenn F ollow-up Study of Small Premature
Infants. Pediatrics. 34: 945 (1964).

Davey, P.R. Hearing Loss in Children of Low Birth Weight. Journal of Laryngology and
Otology. 76: 274-277 (1962).

Doyle, L.W., Keir, E., Kitchen, W.H., Ford, G.W., Rickards, A.L., Kelly, E.A.
Audiologic Assessment of Extremely Low Birth Weight Infants: A Preliminary Report.
Pediatrics. 90: 744-749 (1992).

Drillien, C.M. Aetology and Outcome in Low Birth Weight Infants. Developmental
Medicine and Child Neurology. 14: 563-574 (1972).

Drillien, C.M., Thomson, A.J.M., Burgoyne, K. Low-birthweight Children at Early

School-age: A Longitudinal Study. Developmental Medicine and Child Neurology. 22:
26-47 (1980).

106

Duara, S. Suter, C.M., Bessard, K.K, Gutberlet, R.L. Neonatal Screening with Auditory

Brainstem Responses: Results of Follow-up Audiometry and Risk Factor Evaluation.
The Journal of Pediatrics. 108 (2): 276-281 (1986).

Durkin, M.S., Davidson, L.L., Desai, P., Hasan, M., Khan, N., Strout, P.E., Thomburn,
M.J., Wang, W., Zaman, S.S. Validity of the Ten Questions Screen for Childhood
Disability: Results from Population-Based Studies in Bangladesh, Jamaica, and Pakistan.

Epidemiology. 5(3): 283-289 (1994).

Durkin, M.S., Hasan, Z.M., Hasan, K.Z. The Ten Questions Screen for Childhood
Disabilities: Its Uses and Limitations in Pakistan. Journal of Epidemiology and
Community Health. 49 (4): 431-6 (1995).

Durkin, M.S., Wang, W., Shrout, P.E., Zaman, S.S., Hasan, Z.M., Desai, P., Davidson,
L.L. Evaluating a Ten Questions Screen for Childhood Disability: Reliability and
Internal Structure in Different Cultures. Journal of Clinical Epidemiology. 48 (5): 657-
666 (1995).

Ellsman, S., Matkin, N., Sabo, M. Early Identification of Congenital Sensorineural
Hearing Impairment. The Hearing Journal. 40: 13-17 (1987).

Fisch, L. Hyperbilirubinaemia and Perceptive Deafness. British Medical Journal. 142-
144 (1961).

F isch, L. Congenital Deafness and Haemolytic Disease of the Newborn. Archives of
Disease in Childhood. 309-316 (1954).

F lexer, C. Facilitation Hearing and Listening in Young Children. San Diego, Singular
Publishing Group, Inc. (1994).

Galambos, R., Despland, PA. The Auditory Brainstem Response Evaluates Risk Factors
for Hearing Loss in the Newborn. Pediatric Research. 14: 159-163 (1980).

Galambos, R., Hicks, Wilson, M.J. Hearing Loss in Graduates of a Tertiary Intensive
Care Nursery. Ear and Hearing. 3: 87 (1982).

Gell, F .M., White, B. McC, Newell, K., Machenzie,1., Smith, A., Thompson, S., Hatcher,
J. Practical Screening Priorities for Hearing Impairment among Children in Developing
Countries. Bulletin of the World Health Organization. 70(5): 645-655 (1992).

Gerber, S. E. Review of a High Risk Register for Congenital or Early Onset Deafness.
British Journal of Audiology. 24: 347-56 (1990).

Gerkin, K.P. The High Risk Register for Deafness. American Speech-Language-Hearing
Association. 26: 17-23 (1984).

107

Gravel, IS. Personal Communication. Department of Otolarygology, Albert Einstein
College (1997).

Gravel, J.S., Traquina, D.N. Experience with the Audiologic Assessment of Infants and
Toddlers. International Journal of Pediatric Otorhinolaryngology. 23 (1): 59-71 (1992).

Greenberg, D.B., Wilson, W.R., Moore, J.M., Thompson, G. Visual Reinforcement
Audiometry (VRA) with Young Down's Syndrome Children. Journal of Speech and
Hearing Disorders. 43 (4): 448-458 (1978).

Hall, J.G. A Histopathological Investigation of the Auditory Pathways in Neonatal
Asphyxia. Acta Oto- Laryng. 368-375 (1961).

Halpem, J., Hosford-Dunn, H., Malachoski, N. Four Factors that Accurately Predict
Hearing Loss in “High Risk” Neonates. Bar and Hearing. 8 (1): 21-25 (1987).

Harper, P.A., Weiner, G. Sequalae of Low Birth Weight. Annals Review Medicine. 16:
405-420 (1965).

Harrison, M., Roush, J. Age of Suspicion, Identification, and Intervention for Infants and
Young Children with Hearing Loss: A National Study. Bar and Hearing. 17: 55—62
(1996).

Hegyi, T., Carbone, M. T., Anwar, M., Ostfeld, B., Hiatt, M., Koons, A., Pinto-Martin, J .,
Paneth, N. Blood Pressure Ranges in Premature Infants. The first hours of life. Journal
of Pediatrics. 124: 627-633 (1994).

Hegyi, T., Carbone, M. T., Anwar, M., Ostfeld, B., Hiatt, M., Koons, A., Pinto-Martin, J.,
Paneth, N. Blood Pressure Ranges in Premature Infants. The first week of life.
Pediatrics. 97: 336-342 (1996).

Holzman, C., Paneth, N., Little, R., Pinto-Martin, J. Perinatal Brain Injury in Premature
Infants Born to Mothers Using Alcohol in Pregnancy. Pediatrics. 95: 66-73 (1995).

Jellinek, M.S., Murphy, J .M., Robinson, J., F eins, A., Lamb, 8., F enton, T. Pediatric
Symptom Checklist: Screening School-age Children for Psychosocial Dysfunction.
Journal of Pediatric. 112 (2): 201-9 (1988).

Jacobson, J .T., Seitz, M.R., Mencher, G.T., Parrott, V. Auditory Brainstem Response: A
Contribution to Infant Assessment and Management. Early Management of Hearing
Loss/ Edited by Mencher, G.T., Gerber, S.E. New York, Grune & Stratton. 151-181
(1981)

108

Johnson, A., Ashurst, H. Screening for Sensorineural Deafness by Health Visitors.
Archives of Disease in Childhood. 65: 841—845 (1990).

Johnson, M.A., Cox, M. and McKim, E. Outcome of Infants of Very Low Birthweight: a
Geographically Based Study. Canadian Medical Association Journal. 136: 1157-1165
(1987)

Kelemen, G. Hemorrhage: A Specific Poison to Tissue of Ampullar Cupulae. Archives
of Otolarynogology. 77: 364-375 (1963).

Kelsey, J .L., Thompson, W.D., Evans, A.S. Methods in Observational Epidemiology.
New York, Oxford University Press (1986).

Lansioni, G.E., Coninx, F., Smeets, P.M. A Classical Conditioning Procedure for
the Hearing Assessment of Multiply Handicapped Persons. Journal of Speech and
Hearing Disorders. 54 (1): 88-93 (1989).

Lee, D.L., Gomez-Marin, 0., Lee, H.M. Prevalence of Childhood Hearing Loss.
American Journal of Epidemiology. 144(5): 442-450 (1996).

Leslie, G.I., Kalaw, M.B., Bowen, J.R., Arnold J .D. Risk Factors for Sensorineural
Hearing Loss in Extremely Premature Infants. Journal of Paediatric and Child Health.
31: 312-316(1995).

Liden, G., Harford, E.R. The Pediatric Audiologist: from Magician to Clinician. Ear
and Hearing. 6 (1): 6-9 (1985).

Lilienfeld, D.E., Stolley, P.D. Foundations of Epidemiology. New York, Oxford
University Press (1994).

Lloyd, B.W. Outcome of Very-Low—Birthweight Babies from Wolverhampton. Lancet.
2 (8405): 739-741 (1984).

Lubchenko, L.O., Hornet, F .A., Reed, L.H., Hix, I.E., Metcalf, D., Cohig, R., Elliott,
H.C., Bourq, M. Sequalae of Premature Birth- Evaluation of Premature Infants of Low
Birth Weights at Ten Years of Age. Arrierican Journal of Diseases in Childhood. 106:
101-115 (1963).

McClelland, R.J., Watson, D.R., Lawless, V., Houston, H.G., Adams, D. Reliability and
Effectiveness of Screening for Hearing Loss in High Risk Neonates. British Medical
Journal. 304: 806-809 (1992).

McCormick, B. Screening for Hearing Impairment in Young Children. London, Croom
Helm (1988).

109

McCormick, M. The Credibility of the ACTOBAT Follow-up Study. Pediatrics. 99 (3):
476-478 (1997).

McDonald, A.D. Deafness in Children of Very Low Birth Weight. Archives of Disease
in Childhood. 39: 272-277 (1964).

McMaster Surveys. Hamilton-Wentworth Region Study. Departments of Pediatrics and
Clinical Epidemiology and Biostatistics, Faculty Health Sciences, MacMaster University,
Canada (unpublished).

Martin, J .A.M., Bentzen, O., Colley, J .R.T., Hennebert, D., Holm, C., Iurato, S., Jonge,
de A., McCullen, 0., Meyer, M.I., Moore, W.J., Morgon, A. Childhood Deafiress in the
European Community. Audiology. 21: 185-196 (1982).

Mauk, G., White, K., Mortensen, L., Behrens, T. The Effectiveness of Screening
Programs Based on High-Risk Characteristics in Early Identification of Hearing
Impairment. Ear and Hearing. 12: 312-319 (1991).

Mercer, H.P., Lancaster, P.A., Weiner, T., Gupta, J.M. Very Low Birthweight Infants: a
Follow-up Study. Medical Journal of Australia. 2(13): 581-584 (1978).

Moore, J .M., Thompson, G., Folsom, R.C. Auditory Responsiveness of Premature
Infants Utilizing Visual Reinforcement Audiometry (VRA). Ear and Hearing. 13 (3):
187-194 (1992).

NBH Surveys. The Neonatal Brain Hemorrhage Study. Division of Epidemiology and
Division of Child and Adolescent Psychiatry, Columbia University; University of
Pennsylvania School of Nursing and School of Medicine, Department of Pediatrics,
Division of Biostatistics and Epidemiology (unpublished).

Paneth, N., Rudelli, R., Kazam, E., Monte, W. Brain Damage in the Preterm Infant.
Cambridge, Cambridge University Press (1994).

Paneth, N, Pinto-Martin, J. The Epidemiology of Germinal Matirx Intraventricular
Hemorrhage. Reproductive and Perinatal Epidemiology (Kiely, M). Boca Raton, FL,
CRC Press. 371-399 (1990).

Parving, A., Christensen, B. Children Younger than 4 Years of Age, Referred to an
Audiology Department. International Journal of Pediatric Otorhinolaryngology. 23 (2):
161-170 (1992).

Pinto-Martin, J. Personal Communication. University of Pennsylvania School of
Nursing and School of Medicine, Department of Pediatrics, Division of Biostatistics and
Epidemiology (1997).

110

Pinto, J ., Paneth, N., Kazam, E., Kairam, R., Rose, W., Rosenfeld, D., Witomski, T.,
Stern, 1., Schonfeld, S. Inter-observer Variability in Neonatal Cranial Ultrasonography.
Paediatric and Perinatal Epidemiology. 2: 45-60 (1988).

Pinto-Martin, J., Paneth, N., Witomski, T., Stern, I., Schonfeld, S., Rosenfeld, D., Rose,
W., Kazam, E., Kairam, R., Katsikiotis, V., Susser, M. The Central New Jersey Neonatal
Brain Hemorrhage Study: Design of the Study and Reliability of Ultrasound Diagnosis.
Paediatric and Perinatal Epidemiology. 6: 273-284 (1992).

POPs surveys. The Collaborative Project on Preterm and Small for Gestational Age
Infants in the Netherlands. Departments of Pediatric and Medical Statistics, University
Hospital, Leiden, the Netherlands (unpublished).

Primus, M.A. Infant Thresholds with Enhanced Attention to the Signal in Visual
Reinforcement Audiometry. Journal of Speech and Hearing. 31 (3): 480-484 (1988).

Reuss, M., Clark, C., Paneth, N. Efficiency of Sampling: Birthweight, gestational age
distributions in two cohorts, <31 weeks and 500-1499g. Peadiatric and Perinatal
Epidemiology. 9: 341-350 (1995).

Roberts, J., Ahuja, E.M. Hearing Sensitivity and Related Medical Findings among
Youths 12-17 Years. National Center for Health Statistics (1975).

Roberts, J ., Federico, J .V. Hearing Sensitivity and Related Medical Findings among
Children. National Center for Health (1975).

Roberts, J.L., Davis, H., Phon, G.L., Reichert, T.J., Sturtevant, E.M., Marshall, RE.
Auditory Brainstem Responses in Preterm Neonates: Maturation and Follow-Up. The
Journal of Pediatrics. 101 (2): 257-263 (1982).

Robinson, N.M., Robinson, H.G. A Follow-up Study of Children of Low Birth Weight
and Control Children at School Age. Pediatrics. 35: 425 (1965).

Rudmin, F.W. Brief Clinical Report on Visual Reinforcement Audiometry with
Deaf Infants. Journal of Otolarygology. 13 (6): 367-369 (1984).

Ruth, R.A., Homer, J.S., McCoy, G.S., Chandler, C.R. Comparison of Auditory
Brainstem Response and Behavioral Audiometry in Infants. Scandanavian Audiology
Supplement. 17: 94-98 (1983).

Saigal, S., Rosenbaum, P., Stoskopf, B., Milner, R. Follow-up of Infants 501 to 1,500 gm

Birth Weight Delivered to Residents of Geographically Defined Region with Perinatal
Intensive Care Facilities. The Journal of Pediatrics. 100 (4): 606-613 (1982).

111

Saigal, S., Szatrnari, P., Rosenbaum, R, Campbell, D., King, S. Intellectual and
Functional Status at School Entry of Children Who Weighed 1000 grams or Less at Birth:
A Regional Perspective of Births in the 19803. The Journal of Pediatrics. 116 (3): 409-
416 (1990).

Salamy, A., Eldredge, L, Tooley, W.H. Neonatal Status and Hearing Loss in High-Risk
Infants. The Journal of Pediatrics. 114 (5): 847-852 (1989).

Salwen, K.M., Vikerfors, T., Olcen, P. Increased Incidence of Childhood Bacterial
Meningitis. Scandanavian Journal of Infectious Disease. 19: 1-11 (1987).

Schlesselman, JJ. Case Control Studies : Design, Conduct, Analysis. New York, Oxford
University Press (1982).

Schulte, F .J ., Stennert, E. Hearing Defects in Preterm Infants. Archives of Disease in
Childhood. 53: 269-270 (1978).

Siegel Institute, unpublished.

Spector, G.I., Pettit, W.J., Davis, G., Strauss, M., Rauchbach, E. F etal Respiratory
Distress Causing CNS and Inner Ear Hemorrhage. The Laryngoscope. 88: 764-786
(1978).

Stein, L., Chark, S., Kraus, N. The Hearing Impaired Infant: Patterns of Identifications
and Habituation. Ear and Hearing. 4: 232-236 (1983).

Stein, L., Jabaley, T., Spitz, R., Stoakley, D., McGee, T. The Hearing Impaired Infant:
Patterns of Identification and Rehabilitation Revisited. Ear and Hearing. 11: 210-215
(1990)

Stein, L., Ozdamar, O., Kraus, N., Paton, J. F ollow-Up of Infants Screened by Auditory
Brainstem Response in the Neonatal Intensive Care Unit. The Journal of Pediatrics. 103
(3): 447-453 (1983).

Steiner, E.S., Sanders, E.M., Phillips, E.C.K., Maddock, C.R. Very Low Birth Weight
Children at School Age: Comparision of Neonatal Management Methods. British
Medical Journal. 281: 1237-1240 (1980).

Swigonski, N.S., Bull, M.J., Lemons, J.A. Hearing Screening of High Risk Newborns.
Ear and Hearing. 8 (1): 26-30 (1987).

Talbot, C.B. A Longitudinal Study Comparing Responses of Hearing Impaired Infants to

Pure Tones Using Visual Reinforcement and Play Audiometry. Ear and Hearing. 8
(3): 175-179 (1987).

112

Thompson, M., Thompson, G. Early Identification of Hearing Loss: Listen to Parents.
Clinical Pediatrics. 30: 77-80 (1991).

Thompson, M., Thompson, G., Vethivelu, S. A Comparison of Audiometric Test
Methods for 2-Year-Old Children. Journal of Speech and Hearing Disorders. 54 (2):
174-179 (1989).

Toynbee, J. The Diseases of the Ear. Blanchard and Lea, Philadelphia (1860).

Veen, S., Sassen M.L., Schreuder, A.M., Ens-Dokkum, M.H., Verloove-Vanhorick, S.P.,
Brand, R., Grote J .J , Ruys, J.H. Hearing Loss in Very Preterm and Very Low
Birthweight Infants at the Age of 5 Years in a Nationwide Cohort. International Journal
of Pediatric Otorhinolaryngology. 26: 11-28 (1993).

Volpe, JJ. Intraventricular Hemorrhage and Brain Injury in the Premature Infant.
Neuropathology and Pathogenesis. Clinical Perinatology. 16: 361-386 (1989).

Voss, O. Geburtstrauma und Gehoororgan. Hals, Nase, Und Ohrenheilk. 6: 182-219
(1923)

Watkin, P.M. Parental Suspicion and Identification of Hearing Impairment. Archives of
Disease in Childhood. 65 (8): 846-50 (1990).

Watkin, P.M., Baldwin, M., McEnery, G. Neonatal at Risk Screening and the
Identification of Deafness. Archives of Disease in Childhood. 66: 1130-1135 (1991)

Winkel, S., Bonding, P., Keldegard, L., Roosen, J. Possible Effects of Kanamycin and
Intubation in Newborn Children with Low Birth Weight. Acta Peadiatric Scandanavia.
67: 709-715 (1978).

Wright, F .H., Blough, R. R., Chamberlin, A., Ernest, T., Halstead, W.C., Meier, R,
Moore, R.Y., Naunton, R.F., Newell, F .W. A Controlled F ollow-Up Study of Small
Prematures Born from 1952 through 1956. American Journal of Diseases in Childhood.
124(4): 506-521 (1972).

Yeargin-Allsopp, M., Murphy, C.C., Oakley, G.P., Sikes, R.K. A Multiple-Source
Method for Studying the Prevalence of Developmental Disabilities in Children: The
Metropolitan Atlanta Developmental Disabilities in Children: The Metropolitan Atalanta
Developmental Disabilities Study. Pediatrics. 89: 624-630 (1992).

Zaman, S.S., Khan, N.Z., Islam, 8., Banu, S., Dixit, S., Shrout, P., Durkim-M. Validity

of the “Ten Questions” for Screening Serious Childhood Disability: Results from Urban
Bangladesh. International Journal of Epidemiology. 19(3): 613-20 (1990).

113

A

I" .0

IV. LIBRRR

IES
11111111 11111111
E3

4 8965

MICHIGAN srar: UN
1111111111111111111111111111111
31293020