Citation
Apnea and bradycardia elicited by facial airstream stimulation in healthy infants in the first year of life

Material Information

Title:
Apnea and bradycardia elicited by facial airstream stimulation in healthy infants in the first year of life implications for detection of infants at risk for sudden infant death syndrome
Creator:
Hurwitz, Barry Elliot, 1956- ( Dissertant )
Berg, W. Keith ( Thesis advisor )
Meyer, Merle E. ( Reviewer )
Lang, Peter ( Reviewer )
Posner, Philip ( Reviewer )
Rowland, Neil ( Reviewer )
Place of Publication:
Gainesville, Fla.
Publisher:
University of Florida
Publication Date:
Copyright Date:
1984
Language:
English
Physical Description:
x, 145 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Acceleration ( jstor )
Age groups ( jstor )
Apnea ( jstor )
Breathing ( jstor )
Infants ( jstor )
Neonates ( jstor )
Pediatrics ( jstor )
Reflexes ( jstor )
Respiratory mechanics ( jstor )
Sudden infant death syndrome ( jstor )
Bradycardia ( lcsh )
Dissertations, Academic -- Psychology -- UF
Psychology thesis Ph. D
Sleep apnea syndromes ( lcsh )
Sudden infant death syndrome ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
Cardiorespiratory responses to airstream stimulation (10-seconds, 1.06 psi) were observed in 52 newborn (NB), 2-4 and 8-12 month-old healthy infants. Two areas of stimulation were compared: the facial region, which included the forehead, eyelids and nose, and the abdominal region rostral to the umbilicus, which served as a control stimulation site. The infants were tested while asleep, during which EKG, heart rate and respiration were recorded. The results revealed that, in both NB and 2-4 month infants, facial cooling with an airpuff elicited apnea with greater frequency and duration than did the control airpuff. Apnea occurred on about 27,2% of facial trials compared to '-'.8% of control trials. The duration of apnea was twice as long in the 2-4 month infants as in the NBs. By 8-12 months the facial airpuff still elicited apnea more reliably than did the abdominal airpuff but with reduced frequency of 8.9% and 1.9% respectively. The duration of apnea, however, was maintained at the longer duration seen with the 2-4 month infants. Bradycardia on trials when apnea was elicited differed in magnitude between age groups such that the 2-4 and 8-12 month infants decelerated to about -15 bpm; whereas no bradycardic response to apnea was displayed by the NBs. It has been demonstrated that in healthy sleeping NB and 2-4 month infants an airstream directed to the face can elicit apnea. However, by 8-12 months facial cooling does not have as great an effect on inhibition of respiration. Although apnea can be elicited in the NBs by facial airpuff, this apnea was brief and resulted in no change in cardiac response. Therefore, at 2-4 months, which is the age of greatest vulnerability for Sudden Infant Death Syndrome (SIDS) , the airpuff was likely to produce more frequent, prolonged apnea with concomitant bradycardia. It has been suggested that SIDS infants produce prolonged apnea because they have a higher threshold for hypoxia. Therefore, by eliciting apnea with facial cooling, it may De possible to identify a predisposition to SIDS by analysis of the magnitude of cardiorespiratory change in. infants with a history of prolonged apnea.
Thesis:
Thesis (Ph. D.)--University of Florida, 1984.
Bibliography:
Bibliography: leaves 125-142.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Barry Elliot Hurwitz.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
000479215 ( AlephBibNum )
11795296 ( OCLC )
ACP5939 ( NOTIS )

Downloads

This item has the following downloads:


Full Text














APNEA AND BRADYCARDIA ELICITED BY FACIAL AIRSTREAM
STIMULATION IN HEALTHY INFANTS IN THE FIRST YEAR OF LIFE:
IMPLICATIONS FOR DETECTION OF INFANTS AT RISK
FOR SUDDEN INFANT DEATH SYNDROME




By

BARRY ELLIOT HURWITZ


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA


1984
























Dedicated with love to my wife, Debbie

and my son, Joseph

















ACKNOWLEDGEMENTS

I thank Dr. W. Keith Berg for his guidance, astute

judgement and his infectious positive approach to life.

His active support of our past and present collaborative

efforts has meant a great deal to me. I look forward to a

long and fruitful future association. As well, I thank you

for so willingly being there whenever I needed you and for

patiently allowing me to interrupt your stories of the joys

of fatherhood in order to tell you mine.

I would also like to express my appreciation to my

doctoral committee, Drs. Merle E. Meyer, Peter Lang, Phil

Posner and Neil Rowland, for their assistance throughout my

graduate training. Also, I thank Drs. Don Stehouwer and Paul

Davenport for their helpful comments on this project.

In addition, I thank all those infants who fell asleep

for me and most notably I thank all those little kids who

held their breath for me and even those that didn't. Also,

I am grateful to the parents of the little kids for

generously contributing their time to this effort.

Special thanks are extended to my assistants, Alonso,

Judy, Kristen, Karen, Julio and Yolanda, for their help in

data collection at various times throughout the past two












years. I thank my lab mates Terry Blumenthal, Margaret

Davies and Allen Klempert for their enjoyable association

and patient tolerance.

I thank my parents and the rest of my family for their

tenacious support and everlasting expression of faith and

encouragement. I stand tall because of them. Finally, to

my wife and true friend Debbie, thank you for being there

when I wasn't and also when I was, and more importantly

thank you for my son, Joseph.


















TABLE OF CONTENTS


ACKNOWLEDGEMENTS. . . . . . . . . . . iii

LIST OF TABLES. . . . . . . . . ... ... .vii

LIST OF FIGURES . . . . . . . ... . viii

ABSTRACT. . . . . . . . . ... . . .ix

CHAPTER

1 INTRODUCTION. . . . . . . . . . 1

Epidemiology. . . . . . . . . . 3
Pathology . . . . . . . . ... . 6
Cardiorespiratory Physiology. . . . . . 9
Control of Cardiac Function . . . . .. .10
Control of Respiratory Function . . . .. .13
Respiratory Patterns. . . . . . .. .13
Frequency of apnea. . . . . . .. .15
Control of ventilation. . . . . ... 20
Sleep and Breathing . . . . . . .. 24
Cardiorespiratory Responses To Airstream
Stimulation . . . . . . . . 30

2 METHODS . . . . . . . . ... . 38
Subjects. . . . . . . . .. . . 38
Apparatus . . . . . . . . . . 43
Experimental Manipulations and Design . . .46
Procedure . . . . . . . . . . 47
Data Quantification . . . . . . .. 49
Notes . . . . .. . . . . . . 52

3 RESULTS . . . . . . . . . . 53
Sleep State Incidence and Pattern . . . .. .53
Large Movement Incidence. . . . . . ... 55
Respiration Response to Airstream Stimulation 59
Percent of Elicited Apnea . . . . .. 59
Effect of airstream stimulation site. ... .59
Effect of sleep state . . . . . .. .62
Effect of preceding large movement. ..... 62
Latency to Apnea Onset. . . . . . .. .66
Effect of airstream stimulation site. ... .66
Effect of preceding large movement. ... .66













Duration of Apnea . . . . . . .. 68
Effect of airstream stimulation site. .. .. 68
Effect of preceding large movement. .. .... 71
Summary of the Respiration Response to
Airstream Stimulation . . . . . .. 73
HR Response Topography . . . . . . 74
Prestimulus Levels . . . . . . 75
Post-stimulus Changes . . . . . .. 75
HR response to the airstream stimulus:
face vs. abdomen. . . . . . ... 78
HR response to the facial airstream
stimulus: apnea vs. no apnea. . . .. 81
HR Response to the Facial Airstream Stimulus
from Apnea Onset . . . . . . .. 86
Notes . . . . . . . . . . . 92

4 DISCUSSION . . . . . . . . .. 94
Implication for Dive Reflex as the Mechanism
for Inducing Apnea and Bradycardia. . . 97
Topography of HR Response in the Absence
of Apnea. . . . . . . . . . 104
Detection of Infants At-Risk for SIDS . .. 111
Note. . . . . . . . . .. . 115

APPENDICES

A INFANT BACKGROUND QUESTIONNAIRE . . . .. 116

B INFANT STATE RATING SHEET . . . . . 118

C INFORMED CONSENT . . . . . . . .. 119

D ANALYSIS OF THE PERCENT OF TRIALS WITH QS, TS
AND AS STATES . . . . . . . . .122

E ANALYSIS OF THE PERCENT OF TRIALS WITH
LARGE MOVEMENT. . . . . . . . . 125

F ANALYSIS OF THE HR RESPONSE TO THE AIRSTREAM
STIMULUS: FACE VS. ABDOMEN. . . . . ... 127

REFERENCES . . . . . . . . . . .. 130

BIOGRAPHICAL SKETCH . . . . . . . . . 143













LIST OF TABLES


TABLES
PAGE
1 Description fo the Newborn Subject Pool . . .. 40

2 Description of the 2-4 Month-Old Subject Pool . . 41

3 Description of the 8-12 Month-Old Subject Pool. . 42

4 Percent of Trials in Quiet Sleep, Acitive Sleep and
Transitional Sleep States . . . . . 54

5 Percent of Trials with Large Movement Elicited by
Face and Abdomen Airstream Stimulation. . . .. 58

6 Percent of Trials with Apnea Elicited by Airstream
Stimulation of the Face and Abdomen When
Sleep State Is Ignored and When Infants Were
in Qjiet Sleep. . . . . . . . .. . 60

7 Percent of Trials with Apnea Elicited by Airstream
Stimulation of the Face and Abdomen When
Preceded by Large Movement and When Large
Movement Was Absent . . . . . . ... 63

8 Latency to Apnea Onset in Seconds as Elicited by
Airstream Stimulation of the Face and Abdomen . 67

9 Latency to Apnea in Seconds Elicited by Airstream
Stimulation of the Face When There Was Preceding
Large Movement and When Large Movement Was Absent 69

10 Duration of Apnea in Seconds Elicited by Airstream
Stimulation of the Face and Abdomen . . . .. 70

11 Duration of Apnea in Seconds Elicited by Airstream
Stimulation of the Face When There Was Preceding
Large Movement and When Large Movement Was Absent 72

12 Mean Prestimulus HR Levels in bpm in the Presence
and Absence of Apnea and Large Movement for Facial
and Abdominal Airstream Stimulation . . . .. 76












LIST OF FIGURES


FIGURES
PAGE
1 Mean Sleep State Rating for the First and Last Ten
Trials. . . . . . . . . ... . . . 56

2 Mean HR Response to Facial Airstream Stimulation When
No Apnea or Large Movement Were Elicited. . . .. 79

3 Mean HR Response to Abdominal Airstream Stimulation
When No Apnea or Large Movement Were Elicited ... . 80

4 Mean HR Response to Facial Airstream Stimulation When
Apnea Was Elicited Without Preceding Large Movement . 82

5 Mean HR Response to Facial Airstream Stimulation When
Apnea Was Elicited Without Preceding Large Movement
Adjusted from Apnea Onset . . . . . . . 88

6 Mean HR Response to Facial Airstream Stimulation When
Apnea Occurred Following Large Movement Adjusted from
Apnea Onset . . . . . . . .. . . . 89


viii
















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


APNEA AND BRADYCARDIA ELICITED BY FACIAL AIRSTREAM
STIMULATION IN HEALTHY INFANTS IN THE FIRST YEAR OF LIFE:
IMPLICATIONS FOR DETECTION OF INFANTS AT RISK
FOR SUDDEN INFANT DEATH SYNDROME

By

BARRY ELLIOT HURWITZ

August, 1984


Chairman: W. Keith Berg
Major Department: Psychology



Cardiorespiratory responses to airstream stimulation

(10-seconds, 1.06 psi) were observed in 52 newborn (NB),

2-4 and 8-12 month-old healthy infants. Two areas of

stimulation were compared: the facial region, which

included the forehead, eyelids and nose, and the abdominal

region rostral to the umbilicus, which served as a control

stimulation site. The infants were tested while asleep,

during which EKG, heart rate and respiration were recorded.

The results revealed that in both NB and 2-4 month

infants, facial cooling with an airpuff elicited apnea with

greater frequency and duration than did the control

airpuff. Apnea occurred on about 27.2% of facial trials

compared to 7.8% ot control trials. The duration of apnea












was twice as long in the 2-4 month infants as in the NBs.

By 8-12 months the facial airpuff still elicited apnea

more reliably than did the abdominal airpuff but with

reduced frequency of 8.9% and 1.9% respectively. The

duration of apnea, however, was maintained at the longer

duration seen with the 2-4 month infants. Bradycardia on

trials when apnea was elicited differed in magnitude

between age groups such that the 2-4 and 8-12 month infants

decelerated to about -15 bpm; whereas no bradycardic

response to apnea was displayed by the NBs.

It has been demonstrated that in healthy sleeping NB

and 2-4 month infants an airstream directed to the face can

elicit apnea. However, by 8-12 months facial cooling does

not have as great an effect on inhibition of respiration.

Although apnea can be elicited in the NBs by facial

airpuff, this apnea was brief and resulted in no change in

cardiac response. Therefore, at 2-4 months, which is the

age of greatest vulnerability for Sudden Infant Death

Syndrome (SIDS), the airpuff was likely to produce more

frequent, prolonged apnea with concomitant bradycardia.

It has been suggested that SIDS infants produce prolonged

apnea because they have a higher threshold for hypoxia.

Therefore, by eliciting apnea with facial cooling, it

may be possible to identify a predisposition to SIDS by

analysis of the magnitude of cardiorespiratory change in

infants with a history of prcionged apnea.

x
















CHAPTER 1
INTRODUCTION


"and this woman's child died in the night,
because she overlaid it"


This quote, which is from an account in the Book of

Kings 3:16-28 circa tenth century B.C., in which the word

"overlaid" appeared to imply suffocation of an infant

during the night, has been cited as perhaps the first

recorded instance of Sudden Infant Death Syndrome (SIDS)

(Peterson, 1980). Despite it's probable antiquity, SIDS

was largely ignored by the scientific community until, as

infant mortality rates declined, it became apparent that

these deaths comprised a substantial proportion of all

infant deaths within the first year of life. Currently the

incidence of SIDS throughout the world varies from 0.6 to

3.0 deaths per 1000 live births, accounting for about

7-9,000 deaths each year in the United States alone

(Peterson, 1980; Valdes-Dapena, 1980). About one-third of

all deaths in infants between the ages of one week and one

year are diagnosed SIDS and it is the most frequent cause

of death in this age group (Moore, 1981).

Prior to 1963, deaths that occurred suddenly,

unexpectedly, and inexplicably with no recognizable lethal

pathology at autopsy were generally certified to one or












another cause-of-death category in keeping with the

attending coroner's personal inclination (Peterson, 1980).

This practice circumvented the embarrassment of a factual

"don't know" and at the same time complied with legal

expectations. Following the 1963 conference on causes of

SIDS, awareness and acceptance of sudden unexpected death

as a legitimate diagnostic entity increased. Finally the

designation, Sudden Infant Death Syndrome, emerged from the

Second International SIDS Conference in 1969, wherein an

occurrence of SIDS was defined as the sudden death of any

infant which was unexpected by history, and in which a

thorough post-mortem examination fails to demonstrate an

adequate cause for death (Bergman, Beckwith & Ray, 1970).

The following text will review certain epidemiological,

pathological and physiological findings related to SIDS as

background for the proposed study. For a comprehensive

review of SIDS, the reader is referred to these recently

published reviews (Brooks, 1982; Kelly & Shannon, 1982;

Shannon & Kelly, 1984; Valdes-Dapena, 1980). In addition,

as a basis for comparison, developmental findings in normal

healthy infants will be included where relevant. Based on

this information, an experimental methodology, which could

be used as a screening test for detection of infants at

risk for SIDS, will be proposed to examine the development












of the normal infant's cardiorespiratory response in the

first year of life.

Epidemiology

Epidemiologic evidence has been gathered by the

scientific community in an effort to identify a population

of infants who might be at increased risk for succumbing to

SIDS. In the absence of pathological evidence which could

directly account for SIDS, the epidemiological findings

also have been instrumental in generating hypotheses about

the etiology of SIDS. It is important that any hypothesis

about the cause of SIDS account for these epidemiological

factors. It has been found that SIDS occurs while the

infant is asleep in the early morning hours, between

midnight and six a.m., with no cry or stridor (Peterson,

1980). The incidence of infant mortality due to SIDS peaks

between two to four months of age. Less than ten percent

of all SIDS deaths occur when infants are younger than two

weeks and older than six months of age (Brooks, 1982). It

should be noted, however, that the peak age for infant

mortality of known causes also occurs at two to four months

of postnatal age (Kelly & Shannon, 1982).

About 60% of SIDS victims are male and incidence for

nonwhites is twice that for whites; SIDS is more

common in low birth weight infants (i.e., 11/1000 live

births subsequent siblings of SIDS victims (i.e., 21/1000












live births) and in the surviving twin of SIDS victims

(i.e., 42/1000 live births) (Peterson, 1980). These data

suggest that either genetic factors have promoted the

demise of these infants or an environmental influence has

affected their intrauterine or extrauterine development. It

should be noted, however, that no difference was found

between the concordance of SIDS episodes in monozygotic and

dizygotic twin pairs; in addition, the incidence of SIDS in

full first cousins of SIDS victims did not differ from the

incidence in the general population (Peterson, 1980).

Thus, overall the data do not tend to support a genetic

component to SIDS.

There are several epidemiological perinatal-factors

which indicate an environmental influence on SIDS. SIDS

infants are born with higher incidence to young, unmarried

mothers of lower socioeconomic levels who have had poor

prenatal care, have short interpregnancy intervals, have

previous fetal loss and who are smokers or have narcotic

dependency (Valdes-Dapena, 1980). Other environmental

factors which may play a role in SIDS are altitude of

residence and air temperature. Specifically, SIDS incidence

increases as the altitude of residence increases and

the postnatal age of the infant at death decreased with

altitude from a mean of 17 weeks at 300 meters to nine

weeks at 1200 meters (Getts & Hill, 1982). SIDS exhibits a












seasonal variation being more apt to occur during the

colder winter months. This has been shown in countries

north of the equator peaking between November and March and

countries south of the equator peaking between May and

September (Moore, 1981; Peterson, 1980; Taylor, 1982).

Moreover, a spectral analysis of the SIDS incidence in

Los Angeles County from 1974-79 confirmed this seasonal

peak of SIDS in addition to revealing the coincidence of

decreased air temperature with increases in frequency of

SIDS mortality within this period (Hoppenbrouwers &

Hodgman, 1982).

On the basis of epidemiologic studies, several groups

of infants are believed to be at enhanced risk for SIDS and

have been most extensively tested by physiologic studies.

They are 1) preterm (i.e., low birth weight) infants with

histories of apnea (i.e., cessation of breathing) and

bradycardia; 2) infants who have had a near-miss or aborted

episode of SIDS, including apnea, cyanosis or pallor and

unresponsiveness that responds to intervention of

stimulation or resuscitation; and 3) siblings of SIDS

infants. There has been some controversy regarding the use

of terms like "near-miss" or "aborted" SIDS because they

imply with some certainty that the infant almost died when

in fact there is no way of knowing whether or not the

infant would have spontaneously resumed breathing without












assistance (Brooks, 1982). Therefore, the term "Apnea of

Infancy" or AI, as suggested by Brooks (1982), will be

adopted and used throughout to describe "an unexplained

cessation of breathing for 20 seconds or longer, or a

shorter respiratory pause associated with bradycardia,

cyanosis or pallor." Clearly this term describes a

clinical syndrome of respiratory abnormality similar to the

sleep apnea of adults or the apnea of prematurity.

Pathology

Report of excessive frequency of prolonged apneic

episodes during sleep, in infants who subsequent to their

examination died of SIDS, promoted study of the hypothesis

that SIDS victims were experiencing recurrent periods of

apnea-associated hypoxia prior to their death

(Steinschneider, 1972). Among the plethora of hypotheses

about the cause of SIDS, most investigatory work in the

past decade has centered on the apnea hypothesis (Valdes-

Dapena, 1980). A number of post-mortem studies were

published shortly after Steinschneider's report which

supported the hypothesis that the SIDS is the terminal

event in infants who were suffering from a prolonged

hypoxic process, which possibly resulted from anomalies in

ventilatory control (e.g., Naeye, 1973; 1976). Based on

this hypothesis, one would anticipate anomalies in the anatomic

sites of ventilatory control (brain stem and carotid body),












as well as the organs which are usually affected by chronic

hypoxia (the brain, pulmonary vasculature, liver, adrenal

glands and periadrenal brown fat).

The main pathological findings have demonstrated that

SIDS victims differed from healthy infants who were victims

of accidental injury in that SIDS infants had hypertrophic

right ventricles in direct proportion to hypertrophy and

hyperplasia of pulmonary arteriolar smooth muscle,

increased volume of chromaffin cells in the adrenal

medulla, retention of periadrenal brown fat and presence of

hepatic extramedullary hematopoiesis (Naeye, 1973; Naeye,

1976; Naeye, Whalen, Ryser & Fisher, 1976). Moreover, upon

histologic analysis of SIDS cases, Naeye (1976) found

increases in brain stem astroglial density, particularly in

the midline area of the watershed zone of brain stem

circulation. Other recent findings of increased glomic

volume of the carotid body in those SIDS victims with

evidence of hypertrophy of the pulmonary vasculature have

been reported (Naeye, Fisher, Ryser & Whalen, 1976) and

reduction of myelinated fibers in the cervical vagus nerve

in some SIDS victims (Sachis, Armstrong, Becker & Bryan, 1981).

The finding of increased muscularity and hyperplasia

of the pulmonary arteries has been independently described

in SIDS victims (Shannon, 1980; Valdes-Dapena, Gillane,

Cassady, Catherman & Ross, 1980; Williams, Vawter &












Reid, 1979). These anatomic changes also have been induced

in animals exposed to repeated episodes of hypoxia

(Rabinovitch, Gamble, Nadas, Meittinen & Reid, 1979) and

are typical of children who die after prolonged exposure to

the hypoxic conditions of high altitude (Naeye, 1973). The

increased hypertrophy of the right ventricle may be a result

of the increased pulmonary arteriolar resistance, which forces

the heart to pump against a narrower opening. Brown fat

and hepatic hematopoiesis normally disappears within the

first year of life but its disappearance is delayed in SIDS

infants, which is a common finding also noted in autopsy of

hypoxic-children residing in high altitude (Naeye et al.,

1976). The astroglial proliferation in the brain stem has

affected areas such as the nucleus and tractus of

solitarius, nucleus ambiguous, raphe nucleus and dorsal

vagal nucleus, which are critical in the autonomic control

of the cardiovascular and respiratory systems. Similar and

more extensive CNS damage has been observed in post-mortem

examination of preterm infants with chronic hypoxia due to

respiratory distress syndrome (Brand & Bignami, 1969). In

addition, perinatal asphyxia for 10-15 minutes in monkeys

resulted in a pattern of brain damage which always

involved midline brain stem structures (Myers, 1972).

Takashima, Armstrong, Becker and Bryan (1978) have

postulated that periods of cerebral ischemia may accompany

the prolonged periods of apnea and hypoxemia in SIDS












infants, which in turn may result in the observed medullary

lesions and consequent astroglial proliferation. For this

to occur the medullary lesions would have to result from

discrete ischemia to brain stem otherwise more widespread

damage would be observed.

In sum, a case has been made for the etiology of SIDS,

whereby an infant acutely or chronically exposed to

hypoxic conditions may develop abnormal cardiorespiratory

control due to lesions in the critical brain stem areas.

The resultant prolonged periods of apnea or hypoventilation

could lead to further hypoxia and establish a positive

feedback loop of increasing pathologic tissue damage or at

least impaired function in regions of cardiorespiratory

control. Eventually, perhaps when challenged, the

cardiorespiratory regulatory mechanisms can no longer

compensate for the hypoxemic conditions and respiratory

failure results.

Cardiorespiratory Physiology

Since post-mortem studies can only provide clues to

the etiology of SIDS, physiologic studies have been

performed on living infants. What pre-mortem data on SIDS

victims are available were derived from studies of infants

who have died of SIDS subsequent to physiologic recording.

Other than this chance occurrence, there obviously can be

no systematic study of the physiologic responses of these

infants. However, investigators have studied infants












thought to be at risk for SIDS: infants with AI and

siblings of SIDS victims. They have typically been

contrasted with age-matched, healthy infants. It should

be noted that most studies of SIDS examine the target

physiological response longitudinally over the infant's

development. This is entirely appropriate because the risk

for SIDS seems to be a developmental phenomenon in which

the infant is at increasing risk from birth to 2-4 months

and then with maturation risk declines.

Control of Cardiac Function

Investigations of heart rate (HR) and HR variability

patterns have been performed to determine whether

manifestations of malfunctioning autonomic control

mechanisms could be detected in at-risk infants. During

maternal labor, fetal HR recordings of infants who

subsequently were SIDS victims revealed no differences from

control infants (Hoppenbrouwers, Zanini & Hodgman, 1979).

Siblings of SIDS infants, however, displayed greater fetal

HR variability and more frequent bradycardia than controls

but no difference in overall fetal HR levels when monitored

in-utero while their mothers slept (Hoppenbrouwers,

Hodgman, Harper & Sterman, 1981). The authors suggested

that the increased variability and bradycardic changes in

the siblings may reflect an increased level of intrauterine

hypoxic challenge in the siblings, whereas the effects of

hypoxia may have been concealed by the birthing process in












the SIDS infants. The results in the siblings, however,

should be viewed with caution since they are confounded by

maternal sleep disturbance, which was higher in mothers of

siblings of SIDS than control mothers.

The course of maturation of HR and HR variability for

healthy infants has been documented over the first six

months of life (Harper, Leake, Hodgman & Hoppenbrouwers,

1982). For HR levels, there is an increase in HR from

birth to one month after which rate declines steadily. In

contrast, HR was faster in both siblings of SIDS and in AI

infants during sleep, displaying a sharper rise and slower

decline in the first four months of life (Harper, Leake,

Hoppenbrouwers, Sterman, McGinty & Hodgman, 1978; Harper et

al., 1982; Leistner, Haddad, Epstein, Lai, Epstein, Eng &

Mellins, 1980). These differences were present only in

quiet sleep (QS) state (Harper et al., 1982); whereas

others observed these differences in both QS and active

sleep (AS) state (Leistner et al., 1980). For HR

variability in normals, sleep state had a more marked effect.

Awake variability was greater than asleep; during sleep, QS

was less variable than AS (Harper et al., 1982). Moreover,

during QS, HR variability declined rapidly up to one

month and then increased rapidly between one and three

months after which it became stable. In AS a steady

decline in variability was observed over the first six












months. In contrast, awake HR variability increased up to

two months and then rapidly declined between two and four

months after which it stabilized.

No difference has been observed between normals and

siblings of SIDS in the maturation of HR variability during

sleep up to six months of age, although risk infants

displayed reduced variability in the first two months when

awake (Harper et al., 1982). On the other hand, others

found less HR variability in AI infants than in healthy

controls during AS and QS between two and four months

(Leistner et al., 1980). Harper and colleagues (1978) have

attributed the increased HR levels in the at-risk infants

to a reduction in vagal tone. Leistner and colleagues

(1980) argue, instead, that the elevated HR levels are due

to enhanced sympathoadrenal drive. This is based on the

histologic evidence in SIDS victims of hypertrophy of the

chromaffin cell layer of the adrenal medulla, which may

indicate increased norepinephrine synthesis. Further

support of the latter hypothesis was derived from evidence

of a shorter Q-T electrocardiographic interval in AI

infants than controls over the first four months of life

regardless of state, which suggested sympathetic or adrenal

mediated enhanced-conduction (Haddad, Epstein, Epstein,

Mazza, Mellins & Krongrad, 1979). Although sympathoadrenal

mediation may be a possibility for the shortened Q-T










interval, ventricular repolarization may also be affected

by decreased vagal tone (Prystowsky, Jackman, Rinkenberger,

Heger & Zipes, 1981). Regardless of the autonomic

mediation, it still remains possible that the elevated HR

levels in the at-risk infants represent a compensatory

response to hypoxia; the increased cardiac output could act

to compensate for the oxygen deficiency.

Control of Respiratory Function

Respiratory patterns. The respiratory rates (RR) of

siblings of SIDS have been measured throughout the night

across the first six months of age (Hoppenbrouwers, Jenson,

Hodgman, Harper & Sterman, 1980). In both normals and

siblings, RR declined with increasing age. However,

regardless of sleep state, in each age group the RR rate of

the SIDS siblings was faster than the control infants, which

was interpreted as a compensatory response to hypoxia. Both

groups of infants displayed a developmental trend in RR

pattern throughout the night. In both AS and transitional

sleep (TS), normals and siblings displayed a U-shaped RR

pattern throughout the night at all ages. In QS this

U-shaped RR pattern did not appear in normals until three

months. Siblings, however, displayed this U-shaped pattern

prematurely at one month during QS. The authors posit that

this was due to a transient acceleration of maturation of

the nightly RR pattern in the siblings. Respiratory












variability decreased with increasing age in all infants

although siblings of SIDS displayed enhanced variability

in QS at three months.

In a parallel study of AI infants over the first six

months, the data do not entirely replicate these findings

(Hodgman, Hoppenbrouwers, Geidel, Hadeed, Sterman, Harper &

McGinty, 1982). There was no difference in RR or

respiratory variability between AI infants and controls,

although the decrease in RR with increasing age was

replicated. The only other longitudinal study of AI

infants observed faster RR and reduced tidal volume in the

AI infants at only one and two months. At three months no

differences were observed and at four months AI infants' RR

was less than controls (Haddad, Leistner, Lai & Mellins,

1981).

In studies on preterm infants who were presumably

hypoxemic, faster HR and RR than term infants has been

observed throughout the first six postnatal months, with

maximal differences between 2.5 and 3.5 months, the peak

period of SIDS vulnerability (Katona, Frasz & Egbert,

1980). It was concluded that since preterm HR and RR

maturational patterns were not delayed by the mean

difference in post-conceptional age, that both conceptional

age and postnatal experience were factors in the

development. Therefore, relative to healthy controls all

three SIDS risk groups displayed typical patterns of faster












RR over the first six months of life, which could be

related to maturational differences and/or compensatory

responses to hypoxemia.

Frequency of apnea. There are four types of apnea:

central, periodic, obstructive and mixed. Central apnea is

defined as a cessation of respiration with no chest or air

movement; periodic apnea is also of central origin except

that the apnea appears in constellations of short pauses

usually greater than three seconds in duration and repeated

three or more times with interapnea intervals of 20 seconds

or less; obstructive apnea is a disruption in ventilation

consisting of a lack of air flow but continued respiratory

movements; mixed apnea is initially a central apnea but

towards the end of the apneic episode some chest movements

appear with no air flow indicating some obstruction (Kelly

& Shannon, 1982).

Apnea with duration less than nine seconds is a common

finding in healthy infants over the first six months of

life (Hoppenbrouwers, Hodgman, Arakawa, Harper & Sterman,

1980). In this study short apnea (2-5 seconds) was more

abundant than longer apnea (6-9 seconds) and the incidence

of short apnea increased with age while longer apnea

incidence decreased with age. It was speculated that this

increase with age could probably be accounted for by the

same mechanism which was responsible for the concomitant












increase in interbreath interval. The frequency of apnea

longer than ten seconds was common only in the first week

when incidence of all apnea types was greatest. Otherwise,

prolonged apneic episodes were infrequent. Additionally,

sleep state affected apnea frequency, with more apnea

observed in AS than QS. When apnea does occur it has been

found to be of greater duration in QS than AS (Mellins &

Haddad, 1983).

There are conflicting data on whether the incidence of

apnea is enhanced in at-risk infants. In the fourteen

years since Steinschneider (1972) reported his findings of

prolonged and excessive apnea in infants who subsequently

died of SIDS, a number of studies have also reported

increased incidence of apnea in infants at risk for SIDS.

In another study, Steinschneider (1977) observed longer and

more frequent central apnea as well as increased frequency

of periodic apnea in infants with AI. However, apnea

shorter than 20 seconds was not examined. An increase in

frequency and duration of periodic apnea in both siblings of

SIDS and AI infants has also been reported (Kelly &

Shannon, 1979; Kelly, Walker, Cahen & Shannon, 1980).

In contrast, Guilleminault and colleagues have observed

no significant differences between AI infants and controls in

incidence of periodic apnea or in central apnea greater

than ten seconds (Guilleminault, Ariagno, Korobkin, Nagel,












Baldwin, Coons & Owen, 1979; Guilleminault et al., 1981).

However, they did report an increase in short (i.e., three

to ten seconds) mixed and obstructive apnea frequency in AI

infants from three weeks to four months of age.

Unfortunately, they failed to report or analyze data on

central apneas less than ten seconds. They report the

means of their 1979 study for central apneas from three to

ten seconds in a more recent publication but failed again

to analyse them (Guilleminault, Souquet, Ariagno, Korobkin

& Simmons, 1984). Inspection of these means suggested that

these short central apneas were more frequent than the

combined mixed and obstructive frequencies of similar

duration. In addition, it appeared that the AI infants

had a greater incidence of short central apneas at three

and six weeks of age but not at twelve weeks.

Additional contradictions were evident in other

research. Greater incidence of short apneas in subsequent

siblings of SIDS than controls was observed only when in AS

at two weeks of age (Kelly, Twanmoh & Shannon, 1982). No

difference in these infants was observed at eight and

fourteen weeks of age. Some researchers report no

difference in the frequency or duration of respiratory

pauses in AI and control infants (Mellins & Haddad, 1983).

Conversely, other researchers have found a reduced

incidence of central apnea in siblings of SIDS than in












controls in AS and QS (Hoppenbrouwers, Hodgman, McGinty,

Harper & Sterman, 1980). In this study, the incidence of

mixed and obstructive apnea and central apnea greater than

ten seconds was too low to analyse. However, it was found

that from one week to six months of age, differences

between infant groups were greatest for short apneas of two

to five seconds in QS; whereas for longer apneas of six to

nine seconds differences were greatest in AS and TS. In

another study, using AI and normal infants from one week to

six months of age, this same laboratory reported no

differences in periodic, mixed, obstructive or central

apnea frequency (Hodgman et al., 1982). However, although

the trends were nonsignificant, AI infants did display

lower frequency of short and longer central apnea in this

study.

In summary, both the Shannon and Guilleminault

research groups report excessive incidence of apnea in

infants at risk for SIDS, although they don't agree on

which type of apnea is significant. Shannon claims that

its periodic apnea that is excessive and Guilleminault

emphasizes mixed and obstructive apneas. Alternatively,

the Haddad group reports no difference in apnea frequency

and the Hoppenbrouwers group found frequency reductions in

central apneic episodes less than ten seconds in duration.

Guilleminault and colleagues have recently emphasized that












because of differences in study design, definition of AI,

subject age, time and duration of study, recorded

measurements, analysis of respiratory events, and setting

of session, it is difficult to compare between laboratory

studies (Guilleminault, Ariagno, Korobkin, Coons, Owen-

Boeddiker & Baldwin, 1981). The lack of standardization

may therefore be responsible for the differences or lack

thereof in apnea frequency between studies. It seems clear

from the review above that there has been little

interlaboratory replication and although there may be

differences one way or another in the frequency of a

certain type of apnea, these are restricted to the

particular circumstances of measurement. Hence, the role

of apnea in SIDS remains equivocal.

The finding of an increase or decrease in apnea

incidence has been posited to relate to defective

respiratory control interacting with a myriad of potential

factors such as sleep state and CNS maturation (Shannon &

Kelly, 1982). The lack of consistent replicable finding

suggests that the apnea is of varied origin and probably

constitutes only one of the many triggering stimuli in SIDS

infants. Based upon the reported rates of apnea in the

above studies, it can be calculated that the risk groups

spent approximately only one minute per hour in apnea

(McGinty & Sterman, 1980). At this level the frequency of












apnea can hardly be considered a significant factor in the

induction of hypoxic conditions in risk infants. Apnea may

only represent the final common pathway in a series of

events preceding death.

Control of ventilation. There is evidence that the

mechanisms which control breathing begin to mature and

respond to respiratory stimuli such as level of carbon

dioxide and metabolic acidosis during fetal development.

For a review of the ontogeny of fetal respiratory

development see Maloney and Bowes (1983). The fetus stops

initiating breathing when the partial pressure of arterial

oxygen (PaO2) is lowered to about 15 mmHg (Wilds, 1978).

The adult, however, increases ventilation when PaO2 is

lowered below 50 mmHg (Dejours, 1963). The adult's

hyperpnea or ventilatory increase is sustained until

normoxic levels are achieved. The newborns' response to a

5% decrement in inspired oxygen differs from that of both

the fetus and the adult (Rigatto, 1977). In the newborn,

hypoxia induces a biphasic compensatory response: an

initial ventilatory increase within the first two minutes

followed by a decrement to and below baseline ventilatory

levels. The newly-born preterm infant at 33 and 37 weeks

gestation responds the same as the newborn (Rigatto,

1982). Therefore, gestational age does not seem to be a

factor. However, postnatal maturation may play a role,












since the initial ventilatory increase in the preterm

becomes larger with postnatal age and the adult-like

sustained hyperpnea to hypoxia begins about one month after

birth (Rigatto, 1982). Full term infants, on the other

hand, begin sustained hypoxia-induced hyperpnea at about

two weeks of postnatal age (Rigatto, 1977). The

progressive increase in ventilation seen in the preterm

infants suggests that maturational changes occurred in

either respiratory pump mechanics or peripheral

chemoreception, or both.

The initial increase in ventilation has been

attributed to stimulation of the peripheral chemoreceptors

because of the rapidity of the ventilatory response to

hypoxia onset (Rigatto, 1982). The subsequent fall in

ventilation has been postulated to be due to adjustments in

pulmonary mechanics (e.g., lung compliance decrease),

central depression of respiratory drive, and immature

interaction between carotid sinus chemoreceptive afferent

feedback and the central respiratory mechanisms (Haddad,

Schaeffer & Bazzy, 1983). Hence, it was suggested that in

infants at-risk for SIDS hypoxic depression in an

immature CNS could induce an imbalance of excitatory and

inhibitory influences, which may supercede the

chemoreceptor effect on breathing. This could result in a

state of chronic hypoventilation or be manifested by a













noncompensatory decrement in ventilation to an acute

hypoxic challenge.

Current data suggest that, when compared with adults,

infants have a reduced PaO2, indicating that infants are

normally exposed to a hypoxic challenge (Brooks, Schluete,

Nabelet & Tooley, 1978). More systematic study of this

hypoxic stimulus revealed that periodic apnea could be

induced in preterm infants when subjecting them to mild

hypoxia (Rigatto, 1982). After prolonged exposure,

however, incidence of periodic apnea decreased and central

apnea increased during the decreased ventilation of the

secondary limb of the biphasic response. This suggested

that hypoxia may be an important triggering stimulus for

apnea when the respiratory system is depressed in immature

nervous systems. A similar finding was observed in AI

infants exposed to a 3% decrease in oxygen; most notably,

however, these at-risk infants displayed greater incidence

of periodic apnea and central apnea to this hypoxic

stimulus than did the control infants (Brady, Ariagno,

Watts, Goldman & Dumpit, 1978).

There have been numerous reports of periods of

prolonged apnea in infants at-risk for SIDS (Shannon, Kelly

& O'Connell, 1977; Steinschneider, 1977; Kaun, Blum,

Engleman & Waterschoot, 1982). Increased durations of

apnea in AI infants have been associated with corresponding












greater reductions in the transcutaneous partial pressure

of oxygen (tcPO2), which is a noninvasive measure of PaO2

(Kaun et al., 1982). The prolonged duration of apnea in

the AI infants could be related to a higher peripheral or

central chemoreceptor response threshold to hypoxia or

hypercapnia. That is, during apnea greater hypoxic and/or

hypercapnic levels are accumulated before respiration is

initiated resulting in prolonged apneic episodes.

Therefore, it would be expected that infants with this

elevated threshold would be sluggish to respond to a

hypoxic or hypercapnic challenge.

Support for this suggestion comes from studies of

ventilatory response to hypoxic and hypercapnic stimuli in

at-risk infants. Impaired ventilatory increases to a 5%

reduction in inspired oxygen (i.e., 15% oxygen) were

observed in AI infants (Hunt, McCulloch & Brouillette,

1981). Also, blunted ventilatory responses and a more

pronounced subsequent depression of ventilation were

observed in AI infants to hypoxic challenges of 15% oxygen

compared to controls (Haidmayer, Kurz, Kenner, Wurm & Pfeiffer,

1982). Concomitant with the depression in ventilation,

were more pronounced and rapid declines in tcPO2 values in

the AI infants. Moreover, other studies have found that AI

infants displayed less ventilatory response to hypercapnic

challenges of 4-5% carbon dioxide than controls (Haidmayer












et al., 1982; Hunt et al., 1981; Marotta, Fort, Mondestin,

Hiatt & Hegyi, 1984; Shannon et al., 1977). Therefore,

infants at risk for SIDS have reduced hypoxic and

hypercapnic ventilatory responses, suggesting reduced

sensitivity to chemoreceptive challenge.

In summary, the majority of evidence from

cardiorespiratory physiologic investigations has revealed

that infants at-risk for SIDS display elevated HR levels

and reduced HR variability; RR is also elevated and these

infants exhibit a transient acceleration in the

developmental pattern of RR while sleeping. There are

conflicting data on which type of apnea is present in these

infants and whether the frequency of apnea during sleep is

excessive or diminished. However, when apnea does occur,

the duration is more prolonged and is associated with

corresponding decrement in PaO2. When breathing gas

mixtures of decreased oxygen content or increased carbon

dioxide content, at-risk infants initially displayed a

sluggish ventilatory increase and with prolonged exposure,

subsequently exhibited more pronounced depression of

ventilation.

Sleep and Breathing

The finding that SIDS infants are typically found dead

in the early morning hours, when they were presumed to be

asleep, has led some to speculate that sleep precipitates












the events leading to SIDS (McGinty & Sterman, 1980). There

are rapid maturational changes in sleep and respiratory

physiology in the first six months of life. (See Berg and

Berg (1979) for a review of the ontogeny of sleep,

cardiorespiratory physiology and sensory function in

infancy.) During this time, the sleep states become well

defined and become more organized within the sleeping

periods. There is a consolidation of sleep and waking into

longer and longer uninterrupted periods. By four to six

months, these become coupled so that the longest of the

sleeping periods follows the longest of the waking periods

(Coons & Guilleminault, 1982). By three months stereotypic

elements of adult electroencephalographic (EEG) sleep

patterns can be used to discriminate between AS and QS and

stages within QS (Coons & Guileminault, 1982). For a

recent comprehensive review of the neurophysiology of sleep

the reader may refer to McGinty and Beahm (1984).

In QS breathing is very regular and inspiration is

prolonged. It has been suggested that during QS the

forebrain and the brain stem are uncoupled (Harper, 1983).

However, slow-rhythmic descending discharges from regions

of the forebrain have been documented, which may be pacing

the lower brain stem structures in QS (Harper & Sieck,

1980). Other forebrain areas have been implicated in the

gating of QS onset (Harper, 1983). The influence of the












forebrain in the ontogeny of QS may be evidenced by the

observation of the emergence of a dramatic postnatal

increase in the proportion of total sleep time spent in QS

associated with concomitant increases in postnatal

forebrain development (Parmalee, Stern & Harris, 1972).

During AS, breathing is extremely variable and RR is

increased; desynchronous descending influences of the

forebrain and other areas on brain stem respiratory and

motor neurons affect rhythm and rate of breathing and also

diminish muscle tone (Schulte, Albani, Schmizer & Bentele,

1982). Phasic bursts of rapid eye movement and twitches in

the limbs and peripheral musculature occur. The rib cage

loses stability and during inspiration paradoxical

deflation of the rib cage may be observed due to depression

of intercostal and abdominal muscle activity (Henderson-

Smart & Read, 1978). This might indicate that AS

represents the state of greatest vulnerability to apnea and

hypoxia. However, other observations indicate that AS may

be associated with improved breathing. For example, oxygen

level in the blood of infants is improved in AS compared to

QS (Brooks et al., 1978).

There is evidence accumulating that the state of

greatest vulnerability for apneic episode may be QS. When

kittens were chronically exposed to hypoxic conditions, 28%












of them failed to compensate in QS and exhibited

hypoventilation and prolonged apnea, in some cases

progressing to death; in AS the hypoxemic conditions in

these animals were transiently reversed with faster RR and

decreased incidence of apnea (McGinty & Sterman, 1980).

This evidence suggests that the conflicting reports of

apnea incidence in the at-risk infants may reflect

differential sampling of infants who may or may not be

capable of producing compensatory cardiorespiratory

responses. In support of this possibility, it may be

recalled that the compensatory pattern of elevated HR and

RR with reduced incidence of apnea (which was more

pronounced in QS) was observed in siblings of SIDS who, by

definition, had had a healthy neonatal course with no

documented prolonged apnea attack (Hoppenbrouwers et al.,

1980); whereas no compensatory pattern was observed in AI

infants who had a prior history of prolonged apnea (Hodgman

et al., 1982). This suggests that the siblings of SIDS may

differ from the AI infants in the ability to compensate

with increased cardiac output and ventilation to a hypoxic

challenge; whereas the AI infants having previously

demonstrated some pathophysiologic disorder by virtue of

their prolonged apneic attack, may no longer be capable of

producing compensatory cardiorespiratory responses.












Other evidence suggests some alteration in the

maturation of sleep physiology in at-risk infants.

Siblings of SIDS victims display an accelerated maturation

of 12-15 Hz EEG activity in QS and 4-7 Hz EEG activity in

AS between one and two months of age (Sterman, Harper,

Hoppenbrouwers, McGinty & Hodgman, 1979). It was suggested

that an abnormal stimulus such as hypoxia may have caused

this transient acceleration in EEG development. It has

been speculated that an accelerated maturation of the

forebrain inhibitory influences on medullary respiratory

centers may be present in these infants (Hoppenbrouwers &

Hodgman, 1982). Other evidence indicates that, rather than

being accelerated, the sleep state development in at-risk

infants is delayed. The percent time in QS was less in AI

infants than in controls at three and four months of age,

suggesting a maturational delay or disturbance in the AI

infants (Haddad, Walsh, Leistner, Grodin & Mellins, 1981).

Analysis of the temporal sequencing of sleep-wake epochs in

normal infants and siblings of SIDS, revealed disturbed

patterns of state organization at two to three months of

age (Harper, Frostig, Taube, Hoppenbrouwers & Hodgman,

1983). Siblings were observed to have longer intervals

between periods of wakefulness, a shorter mean QS cycle

period and an increase in the frequency of state

transitions. The prolonged periods of uninterrupted sleep












may reflect an inability to arouse from sleep (Harper et

al., 1983). Hunt (1981) observed that AI infants also

showed a tendency to remain asleep in the face of hypoxic

challenge; whereas control infants displayed an increased

frequency of arousal from sleep when inspired oxygen was

progressively decreased to 15%. The authors posited that

the failure to arouse may be due to a defect in peripheral

chemoreceptor sensitivity. Others have hypothesized that

the chronic hypoxemia of SIDS, and the hypoventilation and

failure to arouse in at-risk infants may be due to tonic

overactive synthesis of endorphins, which have previously

been associated with the CNS suppression of respiration and

wakefulness (Kuich & Zimmerman, 1981). Some support for

this theory was observed in post-mortem examination of

met-enkephalin (an endogenous endorphin) levels in

cerebrospinal fluid, when it was found that SIDS victims

had greater concentrations than age-matched controls

(Rappaport, Kozakewich, Fenton, Mandell, Vawter & Yang,

1984). Regardless of the etiology, the failure to make

transition from sleep to waking may place these infants at

greater risk. The finding of increased incidence of apnea,

occurring antecedent to awakening in infants in the

first six months of life, suggests that arousal from sleep

may be an important mechanism in terminating apneic

episodes (Guilleminault et al., 1981).












Cardiorespiratory Responses to Airstream Stimulation

The evidence from pathological and physiologic

investigations indicate that infants at-risk for SIDS may

be chronically hypoxic, are sluggish to respond to hypoxic

and hypercapnic stimuli with a ventilatory increase and

are less likely to arouse from sleep during such a

challenge; also, while asleep they consistently exhibit

prolonged periods of apnea with concomitant bradycardia.

Hence one of the most striking deficits in these infants

is their higher tolerance of hypoxia and hypercapnea

before responding appropriately. This deficit is

particularly apparent from two to four months of age,

which is the period of most rapid CNS development and also

the age of greatest vulnerability for SIDS. Therefore it

may be possible to detect an infant at risk for SIDS by

eliciting apnea and thereby establishing a hypoxic or

hypercapnic challenge to which they would be expected to

respond sluggishly. The prolonged apneic episodes that are

spontaneously produced by these infants are probably

related to a more general CNS dysfunction of autonomic

depressor reflexes involving the complex interactions of

the medullary chemoreceptors with inputs from the

trigeminal, glossopharyngeal and vagus cranial nerves,

peripheral chemoreceptors, sleep control mechanisms and

cardiorespiratory regulating mechanisms.












Investigations in infants of the ontogeny of reflex

mechanisms which when stimulated can induce apnea are

limited to four areas of research: 1) laryngeal

chemoreceptor stimulation and gastroesophageal reflux;

2) ventilatory response to hypoxic and hypercapnic

exposure; 3) vagal-mediated reflexes in compensation to

inspiratory load; and 4) dive reflex and facial cooling

with an airstream.

In animals stimulation of the laryngeal chemoreceptors

with chemical irritants have been found to evoke prolonged

apnea (Downing & Lee, 1975). There has been a report of an

association of apnea and gastroesophageal reflux in

premature infants, which presumably induces reflex apnea by

acidic reflux stimulation of the laryngeal chemoreceptors

(Herbst, Minton & Book, 1979). However, others have found

no such relationship in premature and full-term infants, and

AI infants (Ariagno, Guilleminault, Baldwin & Owen-

Boeddiker, 1982; Walsh, Farrell, Keenan, Lucas & Kramer,

1981). Some have studied hypercapnic ventilatory

responsiveness and sustained hypoxic hyperpnea and have

observed that at some critical level of PaO2 and PaCO2

periodic apnea is induced and with further exposure central

apnea eventually occurs (e.g., Rigatto, 1982). However,

the role of this mechanism in initiating prolonged apnea

has not been examined. Inflation of infants lungs in order












to activate vagal-inhibition of respiration has revealed

that premature infants are more susceptible to respiratory

inhibition than full-term healthy infants at any age (e.g.,

Boychuk, Rigatto & Sheshia, 1977; Kirkpatrick, Olinsky,

Bryan & Bryan, 1976). Although this technique when studied

developmentally would provide information on the ontogeny

of vagal reflexes in compensation to an inspiratory load,

it is subject to confounding influences of changes in

pulmonary mechanics due to sleep state (Henderson-Smart,

1984). Nevertheless, this test is more likely to provide

information on inspiratory load compensation than

information on the development of mechanisms of prolonged

respiratory cessation. Although research on these first

three areas could be of importance to a better understanding

of apneic episodes in SIDS, apnea induction by the dive

reflex and facial cooling may have the greatest relevance.

The dive reflex is a depressor reflex which appears to

resemble the acute cardiorespiratory response observed in

infants during an apneic episode. Diving in animals like

seals elicits apnea, profound bradycardia and peripheral

vasoconstriction. These responses have the clear

physiological function of conserving oxygen by reducing

circulation to tissues of greater anaerobic capacity (e.g.,

skin, mesentary, muscle) and providing circulation to

tissues less tolerant of the effects of oxygen deprivation












(e.g., heart, brain). The ophthalmic division of the

trigeminal nerve, which in humans innervates the forehead,

eyelids and dorsum of the nose, was found to be primarily

responsible for completing the afferent pathway by which

dive reflexes were elicited in the duck (Anderson, 1963).

The dive reflex has also been observed in humans

(e.g., Elsner, Franklin, Van Citters & Kenney, 1966;

Kawakami, Natelson & DuBois, 1967). More recently,

elaboration of the relevant eliciting stimuli has revealed

that mere submersion of the face in cold water was

sufficient to elicit the bradycardia and peripheral

vasoconstriction typical of the dive reflex (Hurwitz &

Furedy, 1979). Furthermore, it was the combination of cold

facial stimuli and breath holding that produced greater

bradycardia and vasoconstriction than in either control

conditions of face immersion (while breathing through a

snorkel) or breath holding alone. Others have found that

the dive reflex was even more pronounced with colder water

temperatures (Furedy, Morrison, Heslegrave & Arabian, 1983;

Song, Lee, Chung & Hong, 1969). The dive reflex literature

illustrates the significance of cold facial stimulation in

effecting cardiovascular and respiratory responses.

In human infants apneic and bradycardic responses have

been elicited by nasopharyngeal suction and by blowing

oxygen over the infants' face (Brown, Ostheimer, Bell &















topography of the HR response from stimulus onset, which

would have provided more information on the temporal effect

of the stimulus and would have been a less biased measure.

Brief apneic episodes in response to the airstream stimulus

were observed but no difference in apnea duration existed

between the two infant groups. Unfortunately, no further

attempt was made to examine the apnea responses

quantitatively. Moreover, no systematic analysis was

performed to partial out the effects of sleep state,

postnatal age or presence or absence of apnea from their

possible effects on the cardiorespiratory responses

observed.

The proposed study, herein, was designed to establish

the normative response to facial airstream stimulation in

infants throughout the first year of life. This study has

a focus similar to that of McCubbin et al. (1977) but

utilized a more systematic design, including the

appropriate control conditions to compare the

cardiorespiratory responses. The study was designed to

assess in sleeping infants developmental changes in

cardiorespiratory responses to airstream stimulation in the

first year of life by 1) determining whether airstream

stimulation of the facial region supplied by the ophthalmic

branch of the trigeminal nerve (i.e., forehead, eyelids and

dorsum of the nose) would elicit apnea more frequently than











a control airstream stimulus directed to the abdomen (i.e.,

rostral to the umbilicus); 2) measuring the HR response on

a second-by-second basis and average a sufficiently large

number of homogeneous trials to determine the precise

topography of the cardiac response to airstream

stimulation; and 3) examining the effect of apnea

elicited by the airstream stimulus on the HR response

topography.

This study will be informative for the following

reasons. First, the examination of the cardiorespiratory

responses to repeated airstream stimulation will provide

information on the maturation of cardiorespiratory control

in the production of depressor reflexes in normal infants

cross-sectionally throughout the first year of life. It

will be informative to determine whether the cardio-

respiratory responses in infants, during the period of most

rapid CNS development (i.e., two to four months of age) and

vulnerability to SIDS, differ from responses during

earlier or later postnatal age. Second, it will provide

more information on the cardiorespiratory effects of

conditions of facial cooling similar to those which exist

presently in hospital delivery protocol manipulations such

as nasopharyngeal suction and oxygen blown over the face.

Third, since the airstream stimulus is not substantially

different from that which the infant might commonly







37




experience in every day life, this study may shed some

light on the eliciting conditions of apnea and bradycardia

in normal healthy infants. Fourth, this simple reflex test

of the integrity of trigeminal-brainstem-cardiorespiratory

function could have prognostic significance for screening

of cardiorespiratory risk for SIDS. However, baseline

data in a normative sample are required initially so that

the results of this study will provide information against

which data from high risk infants may be compared.















Maternal and infant hospital medical records of the

newborn group were surveyed to insure that all criteria were

met. For the older groups information on the infants'

delivery and neonatal course and the maternal obstetrical

history was obtained in a questionnaire during testing (see

Appendix A). Table 1, 2 and 3, respectively, shows the sex,

gestational age (GA), postnatal age (PNA) at the time of

testing, birth weight (BW), APGAR scores, race (R), maternal

age (MA), and maternal medication (MED) at delivery for

each infant in the three age groups included in this study.

Newborn subjects were recruited by obtaining permission

from their mother or father in the neonatology ward of the

J. Hillis Miller Health Center. These subjects received no

remuneration for their participation. The older subjects

were recruited through the mail. Names of infants born in

the appropriate age group were obtained through the Alachua

County Health Department. Parents were sent a general letter

describing the study and were requested to indicate whether they

were interested in participating in the study on an enclosed

stamped-postcard. Those parents responding affirmatively

were telephoned to schedule an appointment for testing. In

this conversation parents were informed about the purpose

and specifics of the study and an appointment was scheduled

at a time that would correspond to the infant's nap schedule.
1
These subjects received five dollars for their participation.
























































CHAPTER 3
RESULTS

The factors of critical interest in this study were

the site of airstream stimulation and the age of the infant

when tested. Knowledge of these effects indicate whether

cardiorespiratory responses to stimulation of the facial

region supplied by the ophthalmic branch of the trigeminal

nerve may be unique and if so, whether it changes

throughout the first year of life. However, the behavioral

observations revealed that two other factors, sleep state

and presence of elicited-large movement, may also have

influenced the cardiorespiratory response. Before the

effects of these two latter factors on the cardio-

respiratory response could be examined, it was necessary

to assess the infants' sleep state pattern during testing

and to determine the extent to which large movement was

elicited by the airstream stimuli.

Sleep State Incidence and Pattern

Table 4 presents the percentage of trials in which

the sleep state of the infants were rated as QS, TS and

AS across the age groups. The most notable finding was

that the majority of data were collected while the infants

were in QS. However, the NB group tended to exhibit a
























0 0
o

iE N
d H














r- 4-
- ""'r---



















4m
----I




IN-
I : om












IJ I -4 I rL
at -O CD
D7 ji c


o.)-J uLL L- 'C' H- C -- LS


,'" -: I- 1-- I1'


































































Prestimulus Levels

Prior to the analysis of the HR response topography

the prestimulus levels of HR were evaluated to determine

whether the post-stimulus HR response could have been

influenced by initial prestimulus levels. This was of

particular concern since it was to be expected that

prestimulus levels would co-vary with infant age (Harper et

al., 1982). Table 12 depicts the mean HR values in beats

per minute (bpm) for the second preceding stimulus onset

across the age groups during trials with and without

elicited-apnea.

The analysis of the prestimulus HR level in the trials

in which no apnea or LM were elicited used Age as the

between-subject factor and Site of stimulation as the

within-subject factor. Only the Age effect was

significant, F(2/47)=5.9, 2<.01. With no significant

difference between the two younger groups, the difference

between the age groups, therefore, can be accounted for by

diminished prestimulus levels in the 8-12 infants

Post-stimulus Changes

The interpretation of the post-stimulus HR response

topography is complicated by the apparent relationship

between prestimulus HR levels and the infants postnatal

age. Hence, the post-stimulus HR levels and perhaps

topography may be partially determined by the prestimulus













Mean Prestimulus
Apnea and Large


TABLE 12
HR Levels in bpm in the Presence and Absence of
Movement (LM) for Facial (F) and Abdominal (A)
Airstream Stimulation Trials


Apnea Movement Condition NB 2-4 8-12 Mean

No No LM F 117.92 122.76 110.52 117.83

A 118.72 122.17 112.11 118.26


Mean 118.32 122.47 111.32

N 13 22 15


Yes LM F 123.56 130.03 106.65 124.90


N 11 10 2


Yes No LM F 121.21 121.45 117.16 120.31


8 13












HR levels. This statistical problem is known as the Law of

Initial Values as outlined by Wilder (1950). Following the

guidelines suggested by Richards (1980) for countering the

effects of prestimulus HR on postimulus HR, the assumption

of homogeneity of regression coefficients for the data

were tested over age for regression of each of the 25 post-

stimulus seconds on the first prestimulus second. Each of

these analyses revealed that the assumption of homogeneity

of regression coefficients was not violated. Therefore,

the effects of the experimental parameters on the HR

response topography could be compared across age. However,

to control for the effects due to the Law of Initial
5
Values, the analysis of covariance (ANACOVA) was used.

Since the interpretation of HR change was of primary

concern, difference scores were calculated by taking the

algebraic difference between the prestimulus value and the

post-stimulus values. The correlation between prestimulus

and difference scores have typically been found lower than

the correlation between prestimulus and post-stimulus

scores (Benjamin, 1967). Therefore, the use of difference

scores, herein, provided an additional statistical

advantage. Despite lower correlations, however, difference

scores still remain subject to the Law of Initial Values.

The HR difference scores, therefore, were all analysed with

ANACOVAs using the HR value in the second preceding stimulus












onset as the covariate, Age as the between-subject factor

and Seconds as the within-subject factor. It should be

noted that all significant effects reported below remained

significant when the conservative Greenhouse-Geisser

approximation of degrees of freedom was used (Greenhouse &

Geisser, 1959).

HR Response to the Airstream Stimulus: Face vs. abdomen

Figures 2 and 3 depict the HR change scores at each

age in response to facial and abdominal stimulation in the

absence of apnea and LM. For both stimulation sites these

figures show that the a short monophasic acceleration to

the airstream was produced by the NBs. However, the two

older groups displayed a triphasic response to the

airstream stimuli comprised of a brief deceleration

followed by an acceleration and a subsequent deceleration.

After stimulus offset HR levels in these infants gradually

recovered toward prestimulus levels.

More specifically, ANACOVA analyses (see Appendix F)

revealed that an initial deceleration was elicited in the

older infants within one second of stimulus onset but was

not elicited in the NBs. An acceleration followed

maximizing at the third post-stimulus second in all age

groups. This acceleration, which was greater when elicited

by the facial airpuff than by the abdominal airpuff,

differentiated all the age groups such that the 8-12














,dNi

0 1




**--I -






r. 4 a
a






LL







S 0 -- 0.0
r- 1i






I I I :. I : I
0~~~~~~~~L uj;6: '" 'JCO0J ^'1C


rck~~ ~ ~ *zr~iu 'cz cici













LO C) Iq-4
Lo




U
C.







rl
S.4J









-In
k! CO C













- 00
< L.. .




-*-









I I


SCt n a: l-I-'-LU --,- CuJI i.


CIJ

C l I I
:Z:OJCO


ax' 'X Ir r, -':1 c~. ix' ej
I I











infants displayed the fastest acceleratory recruitment

followed by the 2-4 infants and then the NBs. As in the

early deceleration, the NBs were differentiated from the

older infants by the ensuing deceleration from seconds

three to ten, which reached its nadir by the seventh post-

stimulus second in the older infants but merely returned to

pretrial levels in the NBs. Only the 2-4 infants displayed

an effect of the stimulation site during this interval with

greater deceleration to the facial airpuff than the

abdominal airpuff. Following stimulus offset all groups

displayed a cubic trend over seconds 11 to 25, which was

unaffected by the stimulation site, appearing to decrease

slightly 5 to 10 seconds after stimulus offset followed by

a gradual return to pretrials levels.

HR Response to the Facial Airstream Stimulus: Apnea vs. no apnea

Figure 4 shows the response to facial stimulation when

apnea was present without preceding LM. By comparing these

responses with the responses in Figure 2 in which no apnea

was elicited by the facial airstream, the effect of apnea

elicitation on HR can be examined. Although the overall

topography of the response was similar with and without

apnea, when apnea was elicited, the older groups displayed

a more prolonged late deceleration which was of greater

magnitude than when no apnea was elicited. The NBs

response was unaffected by whether or not apnea was induced.




















,.- '..

m I I



4-J








I- *- -A
I- Il H!




-= -4 C'




r OJ 0 -J -t LID I- s
i I I -I - -*


T -"" =' --Ei '-7" .SJ LLI O LU .. "












The ANACOVAs of the overall HR response used seconds 1

to 25 as levels of the within-subject Seconds factor, Age

as the between-subject factor and presence or absence of

Apnea as the second within-subject factor A significant

interaction between Age, Apnea and Seconds, F(48/576)=2.0,

r<.001, was found. The next analysis examined whether the

differences described by this interaction existed during

the period of acceleration across seconds 1 to 3. It did

not, since only a significant Seconds effect, F(2/48)=6.7,

2<.005, emerged. Therefore, up to this point in the

analysis the HR response was unaffected by the age of the

infant or by the presence or absence of apnea.

The age groups were differentiated by the HR response

during the late-deceleration from seconds 3 to 12, as

indicated by a significant interaction between Age, Apnea

and Seconds, F(18/216)=3.1, p<.001. The subanalysis

consisted of examining the NB group alone. Only a

significant linear trend across Seconds, F(1/6)=21.1, p<.005,

was found, indicating that there was a linear HR

decrement toward base level both in trials with apnea and

trials without apnea. Subsequent t-tests revealed that the

decrement in the NBs HR during this interval did not

proceed below prestimulus levels. When the two older groups

were contrasted a significant interaction between Age, Apnea

and Seconds, F(9/162)=4.7, p<.001, emerged. This was due












to changes in both magnitude.and topography of the

deceleration when apnea was elicited. When the two older

groups were compared using only the non-apnea responses,

only a significant quadratic trend over Seconds,

F(1/18)=7.5, p<.05, emerged, indicating that there was no

difference between groups in the magnitude of the

deceleration when no apnea was elicited. However, on

trials when apnea was elicited, the magnitude of

deceleration was increased to a greater extent in the

8-12 infants than in the 2-4 infants, as substantiated by

a significant interaction between Age and Seconds,

F(9/162)=6.3, p<.001. Moreover, the effect of apnea was

to prolong the deceleratory response differentially in

these older infants. As can be seen in Figure 4, the

deceleration in the 2-4 group reached its nadir and began

its recovery toward pretrial levels sooner than the 8-12

infants. This was confirmed by an orthogonal trend

analysis of each age group for apnea data only, with a

significant decreasing quadratic trend over Seconds,

F(1/12)=31.6, p<.001, for the 2-4 infants and a decreasing

linear trend over Seconds, F(1/6 )=39.1, p<.001, for the

8-12 infants.

The analysis of seconds 12 to 25, which represented

the period of HR recovery, yielded a significant main effect

of Apnea, F(1/23)=7.0, p<.05, and an interaction between Age












and a quadratic trend over Seconds, F(2/24)=5.2, p<.05.

Neither the effects of Apnea or Seconds nor their interaction

were significant when analysing only the NB group's HR

recovery, indicating that regardless of whether apnea or no

apnea was elicited the NB HR was stable throughout this

period maintaining a mean HR level of -0.65 bpm. When

comparing the younger groups only a significant Age effect,

F(1/17)=6.2, p<.05, was found, indicating that the 2-4

group operated at a lower level than the NB group, with

mean HR bpm levels of -0.65 and -5.01 for the NB and 2-4

groups respectively. When comparing the 2-4 and 8-12

groups, however, an interaction between Age, Apnea and

quadratic trend over Seconds, F(1/18)=4.5, r<.05, emerged.

No difference was found in the recovery trends of the

trials with no apnea. When apnea did occur, however, the

enhanced late-deceleration and the resultant phase

shift between these groups resulted in interacting

decreasing and increasing quadratic trends over Seconds,

F(1/18)=7.5, p<.05. Inspection of Figure 5 reveals that by

post-stimlulus second 12 the HR response of the 2-4 group

had partially recovered, while the 8-12 group had just

reached its nadir. Hence, HR levels decreased to a mean of

-7.25 bpm for the 2-4 infants while the HR levels increased

to a mean level of -8.58 bpm for the 8-12 infants.












In summary, the NBs HR response when apnea was

elicited was not distinguishable from the response observed
7
when no apnea was elicited. The main effect of apnea on

the HR response appeared in the older groups. Apnea

resulted in an accentuated deceleration during seconds 3 to

12 post-stimulus onset reaching a maximum at seconds 7 and

12 respectively for the 2-4 and 8-12 groups. Apnea also

had an effect on the HR recovery during seconds 12 to 25 in

the older groups, exhibiting less tendency to return to

pretrial levels.

HR Response to the Facial Airstream Stimulus from Apnea Onset

The previous HR analysis used HR values scored from

stimulus onset. When apnea was present, it could occur

anywhere in the trial within the 20-second post-stimulus

observation window. Therefore, the topography of the HR

response in the presence of apnea was difficult to

interpret. This was especially the case, since the latency

to apnea onset was significantly related to age, with older

infants displaying longer apnea latencies. To obtain a

more precise determination of the effect of apnea on the

response topography, HR was aligned relative to apnea onset

rather than stimulus onset. To maintain a constant

standard of reference, these HR values, as in the above HR

statistics, were represented relative to prestimulus

level. For example, the HR value at second zero












represented the HR difference from prestimulus level of the

interval just preceding apnea onset.

The mean HR values observed during apnea as elicited

by the facial airpuff (uncomplicated by LM) were plotted in

Figure 5. The analysis of the HR response from apnea onset

did not examine the effect of preceding LM on the HR

topography. As can be seen in Figure 6, the effect of LM

was to substantially increase HR (relative to prestimulus

level) prior to apnea onset. Otherwise the response

topography remained similar to the response when no LM

preceded apnea. As can be seen in Figure 5, the HR

response of the NBs was not affected by the production of

apnea; whereas, in both of the older groups apnea induced

rapidly-recruiting, large-magnitude decelerations, which

began to recover to prestimulus levels when apnea

terminated. Note that the arrowhead located below each

curve in both Figures 5 and 6 corresponds to the point

where apnea ended and breathing resumed.

ANACOVA of the no LM data (Figure 5) assessed the

pre-apnea HR levels. Seconds -5 to 0 were used as levels

of the Seconds factor, Age as the between-subject factor

and the unadjusted HR value in the second preceding apnea

onset (i.e., second 0) as the covariate. There were no

differences to emerge during this period as neither the

effects of Age or Seconds nor their interaction were

significant.















El -n) 0


mi l





11

0- -o
44 I i UI





S-- C


s^ S-1 a^!
LUJ




n' c '- '.k J LU
-T I s *



-4 -- -

| i I |


1 i C I I I
0.3 s O W .*.sUs.. C' OJ -.'a C
II II
Il~ il


-l- I I'T---I -q C _CL.











:"-
110







S- I



-6
l lG -H-






-
LM =-4 k-H b
" ... ,,- -.,
I II! &
Ifu






iLLI n



?.'. C'i ^- T-L 1 -- j
I I I ID N


*) 1: I5-~ r ? ii












Inspection of the trends in Figure 5 suggested that

there was an initial deceleration upon apnea onset in the

older groups, which was absent in the NB infants. The

analysis of this trend used seconds 0 to 9 as the period of

deceleration after apnea onset. The ANACOVA yielded a

significant Age effect, F(2/24)=4.0, p<.05, a Seconds

effect, F(9/225)=7.7, p<.001 and a two-way interaction of

these factors, F(18/225)=1.9, p<.05. The subanalysis of

the most salient of these effects the Age-by-Seconds

interaction consisted first of assessing whether there

was a Seconds effect in the NB group alone; there was not,

F(9/63)=1.0, indicating that when apnea was elicited it had

no effect on the NBs HR response. An ANACOVA was then

performed on the two older groups. Only a significant

Seconds effect, F(9/162)=4.6, p<.001, was found.

Therefore, from the onset of the apneic period to the ninth

second of the trial, the two older groups exhibited HR

deceleration, which was of similar magnitude and which was

totally absent in the NB group. The 2-4 infants reached a

maximum deceleration of -14.46 bpm at seven seconds post-

apnea onset; whereas the 8-12 infants decelerated to -15.91

bpm by the ninth second after apnea began.

Analysis of seconds 9 to 25 was performed to assess

the period of recovery from the apneic episode. The NB

group displayed no significant change over these seconds,




Full Text
APNEA AND BRADYCARDIA ELICITED BY FACIAL AIRSTREAM
STIMULATION IN HEALTHY INFANTS IN THE FIRST YEAR OF LIFE
IMPLICATIONS FOR DETECTION OF INFANTS AT RISK
FOR SUDDEN INFANT DEATH SYNDROME
By
BARRY ELLIOT HURWITZ
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1384

Dedicated with love to my wife, Debbie
and my son, Joseph

ACKNOWLEDGEMENTS
I thank Dr. W. Keith Berg for his guidance, astute
judgement and his infectious positive approach to life.
His active support of our past and present collaborative
efforts has meant a great deal to me. I look forward to a
long and fruitful future association. As well, I thank you
for so willingly being there whenever I needed you and for
patiently allowing me to interrupt your stories of the joys
of fatherhood in order to tell you mine.
I would also like to express my appreciation to my
doctoral committee, Drs. Merle E. Meyer, Peter Lang, Phil
Posner and Neil Rowland, for their assistance throughout my
graduate training. Also, I thank Drs. Don Stehouwer and Paul
Davenport for their helpful comments on this project.
In addition, I thank all those infants who fell asleep
for me and most notably I thank all those little kids who
held their breath for me and even those that didn't. Also,
I am grateful to the parents of the little kids for
generously contributing their time to this effort.
Special thanks are extended to my assistants, Alonso,
Judy, Kristen, Karen, Julio and Yolanda, for their help in
data collection at various times throughout the past two
i i i

years. I thank my lab mates Terry Blumenthal, Margaret
Davies and Allen Klempert for their enjoyable association
and patient tolerance.
I thank my parents and the rest of my family for their
tenacious support and everlasting expression of faith and
encouragement. I stand tall because of them. Finally, to
my wife and true friend Debbie, thank you for being there
when I wasn't and also when I was, and more importantly
thank you for my son, Joseph.
IV

TABLE OF CONTENTS
ACKNOWLEDGEMENTS iii
LIST OF TABLES vii
LIST OF FIGURES viii
ABSTRACT ix
CHAPTER
1 INTRODUCTION 1
Epidemiology 3
Pathology 6
Cardiorespiratory Physiology. .......... 9
Control of Cardiac Function .......... 10
Control of Respiratory Function , 13
Respiratory Patterns 13
Frequency of apnea 15
Control of ventilation 20
Sleep and Breathing 24
Cardiorespiratory Responses To Airstream
Stimulation 30
2 METHODS 38
Subjects. 38
Apparatus 43
Experimental Manipulations and Design ...... 46
Procedure 47
Data Quantification ..... 49
Notes 52
3 RESULTS 53
Sleep State Incidence and Pattern 53
Large Movement Incidence 55
Respiration Response to Airstream Stimulation . . 59
Percent of Elicited Apnea 59
Effect of airstream stimulation site 59
Effect of sleep state 62
Effect of preceding large movement. ..... 62
latency to Apnea Onset 66
Effect of airstream stimulation site. .... 66
Effect of preceding large movement 66
v

Duration of Apnea 68
Effect of airstream stimulation site 68
Effect of preceding large movement 71
Summary of the Respiration Response to
Airstream Stimulation 73
HR Response Topography 74
Prestimulus Levels 75
Post-stimulus Changes 75
HR response to the airstream stimulus:
face vs. abdomen 78
HR response to the facial airstream
stimulus: apnea vs. no apnea 81
HR Response to the Facial Airstream Stimulus
from Apnea Onset 86
Notes 92
4 DISCUSSION 94
Implication for Dive Reflex as the Mechanism
for Inducing Apnea and Bradycardia 97
Topography of HR Response in the Absence
of Apnea 1C4
Detection of Infants At-Risk for SIDS ..... Ill
Note .115
APPENDICES
A INFANT BACKGROUND QUESTIONNAIRE 116
B INFANT STATE RATING SHEET 118
C INFORMED CONSENT 119
D ANALYSIS OF THE PERCENT OF TRIALS WITH QS, TS
AND AS STATES 122
E ANALYSIS OF THE PERCENT OF TRIALS WITH
LARGE MOVEMENT 125
F ANALYSIS OF THE HR RESPONSE TO THE AIRSTREAM
STIMULUS: FACE VS. ABDOMEN ... 127
REFERENCES 130
BIOGRAPHICAL SKETCH .... 143
vi

TABLES
LIST OF TABLES
PAGE
1 Description fo the Newborn Subject Pool 40
2 Description of the 2-4 Month-Old Subject Pool .... 41
3 Description of the 8-12 Month-Old Subject Pool. ... 42
4 Percent of Trials in Quiet Sleep, Acitive Sleep and
Transitional Sleep States . . 54
5 Percent of Trials with Large Movement Elicited by
Face and Abdomen Airstream Stimulation 58
6 Percent of Trials with Apnea Elicited by Airstream
Stimulation of the Face and Abdomen When
Sleep State Is Ignored and When Infants Were
in Qjiet Sleep . . 60
7 Percent of Trials with Apnea Elicited by Airstream
Stimulation of the Face and Abdomen When
Preceded by Large Movement and When Large
Movement Was Absent 63
8 Latency to Apnea Onset in Seconds as Elicited by
Airstream Stimulation of the Face and Abdomen ... 67
9 Latency to Apnea in Seconds Elicited by Airstream
Stimulation of the Face When There Was Preceding
Large Movement and When Large Movement Was Absent . 69
10 Duration of Apnea in Seconds Elicited by Airstream
Stimulation of the Face and Abdomen 70
11 Duration of Apnea in Seconds Elicited by Airstream
Stimulation of the Face When There Was Preceding
Large Movement and When Large Movement Was Absent . 72
12 Mean Prestimulus HR Levels in bpm in the Presence
and Absence of Apnea and Large Movement for Facial
and Abdominal Airstream Stimulation ........ 76

LIST OF FIGURES
FIGURES
PAGE
1 Mean Sleep State Rating for the First and Last Ten
Trials 56
2 Mean HR Response to Facial Airstream Stimulation When
No Apnea or Large Movement Were Elicited 79
3 Mean HR Response to Abdominal Airstream Stimulation
When No Apnea or Large Movement Were Elicited 80
4 Mean HR Response to Facial Airstream Stimulation When
Apnea Was Elicited Without Preceding Large Movement . . 82
5 Mean HR Response to Facial Airstream Stimulation When
Apnea Was Elicited Without Preceding Large Movement
Adjusted from Apnea Onset 88
6 Mean HR Response to Facial Airstream Stimulation When
Apnea Occurred Following Large Movement Adjusted from
Apnea Onset 89

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
APNEA AND BRADYCARDIA ELICITED BY FACIAL AIRSTREAM
STIMULATION IN HEALTHY INFANTS IN THE FIRST YEAR OF LIFE:
IMPLICATIONS FOR DETECTION OF INFANTS AT RISK
FOR SUDDEN INFANT DEATH SYNDROME
By
BARRY ELLIOT HURWITZ
August, 1984
Chairman: W. Keith Berg
Major Department: Psychology
Cardiorespiratory responses to airstream stimulation
(10-seconds, 1.06 psi) were observed in 52 newborn (NB),
2-4 and 8-12 month-old healthy infants. Two areas of
stimulation were compared: the facial region, which
included the forehead, eyelids and nose, and the abdominal
region rostral to the umbilicus, which served as a control
stimulation site. The infants were tested while asleep,
during which EKG, heart rate and respiration were recorded.
The results revealed that in both NB and 2-4 month
infants, facial cooling with an airpuff elicited apnea with
greater frequency and duration than did the control
airpuff. Apnea occurred on about 27.2% of facial trials
compared to 7.8% of control trials. The duration of apnea
i x

was twice as long in the 2-4 month infants as in the NBs.
By 8-12 months the facial airpuff still elicited apnea
more reliably than did the abdominal airpuff but with
reduced frequency of 8.9% and 1.9% respectively. The
duration of apnea, however, was maintained at the longer
duration seen with the 2-4 month infants. Bradycardia on
trials when apnea was elicited differed in magnitude
between age groups such that the 2-4 and 8-12 month infants
decelerated to about -15 bpm; whereas no bradycardic
response to apnea was displayed by the NBs.
It has been demonstrated that in healthy sleeping NB
and 2-4 month infants an airstream directed to the face can
elicit apnea. However, by 8-12 months facial cooling does
not have as great an effect on inhibition of respiration.
Although apnea can be elicited in the NBs by facial
airpuff, this apnea was brief and resulted in no change in
cardiac response. Therefore, at 2-4 months, which is the
age of greatest vulnerability for Sudden Infant Death
Syndrome (SIDS), the airpuff was likely to produce more
frequent, prolonged apnea with concomitant bradycardia.
It has been suggested that SIDS infants produce prolonged
apnea because they have a higher threshold for hypoxia.
Therefore, by eliciting apnea with facial cooling, it
may be possible to identify a predisposition to SIDS by
analysis of the magnitude of cardiorespiratory change in
infants with a history of prolonged apnea.
x

CHAPTER 1
INTRODUCTION
"and this woman's child died in the night,
because she overlaid it"
This quote, which is from an account in the Book of
Kings 3:16-28 circa tenth century B.C., in which the word
"overlaid" appeared to imply suffocation of an infant
during the night, has been cited as perhaps the first
recorded instance of Sudden Infant Death Syndrome (SIDS)
(Peterson, 1980). Despite it's probable antiquity, SIDS
was largely ignored by the scientific community until, as
infant mortality rates declined, it became apparent that
these deaths comprised a substantial proportion of all
infant deaths within the first year of life. Currently the
incidence of SIDS throughout the world varies from 0.6 to
3.0 deaths per 1000 live births, accounting for about
7-9,000 deaths each year in the United States alone
(Peterson, 1980; Valdes-Dapena, 1980). About one-third of
all deaths in infants between the ages of one week and one
year are diagnosed SIDS and it is the most frequent cause
of death in this age group (Moore, 1981) .
Prior to 1963, deaths that occurred suddenly,
unexpectedly, and inexplicably with no recognizable lethal
pathology at autopsy were generally certified to one or
1

2
another cause--of-death category in keeping with the
attending coroner's personal inclination (Peterson, 1980).
This practice circumvented the embarassment of a factual
"don't know" and at the same time complied with legal
expectations. Following the 1963 conference on causes of
SIDS, awareness and acceptance of sudden unexpected death
as a legitimate diagnostic entity increased. Finally the
designation, Sudden Infant Death Syndrome, emerged from the
Second International SIDS Conference in 1969, wherein an
occurrence of SIDS was defined as the sudden death of any
infant which was unexpected by history, and in which a
thorough post-mortem examination fails to demonstrate an
adequate cause for death (Bergman, Beckwith & Ray, 1970).
The following text will review certain epidemiological,
pathological and physiological findings related to SIDS as
background for the proposed study. For a comprehensive
review of SIDS, the reader is referred to these recently
published reviews (Brooks, 1982; Kelly S Shannon, 1982;
Shannon & Kelly, 1984; Valdes-Dapena, 1980). In addition,
as a basis for comparison, developmental findings in normal
healthy infants will be included where relevant. Based on
this information, an experimental methodology, which could
be used as a screening test for detection of infants at
risk for SIDS, will be proposed to examine the development

3
of the normal infant's cardiorespiratory response in the
first year of life.
Epidemiology
Epidemiologic evidence has been gathered by the
scientific community in an effort to identify a population
of infants who might be at increased risk for succumbing to
SIDS. In the absence of pathological evidence which could
directly account for SIDS, the epidemiological findings
also have been instrumental in generating hypotheses about
the etiology of SIDS. It is important that any hypothesis
about the cause of SIDS account for these epidemiological
factors. It has been found that SIDS occurs while the
infant is asleep in the early morning hours, between
midnight and six a.m., with no cry or stridor (Peterson,
1980). The incidence of infant mortality due to SIDS peaks
between two to four months of age. Less than ten percent
of all SIDS deaths occur when infants are younger than two
weeks and older than six months of age (Brooks, 1982). It
should be noted, however, that the peak age for infant
mortality of known causes also occurs at two to four months
of postnatal age (Kelly & Shannon, 1982).
About 60% of SIDS victims are male and incidence for
nonwhites is twice that for whites; SIDS is more
common in low birth weight infants (i.e., 11/1000 live
births), subsequent siblings of SIDS victims (i.e., 21/1.000

4
live births) and in the surviving twin of SIDS victims
(i.e., 42/1000 live births) (Peterson, 1980). These data
suggest that either genetic factors have promoted the
demise of these infants or an environmental influence has
affected their intrauterine or extrauterine development. It
should be noted, however, that no difference was found
between the concordance of SIDS episodes in monozygotic and
dizygotic twin pairs; in addition, the incidence of SIDS in
full first cousins of SIDS victims did not differ from the
incidence in the general population (Peterson, 1980).
Thus, overall the data do not tend to support a genetic
component to SIDS.
There are several epidemiological perinatal-factors
which indicate an environmental influence on SIDS. SIDS
infants are born with higher incidence to young, unmarried
mothers of lower socioeconomic levels who have had poor
prenatal care, have short interpregnancy intervals, have
previous fetal loss and who are smokers or have narcotic
dependency (Valdes-Dapena, 1980). Other environmental
factors which may play a role in SIDS are altitude of
residence and air temperature. Specifically, SIDS incidence
increases as the altitude of residence increases and
the postnatal age of the infant at death decreased with
altitude from a mean of 17 weeks at 300 meters to nine
weeks at 1200 meters (Getts & Hill, 1982).
SIDS exhibits a

5
seasonal variation being more apt to occur during the
colder winter months. This has been shown in countries
north of the equator peaking between November and March and
countries south of the equator peaking between May and
September (Moore, 1981; Peterson, 1980; Taylor, 1982).
Moreover, a spectral analysis of the SIDS incidence in
Los Angeles County from 1974-79 confirmed this seasonal
peak of SIDS in addition to revealing the coincidence of
decreased air temperature with increases in frequency of
SIDS mortality within this period (Hoppenbrouwers &
Hodgman, 1982).
On the basis of epidemiologic studies, several groups
of infants are believed to be at enhanced risk for SIDS and
have been most extensively tested by physiologic studies.
They are 1) preterm (i.e., low birth weight) infants with
histories of apnea (i.e., cessation of breathing) and
bradycardia; 2) infants who have had a near-miss or aborted
episode of SIDS, including apnea, cyanosis or pallor and
unresponsiveness that responds to intervention of
stimulation or resuscitation; and 3) siblings of SIDS
infants. There has been some controversy regarding the use
of terms like "near-miss" or "aborted" SIDS because they
imply with some certainty that the infant almost died when
in fact there is no way of knowing whether or not the
infant would have spontaneously resumed breathing without

6
assistance (Brooks, 1982). Therefore, the term "Apnea of
Infancy" or AT, as suggested by Brooks (1982) , will be
adopted and used throughout to describe "an unexplained
cessation of breathing for 20 seconds or longer, or a
shorter respiratory pause associated with bradycardia,
cyanosis or pallor." Clearly this term describes a
clinical syndrome of respiratory abnormality similar to the
sleep apnea of adults or the apnea of prematurity.
Pathology
Report of excessive frequency of prolonged apneic
episodes during sleep, in infants who subsequent to their
examination died of SIDS, promoted study of the hypothesis
that SIDS victims were experiencing recurrent periods of
apnea-associated hypoxia prior to their death
(Steinschneider, 1972). Among the plethora of hypotheses
about the cause of SIDS, most investigatory work in the
past decade has centered on the apnea hypothesis (Valdes-
Dapena, 1980). A number of post-mortem studies were
published shortly after Steinschneider's report which
supported the hypothesis that the SIDS is the terminal
event in infants who were suffering from a prolonged
hypoxic process, which possibly resulted from anomalies in
ventilatory control (e.g., Naeye, 1973; 1976). Based on
this hypothesis, one would anticipate anomalies in the anatomic
sites of ventilatory control (brain stem and carotid body),

7
as well as the organs which are usually affected by chronic
hypoxia (the brain, pulmonary vasculature, liver, adrenal
glands and periadrenal brown fat).
The main pathological findings have demonstrated that
SIDS victims differed from healthy infants who were victims
of accidental injury in that SIDS infants had hypertrophic
right ventricles in direct proportion to hypertrophy and
hyperplasia of pulmonary arteriolar smooth muscle,
increased volume of chromaffin cells in the adrenal
medulla, retention of periadrenal brown fat and presence of
hepatic extramedullary hematopoiesis (Naeye, 1973; Naeye,
1976; Naeye, Whalen, Ryser & Fisher, 1976). Moreover, upon
histologic analysis of SIDS cases, Naeye (1976) found
increases in brain stem astroglial density, particularly in
the midline area of the watershed zone of brain stem
circulation. Other recent findings of increased glomic
volume of the carotid body in those SIDS victims with
evidence cf hypertrophy of the pulmonary vasculature have
been reported (Naeye, Fisher, Ryser & Whalen, 1976) and
reduction of myelinated fibers in the cervical vagus nerve
in some SIDS victims (Sachis, Armstrong, Becker & Bryan, 1981).
The finding of increased muscularity and hyperplasia
of the pulmonary arteries has been independently described
in SIDS victims (Shannon, 1980; Valdes-Dapena, Gillane,
Cassady, Catherman & Ross, 1980; Williams, Vawter &

8
Reid, 1979). These anatomic changes also have been induced
in animals exposed to repeated episodes of hypoxia
(Rabinovitch, Gamble, Nadas, Meittinen & Reid, 1979) and
are typical of children who die after prolonged exposure to
the hypoxic conditions of high altitude (Naeye, 1973). The
increased hypertrophy of the right ventricle may be a result
of the increased pulmonary arteriolar resistance, which forces
the heart to pump against a narrower opening. Brown fat
and hepatic hematopoiesis normally disappears within the
first year of life but its disappearance is delayed in SIDS
infants, which is a common finding also noted in autopsy of
hypoxic-children residing in high altitude (Naeye et al.,
1976). The astroglial proliferation in the brain stem has
affected areas such as the nucleus and tractus of
solitarius, nucleus ambiguous, raphe nucleus and dorsal
vagal nucleus, which are critical in the autonomic control
of the cardiovascular and respiratory systems. Similar and
more extensive CNS damage has been observed in post-mortem
examination of preterm infants with chronic hypoxia due to
respiratory distress syndrome (Brand & Bignami, 1969). In
addition, perinatal asphyxia for 10-15 minutes in monkeys
resulted in a pattern of brain damage which always
involved midline brain stem structures (Myers, 1972).
'Takasbima, Armstrong, Becker and Bryan (1978) have
postulated that, periods of cerebral ischemia may accompany
the prolonged periods of apnea and hypoxemia in SIDS

9
infants, which in turn ¡nay result in the observed medullary
lesions and consequent astroglial proliferation. For this
to occur the medullary lesions would have to result from
discrete ischemia to brain stem otherwise more widespread
damage would be observed.
In sum, a case has been made for the etiology of SIDS,
whereby an infant acutely or chronically exposed to
hypoxic conditions may develop abnormal cardiorespiratory
control due to lesions in the critical brain stem areas.
The resultant prolonged periods of apnea or hypoventilation
could lead to further hypoxia and establish a positive
feedback loop of increasing pathologic tissue damage or at
least impaired function in regions of cardiorespiratory
control. Eventually, perhaps when challenged, the
cardiorespiratory regulatory mechanisms can no longer
compensate for the hypoxemic conditions and respiratory
failure results.
Cardiorespiratory Physiology
Since post-mortem studies can only provide clues to
the etiology of SIDS, physiologic studies have been
performed on living infants. What pre-mortem data on SIDS
victims are available were derived from studies of infants
who have died of SIDS subsequent to physiologic recording.
Other than this chance occurrence, there obviously can be
no systematic study of the physiologic responses of these
infants. However, investigators have studied infants

10
thought to be at risk for SIDS: infants with AI and
siblings of SIDS victims. They have typically been
contrasted with age-matched, healthy infants. It should
be noted that most studies of SIDS examine the target
physiological response longitudinally over the infant's
development. This is entirely appropriate because the risk
for SIDS seems to be a developmental phenomenon in which
the infant is at increasing risk from birth to 2-4 months
and then with maturation risk declines.
Control of Cardiac Function
Investigations of heart rate (HR) and HR variability
patterns have been performed to determine whether
manifestations of malfunctioning autonomic control
mechanisms could be detected in at-risk infants. During
maternal labor, fetal HR recordings of infants who
subsequently were SIDS victims revealed no differences from
control infants (Hoppenbrouwers, Zanini & Hodgman, 1979).
Siblings of SIDS infants, however, displayed greater fetal
HR variability and more frequent bradycardia than controls
but no difference in overall fetal HR levels when monitored
in-utero while their mothers slept (Hoppenbrouwers,
Hodgman, Harper & Sterman, 1981). The authors suggested
that the increased variability and bradycardic changes in
the siblings may reflect an increased level of intrauterine
hypoxic challenge in the siblings, whereas the effects of
hypoxia may have been concealed by the birthing process in

11
the SIDS infants. The results in the siblings, however,
should be viewed with caution since they are confounded by
maternal sleep disturbance, which was higher in mothers of
siblings of SIDS than control mothers.
The course of maturation of HR and HR variability for
healthy infants has been documented over the first six
months of life (Harper, Leake, Hodgman & Hoppenbrouwers,
1982). For HR levels, there is an increase in HR from
birth to one month after which rate declines steadily. In
contrast, HR was faster in both siblings of SIDS and in Al
infants during sleep, displaying a sharper rise and slower
decline in the first four months of life (Harper, Leake,
Hoppenbrouwers, Sterman, McGinty & Hodgman, 1978; Harper et
al., 1982; Leistner, Haddad, Epstein, Lai, Epstein, Eng &
Mellins, 1980). These differences were present only in
quiet sleep (QS) state (Harper et al., 1982); whereas
others observed these differences in both QS and active
sleep (AS) state (Leistner et al., 1980). For HR
variability in normals, sleep state had a more marked effect.
Awake variability was greater than asleep; during sleep, QS
was less varible than AS (Harper et al. , 1982). Moreover,
during QS, HR variability declined rapidly up to one
month and then increased rapidly between one and three
months after which it became stable. In AS a steady
decline in variability was observed over the first six

12
months. In contrast, awake HR variability increased up to
two months and then rapidly declined between two and four
months after which it stabilized.
No difference has been observed between normals and
siblings of SIDS in the maturation of HR variability during
sleep up to six months of age, although risk infants
displayed reduced variability in the first two months when
awake (Harper et al., 1982). On the other hand, others
found less HR variability in AI infants than in healthy
controls during AS and QS between two and four months
(Leistner et al., 1980). Harper and colleagues (1978) have
attributed the increased HR levels in the at-risk infants
to a reduction in vagal tone. Leistner and colleagues
(1980) argue, instead, that the elevated HR levels are due
to enhanced sympathoadrenal drive. This is based on the
histologic evidence in SIDS victims of hypertrophy of the
chromaffin cell layer of the adrenal medulla, which may
indicate increased norepinephrine synthesis. Further
support of the latter hypothesis was derived from evidence
of a shorter Q-T electrocardiographic interval in AI
infants than controls over the first four months of life
regardless of state, which suggested sympathetic or adrenal
mediated enhanced-conduction (Haddad, Epstein, Epstein,
Mazza, Mellins & Krongrad, 1979). Although sympathoadrenal
mediation may be a possibility for the shortened Q-T

13
interval, ventricular repolarization may also be affected
by decreased vagal tone (Prystowsky, Jackman, Rinkenberger,
Heger & Zipes, 1981). Regardless of the autonomic
mediation, it still remains possible that the elevated HR
levels in the at-risk infants represent a compensatory
response to hypoxia; the increased cardiac output could act
to compensate for the oxygen deficiency.
Control of Respiratory Function
Respiratory patterns. The respiratory rates (RR) of
siblings of SIDS have been measured throughout the night
across the first six months of age (Hoppenbrouwers, Jenson,
Hodgman, Harper & Sterman, 1980). In both normals and
siblings, RR declined with increasing age. However,
regardless of sleep state, in each age group the RR rate of
the SIDS siblings was faster than the control infants, which
was interpreted as a compensatory response to hypoxia. Both
groups of infants displayed a developmental trend in RR
pattern throughout the night. In both AS and transitional
sleep (TS), normals and siblings displayed a U-shaped RR
pattern throughout, the night at all ages. In QS this
U-shaped RR pattern did not appear in normals until three
months. Siblings, however, displayed this U-shaped pattern
prematurely at one month during QS. The authors posit that
this was due to a transient acceleration of maturation of
the nightly RR pattern in the siblings. Respiratory

14
variability decreased with increasing age in all infants
although siblings of SIDS displayed enhanced variability
in QS at three months.
In a parallel study of AI infants over the first six
months, the data do not entirely replicate these findings
(Hodgman, Hoppenbrouwers, Geidel, Hadeed, Sterman, Harper &
McGinty, 1982). There was no difference in RR or
respiratory variability between AI infants and controls,
although the decrease in RR with increasing age was
replicated. The only other longitudinal study of AI
infants observed faster RR and reduced tidal volume in the
AI infants at only one and two months. At three months no
differences were observed and at four months AI infants' RR
was less than controls (Haddad, Leistner, Lai & Mellins,
1981).
In studies on preterm infants who were presumably
hypoxemic, faster HR and RR than term infants has been
observed throughout the first six postnatal months, with
maximal differences between 2.5 and 3.5 months, the peak
period of SIDS vulnerability (Katona, Frasz & Egbert,
1980). It was concluded that since preterm HR and RR
maturational patterns were not delayed by the mean
difference in post-conceptional age, that both conceptional
age and postnatal experience were factors in the
development. Therefore, relative to healthy controls all
three SIDS risk groups displayed typical patterns of faster

15
RR over the first six months of life, which coaid be
related to maturational differences and/or compensatory
responses to hypoxemia.
Frequency of apnea. There are four types of apnea:
central, periodic, obstructive and mixed. Central apnea is
defined as a cessation of respiration with no chest or air
movement: periodic apnea is also of central origin except
that the apnea appears in constellations of short pauses
usually greater than three seconds in duration and repeated
three or more times with interapnea intervals of 20 seconds
or less; obstructive apnea is a disruption in ventilation
consisting of a lack of air flow but continued respiratory
movements; mixed apnea is initially a central apnea but
towards the end of the apneic episode some chest movements
appear with no air flow indicating some obstruction (Kelly
& Shannon, 1982).
Apnea with duration less than nine seconds is a common
finding in healthy infants over the first six months of
life (Hoppenbrouwers, Hodgman, Arakawa, Harper & Sterman,
1980). In this study short apnea (2-5 seconds) was more
abundant than longer apnea (6-9 seconds) and the incidence
of short apnea increased with age while longer apnea
incidence decreased with age. It was speculated that this
increase with age could probably be accounted for by the
same mechanism which was responsible for the concomitant

16
increase in interbreath interval. Trie frequency of apnea
longer than ten seconds was common only in the first week
when incidence of all apnea types was greatest. Otherwise,
prolonged apneic episodes were infrequent. Additionally,
sleep state affected apnea frequency, with more apnea
observed in AS than QS. When apnea does occur it has been
found to be of greater duration in QS than AS (Mellins &
Haddad, 1983).
There are conflicting data on whether the incidence of
apnea is enhanced in at-risk infants. In the fourteen
years since Steinschneider (1972) reported his findings of
prolonged and excessive apnea in infants who subsequently
died of SIDS, a number of studies have also reported
increased incidence of apnea in infants at risk for SIDS.
In another study, Steinschneider (1977) observed longer and
more frequent central apnea as well as increased frequency
of periodic apnea in infants with AI. However, apnea
shorter than 20 seconds was not examined. An increase in
frequency and duration of periodic apnea in both siblings of
SIDS and AI infants has also been reported (Kelly &
Shannon, 1979; Kelly, Walker, Cahen & Shannon, 1980).
In contrast, Guilleminault and colleagues have observed
no significant differences between AI infants and controls in
incidence of periodic apnea or in central apnea greater
than ten seconds (Guilleminault, Ariagno, Korobkin, Nagel,

17
Baldwin, Coons & Owen, 1979; Guilleminault et al., 1981).
However, they did report an increase in short (i.e., three
to ten seconds) mixed and obstructive apnea frequency in AI
infants from three weeks to four months of age.
Unfortunately, they failed to report or analyze data on
central apneas less than ten seconds. They report the
means of their 1979 study for central apneas from three to
ten seconds in a more recent publication but failed again
to analyse them (Guilleminault, Souquet, Ariagno, Korobkin
& Simmons, 1984). Inspection of these means suggested that
these short central apneas were more frequent than the
combined mixed and obstructive frequencies of similar
duration. In addition, it appeared that the AI infants
had a greater incidence of short central apneas at three
and six weeks of age but not at twelve weeks.
Additional contradictions were evident in other
research. Greater incidence of short apneas in subsequent
siblings of SIDS than controls was observed only when in AS
at two weeks of age (Kelly, Twanmoh & Shannon, 1982). No
difference in these infants was observed at eight and
fourteen weeks of age. Some researchers report no
difference in the frequency or duration of respiratory
pauses in AI and control infants (Mellins & Haddad, 1983) .
Conversely, other researchers have found a reduced
incidence of central apnea in siblings of SIDS than in

L 8
controls in AS and QS (Hoppenbrouwers, Hodgman, McGinty,
Harper & Sterman, 1980). In this study, the incidence of
mixed and obstructive apnea and central apnea greater than
ten seconds was too low to analyse. However, it was found
that from one week to six months of age, differences
between infant groups were greatest for short apneas of two
to five seconds in QS; whereas for longer apneas of six to
nine seconds differences were greatest in AS and TS. In
another study, using AI and normal infants from one week to
six months of age, this same laboratory reported no
differences in periodic, mixed, obstructive or central
apnea frequency (Hodgman et al., 1982). However, although
the trends were nonsignificant, AI infants did display
lower frequency of short and longer central apnea in this
study.
In summary, both the Shannon and Guilleminault
research groups report excessive incidence of apnea in
infants at risk for SIDS, although they don't agree on
which type of apnea is significant. Shannon claims that
its periodic apnea that is excessive and Guilleminault
emphasizes mixed and obstructive apneas. Alternatively,
the Haddad group reports no difference in apnea frequency
and the Hoppenbrouwers group found frequency reductions in
central apneic episodes less than ten seconds in duration.
Guilleminault and colleagues have recently emphasized that

19
because of differences in study design, definition of AI,
subject age, time and duration of study, recorded
measurements, analysis of respiratory events, and setting
of session, it is difficult to compare between laboratory
studies (Guilleminault, Ariagno, Korobkin, Coons, Owen-
Boeddiker & Baldwin, 1981). The lack of standardization
may therefore be responsible for the differences or lack
thereof in apnea frequency between studies. It seems clear
from the review above that there has been little
inter laboratory replication and although there may be
differences one way or another in the frequency of a
certain type of apnea, these are restricted to the
particular circumstances of measurement. Hence, the role
of apnea in SIDS remains equivocal.
The finding of an increase or decrease in apnea
incidence has been posited to relate to defective
respiratory control interacting with a myriad of potential
factors such as sleep state and CNS maturation (Shannon &
Kelly, 1982). The lack of consistent replicable finding
suggests that the apnea is of varied origin and probably
constitutes only one of the many triggering stimuli in SIDS
infants. Based upon the reported rates of apnea in the
above studies, it can be calculated that the risk groups
spent approximately only one minute per hour in apnea
(McGinty & Sterman, 1980), At this level the frequency of

20
apnea can hardly be considered a significant factor in the
induction of hypoxic conditions in risk infants. Apnea may
only represent the final common pathway in a series of
events preceding death.
Control of ventilation. There is evidence that the
mechanisms which control breathing begin to mature and
respond to respiratory stimuli such as level of carbon
dioxide and metabolic acidosis during fetal development.
For a review of the ontogeny of fetal respiratory
development see Maloney and Bowes (1983). The fetus stops
initiating breathing when the partial pressure of arterial
oxygen (PaC^) is lowered to about 15 mmHg (Wilds, 1978).
The adult, however, increases ventilation when PaC^ is
lowered below 50 mmHg (Dejours, 1963). The adult's
hyperpnea or ventilatory increase is sustained until
normoxic levels are achieved. The newborns' response to a
5% decrement in inspired oxygen differs from that of both
the fetus and the adult (Rigatto, 1977). In the newborn,
hypoxia induces a biphasic compensatory response: an
initial ventilatory increase within the first two minutes
followed by a decrement to and below baseline ventilatory
levels. The newly-born preterm infant at 33 and 37 weeks
gestation responds the same as the newborn (Rigatto,
1982). Therefore, gestational age does not seem to be a
factor. However, postnatal maturation tray play a role,

21
since the initial ventilatory increase in the preterm
becomes larger with postnatal age and the adult-like
sustained hyperpnea to hypoxia begins about one month after
birth (Rigatto, 1982). Full term infants, on the other
hand, begin sustained hypoxia-induced hyperpnea at about
two weeks of postnatal age (Rigatto, 1977). The
progressive increase in ventilation seen in the preterm
infants suggests that maturational changes occurred in
either respiratory pump mechanics or peripheral
chemoreception, or both.
The initial increase in ventilation has been
attributed to stimulation of the peripheral chemoreceptors
because of the rapidity of the ventilatory response to
hypoxia onset (Rigatto, 1982). The subsequent fall in
ventilation has been postulated to be due to adjustments in
pulmonary mechanics (e.g., lung compliance decrease),
central depression of respiratory drive, and immature
interaction between carotid sinus chemoreceptive afferent
feedback and the central respiratory mechanisms (Haddad,
Schaeffer & Bazzy, 1983). Hence, it was suggested that in
infants at-risk for SIDS hypoxic depression in an
immature CNS could induce an imbalance of excitatory and
inhibitory influences, which may supercede the
chemoreceptor effect on breathing. This could result in a
state of chronic hypoventilation or be manifested by a

22
noncompensatory decrement in ventilation to an acute
hypoxic challenge.
Current data suggest that, when compared with adults,
infants have a reduced PaC^, indicating that infants are
normally exposed to a hypoxic challenge (Brooks, Schluete,
Nabelet & Tooley, 1978). More systematic study of this
hypoxic stimulus revealed that periodic apnea could be
induced in preterm infants when subjecting them to mild
hypoxia (Rigatto, 1982). After prolonged exposure,
however, incidence of periodic apnea decreased and central
apnea increased during the decreased ventilation of the
secondary limb of the biphasic response. This suggested
that hypoxia may be an important triggering stimulus for
apnea when the respiratory system is depressed in immature
nervous systems. A similar finding was observed in AI
infants exposed to a 3% decrease in oxygen; most notably,
however, these at-risk infants displayed greater incidence
of periodic apnea and central apnea to this hypoxic
stimulus than did the control infants (Brady, Ariagno,
Watts, Goldman & Dumpit, 1978).
There have been numerous reports of periods of
prolonged apnea in infants at-risk for SIDS (Shannon, Kelly
& O'Connell, 1977; Steinschneider, 1977; Kaun, Blum,
Engleman & Waterschoot, 1982). Increased durations of
apnea in AI infants have been associated with corresponding

23
greater reductions in the transcutaneous partial pressure
of oxygen (tcPC^) , which is a noninvasive measure of PaC>2
(Kaun et al., 1982). The prolonged duration of apnea in
the AI infants could be related to a higher peripheral or
central chemoreceptor response threshold to hypoxia or
hypercapnia. That is, during apnea greater hypoxic and/or
hypercapnic levels are accumulated before respiration is
initiated resulting in prolonged apneic episodes.
Therefore, it would be expected that infants with this
elevated threshold would be sluggish to respond to a
hypoxic or hypercapnic challenge.
Support for this suggestion comes from studies of
ventilatory response to hypoxic and hypercapnic stimuli in
at-risk infants. Impaired ventilatory increases to a 5%
reduction in inspired oxygen (i.e., 15% oxygen) were
observed in AI infants (Hunt, McCulloch & Brouillette,
1981). Also, blunted ventilatory responses and a more
pronounced subsequent depression of ventilation were
observed in AI infants to hypoxic challenges of 15% oxygen
compared to controls (Haidmayer, Kurz, Kenner, Wurm & Pfeiffer,
1982). Concomitant with the depression in ventilation,
were more pronounced and rapid declines in tcPC^ values in
the AI infants. Moreover, other studies have found that AI
infants displayed less ventilatory response to hypercapnic
challenges of 4-5% carbon dioxide than controls (Ha)dmayer

2 4
et al., 1982; Hunt et al., 1981; Marotta, Fort, Mondestin,
Hiatt & Hegyi, 1984 ; Shannon et al. , 1977). Therefore,
infants at risk for SIDS have reduced hypoxic and
hypercapnic ventilatory responses, suggesting reduced
sensitivity to chemoreceptive challenge.
In summary, the majority of evidence from
cardiorespiratory physiologic investigations has revealed
that infants at-risk for SIDS display elevated HR levels
and reduced HR variability; RR is also elevated and these
infants exhibit a transient acceleration in the
developmental pattern of RR while sleeping. There are
conflicting data on which type of apnea is present in these
infants and whether the frequency of apnea during sleep is
excessive or diminished. However, when apnea does occur,
the duration is more prolonged and is associated with
corresponding decrement in PaC>2. When breathing gas
mixtures of decreased oxygen content or increased carbon
dioxide content, at-risk infants initially displayed a
sluggish ventilatory increase and with prolonged exposure,
subsequently exhibited more pronounced depression of
ventilation.
Sleep and Breathing
The finding that SIDS infants are typically found dead
in the early morning hours, when they were presumed to be
asleep, has led some to speculate that sleep precipitates

25
the events leading to SIDS (McGinty & Stennan, 1980) . There
are rapid maturational changes in sleep and respiratory
physiology in the first six months of life. (See Berg and
Berg (1979) for a review of the ontogeny of sleep,
cardiorespiratory physiology and sensory function in
infancy.) During this time, the sleep states become well
defined and become more organized within the sleeping
periods. There is a consolidation of sleep and waking into
longer and longer uninterrupted periods. By four to six
months, these become coupled so that the longest of the
sleeping periods follows the longest of the waking periods
(Coons & Guilleminault, 1982). By three months stereotypic
elements of adult electroencephalograpbic (EEG) sleep
patterns can be used to discriminate between AS and QS and
stages within QS (Coons & Guileminault, 1982). For a
recent comprehensive review of the neurophysiology of sleep
the reader may refer to McGinty and Beahm (1984).
In QS breathing is very regular and inspiration is
prolonged. It has been suggested that during QS the
forebrain and the brain stem are uncoupled (Harper, 1983).
However, slow-rhythmic descending discharges from regions
of the forebrain have been documented, which may be pacing
the lower brain stem structures in QS (Harper & Sieck,
1980). Other forebrain areas have been implicated in the
gating of QS onset (Harper, 1983). The influence of the

26
forebrain in the ontogeny of QS may be evidenced by the
observation of the emergence of a dramatic postnatal
increase in the proportion of total sleep time spent in QS
associated with concomitant increases in postnatal
forebrain development (Parmalee, Stern & Harris, 1972).
During AS, breathing is extremely variable and RR is
increased; desynchronous descending influences of the
forebrain and other areas on brain stem respiratory and
motor neurons affect rhythm and rate of breathing and also
diminish muscle tone (Schulte, Albani, Schmizer & Bentele,
1982). Phasic bursts of rapid eye movement and twitches in
the limbs and peripheral musculature occur. The rib cage
loses stability and during inspiration paradoxical
deflation of the rib cage may be observed due to depression
of intercostal and abdominal muscle activity (Henderson-
Smart & Read, 1978). This might indicate that AS
represents the state of greatest vulnerability to apnea and
hypoxia. However, other observations indicate that AS may
be associated with improved breathing. For example, oxygen
level in the blood of infants is improved in AS compared to
QS (Brooks et al., 1978).
There is evidence accumulating that the state of
greatest vulnerability for apneic episode may be QS. When
kittens were chronically exposed to hypoxic conditions, 28%

27
of them failed to compensate in QS and exhibited
hypoventilation and prolonged apnea, in some cases
progressing to death; in AS the hypoxemic conditions in
these animals were transiently reversed with faster RR and
decreased incidence of apnea (McGinty & Sterman, 1980).
This evidence suggests that the conflicting reports of
apnea incidence in the at-risk infants may reflect
differential sampling of infants who may or may not be
capable of producing compensatory cardiorespiratory
responses. In support of this possibility, it may be
recalled that the compensatory pattern of elevated HR and
RR with reduced incidence of apnea (which was more
pronounced in QS) was observed in siblings of SIDS who, by
definition, had had a healthy neonatal course with no
documented prolonged apnea attack (Hoppenbrouwers et al.,
1980) ; whereas no compensatory pattern was observed in AI
infants who had a prior history of prolonged apnea (Hodgman
et al., 1982). This suggests that the siblings of SIDS may
differ from the AI infants in the ability to compensate
with increased cardiac output and ventilation to a hypoxic
challenge; whereas the AI infants having previously
demonstrated some pathophysiologic disorder by virtue of
their prolonged apneic attack, may no longer be capable of
producing compensatory cardiorespiratory responses.

28
Other evidence suggests some alteration in the
maturation of sleep physiology in at-risk infants.
Siblings of SIDS victims display an accelerated maturation
of 12-15 Hz EEG activity in QS and 4-7 Hz EEG activity in
AS between one and two months of age (Sterman, Harper,
Hoppenbrouwers, McGinty & Hodgman, 1979). It was suggested
that an abnormal stimulus such as hypoxia may have caused
this transient acceleration in EEG development. It has
been speculated that an accelerated maturation of the
forebrain inhibitory influences on medullary respiratory
centers may be present in these infants (Hoppenbrouwers &
Hodgman, 1982). Other evidence indicates that, rather than
being accelerated, the sleep state development in at-risk
infants is delayed. The percent time in QS was less in AI
infants than in controls at three and four months of age,
suggesting a maturational delay or disturbance in the AI
infants (Haddad, Walsh, Leistner, Grodin & Mellins, 1981).
Analysis of the temporal sequencing of sleep-wake epochs in
normal infants and siblings of SIDS, revealed disturbed
patterns of state organization at two to three months of
age (Harper, Frostig, Taube, Hoppenbrouwers & Hodgman,
1983). Siblings were observed to have longer intervals
between periods of wakefulness, a shorter mean QS cycle
period and an increase in the frequency of stare
transitions. The prolonged periods of uninterrupted sleep

29
may reflect an inability to arouse from sleep (Harper et
al., 1983). Hunt (1981) observed that AI infants also
showed a tendency to remain asleep in the face of hypoxic
challenge; whereas control infants displayed an increased
frequency of arousal from sleep when inspired oxygen was
progressively decreased to 15%. The authors posited that
the failure to arouse may be due to a defect in peripheral
chemoreceptor sensitivity. Others have hypothesized that
the chronic hypoxemia of SIDS, and the hypoventilation and
failure to arouse in at-risk infants may be due to tonic
overactive synthesis of endorphins, which have previously
been associated with the CNS suppression of respiration and
wakefulness (Kuich & Zimmerman, 1981). Some support for
this theory was observed in post-mortem examination of
met-enkephalin (an endogenous endorphin) levels in
cerebrospinal fluid, when it was found that SIDS victims
had greater concentrations than age-matched controls
(Rappaport, Kozakewich, Fenton, Mandell, Vawter & Yang,
1984). Regardless of the etiology, the failure to make
transition from sleep to waking may place these infants at
greater risk. The finding of increased incidence of apnea,
occurring antecedent to awakening in infants in the
first six months of life, suggests that arousal from sleep
may be an important mechanism in terminating apneic
episodes (Guilleminault et al., 1981).

30
Cardiorespiratory Responses to Airstream Stimulation
The evidence from pathological and physiologic
investigations indicate that infants at-risk for SIDS may
be chronically hypoxic, are sluggish to respond to hypoxic
and hypercapnic stimuli with a ventilatory increase and
are less likely to arouse from sleep during such a
challenge; also, while asleep they consistently exhibit
prolonged periods of apnea with concomitant bradycardia.
Hence one of the most striking deficits in these infants
is their higher tolerance of hypoxia and hypercapnea
before responding appropriately. This deficit is
particularly apparent from two to four months of age,
which is the period of most rapid CNS development and also
the age of greatest vulnerability for SIDS. Therefore it
may be possible to detect an infant at risk for SIDS by
eliciting apnea and thereby establishing a hypoxic or
hypercapnic challenge to which they would be expected to
respond sluggishly. The prolonged apneic episodes that are
spontaneously produced by these infants are probably
related to a more general CNS dysfunction of autonomic
depressor reflexes involving the complex interactions of
the medullary chemoreceptors with inputs from the
trigeminal, glossopharyngeal and vagus cranial nerves,
peripheral chemoreceptors, sleep control mechanisms and
cardiorespiratory regulating mechanisms.

31
Investigations in infants of the ontogeny of reflex
mechanisms which when stimulated can induce apnea are
limited to four areas of research: 1) laryngeal
chemoreceptor stimulation and gastroesophageal reflux;
2) ventilatory response to hypoxic and hypercapnic
exposure; 3) vagal-mediated reflexes in compensation to
inspiratory load; and 4) dive reflex and facial cooling
with an airstream.
In animals stimulation of the laryngeal chemoreceptors
with chemical irritants have been found to evoke prolonged
apnea (Downing & Lee, 1975). There has been a report of an
association of apnea and gastroesophageal reflux in
premature infants, which presumably induces reflex apnea by
acidic reflux stimulation of the laryngeal chemoreceptors
(Herbst, Minton & Book, 1979). However, others have found
no such relationship in premature and full-term infants, and
AI infants (Ariagno, Guilleminault, Baldwin & Owen-
Boeddiker, 1982; Walsh, Farrell, Keenan, Lucas & Kramer,
1981). Some have studied hypercapnic ventilatory
responsiveness and sustained hypoxic hyperpnea and have
observed that at some critical level of PaC>2 and PaCC>2
periodic apnea is induced and with further exposure central
apnea eventually occurs (e.g., Rigatto, 1982). However,
the role of this mechanism in initiating prolonged apnea
has not been examined. Inflation of infants lungs in order

32
to activate vagal-inhibition of respiration has revealed
that premature infants are more susceptible to respiratory
inhibition than full-term healthy infants at any age (e.g.,
Boychuk, Rigatto & Sheshia, 1977; Kirkpatrick, Olinsky,
Bryan & Bryan, 1976). Although this technique when studied
developmentally would provide information on the ontogeny
of vagal reflexes in compensation to an inspiratory load,
it is subject to confounding influences of changes in
pulmonary mechanics due to sleep state (Henderson-Smart,
1984). Nevertheless, this test is more likely to provide
information on inspiratory load compensation than
information on the development of mechanisms of prolonged
respiratory cessation. Although research on these first
three areas could be of importance to a better understanding
of apneic episodes in SIDS, apnea induction by the dive
reflex and facial cooling may have the greatest relevance.
The dive reflex is a depressor reflex which appears to
resemble the acute cardiorespiratory response observed in
infants during an apneic episode. Diving in animals like
seals elicits apnea, profound bradycardia and peripheral
vasoconstriction. These responses have the clear
physiological function of conserving oxygen by reducing
circulation to tissues of greater anaerobic capacity (e.g.,
skin, mesentary, muscle) and providing circulation to
tissues less tolerant of the effects of oxygen deprivation

33
(e.y., heart, brain). The ophthalmic division of the
trigeminal nerve, which in humans innervates the forehead,
eyelids and dorsum of the nose, was found to be primarily
responsible for completing the afferent pathway by which
dive reflexes were elicited in the duck (Anderson, 1963).
The dive reflex has also been observed in humans
(e.g., Eisner, Franklin, Van Citters & Kenney, 1966;
Kawakami, Natelson & DuBois, 1967). More recently,
elaboration of the relevant eliciting stimuli has revealed
that mere submersion of the face in cold water was
sufficient to elicit the bradycardia and peripheral
vasoconstriction typical of the dive reflex (Hurwitz &
Furedy, 1979). Furthermore, it was the combination of cold
facial stimuli and breath holding that produced greater
bradycardia and vasoconstriction than in either control
conditions of face immersion (while breathing through a
snorkel) or breath holding alone. Others have found that
the dive reflex was even more pronounced with colder water
temperatures (Furedy, Morrison, Heslegrave & Arabian, 1983;
Song, Lee, Chung & Hong, 1969). The dive reflex literature
illustrates the significance of cold facial stimulation in
effecting cardiovascular and respiratory responses.
In human infants apneic and bradycardic responses have
been elicited by nasopharyngeal suction and by blowing
oxygen over the infants' face (Brown, Ostheimer, Bell &


35
topography of the HR response from stimulus onset, which
would have provided more information on the temporal effect
of the stimulus and would have been a less biased measure.
Brief apneic episodes in response to the airstream stimulus
were observed but no difference in apnea duration existed
between the two infant groups. Unfortunately, no further
attempt was made to examine the apnea responses
quantitatively. Moreover, no systematic analysis was
performed to partial out the effects of sleep state,
postnatal age or presence or absence of apnea from their
possible effects on the cardiorespiratory responses
observed.
The proposed study, herein, was designed to establish
the normative response to facial airstream stimulation in
infants throughout the first year of life. This study has
a focus similar to that of McCubbin et al. (1977) but
utilized a more systematic design, including the
appropriate control conditions to compare the
cardiorespiratory responses. The study was designed to
assess in sleeping infants developmental changes in
cardiorespiratory responses to airstream stimulation in the
first year of life by 1) determining whether airstream
stimulation of the facial region supplied by the ophthalmic
branch of the trigeminal nerve (i.e., forehead, eyelids and
dorsum of the nose) would elicit apnea more frequently than

36
a control airstream stimulus directed to the abdomen (i.e.,
rostral to the umbilicus); 2) measuring the HR response on
a second-by-second basis and average a sufficiently large
number of homogeneous trials to determine the precise
topography of the cardiac response to airstream
stimulation; and 3) examining the effect of apnea
elicited by the airstream stimulus on the HR response
topography.
This study will be informative for the following
reasons. First, the examination of the cardiorespiratory
responses to repeated airstream stimulation will provide
information on the maturation of cardiorespiratory control
in the production of depressor reflexes in normal infants
cross-sectionally throughout the first year of life. It
will be informative to determine whether the cardio¬
respiratory responses in infants, during the period of most
rapid CNS development (i.e., two to four months of age) and
vulnerability to SIDS, differ from responses during
earlier or later postnatal age. Second, it will provide
more information on the cardiorespiratory effects of
conditions of facial cooling similar to those which exist
presently in hospital delivery protocol manipulations such
as nasopharyngeal suction and oxygen blown over the face.
Third, since the airstream stimulus is not substantially
different from that which the infant might commonly

37
experience in every day life, this study may shed some
light on the eliciting conditions of apnea and bradycardia
in normal healthy infants. Fourth, this simple reflex test
of the integrity of trigeminal-brainstem-cardiorespiratory
function could have prognostic significance for screening
of cardiorespiratory risk for SIDS. However, baseline
data in a normative sample are required initially so that
the results of this study will provide information against
which data from high risk infants may be compared.


39
Maternal and infant hospital medical records of the
newborn group were surveyed to insure that all criteria were
met. For the older groups information on the infants'
delivery and neonatal course and the maternal obstetrical
history was obtained in a questionaire during testing (see
Appendix A). Table 1, 2 and 3, respectively, shows the sex,
gestational age (GA), postnatal age (PNA) at the time of
testing, birth weight (BW), APGAR scores, race (R), maternal
age (MA), and maternal medication (MED) at delivery for
each infant in the three age groups included in this study.
Newborn subjects were recruited by obtaining permission
from their mother or father in the neonatology ward of the
J. Hillis Miller Health Center. These subjects received no
remuneration for their participation. The older subjects
were recruited through the mail. Names of infants born in
the appropriate age group were obtained through the Alachua
County Health Department. Parents were sent a general letter
describing the study and were requested to indicate whether they
were interested in participating in the study on an enclosed
stamped-postcard. Those parents responding affirmatively
were telephoned to schedule an appointment for testing. In
this conversation parents were informed about the purpose
and specifics of the study and an appointment was scheduled
at a time that would correspond to the infant's nap schedule.
These subjects received five dollars for their participation?-














CHAPTER 3
RESULTS
The factors of critical interest in this study were
the site of airstream stimulation and the age of the infant
when tested. Knowledge of these effects indicate whether
cardiorespiratory responses to stimulation of the facial
region supplied by the ophthalmic branch of the trigeminal
nerve may be unique and if so, whether it changes
throughout the first year of life. However, the behavioral
observations revealed that two other factors, sleep state
and presence of elicited-large movement, may also have
influenced the cardiorespiratory response. Before the
effects of these two latter factors on the cardio¬
respiratory response could be examined, it was necessary
to assess the infants’ sleep state pattern during testing
and to determine the extent to which large movement was
elicited by the airstream stimuli.
Sleep State Incidence and Pattern
Table 4 presents the percentage of trials in which
the sleep state of the infants were rated as QS, TS and
AS across the age groups. The most notable finding was
that the majority of data were collected while the infants
were in QS. However, the NR group tended to exhibit a
53



CO ILxJUJÜl CO I— -Xi— LU ilti -XI—1—'
2
Figure 1. Mean Sleep State Rating for the First and Last Ten Trials.
Newborn (NB), 2-4 and 8-12 month infants are contrasted; a rating of two
ras active sleep, one was transitional sleep and zero was quiet sleep.
’-jdT



















Prestimulus Levels
75
Prior to the analysis of the HR response topography
the prestimulus levels of HR were evaluated to determine
whether the post-stimulus HR response could have been
influenced by initial prestimulus levels. This was of
particular concern since it was to be expected that
prestimulus levels would co-vary with infant age (Harper et
al., 1982). Table 12 depicts the mean HR values in beats
per minute (bpm) for the second preceding stimulus onset
across the age groups during trials with and without
elicited-apnea.
The analysis of the prestimulus HR level in the trials
in which no apnea or LM were elicited used Age as the
between-subject factor and Site of stimulation as the
within-subject factor. Only the Age effect was
significant, F(2/47)=5.9, £<.01. With no significant
difference between the two younger groups, the difference
between the age groups, therefore, can be accounted for by
4
diminished prestimulus levels in the 8-12 infants.
Post-stimulus Changes
The interpretation of the post-stimulus HR response
topography is complicated by the apparent relationship
between prestimulus HR levels and the infants postnatal
age. Hence, the post-stimulus HR levels and perhaps
topography may be partially determined by the prestimulus

76
TABLE 12
Mean Prestimulus HR Levels in bpm in the Presence and Absence of
Apnea and Large Movement (LM) for Facial (F) and Abdominal (A)
Airstream Stimulation Trials
Age
Apnea
Movement
Condition
NB
2-4
8-12
Mean
No
No LM
F
117.92
122.76
110.52
117.83
A
118.72
122.17
112.11
118.26
Mean
118.32
122.47
111.32
N
13
22
15
Yes
LM
F
123.56
130.03
106.65
124.90
N
11
10
2
Yes
No LM
F
121.21
121.45
117.16
120.31
N
8
13
7

77
HR levels. This statistical problem is Known as the Law of
Initial Values as outlined by Wilder (1950). Following the
guidelines suggested by Richards (1980) for countering the
effects of prestimulus HR on postimulus HR, the assumption
of homogeneity of regression coefficients for the data
were tested over age for regression of each of the 25 post¬
stimulus seconds on the first prestimulus second. Each of
these analyses revealed that the assumption of homogeneity
of regression coefficients was not violated. Therefore,
the effects of the experimental parameters on the HR
response topography could be compared across age. However,
to control for the effects due to the Law of Initial
Values, the analysis of covariance (ANACOVA) was used^
Since the interpretation of HR change was of primary
concern, difference scores were calculated by taking the
algebraic difference between the prestimulus value and the
post-stimulus values. The correlation between prestimulus
and difference scores have typically been found lower than
the correlation between prestimulus and post-stimulus
scores (Benjamin, 1967). Therefore, the use of difference
scores, herein, provided an additional statistical
advantage. Despite lower correlations, however, difference
scores still remain subject to the Law of Initial Values.
The HR difference scores, therefore, were all analysed with
ANACOVAs using the HR value in the second preceding stimulus

78
onset as the covariate, Age as the between-subject factor
and Seconds as the within-subject factor. It should be
noted that all significant effects reported below remained
significant when the conservative Greenhouse-Geisser
approximation of degrees of freedom was used (Greenhouse &
Geisser , 1959).
HR Response to the Airstream Stimulus: Face vs. abdomen
Figures 2 and 3 depict the HR change scores at each
age in response to facial and abdominal stimulation in the
absence of apnea and LM. For both stimulation sites these
figures show that the a short monophasic acceleration to
the airstream was produced by the NBs. However, the two
older groups displayed a triphasic response to the
airstream stimuli comprised of a brief deceleration
followed by an acceleration and a subsequent deceleration.
After stimulus offset HR levels in these infants gradually
recovered toward prestimulus levels.
More specifically, ANACOVA analyses (see Appendix F)
revealed that an initial deceleration was elicited in the
older infants within one second of stimulus onset but was
not elicited in the NBs. An acceleration followed
maximizing at the third post-stimulus second in all age
groups. This acceleration, which was greater when elicited
by the facial airpuff than by the abdominal airpuff,
differentiated all the age groups such that the 8-12

SECONDS
Figure 2. Mean HR Response to Facial Airstream Stimulation When No Apnea or
Large Movement Were Elicited. The airpuff,as delivered to newborn (NB), 2-4 and
8-12 month infants, was initiated at second zero (ON) and terminated at second
ten (OFF).
VO

ON OFF
SECONDS
Figure 3. Mean HR Response to Abdominal Airstream Stimulation When No Apnea or
Large Movement Were Elicited. The airpuff, as delivered to newborn (NB), 2-4
and 8-12 month infants, was initiated at second zero (ON) and terminated at
second ten (OFF) .

81
infants displayed the fastest accelecatory recruitment
followed by the 2-4 infants and then the NBs. As in the
early deceleration, the NBs were differentiated from the
older infants by the ensuing deceleration from seconds
three to ten, which reached its nadir by the seventh post¬
stimulus second in the older infants but merely returned to
pretrial levels in the NBs. Only the 2-4 infants displayed
an effect of the stimulation site during this interval with
greater deceleration to the facial airpuff than the
abdominal airpuff. Following stimulus offset all groups
displayed a cubic trend over seconds 11 to 25, which was
unaffected by the stimulation site, appearing to decrease
slightly 5 to 10 seconds after stimulus offset followed by
a gradual return to pretrials levels.
HR Response to the Facial Airstream Stimulus: Apnea vs. no apnea
Figure 4 shows the response to facial stimulation when
apnea was present without preceding LM. By comparing these
responses with the responses in Figure 2 in which no apnea
was elicited by the facial airstream, the effect of apnea
elicitation on HR can be examined. Although the overall
topography of the response was similar with and without
apnea, when apnea was elicited, the older groups displayed
a more prolonged late deceleration which was of greater
magnitude than when no apnea was elicited. The NBs
response was unaffected by whether or not apnea was induced.

Figure 4. Mean HR Response to Facial Airstream Stimulation When Apnea Was
Elicited Without Preceding Large Movement. The airpuff, as delivered to na*om
(NB), 2-4 and 8-12 month infants, was initiated at second zero (ON) and
terminated at second ten (OFF).

33
The ANACOVAs of the overall HR response used seconds 1
to 25 as levels of the within-subject Seconds factor. Age
as the between-subject factor and presence or absence of
Apnea as the second within-subject factor® A significant
interaction between Age, Apnea and Seconds, F(48/576)=2.0,
£<.001, was found. The next analysis examined whether the
differences described by this interaction existed during
the period of acceleration across seconds 1 to 3. It did
not, since only a significant Seconds effect, F(2/48)=6.7,
o<.005, emerged. Therefore, up to this point in the
analysis the HR response was unaffected by the age of the
infant or by the presence or absence of apnea.
The age groups were differentiated by the HR response
during the late-deceleration from seconds 3 to 12, as
indicated by a significant interaction between Age, Apnea
and Seconds, F(18/216)=3.1, £<.001. The subanalysis
consisted of examining the NB group alone. Only a
significant linear trend across Seconds, F(l/6)=21.1, £<.005,
was found, indicating that there was a linear HR
decrement toward base level both in trials with apnea and
trials without apnea. Subsequent t-tests revealed that the
decrement in the NBs HR during this interval did not
proceed below prestimulus levels. When the two older groups
were contrasted a significant interaction between Age, Apnea
and Seconds, F(9/162)=4.7, £<.001, emerged. This was due

84
to changes in both magnitude and topography of the
deceleration when apnea was elicited. When the two older
groups were compared using only the non-apnea responses,
only a significant quadratic trend over Seconds,
F(l/18)=7.5, £<.05, emerged, indicating that there was no
difference between groups in the magnitude of the
deceleration when no apnea was elicited. However, on
trials when apnea was elicited, the magnitude of
deceleration was increased to a greater extent in the
8-12 infants than in the 2-4 infants, as substantiated by
a significant interaction between Age and Seconds,
F(9/162)=6.3, £<.001. Moreover, the effect of apnea was
to prolong the deceleratory response differentially in
these older infants. As can be seen in Figure 4, the
deceleration in the 2-4 group reached its nadir and began
its recovery toward pretrial levels sooner than the 8-12
infants. This was confirmed by an orthogonal trend
analysis of each age group for apnea data only, with a
significant decreasing quadratic trend over Seconds,
F(l/12)=31.6, £<.001, for the 2-4 infants and a decreasing
linear trend over Seconds, F(l/6 )=39.1, £<.001, for the
8-12 infants.
The analysis of seconds 12 to 25, which represented
the period of HR recovery, yielded a significant main effect
of Apnea, F(l/23)=7.0, p<.05, and an interaction between Age

85
and a quadratic trend over Seconds, F(2/24)=5.2, £<.05.
Neither the effects of Apnea or Seconds nor their interaction
were significant when analysing only the NB group's HR
recovery, indicating that regardless of whether apnea or no
apnea was elicited the NB HR was stable throughout this
period maintaining a mean HR level of -0.65 bpm. When
comparing the younger groups only a significant Age effect,
F(l/17)=6.2, £<.05, was found, indicating that the 2-4
group operated at a lower level than the NB group, with
mean HR bpm levels of -0.65 and -5.01 for the NB and 2-4
groups respectively. When comparing the 2-4 and 8-12
groups, however, an interaction between Age, Apnea and
quadratic trend over Seconds, F(l/18)=4.5, £<.05, emerged.
No difference was found in the recovery trends of the
trials with no apnea. When apnea did occur, however, the
enhanced late-deceleration and the resultant phase
shift between these groups resulted in interacting
decreasing and increasing quadratic trends over Seconds,
F(l/18)=7.5, £<.05. Inspection of Figure 5 reveals that by
post-stimlulus second 12 the HR response of the 2-4 group
had partially recovered, while the 8-12 group had just
reached its nadir. Hence, HR levels decreased to a mean of
-7.25 bpm for the 2-4 infants while the HR levels increased
to a mean level of -8.58 bpm for the 8-12 infants.

86
In summary, the NBs HR response when apnea was
elicited was not distinguishable from the response observed
when no apnea was elicited? The main effect of apnea on
the HR response appeared in the older groups. Apnea
resulted in an accentuated deceleration during seconds 3 to
12 post-stimulus onset reaching a maximum at seconds 7 and
12 respectively for the 2-4 and 8-12 groups. Apnea also
had an effect on the HR recovery during seconds 12 to 25 in
the older groups, exhibiting less tendency to return to
pretrial levels.
HR Response to the Facial Airstream Stimulus from Apnea Onset
The previous HR analysis used HR values scored from
stimulus onset. When apnea was present, it could occur
anywhere in the trial within the 20-second post-stimulus
observation window. Therefore, the topography of the HR
response in the presence of apnea was difficult to
interpret. This was especially the case, since the latency
to apnea onset was significantly related to age, with older
infants displaying longer apnea latencies. To obtain a
more precise determination of the effect of apnea on the
response topography, HR was aligned relative to apnea onset
rather than stimulus onset. To maintain a constant
standard of reference, these HR values, as in the above HR
statistics, were represented relative to prestimulus
level. For example, the HR value at second zero

87
represented the HR difference from prestimulus level of the
interval just preceding apnea onset.
The mean HR values observed during apnea as elicited
by the facial airpuff (uncomplicated by LM) were plotted in
Figure 5. The analysis of the HR response from apnea onset
did not examine the effect of preceding LM on the HR
topography. As can be seen in Figure 6, the effect of LM
was to substantially increase HR (relative to prestimulus
level) prior to apnea onset. Otherwise the response
topography remained similar to the response when no LM
preceded apnea. As can be seen in Figure 5, the HR
response of the NBs was not affected by the production of
apnea; whereas, in both of the older groups apnea induced
rapidly-recruiting, large-magnitude decelerations, which
began to recover to prestimulus levels when apnea
terminated. Note that the arrowhead located below each
curve in both Figures 5 and 6 corresponds to the point
where apnea ended and breathing resumed.
ANACOVA of the no LM data (Figure 5) assessed the
pre-apnea HR levels. Seconds -5 to 0 were used as levels
of the Seconds factor, Age as the between-subject factor
and the unadjusted HR value in the second preceding apnea
onset (i.e., second 0) as the covariate. There were no
differences to emerge during this period as neither the
effects of Age or Seconds nor their interaction were
significant.

H
R
r.
H
H
N
G
E
I
N
B
P
M
ONSET
SECONDS
Figure 5. Mean HR Response to Facial Airstream Stimulation When Apnea Was
Elicited Without Preceding Large Movsnent Adjusted from Apnea Onset. The
arrowhead below each of the newborn (NB), 2-4 and 8-12 month infants' plots
indicates the point of termination of apnea and resumption of breathing.
co
CD

ONSET
Figure 6. Mean HR Response to Facial Airstream Stimulation When Apnea Occurred
Following Large Movement Mjusted frctn Apnea Onset. Hie arrowhead below each
of the newborn (NB), 2-4 and 8-12 month infants' plots indicates the point of
termination of apnea and resumption of breathing.

90
Inspection of the trends in Figure 5 suggested that
there was an initial deceleration upon apnea onset in the
older groups, which was absent in the NB infants. The
analysis of this trend used seconds 0 to 9 as the period of
deceleration after apnea onset. The ANACOVA yielded a
significant Age effect, F(2/24)=4.0, £<.05, a Seconds
effect, F(9/225)=7.7, £<.001 and a two-way interaction of
these factors, F( 18/22 5)=1.9, £<.05. The subanalysis of
the most salient of these effects - the Age-by-Seconds
interaction - consisted first of assessing whether there
was a Seconds effect in the NB group alone; there was not,
F(9/63)=1.0, indicating that when apnea was elicited it had
no effect on the NBs HR response. An ANACOVA was then
performed on the two older groups. Only a significant
Seconds effect, F(9/162)=4.6, £<.001, was found.
Therefore, from the onset of the apneic period to the ninth
second of the trial, the two older groups exhibited HR
deceleration, which was of similar magnitude and which was
totally absent in the NB group. The 2-4 infants reached a
maximum deceleration of -14.46 bpm at seven seconds post¬
apnea onset; whereas the 8-12 infants decelerated to -15.91
bpm by the ninth second after apnea began.
Analysis of seconds 9 to 25 was performed to assess
the period of recovery from the apneic episode. The NB
group displayed no significant change over these seconds

91
which was not surprising since apnea seemed to have no
effect on HR previous to this point in time. When the two
older groups were analysed, however, a significant
interaction between Age and Seconds, F(16/238)=5.4, £<.001,
emerged. An orthogonal trend analysis revealed an
interaction of linear Seconds and Age, F(1/18)=16.2, £<.001,
indicating that, indeed, the HR levels in both groups
were recovering toward pretrial levels during this period
but differed in the temporal aspects of the recovery.
In summary, when considering all the analyses of the
HR response topography, the onset of the airstream stimulus
to the face evoked a monophasic acceleration in NBs and
a triphasic response in the two older groups. Only the
older age groups displayed a brief initial deceleration
and following the subsequent acceleration a deceleration
occurred proceeding well below prestimulus levels. The
NBs response was unaffected by apnea. Similarly, the
acceleratory component of the older infants' responses were
unaffected by apnea, but not so for the deceleratory
component. This latter deceleration was two to three times
larger on trials when apnea was present. Realignment of
the responses on apnea trials to correspond to apnea onset
revealed that the apnea produced an immediate monophasic
deceleration very similar to the late deceleratory
component in the stimulus aligned functions. The

92
deceleration initiated by apnea onset did not differ
between the two older groups and displayed a recruitment
speed wherein approximately 80% of the deceleration
occurred within 3-5 seconds after apnea onset.
Notes
1. The proportion of trials when infants were in AS or TS
were too small to permit evaluation of the effects of sleep
state. Therefore, the following strategy was adopted
whereby response measures collected while subjects were in
QS were compared to the entire data set collapsed across
state to determine if the contribution of the non-QS
(i.e., AS and TS) data were sufficient to cause a
significant difference in the response being examined. A
nonsignificant difference between these two data sets was
interpreted as indicating that the inclusion of the AS and
TS data were not sufficient to significantly alter the
observed trends. Henceforth, the entire data set was used
in further analyses, thereby improving the statistical
power of the analyses by increasing the sample size that
could be included. It was recognized that a nonsignificant
difference between these data sets did not rule out the
possibility that AS or TS may still have an unique effect
on the data. It should be noted, however, that the
predominance of QS in this data was a function of the fact
that infants in AS tended to awaken when stimulated with
the airstream. This coupled with the probable influence of
background maturational differences in the organization of
sleep states prevented a more in-depth investigation of the
effect of AS and TS states on the response measures.
2. In this case we were assessing the percent of elicited
apnea for those trials in which LM occurred compared to
those trials where no LM occurred. This is quite different
from asking the question, what percent of trials with apnea
was preceded by LM. The latter question will yield an
answer pertaining to the relative proportion of elicited
apneic episodes that were preceded by LM as well as
yielding the proportion of apneic episodes that were not
preceded by LM. As it happens, of the 138 apneic episodes
elicited in this study, LM preceded apnea onset on 67
(i.e., 48%) occasions and only twice did LM follow apnea
offset. Therefore, 50% of all elicited apneic-episodes
were not preceded or followed by any LM.

93
3. Few subjects contributed apnea data for both LM and no
I.M trials. Therefore, we were unable to perform a within-
subjects analysis of the effect of LM or no LM on any of
the apnea response measures. In some cases, indirect
comparisons were possible wherein data from subjects who
contributed LM were assessed separate from all subjects who
contributed no LM results.
4. To parallel the post-stimulus analyses, the prestimulus
levels of trials with apnea preceded by LM and no LM were
also examined. As can be seen in Table 12, the means were
in the same direction as the prestimulus levels exhibited
on trials with no apnea elicited. Whereas, the Age effect
for apnea trials with no preceding LM did not reach
significance, it was significant for trials when apnea was
preceded by LM, F(2/20)=3.8, £<.05.
5. The analysis of covariance (ANACOVA) removes the
effects of the Law of Initial Values by a linear regression
of the post-stimulus HR values on the prestimulus values.
Specifically, it acts to reduce the sum of squares by
covarying out the variance accounted for by the sum of the
deviation crossproduct scores relative to the prestimulus
HR. See Benjamin (1967) for a more in-depth treatment of
the use of ANACOVA to counter the effects of the Law of
Initial Values on HR data.
6. The analysis of the HR response from stimulus onset did
not include an analysis of the data when LM preceded apnea,
since the effect of LM was to result only in a large
monophasic acceleraton preceding the secondary deceleration.
Otherwise the topography of the HR response remained
unaffected by LM preceding apnea.
7. It should be noted that a follow-up analysis was
performed, since sleep state had an effect of the percent
of apnea elicited in the NB subjects. The NB HR response
when apnea was elicited during QS to the facial airpuff
was compared to the HR response of the NBs when both QS and
the effects of AS and TS were combined. No significant
effects due to State nor its interaction with Seconds were
found. This suggested that although the QS state tended
to reduce the proportion of apnea, the HR response remained
unaffected.

CHAPTER 4
DISCUSSION
The results of this study clearly demonstrate the
effects of age and site of airstream stimulation on
cardiorespiratory control. Facial cooling induced more
frequent and prolonged apnea with shorter latency than
abdominal stimulation in healthy, sleeping infants up to
one year of age. There was no change in the frequency of
apnea elicited by the abdominal airpuff over age. With
facial airstream stimulation, the overall incidence of
apnea was fairly level from newborn (NB) through 2-4 months
of age but dropped precipitously by 8-12 months. This was
particularly true in the absence of large movement (LM),
wherein infants 8-12 months of age exhibited an 85%
reduction in apnea incidence relative to the 2-4 month
infants. However, in the presence of LM, the frequency of
apnea increased with age such that infants older than two
months displayed more frequent and prolonged apneic
episodes than newborns. In each age group, LM was
associated with a greater percent of facial trials with
apnea than when no LM occurred. However, since LM was
induced on only 14% or less of the trials in the older
infants, in an absolute sense the apnea consequent to LM
94

95
represented only a small fraction of the total number of
apneic episodes elicited. In contrast, the NBs displayed
LM on about half of the trials, thereby elevating the
absolute incidence of apnea. This suggests that the
mechanisms which resulted in apnea following LM may play a
significant role in increasing apnea frequency in these
infants. When apnea was elicited by the facial airpuff,
the effect of preceding LM was to delay apnea onset
somewhat in all age groups. Regardless of whether or not
LM preceded apnea the duration was much longer in the older
infants than for the NBs.
When no apnea was elicited by the airstream stimulus
to the face or abdomen, a monophasic HR acceleration was
observed in the NBs; however, a triphasic deceleration-
acceleration-deceleration response was produced by the
older infants to airstream stimulation of either site. The
magnitude of the acceleratory component increased slightly
with age of the infant, although the peak was slightly more
prolonged in the NBs (see Figures 2 and 3). The facial
airpuff resulted in larger accelerations than the abdominal
airpuff in all age groups. The late-deceleration, unique
to the older infants, was somewhat greater for facial than
for the abdominal airpuff. The presence of apnea elicited
by the facial airpuff produced no change in the NBs cardiac
response. With the older infants, apnea resulted in a
substantially increased late-bradycardia compared to

96
the no apnea response and did not differ in magnitude
between the 2-4 and 8-12 month infants. Moreover, this
enhanced bradycardia, did not begin to recover until the
apnea was terminated (see Figure 5).
In summary, the results demonstrated an increased
propensity for facial cooling to evoke apnea differentially
over age in healthy, sleeping infants. Evaluation of
infant sleep state revealed that infants in this study were
predominantly in quiet sleep. The non-quiet sleep data did
not appear to have a substantial effect on the parameters
examined, although due to the sparcity of this data its
possible effects could not be evaluated. The infants of
different ages were differentiated based on their
cardiorespiratory responses to facial airstream
stimulation. The NBs displayed frequent apnea to the
facial airpuff but these were of short duration and had no
effect upon their cardiac response. On the other hand, the
facial airpuff elicited apnea infrequently in the 8-12
month infants, although when it occurred it was of longer
duration and accompanied by bradycardia which was
maintained for the duration of the apnea. From the point
of view of vulnerability for SIDS, the 2-4 month infants
displayed a tendency to produce what are presumably the
most potentially deleterious combination of
cardiorespiratory changes; apnea to the facial airpuff in

97
these infants occurred with elevated incidence similar to
that shown by the NBs but with prolonged duration and
concomitant large-magnitude bradycardia similar to that
shown by the 8-12 month infants. Therefore, exposure to
facial cooling by airstream stimulation of the ophthalmic
branch of the trigeminal nerve has revealed that the age
between two to four months is the age of greatest
susceptibility to respiratory inhibition and concomitant
cardiac rate depression in the maturation of
cardiorespiratory control in the first year of life.
Implication for Dive Reflex as the Mechanism
for Inducing Apnea and Bradycardia
This study was not designed to determine what central
mechanisms were involved in the production of apnea in
these infants but clearly by virtue of the enhanced
incidence of apnea to facial airstream cooling in infants
throughout the first year of life some significance must be
attached to the region of the face that is supplied by the
ophthalmic branch of the trigeminal nerve. The trigeminal
nerve mediates pain and temperature sensation of the face
and has been implicated in the mediation of afferent
impulses responsible for the dive reflex (Anderson, 1963) .
Both the apnea and bradycardia of the dive reflex are
present in decerebrate animals, implying that the central
connections between the afferent and efferent limos of the

98
reflex are in the brainstem (Anderson, 1963). Therefore
the increased propensity for cool facial stimulation to
induce apnea in infants probably reflects the preeminant
role in this study of these pathways of central neural
sensory integration which may be similar to those which
mediate the dive reflex in the adult.
The dive reflex has been presented as an adaptive
mechanism for defense of the lungs where the reflex apnea
and laryngeal closure that result serve to prevent
inhalation of the water which evoked the response
(Richardson & Peatfield, 1981) . The concomitant
cardiovascular changes involve a vagal-induced bradycardia
and sympathetically-mediated constriction of systemic
vessels. The circulatory effects of increased PaCC>2 and
decreased PaC>2 further augment the bradycardia and
peripheral vasoconstriction, which effectively acts to
redistribute blood supply toward regions such as the brain,
which are most vulnerable to oxygen deficit (Irving,
Solandt, Solandt & Fisher, 1935). For a mechanism similar
to the dive reflex to be operating, the following must be
present and functional: 1) pathways of sensory transduction
(i.e., temperature and/or tactile); 2) connections to
central pathways that are capable of interrupting ongoing
respiration; 3) connections to central pathways that are
capable of overriding the action of the chemoreceptors on

99
central regions of inspiratory drive so as to maintain
inhibition; 4) mechanisms to allow inspiratory drive to be
disinhibited when stimulation ceases so that breathing can
be resumed; and 5) mechanisms of cardiovascular
compensation such as bradycardia and redistribution of
blood supply away from the periphery and toward the heart
and brain, which act to conserve oxygen and thereby permit
more prolonged apneic episodes. The present results
provide information relevant to each of these points and
each will be discussed in turn.
An examination of the infants' cardiorespiratory
responses to the facial airstream reflects a response of
increasing maturity and adaptability with age. When no
preceding LM occurred the facial airstream evoked frequent
apnea in the NBs and 2-4 month infants demonstrating the
integrity of the trigeminal nerve containing the afferent
limb of the reflex in these infants as well as the ability
of central mechanisms to inhibit respiration. The
mechanism of cessation of breathing to airpuff stimulation
in this study was central in origin and not due to
obstruction of the airways. This was the case because
evidence of obstruction would present itself during
recording as a cessation of nasal respiration but a
continuation of diaphragm movements. This did not occur in
any of the apneic episodes induced in any of the subjects

100
who were monitored with both measures of nasal airflow and
diaphragm movement.
The much greater incidence of LM in the NBs than the
older subjects to the airstream at either stimulation site
probably reflects a greater reactivity in the NBs to the
stimulus rather than indicating any differences in the
maturity of sensory transduction. Active sleep was a state
wherein the airstream stimulation was observed to be more
likely than during QS to induce wakefulness in all infant
groups. This suggests that the threshold for induction of
the response may be altered in AS rendering the infants
more susceptible to respiratory inhibition in QS. Arousal
from sleep during a stimulus presentation was usually
associated with struggling and vigorous directed movements
away from the airstream. Thus arousal from sleep may be
one possible mechanism of preventing or terminating a
reflex apnea. The fact that the NBs were in AS more than
the older infants may also have contributed to their
increased motor reactivity to the airstream.
The shorter latency to apnea onset in the NBs reflects
the greater ease with which breathing can be interrupted
during sleep in these infants. More facial cooling was
required to elicit apnea in the older infants as reflected
by their longer response latencies. Therefore, the
threshold for facial cooling to result in apnea may be

101
higher in these older infants. If this was the case then
the markedly reduced incidence of apnea in the 8-12 month
infants relative to the 2-4 month infants indicates that
the mechanisms driving respiration during sleep in the
oldest infants were more resistant to interruption and
probably more mature. However, when the reflex apnea was
induced in the 8-12 month infants the latency to onset was
about the same as the 2-4 month infants and the response
appeared to be of similar magnitude. If the dive reflex is
actually an adaptive response to protect the lungs from
potential irritants then the infant is clearly subject to
competing interests since the cessation of breathing in the
extreme is also maladaptive. Therefore, it is reasonable
to expect that, as the infant matures and is more capable
of compensating for changes in the composition of inspired
air, the threshold for facial cooling to induce a cessation
of breathing would be elevated.
The NBs demonstrated brief durations of apnea, which
were likely to cease prior to the offset of the facial
airstream stimulus; whereas, the older infants displayed
more prolonged apneic episodes which typically began prior
to and terminated after stimulus offset. The inability of
the stimulus to maintain the induced-apnea for its entire
duration in the NBs suggests that the central mechanisms
which mediated the reflex apnea were not able to sustain

102
the inhibition on respiratory drive. It could indicate
that the arterial and medullary chemoreceptor mechanisms in
the NB have a more substantial influence than those
inducing the apnea at this age. However, the maintenance
of apnea in the older infants for the remaining seconds of
the airpuff suggests more powerful central inhibition of
respiratory drive in these infants. There are several
possible explanations for this finding. One explanation
may be that the increasing duration of apnea with age could
reflect the increasing functional residual capacity with
maturity so that older infants due to their better
buffering systems can tolerate greater hypoxic and
hypercapnic changes. The persistance of apnea after
stimulus offset in the older subjects differs from findings
in adults where the apnea and bradycardia of the dive
reflex is time-locked to the eliciting stimulus (Hurwitz &
Furedy, 1979). However, the dive reflex has only been
recorded in adults while they were awake. Therefore, aside
from the disparity in age and the implied difference in
central neural cardiorespiratory maturation, the
persistance of apnea beyond stimulus offset in infants
could reflect the relative difficulty of reinitiating
breathing during sleep once breathing has been interrupted.
McCubbin and colleagues (1977) reported heart rate
decelerations to facial airstream stimuli occurred
irrespective of whether apnea occurred or not. On the

103
other hand, a clear distinction was observed in this study
between the cardiac response when no apnea was elicited
compared with the large-magnitude changes coincident with
elicited-apnea. In the older infants the facial airstream
elicited large-magnitude bradycardic responses with rapid
onset, which were maintained for the duration of the
concomitant apneic episode. These response-topography
features of rapid recruitment, large magnitude and
sustained response until breathing is resumed were
characteristic of the adult bradycardic response to
simulated dives of apneic-face immersion in water (Hurwitz
& Furedy, 1979). Since peripheral blood flow and oxygen
consumption were not measured in this study there was no
way to determine whether the infants' cardiovascular
changes during apnea resembled those oxygen-conserving
adjustments associated with the dive reflex (i.e., cardiac
output reduction and redistribution of blood flow toward
those regions which cannot sustain anaerobic metabolism for
prolonged durations). Nevertheless, the bradycardia in
these infants probably utilized similar vagal efferent
pathways. The NBs lack of alteration in cardiac rate
during apnea suggests that the mechanisms governing cardiac
rate change at this age may not be physiologically coupled
with brief phasic changes in respiration, although there is
evidence to indicate that this is not so for tonic

104
respiratory activity in newborns (Porges, 1979). Regardless
of whether or not the bradycardic response does represent a
component of an adaptive compensatory response to an
elicited-apnea, it is evident that by two months of
postnatal age the cardiac response is clearly coupled with
the apneic episode.
Topography of HR Response in the Absence of ñpnea
When no apnea was elicited by the facial or abdominal
airstream stimuli the NBs produced a monophasic
acceleratory response maximizing at the third post-stimulus
second; whereas the older infants displayed a triphasic
response composed of an initial brief deceleration at post¬
stimulus second one followed by an acceleration maximizing
at second three and then a larger-magnitude deceleration
which reached a minimum at about seventh post-stimulus
second. Lipton et al. (1966) using a five-second airstream
stimulus directed to the abdomen also reported a monophasic
acceleration in NBs maximizing between the third and fourth
post-stimulus second. The magnitude of acceleration
observed was much greater than displayed by the NBs in this
study but this was probably because the airstream intensity
was 2.5 times greater than that used in the present study.
Despite the elevated intensity of the airpuff employed by
Lipton et al. (1966), it was apparent from the figures
presented that the older infants 2.5 and 5 months of age

IOS
also displayed a triphasic HR change with a similar
topography as that observed in this study. It should be
noted that these infants displayed a much larger
acceleratory magnitude than exhibited by the 2-4 month or
8-12 month infants in this study. Unfortunately, they did
not plot enough post-stimulus data to determine if there
were any effects of stimulus intensity on the magnitude of
the late-deceleration. Thus the difference in stimulus
intensity employed by Lipton et al. (1966) and that used in
the present study may account for the relative difference
in acceleratory response magnitude to the abdominal
airstream.
The slightly larger acceleration to the facial
airstream than to the abdominal airstream in this study
suggests that for a given airstream intensity airstream
stimulation of the face will result in somewhat greater
cardiac rate change. This suggests that there may be
regional differences in the thermal sensitivity to
airstream stimulation. It has been observed that the face
displays greater sensitivity for warm thermal stimuli than
the rest of the body (Stevens, Marks & Simonson, 1974).
Although the appropriate studies have not yet been
performed it is likely that facial preeminance is also the
case for cold thermal sensitivity, since the number of cold
receptors in the face are considerably greater per unit of

106
surface area than in any other region of the body (Guyton,
1981). Hence it is reasonable to suggest that the face was
probably more sensitive to the cooling aspects of the
airstream stimulation than was the abdomen.
Regardless of whether the face was more thermally
sensitive than the abdomen, differences in the magnitude of
the cardiac rate changes in the absence of LM and apnea
were only slight and the overall ontogeny of the HR
response topography over age was the same to either
stimulus. That is, for both stimuli a similar shift in HR
topography emerged whereby the neonates displayed a
monophasic HR acceleration and the older infants exhibited
the triphasic response. Lipton et al. (1966) suggested
that this shift in HR topography over age, which they also
observed, probably reflected maturation of the autonomic
nervous system; in particular, it was posited that the
presence of early and late decelerations in only the older
infants and the observation of increasing magnitude of
deceleration of the early component with age was due to
increasing vagal control over cardiac function. However,
neonates are capable of producing deceleratory responses to
stimuli in a variety of sensory modalities (e.g., Adkinson
& Berg, 1976; Porges, Stamps & Walter, 1974; Pomerleau-
Malcuit & Clifton, 1971; Sameroff, Cashmore & Dykes,
1973). Hence, based solely on the evidence of the lack of

107
deceleratory components, it is difficult to conclude that
vagal control was not sufficiently developed in the NBs.
Berg and Berg (1979) reviewed evidence which suggested
that the di- or tri-phasic HR response represented a
unique cardiac response which differed from the monophasic
long-duration deceleration of the orienting response in
the sensitivity or lack thereof to aspects of the steady-
state portions of the stimulus. Others have reinterpreted
the data to suggest that the direction of the HR change in
neonates was dependent on the intensity and complexity of
the stimulus such that low-intensity, simple stimuli with
controlled slow-rise times resulted in HR deceleration
while highly-intense, complex stimuli with uncontrolled
rise times produced HR acceleration (Graham, Anthony &
Zeigler, in press); these authors maintained that the
diphasic response^was not dependent on either high
intensity or sustained stimulus features. They cite
supporting evidence of a diphasic response evoked by a
range of stimuli intensities from 45 to 105 dB SPL, with
the acceleratory component increasing with increasing
intensity in neonates and one to ten month infants who were
asleep. Other evidence suggests that state may be
particularly relevant in the production of this response.
Pomerleau-Malcuit and Clifton (1973) provided evidence of
the effect of state on the HR response to a ten-second

108
tactile stimulus to the forehead of neonates. They
observed HR deceleration in awake infants before feeding
but this same stimulus induced a monophasic acceleration
while asleep. Berg, Berg and Graham (1971) documented this
HR pattern shift with state change in four-month old
infants to auditory stimulation. They observed that in
response to 50 and 75 dB tones with controlled fast and
slow rise times awake infants produced large monophasic
decelerations but as their state changed from alertness to
drowsiness and then to sleep the HR topography changed to a
triphasic response. Rewey (1973) also observed this same
shift from monophasic deceleration when awake to a
triphasic change when sleepy in response to short and long
75 dB continuous and pulsed tones in three month-old
infants; whereas the six week-old infants' response was
more variable and appeared to shift from a small
deceleration when awake to a diphasic response when
asleep. Adults also display a triphasic HR response to a
broad range of intensity of acoustic stimulation when
asleep (Berg, Jackson & Graham, 1975). It was reported
that as the intensity of the stimulus progressed from 30 to
75 dB the magnitude of each of the components of the
triphasic HR response increased. In sum, it appears that
the shift from a monophasic HR acceleration in NBs to a
triphasic HR response occurs in infancy between six weeks

109
and two months of age and is maintained into adulthood.
This response topography is affected by state such that the
triphasic response will not occur if the infant is awake.
Stimulus intensity appears to alter the magnitude of the
components of the response but otherwise the overall
response topography remains unaffected. Therefore, the
shift in HR topography observed from NB to two months of
age and older in this study in response to both abdominal
and facial airstream stimulation follows a developmental
course similar to that which has been documented previously
in the same and other sensory modalities when the infant is
asleep.
Of some relevance is the cardiorespiratory component
of the startle reflex. Landis and Forbes (1934) reported
that the pulse rate was influenced by the phase of the
electrocardiographic cycle in which the startle stimulus
occurred, with more EKG disturbance occurring when the
startle stimulus occurred during atrial systole before the
R-wave. The difficulty of partialing out the effects of
concomitant body movement prevented them from reporting any
consistent change in cardiac rate. More recently, Berg
(1973) elicited the startle reflex acoustically in awake
adults and observed that as the duration of the startle
stimulus increased from 4 to 32 msec the topography of the
HR response became increasingly triphasic. When startle

110
was elicited by a 50 msec airpuff to the outer canthus of
an eye in awake adults a diphasic HR response, which was
only measured up to five-seconds post-stimulus onset, was
observed (Harbin & Berg, 1983). Therefore, it appears that
when the intensity of the startle stimulus was sufficient
to elicit the eyeblink component of the startle without the
full-blown muscular response the HR response was
triphasic.
It has been suggested that the threshold for eliciting
a startle reflex is lowered during sleep (Berg, Jackson &
Graham, 1975). If this were the case, it could be
postulated that the triphasic responses observed in
sleeping adults to low and moderately intense stimuli were
cardiac components of an elicited startle. Hence it may
be that the stimuli which evoked the triphasic response in
sleeping infants contained sufficiently rapid transients to
elicit startle. Startle may also play a role in explaining
the apnea results. The fact that the abdominal airstream
resulted in apnea, albeit infrequently, in the infants in
this study suggests that perhaps more than one mechanism
had a role in inducing apnea in the infants in this study.
The startle reflex appears to be one possible mechanism
because eyelid contractions and large body movements
resembling the Moro or startle reflex were often observed.
Skaggs (1926) reported that respiration was often briefly

Ill
interrupted by the startle stimulus with an ensuing large
inspiration followed by irregular breathing patterns with
accelerated rate. However, instead of a startle reflex
mechanism, Graham and colleagues (in press) have proposed a
more general mechanism termed the "transient-detecting
interrupt system", which occurs when conditions of stimulus
presentation are inappropriate for orienting responses such
as during sleep. The critical feature of the eliciting
stimulus is its transient aspects and may account for the
triphasic HR responses of sleeping infants to low and
moderately intense stimuli of varying complexity and
duration. Whatever the mechanism, the presence of apneic
episodes following instances of LM especially in the NBs
suggests that there may be some different mechanism other
than the dive reflex which played a role in the production
of apnea in the infants in this study. It may be that both
the transient-detecting and dive reflex mechanisms mediate
the elicited-apnea and that they share a final common
effector pathway.
Detection of Infants At-Risk for SIDS
There is considerable evidence that many SIDS victims
may be derived from a population of susceptible infants,
whereby an infant acutely or chronically exposed to
hypoxic conditions may develop faulty cardiorespiratory
control. Initially, however, the infant would be expected

112
to compensate for the hypoxic conditions. Successful
compensatory response may be challenged by complicating
antagonistic conditions such as respiratory infection,
which by itself or in combination with other factors may
induce decompensatory cardiorespiratory response.
Deficiency of regulatory action may result in impaired
function of mechanisms which normally maintain tonic
ventilation. This could be manifested in either
hypoventilation or exagerration of normal reflex responses.
The data yielded in this study indicates that healthy
infants from two to four months of age may be particularly
responsive during sleep to facial airstream stimulation,
exhibiting prolonged elicited-apnea duration and
bradycardia. This is the age of most rapid CNS development
and also corresponds to the age of highest incidence of
SIDS mortality. Moreover, during this time the infants'
sleep states are consolidated and organized, maternal
immunological protection ceases, sensory and perceptual
capabilities dramatically improve and the infant begins to
replace reflexive behavior with purposeful action.
It is reasonable to expect that vulnerability to external
aggravating conditions would be more potentially
deleterious during a time of most rapid development.
The fact that SIDS is more common in the colder winter
months suggests that the role of cold facial stimulation

113
should receive more attention as a possible contributing
factor in the etiology of hypoxia in SIDS victims. We have
revealed that cool facial stimulation with an airstream
stimulus, which the infant might commonly experience in
every day life, can elicit apnea within seconds of stimulus
onset. It may be possible to discriminate between healthy
infants and infants at-risk for SIDS by eliciting apnea
with a facial airstream stimulus. This would be expected
to induce a hypoxic and hypercapnic challenge. In the
older healthy infants it has been shown that apnea of
moderate duration of about 9-12 seconds was elicited. If
infants at-risk for SIDS are, as the literature suggests,
chronically hypoxic and sluggish to respond to a hypoxic or
hypercapnic challenge then one would anticipate deficits in
medullary chemoreceptors, carotid body chemoreceptors
and/or CNS regions of cardiorespiratory control. It would
be predicted, therefore, that during apnea arterial carbon
dioxide levels would accumulate and oxygen levels would
deplete to a greater extent in these infants than healthy
infants before they would respond with resumption of
breathing; this would result in observations of more
prolonged apnea in the at-risk infants. It has been
speculated that SIDS could be a consequence of an
exaggerated dive reflex (Wolf, 1966). If this were the
case, it might be expected that facial cooling would

114
elicit apnea on a greater percent of airstream
presentations in the at-risk infants than controls but more
so during the postnatal age of greatest risk, two to four
months. Moreover, the at-risk infants might also display
more pronounced bradycardic responses during apnea,
although, this would be expected solely on the basis of the
anticipated increased duration of apnea.
The advantages of this test are fourfold: 1) it is
simple and does not require extensive training to be able
to administer; 2) it is not time consuming; 3) it is
inexpensive and requires very little equipment; and 4) this
is a test which elicits apnea and as such challenges the
infants central mechanisms of apnea termination and
resumption of breathing. Simply exposing the infants to
various atmospheric concentrations of oxygen and carbon
dioxide only challenges the mechanisms of modulation of
ventilation upon a background of ongoing respiration.
Since the facial airstream interrupts breathing, central
mechanisms, which presumably are the mechanisms of ultimate
failure in the final common pathway in SIDS infants, are
required to reinitiate breathing from respiratory
standstill. Therefore, if an excessive apnea and
concomitant bradycardic response in infants can be uniquely
associated with a pathophysiology in the reflex control of
cardiorespiratory function relating to enhanced

115
mortality, then diagnostic screening of this type may be of
utility. Specifically, it may be possible to detect a
predisposition to SIDS by analysis of the magnitude of
cardioresiratory change in infants with a history of
prolonged apnea.
Note
1. Graham and coworkers (in press) described this response
as a diphasic deceleration followed by an acceleration.
There are others that have reported a triphasic response
topography {e.g., Berg, Jackson & Graham, 1975). One of
the reasons there is this discrepency is that some studies
have measured only five seconds post-stimulus onset and
consequently missed the late-deceleration which typically
reaches its nadir by about the seventh or eighth post¬
stimulus second (e.g., Porges, 1974; Harbin & Berg, 1983).
In addition, another reason for the confusion over the form
of the response may be that the late component is the least
reliable and often appears as a secondary accelerative
phase following a brief return toward pre-trial levels
(e.g., Kearsley, 1973; Ver Hoeve & Leavitt, 1983). In the
studies which examined the topography of the HR response
over at least ten seconds post-stimulus onset the response
predominantly appears as a cubic or quartic trend over
seconds.

APPENDIX A
INFANT BACKGROUND QUESTIONAIRE
The data which we collect from your baby can be
affected by a number of factors. The more information we
have about your baby, the easier it will be to interpret
these data. The following questions are for our use in
better understanding your baby's responses, and no
individual information will be reported. If you do not
wish to answer any question, just leave it blank. Please
feel free to ask us to clarify any of the questions.
Mother's name
Infant's name
Today's date
Mother's Age
Birth date
Birth weight
APGAR scores
Has the baby had any recent colds, ear infections,
respiratory difficulties, or any other major health
problems? If so, please specify, and provide the
approximate dates and durations.
Is the baby now taking any medication? If so, what is
it, how often and in what dosage is it taken, and what
problem is it treating?
116

117
3. How many brothers and sisters does the baby have, and
what are their ages?
4. If known, was the baby born early, late, or on time?
What was the due date?
5. If known, what sort of medication did the mother receive
during labor and during birth? If you do not know exactly,
please indicate if there were any pills, injections, or gas
received.
6. Were there any special problems during pregnancy, such
as illnesses, infections, or injuries?
7. Were there any special conditions during birth, such as
breech birth, Cesarean section, or use of forceps?
8. Was the Lamaze method used?
9. Where was the baby born?
10. Did the baby have to stay in the hospital for any extra
period of time? If so, why?
11. Do you breast-feed your baby?
12. If so, are you taking any medication? If so, what is
it, how often and in what dosage is it taken, and what
problem is it treating?

APPENDIX B
INFANT STATE RATING SHEET
Stimulus A
Sequence B
A* F** F A F
A A F A F
Trial No.
1 2 3 4 5
pre post pre post pre post pre post pre post
EYES
open
closed
REMs
MOVEMENT
none
small
large
VOCALIZATION
quiet
soft
loud
STATE
asleep
drowsy
alert
7
fussy
cryinq
Subject: Date tested : Air temp:
Postnatal Age: Rater:
* A denotes abdomen
**F denotes face
The following is a list of the remaining stimulus sequence
from trial number 6 to 20 for sequences A and B, respectively.
Sequence A - A,F,F,A,A,F,F,A,A,F,A,F,A,F,A
Sequence B - A, A, F,F,A,F,A,F,A,F,F,A,A,F , F
118

APPENDIX C
INFORMED CONSENT
We are involved in developing a new procedure to test
the breathing of young infants. Although our eventual aim
is to provide a test that can be used with prematurely
born infants, we are interested in results from full-term
babies as well. Unfortunately, if breathing problems are
present and go undetected, they can have widespread effects
on the infant's development and could possibly result in
difficulties later on. We would like to explain the
procedure we are using to develop the breathing test, and
ask your permission for your baby to participate in the
research. The information obtained will remain strictly
confidential to the extent provided by law. You will have
complete access to all information gathered about your
infant. As well, if you desire, we will be happy to
discuss any conclusions we reach upon completion of this
study and send you a summary of the data we collect. The
results from this study may be published in a scientific
journal concerned with breathing problems of infants.
The basic procedures have been used successfully with
infants in the past. What we do is to record heart rate and
breathing with small sensors taped to the chest. A short
119

120
puff of air will be given to the face or stomach area,
something like you might do when playing a game by lightly
blowing in the baby's face. None of the sensors nor the air
puff produce any discomfort or danger. In fact, in most
circumstances we will want to carry out the test while the
baby is asleep. The entire test lasts about one hour.
Because the test is still being developed, we probably
will not have specific information about your baby's
breathing, but should we find a suggestion of a difficulty we
will inform you. There are no risks or discomforts expected.
You will receive five dollars for participating in this
project.
If you have any further questions now or at any other
point, we will be happy to answer them. If at any time you
decide to withdraw consent or discontinue participation you
are, of course, free to do so.
I understand that if my infant is injured during this study,
and if the investigator is at fault, the University of
Florida and the Board of Regents of the State of Florida
shall be liable only as provided by law. I understand that I
may seek appropriate compensation for injury by contacting
the Insurance Coordinator at 107 Tigert Hall, University of
Florida, telephone number 392-1325.

121
I have read and understand the above described procedure
in which my intent is to participate and have received a copy
of this description. I agree to alllow my infant
child, (baby's name), to participate in the above
described research.
subject date witness date
Investigator date
(relationship to subject)

APPENDIX D
ANALYSIS OF THE PERCENT OF TRIALS WITH QS, TS AND AS STATES
The following analyses were all ANOVAs, which examined
the percent of face and abdomen trials in which infants
were in QS, TS and AS, with Age (i.e., NB vs. 2-4 vs. 8-12)
as the between-subject factor and Site of stimuluation
(i.e., face vs. abdomen) and State (i.e., QS vs. AS vs. TS)
as within-subject factors. The results of this analysis
only revealed a significant interaction between State and
Age, F(4/98)=7.9, £<.001. Subanalysis consisted of
dividing the analyses into QS and non-QS data. The analysis
of QS yielded a significant Age effect, F(2/49)=8.1, £<.001,
which was due to less QS displayed by the NBs than by 2-4
infants, F(l/35)=8.4, £<.01, and no difference between the
2-4 and 8-12 groups. The analysis of the non-QS data
(i.e., percent of trials in AS and TS) revealed a
significant interaction between State and Age, F(2/49)=6.8,
£<.005. Follow-up analyses revealed greater percentage of
AS in the NBs than in the 2-4 infants, F(l/35)=8.4, £<.01;
however, no significant difference in the percentage of
either AS or TS emerged when the two older groups were
compared. Therefore, in sum, infants spent a predominant
amount of time in QS throughout the testing session. The
122

123
NBs displayed a slightly greater tendency to be in AS than
the 2-4 or 8-12 infants. In any case, the infants showed
no tendency to be in any particular state when a face or
abdominal airpuff was delivered.
To determine if there was any sleep state pattern
during the session, an analysis of State was performed
comparing the state ratings (i.e., 2 for AS, 1 for TS and 0
for QS) across Trials, ignoring stimulus Site. Figure 1
depicts these mean state ratings across the 20-trial
session. The initial analysis constrasted State across the
first 10 trials for all age groups. Since each infant had
at least 10 trials, all 52 infants could be included in the
analysis. A significant Age-by-Trial interaction,
F(18/441)=2.2, £<.005, emerged. An independent orthogonal
trend analysis of each age group revealed significant
decreasing linear trends across Trials for NBs,
F(l/13)=7.5, £<.05, and for 2-4s, F(l/22)=8.2, £<.01; but
the analysis of the 8-12 infants did not demonstrate a
significant change across Trials, F(9/126)=1.6, indicating
a greater tendency for the younger infants to be in AS or
TS at the beginning of testing.
The next analysis investigated whether the state of
the infants changed during the last 10 trials. It included
only those subjects in each age group who completed at least
18 trials. This analysis revealed only a significant

124
increasing linear trend across Trials, F(l/36)=4.2, p<.05,
indicating that the infants demonstrated a gradual change
of state toward AS from QS at the end of the session.
Figure 1 displays these trends also.

APPENDIX E
ANALYSIS OF THE PERCENT OF TRIALS WITH LARGE MOVEMENT
Table 5 presents the mean percent of trials in which
LM was observed across age group in response to the
airstream stimulus to the face and the abdomen. The
analyses of the percent of face and abdomen trials in which
LM was observed were all ANOVAs with Age as the between-
subject factor and Site of stimulation, State (QS vs. all
states collapsed) and presence or absence of LM as the
three within-subject factors. Neither the effect of State,
nor any of the interactions of State with Age, Conditions
or LM were significant, indicating that the frequency of LM
was not affected by the inclusion of non-QS data in the
sample. There was, however, a significant interaction
between Age, Site and LM, F(2/49)=3.7, £<.05.
Follow-up analyses consisted of comparing whether
stimulus site differentially influenced the percent of
elicited-LM in the 2-4 and 8-12 infants; it did not,
F(l/36)=0.5. However, while there was greater percent of
LM produced by the 2-4 infants than by the 8-12 infants,
F(1/36)=350.4, £<.001, there was a much greater percent of
trials with no LM than trials with LM, F(1/35)=91.1, £<.001.
This was not the case for the NB infants who displayed
125

126
greater percent of trials with LM than the two older
groups, F(1/35)= 54.6, £<.001; the NBs also produced a
greater proportion of trials with LM relative to the
percent of trials with no LM observed. This was confirmed
in an analysis of only the NB group by a significant Site-
by-LM interaction, F(l/13)=7.1, £<.05. This interaction was
due to a significantly less percent of abdomen trials with
LM than abdomen trials with no LM, F(l/13)=7.4, £<.05;
whereas no differences were found between the percent of
face trials with LM and face trials with no LM.
In sum, there was more LM elicited in the NBs than the
2-4 infants, who in turn displayed more elicited LM than
the 8-12 infants. Overall there were many more trials with
no LM elicited than trials with LM. It was only the NB
group which displayed more sensitivity to the facial
airpuff, with more elicited LM than to the abdominal
airpuff.

APPENDIX F
ANALYSIS OF THE HR RESPONSE TO THE AIRSTREAM STIMULUS:
FACE VS. ABDOMEN
To examine whether there were any differences between
age groups and stimulus sites in the HR response during the
ten seconds of airstream presentation the following
analyses were performed. Within the first second post¬
stimulus onset there was a brief early-deceleration.
An ANACOVA on second one revealed only an Age effect,
F(2/46)=6.4, £<.005. Subanalyses indicated that the two
older groups of infants displayed a greater initial
deceleration than the NBs. This was revealed when the two
younger groups were separately analysed; there was a
significant Age effect, F(l/32)=5.7, £<.05, found, but no
significant difference was found when the two older groups
were compared. Additionally, the NBs' HR for second one
did not significantly change from pretrial levels,
_t (12)=0.21. Therefore, both the 2-4 and the 8-12 groups
decelerated within the first second after airstream onset
to mean HR levels of -1.21 and -2.75 bpm, respectively,
regardless of whether the airstream was directed to the
face or abdomen. The NBs, on the otherhand, displayed no
change in HR level during this time.
127

128
The acceleration component was evaluated by analysing
the third second post-stimulus onset. Only a significant
stimulation Site effect, F(l/46)=4.7, £<.05, was found,
indicating that age did not significantly influence
magnitude but that there was greater HR acceleration to the
facial airpuff than the abdominal airpuff, with mean
adjusted HR bpm values of 4.25 and 3.04 respectively.
However, this conclusion was not quite accurate because the
older infants accelerated from below pretrial level while
the NBs did not. Therefore, the infants displayed
differing rates of acceleration across seconds 1, 2 and 3.
Accordingly, analyses revealed that the 8-12 group
exhibited a faster rate of acceleration than the 2-4 group,
which accelerated faster than the NBs, as indicated by
respective significant two-way interactions of Age and
Seconds of F(2/70)=7.3 and F(2/66)=6.1, both at £<.005.
The subsequent late-deceleration was analysed using
seconds 3 to 10. The initial ANACOVA yielded interactions
between Age and Site, F(2/46)=3.1, £<.05, and Seconds and
Site, F(7/329)=2.5, both significant at £<.05. When the NB
and 8-12 groups were contrasted only a significant
interaction between Age and Seconds, F(7/182)=6.3, £<.001,
remained, indicating that HR decelerated to a greater
magnitude in the 8-12 infants than the NBs. In fact,
separate t-tests over seconds 3 to 10 for the NB infants

129
revealed that HR only returned to pretrial levels from
the previous acceleration and did not significantly
decelerate. Moreover, since none of the effects which
included Site reached significance, it can be concluded
that neither of these groups responded differentially to
either airpuff during this interval. However, when the 2-4
group was separately analysed the significant interaction
between Sites and a quadratic trend over Seconds,
F(l/21)=6.4, £<.05, remained, indicating that the former
interactions with the Sites factor were due to greater
deceleration elicited by the facial airpuff than by the
abdominal airpuff in these infants.
The analysis of the HR trends during the seconds 11 to
25 of stimulus offset only revealed significant main
effects of Age, F(2/46)=7.1, £<.005, Seconds,
F (14/658)=5.7, £<.001, and their two-way interaction,
F(28/658)=2.4, £<.001. An orthogonal trend analysis
revealed that this interaction arose from cubic trends over
Seconds, F(2/47)=3.7, £<.05, indicating that in response to
the offset of either facial or abdominal airpuff each age
group displayed a similar cubic trend in HR (i.e., a
recovery toward pretrial levels, then a brief decrement in
HR followed by continued recovery toward pretrial levels).

REFERENCES
Adkinson, C.D., & Berg, W.K. Cardiac deceleration in
newborns: Habituation, dishabituation, and offset
responses. Journal of Experimental Child Psychology* 1976,
21, 46-60.
Allen, L.G., Howard, G., Smith, J.B., McCubbin, J.A., &
Weaver, R.L. Infant heart rate response to trigeminal
airstream stimulation: Determination of normal and deviant
values. Pediatric Research, 1979, 13, 184-187.
Anderson, H.T. The reflex nature of the physiological
adjustments to diving and their afferent pathway. Acta
Physiologica Scandinavica, 1963, 5J3, 263-273.
Ariagno, R.L., Guilleminault, C., Baldwin, R., & Owen-
Boeddiker, M. Movement and gastroesophageal reflux in awake
term infants with near miss SIDS, unrelated to apnea.
Journal of Pediatrics, 1982, 100, 894-897.
Benjamin, L.S. Facts and artifacts in using analysis of
covariance to "undo" the law of initial values.
Psychophysiology, 1967, 4_, 187-202.
Bergman, A.B., Beckwith, J.B., & Ray, C.G. Part II.
Epidemiology. Proceedings of the second international
conference on causes of sudden death in infants. Seattle:
University of Washington Press, 1970, Pp. 25-79.
Berg, K.M. Elicitation of acoustic startle in the human.
Universtiy of Wisconsin, 1973. Dissertation Abstracts
International, 1974, 34B(10), 5217.
Berg, K.M., Berg, W.K., & Graham, F.K. Infant heart rate
response as a function of stimulus and state.
Psychophysiology, 1971, 8, 30-44.
130

L 31
Berg, W.K., & Berg, K.M. Psychophysiological development in
infancy: State, sensory function and attention. In J.
Osofsky (Ed.) Handbook of Infant Development. New York:
Wiley, 1979, Pp. 283-343.
Berg, W.K., Jackson, J.C., & Graham, F.K. Tone intensity
and rise-decay time effects on cardiac responses during
sleep. Psychophysiology, 1975, _12, 254-261.
Boychuk, R.B., Rigatto, H., & Seshia, M.M.K. The effect of
lung inflation on the control of respiratory frequency of
the neonate. Journal of Physioloqy (London), 1977, 270,
653-660.
Brady, J.P., Ariagno, J.L., Watts, J.C., Goldman, S.L., &
Dumpit, T.M. Apnea, hypoxemia and aborted sudden infant
death syndrome. Pediatrics, 1978, S2_r 686-691.
Brand, M.M., & Bignami, A. The effects of chronic hypoxia
on the neonatal and infantile brain: A neuropathological
study of five premature infants with the respiratory
distress syndrome treated by prolonged artificial
ventilation. Brain, 1969, £2, 233-254.
Brooks, J.G. Apnea of infancy and sudden infant death
syndrome. American Journal of Diseases in Children, 1982,
136, 1012-1023.
Brooks, J.G., Schluete, M.A., Nabelet, Y., & Tooley, W.H.
Sleep state and arterial blood gases and pH in human
newborn and young infants. Journal of Perinatal Medicine,
1978, 6, 280-286.
Brown, W.V., Ostheimer, G.W., Bell, G.C., & Datta, S.S.
Newborn response to oxygen blown over the face.
Anesthesiology, 1976, _44, 535-536.
Coons, M.A., & Guilleminault, M.D. Development of sleep-
wake patterns and non-rapid eye movement sleep stages
during the first six months of life in normal infants
Pediatrics, 1982, 69, 793-798.

132
Cordero, L., Jr., & Hon, E.H. Neonatal bradycardia following
nasopharyngeal stimulation. Journal of Pediatrics, 1971,
28, 441-447.
Dejours, P. Control of respiration by arterial
chemoreceptors. Annals of the New York Academy of Science,
1963, 109, 682-695.
Downing, S.E., & Lee, J.C. Laryngeal chemosensitivity: A
possible mechanism for sudden infant death syndrome.
Journal of Pediatrics, 1981, 98, 791-794.
Eisner, R., Franklin, D.L., Van Citters, R.L., & Kenney, D.W.
Cardiovascular defense against asphyxia. Science, 1966,
153, 141-149.
Furedy, J.J., Morrison, J.W., Heslegrave, R.J., &
Arabian, J.M. Effects of water temperature on some
noninvasively measured components of the human dive reflex
An experimental response-topography analysis.
Psychophysiology, 1983, 20, 569-578.
Getts, A.G., & Hill, F.H. Sudden infant death syndrome:
Incidence at various altitudes. Developmental Medicine
and Child Neurology, 1982, 2_4_, 61-68.
Graham, F.K. Constraints on measuring heart rate and period
sequentially through real and cardiac time.
Psychophysiology, 1978, 15y 492-495.
Graham, F.K., Anthony, B.J., & Zeigler, B.L. The orienting
response and developmental processes. In D. Siddle (Ed.),
The Orienting Response. Sussex, England: Wiley, in press.
Greenhouse, S.W., & Geisser, S. On methods in the analysis
of profile data. Psychometrica, 1959, 2A_, 95-112.
Gregory, G.A. Resuscitation of the newborn. Anesthesioloqy,
1975, 43, 225-237.

133
Guilleminault, C., Ariagno, R., Korobkin, R., Coons, S.,
Owen-Boeddiker , M., & Baldwin, R. Sleep parameters and
respiratory variables in "near miss" sudden infant death
syndrome infants. Pediatrics, 1981, £8, 354-360.
Guilleminault,C., Ariagno,R., Korobkin, R., Nagel, L.,
Baldwin, R., Coons, S., & Owen, M. Mixed and obstructive
sleep apnea and near miss for sudden infant death syndrome:
2. Comparison of near miss and normal control infants by
age. Pediatrics, 1979, 6j5, 882-891.
Guilleminault, C., Souquet, M., Ariagno, D.L.,
Korobkin, R., & Simmons, F.B. Five cases of near-miss
sudden infant death syndrome and development of obstructive
sleep apnea syndrome. Pediatrics, 1984, 7¿, 71-78.
Guyton, A.C. Textbook of Medical Physiology. Philadelphia:
Saunders Co., 1981, Pp. 623.
Haddad, G.G., Epstein, M.A.F., Epstein, R.A., Mazza, N.M.,
Mellins, R.B., & Krongrad, E. The QT interval in aborted
sudden infant death syndrome infants. Pediatric Research,
1979, 13, 136-138.
Haddad, G.G., Leistner, H.L., Lai, T.L., & Mellins, R.B.
Ventilation and ventilatory pattern during sleep in aborted
sudden infant death syndrome. Pediatric Research, 1981, 15,
879-883.
Haddad, G.G., Schaeffer, J.I., & Bazzy, A.R. Postnatal
maturation of the ventilatory response to hypoxia. In
J.R. Sutton, C.S. Houston & N.L. Jones (Eds.), Hypoxia,
exercise, and altitude: Proceedings of the third Banff
international hypoxia symposium. New York: Liss Inc.,
1983, Pp. 39-50.
Haddad, G.G., Walsh, E.M., Leistner, H.L., Grodin, W.K., &
Mellins, R.B. Abnormal maturation of sleep states in
infants with aborted sudden infant death syndrome.
Pediatric Research, 1981, 15, 1055-1057.

134
Haidmayer, R., Kurz, R., Kenner, T. , Wurm, H., & Pfeiffer,
K.P. Physiological and clinical aspects of respiration
control in infants with relation to the sudden infant death
syndrome. Klinische Wochenschr i f t, 1982, 6(), 9-18.
Harbin, T.J., & Berg, W.K. The effects of age and
prestimulus duration upon reflex inhibition.
Psychophysiology, 1983, 20, 603-610.
Harper, R.M. Neurophysiology of sleep. In J.R. Sutton, C.S.
Houston & N.L. Jones (Eds.), Hypoxia, exercise, and
altitude: Proceedings of the third Banff international
hypoxia symposium. New York: Liss Inc., 1983, Pp. 65-73.
Harper, R.M., Frostig, Z., Taube, D., Hoppenbrouwers, T.,
& Hodgman, J.E. Development of sleep-waking temporal
sequencing in infants at risk for the sudden infant death
syndrome. Experimental Neurology, 1983, 7_9' 821-829.
Harper, R.M., Leake, B., Hodgman, J.E., & Hoppenbrouwers, T.
Developmental patterns of heart rate and heart rate
variability during sleep and waking in normal infants and
infants at risk for the sudden infant death syndrome.
Sleep, 1982, 5, 28-38.
Harper, R.M., Leake, B., Hoppenbrouwers, T., Sterman, M.B.,
McGinty, D.J., & Hodgman, J. Polygraphic studies of normal
infants and infants at risk for the sudden infant death
syndrome. Pediatric Research, 1978, 12, 778-785.
Harper, R.M., & Sieck, G.C. Discharge correlations between
neurons in the nucleus parabracbialis medialis during sleep
and waking states. Brain Research, 1980, 199, 343-358.
Henderson-Smart, D.J. Regulation of breathing in the
perinatal period. In N.A. Saunders & C.E. Sullivan (Eds.),
Sleep and breathing, New York: Marcel Dekker Inc., 1984,
Pp. 423-455.
Henderson-Smart, D.J., & Read, D.J.C. Depression of
intercostal and abdominal muscle activity and vulnerability
to asphyxia during active sleep in the newborn. In
C. Guilleminault and W.C. Dement (Eds.), Sleep Apnea
Syndromes, New York: Liss Inc., 1978, 93-117.

135
Herbst, J.J., Minton, S.r & Book, L.S. Gastroesophageal
reflux causing respiratory distress and apnea in newborn
infants. Journal of Pediatrics, 1979, 95, 763.
Hodgman, J.E., Hoppenbrouwers, T., Geidel, S., Hadeed, A.,
Sterman, M.B., Harper, R., & McGinty, D. Respiratory
behavior in near-miss sudden infant death syndrome.
Pediatrics, 1982, £9, 785-792.
Hoppenbrouwers, T., & Hodgman, J.E. Sudden infant death
syndrome (SIDS): An integration of ontogenetic pathologic,
physiologic and epidemiologic factors. Neuropediatrics,
1982, 13, 36-51.
Hoppenbrouwers, T., Hodgman, J.E., Arakawa, K., Harper,
R., & Sterman, M.B. Respiration during the first six months
of life in normal infants. III. Computer identification of
breathing pauses. Pediatric Research, 1980, 14, 1230-1233.
Hoppenbrouwers, T., Hodgman, J.E., Harper, R.M., & Sterman,
M.B. Fetal heart rates in siblings of infants with sudden
infant death syndrome. Obstetrics and Gynecoloqy, 1981, 58,
319-325.
Hoppenbrouwers, T., Hodgman, J.E., McGinty, D., Harper,
R.M., & Sterman, M.B. Sudden infant death syndrome: Sleep
apnea and respiration in subsequent siblinqs. Pediatrics,
1980, 66, 205-214.
Hoppenbrouwers, T., Jenson, D.K., Hodgman, J.E., Harper,
R.M., & Sterman, M.B. The emergence of a circadian pattern
in respiratory rates: Comparison between control infants
and subsequent siblings of SIDS. Pediatric Research, 1980.
14, 345-351.
Hoppenbrouwers, T., Zanini, B., & Hodgman, J.E. Intrapartum
fetal heart rate and sudden infant death syndrome. American
Journal of Obstetrics and Gynecoloqy, 1979, 133, 217-220.

13ó
Hunt, C.E. Abnormal hypercarbic and hypoxic sleep arousal
responses in near-miss SIDS infants. Pediatric Research,
1981, 15, 1462-1464.
Hunt, C.E., McCulloch, K., & Brouillette, R.T. Diminished
hypoxic ventilatory responses in near-miss sudden infant
death syndrome. Journal of Applied Physiology: Respiration,
Environmental and Exercise Physiology, 1981, 50, 1313-1317.
Hurwitz, B., & Furedy, J.J. The human dive reflex: An
experimental topographical and physiological analysis.
Psychophysiology, 1979, 16_r 192. (Abstract)
Irving, L. , Solandt, D.Y., Solandt, O.M., & Fisher, K.C.
The respiratory metabolism of the seal and its adjustments
to diving. Journal of Cellular and Comparative Physioloqy,
1935, 7, 137-151.
Ratona, P.G., Frasz, A., & Egbert, J. Maturation of cardiac
control in full-term and preterm infants during sleep.
Early Human Development, 1980, 4^, 145-159.
Kaun, A., Blum, D., Engelman, E., & Waterschoot, P. Effects
of central apneas on transcutaneous P02 in control
subjects, siblings of victims of sudden infant death
syndrome, and near-miss infants. Pediatrics, 1982, 69,
413-418.
Kawakami, Y. , Natelson, B.H., & DuBois, A.B. Cardiovascular
effects of face immersion and factors affecting diving
reflex in man. Journal of Applied Physioloqy, 1967, 23,
964-970.
Kelly, D.H., & Shannon, D.C. Periodic breathing in infants
with near-miss sudden infant death syndrome. Pediatrics,
1979, 63, 355-360.
Kelly, D.H., & Shannon, D.C. Sudden infant death syndrome
and near sudden infant death syndrome: A review of the
literature, 1964 to 1982. Pediatric Clinics of North
America, 1982, 29, 1241-126T!

137
Kelly, D.H., Twanmoh, J., & Shannon, D.C. Incidence of apnea
in siblings of sudden infant death syndrome victims studied
at home. Pediatrics, 1982, 70, 128-131.
Kelly, D.H., Walker, A.M., Cahen, L. , & Shannon, D.C.
Periodic breathing in siblings of sudden infant death
syndrome victims. Pediatrics, 1980, 6j>, 515-520.
Kirkpatrick, S.M.L., Olinsky, A., Bryan, M.H., & Bryan,
A.C. Effect of premature delivery on the maturation of the
Hering-Breuer inspiratory inhibitory reflex in human
infants. Journal of Pediatrics, 1976, 88, 1010-1014.
Kuich, T.E., & Zimmerman, D. Endorphins, ventilatory control
and sudden infant death syndrome—A review and synthesis.
Medical Hypotheses, 1981, 1_, 1231-1240.
Landis, C., & Forbes, T.W. The relation of startle
reactions to the cardiac cycle. Psychiatric Quarterly,
1934, 8, 235-242.
Leistner, H.L., Haddad, G.G., Epstein, R.A., Lai, T.L.,
Epstein, M.A.F., Eng, S.D., & Mellins, R.B. Heart rate and
heart rate variability during sleep in aborted sudden
infant death syndrome. Journal of Pediatrics, 1980, 97,
51-55.
Lipton, E.L., Steinschneider, A., & Richmond, J.B. Autonomic
function in the neonate: VII. Maturational changes in
cardiac control. Child Development, 1966, 37_, 1-16.
Maloney, J.E., & Bowes, G. The effect of hypoxia on the
fetal respiratory system. In J.R. Sutton, C.S. Houston &
N.L. Jones (Eds.), Hypoxia, exercise, and altitude:
Proceedings of the third Banff international hypoxia
symposium. New York: Liss Inc., 1983, Pp. 17-37.
Marotta, F. , Fort, M., Mondestin, H., Hiatt, I.M., & Hegyi, T.
The response to C02 in infants at risk for SIDS.
Pediatric Research, 1984, lj>, 327A. (Abstract)

138
McCubbin, J.A., Smith, J.B., Allen, L.G. , Wood, F.B., & McGraw,
C.P. Bradycardia in the early neonatal period: The diving
reflex and trigeminal airstream stimulation. Psychophysiology,
1977, 1_4, 111. (Abstract)
McGinty, D.J., & Beahm, E.K. Neurobiology of sleep. In N.A.
Saunders & C.E. Sullivan (Eds.), Sleep and breathing. New
York: Marcel Dekker Inc., 1984, Pp. 1-89.
McGinty, D.J., & Sterman, M.B. Sleep physiology, hypoxemia,
and the sudden infant death syndrome. Sleep, 1980, 3,
361-373..
Mellins, R.B., & Haddad, G.G. Cardiorespiratory control in
sudden infant death syndrome. In J.R. Sutton, C.S.
Houston & N.L. Jones (Eds.), Hypoxia, exercise, and
altitude: Proceedings of the third Banff international
hypoxia symposium. New York: Liss Inc., 1983, Pp. 51-56.
Moore, A. The sudden infant death syndrome. British Journal
of Hospital Medicine, 1981, 2J5, 37-45.
Myers, R.E. Two patterns of perinatal brain damage and
their conditions of occurrence. American Journal of
Obstetrics and Gynecology. 1972, 112, 246-276.
Naeye, R.L. Pulmonary arterial abnormalities in the sudden
infant death syndrome. New England Journal of Medicine.
1973, 289, 1167-1170.
Naeye, R.L. Brain-stem and adrenal abnormalities in the
sudden-infant-death syndrome. American Journal of Clinical
Pathology, 1976, 66, 526-530.
Naeye, R.L., Fisher, R., Ryser, M., & Whalen, P. Carotid
body in sudden infant death syndrome. Science, 1976, 91,
567-569.
Naeye, R.L., Whalen, P., Ryser, M., & Fisher, R. Cardiac and
other abnormalities in the sudden infant death syndrome.
American Journal of Pathology, 1976, 8j2, 1-8.

139
Parmelee, A.H., Stern, E., & Harris, M.A. Maturation of
respiration in prematures and young infants.
Neuropadiatrie, 1972, 3^ 294-303.
Peterson, D.R. Evolution of the epidemiology of sudden
infant death syndrome. Epidemiologic Reviews, 1980, 2,
97-112.
Pomerleau-Malcuit, A., & Clifton, R.K. Neonatal heart-rate
response to tactile, auditory, and vestibular stimulation
in different states. Child Development, 1973, 4_4, 485-496.
Porges, S.W. The application of spectral analysis for the
detection of fetal distress. In T.M. Fields, A.M. Sostek,
S. Goldberg, & H.H. Shuman (Eds.), Infants born at risk.
New York: Spectrum, 1979.
Porges, S.W., Stamps, L.E., & Walter, G.F. Heart rate
variability and newborn heart rate responses to
illumination changes. Developmental Psychology, 1974, 10,
507-513.
Prechtl, H.F.R. The behavioral states of the newborn infant
(a review). Brain Research, 1974, 76, 185-212.
Prystowsky, E.N., Jackman, W.M., Rickenberger, R.L., Heger,
J.J., & Zipes, D.P. Effect of autonomic blockade on
ventricular refractoriness and atrioventricular nodal
conduction in humans: Evidence supporting a direct
cholinergic action on ventricular muscle refractoriness.
Circulation Research, 1981, 49, 511-518.
Rabinovitch, M., Gamble, W., Nadas, A.S., Meittinen, Q.S., &
Reid, L. Rat pulmonary circulation after chronic hypoxia:
Hemodynamic and structural features. American Journal of
Physiology, 1979, 236, 818-827.
Rappaport, L., Kozakewich, H.P.W., Fenton, T. , Mandell, F.,
Vawter, G.F., & Yang, H.Y. Cerebrospinal fluid met-enkephalin
in sudden infant death syndrome. Pediatric Research, 1984,
16, 231A. (Abstract)

140
Rewey, H.H. Developmental change in infant heart rate
response during sleeping and waking states. Developmental
Psychology, 1973, 8^ 35-41.
Richards, J.E. The statistical analysis of heart rate: A
review emphasizing infancy data. Psychophysiology, 1980,
17, 153-166.
Richardson, P.S., & Peatfield, A.C. Reflexes concerned in
the defence of the lungs. Bulletin of European
Physiopathology and Respiration, 1981, 11_, 979-1012.
Rigatto, H. Ventilatory response to hypoxia. Seminar in
Perinatology, 1977, 1, 357-362.
Rigatto, H. Apnea. Pediatric Clinics of North America,
1982, 29, 1105-1116.
Sachis, P.N., Armstrong, D.L., Becker, L.E., & Bryan, A.C.
The vagus nerve and sudden infant death syndrome: A
morphometric study. Journal of Pediatrics, 1981, 98,
278-280.
Sameroff, A.J., Cashmore, T.F., & Dykes, A.C. Heart rate
deceleration during visual fixation in human newborns.
Developmental Psychology, 1973, ^ 117-119.
Schulte, F.J., Albani, M., Schmizer, H., & Bentele, K.
Neuronal control of neonatal respiration-sleep apnea and
the sudden infant death syndrome. Neuropediatrics, 1982,
13, 3-14.
Shannon, D.C. Pathophysiologic mechanisms causing sleep
apnea and hypoventilation in infants. Sleep, 1980, 3,
343-348.
Shannon, D.C., & Kelly, D.H. SIDS and near-SIDS (second of
two parts). New Enqland Journal of Medicine, 1982, 306,
1022-1028.

141
Shannon, D.C., & Kelly, D.H. Sudden Infant Death Syndrome.
In N.A. Saunders & C.E. Sullivan (Eds.), Sleep and
breathinq, New York: Marcel Dekker Inc., 1984, Pp. 457-
484.
Shannon, D.C., Kelly, D.H., & O'Connell, K. Abnormal
regulation of ventilation in infants at risk for sudden
infant death syndrome. New England Journal of Medicine,
1977, 297, 747-750.
Skaggs, E.B. Changes in pulse, breathing and steadiness
under conditions of startledness and excited expectancy.
Journal of Comparative Psychology, 1926, 303-318.
Stevens, J.C., Marks, L.E., & Simonson, D.C. Regional
sensitivity and spatial summation in the warmth sense.
Physiology and Behavior, 1974, 1^, 825-836.
Song, S.H., Lee, W.K., Chung, Y.A., & Hong, S.K. Mechanism
of apneic bradycardia in man. Journal of Applied
Physiology, 1969, 2J_, 323-327.
Steinschneider, A. Prolonged apnea and the sudden infant
death syndrome: Clinical and laboratory observations.
Pediatrics, 1972, 50, 646-654.
Steinschneider, A. Prolonged apnea and respiratory
instability: A discriminative study. Pediatrics, 1977, 59,
962-970.
Sterman, M.B., Harper, R.M., Hoppenbrouwers, T., McGinty,
D.J., & Hodgman, J.E. Quantitative comparison of EEG
development during sleep in infants at high and low risk
for the sudden infant death syndrome. Sleep Research, 1979,
9, 132. (Abstract)
Takashima, S., Armstrong, D., Becker, L.E., & Bryan, A.C.
Cerebral hypoperfusion in the sudden infant death syndrome?
Brainstem gliosis and vasculature. Annals of Neuroloqy,
1978, 4, 257-262.

142
Taylor, W.B. A single risk factor in sudden infant death
and its multiple attack distribution. International Journal
of Epidemiology, 1982, 1_1, 138-145.
Valdes-Dapena, M.A. Sudden infant death syndrome: A review
of the medical literature 1974-1979. Pediatrics, 1980, 66,
597-614.
Valdes-Dapena, M.A., Gillane, M.M., Cassady, J.C.,
Catherman, R., & Ross, D. Wall thickness of small pulmonary
arteries: Its measurement in victims of sudden infant
death syndrome. Archives of Pathological and Laboratory
Medicine, 1980, 104, 621-624.
Ver Hoeve, J.N., & Leavitt, L.A. Neonatal acoustically-
elicited cardiac response: Modulation by state and
antecedent stimulation. Manuscript submitted for
publication, 1983.
Walsh, J.K., Farrell, M.L., Keenan, W.G., Lucas, M., &
Kramer, M. Gastroesophageal reflux in infants: Relation to
apnea. Journal of Pediatrics, 1981, 9j^, 197.
Wilder, J. The law of initial values. Psychosomatic
Medicine, 1950, 12, 392.
Wilds, P.L. Observations of intrauterine fetal breathing
movements--A review. American Journal of Obstetrics and
Gynecology, 1978, 131, 315-338.
Williams, A., Vawter, G., & Reid, L. Increased muscularity
of the pulmonary circulation in victims of the sudden
infant death syndrome. Pediatrics, 1979, 6_3, 18-23.
Wolf, S. Sudden death and the oxygen-conserving reflex.
American Heart Journal, 1966, 71, 840-841.

BIOGRAPHICAL SKETCH
Barry Elliot Hurwitz, the fourth son of William and
Ruth Hurwitz, was conceived in love but with questionable
planning and born on March 18, 1956 in Windsor, Ontario,
Canada, seven years after the birth of number three son,
Michael. Due to the age difference between him and his
brothers (i.e., Paul, Richard and Michael), most of Barry's
formative years were spent as an only child. However,
one shouldn't despair because this meant that Barry was the
sole beneficiary of parents skilled in the art of
manipulating reinforcement contingencies. The diligent use
of these skills resulted in the enhancement of Barry's self-
confidence and self-esteem, which promoted his successful
completion of five years of high school (i.e., grades 9 to
13) in four. Barry graduated from Vincent Massey High
School in June of 1974 and received an Ontario Scholarship
for academic achievement.
Leaving the air-polluted auto-town of Windsor for the
glamor of big city life, Barry enrolled at the University
of Toronto to study what makes people tick. Being
influenced by the strongly behavioristic psychology
department, Barry was biased against the use of
hypothetical constructs and drawn to physiological
143

144
processes in the study of the interdependency of the brain
and the body. One day, while present at a noon-hour brown-
bag seminar, Barry experienced his first taste of the
excitement of the proverbial pursuit of knowledge, which
was to influence his research interests from then until the
present. At this seminar a film was presented by Dr. David
Jones visiting from the University of British Columbia.
Half the screen depicted the forcible immersion of the head
of a duck into a bucket of water, while the other half
displayed the polygraph recordings of the physiological
responses. Upon immersion Barry observed that the duck's
heart rate decelerated to about 10% of its resting level.
This was quite an amazing response, which he later
discovered was termed the diving reflex. Being in the
process of conditioning large-magnitude decelerations in
human adults and meeting with some difficulty, Barry
wondered if he could somehow utilize this reflex in humans,
since it represented a remarkable integration of
cardiorespiratory mechanisms. Ever since this day and
after his graduation from the University of Toronto, with
an Honors B. Sc. degree in psychology and zoology, Barry
has pursued the human dive reflex as it exists in human
adults and infants.
Barry received his M.S. degree from Ohio University in
1978 in psychophysiology and then transferred to the

145
University of Florida to study physiological psychology and
infants. He was awarded a post-graduate scholarship from
the Natural Science and Engineering Research Council of
Canada in 1981, which along with a fellowship from the
Graduate School supported his final years of graduate
work. In April of 1983 Barry's study of infants expanded
into a personal interest, with the birth of his and his wife
Debbie’s first son, Joseph Henry. Joseph attempted to
participate in this study but with true wisdom (probably
inherited from his father) decided that he preferred to
sleep in his own bed rather than in the sterile confines of
the laboratory. Under the astute aegis of Dr. W. Keith
Berg, Barry received his Ph.D. in August of 1984 with an
emphasis on cardiovascular and developmental physiological
psychology. Barry will continue his study of the central
neural basis for cardiorespiratory physiology as a
postdoctoral research associate with Dr. Neil Schneiderman
at the University of Miami in Coral Gables, Florida.

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree Doctor of Philosophy.
r ~~
z*
sO’. .4.
W. Keith Berg, Chairman
Associate Professor of Psychology
I certify that X have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Merle E. Meyer
Professor of Psychology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in/Scope and quality,
as a dissertation for the degreje'^oT .Doctor of Philosophy.
Associate irofesKir
Psychology / /
I /
I certify that I have read this sxudy and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Philip Posner
Associate Professor of Physiology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation an is fully adequate, in scope and quality,
as a dissertation for the degre^, of Doctor of Philosophy.
Neil Row
Associate Professor of Psychology

This dissertation was submitted to the Graduate Faculty of
the Department of Psychology in the College of Liberal Arts
and Sciences and to the Graduate School, and was accepted
as partial fulfillment of the requirements for the degree
of Doctor of Philosophy.
August, 1984
Dean, Graduate School