Citation
Perinatal predictors of cognitive competence at age four

Material Information

Title:
Perinatal predictors of cognitive competence at age four
Creator:
Goldberg, Deborah Cole, 1952-
Publisher:
[s.n.]
Publication Date:
Language:
English
Physical Description:
ix, 72 leaves : ; 28 cm.

Subjects

Subjects / Keywords:
Birth weight ( jstor )
Child psychology ( jstor )
Demography ( jstor )
Gestational age ( jstor )
Infants ( jstor )
Intelligence quotient ( jstor )
Modeling ( jstor )
Mothers ( jstor )
Multiple regression ( jstor )
Obstetrics ( jstor )
Child development ( lcsh )
Children -- Intelligence levels -- Effect of perinatal factors on ( lcsh )
Cognition in children ( lcsh )
Curriculum and Instruction thesis Ph. D
Dissertations, Academic -- Curriculum and Instruction -- UF
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1987.
Bibliography:
Bibliography: leaves 68-71.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Deborah Cole Goldberg.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Deborah Cole Goldberg. 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:
030543337 ( ALEPH )
17497247 ( OCLC )

Downloads

This item has the following downloads:


Full Text















PERINATAL PREDICTORS OF COGNITIVE
COMPETENCE AT AGE FOUR



By

DEBORAH COLE GOLDBERG





















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 1987




























Copyright 1987

by

Deborah Cole Goldberg
























This work is dedicated to my parents,

Mel and Mary Cole,

who with great love, encouraged me to create my own opportunities.















ACKNOWLEDGMENTS



An effort such as this one requires the work and

cooperation of many individuals. First, and most important, in this group of caring friends is my family. My husband, Richard, gave many months of unfailing support not only during the writing of this dissertation, but also through the years of graduate school preparation. I am grateful to my daughter, Rachel, for her tolerance during my long absences from home and for her encouragement. My daughter Jessica, now two, can only be described as inspirational. I am also fortunate to have the support of a loving extended family.

My professional family also deserves credit. This group includes my colleagues at Sacred Heart Hospital, Pensacola, Florida, who have contributed so much to both my professional and personal growth. The "worker bees" who lead the Florida Consortium of Newborn Intervention Programs, especially Linda Stone, helped me persevere and provided a knowledgeable sounding board.

Special thanks must also go to the following members of my academic family: Athol Packer, my chairman, for his wisdom, patience, and tenacity; Michael Resnick, for his fine example as a pioneer in the field of early




iv









intervention; Sam Mathews, for his crisis management skills, his scholarship, and especially his friendship; Dorene Ross, for her keen editorial eye; and Janet Larsen, for her sharing of her vast understanding of young children.

Finally, the technical expertise of several people was instrumental to the completion of the research and manuscript. Thanks go to Juanita Nelson for help with data collection and Mark Littlefield for assistance with data analysis.








































V

















TABLE OF CONTENTS


Page

ACKNOWLEDGMENTS ................. ...................... iv

ABSTRACT..................................................... viii

CHAPTERS

I INTRODUCTION................................... 1

Purpose of the Study ............................ 4
The Sample..................................... 6
Research Questions. ............................... 6

II REVIEW OF RESEARCH... .......................... 8

Overview ....................................... 8
Three Models for Studying Outcome............... 11
The Research in Retrospect ..................... 29

III METHODOLOGY.................................... 33

Pilot Study .... ................................ 34
The Subjects................................... 35
Instrumentation . ................................ 39
Procedures........ ......... .................... 40
Data Analyses.................................. 42
Summary ................... ................... 44

IV RESULTS........................................ 45

Question One: Relationship Between the
Independent Variables and IQ Scores........... 45
Question Two: Contributions of the Independent
Variables to Infant Development .............. 47
Question Three: Prediction of IQ Scores from
Bayley Scores................................. 52
Question Four: Differences Between Follow-Up
and No-Show Groups on Medical Variables...... 52
Question Five: Differences Between Follow-Up
and No-Show Groups on Demographic Variables.. 55
Summary................. ......................... 57






vi










Page

V DISCUSSION .......... .................................. 58

Contribution of the Independent Variables to
Developmental Outcome........................ 58
Prediction of IQ Score from Bayley Scores ...... 60
Differences Between Follow-Up and No-Show
Groups......................................... 61
Comparison to Other Studies...................... 63
Summary ..................... .................. 65

REFERENCES............................................ 68

BIOGRAPHICAL SKETCH................................... 72














































vii










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


PERINATAL PREDICTORS OF COGNITIVE COMPETENCE AT AGE FOUR

by

Deborah Cole Goldberg

August 1987

Chairman: Athol B. Packer
Major Department: Instruction and Curriculum

The purpose of this study was to investigate the relationships among a set of obstetric, perinatal, and demographic variables and developmental outcomes for a sample of 77 infants treated in a regional intensive care nursery and followed over 4 years. The relationship between infant test scores IQ at 4 years of age was also examined. In addition, the characteristics of a sample of 65 children who did not return for follow-up were compared to those who were followed in order to identify differences which may influence development.

The independent variables of interest in this study were mother's gravidity and number of previous preterm births, infant's birth weight, gestational age, Apgar scores, respiratory distress, apnea, hyperbilirubinemia, days hospitalized, inborn vs. outborn status, sex, race, urban vs. rural residence, family income, and parents' age and education. Outcome measures included the Bayley Scales of Infant Development at 6 months, 1 year, and 2 years and the Stanford-Binet Intelligence Scale at 4 years. The



viii










relationships between early risk factors and developmental outcome were investigated utilizing multiple regression analyses.

Analysis of results indicated that the demographic variables of maternal education and race contributed significantly (p < .05) to the prediction of IQ score at age

4. At earlier ages, the medical variables of days hospitalized, gestational age, Apgar scores, and inborn vs. outborn status were predictive of Bayley mental and motor scale scores. Additionally, paternal education was included in the regression models for mental scale scores at 6 months, 1 year, and 2 years.

Another question addressed in this study was the

relationship between infant test scores and preschool IQ score. The results of multiple regression analyses indicated that mental scale scores at 6 months and 2 years and motor scale scores at 2 years were moderately predictive of IQ (R2 = .47).

A comparison of the sample participating in follow-up to a sample which chose not to participate resulted in significant differences (p < .05) on respiratory distress, inborn vs. outborn status, apnea, maternal age, race, and income level. Analysis of these results indicated that the no-show group was at a potential disadvantage for appropriate development.








ix
















CHAPTER I
INTRODUCTION



The relationship between developmental status in infancy and later school performance is one which has interested behavioral scientists for many years. The emergence of the growing population of intensive care nursery (ICN) graduates has increased interest in this issue. There is concern that some of these premature and sick newborns are at risk for future developmental problems, and many researchers are concerned with identifying which medical and environmental variables may be associated with later problems. Although numerous studies have been conducted which assess the relationships between a wide range of medical and demographic variables and subsequent developmental indices, findings have been inconsistent (Caputo, Goldstein, & Taub, 1981; Cohen, Parmelee, Beckwith, & Sigman, 1986; Francis-Williams & Davies, 1974; Holstrum, 1979; Hunt, 1981; Kumar, Anday, Jacks, Ting, & DelivoriaPapadopoulas, 1980; Siegel, 1982). Further, the characteristics of the populations studied have varied a great deal. The intent of the present study was to (a) extend existing research by utilizing a sample population of intensive care nursery (ICN) graduates that was heterogeneous in terms of medical variables, but was largely



-1-







-2


white and from the rural South; and (b) determine if differences in sample population characteristics might be a source of some of the variability in findings.

Interest in the prediction of outcome for at-risk infants was stimulated when premature babies first were cared for in institutions such as the nursery established by Dr. Julian Hess in Chicago in 1922 (Francis-Williams & Davies, 1974). This was the forerunner of the present day ICN. Large-scale longitudinal follow-up studies, many of them retrospective, began appearing in the 1940s and gradually increased through the 1950s and 1960s. Since 1970, many researchers from a variety of disciplines have been working in this area, as is evident by the studies published in a wide variety of medical and behavioral publications. A review of 22 studies of premature infants from around the world demonstrated that until 1960, mortality rates and the incidence of major handicaps were high. Since that time, chances of survival for premature infants weighing less than 1500 grams at birth have increased dramatically due to better understanding of the physiology of prematurity and advances in medical technology (Stewart, Reynolds, & Lipscomb, 1984).

One underlying concern shared by health and educational professionals is the potential increase in the proportion of handicapped individuals due to the decrease in the mortality rate for high risk infants. Analysis of the data gathered in recent years through the Florida Regional Perinatal







-3


Intensive Care Centers program suggests that decreasing mortality has not resulted in increasing morbidity among survivors by birth weight categories (Resnick, Bauer, Cupoli, Ausbon, & Evans, 1983), a finding common to many other studies (Fitzhardinge, 1984; Kafatos & Pantelakis, 1982; Kumar et al., 1980; Pape, Buncic, Ashby, & Fitzhardinge, 1978). However, further analyses are needed to refine the approaches for quickly and accurately identifying those infants most likely to become developmentally disabled or who will require special education services. One of the goals implicit in research in this area is the early identification of the highest risk infants for timely provision of services. A second goal is the identification of those variables most closely linked with disabilities so that medical procedures may be revised and more direct preventive measures be taken. Widespread attention has been directed to these concerns in the face of increasing health care and educational costs and shrinking resources.

Over the past 20 years, the methodology used in followup studies of intensive care nursery graduates has become more complex. Earlier researchers often utilized a singlefactor model, correlating gestational age, birth weight, or other medical variables with some outcome measure, usually IQ (Rubin & Balow, 1979). Later, groups of variables were analyzed using more sophisticated statistical techniques, but still focused on obstetric and neonatal medical events,







-4


as typified by the early work of Knobloch and Pasamanick (1960).

One trend among more recent studies has been to include a broader range of reproductive, perinatal, and demographic variables and to utilize a variety of outcome measures reflecting discrete areas of development. Further, the relative importance of demographic characteristics such as socioeconomic status and age and education of parents is now recognized and the interpretation of some research findings suggests that environmental factors are more influential than perinatal events as the child reaches school age (Broman, Nicholas, & Kennedy, 1975; Hunt, 1981; Sameroff, 1982; Siegel, 1982). Another element of recent research has been an extension of follow-up into early school age investigating school achievement as well as the incidence of major handicaps (Aylward & Kenny, 1979; Caputo et al., 1981; Francis-Williams & Davies, 1974).



Purpose of the Study


The purpose of the present study was to investigate the relationships between a set of obstetric, perinatal, and demographic factors and the cognitive and psychomotor development of a group of ICN graduates. The relationship between infant test scores and IQ scores at 4 years of age was also examined. In addition, the characteristics of a second sample of ICN graduates who did not return for







-5


follow-up were compared to those who did participate in order to identify possible differences.

The scope of this study was defined in part by the

state-wide follow-up effort of Florida's Regional Perinatal Intensive Care Centers program, developmental evaluation component. Staff of this component provide periodic developmental and pediatric evaluations to a cross-section of intensive care nursery graduates beginning at 3 months of age and continuing through 4 years of age. The children in the present study represent a sample of this population drawn from one regional intensive care nursery.

The chief limitation of this study was that it is ex post facto in design, utilizing chart review for data collection. The potential problem with the reliability is offset somewhat by the fact that the hospital and developmental evaluation clinic staff were consistent throughout the time period in question and were available to this researcher for consultation concerning the present study.

The overall goal of the study was to contribute to the body of research concerning the development of high risk infants, particularly with regard to prediction of future disability from data available at the time of hospital discharge. In practical terms, this type of information might be helpful in the timely and effective use of resources needed to maximize the developmental potential of individual children.







-6


The Sample


The sample for this study was drawn from those infants cared for in the Intensive Care Nursery at Sacred Heart Hospital, Pensacola, Florida, between October, 1977, and February, 1980. The sample includes both infants who were inborn and those transported to the ICN following birth. Those who completed 4 years of follow-up were eligible for participation in the study. Since all ICN graduates were invited to participate in the 4-year developmental follow-up during the years in question, the study sample was nonhomogeneous in terms of birth weight, gestational age, and perinatal complications. The sample was predominantly white and middle class and a significant number resided in rural communities. These three demographic characteristics are somewhat unusual compared to other study populations, particularly those in other areas of Florida's Regional Perinatal Intensive Care Centers program. For comparison purposes, a second sample of infants was identified who were also treated in the ICN during the same time period as the follow-up group, but who chose not to participate in follow-up.



Research Questions


The first three questions of interest in this study

were addressed through an examination of the relationships between 19 independent variables and 7 dependent variables







-7


for a sample of ICN graduates. Two additional questions addressed possible differences between the sample Which participated in follow-up and another sample of ICN graduates which did not. The questions were

1. Is there a relationship between a set of obstetric, perinatal, and demographic variables and IQ at 4 years of age?

2. Is there a relationship between a set of obstetric, perinatal, and demographic variables and developmental indices at 6 months, 1 year, and 2 years of age?

3. Are developmental indices at 6 months, 1 year, and

2 years predictive of IQ at 4 years?

4. Are there differences in obstetric and perinatal

variables between those subjects who participated in followup evaluations and those who did not?

5. Are there demographic differences between parents who brought their children for follow-up evaluations and those who did not?

The patterns of relationships observed in this study between the independent variables and the outcome measures were compared to results of studies of other types of high risk populations. Since the sample population in this study differs a priori from other study populations, only descriptive data pertaining to the results are presented.
















CHAPTER II
REVIEW OF RESEARCH



Overview


The majority of research pertinent to this study of perinatal predictors of intellectual status at age 4 has been published or reviewed over the past 10 years. Studies of follow-up of high risk infants have become more prevalent recently. In general, the older studies were based on a single factor approach to prediction of outcome, while more recent research indicates a trend toward multifactor predictors of developmental status. This chapter includes a brief historical background of this complex and rapidly growing field of study, as well as a more detailed exploration of a number of recent studies.

Several researchers have provided interesting

historical overviews of the follow-up research on infants born at risk. Francis-Williams and Davies (1974) described a review published by Benton in 1940 of 30 studies conducted between 1911 and 1940. Benton identified several unsatisfactory elements in the studies including small sample sizes, bias in selection, disregard of socioeconomic status, and lack of objective measures of IQ. The results






-8-







-9


of these studies did not indicate, however, the premature infants had lower IQ scores than full-term infants. Francis-Williams and Davies (1974) also described a study by Wiener, who in reviewing the literature of the next two decades, found Benton's conclusions were not confirmed. The studies reviewed by Wiener showed decreased IQ for low birth weight infants, with the findings of three major studies that IQ scores decreased as birth weight decreased.

Bee and her colleagues described a resurgence of

longitudinal studies of low birth weight infants in the late 1950s and early 1960s (Bee, Barnard, Eyres, Gray, Hammond, Spietz, Snyder, & Clark, 1982). These studies included descriptions of the relationships among prenatal and perinatal variables, home environment, and later outcomes for children. Bee et al. (1982) identified three basic conclusions about the relationships between perinatal status and later outcome. First, medical variables such as birth weight or anoxia had demonstrated small but significant relationships with later cognitive and motor development (Smith, Flick, Ferris, & Fellman, 1972). Second, the effect of the medical variables appeared to be mediated by characteristics of the child's environment (Werner, Simonian, Bierman, & French, 1967). Third, in most predictive studies, the best predictor of later cognitive functioning was not medical or perinatal status, but the level of the mother's education (Smith et al., 1972).







-10


Kafatos and Pantelakis (1982), in reviewing a number of prospective and retrospective studies of pregnancy and infancy in many areas of the world, identified a set of risk factors correlated most strongly with perinatal morbidity and mortality. The most critical risk factors according to their review are related to age of mother, parity, race, previous fetal loss, medical care, poverty, unwanted pregnancy, education of the mother, multiple births, and maternal morbidity.

There now appears to be a consensus among most

researchers that some combination of obstetric, perinatal, and demographic or environmental factors needs to be studied in relation to later developmental outcomes for the population of high risk infants. Because the chance of healthy survival has tripled for these infants since 1960, an increasing population of these children is available for study. Bee et al. (1982) noted that the research climate now exists for more detailed, short-term, longitudinal studies of high risk infants in order to uncover causal links between perinatal indicators, the child's environment, and later cognitive functioning. An emerging goal is the eventual application of this knowledge of causal links to the early identification and screening of children at high risk for later developmental problems so that effective interventions can begin at the earliest opportunity.






-11


Three Models for Studying Outcome


Most of the recent research has focused specifically on defined high risk populations. These high risk infants generally have been cared for in an intensive care nursery and often the populations selected for study are very low birth weight. A number of these studies are reviewed in detail because they are representative of emerging trends in research. Sameroff's (1982) conceptualization of early influences on development is used to provide a framework for discussion of these studies.

Sameroff (1982) identified three models for explaining the relationship between early risk factors and developmental outcome. These are (a) the single factor model, with an emphasis on either constitutional or environmental factors; (b) the interactional model, combining constitutional factors additively with environmental supports; and (c) the transactional model, wherein development results from a continual interplay between a changing organism and changing environment.



Single Factor Model


As mentioned previously, the single factor model has

not been widely used in recent studies of high risk infants. When this approach is utilized, however, medical variables are typically chosen. A primary example of this approach is found in the work of Littman and Parmelee, who are






-12


responsible for a major component of the research conducted regarding clinical predictors of outcome. In a recent study (1978), these researchers expanded on their initial work in order to determine if pediatric complications occurring in the first 9 months of life were more predictive of outcome than perinatal variables. The study included 126 preterm infants followed through the UCLA Center for Health Sciences. No years of birth or follow-up were identified in the report of the research. For Littman and Parmelee's sample the mean birth weight was 1,927 grams and the mean gestational age was 33.1 weeks. Four quantitative scales were used to describe the infants in the study: the obstetric complications scale, the post-natal complications scale, and two pediatric complications scales (1 to 4 months, 4 to 9 months). The obstetric and the post-natal complications scales included measures of physical development, health, behavior, congenital anomalies, and neurological and sensory handicaps. These four medical scales were scored in a summary fashion, with all items receiving equal weight in scoring. All infants were evaluated beginning at their expected date of birth and continuing through 2 years of age. All evaluations were corrected for prematurity and included a newborn neurological examination at term, the Gesell Developmental Tests at 4, 9, and 24 months of age, and the Bayley Scales of Infant Development at 18 and 25 months of age. No demographic measures, including socioeconomic status, were







-13


included in this study. The correlations of medical events scales to outcome measures showed that the pediatric complications scale for 4 to 9 months related most significantly to later performance for the entire sample population. It is important to note the frequency of illness was high during the 4 to 9 months period. By age 9 months, more than 80% of the infants had experienced some type of medical problem. These authors felt that early recognition of developmental disability is quite important and that assessment of health during later infancy should not be overlooked in predicting outcome.

Kumar et al. (1980) studied a sample population of

smaller infants, less than 1,250 grams birth weight, born 1974-1977 and cared for in the Perinatal Center of the University of Pennsylvania Hospital. They followed 50 of 60 survivors for one year to compare the outcomes for this group of high risk infants to findings in other studies where all infants were outborn. At 1-year follow-up these researchers found 3 infants had major neurological problems;

2 infants had retrolental fibroplasia, and severe developmental delay (Gesell Developmental Tests DQ less than 80) was documented in these 5 plus 2 other infants. The incidence of handicap for this sample was 14%. The 7 handicapped infants were then compared to the remainder of the sample on 16 variables using two-tailed t tests for the quantitative variables (birth weight, gestational age, 1and 5-minute Apgar scores, initial pH, initial temperature,







-14


time to regain birth weight, peak bilirubin, hypoxia, and duration of hospital stay), and chi square for qualitative variables (sex, mode of delivery, weight for gestational age, apnea, indication for mechanical ventilation, patent ductus arteriosis, and seizures). Significant between group differences were found only for mode of delivery and duration of hospital stay.

In discussing their results, Kumar et al. (1980)

pointed out that there was a marked increase in survival for infants born less than 1,250 grams, from 33.3% in 1974 to 63.2% in 1978. These figures suggest changes in neonatal management over this time period which may have introduced a bias into the study. The authors noted that it is difficult to compare the results of their study with those of previous researchers because of differences in aspects of patient care as well as racial and socioeconomic differences. Although these results were based on fairly short-term follow-up, the authors believed their findings are encouraging in the area of prognosis for very low birth weight infants in that infants born and treated in a perinatal center, including high risk obstetric management, had fewer medical complications and generally better developmental outcomes as compared with transported neonates.

Another example of the single factor model was a study of first year developmental outcome for premature infants







-15


conducted by Rice and Feeg (1985). In this retrospective study, 57 records were selected from the records of a developmental evaluation clinic at a large teaching hospital. The time period in which the infants were born was not indicated, neither were the criteria for inclusion in the follow-up clinic. In order to review the clinical data, the Categories of Risk Index (CRI) was developed for this study to indicate the number of complications in the perinatal period. The CRI was based on the Littman-Parmelee risk scale. No demographic factors were considered. The Bayley Scales of Infant Development were used as the outcome measure, using the scores obtained closest to 1 year of age. It is notable that 31% of the scores were obtained in the 26 month range. The only criterion noted for inclusion in the study was that the infant was born at 38 weeks gestation or less.

Three hypotheses were proposed for this study: 1. Differences in the Bayley Scales of Infant

Development, using both Mental (MDI) and Motor (PDI) scale scores, would be predicted by the risk status (high, low) and birth weight for gestational age (under or over).

2. Differences in MDI and PDI would be predicted by

gestational age, birth weight, deviation from expected birth weight, and perinatal risk status.

3. The combination of perinatal risk factors would predict MDI and PDI.







-16


Because several hypotheses were of interest in this study, a range of statistical analyses were performed. A

2 x 2 ANOVA was used to test hypothesis #1 with the finding that there were no differences in MDI and PDI between high and low risk groups for over and under birth weight or gestational age groups. For hypothesis #2, multiple regression analyses indicated that combined factors of gestational age, birth weight, deviation birth weight, and total risk were related to uncorrected MDI and PDI. When the Bayley scores were corrected for gestational age, no factors were predictive. Pearson correlation coefficients were computed to analyze the relationship between agecorrected and uncorrected Bayley scores and the eight categories on the risk index. No categories were significant for the uncorrected MDI. The corrected MDI was significantly correlated with surgery in the predicted direction. The uncorrected PDI was negatively correlated with seizures, anomaly-non-infectious illness, and ventilator assistance. The category of seizures was also strongly correlated with corrected PDI.

The most encouraging implication of this study is that infants with high numbers of post-natal risk factors may not necessarily experience more developmental delays than less ill infants. Limitations mentioned by the authors include the research design, including clinical records review for data collection, and a potentially biased sample of preterm infants.







-17


The Interactional Model


The work of a number of researchers can be described according to the interactional model, which explains the relationships between risk factors present in infancy and developmental outcome in terms of a combination of clinical and environmental factors. These studies typically include factors related to family status and the environment as well as the obstetric and neonatal medical events characteristic of the single factor model. Three examples of this approach are found in studies by Ramey et al. (1978), Pape et al. (1978), and Ross, Schechrer, Frayer, and Auld (1982).

Ramey and a group of researchers at the University of North Carolina -at Chapel Hill were interested in-the practicality and effectiveness of using birth certificate information as a mechanism to identify children who were likely to need special education services beginning in the first grade. Their sample population was a group of 1000 first grade students randomly selected from 20 counties in North Carolina. This report of the study does not indicate the birth year or years of these children.

The four outcome measures utilized in this study were the Peabody Picture Vocabulary Test, the Test of Basic Experience, the Developmental Test of Visual-Motor Integration (Beery), and the Myklebust Pupil-Rating Scale. The independent ratings obtained from birth certificates included race, sex, birth order, birth weight, number of







-18


weeks gestation, and legitimacy. Information concerning the mothers was also obtained from the birth certificate, including age, educational level, the month prenatal care began, whether there were previous births now dead, and whether there were any previous fetal deaths.

Multiple regression analyses were utilized to assess the relationship between the predictor variables and the outcome variables. To compare children who performed relatively poorly to those performing at or above the mean, multiple regression analyses were performed on the birth data.

Ramey et al. (1978) found that for the sample

population as a whole, the most predictive characteristics of educational and psychological status at first grade were race and mother's educational level. The results of the multiple regression analyses suggested that several variables significantly discriminated between moderate and low risk children on each of the four criterion variables. These were birth order, education of the mother, birth weight, month prenatal care began, race, and legitimacy. Although causality cannot be implied here, the researchers stated that this study demonstrates it is feasible to identify children utilizing birth certificate data who are likely to need special education services before and during grade school.

Pape et al. (1978) reported on a group of infants with birth weights less than 1000 grams born in 1974. Their






-19


concern, still prevalent among researchers, was the potential relationship between decreasing mortality and increasing morbidity. They studied 46 survivors of 97 premature infants transported to the Hospital for Sick Children in Toronto. Follow-up clinic visits were scheduled at 3 month intervals for the first year and at 18 and 24 months. Each clinic visit included a physical and neurological examination, measurement of growth parameters, and administration of the Bayley Scales of Infant Development corrected for gestational age. Hearing screening and eye examinations were also performed. The socioeconomic status was determined on the basis of the father's occupation and education. Of the 42 infants who received psychometric testing at 18 and 24 months, 13 infants (30%) had a diagnosed significant handicap by 18 months of age including 4 who had severe neurologic defects and 9 others who were developmentally delayed according to the results of the Bayley scales (mean MDI 89.2). To study the relationship between neonatal course and developmental outcome for these infants, 13 handicapped infants were compared with the remaining infants in the sample. Quantitative variables were analyzed by t tests and the'chi square test was used for qualitative variables. Significant associations were shown with the following: birth weight, acidemia, intracranial hemorrhage, and seizures. The variables of complicated pregnancy or delivery, asphyxia, respiratory distress, apnea, prolonged ventilation, or






-20


socioeconomic status were not associated with later delays.

In view of the fact that the 30% incidence of

significant handicap was unusually high, the authors pointed out a number of atypical characteristics of the study sample. First, all infants were outborn and no infant was refused on the basis of pretransport condition. Fifty-nine percent of the survivors suffered severe birth anoxia. There was also a high incidence of severe cold stress and acidemia related to transport. This high risk group also has some very positive characteristics. Ninety-five percent of the intensive care nursery survivors in this group remained in the follow-up program. Seventy-five percent of these were from families in the middle or upper socioeconomic level and living in intact two-parent homes. Early referrals to infant stimulation programs were also made for children showing a delay in any area.

These researchers identified a very positive finding

related to their study. According to their research, before 1970 approximately 75% of infants whose birth weight was less than 1000 grams died and only 15% survived as normal children. In this population born in 1974, 53% died and 33% of the remainder survived without significant handicap. In response to the initial question about the relationship between decreasing mortality and morbidity, these authors found that while mortality decreased 25%, there was a twofold increase in the number of normal survivors and little







-21


change in the absolute number of children who were handicapped.

Ross et al. (1982) studied a group of infants born at or less than 1500 grams in order to assess the predictive value of multiple demographic, perinatal, and neurobehavioral variables for 1-year outcome. The infants in this study were born in an 18-month period beginning March 1978 and admitted to the Perinatal Center of New York Hospital-Cornell Medical College. One hundred two of 120 consecutive survivors whose birth weight was appropriate for gestational age were evaluated at 1, 3, 6, 9, and 12 months corrected ages.

Outcome measures included Bayley scores (MDI and PDI), the presence or absence of cerebral palsy (CP), and pediatrician's rating (normal, suspect, delayed). Developmental status was based on a composite rating in three categories: (a) normal (85 minimum score MDI and PDI, normal by the pediatrician, or free of CP); (b) suspect (7184 MDI or PDI, suspect by the pediatrician, or mild CP); and

(c) abnormal (70 or less MDI or PDI, abnormal by the pediatrician, or moderate or severe CP).

According to this composite system, 52% of the sample was evaluated as normal at one year, 25% suspect, and 23% abnormal. The results of univariate F-tests suggested the following perinatal variables were significant: 1-minute Apgar, patent ductus arteriosis, seizures, intraventricular hemorrhage, sepsis, maximum oxygen required, pneumothorax,







-22


hypernatremia, and duration of stay. No demographic variables (SES, race, sex, maternal age) were found to be related to 1-year outcome. Results of the Amiel-Tison neurobehavioral assessment at 3 and 9 months were most predictive of outcome and added significantly to the predictive value of the perinatal variables overall.

Although the authors acknowledged that development at one year of age may not predict later development, they conjecture that measures at one year may be indicative of underlying neurological impairment before it is influenced by the environment. The demographic characteristics of this population were significant in that the majority were from middle to upper middle class backgrounds and were white. Given the relatively high incidence of handicap, it should be noted that 45% of the sample were outborn and transported to the perinatal center.



The Transactional Model


Ross et al. (1982) in reviewing the literature noted that "as children reach school age, however, environmental factors such as socioeconomic status, parental IQ, and mother's education become far more powerfully related to verbal and visual-motor abilities than perinatal and infant behavioral variables, which may still be associated with aspects of later outcome but to a lesser extent" (p. 318). This viewpoint supports the transactional model of








-23


development, which includes the wide variety of factors of the interactional model and takes into account the plasticity of both the child and the environment over time. Studies of high risk infants which utilize repeated measures or observations over time may be classified by the transactional model.

Two such studies conducted outside the United States were reported by Francis-Williams and Davies (1974) and by Silva et al. (1984). Francis-Williams and Davies (1974) followed 105 of 120 children weighing less than 1500 grams at birth and treated at Hammersmith Hospital, London, between the years of 1961 and 1968. The sample included both inborn and transported infants. Because the time period in which these children were born included significant changes in neonatal care, initial analyses were made of the children in two groups (1961-1964 and 19651968). These analyses revealed, however, that dividing the subjects according to birth years was not helpful.

The children were evaluated at between 4 and 12 years

of age utilizing the Wechsler Preschool and Primary Scale of Intelligence or the Wechsler Intelligence Scale for Children. Reading ability was also assessed. Children older than 5 years were also assessed using the Bender Gestalt Test for Young Children. Correlations were examined between neonatal illness, social class, head size, and developmental outcome as measured by the standardized tests. Children born small for gestational age (SGA) were analyzed






-24


separately from those whose birth weights were appropriate for gestational age (AGA). The only significant correlation found related to neonatal illness was severe asphyxia in the four SGA infants in the study. Social class was found to be the most highly predictive variable. These researchers stated, "this bias and distribution toward the lower social classes is evident in most reports of children of very low birth weights" (p. 718). A significant difference in IQ was also found between infants whose head size was below the 50th percentile. The overall incidence of handicap in this sample was 12%.

These researchers, in comparing their results to

research of the previous 30 years, pointed out significantly improved outcomes for these high risk children overall. They cautioned, however, that one-Eifth of their sample had a performance IQ significantly below their verbal IQ with learning difficulties evident in those children who were school age. Early recognition of learning difficulties and careful follow-up is indicated for these children.

A study conducted in Dunedin, New Zealand, by Silva,

McGee, and Williams (1984) examined the relationship between perinatal problems and cognitive and behavioral development. A sample of 850 infants born between April of 1972 and March of 1973 was classified into three groups (preterm, SGA, and full-term normal birth weight) and compared on IQ at 3, 5, 7, and 9 years of age, and on parent and teacher behavior reports at 5, 7, and 9 years. The sample of 850 children







-25


was selected from 1,661 total births because all relevant information was available for these children. The authors noted that the follow-up group did not differ significantly from the remainder of the population in perinatal characteristics or socioeconomic status.

In contrast to other studies, no significant

differences were identified among the three groups for SES or maternal intelligence. Using an ANOVA, researchers found significant differences on intelligence test results at each age among the groups. The SGA group performed least well. The SGA group also had significantly more behavioral problems than the remainder of the sample according to the parent's behavior rating scale. Although the teacher's behavior rating scale indicated an increasing number of problems for all groups over time, there were no significant differences among the groups.

Several significant studies have come out of the

longitudinal research conducted by Caputo et al. (1981). Information regarding 10-year follow-up of a sample of children born on Staten Island, New York, between July of 1965 and January of 1969 has recently been published. The 64 children included in the follow-up were part of a larger group of 233 infants who were studied at one year. Half of this smaller sample consisted of infants prematurely born weighing less than 2500 grams and the other half of the sample was full term. Almost all of these children were white and middle class. The variables studied included 7







-26


demographic variables, 7 birth and obstetric variables, the sex of the child, prematurity, mother's discomfort during pregnancy and delivery, mother's child rearing attitudes, and mother's IQ. When these children reached the ages between 7 and 9 1/2 years they were evaluated using the Wechsler Intelligence Scale for Children-Revised (WISC-R) and the Bender Gestalt Test for Young Children.

Extremely thorough statistical analyses were made of these data including separate examinations of the WISC-R subtests, verbal IQ, performance IQ, and full scale IQ, as well as the Bender Gestalt results. Briefly, the authors found that in bivariate analyses, both mother's IQ and the social class factor correlated very significantly with verbal IQ and performance IQ in middle childhood. In this study, increasing family size was negatively correlated with IQ. Prematurity was the only birth complication independently related to cognitive development. Specifically, visual perception was negatively affected by prematurity.

The Cattell Infant Intelligence Scale and Gesell Development Tests conducted at one year were compared against the later assessments and were found to be generally poor predictors of later outcome. The Gesell personalsocial score, however, tended to contribute significantly to the WISC-R verbal and full scale IQ scores. The earlier analysis of birth and obstetric complications factors compared to development at one year showed delivery and






-27


related variables and complications to be predictive. These same factors were no longer predictive of outcome at 7 to 9 years, but the obstetric factor, indicative of family size, was significant.

Bee et al. (1982) conducted a longitudinal study of relatively healthy children which addressed several questions. The first was the usefulness of information about the child versus information about the environment. Another question related to the usefulness of infant mental tests for prediction of later IQ. A further aspect of the study was the predictive role of "ecological" family characteristics including maternal education, available social support, amount of life change, and mother's perception of infant. The possible differences in the predictive equations for mothers who differed in level of education was also considered. The authors believed the large sample size, the range of measures used, and the frequency of observations allowed them to appropriately address this wide series of questions.

The population studied consisted of 193 first born infants born during 1973 and 1974 at one hospital in Seattle. In order to achieve a sample including about half mothers who had'a high school education or less and about half who had more than high school maternal education was used as a blocking variable. A second blocking variable was the presence or absence of perinatal risk factors. Multiple births and infants with anomalies were excluded. The authors emphasized that the sample they selected was an






-28



unusually healthy group with only 3 infants with birth weights below 2500 grams and only 23 with Apgar scores of 6 or below at 1 or 5 minutes. The authors also noted that the group as a whole was fairly well educated and economically well off. These characteristics in combination with those previously described classify this sample as a generally low-risk group.

Children in this sample were followed through 4 years of age, with observations and assessments beginning in the eighth month of pregnancy and continuing through birth, 1, 4, 8, 12, 24, 36, and 48 months of age. Home visits were made at all but the prenatal, birth, and the 24-month age levels. A developmental assessment was conducted in a clinical setting at 13, 24, 36, and 48 months.

An expected result of this study was that mothers with more than a high school education had larger, more motorically mature newborns, with significantly higher mental test and language scores beginning at 24 months. A more enriched home environment with more facilitative teaching was evident from the earliest observations. No maternal educational differences were related to mental or language development before 24 months and no difference in psychomotor functioning were found at any age. It should be noted that the size of most correlations was small. The authors stated that no single measure of perinatal status, child outcome, family support characteristics, or







-29


interaction patterns accounts for very much of the variance in the IQ or language scores.

A second analysis was conducted combining variables in each of four clusters with a separate regression analysis for each outcome variable. By means of these analyses, 20% to 50% of the variance in IQ or language development could be predicted by measuring any of three characteristics: child's earlier test performance, maternal infant interaction and environment, or maternal education and family support. The authors pointed out the particular effectiveness of including demographic characteristics about the family in the analyses. They stated "it is possible to gather information at the time of an infant's birth that will tell us as much about his IQ or language as will either direct observation of parent infant interaction or direct measures of the child's cognitive or language development during the first year" (Bee et al., 1982, p. 1048).



The Research in Retrospect


The variety of approaches evident in the studies previously described is characteristic of the field of research concerning at-risk infants. These studies are indicative of the trend toward investigation of multiple predictors of outcome and the increasing complexity of the analyses used. The disparate methodologies and variety of findings suggest that no consensus exists, but the large







-30


number of studies from many professional disciplines indicates that this is a significant area of research.

Although the studies discussed are not directly

comparable, some general strengths and weaknesses in the research can be described. When the methods used to identify the sample population and collect the data were described, they seemed careful and thorough. Most subjects were drawn from the same geographic area and treated at one perinatal center, limiting treatment biases. The instruments utilized as outcome measures are well accepted, and most researchers stated whether or not age corrections for prematurity were made in scoring.

The primary weakness of these studies is related to the lack of generalizability of results. This, unfortunately, is inherent in the methods required to study a defined group of graduates from a specific intensive care nursery (ICN). Results cannot readily be compared across studies because of the initial differences in population as well as medical treatment. Instrumentation varies, as do the statistical analyses employed. Correlational analyses alone do not control for the possible interactions between variables. In general, the large numbers of variables and relatively small sample sizes decreased the power of the multiple regression analyses.

The results of most of these studies contain

indications that some medical and environmental factors are associated with developmental outcome. The relative







-31


significance of these categories seems to vary with the age of the child. A broad trend suggested here is that larger numbers of at risk infants are surviving with minimal or no handicap. The factors most related to healthy survival need to be identified along with those related to poor outcome.

According to Jane Hunt (1981), "the dynamics of the

interaction between the initial deficit, genetic potential, and environmental effect, and intellectual outcome are not well understood" (p. 331). The two major issues involved in continued studies of high risk populations concern the identification of significant developmental problems as early as possible, and the identification of the conditions that caused the problems. The state of the art of follow-up of high risk children is still far from the goal of reliable prediction of the consequences of high risk birth compounded by environment. However, as Sameroff (1982) stated, "the continued study of this problem may help us to find better descriptions of the dynamic processes by which early problems are overcome and later ones created" (p. 393).

Early prediction is certainly important for the care and evaluation of high risk infants. Ross et al. (1982) identified three reasons related to the need for further study of this area: (a) the identification of perinatal factors predictive of poor outcome can result in improved medical treatment and so decrease later problems, (b) early prediction may lead to early identification and






-32


intervention, and (c) identifying predictors can enhance the understanding of factors responsible for poor or good development of low birth weight infants.
















CHAPTER III
METHODOLOGY



This retrospective, ex-post facto study of outcome for infants born at risk is similar in design to others utilizing antecedent data to search for predictors of developmental delay or handicap (Hunt, 1981; Kumar et al., 1980; Littman & Parmelee, 1978; Ramey et al., 1978; Rice & Feeg, 1985). Modeled on the work of Siegel (1982), the current study includes an examination of the relationships between developmental status and a group of obstetric, perinatal, and demographic variables, including (obstetric) gravidity, number of previous preterm births; (perinatal) birth weight, gestational age, 1- and 5-minute Apgar scores, respiratory distress (4-point scale), apnea, hyperbilirubinemia, days hospitalized, inborn/outborn status; and (demographic) sex, race, urban/rural residence (according to HUD guidelines), income (according to Florida's Children's Medical Services categories), maternal age and education, and paternal age and education. Outcome measures included the Bayley Scales of Infant Development (mental and motor scales) at 6, 12, and 24 months adjusted for prematurity when indicated (gestational age less than 40 weeks), and Stanford-Binet IQs at 4 years. The sample,





-33-







-34


design, and procedures for data collection and analyses are described in this chapter.

As noted in Chapter I, the questions posed for this study were

1. Is there a relationship between a set of obstetric, perinatal, and demographic variables and IQ at 4 years of age?

2. Is there a relationship between a set of obstetric, perinatal, and demographic variables and developmental indices measured at 6 months, 1 year, and 2 years of age?

3. Are developmental indices at 6 months, 1 year, and

2 years predictive of IQ at 4 years?

4. Are there differences in obstetric and perinatal

variables between those infants who participated in followup evaluations and those who did not?

5. Are there demographic differences between parents who brought their children for follow-up evaluations and those who did not?



Pilot Study


In 1984, the author conducted a preliminary study using a small sample of infants (N = 43) representing three intensive care nurseries in Florida. Most of the predictor variables of interest in the present study were compared to Bayley scores at 2 years. Correlational analyses of the data indicated that maternal education was most






-35


significantly related to the cognitive scale scores. Gestational age and birth weight also were related to both the cognitive and motor scale scores, but these correlations did not reach statistical significance.

Because the sample for the pilot study was drawn from three intensive care nurseries, results may have been affected by differences in hospital treatment or testing procedures. The present study samples were drawn from one hospital to minimize the potential for this problem. The statistical analyses of the data for the present study utilized multiple regression analyses. The present study also extended follow-up to 4 years of age.



The Subjects


The large data base existing within the developmental evaluation component of Florida's perinatal program was the source of the basic information for this study. The families of all 110 children born between October, 1977, and February, 1980, who participated in the follow-up program for 4 years were contacted and 85 agreed to allow review of hospital records for this study. Such a large response rate is unusual and probably reflects the positive relationship between families and hospital staff developed in the ICN and over the 4 years of follow-up. Twins and children with congenital anomalies were excluded. Data sets were compiled for 77 of the subjects. All the children in this sample







-36


were treated at Sacred Heart Hospital's intensive care nursery and completed 4 years of developmental follow-up.

Sacred Heart's ICN is a regional referral center

serving the 17 rural counties of the Florida Panhandle. From 1978 through 1980, approximately 600 infants per year were admitted to the ICN and the average mortality rate was 10%. In the present sample, 58% of the infants were born elsewhere and transported to the ICN. This percentage is consistent with the total ICN admissions for that time period.

During the years indicated, all infants treated at this tertiary care center were invited to participate at no charge in the follow-up program. Of those included in this sample (N = 77), about half the families were middle class (53% with incomes greater than $12,000) and 44% resided in rural communities. Additionally, the average age and educational level of mothers were 26.2 and 12.8, respectively, and for fathers, 30.2 and 12.9. According to these data, this sample was at potentially less environmental risk due to factors such as low income and low education than others often described in the literature. The infant population in this study was 58% male and 86% caucasian. Slightly more than half (58%) were born at outlying hospitals and transported to the ICN. Respiratory distress was experienced by 69% of the infants (mild--28%, moderate--28%, severe--ll%). Fifty-four percent were treated for hyperbilirubinemia and 78% experienced apnea.







-37


Descriptive data pertaining to this sample of infants who participated in follow-up are reported in Table 1.


Table 1

Descriptive Data for Follow-Up Group



Standard
Variable N Mean Deviation


Gravidity 75 2.2 1.4 Previous Preterm Births 75 .12 .37 Birth Weight 77 2533 Grams 943.7 Grams Gestational Age 77 36 Weeks 4.5 Weeks Apgar--1 Minute 74 6.2 2.4 Apgar--5 Minutes 72 8.0 1.9 Days Hospitalized 77 26.8 30.6 Maternal Age 77 26.2 5.3 Maternal Education 77 12.8 2.3 Paternal Age 71 30.2 6.3 Paternal Education 71 12.9 2.6





Over time, 45% of the families invited to participate in the follow-up program did not participate. Nonparticipation may be indicative of parental resources and/or attitudes, so the characteristics of the families who chose not to attend the follow-up clinic were of some interest to this researcher. In order to identify possible differences, the follow-up sample was compared to a "no-show" sample







-38


(N = 64) of children born during the same time period on most of the predictor variables included in the primary analysis. Information regarding maternal education and paternal age and education was unfortunately not available for many of the no-show cases. In this sample, 60% of the infants were outborn, 62% were male, and 35% caucasian. Their families were primarily low income (66%) and lived in rural areas (66%). Respiratory distress was experienced by 58% of these infants (mild--34%, moderate--14%, severe--5%) Treatment for hyperbilirubinemia was required for 29%, and 14% experienced apnea. Descriptive data pertaining to this sample of infants who did not participate in follow-up are reported in Table 2.


Table 2

Descriptive Data for No-Show Group



Standard
Variable N Mean Deviation Gravidity 64 2.23 1.4 Previous Preterm Births 63 .03 .25 Birth Weight 62 2382 Grams 1133 Grams Gestational Age 63 35.3 Weeks 3.5 Weeks Apgar--1 Minute 63 6.6 2.4 Apgar--5 Minutes 61 8.3 1.5 Days Hospitalized 64 20.4 17.6 Maternal Age 64 23.5 5.9






-39


Instrumentation


Developmental outcome was measured at 6, 12, and 24

months using the Bayley Scales of Infant Development. This instrument is composed of two scales which measure mental abilities and psychomotor abilities of infants from birth to 30 months of age.

Several aspects of reliability were reported in the test manual (Bayley, 1969). Split-half reliability coefficients for the mental scale ranged from .81 to .93 and for the motor scale from .68 to .92. Results from a sample of 8-month-old infants were used to examine tester-observer and test-retest reliability. Mean percentages of testerobserver agreement were 89.4 for the mental scale and 93.4 for the motor scale. Mean percentages of test-retest agreement for the mental and motor scales were 76.4 and 75.3, respectively. Bayley stated that the percentage of tester-observer agreement was markedly higher than that for test-retest because the former is free from problems of the stability of behavior over time (p. 21).

Validity per se is not discussed in the manual. A report is included, however, of a study of the degree of correspondence between the Bayley MDI and the Stanford Binet IQ score. The coefficient of correlation between scores obtained on the two measures by the total group of children (age range 18-30 months) was .57, which Bayley believed indicates a substantial degree of agreement.






-40


The Stanford-Binet Intelligence Scale, Form L-M, was

used as a measure of the children's cognitive competence at

4 years of age. Reliability coefficients are reported to range from .83 to .91 for 2 1/2 to 5 1/2 year olds (Terman & Merrill, 1973). The reliability data were derived from the scores of the standardization group of 3,184 children and appear to be adequate evidence for the dependability of the scale. The item selection procedures used provide a high degree of internal consistency.

The manual includes reports of the percentages of

subjects passing each item and the biserial correlations of each subtest with the total score. Analysis of the correlations suggests that several specific kinds of abilities contributing to overall mental ability are sampled. These are verbal, nonverbal, and manipulative skills.

Evidence that Form L-M of the Stanford-Binet is valid

is based on the fact that the same type of tests are used as in the 1937 version of the instrument. "Because of the great amount of overlap and the careful selection of subtests to be used in the revision, the probability is high that the validity of the revision will be at least equal to if not greater than the 1937 version" (Balinsky, 1965, p. 832).



Procedures


Because the state-wide developmental evaluation

component of the Regional Perinatal Intensive Care Centers






-41


program originated as a research program, the data available for this study conformed to previously established protocols. Families were contacted by mail with an appointment data and time. If that appointment was not kept, one further attempt was made to re-schedule the appointment by mail and/or phone. All children were evaluated at 6, 12, and 24 months using the Bayley Scales of Infant Development (Bayley, 1969). Scores were adjusted for prematurity at 6, 12, and 24 months by subtracting the number of weeks short of full-term gestation (40 weeks) from the chronological age of the child. The Stanford-Binet Intelligence Scale, Form L-M (Terman & Merrill, 1973), was used to assess each child at 4 years of age.

Parents accompanied their children throughout the

evaluation process, which included a pediatric examination following the developmental assessment at each visit. Referrals were made as needed for further diagnostic and/or remedial services, such as opthamology, physical therapy, speech and hearing, and developmental programming. Parents were informed of the test results, and their concerns about the child's health and development were discussed. In this sample population, a number of children who had developmental delays received intervention services.

Three sources were utilized to accomplish data collection for this study. The primary source of information on obstetric and perinatal variables was the medical record. Records from the follow-up clinic provided






-42


demographic information and test scores. When information was missing, intensive care nursery weekly staffing notes were reviewed.



Data Analyses


The variety of analytical approaches used in this research is evident in current literature on the developmental outcome for at-risk infants. Some interpretations of data were based on correlations alone (Kumar et al., 1980; Littman & Parmelee, 1978; Pape et al., 1978). Some utilized only discriminant function analyses (Caputo et al., 1981; Ross et al., 1982; Silva et al., 1984), and others utilized t tests, correlational analyses, and multiple regression analyses in various combinations (Bee et al., 1982; Francis-Williams & Davies, 1974; Ramey et al., 1978; Rice & Feeg, 1985). The basic design of the present study was drawn from Siegel's efforts to develop a risk index useful in predicting outcome for an individual child, utilizing multiple regression in order to control for interactions among the predictor variables.

The first question, concerning the relationship between reproductive, perinatal, and demographic variables and cognitive competence at 4 years for a population of ICN graduates, was tested utilizing a forward stepwise multiple regression. The procedure accommodated the large number of independent variables (19) compared to the sample size






-43


by identifying those variables most strongly associated with the outcome measure. Stepwise regression analyses were also conducted on the complete set of independent variables for each previous assessment (6, 12, and 24 months) to investigate the second question. Because of the question of the stability of the results of the multiple regression analyses due to the large number of independent variables considered, a cross-validated correlation coefficient was computed for each of these analyses. Crocker and Algina (1986) described this statistical procedure as a means to investigate the accuracy of the sample prediction equation.

The third question regarding the relationship between Bayley test scores at 6, 12, and 24 months and StanfordBinet IQ at age 4 was investigated utilizing multiple regression analysis. The "min r" procedure was used in this case to test all possible regression models.

The fourth and fifth questions, addressing the

potential differences between the follow-up group and the no-show group, were investigated by computing chi squares for the dichotomous and categorical variables and t tests for the continuous variables which were significant in the regression analyses (p < .05). All analyses were executed using the Statistical Analysis System (SAS) (Barr et al., 1985).







-44


Summary


In summary, the data were collected and analyzed in order to assess the relationships between obstetric, perinatal, and demographic factors and the mental and psychomotor development of a group of ICN graduates who were generally considered "at risk" because premature birth, other medical problems, and/or environmental problems that may negatively influence later development. The ability of infant test scores to predict IQ at 4 years of age was also examined. In addition, the characteristics of a sample of children who did not return for follow-up were compared to those who did participate in order to identify differences which may influence development. The results of the analyses are described in Chapter IV.
















CHAPTER IV
RESULTS



Data regarding obstetric, perinatal, and demographic characteristics of a sample of intensive care nursery graduates were collected to examine the relationships between those factors and developmental outcome at intervals from 6 months through 4 years. These infants were considered at risk for developmental delay because of premature birth or other medical complications during the newborn period. Comparable data were collected for a second sample of infants also treated in the intensive care nursery but who did not participate in follow-up. The obstetric, perinatal, and demographic characteristics of the two samples were then compared to identify possible differences. The results of the statistical analyses of the data are presented in this chapter.



Question One: Relationship Between the Independent Variables and IQ Scores


The first question for analysis was "Is there a

relationship between a set of obstetric, perinatal, and demographic variables and IQ score at 4 years of age?" A stepwise multiple regression procedure was used to regress the IQ scores on variables representing gravidity and number



-45-







-46


of previous preterm births of the mother, birth weight, gestational age, 1- and 5-minute Apgar scores, respiratory distress, apnea, hyperbilirubinemia, days hospitalized, inborn/outborn status, sex and race of the infant, urban/rural residence, income, and age and education of the parents. The significant independent variables, their regression weights, and the squared partial correlations are listed in Table 3.



Table 3

Regression Weights and Squared Partial Correlations for the Three Variables Predicting IQ


Independent Regression Squared p*
Variable Weight Partial
Correlation


Maternal Education 4.764 .405 .0001 Race -13.573 .045 .0273 Respiratory Distress -9.042 .023 ns



R2 for Full Model = .473, p = .0001 R2cy = .44 (cross-validated correlation coefficient) N = 66

*p value for testing the hypothesis that the corresponding partial correlation is zero. ns = not significant, p > .05


Maternal education and race contributed significantly (p < .05) to the prediction of IQ score at age 4. Respiratory distress also contributed to the full model, but







-47


did not reach individual significance. The small difference between the R2 and the cross-validated correlation coefficient (R2cv) indicated that the former is a stable value (Crocker & Algina, 1986).



Question Two: Contributions of the Independent
Variables to Infant Development


The second question for analysis was "Is there a

relationship between a set of obstetric, perinatal, and demographic variables and developmental indices measured at

6 months, 1 year, and 2 years?" Stepwise multiple regression analyses were conducted for the full set of independent variables and Bayley scores at each age. The significant independent variables, their regression weights, and the squared partial correlations are listed in Tables 4, 5, and 6.



Six-Month Data


Days hospitalized, inborn vs. outborn status, and gestational age contributed significantly to the Bayley mental scale scores at 6 months of age. Father's educational level and 1-minute Apgar score also added to the prediction but were not significant individually.

Days hospitalized and urban vs. rural residence were

predictive of motor scale scores at 6 months. The 1-minute Apgar score contributed to the strength of the full







-48


regression model but did not reach significance. These data are summarized in Table 4.


Table 4

Regression Weights and Squared Partial Correlations for the Variables Predicting 6-Month Bayley Scores Mental Scale


Independent Regression Squared p*
Variable Weight Partial
Correlation


Days Hospitalized -.263 .1266 .0002 Inborn/Outborn -10.215 .1012 .017 Gestational Age -.983 .0326 .0467 Paternal Education 1.265 .0381 ns Apgar--1 Minute 1.669 .0407 ns R2 for Full Model = .34, p = .0002 R2cv = .28 (cross-validated correlation coefficient) N = 63

Motor Scale


Independent Regression Squared p*
Variable Weight Partial
Correlation


Days Hospitalized -.23 .1677 .0006 Urban/Rural -9.976 .0430 .0299 Apgar--1 Minute 1.514 .0333 ns R2 for Full Model = .24, p = .0009 R2c = .18 (cross-validated correlation coefficient) N = 63

*p value for testing the hypothesis that the corresponding partial correlation is zero. ns = not significant, p > .05







-49


One-Year Data


The results of a stepwise regression analysis of the independent variables on Bayley mental scale scores at 1 year showed a number of variables contributed significantly to the prediction. These were paternal education, gravidity, days hospitalized, and income. Gestational age (number of weeks in utero) also entered the regression model but did not reach individual significance.

The relationship between days hospitalized, income, inborn vs. outborn status, and the motor scale scores is reliable (p = .01), but relatively weak (R2 = .171). These regression analyses are summarized in Table 5.



Two-Year Data


Paternal education, hyperbilirubinemia, and apnea

contributed significantly to the prediction of mental scale scores at 2 years. Gestational age was also included in the full regression model but did not reach individual significance.

The multiple regression on the motor scale scores showed that previous preterm births, the 1-minute Apgar score, and days hospitalized were related to the 2-year scores. These data are summarized in Table 6.







-50


Table 5

Regression Weights and Squared Partial Correlations for the variables Predicting 1-Year Bayley Scores Mental Scale


Independent Regression Squared p*
Variable Weight Partial
Correlation


Paternal Education 2.03 .1154 .0087 Gravidity -2.8991 .0725 .0421 Days Hospitalized -.179 .0447 .0081 Income (Medium-High) -10.875 .0491 .0421 Gestational Age -.885 .0411 ns


R2 for Full Model = .32, p = .0003 R2cv = 22 (cross-validated correlation coefficient) N = 64

Motor Scale


Independent Regression Squared p*
Variable Weight Partial
Correlation


Days Hospitalized -.139 .0749 .0078 Inborn/Outborn Status -5.825 .0565 ns Income (Medium-High) -6.658 .0396 ns



R2 for Full Model = .171, p = .0101 N = 64

*p value for testing the hypothesis that the corresponding partial correlation is zero. ns = not significant, p > .05







-51


Table 6

Regression Weights and Squared Partial Correlations for the Variables Predicting 2-Year Bayley Scores Mental Scale


Independent Regression Squared p*
Variable Weight Partial
Correlation


Paternal Education 2.447 .1649 .0010 Hyperbilirubinemia -8.226 .0392 .0346 Apnea 15.397 .0373 .0126 Gestational Age -1.035 .0444 ns R2 for Full Model = .286, p = .0002 R2cv = 20 (cross-validated correlation coefficient) N = 64

Motor Scale


Independent Regression Squared p*
Variable Weight Partial
Correlation


Apgar--1 Minute 2.709 .1249 .0012 Days Hospitalized -.157 .1020 .0054 Previous Preterm Births 10.263 .0527 .0387


R2 for Full Model = .280, p = .0002 R2cy = .22 (cross-validated correlation coefficient) N = 65

*p value for testing the hypothesis that the corresponding partial correlation is zero. ns = not significant, p > .05







-52


Question Three: Prediction of IQ Scores from Bayley Scores

The third question was "Are any of the developmental indices measured at 6 months, 1 year, and 2 years of age predictive of IQ score at 4 years?" A forward stepwise multiple regression analysis was performed on the data using the "min r" option in order to consider every possible combination of independent variables. Of the three sets of Bayley scores considered, the single best predictor of IQ at

4 years was the mental scale score at 2 years (R2 = .36). The next variable to enter the regression model was mental scale score at 6 months, which increased the predictive power (R2 = .42). The best 3-variable model added the motor scale score at 2 years, with another increase in predictive power (R2 = .47).

The addition of PDI at 6 months and MDI and PDI scores at 1-year resulted in a minimal increase in the R2 coefficient. The results of these multiple regression analyses are summarized in Table 7. For the purpose of interpretation of the results, the means and standard deviations for the outcome measures are reported in Table 8.



Question Four: Differences Between Follow-Up
and No-Show Groups on Medical Variables


The fourth question posed for this study was "Are there differences in obstetric and perinatal variables between






-53


Table 7

Regression Weights and Correlation Coefficients for Prediction of IQ at Age Four



Independent Regression R2 for p*
Variable Weight Full Model


MDI--2 yr**
(best 1 variable model) .652 .368 .0001

MDI--6 mos .261 .0135 MDI--2 yrs .533 .0001
(best 2 variable model) .424

MDI--6 mos .338 .0018 MDI--2 yrs .630 .0001 PDI--2 yrs*** -.289 .0156
(best 3 variable model) .473

MDI--6 mos .299 .0757 PDI--6 mos .067 .6554 MDI--1 yr -.128 .4848 PDI--1 yr .137 .4827 MDI--2 yrs .697 .0001 PDI--2 yrs -.334 .0121
(best 6 variable model) .481



N = 70

*p value for testing the hypothesis that the corresponding
partial correlation is zero.

**MDI = Bayley mental scale score

***PDI = Bayley motor scale score






-54






Table S

Means and Standard Deviations for Outcome Measures



Variable N Mean Standard Deviation MDI--6 mos* 73 106 17.9 PDI--6 mos** 73 106.4 18.3 MDI--1 yr 75 105.2 17.5 PDI--1 yr 75 101 13.3 MDI--2 yrs 74 103.7 16.9 PDI--2 yrs 75 97.9 16.1 IQ--4 yrs 77 103.6 18.3



*Bayley mental scale score

**Bayley motor scale score







-55


those infants who participated in follow-up and those who did not?" The variables found in the regression models previously described and predictive of outcome were compared for the follow-up and no-show groups to identify possible differences. Chi square tests were computed for respiratory distress, inborn/outborn status, and apnea. Significant differences were found between the groups for all three of these variables. The t tests computed for the two groups resulted in no significant differences for gestational age or days hospitalized. These results are summarized in Table 9.



Question Five: Differences Between Follow-Up and
No-Show Groups on Demographic Variables


The fifth question for analysis was "Are there

demographic differences between parents who brought their children in for follow-up and those who did not?" To examine these relationships, chi-squares were computed for race and income level and a t test was computed for maternal age. Significant differences between the groups were found for all three of these variables. Unfortunately, information about parents' education and paternal age were not available for many subjects in the no-show group so no analyses were done. These data are summarized in Table 9.







Table 9

Differences Between Follow-Up and No-Show Groups



Follow-Up Group No-Show Group Statistic Independent Mean S.D. Mean S.D. Variable N (or % in category) N (or % in category) X2 t p Value Medical Variables

Respiratory 77 (12% severe) 62 (22% severe) 12.55 .001
Distress

Inborn/Outborn 77 (58% outborn) 63 (40% outborn) 7.75 .01 Apnea 77 (22% yes) 63 (14% yes) 2.3 .01 1 Gestational Age 77 35.86 4.28 63 35.33 3.54 .78 ns Days 75 22.88 18.0 63 19.24 14.9 1.28 ns
Hospitalized

Demographic Variables

Race 77 (86% Caucasian) 63 (65% Caucasian) 21.91 .001 Income 77 (53% high) 56 (12% high) 131.89 .001 Maternal 77 26.25 5.3 64 23.52 5.89 2.9 .001
Age


ns = not significant, p > .05








-57


Summary


In summary, a variety of obstetric, perinatal, and demographic variables were predictive of developmental indices from 6 months through 4 years of age. No consistent group of variables can be identified across time, although days hospitalized appeared in five of the seven regression models. The parents' level of education also appeared in three of the analyses.

According to the regression results reported in

Table 6, the prediction of IQ scores at 4 years from Bayley scores can be done best from the tests administered most closely in time to age 4. The correlations between the developmental indices for this sample are similar to findings of several other studies (Wolking, Packer, Carter, & Resnick, 1985) which found moderately strong correlations between 2-year and 4-year test scores.

Comparison of the characteristics of those infants who returned for follow-up and those who did not indicates differences existed on respiratory distress, inborn vs. outborn status, and apnea. Differences were also found on the demographic factors of race, income, and maternal age. Discussion and implications of these results are presented in Chapter V.
















CHAPTER V
DISCUSSION



In this study, the relationships between a group of obstetric, perinatal, and demographic variables and developmental outcome were investigated for a group of intensive care nursery graduates. Characteristics of that sample were also compared to those of an additional sample of ICN graduates who were not followed after discharge. The discussion of the results of the data analyses and comparisons with previous findings are presented in this chapter.



Contribution of the Independent Variables to Developmental Outcome


The first two questions of interest in this study

concerned the relationship between a group of obstetric, perinatal, and demographic variables and developmental outcome. Seven multiple regression analyses were conducted using Bayley mental and motor scale scores at 6 months, 1 year, and 2 years and Stanford-Binet IQ scores at 4 years as the dependent measures. No consistent group of variables emerged which was predictive of developmental outcome across ages, although several different groups of variables were predictive at each age.



-58-






-59


Each of the obstetric variables was included in one

regression model. Gravidity was included in the regression model that predicted 1-year mental scale scores and previous preterm births contributed to the prediction of 2-year motor scale scores.

The .perinatal variables were more prevalent than

obstetric or demographic variables in the prediction of Bayley scores, with six of the nine variables present in one or more of the analyses. Days hospitalized was included in all except the models predictive of 2-year mental scale scores and IQ at 4 years. The 1-minute Apgar score contributed to the prediction of both mental and motor scores at 6 months and motor score at 2 years. All three of the regression models for mental scale scores included gestational age. Additionally, inborn/outborn status appeared in two analyses and apnea and hyperbilirubinemia each appeared in one.

Demographic variables were not consistent in their

contributions to the outcome measures, with the exception of paternal education, which was included in the prediction equations for mental scale scores at 6 months, 1 year, and 2 years. Income was included in the models for both mental and motor scores at 1 year. Urban vs. rural residence occurred once, in the model for 6-month motor scores. Interestingly, none of the variables in the models predicting outcome during infancy were included in the prediction of Stanford-Binet scores at 4 years. The







-60


demographic variables of maternal education and race were highly related to IQ (R2 = .45), with respiratory distress included in the model but not individually significant.

In summary, there was a greater tendency for the

perinatal factors to be significant predictors of mental and motor development in infancy than at 4 years of age. Obstetric factors did not appear to be especially powerful predictors. Demographic factors, specifically education of the parents, were consistently influential over time and emerged as the most significant predictors of cognitive competence at 4 years of age. This suggests that medical high risk factors may be mediated over time by environmental factors. An issue remaining for future study relates to the heterogeneity of this sample population. The methodology used in this study did not include analyses of relationships between the predictor variables and outcome for specific groups, such as young mothers. Different patterns of prediction could emerge for homogenous populations.



Prediction of IQ Scores from Bayley Scores


Another issue addressed in this study was the

relationship between infant test scores and IQ scores at 4 years. It is not surprising that the single best predictor of IQ was the measure of cognitive functioning closest to it in time, the 2-year Bayley mental scale score. This finding is consistent with other studies of high risk populations (Siegel, 1983; Wolking et al., 1985).







-61


A less typical finding was that the 6-month mental score entered the multiple regression equation next, followed by the motor score at 2 years (R2 = .47). It would appear that the 1-year Bayley scores were the "weak link" here.

The means and standard deviations calculated for the

developmental indices were anet4he-r unexpected. f-inding. They were quite similar to the means and standard deviations of the Bayley scales and the Stanford-Binet, which is an indication that the developmental outcome of this study sample fell within normal ranges.



Differences Between Follow-Up and No-Show Groups


The final issue addressed in this study concerned

potential differences between the sample of 77 children who participated in 4 years of follow-up and the sample of 64 who did not, presumably by parental choice. Although significant differences were found between the two groups for respiratory distress, inborn vs. outborn status, and apnea, no differences were found for gestational age or days hospitalized which appeared most consistently in the regression models predictive of developmental outcome. It was not clear based on these variables whether or not the no-show group was more acutely ill in the perinatal period than the follow-up group.

Since demographic factors were strongly related to

developmental outcome in the multiple regression analyses,








-62


maternal age, race, and income level were compared for the two groups. The no-show group contained a significantly younger group of mothers. It was difficult to determine if the no-show group was at a disadvantage based on mother's age since this variable was not predictive of outcome in this study. Race and income level were also significantly different between the two groups, however. The follow-up group contained significantly higher numbers of white families and of higher income families. These results indicated that on two of the three variables present in the regression model predictive of IQ, the no-show group was at a potential disadvantage.

These findings raise several questions. Could the mean test scores of high risk populations reported here and in the literature be overestimated because of the number of potentially low-scoring or handicapped infants lost to follow-up? Is the follow-up process itself an intervention that supports normal development?

One reason that low-income parents who may be less educated may not return for follow-up could be that they feel more threatened by the ICN environment during their child's hospitalization. If they feel alienated at this early stage, they could be less likely to perceive any benefits of developmental follow-up. Other possible interpretations of these data are that younger mothers of less economic means are less able to organize their resources to participate in noncritical services for their







-63


children or that these services are more appealing to upper income, educated parents. These issues warrant consideration both in future research and in clinical practice.



Comparison to Other Studies


Since the variables chosen for this study were

suggested by the risk index developed by Siegel (1982), it is of interest to compare these results to those obtained in her study. The sample population for Siegel's study consisted of 42 preterm infants (birth weight less than 1501 grams) from a working-class community in Ontario, Canada. Comparisons of the demographic characteristics of the Canadian group and the sample of the current study suggest that the average socioeconomic status of the former was lower than that of the current study. All of the participants in the Canadian study were white, in contrast to 14% black participants in this study, and from an urban community, in contrast to the large rural region where 44% of the sample in the present study resided. Maternal age was the same for both groups. Because Siegel's study focused on preterm infants, the average birth weight of 1236 grams and gestational age of 30.3 weeks were different from the nonhomogeneous sample in the present study which were 2533 grams and 36 weeks, respectively.

Despite the differences in the sample populations,

results of the multiple regression analyses in both studies







-64


have some similarities. For Siegel's group, evaluated at 5 years of age using the McCarthy Scales of Children's Abilities, the variables of SES, maternal education, previous spontaneous abortions, birth order, and respiratory distress contributed to the prediction of the General Cognitive Index (R2 = .51). Two of these variables (maternal education and respiratory distress) were predictive of Stanford-Binet IQ at 4 years in the present study.

The shift toward environmental factors more strongly predicting outcome at 4 years than medical factors is consistent with Sameroff's (1982) transactional model. The results of the current study are consistent with those of previous studies in which it was found that environmental factors are more predictive of developmental outcome as children reach school age. Social class was found to be the most highly predictive variable of IQ at between 4 and 12 years of age in a study of premature babies by FrancisWilliams and Davies (1974). For another group of premature and full-term children, also evaluated in middle childhood, mother's IQ and social class correlated with IQ (Caputo et al., 1981). Bee et al. (1982) followed a group of low-risk first-borns over four years. They found that children whose mothers had more than a high school education had significantly higher mental test scores beginning at 2 years. While these authors found that no individual variable was highly predictive of IQ score, the child's






-65


previous test performance, or a combination of maternal education and family support accounted for 20% to 50% of the variance in IQ. In the present study, a similar R2 was obtained (.47) using either infant test scores or a combination of maternal education, race, and respiratory distress. In contrast, Silva et al. (1984) found that neither SES nor maternal intelligence predicted cognitive development for a sample of preterm babies or a sample of small for gestational age babies.



Summary


Although causality cannot be implied, the

interpretation of the results of this study was that a combination of obstetric, perinatal, and demographic variables were moderately predictive of developmental outcome in infancy and early childhood. In addition, the prediction of IQ score at 4 years utilizing infant test scores supports other current findings that such prediction is more likely to be successful for at-risk populations such as ICN graduates than for normal populations (Wolking et al., 1985). Two unique characteristics of this sample population, that it was nonhomogeneous in regard to medical variables and nearly half of the subjects resided in rural areas, are key aspects of the contribution of this study to understanding of the field.

In many follow-up studies of high risk infants, the samples were selected on the basis of a specific medical






-66


criterion, such as low birth weight, intraventricular hemorrhage, or bronchopulmonary dysplasia. Since the findings in the literature do not support that these medical variables generally predict developmental outcome beyond 2 years of age, this study sample was selected on the basis of at-risk birth status alone, in order to insure a variety of medical complications. It is important to learn more about this population of intensive care nursery graduates because they are entering community services and public schools in growing numbers.

Rural populations are also not widely represented in

the literature probably because resources for research tend to be located in urban centers. Urban vs. rural residence was not specifically related to outcome in this study. In the follow-up sample, a higher proportion of rural infants were white and were transported to the ICN. Apparently these families had both the resources and the interest to participate in 4 years of follow-up, however.

The primary clinical implication of this study is that consideration regarding allocation of intervention resources should include environmental factors such as the parents' educational level as well as medical variables indicative of severity of illness. These aspects of developmental risk need further study to more precisely identify the factors related to outcome for individual children. In particular, the specific components of parent-child interaction and the home environment that support early learning need further







-67


exploration. This knowledge could contribute to improved developmental care in the ICN as well as more effective early intervention strategies.

Much has been accomplished through improved prenatal care to reduce the incidence of high risk birth. These efforts for primary prevention should continue, along with expanded secondary prevention services as early in the life of the at-risk infant as possible.

The current trend in public education to extend

special education services to the preschool population is encouraging in that it may result in the provision of more comprehensive services to at-risk infants and young children. It is important, however, to consider the family and the environment of the child in planning and implementing intervention services. Communities must organize resources from many agencies and disciplines, including health care, education, and social services, to reduce the incidence and severity of long-term handicap.
















REFERENCES



Aylward, G.P., & Kenny, T.J. (1979). Developmental followup: Inherent problems and a conceptual model. Journal
of Pediatric Psychology, 4, 331-343.

Balinsky, B. (1965). Review of Stanford-Binet Intelligence
Scale. In O.K. Buros (Ed.), Sixth mental measurements
yearbook (pp. 831-832). Highland Park, NJ: Gryphon
Press.

Barr, A., Goodnight, J., Scott, J., & Helwig, J. (1985).
Users guide to SAS. Raleigh, NC: SAS Institute.

Bayley, N. (1969). Bayley scales of infant development.
New York: The Psychological Corporation.

Bee, H.L., Barnard, K.E., Eyres, S.J., Gray, C.A., Hammond,
M.A., Spietz, A.L., Snyder, C., & Clark, B. (1982).
Prediction of IQ and language skill from perinatal
status, child performance, family characteristics, and
mother-infant interaction. Child Development, 53,
1134-1156.

Bozynski, M., Nelson, M., Rosati-Skertich, C., O'Donnell,
D., & Naughton, P. (1984). Two year longitudinal
follow-up of premature infants weighing 1,200 grams at
birth: Sequelae of intracranial hemorrhage. Journal
of Developmental and Behavioral Pediatrics, 5, 346-352.

Broman, S.H., Nichols, P.L., & Kennedy, W.L. (1975).
Preschool IQ: Prenatal and early developmental
Correlates. New York: John Wiley & Sons.

Caputo, D.V., Goldstein, K.M., & Taub, H.B. (1981).
Neonatal compromise and later psychological
development: A 10-year longitudinal study. In S.
Friedman & M. Sigman (Eds.), Preterm birth and
psychological development (pp. 353-386). New York:
Academic Press.

Cohen, S., Parmelee, A., Beckwith, L., & Sigman, M. (1986).
Cognitive development in preterm infants: Birth to 8
years. Journal of Developmental and Behavioral
Pediatrics, 7, 102-110.




-68-







-69


Coolman, R., Bennett, F., Sells, C., Swanson, M., Andrews,
M., & Robinson, N. (1985). Neuromotor development of
graduates of the neonatal intensive care unit:
Patterns encountered in the first two years of life.
Journal of Developmental and Behavioral Pediatrics, 6,
327-333.

Crocker, L., & Algina, J. (1986). Introduction to
classical and modern test theory. New York: Holt,
Rinehart, & Winston.

Fitzhardinge, P. (1984, May). Follow-up of high-risk
newborns. Paper presented at the First Annual Eric
Denoff Memorial Symposium on Child Development,
Providence, RI.

Francis-Williams, J., & Davies, P. (1974). Very low birth
weight and later intelligence. Developmental Medicine
and Child Neurology, 16, 709-728.

Hofheimer, J.A. (1979). The adolescent mother and her
infant: Correlates of transaction and development
(Doctoral dissertation, University of Florida, 1979).
Dissertation Abstracts International, 40, 5717A.

Holstrum, W.J. (1979). The prediction of three year
developmental status of high risk infants (Doctoral
dissertation, University of Florida, 1979).
Dissertation Abstracts International, 40, 4882A.

Hunt, J.V. (1981). Predicting intellectual disorders in
childhood for preterm infants with birth weights below
1501 gm. In S. Friedman & M. Sigman (Eds.), Preterm
birth and psychological development (pp. 329-351).
New York: Academic Press.

Goldstein, K. (1976). The effects of prenatal and
perinatal complications on development at one year of
age. Child Development, 47, 613-621.

Kafatos, A.B., & Pantelakis, S.N. (1982). Factors related
to perinatal morbidity and mortality. Paediatrician,
11, 27-44.

Knobloch, H., & Pasamanick, B. (1960). Brain damage and
reproductive casualty. American Journal of
Orthopsychiatry, 30, 293-305.

Kopp, C.B., & Krakow, J.B. (1983). The developmentalist
and the study of biological risk: A view of the past with an eye toward the future. Child Development, 54,
1086-1108.







-70


Kumar, S.P., Anday, E.K., Sacks, L.M., Ting, R.Y., &
Delivoria-Papadopoulas, M. (1980). Follow-up studies of very low birth weight infants (1,250 grams or less)
born and treated within a perinatal center.
Pediatrics, 66, 438-444.

Landry, S.H. (1984). Differential outcomes associated with
early medical complications in premature infants.
Journal of Pediatric Psychology, 9, 385-401.

Littman, B., & Parmelee, A.H. (1978). Medical correlates
of infant development. Pediatrics, 61, 470-474.

McCall, R.B. (1981). Early predictors of later IQ: The
search continues. Intelligence, 5, 141-147.

McCormick, M.C., Shapiro, S., & Starfield, B. (1984). High
risk young mothers: Infant mortality and morbidity in four areas in the United States, 1974-1978. American
Journal of Public Health, 74, 18-23.

Pape, K.E., Buncic, R.J., Ashby, S., & Fitzhardinge, P.M.
(1978). The status at two years of low-birth weight infants born in 1974 with birth weights of less than
1,001 gm. Journal of Developmental and Behavioral
Pediatrics, 3, 22-24.

Ramey, C., Stedman, D., Borders-Patterson, A., & Mengel, W.
(1978). Predicting school failure from information
available at birth. American Journal of Mental
Deficiency, 82, 525-534.

Resnick, M., Bauer, C., Cupoli, M., Ausbon, W., & Evans, J.
(1983). Florida Regional Intensive Care Program
developmental evaluation component-early developmental
outcome. Journal of Florida Medical Association, 70,
833-837.

Rice, B., & Feeg, V. (1985, Jan.-Feb.). First-year
developmental outcomes for multiple-risk premature
infants. Pediatric Nursing, pp. 30-35.

Ross, G., Schechrer, S., Frayer, W., & Auld, P. (1982).
Perinatal and neurobehavioral predictors of one-year
outcome in infants < 1500 grams. Seminars in
Perinatology, 6, 317-326.

Rubin, R., & Balow, B. (1979). Measures of infant
development and socioeconomic status as predictors of
later intelligence and school achievement.
Developmental Psychology, 15, 225-227.







-71


Sameroff, A. (1982). The environmental context of
developmental disabilities. In D. Bricker (Ed.), Intervention with at-risk and handicapped infants:
From research to application (pp. 141-152). Baltimore:
University Park Press.

Sameroff, A., & Chandler, M. (1975). Reproductive risk and
the continuum of caretaking casualty. In F. Horowitz,
E. Hetherington, S. Scarr-Salapatek, & G. Siegel
(Eds.), Review of child development research, Vol. 4
(pp. 187-244). Chicago: University of Chicago Press.

Siegel, L.S. (1982). Reproductive, perinatal, and
environmental variables as predictors of development of
preterm (< 1501 grams) and full term children at 5
years. Seminars in Perinatology, 6, 274, 278.

Siegel, L.S. (1983). The prediction of possible learning
disabilities in preterm and full term children. In T.
Field & A. Sostek (Eds.), Infants born at risk:
Physiological, perceptual, and cognitive processes (pp.
295-315). Orlando, FL: Grune & Stratton.

Silva, P., McGee, R., & Williams, S. (1984). A
longitudinal study of the intelligence and behavior of
preterm and small for gestational age children.
Journal of Developmental and Behavioral Pediatrics, 5,
1-5.

Smith, A., Flick, G., Ferris, G., & Fellman, A. (1972).
Prediction of developmental outcome at seven years from
prenatal, perinatal, and postnatal events. Child
Development, 43, 495-507.

Stewart, A., Reynolds, E., & Lipscomb, A. (1984, May).
Outcome for infants of very low birth weight: Survey
of world literature. The Lancet, pp. 638-1041.

Terman, L., & Merrill, M. (1973). Stanford-Binet
Intelligence Scale. New York: Houghton Mifflin Co.

Werner, W., Simonian, K., Bierman, J., & French, F. (1967).
Cumulative effects of perinatal complication and
deprived environment on physical, intellectual and
social development of preschool children. Pediatrics,
39, 490-505.

Wolking, B., Packer, A., Carter, R., & Resnick, M. (1985).
Predicting preschool Binet IQs from infant Bayley MDIs
for high-risk infants. Manuscript submitted for
publication.
















BIOGRAPHICAL SKETCH



Deborah Cole Goldberg was born August 16, 1952, near Boston, Massachusetts. Her family moved South in 1965 and she completed high school in Atlanta, Georgia, in 1970.

She attended Eckerd College (formerly Florida

Presbyterian College) where she was active in student government and performing arts. In 1974 she graduated with a B.A. in literature and relocated to Pensacola where she earned an M.A. in early childhood education at the University of West Florida. She then worked for 5 years in the Child Development Program of the Escambia Community Mental Health Center.

Deborah continued graduate studies at the University of Florida and received further clinical training through the Department of Pediatrics. She completed the Ph.D. in instruction and curriculum in 1987.

Deborah was married in 1978 to Richard M. Goldberg, a clinical psychologist. They have two children, Rachel and Jessica. She currently works as a child development specialist in private practice with her husband. She has been Coordinator of Sacred Heart Hospital's Developmental Evaluation Clinic and consultant to the Intensive Care Nursery since 1981.




-72-











I certify that I have read this study and that in omy 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.




Athol B. Packer, Chairman
Associate Professor of Instruction and Curriculum






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 of Doctor of Philosophy.




Ja n6 Largen
P sor of Coutselor Education





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 of Doctor of Philosophy.




Samuel R. athews II
Associate Professor of Psychology University of West 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 of Doctor of Philosophy.




MicHael B. ResnT&k
Assistant Professor of Pediatrics






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 of Doctor of Philosophy.




Dorene Ross
Associate Professor of Instruction and Curriculum





This dissertation was submitted to the Graduate Faculty of the College of Education and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy.

August 1987


ean, -College of Edbition




Dean, Graduate School




Full Text

PAGE 1

PERINATAL PREDICTORS OF COGNITIVE COMPETENCE AT AGE FOUR By DEBORAH COLE GOLDBERG 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 1987

PAGE 2

Copyright 1987 by Deborah Cole Goldberg

PAGE 3

This work is dedicated to my parents, Mel and Mary Cole, who with great love, encouraged me to create my own opportunities.

PAGE 4

ACKNOWLEDGMENTS An effort such as this one requires the work and cooperation of many individuals. First, and most important, in this group of caring friends is my family. My husband, Richard, gave many months of unfailing support not only during the writing of this dissertation, but also through the years of graduate school preparation. I am grateful to my daughter, Rachel, for her tolerance during my long absences from home and for her encouragement. My daughter Jessica, now two, can only be described as inspirational. I am also fortunate to have the support of a loving extended family. My professional family also deserves credit. This group includes my colleagues at Sacred Heart Hospital, Pensacola, Florida, who have contributed so much to both my professional and personal growth. The "worker bees" who lead the Florida Consortium of Newborn Intervention Programs, especially Linda Stone, helped me persevere and provided a knowledgeable sounding board. Special thanks must also go to the following members of my academic family: Athol Packer, my chairman, for his wisdom, patience, and tenacity; Michael Resnick, for his fine example as a pioneer in the field of early i V

PAGE 5

intervention; Sam Mathews, for his crisis management skills, his scholarship, and especially his friendship; Dorene Ross, for her keen editorial eye; and Janet Larsen, for her sharing of her vast understanding of young children. Finally, the technical expertise of several people was instrumental to the completion of the research and manuscript. Thanks go to Juanita Melson for help with data collection and Mark Littlefield for assistance with data analysis. V

PAGE 6

TABLE OF CONTENTS Page ACKNOWLEDGMENTS iv ABSTRACT viii CHAPTERS I INTRODUCTION 1 Purpose of the Study 4 The Sample 6 Research Questions 6 II REVIEW OF RESEARCH 8 Overview 8 Three Models for Studying Outcome 11 The Research in Retrospect 29 III METHODOLOGY 33 Pilot Study 34 The Subjects 35 Instrumentation 39 Procedures 40 Data Analyses 42 Summary 44 IV RESULTS 45 Question One: Relationship Between the Independent variables and IQ Scores 45 Question Two: Contributions of the Independent Variables to Infant Development 47 Question Three: Prediction of IQ Scores from Bayley Scores 52 Question Four: Differences Between Follow-Up and No-Show Groups on Medical Variables 52 Question Five: Differences Between Follow-Up and No-Show Groups on Demographic Variables.. 55 Summary 57 vi

PAGE 7

Page V DISCUSSION 58 Contribution of the Independent Variables to Developmental Outcome 58 Prediction of IQ Score from Bayley Scores 60 Differences Between Follow-Up and No-Show Groups 61 Comparison to Other Studies 63 Summary 65 REFERENCES 68 BIOGRAPHICAL SKETCH 72 vii

PAGE 8

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 PERINATAL PREDICTORS OF COGNITIVE COMPETENCE AT AGE FOUR by Deborah Cole Goldberg August 1987 Chairman: Athol B. Packer Major Department: Instruction and Curriculum The purpose of this study was to investigate the relationships among a set of obstetric, perinatal, and demographic variables and developmental outcomes for a sample of 77 infants treated in a regional intensive care nursery and followed over 4 years. The relationship between infant test scores IQ at 4 years of age was also examined. In addition, the characteristics of a sample of 65 children who did not return for follow-up were compared to those who were followed in order to identify differences which may influence development. The independent variables of interest in this study were mother's gravidity and number of previous preterm births, infant's birth weight, gestational age, Apgar scores, respiratory distress, apnea, hyperbilirubinemia, days hospitalized, inborn vs. outborn status, sex, race, urban vs. rural residence, family income, and parents' age and education. Outcome measures included the Bayley Scales of Infant Development at 6 months, 1 year, and 2 years and the Stanf ord-Binet Intelligence Scale at 4 years. The V i i i

PAGE 9

relationships between early risk factors and developmental outcome were investigated utilizing multiple regression analyses . Analysis of results indicated that the demographic variables of maternal education and race contributed significantly (p < .05) to the prediction of IQ score at age 4. At earlier ages, the medical variables of days hospitalized, gestational age, Apgar scores, and inborn vs. outborn status were predictive of Bayley mental and motor scale scores. Additionally, paternal education was included in the regression models for mental scale scores at 6 months, 1 year, and 2 years. Another question addressed in this study was the relationship between infant test scores and preschool IQ score. The results of multiple regression analyses indicated that mental scale scores at 6 months and 2 years and motor scale scores at 2 years were moderately predictive of IQ (r2 = .47) . A comparison of the sample participating in follow-up to a sample which chose not to participate resulted in significant differences (p < .05) on respiratory distress, inborn vs. outborn status, apnea, maternal age, race, and income level. Analysis of these results indicated that the no-show group was at a potential disadvantage for appropriate development. ix

PAGE 10

CHAPTER I INTRODUCTION The relationship between developmental status in infancy and later school performance is one which has interested behavioral scientists for many years. The emergence of the growing population of intensive care nursery (ICN) graduates has increased interest in this issue. There is concern that some of these premature and sick newborns are at risk for future developmental problems, and many researchers are concerned with identifying which medical and environmental variables may be associated with later problems. Although numerous studies have been conducted which assess the relationships between a wide range of medical and demographic variables and subsequent developmental indices, findings have been inconsistent (Caputo, Goldstein, & Taub , 1981; Cohen, Parmelee, Beckwith, & Sigman, 1986; Francis-Williams & Davies, 1974; Holstrum, 1979; Hunt, 1981; Kumar, Anday, Jacks, Ting, & DelivoriaPapadopoulas, 1980; Siegel, 1982). Further, the characteristics of the populations studied have varied a great deal. The intent of the present study was to (a) extend existing research by utilizing a sample population of intensive care nursery (ICN) graduates that was heterogeneous in terms of medical variables, but was largely -1-

PAGE 11

-2white and from the rural South; and (b) determine if differences in sample population characteristics might be a source of some of the variability in findings. Interest in the prediction of outcome for at-risk infants was stimulated when premature babies first were cared for in institutions such as the nursery established by Dr. Julian Hess in Chicago in 1922 (Francis-Williams & Davies, 1974). This was the forerunner of the present day ICN. Large-scale longitudinal follow-up studies, many of them retrospective, began appearing in the 1940s and gradually increased through the 1950s and 1960s. Since 1970, many researchers from a variety of disciplines have been working in this area, as is evident by the studies published in a wide variety of medical and behavioral publications. A review of 22 studies of premature infants from around the world demonstrated that until 1960, mortality rates and the incidence of major handicaps were high. Since that time, chances of survival for premature infants weighing less than 1500 grams at birth have increased dramatically due to better understanding of the physiology of prematurity and advances in medical technology (Stewart, Reynolds, & Lipscomb, 1984). One underlying concern shared by health and educational professionals is the potential increase in the proportion of handicapped individuals due to the decrease in the mortality rate for high risk infants. Analysis of the data gathered in recent years through the Florida Regional Perinatal

PAGE 12

-3Intensive Care Centers program suggests that decreasing mortality has not resulted in increasing morbidity among survivors by birth weight categories (Resnick, Bauer, Cupoli, Ausbon, & Evans, 1983), a finding common to many other studies (Fitzhardinge, 1984; Kafatos & Pantelakis, 1982; Kumar et al . , 1980; Pape, Buncic, Ashby, & Fitzhardinge, 1978). However, further analyses are needed to refine the approaches for quickly and accurately identifying those infants most likely to become developmentally disabled or who will require special education services. One of the goals implicit in research in this area is the early identification of the highest risk infants for timely provision of services. A second goal is the identification of those variables most closely linked with disabilities so that medical procedures may be revised and more direct preventive measures be taken. Widespread attention has been directed to these concerns in the face of increasing health care and educational costs and shrinking resources . Over the past 20 years, the methodology used in followup studies of intensive care nursery graduates has become more complex. Earlier researchers often utilized a singlefactor model, correlating gestational age, birth weight, or other medical variables with some outcome measure, usually IQ (Rubin & Balow, 1979). Later, groups of variables were analyzed using more sophisticated statistical techniques, but still focused on obstetric and neonatal medical events.

PAGE 13

as typified by the early work of Knobloch and Pasamanick (1960) . One trend among more recent studies has been to include a broader range of reproductive, perinatal, and demographic variables and to utilize a variety of outcome measures reflecting discrete areas of development. Further, the relative importance of demographic characteristics such as socioeconomic status and age and education of parents is now recognized and the interpretation of some research findings suggests that environmental factors are more influential than perinatal events as the child reaches school age (Broman, Nicholas, & Kennedy, 1975; Hunt, 1981; Saraeroff, 1982; Siegel, 1982). Another element of recent research has been an extension of follow-up into early school age investigating school achievement as well as the incidence of major handicaps (Aylward & Kenny, 1979; Caputo et al . , 1981; Francis-Williams & Davies, 1974). Purpose of the Study The purpose of the present study was to investigate the relationships between a set of obstetric, perinatal, and demographic factors and the cognitive and psychomotor development of a group of ICN graduates. The relationship between infant test scores and IQ scores at 4 years of age was also examined. In addition, the characteristics of a second sample of ICN graduates who did not return for

PAGE 14

-5follow-up were compared to those who did participate in order to identify possible differences. The scope of this study was defined in part by the state-wide follow-up effort of Florida's Regional Perinatal Intensive Care Centers program, developmental evaluation component. Staff of this component provide periodic developmental and pediatric evaluations to a cross-section of intensive care nursery graduates beginning at 3 months of age and continuing through 4 years of age. The children in the present study represent a sample of this population drawn from one regional intensive care nursery. The chief limitation of this study was that it is ex post facto in design, utilizing chart review for data collection. The potential problem with the reliability is offset somewhat by the fact that the hospital and developmental evaluation clinic staff were consistent throughout the time period in question and were available to this researcher for consultation concerning the present study. The overall goal of the study was to contribute to the body of research concerning the development of high risk infants, particularly with regard to prediction of future disability from data available at the time of hospital discharge. in practical terms, this type of information might be helpful in the timely and effective use of resources needed to maximize the developmental potential of individual children.

PAGE 15

-6The Sample The sample for this study was drawn from those infants cared for in the Intensive Care Nursery at Sacred Heart Hospital, Pensacola, Florida, between October, 1977, and February, 1980. The sample includes both infants who were inborn and those transported to the ICN following birth. Those who completed 4 years of follow-up were eligible for participation in the study. Since all ICN graduates were invited to participate in the 4-year developmental follow-up during the years in question, the study sample was nonhomogeneous in terms of birth weight, gestational age, and perinatal complications. The sample was predominantly white and middle class and a significant number resided in rural communities. These three demographic characteristics are somewhat unusual compared to other study populations, particularly those in other areas of Florida's Regional Perinatal Intensive Care Centers program. For comparison purposes, a second sample of infants was identified who were also treated in the ICN during the same time period as the follow-up group, but who chose not to participate in fol low-up . Research Questions The first three questions of interest in this study were addressed through an examination of the relationships between 19 independent variables and 7 dependent variables

PAGE 16

-7for a sample of ICN graduates. Two additional questions addressed possible differences between the sarnple which participated in follow-up and another sample of ICN graduates which did not. The questions were 1. Is there a relationship between a set of obstetric, perinatal, and demographic variables and IQ at 4 years of age? 2. Is there a relationship between a set of obstetric, perinatal, and demographic variables and developmental indices at 5 months, 1 year, and 2 years of age? 3. Are developmental indices at 6 months, 1 year, and 2 years predictive of IQ at 4 years? 4. Are there differences in obstetric and perinatal variables between those subjects who participated in followup evaluationsand those who did not? 5. Are there demographic differences between parents who brought their children for follow-up evaluations and those who did not? The patterns of relationships observed in this study between the independent variables and the outcome measures were compared to results of studies of other types of high risk populations. Since the sample population in this study differs a priori from other study populations, only descriptive data pertaining to the results are presented.

PAGE 17

CHAPTER II REVIEW OF RESEARCH Overview The majority of research pertinent to this study of perinatal predictors of intellectual status at age 4 has been published or reviewed over the past 10 years. Studies of follow-up of high risk infants have become more prevalent recently. In general, the older studies were based on a single factor approach to prediction of outcome, while more recent research indicates a trend toward multifactor predictors of developmental status. This chapter includes a brief historical background of this complex and rapidly growing field of study, as well as a more detailed exploration of a number of recent studies. Several researchers have provided interesting historical overviews of the follow-up research on infants born at risk. Francis-Williams and Davies (1974) described a review published by Benton in 1940 of 30 studies conducted between 1911 and 1940. Benton identified several unsatisfactory elements in the studies including small sample sizes, bias in selection, disregard of socioeconomic status, and lack of objective measures of IQ. The results -8-

PAGE 18

-9of these studies did not indicate, however, the premature infants had lower IQ scores than full-term infants. Francis-Williams and Davies (1974) also described a study by Wiener, who in reviewing the literature of the next two decades, found Benton's conclusions were not confirmed. The studies reviewed by Wiener showed decreased IQ for low birth weight infants, with the findings of three major studies that IQ scores decreased as birth weight decreased. Bee and her colleagues described a resurgence of longitudinal studies of low birth weight infants in the late 1950s and early 1960s (Bee, Barnard, Eyres, Gray, Hammond, Spietz, Snyder, & Clark, 1982). These studies included descriptions of the relationships among prenatal and perinatal variables, home environment, and later outcomes for children. Bee et al . (1982) identified three basic conclusions about the relationships between perinatal status and later outcome. First, medical variables such as birth weight or anoxia had demonstrated small but significant relationships with later cognitive and motor development (Smith, Flick, Ferris, & Fellman, 1972). Second, the effect of the medical variables appeared to be mediated by characteristics of the child's environment (Werner, Simonian, Bierman, & French, 1967). Third, in most predictive studies, the best predictor of later cognitive functioning was not medical or perinatal status, but the level of the mother's education (Smith et al . , 1972).

PAGE 19

-10Kafatos and Pantelakis (1982) , in reviewing a number of prospective and retrospective studies of pregnancy and infancy in many areas of the world, identified a set of risk factors correlated most strongly with perinatal morbidity and mortality. The most critical risk factors according to their review are related to age of mother, parity, race, previous fetal loss, medical care, poverty, unwanted pregnancy, education of the mother, multiple births, and maternal morbidity. There now appears to be a consensus among most researchers that some combination of obstetric, perinatal, and demographic or environmental factors needs to be studied in relation to later developmental outcomes for the population of high risk infants. Because the chance of healthy survival has tripled for these infants since 1960, an increasing population of these children is available for study. Bee et al . (1982) noted that the research climate now exists for more detailed, short-term, longitudinal studies of high risk infants in order to uncover causal links between perinatal indicators, the child's environment, and later cognitive functioning. An emerging goal is the eventual application of this knowledge of causal links to the early identification and screening of children at high risk for later developmental problems so that effective interventions can begin at the earliest opportunity .

PAGE 20

-11Three Models for Studying Outcome Most of the recent research has focused specifically on defined high risk populations. These high risk infants generally have been cared for in an intensive care nursery and often the populations selected for study are very low birth weight. A number of these studies are reviewed in detail because they are representative of emerging trends in research. Sameroff's (1982) conceptualization of early influences on development is used to provide a framework for discussion of these studies. Sameroff (1982) identified three models for explaining the relationship between early risk factors and developmental outcome. These are (a) the single factor model, with an emphasis on either constitutional or environmental factors; (b) the interactional model, combining constitutional factors additively with environmental supports; and (c) the transactional model, wherein development results from a continual interplay between a changing organism and changing environment. Single Factor Model As mentioned previously, the single factor model has not been widely used in recent studies of high risk infants. VJhen this approach is utilized, however, medical variables are typically chosen. A primary example of this approach is found in the work of Littman and Parmelee, who are

PAGE 21

-12responsible for a major component of the research conducted regarding clinical predictors of outcome. In a recent study (1978), these researchers expanded on their initial work in order to determine if pediatric complications occurring in the first 9 months of life were more predictive of outcome than perinatal variables. The study included 125 preterm infants followed through the UCLA Center for Health Sciences. No years of birth or follow-up were identified in the report of the research. For Littman and Parmelee's sample the mean birth weight was 1,927 grams and the mean gestational age was 33.1 weeks. Four quantitative scales were used to describe the infants in the study: the obstetric complications scale, the post-natal complications scale, and two pediatric complications scales (1 to 4 months, 4 to 9 months) . The obstetric and the post-natal complications scales included measures of physical development, health, behavior, congenital anomalies, and neurological and sensory handicaps. These four medical scales were scored in a summary fashion, with all items receiving equal weight in scoring. All infants were evaluated beginning at their expected date of birth and continuing through 2 years of age. All evaluations were corrected for prematurity and included a newborn neurological examination at term, the Gesell Developmental Tests at 4, 9, and 24 months of age, and the Bayley Scales of Infant Development at 18 and 25 months of age. No demographic measures, including socioeconomic status, were

PAGE 22

-13included in this study. The correlations of medical events scales to outcome measures showed that the pediatric complications scale for 4 to 9 months related most significantly to later performance for the entire sample population. it is important to note the frequency of illness was high during the 4 to 9 months period. By age 9 months, more than 80% of the infants had experienced some type of medical problem. These authors felt that early recognition of developmental disability is quite important and that assessment of health during later infancy should not be overlooked in predicting outcome. Kumar et al . (1980) studied a sample population of smaller infants, less than 1,250 grams birth weight, born 1974-1977 and cared for in the Perinatal Center of the university of Pennsylvania Hospital. They followed 50 of 60 survivors for one year to compare the outcomes for this group of high risk infants to findings in other studies where all infants were outborn. At 1-year follow-up these researchers found 3 infants had major neurological problems; 2 infants had retrolental fibroplasia, and severe developmental delay (Gesell Developmental Tests DQ less than 80) was documented in these 5 plus 2 other infants. The incidence of handicap for this sample was 14%. The 7 handicapped infants were then compared to the remainder of the sample on 16 variables using two-tailed _t tests for the quantitative variables (birth weight, gestational age, 1and 5-minute Apgar scores, initial pH, initial temperature.

PAGE 23

-14time to regain birth weight, peak bilirubin, hypoxia, and duration of hospital stay) , and chi square for qualitative variables (sex, mode of delivery, weight for gestational age, apnea, indication for mechanical ventilation, patent ductus arteriosis, and seizures). Significant between group differences were found only for mode of delivery and duration of hospital stay. In discussing their results, Kumar et al . (1980) pointed out that there was a marked increase in survival for infants born less than 1,250 grams, from 33.3% in 1974 to 63.2% in 1978. These figures suggest changes in neonatal management over this time period which may have introduced a bias into the study. The authors noted that it is difficult to compare the results of their study with those of previous researchers because of differences in aspects of patient care as well as racial and socioeconomic differences. Although these results were based on fairly short-term follow-up, the authors believed their findings are encouraging in the area of prognosis for very low birth weight infants in that infants born and treated in a perinatal center, including high risk obstetric management, had fewer medical complications and generally better developmental outcomes as compared with transported neonates . Another example of the single factor model was a study of first year developmental outcome for premature infants

PAGE 24

-15conducted by Rice and Feeg (1985). In this retrospective study, 57 records were selected from the records of a developmental evaluation clinic at a large teaching hospital. The time period in which the infants were born was not indicated, neither were the criteria for inclusion in the follow-up clinic. In order to review the clinical data, the Categories of Risk Index (CRI) was developed for this study to indicate the number of complications in the perinatal period. The CRI was based on the L i ttrnan-Parmelee risk scale. No demographic factors were considered. The Bayley Scales of Infant Development were used as the outcome measure, using the scores obtained closest to 1 year of age. It is notable that 31% of the scores were obtained in the 26 month range. The only criterion noted for inclusion in the study was that the infant was born at 38 weeks gestation or less. Three hypotheses were proposed for this study: 1. Differences in the Bayley Scales of Infant Development, using both Mental (MDI) and Motor (PDI) scale scores, would be predicted by the risk status (high, low) and birth weight for gestational age (under or over) . 2. Differences in MDI and PDI would be predicted by gestational age, birth weight, deviation from expected birth weight, and perinatal risk status. 3. The combination of perinatal risk factors would predict MDI and PDI.

PAGE 25

-16Because several hypotheses were of interest in this study, a range of statistical analyses were performed. A 2x2 ANOVA was used to test hypothesis #1 with the finding that there were no differences in MDI and PDI between high and low risk groups for over and under birth weight or gestational age groups. For hypothesis #2, multiple regression analyses indicated that combined factors of gestational age, birth weight, deviation birth weight, and total risk were related to uncorrected MDI and PDI. When the Bayley scores were corrected for gestational age, no factors were predictive. Pearson correlation coefficients were computed to analyze the relationship between agecorrected and uncorrected Bayley scores and the eight categories on the risk index. No categories were significant for the uncorrected MDI. The corrected MDI was significantly correlated with surgery in the predicted direction. The uncorrected PDI was negatively correlated with seizures, anomaly-noninfectious illness, and ventilator assistance. The category of seizures was also strongly correlated with corrected PDI. The most encouraging implication of this study is that infants with high numbers of post-natal risk factors may not necessarily experience more developmental delays than less ill infants. Limitations mentioned by the authors include the research design, including clinical records review for data collection, and a potentially biased sample of preterm i nf ants .

PAGE 26

-17The Interactional Model The work of a number of researchers can be described according to the interactional model, which explains the relationships between risk factors present in infancy and developmental outcome in terms of a combination of clinical and environmental factors. These studies typically include factors related to family status and the environment as well as the obstetric and neonatal medical events characteristic of the single factor model. Three examples of this approach are found in studies by Ramey et al . (1978), Pape et al . (1978), and Ross, Schechrer , Frayer, and Auld (1982). Ramey and a group of researchers at the University of North Carolina at Chapel Hill were interested in the practicality and effectiveness of using birth certificate information as a mechanism to identify children who were likely to need special education services beginning in the first grade. Their sample population was a group of 1000 first grade students randomly selected from 20 counties in North Carolina. This report of the study does not indicate the birth year or years of these children. The four outcome measures utilized in this study were the Peabody Picture Vocabulary Test, the Test of Basic Experience, the Developmental Test of Visual-Motor Integration (Beery) , and the Myklebust Pupil-Rating Scale. The independent ratings obtained from birth certificates included race, sex, birth order, birth weight, number of

PAGE 27

-18weeks gestation, and legitimacy. Information concerning the mothers was also obtained from the birth certificate, including age, educational level, the month prenatal care began, whether there were previous births now dead, and whether there were any previous fetal deaths. Multiple regression analyses were utilized to assess the relationship between the predictor variables and the outcome variables. To compare children who performed relatively poorly to those performing at or above the mean, multiple regression analyses were performed on the birth data . Ramey et al . (1978) found that for the sample population as a whole, the most predictive characteristics of educational and psychological status at first grade were race and mother's educational level. The results of the multiple regression analyses suggested that several variables significantly discriminated between moderate and low risk children on each of the four criterion variables. These were birth order, education of the mother, birth weight, month prenatal care began, race, and legitimacy. Although causality cannot be implied here, the researchers stated that this study demonstrates it is feasible to identify children utilizing birth certificate data who are likely to need special education services before and during grade school. Pape et al. (1978) reported on a group of infants with birth weights less than 1000 grams born in 1974. Their

PAGE 28

-19concern, still prevalent among researchers, was the potential relationship between decreasing mortality and increasing morbidity. They studied 46 survivors of 97 premature infants transported to the Hospital for Sick Chirldren in Toronto. Follow-up clinic visits were scheduled at 3 month intervals for the first year and at 18 and 24 months. Each clinic visit included a physical and neurological examination, measurement of growth parameters, and administration of the Bayley Scales of Infant Development corrected for gestational age. Hearing screening and eye examinations were also performed. The socioeconomic status was determined on the basis of the father's occupation and education. Of the 42 infants who received psychometric testing at 18 and 24 months, 13 infants (30%) had a diagnosed significant handicap by 18 months of age including 4 who had severe neurologic defects and 9 others who were developmentally delayed according to the results of the Bayley scales (mean MDI 89.2) . To study the relationship between neonatal course and developmental outcome for these infants, 13 handicapped infants were compared with the remaining infants in the sample. Quantitative variables were analyzed by t tests and the'chi square test was used for qualitative variables. Significant associations were shown with the following: birth weight, acidemia, intracranial hemorrhage, and seizures. The variables of complicated pregnancy or delivery, asphyxia, respiratory distress, apnea, prolonged ventilation, or

PAGE 29

-20soc ioeconomic status were not associated with later delays . In view of the fact that the 30% incidence of significant handicap was unusually high, the authors pointed out a number of atypical characteristics of the study sample. First, all infants were outborn and no infant was refused on the basis of pretransport condition. Fifty-nine percent of the survivors suffered severe birth anoxia. There was also a high incidence of severe cold stress and acidemia related to transport. This high risk group also has some very positive characteristics. Ninety-five percent of the intensive care nursery survivors in this group remained in the follow-up program. Seventy-five percent of these were from families in the middle or upper socioeconomic level and living in intact two-parent homes. Early referrals to infant stimulation programs were also made for children showing a delay in any area. These researchers identified a very positive finding related to their study. According to their research, before 1970 approximately 75% of infants whose birth weight was less than 1000 grams died and only 15% survived as normal children. in thispopulat ion born in 1974, 53% died and 33% of the remainder survived without significant handicap. in response to the initial question about the relationship between decreasing mortality and morbidity, these authors found that while mortality decreased 25%, there was a twofold increase in the number of normal survivors and little

PAGE 30

-21change in the absolute number of children who were hand icapped . Ross et al . (1982) studied a group of infants born at or less than 1500 grams in order to assess the predictive value of multiple demographic, perinatal, and neurobehavioral variables for 1-year outcome. The infants in this study were born in an 18-month period beginning March 1978 and admitted to the Perinatal Center of New York Hospital-Cornell Medical College. One hundred two of 120 consecutive survivors whose birth weight was appropriate for gestational age were evaluated at 1, 3, 6, 9, and 12 months corrected ages. Outcome measures included Bayley scores (MDI and PDI) , the presence or absence of cerebral palsy (CP) , and pediatrician's rating (normal, suspect, delayed). Developmental status was based on a composite rating in three categories: (a) normal (85 minimum score MDI and PDI, normal by the pediatrician, or free of CP); (b) suspect (7184 MDI or PDI, suspect by the pediatrician, or mild CP); and (c) abnormal (70 or less MDI or PDI, abnormal by the pediatrician, or moderate or severe CP). According to this composite system, 52% of the sample was evaluated as normal at one year, 25% suspect, and 23% abnormal. The results of univariate F-tests suggested the following perinatal variables were significant: 1-minute Apgar, patent ductus arteriosis, seizures, intraventricular hemorrhage, sepsis, maximum oxygen required, pneumothorax.

PAGE 31

-22hypernatremia , and duration of stay. No demographic variables (SES, race, sex, maternal age) were found to be related to 1-year outcome. Results of the Amiel-Tison neurobehavioral assessment at 3 and 9 months were most predictive of outcome and added significantly to the predictive value of the perinatal variables overall. Although the authors acknowledged that development at one year of age may not predict later development, they conjecture that measures at one year may be indicative of underlying neurological impairment before it is influenced by the environment. The demographic characteristics of this population were significant in that the majority were from middle to upper middle class backgrounds and were white. Given the relatively high incidence of handicap, it should be noted that 45% of the sample were outborn and transported to the perinatal center. The Transactional Model Ross et al . (1982) in reviewing the literature noted that "as children reach school age, however, environmental factors such as socioeconomic status, parental IQ, and mother's education become far more powerfully related to verbal and visual-motor abilities than perinatal and infant behavioral variables, which may still be associated with aspects of later outcome but to a lesser extent" (p. 318) . This viewpoint supports the transactional model of

PAGE 32

-23developraent, which includes the wide variety of factors of the interactional model and takes into account the plasticity of both the child and the environment over time. Studies of high risk infants which utilize repeated measures or observations over time may be classified by the transactional model. Two such studies conducted outside the United States were reported by Francis-Williams and Davies (1974) and by Silva et al . (1984). Francis-Williams and Davies (1974) followed 105 of 120 children weighing less than 1500 grains at birth and treated at Hammersmith Hospital, London, between the years of 1961 and 1968. The sample included both inborn and transported infants. Because the time period in which these children were born included significant changes in neonatal care, initial analyses were made of the children in two groups (1961-1964 and 19651968). These analyses revealed, however, that dividing the subjects according to birth years was not helpful. The children were evaluated at between 4 and 12 years of age utilizing the Wechsler Preschool and Primary Scale of intelligence or the Wechsler Intelligence Scale for Children. Reading ability was also assessed. Children older than 5 years were also assessed using the Bender Gestalt Test for Young Children. Correlations were examined between neonatal illness, social class, head size, and developmental outcome as measured by the standardized tests. Children born small for gestational age (SGA) were analyzed

PAGE 33

separately from those whose birth weights were appropriate for gestational age (AGA) . The only significant correlation found related to neonatal illness was severe asphyxia in the four SGA infants in the study. Social class was found to be the most highly predictive variable. These researchers stated, "this bias and distribution toward the lower social classes is evident in most reports of children of very low birth weights" (p. 718). A significant difference in IQ was also found between infants whose head size was below the 50th percentile. The overall incidence of handicap in this sample was 12%. These researchers, in comparing their results to research of the previous 30 years, pointed out significantly improved outcomes for these high risk children overall. They cautioned, however, that onefifth of their sample had a performance IQ significantly below their verbal IQ with learning difficulties evident in those children who were school age. Early recognition of learning difficulties and careful follow-up is indicated for these children. A study conducted in Dunedin, New Zealand, by Silva, McGee, and Williams (1984) examined the relationship between perinatal problems and cognitive and behavioral development. A sample of 850 infants born between April of 1972 and March of 1973 was classified into three groups (preterm, SGA, and full-term normal birth weight) and compared on IQ at 3, 5, 7, and 9 years of age, and on parent and teacher behavior reports at 5, 7, and 9 years. The sample of 850 children

PAGE 34

-25was selected from 1,561 total births because all relevant information was available for these children. The authors noted that the follow-up group did not differ significantly from the remainder of the population in perinatal characteristics or socioeconomic status. In contrast to other studies, no significant differences were identified among the three groups for SES or maternal intelligence. Using an ANOVA, researchers found significant differences on intelligence test results at each age among the groups. The SGA group performed least well. The SGA group also had significantly more behavioral problems than the remainder of the sample according to the parent's behavior rating scale. Although the teacher's behavior rating scale indicated an increasing number of problems for all groups over time, there were no significant differences among the groups. Several significant studies have come out of the longitudinal research conducted by Caputo et al . (1981). Information regarding 10-year follow-up of a sample of children born on Staten Island, New York, between July of 1965 and January of 1969 has recently been published. The 64 children included in the follow-up were part of a larger group of 233 infants who were studied at one year. Half of this smaller sample consisted of infants prematurely born weighing less than 2500 grams and the other half of the sample was full term. Almost all of these children were white and middle class. The variables studied included 7

PAGE 35

-26demographic variables, 7 birth and obstetric variables, the sex of the child, prematurity, mother's discomfort during pregnancy and delivery, mother's child rearing attitudes, and mother's IQ. When these children reached the ages between 7 and 9 1/2 years they were evaluated using the Wechsler Intelligence Scale for Children-Revised (WISC-R) and the Bender Gestalt Test for Young Children. Extremely thorough statistical analyses were made of these data including separate examinations of the WISC-R subtests, verbal IQ, performance IQ, and full scale IQ, as well as the Bender Gestalt results. Briefly, the authors found that in bivariate analyses, both mother's IQ and the social class factor correlated very significantly with verbal IQ and performance IQ in middle childhood. In this study, increasing family size was negatively correlated with IQ. Prematurity was the only birth complication independently related to cognitive development. Specifically, visual perception was negatively affected by prematurity. The Cattell Infant Intelligence Scale and Gesell Development Tests conducted at one year were compared against the later assessments and were found to be generally poor predictors of later outcome. The Gesell personalsocial score, however, tended to contribute significantly to the WISC-R verbal and full scale IQ scores. The earlier analysis of birth and obstetric complications factors compared to development at one year showed delivery and

PAGE 36

-27related variables and complications to be predictive. These same factors were no longer predictive of outcome at 7 to 9 years, but the obstetric factor, indicative of family size, was significant. Bee et al . (1982) conducted a longitudinal study of relatively healthy children which addressed several questions. The first was the usefulness of information about the child versus information about the environment. Another question related to the usefulness of infant mental tests for prediction of later IQ. A further aspect of the study was the predictive role of "ecological" family characteristics including maternal education, available social support, amount of life change, and mother's perception of infant. The possible differences in the predictive equations for mothers who differed in level of education was also considered. The authors believed the large sample size, the range of measures used, and the frequency of observations allowed them to appropriately address this wide series of questions. The population studied consisted of 193 first born infants born during 1973 and 1974 at one hospital in Seattle. In order to achieve a sample including about half mothers who had 'a high school education or less and about half who had more than high school maternal education was used as a blocking variable. A second blocking variable was the presence or absence of perinatal risk factors. Multiple births and infants with anomalies were excluded. The authors emphasized that the sample they selected was an

PAGE 37

-28unusually healthy group with only 3 infants with birth weights below 2500 grains and only 23 with Apgar scores of 6 or below at 1 or 5 minutes. The authors also noted that the group as a whole was fairly well educated and economically well off. These characteristics in combination with those previously described classify this sample as a generally lov/-cisk group. Children in this sample were followed through 4 years of age, with observations and assessments beginning in the eighth month of pregnancy and continuing through birth, 1, 4, 8, 12, 24, 36, and 48 months of age. Home visits were made at all but the prenatal, birth, and the 24-raonth age levels. A developmental assessment was conducted in a clinical setting at 13, 24, 36, and 48 months. An expected result of this study was that mothers with more than a high school education had larger, more raotorically mature newborns, with significantly higher mental test and language scores beginning at 24 months. A more enriched home environment with more facilitative teaching was evident from the earliest observations. No maternal educational differences were related to mental or language development before 24 months and no difference in psychomotor functioning were found at any age. It should be noted that the size of most correlations was small. The authors stated that no single measure of perinatal status, child outcome, family support characteristics, or

PAGE 38

-29interaction patterns accounts for very much of the variance in the IQ or language scores. A second analysis was conducted combining variables in each of four clusters with a separate regression analysis for each outcome variable. By means of these analyses, 20% to 50% of the variance in IQ or language development could be predicted by measuring any of three characteristics: child's earlier test performance, maternal infant interaction and environment, or maternal education and family support. The authors pointed out the particular effectiveness of including demographic characteristics about the family in the analyses. They stated "it is possible to gather information at the time of an infant's birth that will tell us as much about his IQ or language as will either direct observation of parent infant interaction or direct measures of the child's cognitive or language development during the first year" (Bee et al . , 1982, p. 1048). The Research in Retrospect The variety of approaches evident in the studies previously described is characteristic of the field of research concerning at-risk infants. These studies are indicative of the trend toward investigation of multiple predictors of outcome and the increasing complexity of the analyses used. The disparate methodologies and variety of findings suggest that no consensus exists, but the large

PAGE 39

-30number of studies from many professional disciplines indicates that this is a significant area of research. Although the studies discussed are not directlycomparable, some general strengths and weaknesses in the research can be described. When the methods used to identify the sample population and collect the data were described, they seemed careful and thorough. Most subjects were drawn from the same geographic area and treated at one perinatal center, limiting treatment biases. The instruments utilized as outcome measures are well accepted, and most researchers stated whether or not age corrections for prematurity were made in scoring. The primary weakness of these studies is related to the lack of generalizability of results. This, unfortunately, is inherent in the methods required to study a defined group of graduates from a specific intensive care nursery (ICN) . Results cannot readily be compared across studies because of the initial differences in population as well as medical treatment. Instrumentation varies, as do the statistical analyses employed. Correlational analyses alone do not control for the possible interactions between variables. In general, the large numbers of variables and relatively small sample sizes decreased the power of the multiple regression analyses . The results of most of these studies contain indications that some medical and environmental factors are associated with developmental outcome. The relative

PAGE 40

-31significance of these categories seems to vary with the age of the child. A broad trend suggested here is that larger numbers of at risk infants are surviving with minimal or no handicap. The factors most related to healthy survival need to be identified along with those related to poor outcome. According to Jane Hunt (1981) , "the dynamics of the interaction between the initial deficit, genetic potential, and environmental effect, and intellectual outcome are not well understood" (p. 331) . The two major issues involved in continued studies of high risk populations concern the identification of significant developmental problems as early as possible, and the identification of the conditions that caused the problems. The state of the art of follow-up of high risk children is still far from the goal of reliable prediction of the consequences of high risk birth compounded by environment. However, as Sameroff (1982) stated, "the continued study of this problem may help us to find better descriptions of the dynamic processes by which early problems are overcome and later ones created" (p. 393) . Early prediction is certainly important for the care and evaluation of high risk infants. Ross et al . (1982) identified three reasons related to the need for further study of this area: (a) the identification of perinatal factors predictive of poor outcome can result in improved medical treatment and so decrease later problems, (b) early prediction may lead to early identification and

PAGE 41

intervention, and (c) identi the understanding of factors development of low birth wei 32fying predictors can enhance responsible for poor or good ght infants.

PAGE 42

CHAPTER III METHODOLOGY This retrospective, ex-post facto study of outcome for infants born at risk is similar in design to others utilizing antecedent data to search for predictors of developmental delay or handicap (Hunt, 1981; Kumar et al . , 1980; Littman & Parmelee, 1978; Ramey et al . , 1978; Rice & Feeg, 1985). Modeled on the work of Siegel (1982), the current study includes an examination of the relationships between developmental status and a group of obstetric, perinatal, and demographic variables, including (obstetric) gravidity, number of previous preterm births; (perinatal) birth weight, gestational age, 1and 5-minute Apgar scores, respiratory distress (4-point scale), apnea, hyperbilirubinemia, days hospitalized, inbor n/outborn status; and (demographic) sex, race, urban/rural residence (according to HUD guidelines) , income (according to Florida's Children's Medical Services categories), maternal age and education, and paternal age and education. Outcome measures included the Bayley Scales of infant Development (mental and motor scales) at 6, 12, and 24 months adjusted for prematurity when indicated (gestational age less than 40 weeks) , and Stanf ord-B inet IQs at 4 years. The sample. -33-

PAGE 43

-34design, and procedures for data collection and analyses are described in this chapter. As noted in Chapter I, the questions posed for this study were 1. Is there a relationship between a set of obstetric, perinatal, and demographic variables and IQ at 4 years of age? 2. Is there a relationship between a set of obstetric, perinatal, and demographic variables and developmental indices measured at 6 months, 1 year, and 2 years of age? 3. Are developmental indices at 6 months, 1 year, and 2 years predictive of IQ at 4 years? 4. Are there differences in obstetric and perinatal variables between those infants who participated in followup evaluations and those who did not? 5. Are there demographic differences between parents who brought their children for follow-up evaluations and those who did not? pilot Study In 1984, the author conducted a preliminary study using a small sample of infants (N = 43) representing three intensive care nurseries in Florida. Most of the predictor variables of interest in the present study were compared to Bayley scores at 2 years. Correlational analyses of the data indicated that maternal education was most

PAGE 44

-35signif icantly related to the cognitive scale scores. Gestational age and birth weight also were related to both the cognitive and motor scale scores, but these correlations did not reach statistical significance. Because the sample for the pilot study was drawn from three intensive care nurseries, results may have been affected by differences in hospital treatment or testing procedures. The present study samples were drawn from one hospital to minimize the potential for this problem. The statistical analyses of the data for the present study utilized multiple regression analyses. The present study also extended follow-up to 4 years of age. The Subjects The large data base existing within the developmental evaluation component of Florida's perinatal program was the source of the basic information for this study. The families of all 110 children born between October, 1977, and February, 1980, who participated in the follow-up program for 4 years were contacted and 85 agreed to allow review of hospital records for this study. Such a large response rate is unusual and probably reflects the positive relationship between families and hospital staff developed in the ICN and over the 4 years of follow-up. Twins and children with congenital anomalies were excluded. Data sets were compiled for 77 of the subjects. All the children in this sample

PAGE 45

-36were treated at Sacred Heart Hospital's intensive care nursery and completed 4 years of developmental follow-up. Sacred Heart's ICN is a regional referral center serving the 17 rural counties of the Florida Panhandle. From 1978 through 1980, approximately 600 infants per year were admitted to the ICN and the average mortality rate was 10%. In the present sample, 58% of the infants were born elsewhere and transported to the ICN. This percentage is consistent with the total ICN admissions for that time period . During the years indicated, all infants treated at this tertiary care center were invited to participate at no charge in the follow-up program. Of those included in this sample (N = 77), about half the families were middle class (53% with incomes greater than $12,000) and 44% resided in rural communities. Additionally, the average age and educational level of mothers were 26.2 and 12.8, respectively, and for fathers, 30.2 and 12.9. According to these data, this sample was at potentially less environmental risk due to factors such as low income and low education than others often described in the literature. The infant population in this study was 58% male and 86% Caucasian. Slightly more than half (58%) were born at outlying hospitals and transported to the ICN. Respiratory distress was experienced by 69% of the infants (mild--28%, moderate — 28%, severe — 11%). Fifty-four percent were treated for hyperbilirubinemia and 78% experienced apnea.

PAGE 46

-37Descriptive data pertaining to this sample of infants who participated in follow-up are reported in Table 1. Table 1 Descriptive Data for Follow-Up Group Var i able N Mean Standard Deviation Gravidity 75 2.2 1.4 Previous Preterm Births 75 .12 .37 Birth Weight 77 2533 Grains 943.7 Grams Gestational Age 77 36 Weeks 4.5 Weeks Apgar — 1 Minute 74 6.2 2.4 Apgar — 5 Minutes 72 8.0 1.9 Days Hospitalized 77 26.8 30.6 Maternal Age 77 26.2 5.3 Maternal Education 77 12.8 2.3 Paternal Age 71 30.2 6.3 Paternal Education 71 12.9 2.6 Over time, 45% of the families invited to participate in the follow-up program did not participate. Nonparticipation may be indicative of parental resources and/or attitudes, so the characteristics of the families who chose not to attend the follow-up clinic were of some interest to this researcher. in order to identify possible differences, the follow-up sample was compared to a "no-show" sample

PAGE 47

-38(N = 64) of children born during the same time period on most of the predictor variables included in the primary analysis. Information regarding maternal education and paternal age and education was unfortunately not available for many of the no-show cases. In this sample, 60% of the infants were outborn, 62% were male, and 35% Caucasian. Their families were primarily low income (66%) and lived in rural areas (66%) . Respiratory distress was experienced by 58% of these infants (mild — 34%, moderate — 14%, severe — 5%) . Treatment for hyperbilirubinemia was required for 29%, and 14% experienced apnea. Descriptive data pertaining to this sample of infants who did not participate in follow-up are reported in Table 2. Table 2 Descriptive Data for No-Show Group Variable N Mean Standard Deviation Gravidity 64 2. 23 1.4 Previous Preterm Births 63 .03 .25 Birth Weight 62 2382 Grams 1133 Grams Gestational Age 63 35.3 Weeks 3.5 Weeks Apgar — 1 Minute 63 6.6 2.4 Apgar — 5 Minutes 61 8.3 1.5 Days Hospitalized 64 20.4 17.6 Maternal Age 64 23. 5 5.9

PAGE 48

-39Instrumentation Developmental outcome was measured at 8, 12, and 24 months using the Bayley Scales of Infant Development. This instrument is composed of two scales which measure mental abilities and psychomotor abilities of infants from birth to 30 months of age. Several aspects of reliability were reported in the test manual (Bayley, 1969) . Split-half reliability coefficients for the mental scale ranged from .81 to .93 and for the motor scale from .68 to .92. Results from a sample of 8-month-old infants were used to examine tester-observer and test-retest reliability. Mean percentages of testerobserver agreement were 89.4 for the mental scale and 93.4 for the motor scale. Mean percentages of test-retest agreement for the mental and motor scales were 76.4 and 75.3, respectively. Bayley stated that the percentage of tester-observer agreement was markedly higher than that for test-retest because the former is free from problems of the stability of behavior over time (p. 21) . Validity per se is not discussed in the manual. A report is included, however, of a study of the degree of correspondence between the Bayley MDI and the Stanford Binet IQ score. The coefficient of correlation between scores obtained on the two measures by the total group of children (age range 18-30 months) was .57, which Bayley believed indicates a substantial degree of agreement.

PAGE 49

-40The Stanf otd-Binet Intelligence Scale, Form L-M, was used as a measure of the children's cognitive competence at 4 years of age. Reliability coefficients are reported to range from .83 to .91 for 2 1/2 to 5 1/2 year olds (Terman & Merrill, 1973). The reliability data were derived from the scores of the standardization group of 3,184 children and appear to be adequate evidence for the dependability of the scale. The item selection procedures used provide a high degree of internal consistency. The manual includes reports of the percentages of subjects passing each item and the biserial correlations of each subtest with the total score. Analysis of the correlations suggests that several specific kinds of abilities contributing to overall mental ability are sampled. These are verbal, nonverbal, and manipulative skills. Evidence that Form L-M of the Stanf ord-Binet is valid is based on the fact that the same type of tests are used as in the 1937 version of the instrument. "Because of the great amount of overlap and the careful selection of subtests to be used in the revision, the probability is high that the validity of the revision will be at least equal to if not greater than the 1937 version" (Balinsky, 1965, p. 832). Procedures Because the state-wide developmental evaluation component of the Regional Perinatal intensive Care Centers

PAGE 50

-41program originated as a research program, the data available for this study conformed to previously established protocols. Families were contacted by mail with an appointment data and time. If that appointment was not kept, one further attempt was made to re-schedule the appointment by mail and/or phone. All children were evaluated at 6, 12, and 24 months using the Bayley Scales of Infant Development (Bayley, 1969) . Scores were adjusted for prematurity at 6, 12, and 24 months by subtracting the number of weeks short of full-term gestation (40 weeks) from the chronological age of the child. The Stanf ord-Binet Intelligence Scale, Form L-M (Terman & Merrill, 1973), was used to assess each child at 4 years of age. Parents accompanied their children throughout the evaluation process, which included a pediatric examination following the developmental assessment at each visit. Referrals were made as needed for further diagnostic and/or remedial services, such as opthamology, physical therapy, speech and hearing, and developmental programming. Parents were informed of the test results, and their concerns about the child's health and development were discussed. in this sample population, a number of children who had developmental delays received intervention services. Three sources were utilized to accomplish data collection for this study. The primary source of information on obstetric and perinatal variables was the medical record. Records from the follow-up clinic provided

PAGE 51

-42demographic information and test scores. When information was missing, intensive care nursery weekly staffing notes were reviewed. Data Analyses The variety of analytical approaches used in this research is evident in current literature on the developmental outcome for at-risk infants. Some interpretations of data were based on correlations alone (Kumar et al . , 1980; Littman & parmelee, 1978; Pape et al . , 1978). Some utilized only discriminant function analyses (Caputo et al., 1981; Ross et al . , 1982; Silva et al . , 1984), and others utilized t tests, correlational analyses, and multiple regression analyses in various combinations (Bee et al . , 1982; Francis-Williams & Davies, 1974; Ramey et al . , 1978; Rice & Feeg, 1985). The basic design of the present study was drawn from Siegel's efforts to develop a risk index useful in predicting outcome for an individual child, utilizing multiple regression in order to control for interactions among the predictor variables. The first question, concerning the relationship between reproductive, perinatal, and demographic variables and cognitive competence at 4 years for a population of ICN graduates, was tested utilizing a forward stepwise multiple regression. The procedure accommodated the large number of independent variables (19) compared to the sample size

PAGE 52

-43by identifying those variables most strongly associated with the outcome measure. Stepwise regression analyses were also conducted on the complete set of independent variables for each previous assessment (6, 12, and 24 months) to investigate the second question. Because of the question of the stability of the results of the multiple regression analyses due to the large number of independent variables considered, a cross-validated correlation coefficient was computed for each of these analyses. Crocker and Algina (1986) described this statistical procedure as a means to investigate the accuracy of the sample prediction equation. The third question regarding the relationship between Bayley test scores at 6, 12, and 24 months and StanfordBinet IQ at age 4 was investigated utilizing multiple regression analysis. The "min r" procedure was used in this case to test all possible regression models. The fourth and fifth questions, addressing the potential differences between the follow-up group and the no-show group, were investigated by computing chi squares for the dichotomous and categorical variables and t tests for the continuous variables which were significant in the regression analyses (p < .05). All analyses were executed using the Statistical Analysis System (3AS) (Barr et al . , 1985) .

PAGE 53

-44Summary In summary, the data were collected and analyzed in order to assess the relationships between obstetric, perinatal, and demographic factors and the mental and psychomotor development of a group of ICN graduates who were generally considered "at risk" because premature birth, other medical problems, and/or environmental problems that may negatively influence later development. The ability of infant test scores to predict IQ at 4 years of age was also examined. in addition, the characteristics of a sample of children who did not return for follow-up were compared to those who did participate in order to identify differences which may influence development. The results of the analyses are described in Chapter iv.

PAGE 54

CHAPTER IV RESULTS Data regarding obstetric, perinatal, and demographic characteristics of a sample of intensive care nursery graduates were collected to examine the relationships between those factors and developmental outcome at intervals from 6 months through 4 years. These infants were considered at risk for developmental delay because of premature birth or other medical complications during the newborn period. Comparable data were collected for a second sample of infants also treated in the intensive care nursery but who did not participate in follow-up. The obstetric, perinatal, and demographic characteristics of the two samples were then compared to identify possible differences. The results of the statistical analyses of the data are presented in this chapter. Question One; Relationship Between the Independent Variables and IQ Scores The first question for analysis was "Is there a relationship between a set of obstetric, perinatal, and demographic variables and IQ score at 4 years of age?" A stepwise multiple regression procedure was used to regress the IQ scores on variables representing gravidity and number -45-

PAGE 55

-46of previous preterm births of the mother, birth weight, gestational age, 1and 5-minute Apgar scores, respiratory distress, apnea, hyperbilirubinemia, days hospitalized, inborn/outborn status, sex and race of the infant, urban/rural residence, income, and age and education of the parents. The significant independent variables, their regression weights, and the squared partial correlations are listed in Table 3. Table 3 Regression Weights and Squared Partial Correlations for the Three Variables Predicting IQ Independent Regression Squared Variable Weight Partial Correlation Maternal Education 4.764 .405 .0001 Race -13.573 .045 .0273 Respiratory Distress -9.042 .023 ns R^ for Full Model = .473, p = .0001 ^ ^ '^^ (cross-validated correlation coefficient) N = 66 *p value for testing the hypothesis that the corresponding partial correlation is zero. ns = not significant, p > .05 Maternal education and race contributed significantly (p < .05) to the prediction of IQ score at age 4. Respiratory distress also contributed to the full model, but

PAGE 56

-47did not reach individual significance. The small difference between the R'^ and the crossval idated correlation coefficient (R^^v^ indicated that the former is a stable value (Crocker & Algina, 1986). Question Two; Contributions of the Independent Variables to Infant Development The second question for analysis was "Is there a relationship between a set of obstetric, perinatal, and demographic variables and developmental indices measured at 6 months, 1 year, and 2 years?" Stepwise multiple regression analyses were conducted for the full set of independent variables and Bayley scores at each age. The significant independent variables, their regression weights, and the squared partial correlations are listed in Tables 4, 5, and 6. Six-Month Data Days hospitalized, inborn vs. outborn status, and gestational age contributed significantly to the Bayley mental scale scores at 6 months of age. Father's educational level and 1-minute Apgar score also added to the prediction but were not significant individually. Days hospitalized and urban vs. rural residence were predictive of motor scale scores at 6 months. The 1-minute Apgar score contributed to the strength of the full

PAGE 57

-48regcession model but did not reach significance. These data are summarized in Table 4. Table 4 Regression Weights and Squared Partial Correlations for the Variables Predicting 6-Month Bayley Scores Mental Scale Independent Regress ion Squared p* Var iable VJeight Partial Correlation Days Hospitalized -.263 .1266 .0002 Inborn/Outborn -10. 215 . 1012 .017 Gestational Age -.983 .0326 .0467 Paternal Education 1.265 .0381 ns Apgar — 1 Minute 1.669 .0407 ns R'^ for Full Model = .34, p = .0002 ^ cv ~ '^^ (cross-validated correlation coefficient) N = 63 Motor Scale Independent Var iable Regression We ight Squared Partial Correlation P* Days Hospitalized Urban/Rural Apgar — 1 Minute -.23 -9.976 1.514 .1677 .0430 .0333 .0006 .0299 ns R^ for Full Model = .24, p = .0009 R^cv (crossvalidated correlation coefficient) N = 63 *p value for testing the hypothesis that the cor resoonding partial correlation is zero. ns = not significant, p > .05

PAGE 58

-49One-Year Data The results of a stepwise regression analysis of the independent variables on Bayley mental scale scores at 1 year showed a number of variables contributed significantly to the prediction. These were paternal education, • gravidity, days hospitalized, and income. Gestational age (number of weeks in utero) also entered the regression model but did not reach individual significance. The relationship between days hospitalized, income, inborn vs. outborn status, and the motor scale scores is reliable (p = .01), but relatively weak (R^ = .171). These regression analyses are summarized in Table 5. Two-Year Data Paternal education, hyperbilirubinemia, and apnea contributed significantly to the prediction of mental scale scores at 2 years. Gestational age was also included in the full regression model but did not reach individual significance . The multiple regression on the motor scale scores showed that previous preterm births, the 1-minute Apgar score, and days hospitalized were related to the 2-year scores. These data are summarized in Table 6.

PAGE 59

-50Table 5 Regression Weights and Squared Partial Correlations for the Variables Predicting 1-Year Bayley Scores Mental Scale Independent Regress ion Squared P* Var iable Weight Partial Correlation Paternal Education 2.03 .1154 .0087 Gravidity -2.8991 .0725 . 0421 Days Hospitalized -.179 .0447 .0081 Income (Medium-High) -10.875 . 0491 .0421 Gestational Age -.885 .0411 ns R-^ for Full Model = .32, p = .0003 ^ cv ~ '^^ (cross-validated correlation coefficient) N = 64 Motor Scale Independent Regression Squared p* Variable Weight Partial Correlation Days Hospitalized -.139 Inborn/Outborn Status -5.825 Income (Medium-High) -6.658 .0749 .0078 .0565 ns .0396 ns R^ for Full Model = .171, p = .0101 N = 64 *p value for testing the hypothesis that the corresponding partial correlation is zero. ns = not significant, p > .05

PAGE 60

-51Table 6 Regression Weights and Squared Partial Correlations for the Variables Predicting 2-Year Bayley Scores Mental Scale Independent Regress ion Squared P* vai; 1 aoie We ight Par t i al Correlation Paternal Education 2. 447 . 1649 .0010 Hyperbil irubinemia -8. 226 .0392 .0346 Apnea 15.397 .0373 .0126 Gestational Age -1.035 .0444 ns r2 for Full Model = . 286 , p = . 0002 QY . across— validated correlation coef f ic ient) M — C /I IN — & 4 Motor Scale Independent Regress ion Squared 9 Var iable Weight Partial Correlation Apgar--1 Minute 2.709 .1249 .0012 Days Hospitalized -.157 . 1020 .0054 Previous preterm Births 10.263 .0527 .0387 R^ for Full Model = .280, p = .0002 R cv ^ (cross-validated correlation coefficient) N = 65 *p value for testing the hypothesis that the corresponding partial correlation is zero. ns = not significant, p > .05

PAGE 61

-52Question Three; Prediction of IQ Scores from Bayley Scores The third question was "Are any of the developmental indices measured at 6 months, 1 year, and 2 years of age predictive of IQ score at 4 years?" A forward stepwise multiple regression analysis was performed on the data using the "min r" option in order to consider every possible combination of independent variables. Of the three sets of Bayley scores considered, the single best predictor of IQ at 4 years was the mental scale score at 2 years (R^ = .36). The next variable to enter the regression model was mental scale score at 6 months, which increased the predictive power (r2 = .42) . The best 3-variable model added the motor scale score at 2 years, with another increase in predictive power (R^ = ,47) . The addition of PDI at 6 months and MDI and PDI scores at 1-year resulted in a minimal increase in the R^ coefficient. The results of these multiple regression analyses are summarized in Table 7. For the purpose of interpretation of the results, the means and standard deviations for the outcome measures are reported in Table 8. Question Four; Differences Between Follow-Up and No-Show Groups on Medical Variables The fourth question posed for this study was "Are there differences in obstetric and perinatal variables between

PAGE 62

-53Table 7 Regression Weights and Correlation Coefficients for Prediction of IQ at Age Four Independent Var iable Regress ion Weight R^ for Full Model MDI — 2 yr** (best 1 variable model) .652 MDI--6 mos .261 MDI — 2 yrs .533 (best 2 variable model) MDI — 6 mos .338 MDI— 2 yrs .630 PDI — 2 yrs*** -.289 (best 3 variable model) MDI--6 mos .299 PDI — 6 mos .067 MDI — 1 yr -.128 PDI— 1 yr .137 MDI — 2 yrs .697 PDI--2 yrs -.334 (best 6 variable model) 368 424 .473 .0001 .0135 .0001 .0018 ,0001 ,0156 ,0757 ,6554 ,4848 ,4827 ,0001 0121 .481 N = 70 *p value for testing the hypothesis that the corresponding partial correlation is zero. **MDI = Bayley mental scale score ***PDI = Bayley motor scale score

PAGE 63

-54Table 3 Means and Standard Deviations for Outcome Measures N Mean Standard Deviation MDI — 5 mos* 73 106 17.9 PDI — 6 mos** 73 106.4 18.3 MDI — 1 yr 75 105.2 17.5 PDI— 1 yr 75 101 13.3 MDI — 2 yrs 74 103.7 15.9 PDI — 2 yrs 75 97.9 16 . 1 IQ — 4 yrs 77 103.6 18 . 3 *Bayley mental scale score *Bayley motor scale score

PAGE 64

-55those infants who participated in follow-up and those who did not?" The variables found in the regression models previously described and predictive of outcome were compared for the follow-up and no-show groups to identify possible differences. Chi square tests were computed for respiratory distress, inborn/outborn status, and apnea. Significant differences were found between the groups for all three of these variables. The t tests computed for the two groups resulted in no significant differences for gestational age or days hospitalized. These results are summarized in Table 9. Question Five; Differences Between Follow-Up and No-Show Groups on Demographic Variables The fifth question for analysis was "Are there demographic differences between parents who brought their children in for follow-up and those who did not?" To examine these relationships, chi-squares were computed for race and income level and a _t test was computed for maternal age. Significant differences between the groups were found for all three of these variables. Unfortunately, information about parents' education and paternal age were not available for many subjects in the no-show group so no analyses were done. These data are summarized in Table 9.

PAGE 65

-56m a D O u u o x; w 1 o 2 c D 1 3 O i-H iH o C Q) (U 3 -(J (U CQ w 0) 0 c (U d) d) i-H X3 u-l m •H Eh Q CO •iH -U tt3 a a o >-i a 3 O s: Ui I o 2 D O I 3 O O 0 a I— I IT! > >l o • (T3 c Q) O M O • 0) Q -P • 03 c 0) S u o -p c c ^ q; >-< c > rH rH rH o 1— 1 CO o o o o o o c c o o o • * • • • • 00 00 CN • • CM un rH in CO CTi 00 • • • • • CN rH i-H CN ro rH _-_ (0 c •rt QJ 0) u O m 0) in o 00 > uT • • D CTi • 3 I" •H in rn VJ rH OVO o\o o\o <1\° CM O ro in CN CN CN >;)' rH CN >^ rH in • in ro m r-l CN ro ro ro ro ro >^ in ^ ^ R3 — C M CO »j O 00 fO OJ XI 4J cn • • a • 0) 3 a; 00 (0 •r< in U3 O >i iH u x: o\o o\o (N 00 CN 00 VD ro in iH in CN 00 00 00 in CN • • • in ro CN CM cn r~ in 0) r~ rH 03 0) •1-4 rH QJ u XI c cn (0 (0 u < QJ > •rH o N U >l Xi rH •rH u u cn 4J (0 rH •H > o tn D c ro 4J QJ O o -p a rH ro SH \ rH -rH ro c -U Qj u M tn u ro ro cn cn -rH o QJ -P cn o o cn Q ^ C tn >i I e QJ QJ c a QJ to QJ s OS 1— 1 <: u Q Q rH -p to o QJ C c g QJ QJ O QJ cn II O U -P < to c ro tn 05 I-H ;2 c in o A p c ro O c

PAGE 66

-57Summary In summary, a variety of obstetric, perinatal, and demographic variables were predictive of developmental indices from 6 months through 4 years of age. No consistent group of variables can be identified across time, although days hospitalized appeared in five of the seven regression models. The parents' level of education also appeared in three of the analyses. According to the regression results reported in Table 6, the prediction of IQ scores at 4 years from Bayley scores can be done best from the tests administered most closely in time to age 4. The correlations between the developmental indices for this sample are similar to findings of several other studies (Wolking, Packer, Carter, & Resnick, 1985) which found moderately strong correlations between 2-year and 4-year test scores. Comparison of the characteristics of those infants who returned for follow-up and those who did not indicates differences existed on respiratory distress, inborn vs. outborn status, and apnea. Differences were also found on the demographic factors of race, income, and maternal age. Discussion and implications of these results are presented in Chapter v.

PAGE 67

CHAPTER V DISCUSSION In this study, the relationships between a group of obstetric, perinatal, and demographic variables and developmental outcome were investigated for a group of intensive care nursery graduates. Characteristics of that sample were also compared to those of an additional sample of ICN graduates who were not followed after discharge. The discussion of the results of the data analyses and comparisons with previous findings are presented in this chapter . Contribution of the Independent Variables to Developmental Outcome The first two questions of interest in this study concerned the relationship between a group of obstetric, perinatal, and demographic variables and developmental outcome. Seven multiple regression analyses were conducted using Bayley mental and motor scale scores at 6 months, 1 year, and 2 years and Stanf ord-Binet IQ scores at 4 years as the dependent measures. No consistent group of variables emerged which was predictive of developmental outcome across ages, although several different groups of variables were predictive at each age. -58-

PAGE 68

-59Each of the obstetric variables was included in one regression model. Gravidity was included in the regression model that predicted 1-year mental scale scores and previous preterm births contributed to the prediction of 2-year motor scale scores. The perinatal variables were more prevalent than obstetric or demographic variables in the prediction of Bayley scores, with six of the nine variables present in one or more of the analyses. Days hospitalized was included in all except the models predictive of 2-year mental scale scores and IQ at 4 years. The 1-minute Apgar score contributed to the prediction of both mental and motor scores at 6 months and motor score at 2 years. All three of the regression models for mental scale scores included gestational age. Additionally, inborn/outborn status appeared in two analyses and apnea and hyperbilirubinemia each appeared in one. Demographic variables were not consistent in their contributions to the outcome measures, with the exception of paternal education, which was included in the prediction equations for mental scale scores at 6 months, 1 year, and 2 years. Income was included in the models for both mental and motor scores at 1 year. Urban vs. rural residence occurred once, in the model for 6-month motor scores. Interestingly, none of the variables in the models predicting outcome during infancy were included in the prediction of Stanf ord-B inet scores at 4 years. The

PAGE 69

-60demographic variables of maternal education and race were highly related to IQ (R^ = .45) , with respiratory distress included in the model but not individually significant. In summary, there was a greater tendency for the perinatal factors to be significant predictors of mental and motor development in infancy than at 4 years of age. Obstetric factors did not appear to be especially powerful predictors. Demographic factors, specifically education of the parents, were consistently influential over time and emerged as the most significant predictors of cognitive competence at 4 years of age. This suggests that medical high risk factors may be mediated over time by environmental factors. An issue remaining for future study relates to the heterogeneity of this sample population. The methodology used in this study did not include analyses of relationships between the predictor variables and outcome for specific groups, such as young mothers. Different patterns of prediction could emerge for homogenous populations. Prediction of IQ Scores from Bayley Scores Another issue addressed in this study was the relationship between infant test scores and IQ scores at 4 years. it is not surprising that the single best predictor of IQ was the measure of cognitive functioning closest to it in time, the 2-year Bayley mental scale score. This finding is consistent with other studies of high risk populations (Siegel, 1983; Wolking et al . , 1985).

PAGE 70

-61A less typical finding was that the 6-month mental score entered the multiple regression equation next, followed by the motor score at 2 years (R^ = .47). It would appear that the 1-year Bayley scores were the "weak link" here . The means and standard deviations calculated for the developmental indices were another unexpectedf-i-nd-ing . They were quite similar to the means and standard deviations of the Bayley scales and the Stanf ord-Binet , which is an indication that the developmental outcome of this study sample fell within normal ranges. Differences Between Follow-Up and No-Show Groups The final issue addressed in this study concerned potential differences between the sample of 77 children who participated in 4 years of follow-up and the sample of 64 who did not, presumably by parental choice. Although significant differences were found between the two groups for respiratory distress, inborn vs. outborn status, and apnea, no differences were found for gestational age or days hospitalized which appeared most consistently in the regression models predictive of developmental outcome. It was not clear based on these variables whether or not the no-show group was more acutely ill in the perinatal period than the follow-up group. Since demographic factors were strongly related to developmental outcome in the multiple regression analyses.

PAGE 71

-62maternal age, race, and income level were compared for the two groups. The no-show group contained a significantly younger group of mothers. It was difficult to determine if the no-show group was at a disadvantage based on mother's age since this variable was not predictive of outcome in this study. Race and income level were also significantly different between the two groups, however. The follow-up group contained significantly higher numbers of white families and of higher income families. These results indicated that on two of the three variables present in the regression model predictive of IQ, the no-show group was at a potential disadvantage. These findings raise several questions. Could the mean test scores of high risk populations reported here and in the literature be overestimated because of the number of potentially low-scoring or handicapped infants lost to follow-up? Is the follow-up process itself an intervention that supports normal development? One reason that low-income parents who may be less educated may not return for follow-up could be that they feel more threatened by the ICN environment during their child's hospitalization. If they feel alienated at this early stage, they could be less likely to perceive any benefits of developmental follow-up. Other possible interpretations of these data are that younger mothers of less economic means are less able to organize their resources to participate in noncritical services for their

PAGE 72

-63children or that these services are more appealing to upper income, educated parents. These issues warrant consideration both in future research and in clinical practice . Comparison to Other Studies Since the variables chosen for this study were suggested by the risk index developed by Siegel (1982), it is of interest to compare these results to those obtained in her study. The sample population for Siegel 's study consisted of 42 preterm infants (birth weight less than 1501 grams) from a working-class community in Ontario, Canada. Comparisons of the demographic characteristics of the Canadian group and the sample of the current study suggest that the average socioeconomic status of the former was lower than that of the current study. All of the participants in the Canadian study were white, in contrast to 14% black participants in this study, and from an urban community, in contrast to the large rural region where 44% of the sample in the present study resided. Maternal age was the same for both groups. Because Siegel 's study focused on preterm infants, the average birth weight of 1236 grams and gestational age of 30.3 weeks were different from the nonhomogeneous sample in the present study which were 2533 grams and 36 weeks, respectively. Despite the differences in the sample populations, results of the multiple regression analyses in both studies

PAGE 73

-64have some similarities. For Siegel's group, evaluated at 5 years of age using the McCarthy Scales of Children's Abilities, the variables of SES, maternal education, previous spontaneous abortions, birth order, and respiratory distress contributed to the prediction of the General Cognitive Index (R^ = .51). Two of these variables (maternal education and respiratory distress) were predictive of Stanf ord-Binet IQ at 4 years in the present study. The shift toward environmental factors more strongly predicting outcome at 4 years than medical factors is consistent with Sameroff's (1982) transactional model. The results of the current study are consistent with those of previous studies in which it was found that environmental factors are more predictive of developmental outcome as children reach school age. Social class was found to be the most highly predictive variable of IQ at between 4 and 12 years of age in a study of premature babies by FrancisWilliams and Davies (1974). For another group of premature and full-term children, also evaluated in middle childhood, mother's IQ and social class correlated with IQ (Caputo et al., 1981). Bee et al . (1982) followed a group of low-risk first-borns over four years. They found that children whose mothers had more than a high school education had significantly higher mental test scores beginning at 2 years, while these authors found that no individual variable was highly predictive of IQ score, the child's

PAGE 74

-65previous test performance, or a combination of maternal education and family support accounted for 20% to 50% of the 2 variance in IQ. In the present study, a similar R was obtained (.47) using either infant test scores or a combination of maternal education, race, and respiratory distress. In contrast, Silva et al . (1984) found that neither SES nor maternal intelligence predicted cognitive development for a sample of preterm babies or a sample of small for gestational age babies. Summary Although causality cannot be implied, the interpretation of the results of this study was that a combination of obstetric, perinatal, and demographic variables were moderately predictive of developmental outcome in infancy and early childhood. In addition, the prediction of IQ score at 4 years utilizing infant test scores supports other current findings that such prediction is more likely to be successful for at-risk populations such as ICN graduates than for normal populations (Wolking et al., 1985). Two unique characteristics of this sample population, that it was nonhomogeneous in regard to medical variables and nearly half of the subjects resided in rural areas, are key aspects of the contribution of this study to understanding of the field. In many follow-up studies of high risk infants, the samples were selected on the basis of a specific medical

PAGE 75

-66criterion, such as low birth weight, intraventricular hemorrhage, or bronchopulmonary dysplasia. Since the findings in the literature do not support that these medical variables generally predict developmental outcome beyond 2 years of age, this study sample was selected on the basis of at-risk birth status alone, in order to insure a variety of medical complications. It is important to learn more about this population of intensive care nursery graduates because they are entering community services and public schools in growing numbers. Rural populations are also not widely represented in the literature probably because resources for research tend to be located in urban centers. Urban vs. rural residence was not specifically related to outcome in this study. In the follow-up sample, a higher proportion of rural infants were white and were transported to the ICN. Apparently these families had both the resources and the interest to participate in 4 years of follow-up, however. The primary clinical implication of this study is that consideration regarding allocation of intervention resources should include environmental factors such as the parents' educational level as well as medical variables indicative of severity of illness. These aspects of developmental risk need further study to more precisely identify the factors related to outcome for individual children. In particular, the specific components of parent-child interaction and the home environment that support early learning need further

PAGE 76

-67exploration. This knowledge could contribute to improved developmental care in the ICN as well as more effective early intervention strategies. Much has been accomplished through improved prenatal care to reduce the incidence of high risk birth. These efforts for primary prevention should continue, along with expanded secondary prevention services as early in the life of the at-risk infant as possible. The current trend in public education to extend special education services to the preschool population is encouraging in that it may result in the provision of more comprehensive services to at-risk infants and young children. it is important, however, to consider the family and the environment of the child in planning and implementing intervention services. Communities must organize resources from many agencies and disciplines, including health care, education, and social services, to reduce the incidence and severity of long-term handicap.

PAGE 77

REFERENCES Aylward, G.P., & Kenny, T.J. (1979). Developmental followup: Inherent problems and a conceptual model. journal of Pediatric Psychology , _4, 331-343. Balinsky, B. (1965) . Review of Stanf ord-Binet Intelligence Scale. In O.K. Euros (Ed.) , Sixth mental measurements yearbook (pp. 831-832). Highland Park, N J : Gryphon Press . Barr, A., Goodnight, J., Scott, J., & Helwig, J. (1985). Users guide to SAS . Raleigh, NC: SAS Institute. Bayley, N. (1969) . Bayley scales of infant development . New York: The Psychological Corporation. Bee, H.L., Barnard, K.E., Eyres, S.J., Gray, C.A. , Hammond, M.A., Spietz, A.L., Snyder, C, & Clark, B. (1982). Prediction of IQ and language skill from perinatal status, child performance, family characteristics, and motherinfant interaction. Child Development , 53 , 1134-1156. Bozynski, M. , Nelson, M. , Rosati-Skertich, C. , O'Donnell, D., & Naughton, P. (1984). Two year longitudinal follow-up of premature infants weighing 1,200 grams at birth: Sequelae of intracranial hemorrhage. Journal of Developmental and Behavioral Pediatrics , _5, 346-352. Broman, S.H., Nichols, P.L., & Kennedy, W.L. (1975). Preschool IQ: Prenatal and early developmental Correlates . New York: John Wiley & Sons. Caputo, D.V. , Goldstein, K.M., & Taub , H.B. (1981). Neonatal compromise and later psychological development: A 10-year longitudinal study. In S. Friedman & M. Sigman (Eds.), Preterm birth and psychological development (pp. 353-386). New York: Academic Press. Cohen, S., Parmelee, A., Beckwith, l., & Sigman, M. (1986). Cognitive development in preterm infants: Birth to 8 years. Journal of Developmental and Behavioral Pediatrics, 7, 102-110. ~ -68-

PAGE 78

-69Coolman, R., Bennett, F., Sells, C, Swanson, M. , Andrews, M., & Robinson, N. (1985). Neuromotor development of graduates of the neonatal intensive care unit: Patterns encountered in the first two years of life. Journal of Developmental and Behavioral Pediatrics , _6 , 327-333. Crocker, L., & Algina, J. (1986). Introduction to classical and modern test theory . New York: Holt, Rinehart, & Winston. Fitzhardinge , P. (1984, May). Follow-up of high-risk newborns. Paper presented at the First Annual Eric Denoff Memorial Symposium on Child Development, Providence, RI. Francis-Williams, J., & Davies, P. (1974). Very low birth weight and later intelligence. Developmental Medicine and Child Neurology , 16 , 709-728. Hofheimer, J. A. (1979). The adolescent mother and her infant: Correlates of transaction and development (Doctoral dissertation. University of Florida, 1979) . Dissertation Abstracts International , 40 , 5717A. Holstrum, W.J. (1979). The prediction of three year developmental status of high risk infants (Doctoral dissertation, university of Florida, 1979) . Dissertation Abstracts International , 40 , 4882A. Hunt, J.V. (1981) . Predicting intellectual disorders in childhood for preterm infants with birth weights below 1501 gm. In S. Friedman & M. Sigman (Eds.), Preterm birth and psychological development (pp. 329-351) . New York: Academic Press. Goldstein, K. (1976) . The effects of prenatal and perinatal complications on development at one year of age. Child Development , 47 , 613-621. Kafatos, A.B., & Pantelakis, S.N. (1982). Factors related to perinatal morbidity and mortality. Pae diatrician, 11, 27-44. Knobloch, H. , & Pasamanick, B. (1960). Brain damage and reproductive casualty. American journal of Orthopsychiatry , 30 , 293-305. Kopp, C.B., & Krakow, J.B. (1983). The developmental ist and the study of biological risk: A view of the past with an eye toward the future. Child Developmen t, 54, 1086-1108. —

PAGE 79

-70Kumar , S.P., Anday, E.K., Sacks, L.M., Ting, R.Y., & Del ivor ia-Papadopoulas , M. (1980). Follow-up studies of very low birth weight infants (1,250 grams or less) born and treated within a perinatal center. Pediatrics , 66 , 438-444. Landry, S.H. (1984) . Differential outcomes associated with early medical complications in premature infants. Journal of Pediatric Psychology , 9, 385-401. Littman, B., & Parmelee, A.H. (1978). Medical correlates of infant development. Pediatrics , 61 , 470-474. McCall , R.B. (1981). Early predictors of later IQ: The search continues. Intelligence , _5, 141-147. Mccormick, M.C., Shapiro, S., & Starfield, B. (1984). High risk young mothers: Infant mortality and morbidity in four areas in the United States, 1974-1978. Amer ican Journal of public Health , 74 , 18-23. Pape, K.E., Buncic, R.J., Ashby, S., & Fitzhardinge, P.M. (1978) . The status at two years of low-birth weight infants born in 1974 with birth weights of less than 1,001 gm . Journal of Developmental and Behavioral Pediatrics , 3, 22-24. Ramey, C. , Stedman, D., Borders-Patterson, A., & Mengel , W. (1978). Predicting school failure from information available at birth. American Journal of Mental Deficiency , 82, 525-534. Resnick, M. , Bauer, C. , Cupoli, M. , Ausbon, W. , & Evans, J. (1983). Florida Regional Intensive Care Program developmental evaluation component-early developmental outcome. journal of Florida Medical Associatio n, 7 0, 833-837. Rice, B., & Feeg, V. (1985, Jan. -Feb.). First-year developmental outcomes for multiple-risk premature infants. Pediatric Nursing , pp. 30-35. Ross, G., Schechrer, S., Frayer , W. , & Auld , P. (1982). Perinatal and neurobehavioral predictors of one-year outcome in infants < 1500 grams. Seminars in Perinatology , 6, 317-326. Rubin, R., & Balow, B. (1979). Measures of infant development and socioeconomic status as predictors of later intelligence and school achievement. Developmental Psychology , 15 , 225-227.

PAGE 80

-71Sameroff, A. (1982) . The environmental context of developmental disabilities. In D. Bricker (Ed.)» Intervention with at-risk and handicapped infants; From research to application (pp. 141-152). Baltimore: University Park Press. Sameroff, A., & Chandler, M. (1975). Reproductive risk and the continuum of caretaking casualty. In F. Horowitz, E. Hether ington , S. Scar r-Salapatek , & G. Siegel (Eds.), Review of child development research , Vol. 4 (pp. 187-244). Chicago: University of Chicago Press. Siegel, L.S. (1982). Reproductive, perinatal, and environmental variables as predictors of development of preterm (< 1501 grams) and full term children at 5 years. Seminars in Perinatology , 274, 278. Siegel, L.S. (1983). The prediction of possible learning disabilities in preterm and full term children. In T. Field & A. Sostek (Eds.), Infants born at risk; Physiological, perceptual, and cognitive processes (pp . 295-315). Orlando, FL: Grune & Stratton. Silva, P., McGee, R. , & Williams, S. (1984). A longitudinal study of the intelligence and behavior of preterm and small for gestational age children. Journal of Developmental and Behavioral Pediatrics , _5, 1-5. Smith, A., Flick, G., Ferris, G., & Fellman, A. (1972). Prediction of developmental outcome at seven years from prenatal, perinatal, and postnatal events. Chi Id Development , 43 , 495-507. Stewart, A., Reynolds, E. , & Lipscomb, A. (1984, May). Outcome for infants of very low birth weight: Survey of world literature. The Lancet , pp. 638-1041. Terman, L., & Merrill, M. (1973). Stanf ord-B inet Intelligence Scale . New York: Houghton Mifflin Co. Werner, W. , Simonian, K., Bierman, J., & French, F. (1967). Cumulative effects of perinatal complication and deprived environment on physical, intellectual and social development of preschool children. Pediatr ics , 39, 490-505. Wolking, B. , Packer, A., Carter, R. , & Resnick, M. (1985). Predicting preschool Binet IQs from infant Bayley MDls for high-risk infant"i~ [ Manuscript submitted for publ icat ion .

PAGE 81

BIOGRAPHICAL SKETCH Deborah Cole Goldberg was born August 16, 1952, near Boston, Massachusetts. Her family moved South in 1965 and she completed high school in Atlanta, Georgia, in 1970. She attended Eckerd College (formerly Florida Presbyterian College) where she was active in student government and performing arts. In 1974 she graduated with a B.A. in literature and relocated to Pensacola where she earned an M.A. in early childhood education at the University of West Florida. She then worked for 5 years in the Child Development Program of the Escambia Community Mental Health Center. Deborah continued graduate studies at the University of Florida and received further clinical training through the Department of Pediatrics. She completed the Ph.D. in instruction and curriculum in 1987. Deborah was married in 1978 to Richard M. Goldberg, a clinical psychologist. They have two children, Rachel and Jessica. She currently works as a child development specialist in private practice with her husband. She has been Coordinator of Sacred Heart Hospital's Developmental Evaluation Clinic and consultant to the intensive Care Nursery since 1981. -72-

PAGE 82

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 of Doctor of Philosophy. C C C.^l ^< / / ..c Athol B. Packer, Chairman Associate Professor of Instruction and Curriculum 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 of Doctor of Philosophy. JaTH^t Larsen "^-^ Profbsaor o^ Counselor Education 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 of Doctor of Philosophy. Associate Professor of Psychology University of West Florida

PAGE 83

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 of Doctor of Philosophy. MichaelXB. Resnick Assistant Professor of pediatrics 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 of Doctor of Philosophy. Dorene Ross Associate professor of Instruction and Curriculum This dissertation was submitted to the Graduate Faculty of the College of Education and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August 198 7 )ea'h',' 'College of' Ed"i Dean, Graduate School


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E4TICEPHA_VX8702 INGEST_TIME 2015-03-09T20:58:59Z PACKAGE AA00028864_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES