Previous - Next
Problems of definition and interpretation
Cerebral palsy
Minimal neurologic dysfunction
Sensory loss or handicap
Conclusions
References
Discussion
RL Goldenberg¹, HJ Hoffman² and SP Cliver¹
Correspondence: Robert L. Goldenberg
¹University of Alabama at
Birmingham, 618 South 20th Street, OHB 560, Birmingham, AL 35294-7333, USA
²Epidemiology Statistics, and Data System
Branch, National Institute on Deafness and other Communication Disorders, National
Institute of Health, Bethesda, MD 20892-7180, USA
Despite many methodological difficulties, studies evaluating the relationship between being small for gestational age (SGA) at birth and various measures of adverse neurologic outcome generally show significant associations. Nevertheless, cerebral palsy is rarely found in SGA infants. Minimal neurologic dysfunction is more commonly seen in males and lower socioeconomic SGA children, and is often associated with attention deficits, hyperactivity, clumsiness, and poor school performance. Vision and hearing are generally not disturbed in SGA infants.
The major purpose of this paper is
to explore the question as to whether infants who are born small for gestational age (SGA)
are neurodevelopmentally different than infants who are born with appropriate size for
gestational age (AGA). Although much has been written on this subject, there remains a
great deal of controversy about whether or not SGA status is associated with neurologic
damage (Chard et al, 1993; Aylward et al, 1989; Breart and Poisson-Salomon,
1988; Allen, 1984).
One of the major reasons for the controversy
surrounding the relationship of SGA to adverse neurologic outcomes is that infants who are
considered SGA in one study are not deemed SGA in other studies (Goldenberg et al,
1989). In addition, some studies include only the most severely SGA infants, while in
others the entire spectrum of SGA infants is included in the analysis. These differences
in study population definitions are potentially magnified by the fact that even though
most authors define SGA as including infants born below the tenth percentile birth weight
for gestational age, the standards used to define this tenth percentile cutoff are very
different from one study to the next. In addition, not only are different cutoffs used,
but some standards are race- or sex-specific while others are not. Because female infants
generally weight less than male infants do, studies not using sex-specific standards will
have a larger proportion of the female population and a smaller proportion of the male
population. Similarly, since black infants tend to weigh less than white infants, using
standards that are not race-specific will identify more black infants than white infants
as SGA. Not only are the standards different, but the gestational ages of the infants
studied are often different as well. Since it is likely that preterm SGA infants will have
different outcomes than term SGA infants, it is important to define the gestational age
ranges in populations under study.
Infants with various types of chromosomal or structural anomalies tend to weigh less than normal infants (Jones, 1978). The SGA infant population, therefore, potentially includes more infants with chromosomal and structural anomalies than the control population. When studying neurologic outcome associated with SGA, it is important to define whether these infants are included or excluded from the study population. Similarly, it is important to determine whether infants who are severely disabled were included or not. As an example, infants with severe disabilities are often excluded from the population when minor neurologic disabilities or IQ scores are presented - potentially giving a falsely high impression of the cognitive or neurologic capacity of the entire SGA population.
Many study designs have been used to evaluate neurodevelopment outcome in SGA infants. One of the most common is a cohort study in which the entire birth population is evaluated prospectively and the outcome associated with being in the lower tenth percentile birthweight for gestational age is compared to either the entire non-SGA population or a representative sample. This type of study is labor-intensive, in that only 10% of the children are SGA, and the outcomes of interest, which may include cerebral palsy or some other relatively rare outcome, occur only occasionally. Therefore, these studies generally do not have sufficient power to define significant relationships between SGA and major but rare outcomes, which are dichotomous in nature. Instead, many of these cohort studies tend to look at continuous outcomes such as the mean IQ score.
A retrospective case control study is another type of study used to determine adverse outcomes associated with SGA status. In these studies, children with the outcome of interest such as mental retardation or cerebral palsy are first identified. Then the rate of SGA in the cases is compared to the rate of SGA in the control group, and the relative risk of the association between SGA and the neurodevelopment handicap is determined. This type of study provides much of the evidence for the relationship between cerebral palsy and SGA.
One major difficulty in dealing with adverse neurologic outcomes related to SGA is that the outcomes are not necessarily stable over the child's lifetime. As an example, minimal neurologic dysfunctions such as hyperactivity and poor concentration are not apparent early in the child's life, and poor school performance obviously cannot be determined until the child enters school. Some poor outcomes, such as hyperactivity, may improve over time. It is, therefore, very important to specify the age at which an adverse outcome is observed, and to make sure the study age is comparable between the SGA infants and their controls.
There are many neuro-development outcomes one could assess in relationship to SGA. The most frequently measured outcome is cognitive function, which is usually determined by an IQ test. The outcome of interest is either dichotomized into mental retardation defined by an IQ below 70 or 75, or is presented as the continuous variable of IQ. Abnormalities of motor function include cerebral palsy as well as various measurements of decreased fine and gross motor skills and clumsiness. Minimal neurologic dysfunction is often defined as a combination of hyperactivity, poor attention span, and clumsiness, often associated with poor school performance. Finally, another outcome of some interest is sensory impairment, particularly related to vision and hearing. Each of these outcomes, except cognitive function (which will be discussed in McGregor's paper), will be evaluated for its association with SGA status.
Before evaluating the specific outcomes associated with SGA, it is important to emphasize that many of the factors associated with, or causative for, SGA may have a direct effect on neurologic or sensory development, independent of their effect on fetal growth. As an example, many babies with chromosomal abnormalities are small for gestational age (Breart and Poisson-Salomon, 1988). These infants also have several different types of neuro-development handicaps, but it would seem inappropriate to consider their being SGA as causative for the neurodevelopment handicaps. Instead, the chromosome anomaly itself is likely responsible for both the SGA and the neuro-development handicap. Several other risk factors, such as congenital infection, various types of structural abnormalities, drug use, alcohol use, and smoking all may cause growth retardation and may cause neurodevelopment problems, but the growth retardation itself may not be in the causative pathway for the neurologic dysfunction.
Finally, it is extremely important to understand that being born SGA is only one factor of many which may contribute to neurodevelopment variability in childhood. These factors may include maternal intelligence, maternal education, socioeconomic status, home environment and attendance at preschool (Chard et al, 1993; Aylward et al, 1989; Breart and Poisson-Salomon, 1988; Allen, 1984; Goldenberg, 1996). Since at least some of the factors associated with the baby being born SGA also may influence the child's neurodevelopment functioning after birth independently of the risk factor's effect on infant size, understanding the direct causative relationship between being born SGA and a neurodevelopment problem later in life is often very difficult.
One of the very difficult issues dealing with adverse neurologic outcome in SGA infants is the relationship among SGA, hypoxia, and the adverse outcome. For example, it is often not clear if any specific poor neurologic outcome is related to the SGA or to an associated hypoxia. Conceptually, chronically poor placental function, is a cause of at least a portion of the SGA. If the SGA is caused by poor transport of essential nutrients and oxygen across the placenta, one can picture that poor fetal oxygenation, especially during labor, is a likely phenomenon. What is known, is that poor oxygenation and the resultant acidemia is seen more frequently in SGA infants than in other infants (Low et al, 1972). However, there is not a universally agreed upon definition for hypoxia, asphyxia, or acidemia. The nature of the association between these conditions and cerebral palsy, as well as other neurologic outcomes, is therefore currently under debate. The question that naturally arises is whether SGA infants who are not subject to decreased oxygenation in the perinatal period are at risk for poor neurologic performance.
In a study which addressed this issue, Berg (1989), using data from the National Collaborative Perinatal Project, found that in the absence of hypoxia-related factors, neither symmetric nor asymmetric IUGR children were at higher risk for neurologic morbidity compared to non-IUGR children at seven years of age. However, in the presence of perinatal hypoxia-related factors, IUGR children were more likely to be neurologically abnormal compared to non-IUGR children. Similar conclusions have been reached by a number of other authors, including Uvebrant and Hagberg (1992), who agreed that at least part of the increase in cerebral palsy in term SGA infants was associated with asphyxia. Other authors have confirmed that low ponderal index SGA infants generally had normal outcomes unless they were asphyxiated at birth (Low et al, 1978; 1982; 1992). Asphyxia more commonly occurred in primiparous women who develop severe preeclampsia and poor uterine blood flow, especially prior to term. This may be the reason that in certain studies SGA infants born in association with maternal hypertension were more likely to have low IQs and mild neurologic handicaps than were other SGA infants. Ounsted et al (1984) evaluated neurologic outcome in SGA infants at seven years of age or more, and noted that given good obstetric and neonatal care and a favorable environment in which to grow up, the outlook for normal development of SGA infants born in the 1970's was much improved over that for SGA children born earlier. At least part of this improvement has been associated with less asphyxial injury in infants born to preeclamptic mothers.
A related issue is whether SGA,
which occurs solely on the basis of acute maternal nutritional deprivation, is associated
with long-term neurologic deficiencies. One of the few natural experiments in which this
issue can be evaluated, the Dutch famine during World War II, showed that there was no
increase in adverse neurologic outcomes with nutritional deprivation alone (Susser and
Stein, 1977). These data suggest that most of the adverse neurologic outcomes associated
with SGA in otherwise healthy populations are due to non-nutritional causes.
Cerebral palsy is found in 1 to 2 per 1000 children. It has been well known for nearly 150 years that low birthweight infants are at greater risk for this outcome than more average size infants are. However, it has only been in the last 40 years that a relatively clear distinction has been made between babies born preterm and those born small for gestational age, whether born at term or preterm. As emphasized above, the boundaries between SGA and AGA are not well defined, and have shifted over time. The incidence of cerebral palsy is highly dependent on gestational age at birth, ranging from as high as 250/1000 for infants born at 23 or 24 weeks to 30/1000 at 30 weeks, to 5/1000 at 36 weeks, to 1/1000 for infants born at 40 weeks (Escobar et al, 1991). It is therefore important to define the impact of SGA on cerebral palsy against this background. In doing so, it becomes very obvious that evaluating the relationship between SGA and cerebral palsy in term infants requires huge sample sizes and the answer is not likely to be elucidated by relatively small prospective cohort studies.
We should also emphasize that the composition of the preterm population is changing, especially as more and more very low birthweight infants are surviving. In addition, the preterm population consists of infants that are not only preterm, but SGA as well. The outcome for these latter infants may be different than for infants who are only preterm.
Despite these considerations, review of the available data regarding the relationship between SGA and cerebral palsy does allow us to reach some conclusions. As an example, babies born at term and later diagnosed with cerebral palsy tend to be lighter at birth than those who are not diagnosed as having cerebral palsy (Ellenberg and Nelson, 1979; Blair and Stanley, 1990). Nevertheless, most small babies, even if they are very SGA, do not end up with cerebral palsy, and most children with cerebral palsy are of normal birthweight. Translated into a measure of risk, being SGA at term appears to double or triple the risk of cerebral palsy, i.e. from 1 or 2 per 1000 live births to 2 to 6 per 1000 live births. Factors that appear to be associated with increased risk are the severity of the SGA, male sex, and whether asphyxia or other risk factors for cerebral palsy were present.
In many studies of preterm infants
and cerebral palsy, preterm SGA infants appear to have less cerebral palsy than AGA
preterm infants (Hack et al, 1989; Robertson et al, 1990; Saigal et al,
1990; Sung et al, 1993). These SGA preterm infants are frequently born in
conjunction with maternal preeclampsia, and most studies suggest a protective effect
against cerebral palsy of preeclampsia in preterm SGA infants. However, even this
conclusion is not unanimous. Robertson et al (1990), for example, studying £ 1500 g, less than
tenth percentile infants, found that preterm SGA infants were not at greater risk to be
disabled with cerebral palsy than were AGA birthweight or gestational age matched
controls, but all were at substantially greater risk than a term AGA comparison group. On
the other hand, Hack et al (1989), studying £ 1500 g infants, observed an 8% rate of cerebral palsy in
the AGA group, but only a 3% rate (p <.01) in the SGA group. Veelken et al, also
showed that < 1500 g SGA infants had less cerebral palsy than < 1500 g preterm AGA
infants (Veelken et al, 1992). We interpret these data to indicate that in cohorts
defined by birthweight, SGA infants have less cerebral palsy. However, in cohorts defined
by gestational age, SGA infants had similar or higher rates of cerebral palsy.
The relationship of SGA to minimal neurologic dysfunction (MND) is clearer than its relationship to cerebral palsy, predominantly because MND is more common, occurring in 10 to 35% of SGA populations, and very large sample sizes are not necessary to define this relationship (Chiswick, 1985). On the other hand a major problem in quantifying the relationship is that MND is not very well defined, and the term is used differently in different studies. In one of the first studies to evaluate outcome of SGA infants, Fitzhardinghe and Steven (1972) found very high rates of MND in a group of SGA children, but did not have appropriate controls. Hadders-Algra et al (1988), evaluated MND in four groups. Fifteen percent of full term AGA infants had MND compared to 24% of full term SGA infants. Preterm AGA infants had a 26% rate of MND while 37% of preterm SGA infants had MND. In a later study, Hadders-Algra and Touwen (1990) found little difference in the rate of MND between term AGA infants (16%) and the less severe term SGA infants (22%). On the other hand, the most severe SGA infants had more than double the rate of MND (41%) compared to the term AGA infants (16%). In comparison, both the preterm AGA infants (30%) and the preteen SGA infants (40%) had higher rates of MND.
Ounsted et al (1984), found decreased verbal ability, practical reasoning and motor coordination in SGA infants compared to AGA infants, and, interestingly, found less of these conditions in LGA infants. Therefore, there appears to be a continuum in teems of infant size in relationship to MND. On the other hand, Low et al (1978; 1982; 1992), reported no significant differences in motor, cognitive handicap or developmental delay between term SGA and control infants. Martikainen (1992) found that preterm SGA infants were at a two-fold greater risk for abnormal neurologic development than term infants, and symmetric SGA infants were at greater risk for MND than asymmetric infants. Evaluating hypotonia as the outcome, Touwen et al (1988), found this condition in 11% of full-term SGA children at age six, but in none of the control term AGA children. Walther (1988), found a moderate increase in school failure in SGA infants, probably secondary to 'soft' neurologic signs and behavioral characteristics such as hyperactivity, poor concentration and clumsiness.
Veelken et al (1992), found an increase in minor neurologic abnormalities in < 1500 g SGA infants compared to control infants. Similarly, Smedler et al (1992), studying 15 < 1500 g SGA infants at 10 years of age, found that these infants scored significantly lower on measures of visio-spatial ability, non-verbal reasoning, strategy formation and gross motor coordination. Low et al (1992), evaluating learning deficits at 9-11 years in both term and preterm SGA infants, found an approximate doubling of learning deficits compared to gestational age matched controls.
Spinillo et al (1993, 1995),
in a series of studies, found that SGA infants of hypertensive mothers were snore likely
to have mild neurologic findings such as hypotonia than idiopathic SGA infants. Severely
SGA term infants (mean birthweight 1947 g) had more transient neurologic signs than less
severely SGA infants. In a very interesting study of Israeli Army recruits, those born SGA
had about a 50% increase in minor neurologic abnormalities even after adjustment for
socioeconomic status, birth order and intrapartum events (Paz et al, 1995). In many
studies, the severity of school problems in SGA children was influenced by the child's
social class, sex, parental abilities, home environment, and preschool education
(Parkinson et al, 1981). We interpret these studies to indicate that SGA infants,
whether at term or preterm, are more likely to be diagnosed with various components of
minimal neurologic dysfunction than non-SGA infants.
There are some, but not many,
studies in which sensory loss has been evaluated in SGA infants. For example, visual
acuity has generally not been found to be reduced in SGA infants (Hack et al, 1989;
Hermans et al, 1992). Similarly, hearing acuity has not generally been found to be
diminished in SGA infants (Low et al, 1982; Hack et al, 1989). On the other
hand, integration of these sensory inputs into overall brain function seems affected by
SGA status. As an example, Jiang et al (1991), used brainstem auditory-evoked
response as a measure of functional integrity of the brainstem auditory pathway, which was
reduced in SGA infants. Fried and Watkinson (1988, 1990), found that maternal prenatal
smoking, a major risk factor for SGA, was associated with altered auditory response at 12
to 36 months. Saxton (1978) also found that smokers' infants had diminished auditory
senses compared to other infants. Martikainen (1992) showed that asymmetric SGA infants
had lower visuo-auditory perception scores than other infants. Similarly, Todorovich, et
al (1987), found that the response to auditory stimulation in term SGA infants (n =
22) was significantly retarded compared to auditory response in term AGA infants (n = 50).
In summary, it appears that, overall, being born SGA is associated with an increase in various measures of MND (Aylward et al, 1989; Breart and Poisson-Salomon, 1988; Allen, 1984; Teberg et al, 1988). Major motor and cognitive disability is rare in SGA infants, but is probably significantly increased when evaluated in large sample sizes. Boys seem to be more affected by being SGA than girls, and children born into the low social classes seem to be more affected than other children. If the SGA develops early so that it affects head growth before 26 weeks, there seems to be more of an impact on neurologic function than SGA which develops later (Harvey et al, 1982). Certainly SGA which is accompanied by asphyxia is more commonly associated with neurodevelopment abnormalities than SGA which is not. There does not seem to be a major relationship between SGA status and vision or hearing deficits, although there are some studies to suggest that SGA infants have more difficulty in achieving 'normal' responses to visual or auditory stimuli than do non-SGA infants.
This review suggests that SGA is a heterogeneous condition, at times but not usually associated with various types of neurodevelopment dysfunction. There are no studies we identified showing that improvements in these neurologic outcomes can be achieved by any particular course of action. Logically, however, preventing asphyxia in SGA infants should reduce the prevalence of major and minor handicaps, especially cerebral palsy and mental retardation, seen in some of these infants, particularly those who are asymmetric and associated with maternal hypertension. It is less clear if interventions directed at the more symmetric or uniformly under-grown infants will be able to improve outcome in the relatively small percentage of these children who have major developmental problems associated with being SGA.
Acknowledgement - This
project is funded by the Agency for Health Care Policy Research (AHCPR) Contract
#290-92-0055.