| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Presidential Address. Prediction of Function From Infancy to Early Childhood: Implications for Pediatric Psychology
Departments of Pediatrics and Psychiatry, Southern Illinois University School of Medicine Division 54 Presidential Address, presented at the 111th annual convention of the American Psychological Association, Toronto, Canada, August 7–10, 2003.
All correspondence should be sent to Glen P. Aylward, SIU School of Medicine, Pediatrics, P.O. Box 19658, Springfield, Illinois 62794-9658. E-mail: Gaylward{at}siumed.edu
 |
Abstract |
|---|
 Top Abstract MethodResultsDiscussionReferences |
|---|
Objective To determine whether item groupings derived from the Bayley Infant Neurodevelopmental Screener (BINS) are stable and predictive of 36-month cognitive and motor outcome. Methods BINS was administered at 6, 12, and 24 months, and the Bayley-II or McCarthy scales at 36 months. The BINS was factor analyzed, and factors, biomedical and environmental variables, were related to 36-month outcomes. Results Three factors were identified at each age, accounting for 52% to 64% of the variance. Continuity in factors over infancy and predictive utility of similar functions at 36 months were found. Optimal factor scores (
75th percentile) increased the likelihood of later normal cognitive or motor outcome (ORs 2.14–7.94). Conclusions Stability and continuity over time exist in specific subdomains of function on a neurodevelopmental screening test. Key words: high-risk infants; prediction; Bayley Infant Neurodevelopmental Screener; outcomes.
There is increased interest in prediction of outcome from infancy to early childhood and later ages, particularly in infants at biological risk (Aylward, 2002a
, 2002b). As a result, pediatric psychologists are frequently involved in the assessment and care of these children, whose risks include prematurity and low birth weight, as well as exposure to central nervous system (CNS) insults such as asphyxia, prenatal drug use, or maternal infections during pregnancy. Moreover, improvements in medical technology have increased survival rates of infants as young as 23 weeks’ gestational age, as well as in larger babies who have sustained CNS insults. These increased survival rates have raised concerns that there may be a higher incidence of later neurodevelopmental disability in these children (Bregman, 1998).
Prediction rates from early infant assessments to later cognitive function have generally been low, with some estimates indicating that 1% to 6% of the variance in later IQ is explained by early infant measures (Colombo, 1993; McCall, 1979, 1983). These rates have improved somewhat with inclusion of laboratory measures such as infant recognition memory (Colombo, 1993; Colombo & Mitchell, 1990; Lewis & Brooks-Gunn, 1981). Therefore, the pragmatic issue exists as to whether "lagged prediction" (Colombo, 1993) of function from infancy to later ages can be enhanced. Improved prediction would enable provision of feedback to parents, help identify those children who are in need of early intervention, specify the areas of function that require intervention, allow evaluation of short- and long-term outcomes of innovative medical procedures, and facilitate development of risk phenotypes (Aylward, 1997a).
There are several ways to enhance prediction: (1) Implement changes in interpretation of existing tests on the "front end," where critical items or groups of items are identified and used to predict outcomes; (2) employ changes at the "tail end" by reducing outcome from a global score to specific subdomains, thereby better defining outcomes of interest; or (3) apply a combination of the two methods, which conceptually would seem to be the best approach. Inherent in the concept of prediction is the underlying assumption that continuity must exist in specific developmental functions over time. This continuity would require stability of underlying processes as well as stability of developmental status that would extend beyond infancy (Bornstein & Sigman, 1986).
Children with biological risks such as prematurity have specific areas of deficit (Aylward, 2002a) as well as strengths, and the same is true with other potential CNS insults. Therefore, use of a single, composite measure such as a developmental index (e.g., Bayley, 1993) may not be appropriate as either a baseline or outcome measure because it would not parse out discrete developmental functions. The assumption that all areas of development progress simultaneously at an equal rate, or, conversely, would be equally delayed, is not supported in either normal or at-risk populations (Harris & Langkamp, 1994). While more severe handicaps would have a suppressor effect on all domains of development and thereby validate a global measure of delay, children with later high prevalence/low severity dysfunctions (learning disabilities, attention deficit hyperactivity disorders, borderline to low average IQ, behavioral problems, or neuropsychological deficits [visual motor integration, executive function]) would display much more variability in their initial evaluation results as well as in later outcome.
This is a particular concern, given that the rate of high prevalence/low severity dysfunction is much higher than the rate of severe handicaps, the latter being in the 15% to 20% range, depending on birth weight (Aylward, 2002a). It is imperative that cognitive, neuromotor, and functional deficits be identified early, particularly in light of the possibility that prematurity and similar biological risks act through the association with health, cognitive, and neuromotor function to explain behavioral and other problems at school age (Nadeau, Boivin, Tessier, Lefebvre, & Robaey, 2001; Taylor, Burant, Holding, Klein, & Hack, 2002). Moreover, the link in developmental pathways between biological risks such as extreme prematurity and later problems is indirect, and different types of early disabilities do not lead to the same problematic behavioral or cognitive outcomes at later ages (Nadeau et al., 2001).
This situation is compounded by the fact that children are more likely to receive screening tests than full assessments during infancy, due to time and economic issues as well as sheer volume (Glascoe, 1997). Screening tests typically have a reduced number of items, and it is questionable whether brevity of administration and the restricted sampling range of developmental functions would allow for a more detailed, front-end delineation of critical items or groups of items.
The Bayley Infant Neurodevelopmental Screener (BINS) (Aylward, 1995) is a screening test designed to identify infants between the ages of 3 and 24 months who are developmentally delayed or have neurodevelopmental impairments. It consists of six item sets: 3–4 months, 5–6 months, 7–10 months, 11–15 months, 16–20 months, and 21–24 months, each with 11–13 items that are scored in a binary optimality format (optimal/nonoptimal). The BINS takes approximately 10 minutes to administer. The instrument allows screening of posture, tone, quality of movement, developmental status, and basic neurological intactness by assessing four conceptual areas that are based on an early developmental, neuropsychological framework. The BINS risk scores have acceptable concurrent validity when compared with the Bayley Scales of Infant Development–II (BSID-II): a 68% to 96% rate of agreement for high risk status, depending on item set for the mental developmental index, and 78% to 86% agreement for the psychomotor developmental index (Aylward, 1995). Predictive validity has also been established (Aylward & Verhulst, 2000; Creighton, Dewey, & Suave, 1997; Leonard, Piecuch, & Cooper, 2001; Macias et al., 1998). However, the issue of identification of specific developmental subdomains on the test and prediction of specific later outcomes has not been investigated adequately.
Therefore, the purpose of the current investigation is to determine whether identification during infancy of conceptual groupings of items on the BINS, in conjunction with medical/biological and environmental marker variables, will enhance prediction of 36-month cognitive and motor outcome. Furthermore, inherent in the pursuit of prediction is the assumption that specific functional subdomains, identified in infancy, will be associated with related functions in early childhood, thereby supporting the concept of continuity.
|
Method |
|---|
|
TopAbstractMethodResultsDiscussionReferences |
|---|
Participants and Procedures
Participants include infants who graduated from a tertiary care neonatal nursery and are followed longitudinally in a university-based, multidisciplinary developmental continuity clinic. Eligibility criteria for enrollment included any of the following: asphyxia (5-minute Apgar score
5), neonatal seizures, birth weight < 1250 g, mechanical ventilation > 28 days, intraventricular hemorrhage, or persistent fetal circulation. The subject sample may be considered representative of a typical high-risk nursery follow-up program and contains primarily preterm infants as well as term babies who experienced perinatal complications. Infants were enrolled in the follow-up program after the primary caregiver gave consent, according to requirements of the institutional review board. Clinic dropout is approximately 10% per year. Infants are typically evaluated at 6, 12, 24, and 36 months of age. Conceptional age (age corrected for prematurity) is used at the first three evaluations, while chronological age (not corrected) is employed at the 3-year visit. The total sample consisted of 1,332 infants, 54% male and 46% female. Approximately 65% were white, 30% African American, and 5% of other ethnic groups (Hispanic, Asian). The mean birth weight was 1731 g (±997 g), with 25% of the infants being < 1000 g, and 50% < 1292 g. The mean gestational age was 31.79 weeks (± 4.72), with 30% of the babies being < 28 weeks, and 56% being less than 31 weeks. The mean days of hospitalization after birth was 54 days, the mean 5-minute Apgar score was 7.4, and 79% of the infants had respiratory distress syndrome.
Of these infants, 569 were evaluated at 6 months, 56 at 9 months, 458 at 12 months, 16 at 18 months, 358 at 24 months, and 489 at 36 months (for a total of 1,946 examinations). With respect to longitudinal follow-up, 161 were seen at 6 and 36 months, 186 at 12 and 36 months, and 201 at 24 and 36 months. Data collection is ongoing and as a result, the sample should be considered mixed, that is, it is longitudinal and cross-sectional due to dropout, missed appointments and resumption of participation, and infants whose families kept all four appointments in the follow-up schedule.
At the 6-, 12-, and 24-month visits, the BINS (Aylward, 1995), a speech/language evaluation, occupational/physical therapy evaluation, and a pediatric examination were administered. At 36 months, either the McCarthy Scales of Children’s Abilities (MSCA) (McCarthy, 1972) or the BSID-II (Bayley, 1993) were given. Environmental socioeconomic status (SES) was assessed with the SES-Composite Index (SES-Comp) (Aylward, Dunteman, Hatcher, Gustafson, & Widmayer, 1985) or the SES-Composite Index–Revised (SES-Comp-R) (Aylward, 1997b), which was completed by clinic research staff at the first clinic visit. The Neonatal Medical Index (NMI) (Korner et al., 1993) was scored from the infant’s medical record by a member of the nursing staff to document biological risk.
Measures
Bayley Infant Neurodevelopmental Screener
The BINS was designed to identify infants at risk for developmental delays or neurodevelopmental problems by assessing four a priori conceptual areas of ability: (1) basic neurological functions/intactness (posture, muscle tone, movement, asymmetries, abnormal indicators), (2) expressive functions (gross motor, fine motor, oral motor/verbal), (3) receptive functions (visual, auditory, verbal), and (4) cognitive processes (object permanence, goal directedness, imitative abilities, problem solving). The BINS consists of six item sets, each containing 11 to 13 items. Three item sets were used in this study as per the clinic protocol: 5 to 6 months (13 items), 11 to 15 months (11 items), and 21 to 24 months (13 items). Each item was scored optimal or nonoptimal, based on a priori decision rules. The numbers of optimal responses for a given item set were added to provide a summary score. For each item set, three summary cut scores were established via comparison of clinical and normative standardization samples, to identify a given infant’s level of risk for developmental problems: low risk, moderate risk, and high risk (Aylward, 1995).
In addition, BINS total scores were analyzed as a dichotomous or binary variable (Aylward, 1998; Aylward & Verhulst, 2000) by subdividing the moderate risk group. This division was based on the cutoff score in the BINS manual that offered the best measures of sensitivity and specificity, indicated by a dashed line on the BINS form. Moderate risk scores falling between the cutoff point and the BINS high risk category were combined into a moderate/high risk group (HIGHRISK). Moderate risk scores falling between the cutoff point and the BINS low risk category were combined with the low risk category to form a LOWRISK group.
Socioeconomic Status–Composite Index
The SES-Comp (Aylward et al., 1985) includes six marker variables: maternal education, paternal education, family occupation, availability of car or phone, integration (involvement) of the male into the family, and nonparticipation in public assistance. Scores, ranging from 0 to 21 are summed and categorized into three levels—upper, middle two, and lower quartiles (Aylward, Verhulst, & Bell, 1994). The SES-Comp-R was developed by Aylward and Berger as a revision to the original instrument (Aylward, 1997b). Changes included: occupation expanded to eight levels, a labor force multiplier based on hours per week of employment, eight levels of education (in .5 increments), and the requirement of actually possessing a car, and possessing a phone (versus simple access to either). Items were added to document supplemental income (disability, child support), housing tenure (owning, renting, staying with someone), and health insurance (yes/no). The total score could range from 0.5 to 19 points.
Neonatal Medical Index
The NMI (Korner et al., 1993) was designed to summarize the prior medical course of high-risk infants at the time of hospital discharge. NMI classifications range from I to V, with a score of I describing an infant who is free of significant medical problems, and V characterizing an infant with the most serious complications. Birth weights of > 1000 g versus < 1000 g, and duration of ventilation are two major determinants of the NMI score. Other determining factors include seizures, intraventricular/periventricular hemorrhage, patent ductus arteriosis, hyperbilirubinemia, meningitis, apnea/bradycardia, and periventricular leukomalacia.
Three-Year Outcome Measures
The MSCA (McCarthy, 1972) consist of five subscales: verbal, perceptual/performance, quantitative, memory, and motor (M = 50, SD = 10). The verbal, perceptual/performance, and quantitative scales are combined to produce a general cognitive index (GCI) (M = 100, SD = 16). MSCA scores were analyzed as continuous as well as dichotomous variables. In the latter analyses, a GCI > 1 SD below average (< 84) was considered nonoptimal. Similarly, on the motor subscale, scores > 1 SD below average (< 40) were nonoptimal. The BSID-II, applicable for ages 1–42 months (Bayley, 1993), yields a mental Developmental Index (MDI) and a Psychomotor Developmental Index (PDI) (M = 100, SD = 15). The MDI and PDI were analyzed as both continuous and binary variables (a score < 85 on either index was considered nonoptimal). The MSCA GCI and the BSID-II MDI were used to determine cognitive outcome, while the MSCA motor scale and the BSID-II PDI were measures of 36-month motor outcome. Use of the two outcome measures was mandated by the change in assessment instruments during data collection. Because the BSID-II PDI and the MSCA motor score were recorded as standard scores and T-scores, respectively, they were converted to a common metric (based on percentiles) when these data were analyzed as continuous variables.
Statistics
Chi-square analyses were used to generate sensitivity (copositivity) and specificity (conegativity) values and odds ratios for the 6-, 12-, and 24-month BINS and 36-month cognitive and motor outcome reference standard. Pairwise principal components factor analysis with varimax rotation was employed to explore the factor structure of the BINS at each age. This technique was selected because it uses 1’s in diagonals, versus squared multiple correlations, and it contains the assumption that relationships between variables are equal. Varimax rotation optimizes the independence of factors. Moreover, this technique has been used previously with infant tests (McCall, Hogarty, & Hurlburt, 1972). Factors with eigenvalues > 1 were retained. Relationships among factors, factors and demographic variables, and factors and 36-month outcome, were explored with Pearson correlation coefficients. Stepwise linear regression was employed to evaluate continuity between the BINS factors and 36-month outcome using continuous variables, while forward, stepwise logistic regression (likelihood ratio) was employed to generate more clinically meaningful likelihood ratios regarding factor scores and later probability of optimal cognitive and motor outcome. In these latter analyses, BINS factor scores were divided into the upper quartile (optimal) versus the remaining 3 quartiles (nonoptimal).
|
Results |
|---|
|
TopAbstractMethodResultsDiscussionReferences |
|---|
Sensitivity and Specificity Values
The accuracy in prediction of the BINS binary HIGHRISK and LOWRISK status and 36-month cognitive and motor outcome is found in Table I. In actuality, sensitivity should be considered as copositivity, because the so-called gold standard (MSCA or BSID-II) is actually a reference standard. Similarly, specificity should be conceptualized as conegativity because of the same issue.
|
Sensitivity (copositivity) with respect to cognitive outcome is moderate, improving to acceptable levels from 12 months onward. Specificity (conegativity) reaches an acceptable level by 24 months. However, if a child had LOWRISK score at 6 months, she was 2.7 times more likely to have low risk cognitive outcome at 36 months than if a HIGHRISK BINS-6 score were obtained (95% CI = 1.38–5.33). The odds ratio was 2.2 for 12-month LOWRISK BINS scores, increasing to 7.7 by 24 months.
Similarly, sensitivity values with respect to 6-month motor function were in the acceptable range (i.e., > .70) for the 6- and 12-month BINS scores, with moderate specificity values. A definite trade-off occurred at 24 months, with a decrease in sensitivity and a corresponding increase in specificity. Once again, however, if the child received a LOWRISK BINS score at 6, 12, or 24 months, he was 3.5 to 5.9 times more likely to have good motor outcome at age 3 than if a HIGHRISK score had been obtained.
Factor Analysis
While the four original BINS conceptual areas could have been used, these were developed in an a priori fashion. As a result, the clusters were not orthogonal; it is expected that for any one item, the abilities represented in any one cluster are involved along with abilities from another cluster (Aylward, 1995). Therefore, a factor analytic approach was selected. In order to determine whether changes at the front end would enhance prediction, the BINS was subject to principal components factor analysis with a varimax rotation. The varimax rotation was employed to produce orthogonal factors (see Table II). At 6 months, using 569 infants, a three-factor solution was extracted. The factors were (1) cognitive/fine motor, (2) neuromotor, and (3) gross motor. Eigenvalues were 4.78, 1.20, and 1.18, respectively. This solution accounted for 52% of the common variance. Only two items—looks for spoon and sits with support for 10 seconds—loaded on more than one factor, indicating minimal communalities.
|
Using a sample of 458 infants at 12 months, a three-factor solution again was derived: (1) motor, (2) verbal/language, and (3) cognitive processes (eigenvalues of 3.53, 1.26, and 1.00, respectively). This solution accounted for 53% of the common variance. Only one item, walks without assistance, loaded on more than one factor.
Employing a sample of 358 babies at 24 months, a three-factor solution was obtained: (1) verbal expressive, (2) motor, and (3) cognitive/verbal receptive (eigenvalues of 5.53, 1.71, and 1.10). These three factors accounted for 64% of the variance. Of note is the fact that two verbal items (identifies four pictures, points to body parts of doll) and one motor item (kicks ball) loaded on more than one factor.
Correlations
Continuity was first evaluated by exploring the associations among the factor scores over time. The items in each factor were summed (an optimal response on an item = 1; a nonoptimal response = 0) to produce factor scores, these ranging from 0 to a maximum of 7, depending on the specific factor. Correlations are depicted in Table III. Continuity in cognitive function was apparent across the three ages, as was continuity in gross motor and verbal skills. The 6-month neuromotor factor was also moderately correlated with later motor function.
|
Correlations among the BINS factors and demographic variables were also examined. Only the NMI, days of hospitalization, SES-Comp-R, highest level of occupation in the household, and highest level of education were modestly (but significantly) related to the BINS factors (see Table IV). Surprisingly, 5-minute Apgar scores, birth weight, and gestational age were not significantly related to the BINS factors. This may be due to the study population (most were premature and had low birth weight) or it could be an artifact of correction for prematurity which could essentially negate gestational age effects.
|
Factor scores were then correlated with 36-month outcome (Table V). While all factors were significantly correlated with cognitive outcome, the correlations were greater with the 6-, 12-, and 24-month BINS cognitive and verbal factor scores, again suggesting continuity. Moreover, with respect to motor outcome, the reverse was true; BINS motor factor scores were more strongly correlated with later motor outcome than were cognitive or verbal factor scores (although the latter correlations were still significant).
|
This continuity pattern was supported further by stepwise linear regression. When predicting 36-month cognitive function, only the cognitive/fine motor factor at 6 months was retained in the model (R2 = .19, p = .001). At 12 months, the verbal/language and cognitive-processes factors were included (R2 = .21, p = .005), while at 24 months, all three factors (verbal expressive, motor, and cognitive/verbal receptive) were retained in the model (R2 = .57, p = .002).
With regard to 36-month motor outcome, only the gross motor factor was retained at 6 months (R2 = .12, p = .001), the motor factor was predictive at 12 months (R2 = .28, p = .0001), while the motor and verbal expressive factors were included in the model at 24 months (R2 = .38, p = .002).
Logistic Regression
To enhance the clinical utility of these data, factor scores were converted to a binary format, with scores below the 75th percentile being considered nonoptimal, and those at or above the 75th percentile being optimal; 36-month outcome was also considered normal/not normal, based on criteria listed previously. Forward, stepwise logistic regression (likelihood ratio) was employed in these analyses. With respect to cognitive outcome, at 6 months the BINS cognitive/fine motor factor was included with a resultant OR of 3.04 (95% CI = 1.33–6.94, p = .008). Stated differently, if the cognitive/fine motor factor score was 7 ( 75th percentile) at 6 months, the infant was three times more likely to have a normal 3-year cognitive outcome than if the factor score was below the 75th percentile. At 12 months, the verbal/language (OR = 2.51, 95% CI = 1.33–4.76; p = .005) and cognitive processes factors (OR = 2.43, 95% CI = 1.30–4.55; p = .006) were included in the model. The top quartile scores for these factors were 4 and 2, respectively. At 24 months, the verbal expressive (OR = 7.69, CI = 4.67-18.87; p = .001) and motor factors (OR = 2.58, CI = 1.27-5.35; p = .009) were included. Top quartile scores were 4 and 5, respectively.
With respect to 36-month motor outcome, the 6-month BINS gross motor factor was predictive (OR = 7.94, 95% CI = 1.71–37.04; p = .008), with a score of 1 being in the top quartile. At 12 months, the motor factors (OR = 4.00, 95% CI = 2.02–7.87; p = .001) and cognitive processes factors (OR = 2.14, CI = 1.09–4.20; p = .003) were included in the model (scores of 4 and 3, respectively). Finally, at 24 months, the motor factor, with a top quartile score of 5, was retained in the model (OR = 5.78, 95% CI = 2.94–11.36; p = .001).
|
Discussion |
|---|
|
TopAbstractMethodResultsDiscussionReferences |
|---|
Traditionally, prediction or continuity from infant tests to later outcome has been poor (Colombo, 1993
; McCall, 1979, 1983). Results of the present study suggest that stability and continuity over time do exist in a biologically at-risk sample of infants with respect to specific subdomains of function. Moreover, it appears that prediction of later function can be enhanced by better delineation of item groupings at the front end.
While this study was restricted to the BINS, it is quite likely that this same approach can be applied to other infant tests as well. In fact, a similar recommendation was made 30 years ago, when McCall et al. (1972) reported that although overall Gesell infant test scores (Gesell, Thompson, & Amatruda, 1934) obtained during the first years of life were not related to later intelligence test scores, certain clusters of items were indeed related to later functioning as assessed by the Stanford-Binet. This is especially important given the more recent emphasis on more subtle, high-prevalence/low-severity dysfunctions found in biologically at-risk infants (Aylward, 2002a; Taylor et al., 2002; Taylor, Klein, & Hack, 2000). Hence, parsing out functional subdomains throughout infancy, particularly verbal, cognitive, and motor abilities, may yield more precise prediction than tracking global measures. Similarly, later outcomes should be better defined. At minimum, cognitive and motor abilities should be separated.
Developmental functions may not be differentiated as well at 6 months or earlier, particularly with respect to the cognitive/fine motor subdomain. This is possibly due to the gradual divergence of cognitive, motor, and neurological functions during the period of infancy, coupled with strongly canalized behaviors of a sensorimotor nature. Later, complex intellectual function may unfold from an integration of basic sensorimotor skills that emerge and dominate behavioral patterns during infancy. These functions are thought to be built from the bottom up and provide the foundation for later cognitive acquisitions (Colombo, 1993). In the present study, this is evident in the combined cognitive/fine motor factor extracted at 6 months. Items range from simple sensorimotor tasks such as hand-to-hand transfer, picking up a pellet, and vocalizing, to goal-directed behavior and higher-order processing necessary for imitation and precursory object permanency. Medical issues (i.e., recovery from medical problems), recovery of neurodevelopmental function, and the fact that some abnormal findings are indicative of lags versus deficits all serve to compound the issue. The current data suggest that functional subdomains separate out more clearly by 12 months.
The distinction between verbal expressive and cognitive/verbal receptive factors at 24 months is noteworthy. Although the two verbal receptive items also have loadings on the verbal expressive factor, these were modest. This may be reflective of the language burst phenomenon that occurs around 2 years: Some children simply will have more accelerated language production than others, although those with slower production still have the basic language infrastructure in place as evident by their verbal receptive capabilities. These children will overcome this temporary lag in expression and display "catch-up." Such acceleration in verbal production would not be possible in the absence of an underlying language-based infrastructure.
Medical/biological variables and SES markers were not as predictive of later outcome as were factor scores. In fact, when medical/biological and SES variables were also entered into stepwise regression models, they did not account for any additional variance above that provided by BINS factor scores. In separate analyses (without inclusion of BINS factor scores), the OR for days of hospitalization in prediction of cognitive function was 1.03 (ns) and the OR for SES was 1.17 (ns). This lack of significant, additional explanatory power could be a function of the BINS test items, age at time of evaluation (environment is influential from 2 years onward), or the fact that the nature of the population was positively skewed with respect to biological risk. Nonetheless, days of hospitalization is a modest marker variable for medical/biological risk in this population.
The weak influence of the SES-Comp measures in the prediction of outcome may be due to several reasons. First, these measures were administered during the first clinic visit, and it is possible that some aspects could have changed over the course of the follow-up period. Second, the items measured are of a more "distal" or status type, and their effects may not be as strong at younger ages as are other environmental influences such as the mother–child interaction (Aylward, 1996).
It should also be noted that many of the children in this study were afforded early intervention (EI) services. The clinic procedure holds that any child who is identified as at risk by the psychologist, speech/language specialist, occupational/physical therapist, or physician is referred for services. Because of the large geographical area, EI services are not uniform in frequency, type (developmental, speech, occupational/physical therapy), or utilization. In addition, some children were referred earlier by their primary care physician or hospital social worker. Although specific numbers of EI participants cannot be determined, the majority of babies did receive some type of service. While this undoubtedly adds more variance to the study, EI services are essential. In actuality, positive EI effects should have improved the status of children in the nonoptimal group, thereby reducing prediction. Nonetheless, despite this potential influence, the degree of continuity was still impressive. Also of note is the fact that because of the optimality approach inherent in the BINS (Aylward, 1995), emphasis is placed on prediction of optimal versus nonoptimal (not normal) scores and outcomes.
A final caveat is that there is a need for a cross-validation sample to determine whether the same factor structures and continuity would exist in other cohorts. This should be undertaken in subsequent studies.
These findings underscore the need to serially evaluate children at biological risk over the course of infancy so as to enhance lagged prediction. Early measures of cognitive operations or processes do seem to correlate with so-called product measures of intelligence (Underwood, 1975). However, no general factor of intelligence or underlying "g" can be identified at this point (Bornstein & Sigman, 1986); rather, more stable, specific functions or an "s" appears to exist. While cost and time factors are considerations, it does appear possible to use screening tests for monitoring purposes and to predict later outcome, provided that subdomain scores, in addition to overall composite scores, are used. Infants at risk for cognitive and motor deficits in later childhood can thus be identified by delineating and better understanding processes or subdomains that contribute to these deficits. Identification and understanding of these functional processes would also enable design and implementation of early interventions and provision of feedback to parents, both of which fall under the purview of pediatric psychologists.
|
Acknowledgements |
|---|
Special thanks to Steven J. Verhulst, PhD, for his helpful statistical review.
Received October 24, 2003; revision received December 15, 2003; accepted January 5, 2004
|
References |
|---|
|
TopAbstractMethodResultsDiscussionReferences |
|---|
Aylward, G. P. (1995). Bayley Infant Neurodevelopmental Screener. San Antonio, TX: The Psychological Corporation.
Aylward, G. P. (1996). Environmental risk, intervention, and developmental outcome. Ambulatory Child Health, 2, 161–170.
Aylward, G. P. (1997a). Conceptual issues in developmental screening and assessment. Journal of Developmental and Behavioral Pediatrics, 18, 340–349.[CrossRef][Web of Science][Medline]
Aylward, G. P. (1997b). Environmental influences: Consideration for early assessment and intervention. In S. M. C. Dollinger & L. F. DiLalla (Eds.), Intervention and assessment across the life span (pp. 9–34). New York: Erlbaum.
Aylward, G. P. (1998). Alternative risk grouping method for the Bayley Infant Neurodevelopmental Screener (BINS) in the prediction of later dysfunction. Journal of Developmental and Behavioral Pediatrics, 19, 386–387.
Aylward, G. P. (2002a). Cognitive and neuropsychological outcomes: More than IQ scores. Mental Retardation and Developmental Disabilities Research Reviews, 8, 234–240.[CrossRef][Web of Science][Medline]
Aylward, G. P. (2002b). Methodological issues in outcome studies of at-risk infants. Journal of Pediatric Psychology, 27, 37–45.
Aylward, G. P., Dunteman, G., Hatcher, R. P., Gustafson, N., & Widmayer, S. (1985). The SES-Composite Index: A tool for developmental outcome studies. Psychological Documents, 15, ms. no. 2628.
Aylward, G. P., & Verhulst, S. J. (2000). Predictive utility of the Bayley Infant Neurodevelopmental Screener (BINS) risk status classifications: Clinical interpretation and application. Developmental Medicine and Child Neurology, 42, 25–31.[CrossRef][Web of Science][Medline]
Aylward, G. P., Verhulst, S. J., & Bell, S. (1994). Enhanced prediction of later normal outcome using infant neuropsychological assessment. Developmental Neuropsychology, 10, 377–393.
Bayley, N. (1993). The Bayley Scales of Infant Development–II. San Antonio, TX: The Psychological Corporation.
Bornstein, M. H., & Sigman, M. S. (1986). Continuity in mental development from infancy. Child Development, 57, 251–274.[CrossRef][Web of Science][Medline]
Bregman, J. (1998). Developmental outcome in very low birthweight infants. Current status and future trends. Pediatric Clinics of North America, 45, 673–690.[CrossRef][Web of Science][Medline]
Colombo, J. (1993). Infant cognition: Predicting later intellectual functioning. Newbury Park, CA: Sage Publications.
Colombo, J., & Mitchell, D. W. (1990). Individual differences in early visual attention: Fixation time and information processing. In J. Colombo & J. W. Fagen (Eds.), Individual differences in infancy (pp. 193–228). Hillsdale, NJ: Lawrence Erlbaum.
Creighton, D. E., Dewey, D., & Sauve, R. S. (1997). A comparison of three infant developmental screening tests. Journal of Developmental and Behavioral Pediatrics, 18, 361–362.
Gesell, A. L., Thompson, H., & Amatruda, C. S. (1934). An atlas of infant behavior. New Haven, CT: Yale University Press.
Glascoe, F. P. (1997). Parents’ concerns about children’s development: Prescreening technique or screening test? Pediatrics, 99, 522–528.
Harris, S. R., & Langkamp, D. L. (1994). Predictive value of the Bayley mental scale in the early detection of cognitive delays in high-risk infants. Journal of Perinatology, 14, 275–279.
Korner, A. F., Stevenson, D. K., Kraemer, H. C., Spiker, D., Scott, D., Constantinou, J., et al. (1993). Prediction of the development of low birth weight preterm infants by a new neonatal medical index. Journal of Developmental and Behavioral Pediatrics, 14, 106–111.[Web of Science][Medline]
Leonard, C. H., Piecuch, R. E., & Cooper, B. A. (2001). Use of the Bayley Infant Neurodevelopmental Screener with low birth weight infants. Journal of Pediatric Psychology, 26, 33–40.
Lewis, M., & Brooks-Gunn, J. (1981). Visual attention at three months as a predictor of cognitive functioning at two years of age. Intelligence, 5, 131–140.[CrossRef]
Macias, M. M., Saylor, C. F., Greer, M. K., Charles, J. M., Bell, N., & Katikaneni, L. D. (1998). Infant screening: The usefulness of the Bayley Infant Neurodevelopmental Screener and the Clinical Adaptive Test/Clinical Linguistic Auditory Milestone Scale. Journal of Developmental and Behavioral Pediatrics, 19, 155–161.[Web of Science][Medline]
McCall, R. B. (1979). The development of intellectual functioning in infancy and the prediction of later IQ. In J. D. Osofsky (Ed.), Handbook of infant development (pp. 707–741). New York: Wiley.
McCall, R. B. (1983). A conceptual approach to early mental development. In M. Lewis (Ed.), Origins of intelligence (2nd ed., pp. 67–106). New York: Plenum.
McCall, R. B., Hogarty, P. S., & Hurlburt, N. (1972). Transitions in infant sensorimotor development and the prediction of childhood IQ. American Psychologist, 27, 728–748.[CrossRef][Medline]
McCarthy, D. (1972). McCarthy Scales of Children’s Abilities. New York: The Psychological Corporation.
Nadeau., L., Boivin, M., Tessier, R., Lefebvre, F., & Robaey, P. (2001). Mediators of behavioral problems in 7-year-old children born after 24 to 28 weeks of gestation. Journal of Developmental and Behavioral Pediatrics, 22, 1–10.[Web of Science][Medline]
Taylor, H. G., Burant, C. J., Holding, P. A., Klein, N., & Hack, M. (2002). Sources of variability in sequelae of very low birth weight. Child Neuropsychology, 8, 163–178.
Taylor, H. G., Klein, N., & Hack, M. (2000). School-age consequences of birth weight less than 750 g: A review and update. Developmental Neuropsychology, 17, 289–321.[CrossRef][Web of Science][Medline]
Underwood, B. J. (1975). Individual differences as a crucible for theory construction. American Psychologist, 30, 128–134.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Hack, H. G. Taylor, D. Drotar, M. Schluchter, L. Cartar, D. Wilson-Costello, N. Klein, H. Friedman, N. Mercuri-Minich, and M. Morrow Poor Predictive Validity of the Bayley Scales of Infant Development for Cognitive Function of Extremely Low Birth Weight Children at School Age Pediatrics, August 1, 2005; 116(2): 333 - 341. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||