Evaluating Features and Application of Neurodevelopmental Tests in Epidemiological Studies
5
however, to the fact that localized lesions with predictable behavioral anomalies are more
common in adults. When a lesion or disorder appears in the developed adult brain, structure-
function relationships are well established, and it is often clear where the problem might be.
Neurodevelopment is more complicated. During child development, the brain is more plastic
with regard to how specific structures mediate behaviors, is capable of compensating when a
structure or brain area is diseased (or even absent), can be influenced by acute or chronic
exposures during susceptible periods, and develops and expresses behavior in a dynamic fashion
given the circumstances occurring at any stage of development. Because of all these
considerations, structure-function relationships in early brain development are more diffuse and
less “focal” than in adults, and insults to the developing brain—both toxicants and other
neurological conditions—may have different effects than would insults on the mature brain.
Given these circumstances, neurodevelopmental assessments have used a combination of
existing tests for children (e.g., IQ, developmental, academic); adaptations of adult tests for
children; and specialized tests that have been developed, standardized, normed, and validated in
clinical populations (e.g., Developmental Neuropsychological Assessment, Behavior Assessment
System for Children, Child Behavior Checklist, tests that assess clinical conditions in children)
(White 2004).
Almost all psychometric tests provide raw scores that can be converted into standard or scaled
scores (mean = 10, SD = 3), T-scores (mean = 50, SD = 10), standard scores (mean = 100,
SD = 15), and corresponding percentiles using normative or reference data. This practice allows
participants’ scores to be compared with one another after removing the effects of age, sex, or
other characteristics on each participant’s raw score. For instance, even within narrow age
intervals, older children have higher average raw scores on tests of mental and psychomotor
development than younger children. By using standard scores, the traits of participants who are
different ages (or other characteristics) can be compared. Moreover, an individual’s performance
over time can be examined in relation to reference or normative data. Such comparisons between
groups or within individuals over time are often made in terms of number of points or in units of
SDs away from the average score.
Note that normative data are not required for comparison of scores between individuals when
appropriate measures are taken to validly analyze raw scores; however, normative data are
necessary and routinely used when making decisions about an individual’s health care, clinical
diagnosis, vocational services, education, legal status, etc. In addition, when appropriate
normative data are available, they can be used as the outcome in statistical models.
There is a longstanding tension regarding the interpretation of results from population-based
studies that examine neurodevelopmental toxicity. Generally, this friction arises from the lack of
consensus regarding the “clinical significance” of effects observed in epidemiological studies. In
clinical situations, it is appropriate to refer to “abnormalities” or “deficits” in a functional area
when an examinee performs poorly on a test (usually 1–2 SDs away from the average score). In
epidemiological studies, effect sizes often fall short of being clinically significant in that they
occur within the “normal range” of test performance and are typically less than half of an SD
away from the average score. However, subtle shifts in a continuous trait can have profound
impacts on the tails of a distribution in a population (Needleman 1990; Rose 1985; 2001; Weiss
2000). For instance, a 5-point decrease in a population’s IQ would nearly double the number of
people classified as intellectually disabled (Braun 2016). Finally, it is important to refer to
subclinical or preclinical findings of lower scores in these situations as “decrements” in