Infant motor development is an important indicator of neurological integrity1 and can be used to identify infants at risk for neurodevelopmental delay, enabling early referral and intervention.2 Infant neuromotor development is thought to be associated with maternal socioeconomic status (SES), but the nature of the association has varied across different studies.
The relationship of maternal SES has been studied primarily in high-risk populations or in studies also exploring the relationship of race or ethnicity to motor development. Few studies have examined the relationship between maternal education and infant development in infants born at term and considered healthy.
One could assume that if maternal education relates to infant motor development in infants who are healthy, it would be critical that normative data collected on standardized motor tests should include a sample of infants of mothers with educational levels that are representative of the normative population. Because standardized tests are used frequently to discriminate between infants who are developing typically and those who show delays, it is imperative that these tests control for possible confounding factors, such as maternal education. The rationale for our study, therefore, was to explore the relationship between maternal education and infant neuromotor development in a normative sample of Canadian infants who were born at term and considered healthy.
Berkman and Macintyre identified income, occupation, and education as the three most common internationally used indicators of SES, noting that the validity of these measures is borne out by “the consistency and strength of the association between SES and morbidity and mortality, both within and across countries.”3(p. 54) They observed that education is the most easily and consistently measured as well as the most reliable and valid indicator of SES. Unfortunately, many health studies reporting SES have failed to operationally define the construct.4 In studies examining infant development that did operationally define SES, one of the most common measures used was parental educational attainment, usually that of mothers.
Maternal education is but one of a number of factors identified that may affect child development during infancy, childhood and adolescence. Social and environmental factors include (1) maternal indicators (eg, education, occupation, ethnicity, income, smoking, drug and alcohol use, high-risk medical conditions, and age); (2) presence of siblings, and (3) infant indicators (eg, gestational age, birth weight, gender, high-risk medical condition(s), and occurrence of otitis media).2,5–10 A number of these factors have been established, in high-risk infant populations (eg, premature birth, low-birth weight, maternal smoking, etc.), to have an effect on infant or childhood development.
In one of the earliest studies to examine the relationship of SES to infant mental and motor development, the researchers postulated, based on Piaget’s concept of the sensorimotor period, that factors that influence mental development during infancy also would influence motor development.8 In this study, SES included the components of paternal and maternal education, family income, and fathers’ occupational status. The authors of this study of 27 two year-old infant/mother dyads reported significant, albeit modest, relationships between all four SES components and infants’ mental development (rs = 0.50, 0.52, 0.41, and 0.48, respectively, p < 0.05). Only maternal education was significantly but modestly related to the infants’ motor development (rs = 0.41, p < 0.05).
The majority of studies examining the relationship between maternal education and infant development focused on infants who were preterm or small-for-gestational age,5,6,11–15 infants’ exposed to toxic substances,15–17 or maternal conditions (eg, human immunodeficiency virus,18 depression,7 or poverty).7 While lower levels of maternal education have been shown repeatedly to be related to lower cognitive development in infants and young children,5,6,11,16–18 results of the few studies examining the relationship of maternal education to neuromotor development from infancy through later childhood are contradictory. These results range from showing an association of more advanced motor development (including visuo-motor coordination in adolescence) with higher maternal education3,12–14 to no relationship5,6,15,18 or, in one study, an inverse relationship.7 Clearly, further exploration is required to add to this limited and incomplete understanding of the relationship between maternal education and infant motor development.
The following analysis and discussion focus on the relationship of maternal level of education to neuromotor development during the first year of life as measured by the Harris Infant Neuromotor Test (HINT). The purpose of this study was to determine whether there was a relationship between maternal level of education and infant neuromotor development in infants who are developing typically. On the basis of the limited literature available, as well as on Piaget’s conceptualization of the sensorimotor period, we hypothesized that higher levels of maternal education would be associated with higher levels of motor performance, as assessed by the HINT.
Screening Instrument: Harris Infant Neuromotor Test (HINT)
Harris and colleagues2 described the HINT as a new neurodevelopmental tool developed for screening infants who are healthy or at high risk between the ages of three and 12 months in either clinical or research settings. While other recently developed tools, such as the Test of Infant Motor Performance (TIMP)19 and the Alberta Infant Motor Scale (AIMS)20 screen for motor development, the HINT also screens for cognitive and behavioral development, although it is primarily a neuromotor-screening test. The HINT also allows for inclusion of parental assessment of infant development, which the AIMS and TIMP do not.
The HINT is designed for administration and scoring in less than 30 minutes by health care professionals involved in early infant screening (eg, community nurses, physicians, occupational therapists, physical therapists, etc.). The instrument has three parts: (1) background information about the infant and caregiver, including parents’ level of education, occupation, age, etc., (2) the caregiver’s responses to a series of questions about the infant’s development, and (3) a 21-item professional assessment of the infant. The HINT is designed such that the total HINT score diminishes with increasing neuromotor development.
Good content validity has been established,21 and previous evaluation has shown the test is acceptable to infants, parents, and the health care professionals trained to use it.2 Assessment of the scale has demonstrated good inter- and intra-rater reliability and stronger predictive validity than the Bayley-II Motor Scale, which is more costly and time-consuming to administer.2,22
Infant Data Collection
The data for this study were collected from 412 healthy infants and their caregivers between October 2000 and December 2002 to establish Canadian norms for the HINT. Infants were recruited from British Columbia, Manitoba, Nova Scotia, Ontario, and Quebec and the data were collected by a group of examiners, primarily physical therapists and occupational therapists, with experience in assessing infants and young children. Examiners were trained by the second author in a two-day workshop on administration and scoring of the HINT. Inter-rater reliability with the trainer was established for each of the examiners for HINT Total scores (with ICCs ranging from 0.93 to 0.99). Infants were recruited by contacting staff at childcare centers and well baby clinics as well as community health nurses, using a letter of contact approved by the Clinical Ethics Review Board of the University of British Columbia (UBC). Informed consent was obtained from each infant’s parent(s) in accordance with the UBC ethics approval.
To ensure representation of Canadian infants, recruitment considerations included: (1) equal representation of infant gender at each of the 10 monthly age intervals from three to 12 months; (2) representation of Canadian ethnic diversity, particularly First Nations infants; and (3) parents with varying levels of education representative of Canadian demographics. Four levels of parent education were considered: zero to 11 years (no high school diploma), 12 years (high school diploma), 13–15 years (college diploma or technical/trades certificate), and more than 16 years (university degree).
For inclusion in the original study, infants had to be born at term (38 to 42 weeks’ gestational age [GA]), weigh more than 2500 g at birth, and have no history of major prenatal, perinatal, or postnatal medical complications or maternal complications. To remove a number of potential confounds, the exclusion criteria included premature birth (before 38 weeks’ GA), low birth weight (less than 2500 g), reported history of maternal alcohol or drug use during pregnancy, and mother or infant having a high-risk condition (eg, congenital heart defect or chromosomal abnormality).
Initial exploration of the data showed that the infant sample was almost equally split by gender with 204 (49.5%) female infants and 208 (50.5%) male infants. The number of infants (n = 412) was evenly split across each of the 10 one-month age levels (Table 1). The means and standard deviations for the total HINT scores are presented in Table 2 and, as expected, decrease with increasing age. Maternal age ranged from 16 to 43 years with a mean of 29 years. Mothers with education at all levels were included in the data set (Table 3). All infants and mothers met the inclusion criteria. There were no missing data for the variables of interest for the 412 infant-mother dyads.
General linear modeling, specifically multiple regression analysis, was used to allow for statistical control of the strong positive correlation between infant age and infant neuromotor development (ie, neuromotor development typically increases with increasing infant age). The data were observational, and the variables of interest to this analysis were (1) infant neuromotor development, indicated by the total HINT score for each infant (HINT), (2) infant age in months, and (3) maternal education level grouped according to the four levels outlined earlier. The hypothesis was that higher levels of maternal education would be associated with more favorable infant neuromotor development, that is, lower total HINT scores.
The data analysis included (1) preliminary exploration of the variables of interest, (2) testing of the assumptions underlying linear regression modeling, (3) consideration of data transformation, and (4) multiple linear regression of the data. The computer software package SPSS Version 12.0 was chosen to assist with computation.
Frequency histograms and box-plots were used to visualize and examine the central tendency, variability, range, and shape of distribution for each variable of interest (HINT, infant age, and maternal education).23 There were no outliers, and the data were positively skewed for the HINT, negatively skewed for maternal education and, as anticipated, uniformly distributed for infant age. Bivariate analysis indicated: (1) a highly negative relationship between infant age and HINT (r = −0.898, p = 0.000, one-tailed), that is, HINT scores decreased with increasing infant age, and 2) a weak positive relationship between maternal education and HINT (r = 0.147, p = 0.001, one-tailed).
The data were examined with respect to the key assumptions of the regression model (ie, linearity, independence, normality, and equality of variance) with regard to the properties of the errors (or equivalently, the distribution of the response variable conditional on the explanatory variable). Violation of these assumptions can lead to one of two problems: (1) bias of the estimate of the regression coefficients, thereby biasing R2, significance tests, and confidence intervals, or (2) bias of the estimate of the standard error of the regression coefficients leading to incorrect hypothesis tests and confidence intervals.23
The scatter-plot for the residuals against infant age suggested that the relationship was not linear (R2 = −2.22E-16) but rather quadratic (R2 = 0.166). The scatter-plot of the residuals against maternal education suggested that the relationship might be cubic (R2 = 0.002) rather than linear (R2 = 0.000). No obvious violations of independence or normality were noted. The scatter-plots of the residuals against infant age and maternal education appeared curved rather than linear, suggesting that there might be some unequal variance, particularly for infant age. The box-plot of the standardized residuals indicated five outliers with three cases above the maximum value and two cases below the minimum value.
Transformation of Data
Consideration was given to transforming the data to improve normality and equality of variance. However, the data were not transformed for reasons of maintaining the ability to interpret the results.
Taking into account the theoretically grounded need to covary for infant age, two approaches (sequential regression modeling and linear transformation) were taken to the multiple regression modeling. First, using sequential regression modeling to co-vary for infant age, there was a significant effect for infant age, t (410) = 0.097, p = 0.000, with older infants receiving lower HINT scores. There was a significant main effect for infant age, F (1, 410) = 1710.348, p = 0.000, with older infants receiving lower HINT scores. For every one-month increase in infant age the HINT score decreased by 4.010. The R2 was 0.807 indicating that the proportion of variance in HINT accounted for by infant age was 80.7%. Using Cohen’s24 criteria for effect size (R2: 0.0196, small; 0.1305, medium; 0.2592, large), the size of this effect was very large. With the addition of the explanatory variable maternal education, the R2 was unchanged at 0.807 and the R2 change was 0.000, that is, there was no effect of maternal education on HINT scores if the infants’ age was already in the model.
The variables in the sequential regression model were examined further. The maximum centered leverage point was 0.016. Huber25 suggested that a maximum leverage point of less than or equal to 0.20 is safe. Therefore, the maximum leverage point was retained. The variance inflation factor values at less than 2.0 indicated acceptable levels of co-linearity among the explanatory variables. The analysis was repeated excluding the outliers identified from the box-plot of the standardized residuals earlier. The outliers were not extreme and, as expected, there was no change to the results found with the outliers included.
Second, given the earlier finding that the HINT variable was positively skewed, the individual HINT scores were linearly transformed to z-scores (HINTz) using the mean and standard deviation (SD) of the HINT scores (Table 2) for each one-month age group. The distribution of the transformed variable was more normal. The scatterplots did not appear markedly different but the R2 for HINTz was 7.85E-6, indicating that maternal education accounted for only 0.0000785% of the variance in HINTz. The regression computation for HINTz to maternal education confirmed an expectation that maternal education was minimally associated with infant motor development, F (1, 410) = 0.003, p = 0.955.
The normative data set analysis described in the last section clearly demonstrated minimal association between maternal education and infant neuromotor development as measured by the Total HINT scores. The results failed to support our hypothesis that higher levels of maternal education would be associated with enhanced infant neuromotor development. The null hypothesis holds true for these data. This conclusion is congruent with findings of other researchers6,15,18 and contradicts the finding of a negative relationship reported by Petterson and Albers.7
Why did our study findings contradict those of several other studies?8,12–14 In the four studies in which a relationship between SES and children’s motor development was reported, three different measures of motor development were used. Only one of the studies examined motor outcomes exclusively during the first year of life and included only infants born preterm from low-income African-American (n = 41) and Hispanic families (n = 82).13 Further, motor development on the Bayley Scales of Infant Development was related to maternal education only for the infants who were African-American. Although the study by Hediger and colleagues used a large representative sample of infants from the United States (n = 4621) and was therefore more similar to our sample,14 81% of the participants were older than one year at testing (on the Motor and Social Development scale). The third study assessed visual-motor coordination of 90 adolescents who were Israeli, and prematurely born with very low birth weight.12 In the fourth study, infants who were healthy comprised the sample but were assessed between 24 and 29 months of age on the original version of the Bayley Mental and Motor Scales.8 Based on the results of the latter three studies, it could be speculated that the influence of SES on motor development might not appear until after the first year of life.8,12,14 On the basis of the findings in the one study13 that did exclusively involve infants, it is plausible that the relationship may have stemmed from the fact that the infants in that study were Black (an ethnic group under-represented in our Canadian sample) as well as pre-term and low-income.
Another important difference between those studies that found a relationship between SES and motor development and our study was in the measurement of the construct of SES. Of the four studies, only two used maternal education as a measure of SES8,13 (as in our study), while Hediger and colleagues used income of the “family reference person” as the measure for SES.14 As the Committee on Pediatric Research of the American Academy of Pediatrics stated: SES is “a complicated construct in its own right.”26(p. 1349) Although family income and parental education are the two most common measures used to operationalize SES in the health literature, Braveman and colleagues4 reported that the correlations between these two measures were only moderate, thus suggesting that they measure slightly different constructs.
The findings of our study were therefore not surprising given the complexity of the concept of SES and the many other factors that may influence infant neuromotor development. Additional model building to account for multiple other potential confounds was not possible with the data available.
A potential limitation of our study was the use of a screening test, the HINT, as a measure of neuromotor status, rather than using a more comprehensive developmental motor assessment tool. Nonetheless, when the original version of the HINT was subjected to content analysis by an interdisciplinary panel of 22 international experts in child development, 79% rated the test as good or excellent in its ability to screen for early neuromotor concerns and 73.7% rated it as good or excellent in the completeness of items for sampling neuromotor disability or developmental delay.21 Because construct validity is based, in part, on content validity, these results also support the premise that the construct of “neuromotor development” is imbedded within the HINT.27
An additional type of evidence that provides construct validity is based on the known groups method.27 A manuscript based on a study supporting the ability of the HINT to discriminate between low-risk and high-risk infants using the known groups method is currently in press in Physical Therapy.
CLINICAL IMPLICATIONS AND FUTURE RESEARCH
On the basis of our study findings, the need for standardized, discriminative tests of infant motor development to include normative samples that are representative of different levels of maternal education does not appear to be as critical as it would be for tests of cognitive development. Although the HINT normative sample included mothers from a variety of different educational levels and ethnic backgrounds representative of the Canadian population,2 the AIMS did not include ethnic or socioeconomic characteristics, for instance, maternal education, in their standardization sample.20 Our findings would suggest, therefore, that pediatric clinicians could use either test in trying to discriminate motor development differences in typically developing or high-risk infants, regardless of maternal education level.
Further study is recommended, however, to explore the relationship of maternal education to neuromotor development in typical infants. In particular, consideration needs to be given to greater delineation of level of maternal education, identification and inclusion of other indicators of SES, and identification and measurement of other variables known to influence infant motor development, such as aspects of the home environment.28 Further study of normative samples of infants, such as the one used in this analysis, is recommended given that the primary focus of previous research has been high-risk infant populations.
1. Capute AJ, Shapiro BK, Palmer FB, et al. Normal gross motor development: the influences of race, sex and socio-economic status. Dev Med Child Neurol
2. Harris SR, Megens AM, Backman CL, et al. Development and standardization of the Harris Infant
Neuromotor Test. Infants & Young Children
3. Berkman LF, Macintyre S. The measurement of social class in health studies: old measures and new formulations. LARC Sci Publ
4. Braveman P, Cubbin C, Marchi K, et al. Measuring socioeconomic status/position in studies of racial/ethnic disparities: maternal and infant
health. Public Health Rep
5. Hutton JL, Pharoah POD, Cooke RWI, et al. Differential effects of preterm birth and small gestational age on cognitive and motor development. Arch Dis Child
6. Kierscht MS, Meuwissen J. Follow-up of preschool age survivors of neonatal intensive care. J Psychol
7. Petterson SM, Albers AB. Effects of poverty and maternal depression on early child development
. Child Dev
8. Poresky RH, Henderson ML. Infants’ mental and motor development: effects of home environment, maternal attitudes, marital adjustment, and socioeconomic status. Percept Mot Skills
9. Badr Zahr LK. Quantitative and qualitative predictors of development for low-birth weight infants of Latino background. Applied Nurs Res
10. Trasti N, Vik T, Jacobsen G, Bakketeig LS. Smoking in pregnancy and children’s mental and motor development at age 1 and 5 years. Early Hum Dev
11. Black MM, Sazawal S, Black RE, et al. Cognitive and motor development among small-for-gestational age infants: impact of zinc supplementation, birth weight, and caregiving practices. Pediatrics
12. Levy-Shiff R, Einat G, Mogilner MB, et al. Biological and environmental correlates of developmental outcomes of prematurely born infants in early adolescence. J Pediatr Psychol
13. Zahr LK. Predictors of development in premature infants from low-income families: African Americans and Hispanics. J Perinatol
14. Hediger ML, Overpack MD, Ruan WJ, et al. Birthweight and gestational age effects on motor and social development. Pediatri Perinat Epi
15. Shankaran S, Papile LA, Wright LL, et al. Neurodevelopmental outcome of premature infants after antenatal phenobarbitol exposure. Am J Obstet Gynecol
16. Singer LT, Arendt R, Minnes S, et al. Cognitive and motor outcomes of cocaine-exposed infants. JAMA
17. Ornoy A. The impact of intrauterine exposure versus postnatal environment in neurodevelopmental toxicity: long-term neurobehavioral studies in children at risk for developmental disorders. Toxicol Lett
18. Chase C, Ware J, Hittelman J, et al. Early cognitive and motor development among infants born to women infected with human immunodeficiency virus. Pediatrics
19. Campbell SK. The Test of Infant Motor Performance: Test User’s Manual Version 1.4
. Chicago, IL: Infant
Motor Performance Scales; 2001.
20. Piper MC, Darrah J. Motor Assessment of the Developing Infant
. Philadelphia: WB Saunders; 1994.
21. Harris SR, Daniels LE. Content validity of the Harris Infant
Neuromotor Test. Phys Ther
22. Harris SR, Daniels LE. Reliability and validity of the Harris Infant
Neuromotor Test. J Pediatr
23. Cohen J, Cohen P, West SG, Aiken LS. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences
. 3rd ed. Mahwah, NJ: Lawrence Erlbaum Associates; 2003.
24. Cohen J. A power primer. Psychol Bull
25. Huber PJ. Robust Statistics
. New York, NY: Wiley; 1981.
26. American Academy of Pediatrics. Committee on Pediatric Research. Race/ethnicity, gender, socioeconomic status – research exploring their effects on child health: a subject review. Pediatrics
27. Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice
. 2nd ed. Stamford, CT: Appleton & Lange; 2000:83–89.
28. Abbott AL, Bartlett DJ, Fanning JEK, et al. Infant
motor development and aspects of the home environment. Pediatri Phys Ther