Early Working Memory in Children Born With Extremely Low Birth Weight: Assessed by Object Permanence

Abstract

Object permanence is considered the earliest method for assessing working memory. Factors affecting object permanence performance in a sample of two hundred and thirty-three 18- to 22-month olds born with extremely low birth weight were examined. It was hypothesized that object permanence would be directly related to emotional and attention regulation, that children with lower birth weight and higher illness severity would have more difficulty on the object permanence task, and that no ethnic/racial differences would be found, as this is considered a culturally unbiased task. Attainment of object permanence had a significant positive association with emotional and attention regulation, even after controlling the medical severity and socioeconomic factors. More girls than boys passed the object permanence items. There was no ethnic/racial difference on object permanence. Our findings indicate that object permanence may be a culturally fair way of assessing development, and emotional and attention regulation are potential avenues of intervention for such skill.

It is well documented that infants born with extremely low birth weight (<1000 g) are at risk of displaying cognitive, attention regulation, and self-regulation difficulties.¹ These difficulties can persist throughout childhood and are associated with an increased incidence of learning difficulties, attention-deficit/hyperactivity disorder, and behavioral problems throughout childhood.^2,3

Recently, an association with working memory (as measured by tasks such as the A not B task) has been shown with cognitive, behavioral, and academic outcomes of children born preterm.⁴ Working memory refers to the process of holding task-relevant information in mind for brief intervals so that the information can be used to guide future actions⁵ and is considered essential for higher order cognitive functioning.⁶ Studies examining early working memory have shown that children born preterm have impaired working memory throughout childhood.^4,7–11 Impairment in this skill is significantly associated with later risks of global intellectual and academic difficulties at school age in children born preterm.^12,13

Studies examining the working memory impairment have highlighted the role of the dorsolateral prefrontal cortex. Goldman-Rakic¹⁴ first documented that lesions in the dorsolateral prefrontal cortex of monkeys resulted in impairments on spatial delayed response, an essential skill for the A not the B task. Further investigation by Goldman-Rakic⁵ substantiated that the ability to perform delayed response tasks such as the A not B task coincided with the period of highest synaptic density growth in the cortical regions of nonhuman primates.

There is increasing evidence that self-regulation and attention regulation play an essential role in working memory performance.^6,15 Previous studies have shown that the infants who demonstrate self-regulatory problems have more difficulty exploring and attending to the environment, limiting their ability to engage effectively in working memory tasks.^6,15 Although this association between self-regulation and working memory performance has been demonstrated in infants born full-term,^6,15 no study to date has examined this relationship in a population of infants born with extremely low birth weight.

The purpose of this study was to better understand the early working memory as measured by object permanence tasks in 18- to 22-month olds born with extremely low birth weight, and its relationship to measures of cognition (ie, Mental Developmental Index) and self-regulation (ie, emotional and attention regulation). We hypothesized that children born with extremely low birth weight would display difficulties on the working memory task (ie, object permanence) directly associated with difficulties in emotional and attention regulation. The relationship between object permanence performance and medical factors such as birth weight, gestation, illness severity, and hydrocortisone therapy and sociodemographic factors such as gender and maternal education was also explored.

Additionally, we postulated that object permanence performance would be directly related to a measure of cognition (Mental Developmental Index) but that unlike the Mental Developmental Index scores, performance would not differ by ethnicity, as this should be a culturally neutral, unbiased test. In contrast, we expected that the Mental Developmental Index would be affected by race and ethnicity, as it has previously been shown to be ethnically biased.¹⁶

Methods

Infants eligible for this follow-up study were the surviving infants who had been enrolled in a multicenter study of low-dose hydrocortisone therapy for prophylaxis of early adrenal insufficiency.¹⁷ Singletons and twins between 500 and 999 g birth weight were eligible if they were mechanically ventilated at the study entry (12–48 hours postnatal age), as this was a requirement for the larger study. The study protocol was approved by the institutional review boards at all participating institutions, and parental consent was obtained prior to enrollment. At the evaluation, demographic and medical histories were obtained, and weight, height, and head circumference were also recorded.¹⁷

Development was assessed with the Bayley Scales of Infant Development II¹⁸ with a Mental Developmental Index calculated as a measure of cognition. All examiners administering the Bayley Scales of Infant Development II were trained and certified.

Emotional and attention regulation were assessed using the Emotional Regulation and Orientation/Engagement scales of the Bayley Scales of Infant Development II and the Behavior Rating Scale, respectively. The examiner who administered the Bayley Scales of Infant Development II scored the Behavior Rating Scale, based on the observation of the child during testing. The 30-item Behavior Rating Scale contains questions regarding the child’s sustained attention, distractability, task reactivity, task persistence, affect, and social orientation. The rating scale was completed by the examiner after administering the Bayley Scales of Infant Development II Mental Scale, and each item was rated on a 5-point likert scale. The examiners completing the Behavior Rating Scale were blinded to the object permanence study, not knowing how the object permanence items would be used.

Items 84, 96, and 102 of the Bayley Scales of Infant Development II Mental Scale were used as measures of object permanence. The object permanence items sequentially increased in difficulty. First, the child was asked to find a toy hidden under 1 of 2 cups (item 84). Second, the cups were reversed after the toy was hidden (item 96). Third, double visual displacement was used; the toy was hidden under 1 cup, removed and hidden for the second time under the second cup (item 102). The number of object permanence items correctly completed was calculated for each child and used as an ordinal measure. Birth weight, gestational age, and the Clinical Risk Index for Babies¹⁹ (CRIB), a measure of illness severity, were used as indicators of medical risk, whereas maternal education as a proxy for family socioeconomic status. Head circumference (occipital-frontal circumference) at testing was measured as a measure of medical risk (ie, growth restriction).

Statistical Analysis

Achievement of object permanence was analyzed using ordinal logistic regression. We first used an unadjusted model, with each variable separately entered into the regression equation. Then, a full model included adjustment for the following characteristics: gestational age, birth weight, gender, ethnicity, CRIB score, hydrocortisone treatment, maternal education, and emotional and attention regulation. Unless otherwise noted, all hypotheses tests were 2-sided and used a significance level of .05. All statistical analyses were conducted using SAS Version 9 (SAS Institute Inc, Cary, North Carolina).

Results

Of the surviving 291 infants of the original multicenter study,¹⁷ 250 (87%) were assessed at follow-up. There were insufficient numbers of several ethnic/racial groups for adequate analysis; therefore, 17 children (4 native Americans, 2 black Hispanics, and 11 Asians) were excluded from analysis. A total of 233 (80%) of the surviving cohort are the participants of these data. Demographic characteristics are shown in Table 1.

Table 1.

Family Demographic Characteristics

Characteristic	Summary Statistic
Adjusted age at testing, in months	19.9 (2.0)
Birth weight in grams, mean (SD)	736 (122)
Gestational age in weeks, mean (SD)	25.3 (1.6)
CRIB score, median (25–50 percentile)	8 (5–10)
Head circumference, mean (SD)	47.4 (1.85)
Male gender, n (%)	121 (52)
Ethnicity, n (%)
Caucasian	90 (39)
African American	118 (51)
Hispanic	25 (11)
Maternal education, n (%)
Less than high school	45 (19)
High-school degree	48 (21)
Greater than high school	73 (31)
College degree or more	60 (26)
Unknown	7 (3)

CRIB, Clinical Risk Index for Babies; SD, standard deviation.

The object permanence scores ranged from 0 to 3 in the following manner: 15% completed no items, 15% completed 1 item correctly, 14% got 2 items correct, and the remaining 56% correctly answered all 3 items. Object permanence showed significant positive correlations with Orientation/Engagement and Emotional Regulation scores on the Bayley Scales of Infant Development II, Behavior Record (P = .002 and P < .001, respectively). After controlling the medical and sociodemographic variables, only the relationship between object permanence and emotional regulation remained significant (P = .018).

For the simple 1-factor analysis, girls performed significantly better than boys on the object permanence task (P = .004). After adding medical, sociodemographic, emotional and attention regulation variables as covariates, gender remained significant (P = .007). No significant differences were found between racial and ethnic groups in the achievement of object permanence. Medical risk factors, head circumference at 18 to 22 months, and maternal education were not related to the object permanence score.

Object permanence was significantly associated with cognition (Bayley Scales of Infant Development II Mental Developmental Index), such that the Mental Developmental Index scores increased as the ordinal object permanence score increased (P < .001). Birth weight, gestational age, CRIB score, and gender were all significantly related to the Mental Developmental Index score in an unadjusted model but were not significant in the full model analysis. Ethnic and racial group was significantly related to the Mental Developmental Index score (P < .001), that is the Hispanic and black children had significantly lower Mental Developmental Index than the white children had (mean ± SD: Hispanic and black 77.4 ± 16.5 and 72.5 ± 16.0 vs white children 85.9 ± 20.0). The racial/ethnic differences remained significant after controlling the medical, sociodemographic, and emotional and attention regulation variables (P < .001). Girls scored significantly higher than boys on the Mental Developmental Index score (mean ± SD: girls 84.82 ± 18.89 and boys 74.81 ± 18.09), when a full model was used with sociodemographic, medical, emotional and attention covariates (P < .001; Table 2).

Table 2.

Unadjusted and Full Model for Object Permanence

Variable	Unadjusted Model	P Value	Full Modela	P Value
	Odds Ratio (95% Confidence Interval)		Odds Ratio (95% Confidence Interval)
Birth weight, g	1.000 (0.998–1.002)	.810	1.000 (0.996–1.003)	.858
Gestational age, weeks	0.911 (0.772–1.075)	.271	1.024 (0.821–1.276)	.834
CRIB score	1.014 (0.943–1.090)	.705	1.014 (0.905–1.136)	.816
Genderb	2.170 (1.288–3.656)	.004	2.193 (1.238–3.891)	.007
Ethnicityc
African-American	1.129 (0.655–1.948)	.662	1.069 (0.573–1.994)	.834
Hispanic	1.935 (0.853–4.388)	.114	1.844 (0.673–5.052)	.234
Maternal educationd	1.073 (0.882–1.304)	.482	1.062 (0.847–1.332)	.603
Emotional regulation	0.980 (0.969–0.990)	<.001	0.983 (0.970–0.997)	.018
Orientation/Engagement	0.984 (0.974–0.994)	.002	0.991 (0.978–1.004)	.174
Hydrocortisone therapye	0.767 (0.461–1.276)	.307	0.771 (0.441–1.346)	.360

CRIB, Clinical Risk Index for Babies.

^aThe full model includes birth weight, gestational age, CRIB score, gender, ethnicity, maternal education, emotional regulation, Orientation/Engagement, and hydrocortisone therapy as randomized in the original trial.

^bFemale is the reference category.

^cCaucasian is the reference category.

^dMaternal education is defined as the highest grade.

^ePlacebo is the reference category.

Object permanence was significantly associated with adjusted age at the time of testing in both the unadjusted model (P = .01) and the full model (P = .03). As the age at the time of testing increased so did the odds of having higher object permanence score. The findings of the full model, however, remained the same with gender and emotional regulation score maintaining significance in relation to the object permanence score.

Discussion

The primary purpose of this study was to better understand the early working memory in 18- to 22-month olds born with extremely low birth weight by examining the association between object permanence and self-regulation (ie, emotional and attention regulation) as well as the impact of ethnicity, gender, illness severity, and family socioeconomic status on object permanence performance. We found that the object permanence mastery was highly associated with measures of self-regulation, indicating that emotional and regulatory difficulties in children born with extremely low birth weight are associated with poorer performance on early working memory measures. In addition, we found a significant gender difference on object permanence performance, with girls having twice the odds of achieving object permanence mastery as boys. Adjusted age at the time of test was also significantly associated with the object permanence scores but did not affect the results regarding the significant gender findings, the association with self-regulation measures, and the lack of association with ethnicity. As expected, the Mental Developmental Index scores were highly associated with the object permanence scores, as 3 of the 25 cognitive items (ie, 12%) were used as the object permanence measure. However, despite the overlap in items, there still remained a clear contrast between the Mental Developmental Index and the object permanence scores as it regards to the impact of gender and ethnicity. As hypothesized, object permanence performance was not affected by ethnicity or race, in contrast to the Mental Developmental Index scores that were significantly affected by ethnic or racial group. Contrary to expectations, medical illness severity and family socioeconomic variables were not significantly associated with the object permanence mastery.

Recent studies indicate that working memory is a significant area of impairment in children born preterm.^4,7–11 Woodward et al⁴ found that 2-year-old children born preterm had difficulty encoding new information in working memory and showed bilateral reduction in cerebral tissue volume in areas specifically related to working memory (ie, dorsolateral prefrontal cortex, parietooccipital, and premotor areas) on MRI scans, compared with children born full-term. Studies also indicate that working memory deficits in this population contribute significantly to the cognitive, behavioral, and academic outcomes of these children.^12,13 Rose et al¹² found that working memory impairments at 12 months predicted the academic achievement and cognitive outcomes at 6 years of age in a sample of children born with very low birth weight (<1500 g). Given the importance of working memory impairment in predicting future outcomes, understanding the correlates of working memory deficits and the impact of factors such as ethnicity, gender, illness severity, and family socioeconomic status on such skills may help in early identification of such deficits, possibly permitting the use of early intervention techniques to ameliorate the working memory skills and related childhood outcomes.

The significant correlation we found in this study between working memory tasks and measures of self-regulation indicates that children who demonstrate difficulty in regulating are more likely to show difficulty on early working memory tasks, such as the object permanence tasks. Children who display attention and/or emotional regulation difficulties may be less able to maintain their attention and encode critical aspects of the object permanence task and, as a result, have difficulty completing the task successfully. This is supported by findings indicating that children who have regulatory difficulties will, in the course of their attempts to regulate, decrease their attention to the environment, show less exploration of the environment, and show a diminished motivation to engage in exploration.²⁰ The relationship found between self-regulation and early working memory performance highlights the potential avenues of intervention. Treatment aimed at ways to enhance attention and emotional regulation in early childhood may help set the foundation for more advanced working memory skills.

Our finding that girls did better than the boys on a test of working memory has not been reported in previous studies, although previous studies have reported poorer cognitive outcomes among boys.^1,21 Vohr et al¹ found that among the 18- to 22-month-old infants born with extremely low birth weight, male gender was a significant predictor of a subnormal Mental Developmental Index score (<70). Studies examining the school-age outcomes of children born with extremely low birth weight also indicate higher rates of cognitive impairments, learning disorders, and behavioral problems in boys.^22–24

The impact of ethnicity on intelligence testing has been explored since the 1960s when Arthur Jensen began the scholarly debate on race and intelligence.¹⁶ Findings from outcome studies have been mixed, with some indicating that maternal race or ethnicity adds prognostic information to poorer developmental outcome,²⁵ whereas others attribute differences to socioeconomic status,²⁶ maternal education,²⁷ or non-white race.^1,28 Our findings that the working memory items from the Bayley Scales of Infant Development II are not different among ethnic groups, whereas the Mental Developmental Index scores are, provide an additional reason to explore the measures of early working memory and other executive function measures. These types of measures may potentially be ethnically unbiased for minorities in contrast to traditional tests of cognition.

Our finding that the adjusted age at testing was related to the object permanence score was the one that we expected. It has been well documented^5,29 that object permanence is developmental in nature. In fact, children are unable to do this task before they have mastered the ability to recall that an object exists when out of view. The object permanence tasks presented in the Bayley Scales and also in tests such as the A not B task³⁰ increase in difficulty to measure various levels of spatial memory that are developmental in nature.

Limitations of our study include the absence of a full-term control group and the small numbers within our ethnic groups, especially Hispanic. Additionally, the lack of effect found for gestational age, birth weight, and illness severity may be because of the narrow gestational age and birth weight window and high illness severity of the original study cohort. Because the object permanence items were taken from the Bayley Scales of Infant Development, our object permanence task did not have the increasing delays in item retrieval that is often included in the A not B task (the standard object permanence task). These items also comprised part of the overall Mental Developmental Index score, though they represented only 3 of more than 25 items generally administered.

Future studies should examine ways to better assess the working memory in children born with extremely low birth weight at young ages, to assist in better identification of those children at greater risk for later attention and learning problems. Measures of working memory should be included in future studies of developmental outcomes, allowing us to go beyond measures of cognition which can be ethnically biased and have been shown to be poor predictors of later cognitive function.³¹ Curricula are being developed which specifically target ways to enhance the early working memory.³² These curricula, coupled with better ways to identify children with early working memory problems, could provide a framework for earlier, more comprehensive intervention programs to improve the outcomes for this very high-risk population.