Study population
VLBW infants born at Seoul Hanyang University Hospital in South Korea between December 2013 and March 2015 and admitted to the level 3 Neonatal Intensive Care Unit, were eligible for the study. The inclusion criteria was the VLBW preterm infants who were recruited from the Hanyang Developmental Medical Center for premature infants at the corrected age of 6–10 months or 16–20 months for follow-up because this is the standard practice for neurodevelopmental assessment of VLBW preterm infants [9, 15, 16]. For the control group, we recruited term infants with a gestational age of ≥37 weeks, born at Hanyang University Hospital. These controls were age-matched with VLBW preterm infants for corrected age, and they participated in the study at the well-baby clinic. The exclusion criteria were major congenital malformations, severe brain injury (periventricular leukomalacia on brain magnetic resonance imaging (MRI) or intraventricular haemorrhage ІІІ-ІV grade), metabolic disorder, high risk of developmental delay, retinopathy of prematurity ІІІ-ІV grade or any sign of neonatal encephalopathy or seizure. A total of 45 VLBW preterm infants were recruited from Hanyang Developmental Medical Center for follow-up at the corrected age of 6–10-month or 16–20-month visit. Neurodevelopmental outcomes were assessed in eligible study infants at the 6–10 or 16–20 months visit in the VLBW preterm follow-up program, using the Bayley Infant Neurodevelopmental Screener (BINS), which assesses cognitive capability, language, gross motor skills and fine motor skills. The risk status classifications of BINS is minimally affected by environmental variables, when compared with the Bayley Scales of Infant Development-II, suggesting that it has predictive utility. The age corrected for prematurity was used in the BINS evaluation. Using the BINS score, we categorized the infants in groups at low risk and high risk of developmental delay or neurodevelopmental impairment. The VLBW preterm infants at low risk were eligible for the study. Of the original 45 infants, 8 were high risk based on the BINS evaluation, one parent refused consent, 4 were lost to follow-up and 2 had abnormal brain MRIs. The control group was screened for neurodevelopment at the well-baby clinic at the 6–10 or 16–20 months, using BINS. The control group were born healthy and developed normally, with comparable groups based on the BINS test. Thus, a total of 30 VLBW preterm infants were eligible and 25 full-term infants were matched with the VLBW preterm infants for corrected age to form a control group. However, 10 of the 55 infants were excluded due to incomplete interventions and 1 was excluded due to insufficient data quality. Eye-tracking was finally completed for 26 VLBW preterm infants and 18 term infants (Fig. 1). The object permanence test was assessed in 19 infants at the corrected age of 6–10 months, and attention capacities were compared in 44 infants at corrected ages of 6–10 or 16–20 months in the VLBW preterm cohort, or at these chronologic ages in the term infants. This study was approved by the Hanyang University Institutional Review Board [No. 20141226]. The parents were given a full explanation of the purpose and nature of all procedures, and informed parental consent was obtained before data collection.
Neonatal risk factors in VLBW preterm infants
Prenatal and neonatal data were based on medical records, including gestational age (GA), birth weight, delivery mode and sex. Small for gestational age, maternal chorioamnionitis, prenatal steroid use, bronchopulmonary dysplasia (BPD, ≥ moderate), retinopathy of prematurity (ROP) and intraventricular haemorrhage I-II were recorded for VLBW preterm infants. Chorioamnionitis was defined by histologic chorioamnionitis or umbilical cord vasculitis of grade 2 or greater, according to the grading system suggested by Salafia et al. [17]. The diagnosis and severity of BPD were determined by assessing the need for supplementary oxygen at 28 days of age and 36 weeks postmenstrual age, the infants breathing air had mild BPD, those who needed <30% supplementary oxygen had moderate BPD, and those needing >30% supplementary oxygen and/or continuous positive airway pressure or a ventilator were defined as having severe BPD [18]. IVH was classified according to Volpe. [19]. Maternal education was categorized as high (more than 10 years), middle (6 to 10 years) or low (less than 6 years), based on the number of years of post-elementary education.
Eye-tracking assessment
The experiment was conducted in a assigned room (see Fig. 2), where two 19″ monitors and one eye-tracker (Tobii© X2–60, Tobii, Stockholm, Sweden) were set up. The stimuli for object permanence were presented on one of the monitors (i.e., the stimulus monitor), and the eye-tracker was attached at the bottom of the monitor. The other monitor (i.e., the experimenter monitor) was used by the experimenter to control the experimental session. The infant was seated on the parent’s lap approximately 60 cm from the stimulus monitor. If the infant was not able to maintain this 60 cm distance, he or she was seated on the desk and a parent held him or her from the back (Fig. 2). All eye movements were recorded using the eye-tracking system, which had an accuracy of 0.4 degrees at a rate of 60 Hz. Prior to data collection, a 2-min calibration of the eye-tracking system was carried out. An appropriate sound intensity level (dB) was selected, and the eye-tracking calibration was carried out while the participant was watching an infant-friendly movie. The main experimental session with infants at 6–10 month involved 3 tasks that were counterbalanced and took approximately 10 min.
Tasks
We designed the experimental tasks to induce attention and object permanence based on previous reports, using a modification of the methods of de Jong et al. and Lowe et al. [1, 9]. The tasks consisted of watching 3 video clips in which an actress presented different stimuli involving 2 cups and a yogurt bottle.
Prior to starting the three tasks, cartoon images of toys with sound effects were used to draw the infants’ attention to the screen. The root mean square (RMS) of the noise of the eye-tracking signals is a measure of data quality. There was no significant difference between the RMS noise in the VLBW preterm and term groups, indicating that the quality of the eye-tracking data was amenable to statistical analysis (Wilk’s Λ = 0.93, F8,190 = 1.88, P = 0.07).
Task 1: Basic object permanence test (Fig. 3a). The actress picks up the yogurt bottle and hides it under the left cup, then exchanges the two cups on the table. She says, “Where is the yummy yogurt bottle?” while hiding the yogurt bottle under the left cup, and then exchanges the position of the two cups again. Here, the infants believe that the left cup and the yogurt bottle are the same object, so following the moving object (i.e., both the left cup and the yogurt bottle) is seen as a basic level of object permanence. Each stimulus was displayed for 10 s. Two stimuli were administered to each child (two trials per task).
Task 2: Advanced object permanence test (Fig. 3b). The actress picks up the yogurt bottle. She initially hides it under the left cup. A few second later, she hides it under the other cup, saying “I am hiding it again.” Here, there is no reversal of the two cups, only a change in the location of the yogurt bottle. The infant needs to employ a higher level of cognitive capability to simultaneously manage two hiding steps (i.e., the yogurt bottle under the left cup, and the yogurt bottle under the right cup) and object permanence (i.e., while they observe that the yogurt bottle is moving from the left cup to the right cup) as they retrieve where the hidden object is. Each stimulus was displayed for 10 s. Two stimuli were administered to each child (two trials per task).
Task 3: Attention test (Fig. 3c). The stimulus video shows the actress’s face for 1 s; then she lifts a yogurt bottle to shoulder height and shifts her gaze to the bottle with a verbal indication (“Let’s look at this. It is a yummy yogurt”). Gaze direction was not altered in referring to the object. The stimulus is displayed for 4 s. Two stimuli were administered to each child (two trials per task).
Analysis of the eye-tracking data
For analysis, we developed several areas of interest (AOIs) representing the primary analytic regions where the infants looked and how they responded to the various stimuli. In general, all of the AOIs were rectangular areas that covered the movement of a special object used as the stimulus.
Object permanence test: In Task 1, the left cup was the only AOI. The left cup hiding the yogurt bottle moved to the right position where the right cup was originally located. To analyze the infants’ gaze and object choice followed by basic object permanence, we assessed the infants’ gaze shift at the left cup while the yogurt bottle was hiding, and after moving the left cup. Hence, the infant was considered to have basic object permanence if he or she maintained the gaze on the left cup even after the cup was moved. In Task 2, the stimuli consisted of two types of object permanence: i) the yogurt bottle is hidden under the left cup (then, the actress lifts the left cup and hides the yogurt bottle under the right cup, saying “I am hiding it again”), ii) the yogurt bottle is hidden under the right cup. The accuracy of all gaze shifts was scored by (the number of accurate gaze shifts)/(the total number of trials). The maximum object permanence score was 2, and the minimum score was 0.
Attention test: The yogurt bottle was the main AOI, starting from the actress saying “Let’s look at this. It is a yummy yogurt.” The distractor was the actress’s face, another AOI. The analysis of Task 3 involved checking how well our participants paid attention to the referential object, the yogurt bottle. To analyze the infants’ referential gaze, we assessed the looking time of the target objects by quantifying sustained attention after attention shifted. We calculated the relative proportion of the former (looking time at the yogurt bottle/looking time at both the face and the yogurt bottle). The categorical definition of ‘sustained attention’ was that the infants had to shift their eyes and consistently fix on the target for at least 0.33 s [20].
Statistical analysis
All the analyses were carried out with SPSS 22 (IBM, Armonk, NY, USA.). We checked data for normal distribution. Data were analyzed for normality of distribution using the Kolmogorov-Smirnov test. Continuous measures were summarized and analyzed using parametric statistics. Normally distributed variables are presented as mean ± SD, and non–normally distributed variables as median value and range. The proportions of looking time, and gaze shift scores, were compared by t-tests or Mann-Whitney U tests in order to identify differences in attention performance and object permanence between VLBW preterm and term infants. A multivariate linear regression analysis was conducted to determine risk factors associated with attention in the VLBW preterm groups. Because looking time during the attention test is a continuous measure, a logistic regression model using the categorical definition of attention was used to compare attention function in VLBW preterm infants with different medical morbidities, while controlling for gestational age, sex, and adjusted age at testing P values less than 0.05 were considered statistically significant.