Language development of low risk preterm infants up to the age of 30 months

Early Human Development 90 (2014) 649–656

Contents lists available at ScienceDirect

Early Human Development journal homepage: www.elsevier.com/locate/earlhumdev

Language development of low risk preterm infants up to the age of 30 months Miguel Pérez-Pereira a,⁎, Pilar Fernández a, María Luisa Gómez-Taibo b, Mariela Resches a a b

University of Santiago de Compostela, Spain University of A Coruña, Spain

a r t i c l e

i n f o

Article history: Received 23 March 2014 Received in revised form 8 August 2014 Accepted 12 August 2014 Keywords: Preterm children Language development Vocabulary development Grammatical development Determinant factors Environmental factors Biomedical factors CDI

a b s t r a c t Purpose: To compare the lexical and grammatical development of a group of low risk preterm children with a group of full-term children at 10, 22, and 30 months of corrected age. In addition, the effect of possible determinant factors on linguistic development was investigated. Method: An initial group of 150 low-risk PR children (mean GA: 32.62) and 49 FT children (mean GA: 39.70) recruited at birth were assessed at 10, 22, and 30 months of age. Communicative and linguistic abilities were measured at these three points in time through the CDI. Cognitive development and quality of family environment of the children, among other variables, were also assessed at 22 months of age. Hierarchical regression analyses were performed in order to test those factors which may contribute to prediction of language outcomes. Results: There was no significant delay in communicative, lexical or grammatical development of PR children. Even when comparisons were performed between fullterm and very preterm children, differences were not significant. Regression analyses indicate that gestures and early word comprehension predict very early word production development, but their effect disappears with time. The most important factors which predict language development at 30 months of age are previous cognitive scores and word production at 22 months of age. The results coming from group comparisons and from hierarchical regression analyses indicate that GA does not significantly affect language development from 10 to 30 months of age. Conclusions: Low risk preterm toddlers do not seem to be delayed in their linguistic development. © 2014 Elsevier Ireland Ltd. All rights reserved.

1. Introduction In recent years the study of first language acquisition by preterm (PR) children has received increasing attention. This interest is twofold: practical and theoretical. From a practical point of view, preterm (PR) children are a group that has a higher risk of suffering developmental problems due to their immaturity. Therefore the study of this population is an area of great concern for researchers and practitioners in order to promote efficient intervention. From a theoretical perspective, scholars' interests have been focused on discovering if PR children show atypical trajectories in their development [1] and on discovering those factors which may predict their linguistic development. However, PR children are not a homogeneous group, and therefore not at equal risk of developmental delays [2]. Usually, PR children are classified according to gestational age (GA) into: late preterm (GA 34–36 weeks), moderately preterm (GA 32–33 weeks), very preterm (VPR) (GA between 28 and 31 weeks), and extremely preterm ⁎ Corresponding author at: Department of Developmental Psychology, Faculty of Psychology, University of Santiago de Compostela, 15782 Santiago de Compostela, Spain. E-mail address: [email protected] (M. Pérez-Pereira).

http://dx.doi.org/10.1016/j.earlhumdev.2014.08.004 0378-3782/© 2014 Elsevier Ireland Ltd. All rights reserved.

children (EPR) (GA below 28 weeks). Different studies have shown that not only gestational age (GA) and birth weight (BW) (which are usually correlated: the shorter the GA the lower the BW) are factors that predict later linguistic outcomes, but also medical complications, such as bronchopulmonary dysplasia or periventricular leukomalacia, along with environmental factors are important determining factors. The risk of medical complications increases as GA and BW are lower [3]. Extremely and very preterm children have a greater probability of being affected by them than late preterm children. Therefore it would be expected that PR children with different GAs or BWs, with a different incidence of medical complications, and coming from different family environments, should have different linguistic (developmental) outcomes. Previous studies tended to confirm that PR children show language delays in relation to FT children as measured through vocabulary production or grammar scores [4–11]. Differences were greater when comparisons were performed between the extreme groups of very preterm or very low birth weight (VLBW) children and FT children, indicating that GA and BW affect language development [6,12]. A few studies found differences only after a given point in time (around age 18 months), although not earlier [13–15]. In contrast, other studies

650

M. Pérez-Pereira et al. / Early Human Development 90 (2014) 649–656

did not find significant differences between PR and FT children's language development [16–21]. Most of these later studies were carried out with a wider range of PR children without health problems. Discrepant results among previous studies may be due to different reasons:

and FT children [28,29], and on those investigations which indicated that early language abilities are the main predictor of later language abilities [10,30].

1) Differences in the characteristics of the samples studied in terms of their GAs. 2) Exclusion criteria are not always carefully described and controlled a reason why the role played by medical problems is difficult to ascertain. This circumstance is a menace to internal validity, and may lead to confounding the effects of GA with those of medical problems. 3) Age of assessment varies among studies, and this could also have affected the results obtained. 4) Finally, the use of different instruments to assess language development may also have resulted in apparent differences between studies.

2.1. Participants

Several factors were found to affect language outcomes in preterm children. A few studies found that biomedical factors seem to have an effect on later language scores. Among those factors are Apgar score at birth [16,20–23], length of stay in Neonatal Intensive Care Unit (NICU) [21,24,25], gestational age [5,6,11], gender [10,11,13,17], and medical complications such as bronchopulmonary dysplasia, intraventricular hemorrhage higher than grade II, or periventricular leukomalacia [2, 22,26–30]. Other studies found an influence of environmental factors such as mothers' education [16,18,30,31] or quality of family environment [26,32,33]. Finally, there were other studies which found that previous cognitive development [9,18,30], or previous communicative (use of gestures) or linguistic (word comprehension and production) development [9–11,34] are good predictors of later linguistic development. Connected with this, a number of studies indicate that PR children obtain significantly lower results in cognitive scales than FT children [4, 9,10,27,29,35]. Approximately 18% of EPR children with cognitive delay also show language delay, although another 14% of EPR children show specific language delay without associated cognitive delay [10]. Most studies on language development of PR children were carried out with very or extremely preterm (VPR or EPR) children, or with samples of PR children in which medical complications and environmental circumstances were not carefully controlled. It is important to note that EPR or VPR children constitute around 20% of the entire population of PR children [3]. Therefore it is of extreme relevance to carry out studies with PR children with characteristics more representative of the entire PR population. The present study investigates language development up to 30 months of age in a sample of PR children with a relatively wide variety of GA, and without associated medical problems. Given the characteristics of the sample, this sample could be defined as a low risk sample. The PR children's development will be compared to a control group of FT children of similar characteristics. Since this is a longitudinal study, it is possible to observe whether differences between PR and FT children appear at a given point in time or not. The second aim of this research is to identify factors that may predict language development. The effect of most of the former biomedical, contextual and personal factors on language outcomes at different ages will be also investigated in this study. Our hypotheses are 1) that low risk PR children will not show significantly lower results than those of the PR children, 2) that therefore GA will not have any predictive role on children's language development, which will mostly depend on previous language abilities. The first hypothesis is based on the idea that PR children are not a homogeneous group [2] as well as on the findings obtained in previous studies carried out with low risk PR children [16,17,19–21], who obtained similar results to those of FT children. The second hypothesis is based on those studies which found significant differences in language development between PR children with associated medical complications and FT children, but not between PR children without medical complications

2. Method

A group of 150 PR children, and another group of 49 FT children were recruited for a longitudinal project just after birth from 4 different hospitals in Galicia (Spain). Parents' consent, and approval by the Galician Ethics Committee of Clinical Research were obtained before the beginning of the research. PR children with further serious complications were excluded from the study. Among the exclusion criteria were babies suffering from cerebral palsy (as diagnosed up until 9 months of age), periventricular leukomalacia (PVL), intraventricular hemorrhage (IVH) greater than grade II, hydrocephalus, encephalopathy, genetic malformations, chromosomal syndromes, metabolic syndromes associated to mental retardation, or important motor or sensorial impairments. Newborn children with Apgar scores below 6 at 5 min were also excluded. The initial sample was recruited at birth, and there were 150 PR children and 49 FT children. When the children were 10 months of age they were assessed on language and communicative development for the first time. The sample at this time comprised 142 PR children, and 49 FT children. The next assessment occasion took place when the children were 22 months of age. At this moment, there were 138 PR children, and 43 FT children. At 30 months the children were assessed again. At this time, the PR sample consisted of 115 children, and the FT sample of 37 children. Descriptive data of the children who initially entered in the study are the following: The group of PR children had a mean GA of 32.60 (SD = 2.43; range 26–36), and the FT group had a mean GA of 39.84 (SD = 1.44, range 37–42). The mean Apgar scores (1 min) of the PR and FT children were similar (t(197) = − .909, p = .365): PR mean = 7.87 SD = 1.43, and FT mean = 8.08, SD = 1.25. Both groups were similar in terms of distribution by gender (X2(1) = .000, p = .997), and mothers' education (X2(6) = 8.66, p = .194). The 138 PR children participating on the next assessment occasion (22 months of age) had a mean GA (and SD) of 32.62 (2.41), a mean BW of 1721.70 (435.36), and a mean Apgar score (first minute) of 7.94 (1.30). As for the 43 FT children, they had a mean GA of 39.70 (1.48), a mean BW of 3373.83 (433.09), and a mean Apgar score (1 min) of 8.13 (1.20). At 30 months of age, the characteristics of the 115 PR and the 37 FT children were very similar to those of the sample at the beginning of the study. The mean GA (and SD) of the PR children was 32.56 (2.49), the mean BW was 1712 g (428), and the mean Apgar score was 7.94 (1.27). For the FT group, the mean GA (and SD) was 39.76 (1.49), the mean BW was 3377 g (443), and the mean Apgar score was 8.16 (1.25). The former data indicate that the children who still continued in the project at 30 months of age had similar characteristics to the original sample. Thus there was no selective mortality. The sample of PR children may be considered as a low risk sample if we consider the Apgar mean score, the inexistence of children with serious medical complications, and the characteristics of their families (mother's education). 2.2. Instruments From the battery of instruments used to assess the children, the following instruments were taken into consideration for the present study. Inventario do Desenvolvemento de Habilidades Comunicativas (IDHC) [36,37], which is the Galician version of the MacArthur–Bates Communicative Development Inventories (CDI) [38]. The form for children between 8 and 15 months (Palabras e Xestos ‘Words and Gestures’) of this

M. Pérez-Pereira et al. / Early Human Development 90 (2014) 649–656

parental inventory, which evaluates different aspects of communicative abilities and first language, was applied when the children were 10 months of age. From the results obtained, the following measurements were taken into account: phrases, word comprehension, word production, and total score for gestures and actions. Possible maximum score for phrases is 27, for word comprehension and production is 384, while the maximum score for total gestures and actions is 65. Total score in gestures and actions is a composite score which is the sum of first communicative gestures, games and routines, actions with objects, pretending to be a parent, and imitating other adult actions, from the IDHC. The following instruments were applied when the children were 22 months of age. Inventario do Desenvolvemento de Habilidades Comunicativas (IDHC) [36,37]. The form for children between 16 and 30 months (Palabras e Oracións ‘Words and Sentences’) was applied. The following sections were taken into consideration in this study: Word production (maximum score 700), MLU of the 3 longest sentences in words (MLU3) and sentence complexity (maximum score 37). The Batelle Developmental Inventory [39] was also used at 22 months of age. This scale measures a child's progress in global cognitive development and in discrete skill sets. The skills assessed by the Batelle scale are adaptive, personal–social, communication, motor, and cognitive. The overall raw score was used for the analysis. The Home Observation for Measurement of the Environment (HOME) [40] in its Spanish version for infants and toddlers was applied [41]. The HOME is designed to measure the quality and quantity of stimulation and support available to a child in the home environment. This inventory is composed of 45 items that are presented as statements to be scored as YES or NO. The HOME includes 6 subscales: Emotional and verbal responsiveness of the primary caregiver, avoidance and restriction of punishment, organization of the physical and temporal environment, provision of appropriate play materials, parental involvement with the child, and opportunities for variety in daily stimulation. Higher total HOME scores indicate a more enriched home environment. Finally, when the children were 30 months of age, their linguistic development was again assessed through the Inventario do Desenvolvemento de Habilidades Comunicativas (IDHC) [36,37]. The form for children between 16 and 30 months (Palabras e Oracións ‘Words and Sentences’) of this parental inventory was applied, and the same sections as reported for 22 months were taken into consideration. 2.3. Procedure At 15 days of age (corrected age for prematurity) the children were evaluated with the Neonatal Behavioral Assessment Scale (NBAS) [42]. The data obtained with the NBAS have already been reported elsewhere [43], and will not be reported here. At that time the mothers of the children participated in a long interview, which permitted us to gather data on the family environment and health of the children. Some of them were used in the analyses performed. When the children were 10 months old, all of them were assessed with different instruments, see [20,21]. For the purposes of the present study, we will take into account part of the data obtained with the Galician version of the CDI [38] or Inventario do Desenvolvemento de Habilidades Comunicativas (IDHC Word and Gestures) [36,37]. When the children were 22 months of age (±7 days) (corrected age for PR children), they were assessed with different instruments. The children's cognitive development (Batelle inventory), and the quality of their home environments (HOME scale), were assessed by a trained psychologist, who visited their homes. Their linguistic abilities were assessed through the IDHC Word and Sentences, which the mothers filled in a few days before the visit, or sent within the first week after the visit. (The mother of one PR infant did not send the parental inventories on time and the data were not considered.)

651

Finally, at 30 months of age (±15 days) the children were assessed through the Galician CDI (IDHC Words and Sentences) following the same procedure as at 22 months of age. 2.4. Analyses performed 1) First, t tests were performed to assess differences in the IDHC Word and Gestures and Word and Sentences scores between the group of PR children (GA below 37 weeks) and the group of FT children (GA above 36 weeks). 2) Secondly, all the children were additionally grouped into the following GA groups to assess whether more subtle differences in GA may have an effect on the IDHC Word and Sentences: 1) ≤31 weeks, 2) 32 to 33 weeks, 3) 34 to 36 weeks, and 4) ≥37 weeks of GA (the FT group). ANOVA analyses with F statistics (or the Brown–Forsythe test when the homogeneity of variances' condition was not fulfilled) were performed to compare the 4 groups' results. Post-hoc Tukey's HSD or Games Howell tests were performed to analyze differences between groups of different GAs. Since there were only 3 children in the usual classification of EPR children (GA b 28 weeks), they were integrated into the GA group (b31 weeks), which was made up by EPR and VPR children. For analyses 1 and 2, GA was considered as an independent variable (IV) and the IDHC scores as dependent variables (DV). 3) Finally, a series of hierarchical regression analyses were performed to test which predictors may contribute to explain language scores at 10, 22 and 30 months of age. The variables used as predictors were chosen because of their theoretical significance according to previous studies, and because a significant correlation had been found between them and word production scores. There were three types of predictive factors: biomedical (GA, stay in intensive care unit (ICU), and gender), environmental (maternal education and the HOME score), or personal traits of the children (previous language, communicative or cognitive development). In order to compute stay in the intensive care unit (ICU), all the children (PR and FT) were divided into three groups depending on the number of days they stayed in the ICU: 1) no stay; 2) less than 15 days; 3) 15 days or more. Stay in the ICU may be considered an index of the children's initial health risk, but always taking into consideration that children with serious risks were excluded. To compute mother's education, the children were divided into three groups: 1) basic education (below or up to age 16), 2) high school, technical school education, and 3) university degree. The characteristics of the sample regarding distribution by gender, stay in NICU, and maternal education are shown in Table 1 for the PR and the FT groups at the beginning of the longitudinal study and at 30 months of age. A) The first three hierarchical regression analyses performed took as dependent variable (DV) word production at 10, 22, and 30 months of age. Predictive variables were the same for the

Table 1 Demographic information of the PR and FT sample at birth and at 30 months of age: Frequency (percentage). Variable

Values

Initial PR group

Initial FT group

PR group at 30 m

FT group at 30 m

Gender

Male Female No stay 1–15 days N15 days Basic High school and technical school University degree

78 (52.1) 72 (47.9) 42 (28) 61 (40.7) 47 (31.3) 38 (25.3) 59 (39.3)

25 (51) 24 (49) 46 (93.9) 2 (4.1) 1 (2.0) 19 (38.8) 13 (26.5)

65 (56.5) 50 (43.5) 30 (26.1) 48 (41.7) 37 (32.2) 26 (22.6) 52 (45.2)

19 (51.4) 18 (48.6) 34 (91.9) 2 (5.4) 1 (2.7) 14 (37.8) 10 (27.0)

53 (35.3)

17 (34.7)

37 (32.2)

13 (35.1)

Stay in NICU

Maternal education

652

M. Pérez-Pereira et al. / Early Human Development 90 (2014) 649–656

three analyses, in order to compare the effect of those variables on word production over time. Two accumulative models were defined: Model 1: Word understanding score at 10 months of age. Total score in gestures and actions at 10 months of age. Model 2: Word understanding at 10 months of age. Total score in gestures and actions at 10 months of age. Gender Maternal education (3 groups). Gestational age in weeks. Stay in NICU (3 groups). B) In the light of the results obtained in the three former analyses, additional regression analyses were performed in order to identify the factors which most predicted language scores (word production and sentence complexity) at 30 months of age. Dependent variables were word production at 30 months, and sentence complexity at 30 months of age, respectively in two separate analyses. Predictive variables were the same for both analyses, and two accumulative models were defined: Model 1: Word production score at 22 months of age. Batelle total raw score at 22 months of age. Model 2: Word production score at 22 months of age. Batelle total raw score at 22 months of age. Gender Maternal education (3 groups). HOME total score. Gestational age in weeks. Stay in NICU (3 groups). The program IBM SPSS Statistics 20.0 was used for the analyses. 3. Results 1) The mean and standard deviation scores with the performance of the PR and the FT groups at 10, 22 and 30 months of age in the IDHC measures are shown in Table 2. Table 2 also shows the results of the comparisons between the PR and the FT groups (t test). The t tests carried out indicate that there were no significant differences between the two groups in any of the IDHC measures taken at any age. In relation to the norms of the IDHC [36,37], the mean raw scores obtained by the PR and the FT groups in all the IDHC sections at 10, 22 or

Table 2 IDHC scores of PR and FT children at 10, 22 and 30 months of age, and the mean comparison results. Age

IDHC scores

10 months Phrases

Group N

PR FT Word understanding PR FT Word production PR FT Total gestures and PR actions FT 22 months Word production PR FT MLU3 PR FT Sentence complexity PR FT 30 months Word production PR FT MLU3 PR FT Sentence complexity PR FT

142 49 142 49 142 49 142 49 137 43 137 43 137 43 115 37 115 37 115 37

Mean

SD

t

p

13.70 14.45 79.17 71.86 5.30 6.39 21.13 23.60 158.65 173.77 2.65 2.83 2.53 2.35 419.53 411.86 6.97 7.62 20.92 20.49

6.49 −.693 .489 6.44 74.14 .625 .533 58.88 7.70 −.510 .610 21.96 9.27 −1.540 .125 10.15 147.28 −.597 .552 137.18 2.11 −.529 .597 1.63 4.96 .218 .827 4.33 175.44 .229 .819 173.75 4.48 −.699 .485 5.58 14.35 .158 .874 13.32

30 months of age are well within the normal variation, and correspond to percentile scores between 25 and 75, and most of them were around percentile 50. When the children were sorted into more subtle GA groups, the descriptive results they obtained are given in Table 3 for 10, 22 and 30 months of age. This table also shows the results of the one way ANOVA performed to compare the results of the different GA groups. As can be observed, the results obtained by the 4 GA groups were very similar, and no significant difference was observed in any case. The results obtained for cognitive development and quality of home environment (measured at 22 months of age), two of the predictive variables introduced in the regression models, were the following for the PR and the FT groups. The mean total score (and SD) in the Batelle scale for the FT group was 260.00 (22.28), and for the PR group was 245.96 (21.63). The t test mean comparison indicates that FT children obtained significantly higher scores than the PR children (t (179) = −3.69, p = .001). The mean raw scores obtained by the PR and the FT groups were, however, within the standard deviation for the Spanish normative sample of the Batelle inventory of the corresponding age (M = 247.29, SD = 23.57), and correspond to percentile 50 (IQ = 100) and percentile 70 (IQ = 108), respectively [39]. The mean (and SD) total score in the HOME scale for the FT group was 38.70 (3.97), and for the PR group was 38.23 (4.32). There was no significant difference between groups (t = − .625, p = .533). The mean results obtained by the two groups in the Home scale fit in the fourth quartile, indicating relatively good quality of the children's environments [41]. To check the possible effect of cognitive development on the results obtained in the former ANOVAs, we introduced the Batelle total score as covariable in a series of ANCOVAs where the independent variable was the GA of the four groups, and the dependent variables were the results obtained in the different measurements of the IDHC. The results paralleled those obtained before with the ANOVAs, indicating that the effect of the independent variable remains unchanged. Therefore, although there were relationships which reached significance between the covariable and the dependent variables, these do not affect the relationship between the independent variable (GA groups) and the dependent variables (scores obtained in the IDHC measures). 2) Finally, different hierarchical regression analyses were performed in order to identify the factors which predict linguistic development. As commented on above, a series of variables were taken as predictive factors for the regression analysis. The first hierarchical regression analyses aimed to obtain those factors predictive of word production, usually considered a good measure of language development, at 10, 22 and 30 months of age. The results obtained are shown in Table 4 for each age. The most important results indicate that word understanding and gestures and actions at 10 months are good predictors of language production at 10 months, and even at 22 months of age. In the first case Model 1 predicts 25.4% of variance of word production at 10 months, and 7.6% at 22 months of age. The inclusion of other variables in Model 2 hardly increments the variance explained (3.4% at 10 months and 4.2% at 22 months of age). However neither word comprehension nor gestures and actions seem to predict the results obtained in word production at 30 months of age, and the introduction of the other variables in Model 2 does not change this fact. No other variable had a significant predictive effect. Given that neither gestures nor word comprehension has any significant effect on language measures at 30 months, the effects of other variables were tested in subsequent hierarchical regression analyses. In the search for predictors of language scores at 30 months of age, new regression analyses were performed. First, cognitive development measured at 22 months of age through the Batelle scale and vocabulary production at this same age measured with the IDHC were entered in

M. Pérez-Pereira et al. / Early Human Development 90 (2014) 649–656

653

Table 3 Mean scores (and SD) of children with different GAs on the IDHC at 10, 22 and 30 months of age, and ANOVA results. GA ≤31 weeks

GA 32–33 weeks

GA 34–36 weeks

GA ≥37 weeks

F

p

N Phrases Word understanding Word production Total gestures

45 12.78 (5.82) 73.31 (73.39) 4.51 (7.40) 20.13 (8.08)

36 12.25 (6.87) 71.50 (70.05) 6.00 (7.67) 23.36 (10.60)

61 15.16 (6.61) 88.02 (77.23) 5.46 (8,00) 20.56 (9.20)

49 14.45 (6.44) 71.86 (58.88) 6.39 (21.96) 23.60 (10.15)

2.142 .694 .179 1.697

.096 .557 .910 .169

N Word production MLU3 Sentence complexity

43 140.93 (137.87) 2.19 (1.38) 1.19 (2.47)

36 154.25 (130.12) 3.13 (3.23) 2.83 (5.62)

58 174.52 (163.84) 2.67 (1.62) 3.34 (5.72)

43 173.77 (137.18) 2.83 (1.63) 2.35 (4.33)

.573 1.533 1.815a

.633 .208 .147a

N Word production MLU3 Sentence complexity

35 417.09 (182.58) 7.09 (3.09) 22.03 (13.44)

32 432.63 (147.50) 8.40 (6.59) 21.65 (14.13)

47 421.36 (181.72) 6.25 (3.15) 20.04 (15.20)

36 411.86 (173.75) 7.62 (5.58) 20.49 (13.32)

.087 1.210a .165

.967 .311a .920

Age 10 months of age

22 months of age

30 months of age

a

Brown–Forsythe test.

Model 1. As Table 5 shows (upper part), nearly 40% of variance of word production at 30 months of age was explained by these variables. When maternal education, stay in NICU, gender, HOME total score and gestational age were added (Model 2) the variance explained for word production at 30 months barely increased 2%, and none of them seemed to have a significant effect. Finally we tried to test whether previous cognitive development and word production at 22 months of age were also good predictors of other linguistic measures, this time more closely linked to morphosyntactic development. In this case the dependent variable was sentence complexity at 30 months of age. The results (which are shown in Table 5, lower part) paralleled those obtained before with word production as the dependent variable. The variance explained by cognitive development and word production at 22 months reached nearly 37%. When we additionally entered maternal education, stay in NICU, gender, HOME total score and GA in Model 2, the variance explained increased 6.2% and maternal education had a significant effect (p b .01), which means that

mothers with a higher level of studies tended to have children with better scores in sentence complexity. 4. Discussion The results found indicate that the PR children do not show significant delays in relation to the FT children in any measures obtained at 10 months of age with the IDHC Words and Gestures. These results contradict other findings obtained with very or extremely preterm infants, who obtained lower scores in gesture development [34], and also in word comprehension [30,44] at a similar age, but at the same time are in agreement with the studies which found that PR children do not show lower development than FT children in gestures and actions [30], word understanding [14,19] and word production [13–15,19,30] at around 12 months of age. Similar results were obtained for vocabulary development or any of the grammatical development scores reported (the mean length of the

Table 4 Hierarchical regression analysis: Gestures and actions, word comprehension and biological and environmental factors as predictors of word production at different ages. Dependent Variable

Predictors

R2

Standardized β

F

df

p

Word production 10 months

Model 1: Gestures and actions Word understand Model 2: Gestures and actions Word understand GA Stay in NICU Gender Maternal education Model 1: Gestures and actions Word understand Model 2: Gestures and actions Word understand GA Stay in NICU Gender Maternal education Model 1: Gestures and actions Word understand Model 2: Gestures and actions Word understand GA Stay in NICU Gender Maternal education

.254

.315 .271

31.691

2,186

.000

.288

.315 .266 .001 −.012 −.069 −.171

12.243

6,182

.000

.076

.301 −.075

7.173

2,175

.001

.118

.267 −.068 .086 −.044 −.127 .144

3.797

6,171

.001

.011

.034 .085

.795

2,146

.454

.050

.010 .100 −.029 −.067 −.133 .140

1.244

6,142

.288

Word Production 10 months

Word production 22 months

Word production 22 months

Word production 30 months

Word production 30 months

654

M. Pérez-Pereira et al. / Early Human Development 90 (2014) 649–656

Table 5 Hierarchical regression analysis: Cognitive development, word production at 22 months and biological and environmental factors as predictors of word production, and sentence complexity at 30 months of age. Dependent Variable

Predictors

R2

standardized β

F

df

p

Word production 30 months

Model 1: Total Batelle Word product. 22 m Model 2: Total Batelle Word product. 22 m Total HOME GA Maternal education Gender Stay in NICU Model 1: Total Batelle Word product. 22 m Model 2: Total Batelle Word product. 22 m Total HOME GA Maternal education Gender Stay in NICU

.392

.227 .469

47.155

2,146

.000

.412

.263 .455 .010 −.139 .040 .013 −.014

14.088

7,141

.000

.369

.098 .546

41.157

2,141

.000

.431

.229 .523 −.120 −.024 .182 .117 .089

14.738

7,136

.000

Word production 30 months

Sentence complexity 30 months

Sentence complexity 30 months

3 longest utterances, or sentence complexity), as measured through the IDHC Words and Sentences at 22 and 30 months of age. These results do not coincide with a series of studies which found that preterm children around 24 or 30 months of age were delayed in their language abilities (lexical as well as grammatical) in relation to full term children [5–7,10–15]. Other studies, however, did not find significant differences between these populations [7,9,16–19], when corrected age for prematurity is applied. The discrepancy between the abovementioned studies is difficult to resolve, and the results are complex to explain. In some longitudinal studies discrepant results were even found at different ages after 24 months of age [9]. In most cases the studies used similar assessment instruments (the Mac-Arthur–Bates scales—CDI), with Menyuk being the only exception [16]. For this reason the variability in the findings cannot be explained by the use of different measurement instruments. One important source of variability in the results found may be related with the characteristics of the samples. In many of these studies which found that PR children were delayed in their linguistic development in relation to FT children, the PR children were of very low birth weight or very preterm children [6,7,13,15]. In other cases, when the sample of PR children included a wider range of GAs, differences between PR and FT children were higher when very PR children and VLBW children were compared to FT children [5,6]. The fact that our sample was made up of PR children without additional impairments (which was not the case in many other studies) and who were not at high risk – as indicated by the Apgar score, the relatively low stay at ICU, and their average mother's educational level – could probably explain the absence of major differences in linguistic development. The use of corrected age for prematurity may also favor the results obtained by the PR children [44], although adjustment for age is a current practice when studying PR children under 2 years of age. In any case, this compensatory effect of age correction should theoretically become less important with age, and the benefit of PR over FT children should become less obvious as children get older. The results we found, however, do not point to this hypothesis, since at 30 months of age the PR children obtained relatively better results than on previous assessment occasions when compared to the FT children. We also carried out more detailed comparisons of linguistic development by GA groups, with a similar result. No significant differences were found in lexical or grammatical development between any GA groups, even when comparisons were performed between the group of GA

≥ 37 weeks (FT children) and those EPR and VPR children with GA ≤ 31 weeks. Therefore the results we found are different to those found by other authors [5,6] who found that differences were greater when comparisons were made between FT children and VPR or EPR children. Again, the characteristics of the sample of preterm children of our study may be the explanation for these results. Therefore, the results we obtained comparing groups of PR children with different GA and FT children seem to point to a non-significant effect of GA on language development, concerning both lexical and grammatical development. The results obtained with hierarchical regression analyses, which will be commented on below, confirm this hypothesis. In relation to the results obtained through the regression analyses, the first results obtained seem to indicate that gestures and word comprehension at 10 months lose their predictive power as explanatory factors of later linguistic development after 24 months of age. Previous studies found that gestures measured at around 12 months [10,34] do have a predictive value on later language development. However, the latest measures of language taken in those studies were at 2.0 years of age while gesture measures were taken at 12 months [30], or at 9, 12 and 15 months of age [34]. In our study, when language measures were taken a few months later (30 months of age), the effect that existed on language scores at 10 and 22 months of age disappeared. Setting apart the case of word production (DV) measured at the same age as the predictive variables (10 months), the former results indicate that the predictive capacity of gestures on later linguistic measures fades when distance between measures increases. Findings like these are not unusual. In fact, Stolt et al. [34] practically found no significant correlations between gesture measures at 9 months and later linguistic measures taken at 24 months of age when the distance between measurement points increased. Similarly no significant predictive value of gestures at 9 months on language at 24 months of age was found, although significant values were found when gestures were assessed at the ages of 12 and 15 months [34]. Considerations of the same type can be made on word comprehension at 10 months as a predictor of (later) language production. Its predictive effect seems to progressively dissolve with time and disappears at 30 months of age. Stolt and colleagues [34] also observed that the capacity of prediction of poor language performance at 2.0 was higher when word comprehension was measured closer in time, at 1.3, but descended progressively as word comprehension measurements were further apart in time, at 1.0 and 0.9. At this earlier time, word comprehension was a weak predictor of poor language performance at 2.0.

M. Pérez-Pereira et al. / Early Human Development 90 (2014) 649–656

Therefore, language and communicative measures taken at 10 months do not predict language outcomes at 30 months, which points to the idea that measures of first linguistic and communicative abilities have a limited capacity of prediction, which is reduced to language measures taken only a few months later and does not seem to extend beyond 24 months of age. In the search for predictors of later language development, additional regression analyses confirmed that cognitive measures and word production at 22 months are good predictors of lexical and grammatical development at 30 months. The percentage of variance explained was high for those measures (around 40% and 37% in word production and sentence complexity, respectively). These results are similar to those obtained by Sansavini et al. [9] with children between 2.6 and 3.6 years of age. In addition the results of the present study indicate that the amount of variance explained increases a few points when other variables are included in the model (Model 2), particularly in the case of sentence complexity. In this case the introduction of those variables increases the capacity of explanation of the model by more than 6 points, explaining more than 43% of the variance in sentence complexity. Particularly interesting is the effect of maternal education which reaches a significant level; this finding suggests that the effect of maternal education might increase as children grow older, reinforcing the hypothesis drawn from other studies [26,30,31,45]. It is important to note that variables which were found to have an effect on linguistic measures in other studies do not seem to have any significant effect in our study. This is the case with gestational age, in particular, as well as with stay in NICU, maternal education and gender. The case of gestational age, which does not seem to have any significant predictive effect on language scores at 30 months of age, or even at 22 or 10 months of age, is particularly noticeable. These results may indicate that preterm birth on its own, in the absence of other serious biomedical complications which increase the risk factor, does not necessarily determine delays in language development [2,28]. Future studies should pay more attention to the criteria of selection of the samples in order to clarify this point. Finally, gender does not seem to have a relevant effect on language development at the ages studied, nor do other biomedical (stay in NICU) or environmental (quality of home environment) variables seem to affect language outcomes. The case of gender is particularly noteworthy, since previous studies had found that girls obtained significantly better results than boys, and gender was found to be a predictor of language outcomes [10,11,13,17]. Probably, gender increases its effects when associated to additional risk factors, such as bronchopulmonary dysplasia or extremely preterm birth. For this reason no effect of gender was found in our sample during the time period studied. In addition, slight variations in the effect of gender were also found across language communities [46]. 5. Conclusions The results found indicate that communicative, lexical and grammatical development of PR children is not significantly delayed in relation to that of their FT peers at the ages studied. Differences were not even found when comparing the most premature children of the sample to the FT children. The results coming from group comparisons and from hierarchical regression analyses indicate that gestational age does not significantly affect language development from 10 to 30 months of age in this low risk sample of PR children. The effect of gestures and early word comprehension on later language development seems to fade with time, and the most important factors which predict language development at 30 months of age turn out to be previous cognitive and linguistic development. The possible explanation for the controversial results we have found is probably linked to the characteristics of the sample, which was free of PR children with serious medical problems and could be described as a low risk sample by virtue of its other features. Given the apparent

655

relevance of medical complications on developmental outcomes, factors leading to decreased medical complication incidence deserve further study. Future research should carefully consider the sample selection criteria in an attempt to disentangle the controversial role of prematurity on language acquisition. Conflict of interest statement None declared. Acknowledgments This research was funded by the Ministerio de Ciencia e Innovación of the Spanish Government (grant PSI2008-03905 to the first author), and the Xunta de Galicia (INCITE) (grant PGIDIT07PXIB211044PR to the first author). References [1] Guarini A, Sansavini A, Fabbri C, Alessandroni R, Faldella G, Karmiloff-Smith A. Reconsidering the impact of preterm birth on language outcome. Early Hum Dev 2009;85:639–45. [2] Singer LT, Siegel AC, Lewis B, Hawkins S, Yamashita T, Baley J. Preschool language outcomes of children with history of bronchopulmonary dysplasia and very low birth weight. Dev Behav Pediatr 2001;22(1):19–26 [doi: 0196-206X/00/2201-0019]. [3] Johansson S, Cnattigius S. Epidemiology of preterm birth. In: Nosarti C, Murray RM, Hack M, editors. Neurodevelopmental outcomes of preterm birth from childhood to adult life. Cambridge: Cambridge University Press; 2010. p. 1–16. [4] Magill-Evans J, Harrison MJ. Parent–child interactions and development of toddlers born preterm. West J Nurs Res 1999;21:292–307. [5] Kern S, Gayraud F. Influence of preterm birth on early lexical and grammatical acquisition. First Lang 2007;27(2):159–73. [6] Foster-Cohen S, Edgin JO, Champion PR, Woodward LJ. Early delayed language development in very preterm infants: evidence from the MacArthur–Bates CDI. J Child Lang 2007;34:655–75. [7] Stolt S, Lehtonen L, Haataja L, Lapenleimu H. The language used in early mother– child interaction by very-low-birth-weight children, with a focus on the emergence of grammar. Revista de logopedia, foniatría y audiología 2011;31(3):115–24. [8] Stolt S, Lehtonen L, Haataja L, Lapinleimu H. The development and predictive value of early vocalizations in prematurely born very-low-birth-weight children: a longitudinal study. Clin Linguist Phon 2012;26(5):414–27. http://dx.doi.org/10. 3109/02699206.2011.648365. [9] Sansavini A, Savini S, Guarini A, Broccoli S, Alessandroni R, Faldella G. The effect of gestational age on developmental outcomes: a longitudinal study in the first two years of life. Child: Care, Health & Development 2010;37:26–36. [10] Sansavini A, Guarini A, Savini S. Linguistic and cognitive delays in very preterm infants at 2 years: general or specific delays? Revista de Logopedia Foniatría y Audiología 2011;31(3):133–47. [11] Schults A, Tulviste T, Haan E. Early vocabulary in full term and preterm Estonian children. Early Hum Dev 2013;89(9):721–6. http://dx.doi.org/10.1016/j.earlhumdev. 2013.05.004. [12] Foster-Cohen SH, Friesen MD, Champion PR, Woodward LJ. High prevalence/low severity language delay in preschool children born very preterm. J Dev Behav Pediatr 2010;31(8):658–67. [13] Bosch L, Ramon Casas M, Solé J, Nícar L, Iriondo Sanz M. Desarrollo léxico en el prematuro: medidas del vocabulario expresivo en el segundo año de vida. Revista de logopedia, foniatría y audiología 2011;31(3):169–79. [14] Jansson-Verkasalo E, Ruusuvirta T, Huotilainen M, Alku P, Kushnerenko E, Suominen K, et al. Atypical perceptual narrowing in prematurely born infants is associated with compromised language acquisition at 2 years of age. BMC Neurosci 2010;11(88) [doi: 8810.1186/1471-2202-11-88]. [15] Stolt S, Haataja L, Lapinleimu H, Lehtonen L. The early lexical development and its predictive value to language skills at 2 years in very-low-birth-weight children. J Commun Disord 2009;42:107–23. [16] Menyuk P. Patterns of early lexical and cognitive development in premature and fullterm infants. J Speech Lang Hear Res 1991;34:88–94. [17] Sansavini A, Guarini A, Alessandroni R, Faldella G, Giovanelli G, Salvioli G. Early relations between lexical and grammatical development in very immature Italian preterms. J Child Lang 2006;33:199–216. [18] Stolt S, Klippi A, Launonen K, Munck P, Lehtonen L, Lapinleimu H, et al. Size and composition of the lexicon in prematurely born very-low-birth-weight and full-term Finnish children at two years of age. J Child Lang 2007;34:283–310. [19] Cattani A, Bonifacio S, Fertz M, Iverson JM, Zocconi E, Caselli MC. Communicative and linguistic development in preterm children: a longitudinal study from 12 to 24 months. Int J Lang Commun Disord 2010;45:162–73. [20] Pérez-Pereira M, Fernández P, Díaz C, Resches M, Gómez-Taibo ML, Peralbo M. Desarrollo lingüístico y comunicativo temprano de niños prematuros. Revista de logopedia, foniatría y audiología 2011;31(3):148–59.

656

M. Pérez-Pereira et al. / Early Human Development 90 (2014) 649–656

[21] Pérez-Pereira M, Fernández P, Resches M, Gómez-Taibo ML. Determinants of early language and communication in preterm and fullterm infants: a comparative study. Enfance 2013;1:59–76. [22] Casiro OG, Moddemann DM, Stanwick RS, Cheang MS. The natural history and predictive value of early language delays in very-low-birth-weight infants. Early Hum Dev 1991;26(1):45–50. http://dx.doi.org/10.1016/0378-3782(91)90042-2. [23] Cusson RA. Factors influencing language development in preterm infants. Jognn-J Obstet Gynecol Neonatal Nurs 2003;32(3):402–9. http://dx.doi.org/10.1177/ 0884217503253530. [24] Jennische M, Sedin G. Speech and language skills in children who required neonatal intensive care: evaluation at 6.5 y of age based on interviews with parents. Acta Paediatr 1999;88(9):975–82. [25] Marston L, Peacock JL, Calvert SA, Greenough A, Marlow N. Factors affecting vocabulary acquisition at age 2 in children born between 23 and 28 weeks' gestation. Dev Med Child Neurol 2007;49(8):591–6. [26] Molfese VJ, Holcomb L, Helwig S. Biomedical and social–environmental influences on cognitive and verbal abilities in children 1 to 3 years of age. Int J Behav Dev 1994;17(2):271–87. http://dx.doi.org/10.1177/016502549401700203. [27] Vohr B, Wright LL, Dusick AM, Mele L, Verter J, Steichen JJ, et al. Neurodevelopmental and functional outcomes of extremely low birth weight infants in the National Institute of Health and Human Development Neonatal Research Network, 1993–1994. Pediatrics 2000;105(6):1216–26. [28] Lewis BA, Singer LT, Fulton S, Salvator A, Short EJ, Klein N, et al. Speech and language outcomes of children with bronchopulmonary dysplasia. [doi: DOI: 10.1016/S00219924(02)00085-0]. J Commun Disord 2002;35(5):393–406. [29] Baron IS, Ahronovich MD, Erickson K, Larson JCG, Litman FR. Age-appropriate early school age neurobehavioral outcomes of extremely preterm birth without severe intraventricular hemorrhage: a single center experience. Early Hum Dev 2009; 85(3):191–6. http://dx.doi.org/10.1016/j.earlhumdev.2008.09.411. [30] Sansavini A, Guarini A, Savini S, Broccoli S, Justice L, Alessandroni R, et al. Longitudinal trajectories of gestural and linguistic abilities in very preterm infants in the second year of life. Neuropsychologia 2011;49(13):3677–88. http://dx.doi.org/10.1016/ j.neuropsychologia.2011.09.023. [31] Sansavini A, Guarini A, Alessandroni R, Faldella G, Giovanelli G, Salvioli G. Are early grammatical and phonological working memory abilities affected by preterm birth? J Commun Disord 2007;40(3):239–56. http://dx.doi.org/10.1016/j.jcomdis. 2006.06.009. [32] Murray D, Yingling JL. Competence in language at 24 months: relations with attachment security and home stimulation. J Genet Psychol 2000;161(2):133–40.

[33] Weisglaskuperus N, Baerts W, Degraaf MA, Vanzanten GA, Sauer PJJ. Hearing and language in preschool very-low-birth-weight children. Int J Pediatr Otorhinolaryngol 1993;26(2):129–40. http://dx.doi.org/10.1016/0165-5876(93) 90018-x. [34] Stolt S, Makila AM, Matomaki J, Lehtonen L, Lapinleimu H, Haataja L. The development and predictive value of gestures in very-low-birth-weight children: a longitudinal study. Int J Speech Lang Pathol 2013;16(2):121–31. http://dx.doi.org/10.3109/ 17549507.2013.794861. [35] Salt A, Redshaw M. Neurodevelopmental follow up after preterm birth: follow up after two years. Early Hum Dev 2006;82:185–97. [36] Pérez Pereira M, García Soto XR. El diagnóstico del desarrollo comunicativo en la primera infancia: adaptación de las escalas MacArthur al gallego. Psicothema 2003;15(3):352–61. [37] Pérez-Pereira M, Resches M. Concurrent and predictive validity of the Galician CDI. J Child Lang 2011;38(1):121–40. [38] Fenson L, Marchman VA, Thal DJ, Dale PS, Reznick JS, Bates E. MacArthur–Bates Communicative Development Inventories (CDIs). 2nd ed. Baltimore: Paul H. Brookes; 2007. [39] Newborg J, Stock JR, Wnek L. Batelle, Inventario de Desarrollo. Madrid: TEA ediciones; 1996. [40] Caldwell BM, Bradley RH. Home observation for measurement of the environment. Revised ed. Little Rock: University of Arkansas at Little Rock; 1984. [41] Moreno C, Escala HOME. (Home observation for measurement of the environment; Caldwell y Bradley, 1984) y C.V.C. Cuestionario de la Vida Cotidiana. Sevilla: University of Sevilla; 1992. [42] Brazelton TB, Nugent JK. Escala para la evaluación del comportamiento neonatal. Paidós: Barcelona; 1997. [43] Perez-Pereira M, Fernández P, Gómez-Taibo ML, González ML, Trisac JL, Casares J, Domínguez M. Neurobehavioral development of preterm and full term children: biomedical and environmental influences. Early Hum Dev 2013;89:401–9. [44] Wilson SL, Cradock MM. Review: accounting for prematurity in developmental assessment and the use of age-adjusted scores. J Pediatr Psychol 2004;29(8):641–9. [45] Hoff E. How social contexts support and shape language development. Dev Rev 2006;26(1):55–88. [46] Eriksson M, Marschik PB, Tulviste T, Almgren M, Pérez-Pereira M, Wehberg S, et al. Differences between girls and boys in emerging language skills: evidence from 10 language communities. Br J Dev Psychol 2012;30(2):326–43. http://dx.doi.org/10. 1111/j.2044-835X.2011.02042.x.