Reading and spelling skills of prematurely born children in light of the underlying cognitive factors

Premature birth (before the 37 week of gestation) is the most common perinatal risk; therefore, the heightened research interest in the development of preterm children is not surprising. In spite of the ever-growing bulk of research evidence, the picture is far from being clear. As far as the long-term outcome is concerned, the majority of the studies reported IQ’s of moderate-risk preterm children significantly lower as compared to those of full-term comparison groups (Aylward 2002; Hadders-Algra 2005; Kalmár 1996; Rose et al. 2005), mostly falling into the lower third of the average zone (Breslau and Chilcoat 1996; Breslau et al. 2001; Rose et al. 2011). Several researchers found cognitive deficits in various domains associated with structural disorders of the brain (Inder et al. 2005; Bradley et al. 2000; Skranes et al. 2012). School-age preterm children even with normal IQs and without neurological impairment often display cognitive dysfunctions mainly related to visual processing and executive functions (Larsson et al. 2005).

Prematurity is a serious risk factor for learning difficulties (Aarnoudse-Moens et al. 2006; Breslau et al. 2001; Saigal et al. 1991). Within the academic skills reading is likely to have the greatest impact on the prospects of the students. Reading deficits hamper school performance across various domains and, hence, often lead to the need of special education as well as to failures in the adult life (Kovachy et al. 2014; Samuelsson et al. 2006). Studying the reading skills in the risk populations is therefore a research issue of special importance.

Reading has two fundamental components: decoding (single-word reading) and reading comprehension, i.e., to derive meanings from and form interpretations of written words and sentences. To acquire reading, first one should learn the code for representing speech as a series of visual symbols. It is a process of matching visual symbols (in many languages letters) to units of sound. Decoding stems from primary linguistic skills such as phonological awareness and alphabet knowledge. Efficient reading also requires the recognition of the orthographic patterns of whole words not only accurately but also fluently, that is, automatically. The acquisition of this skill which is a basis of comprehension is a further step in learning to read. Comprehension is more complex as it requires the integration of these linguistic skills with higher-order cognitive processes, e.g., working memory (Kovachy et al. 2014; Shaywitz and Shaywitz 2008; Ziegler and Goswami 2005).

A recent meta-analysis by Kovachy et al. (2014) covering studies which compared the reading abilities in school-age children who were born preterm to those in full-term comparison groups demonstrated that the performances of the preterm children were significantly worse on both major components of reading, i.e., on decoding and reading comprehension. The preterm samples of the studies included in the meta-analysis were rather heterogeneous. The upper limit of gestational age was 32 weeks, and the mean birthweights varied between 740 and 1256 g. Gestational age was significantly associated with the reading performances in the preterms: Lower gestational ages increased the disadvantages. However, the group differences remained the same after the exclusion of children with major disabilities or intellectual impairments as well as after controlling for the inequalities of the socio-economic status. In the above meta-analysis the effect of birth weight was not checked, although it is generally acknowledged to contribute to the degree of risk in prematurely born children. A considerable proportion of research limits their samples to very low birthweight (< 1500 g, VLBW) preterms. Two research teams (Samuelsson et al. 2006; Takeuchi et al. 2016) which investigated the reading abilities in elementary school-age VLBW children with very different native languages reported remarkably similar findings: Roughly one-third of both the Swedish and the Japanese VLBW children with normal intelligence had reading difficulties. The 9-year-old Swedish VLBW children were more disadvantaged in the orthographic (spelling-based) reading skills than in the phonological (sound-based) skills (Samuelsson et al. 2006). Language functions such as vocabulary and receptive language were found relatively intact in VLBW children (Aylward 2002; Ment et al. 2003) which suggest that the mechanism of reading difficulty in VLBW preterms differs from that in full-term children. A potential complication of VLBW preterm birth is a periventricular white matter injury which may compromise the visual cognition pathway (Downie et al. 2003). Problems in visual cognition, in addition to phonological weakness, might play an important role in the reading difficulties in VLBW children (Takeuchi et al. 2016).

However, the literature on the reading abilities of preterm children is not consistent. A number of studies did not find the reading performances of the preterm groups significantly different from those of the full-term comparison groups (Kesler et al. 2004; Lee et al. 2017; Takeuchi et al. 2016; Taylor et al. 2011). It is also notable that the reading deficits in the Swedish VLBW children followed up by Samuelsson et al. (2006) did not persist into adolescence. By 15 years of age the VLBW subjects free of identified brain insults caught up with their non-risk counterparts on all reading measures, and the comparison between the total VLBW sample and the full-term group remained significant on one measure only.

The specific characteristics of the different languages are likely to influence the learning processes in the acquisition of reading (Ziegler and Goswami 2005); therefore, the research in foreign languages cannot provide full guidelines for the measurement of the reading skills and the interpretation of the potential faults and deficits.

The aim of our study was to investigate certain reading and spelling skills of school-age prematurely born Hungarian children. Based on the majority of research evidence available in the literature our hypothesis was that the reading and spelling performances of the preterm children will be poorer than those of the same-age non-risk, problem-free children. According to our knowledge no research has so far compared the development of reading and spelling skills in preterm children and that of children diagnosed with dyslexia. We expected that the reading deficits in the preterm children will not be as severe as those in the dyslexics. We were also interested in the cognitive mechanisms underlying the reading and spelling performances and whether those mechanisms differ depending on the children’s birth status (full-term—preterm) and reading disability diagnosis.

The target group consisted of school-age children (n = 23) who were born preterm with very low birthweights (990–1350 g, mean: 1211.7 g), at 28–33 weeks (mean: 30 weeks) of gestation and cared for in the Neonatology Ward of the Péterfy Sándor Hospital in Budapest. They had no neurological impairments and were considered at moderate risk. The preterm children were compared to two groups of full-term children. Thirty of the full-term children were dyslexics, and 27 of them were good readers. Children with seizures, attention disorders (scoring > 70 on the attention problem subscale of the CBCL, Achenbach and Rescorla 2001), uncorrected visual problems, specific language impairment, and language perception disorders were not included in the study. The groups were matched on age. The age means and ranges in the three groups were the following: preterms mean 9.56 years (7.6–11.1); full-term dyslexics mean 10.21 years (7.8–12.8); full-term good readers mean 9.67 years (7.1–10.9).

The children attended Hungarian-language schools (grade 2–4) and were not absent from school more than 3 months. At least one of the parents used Hungarian in communicating with the child since his/her birth. None of the children came from socially disadvantaged backgrounds, and all the mothers completed at least 8 years of general school. All of the children scored > 85 both on the verbal comprehension index and the perceptual reasoning index of the Hungarian version of the WISC-IV. The aim of the study did not require a more precise matching of the groups on IQ as it was a question to what extent and which way the various aspects of intelligence explained the individual variances in reading and spelling performances.

The Hungarian version of the 3DM¹ (Dyslexia Differential Diagnosis, Maastricht; Blomert and Vaessen 2009; 3DM-H: copyright Csépe et al. 2009; Ziegler et al. 2010) was used to test reading and spelling abilities. The reading test consists of common and rare real words as well as pseudo-words appearing on a computer screen, the difficulty of which both in length and orthography are gradually increasing. In the spelling test incomplete words appear on the screen which are to be completed by the subject, using letters offered by the computer. Dyslexia was established using the reading test of the 3DM-H. The “good readers” and the “dyslexics” were identified on the basis of z-scores of accuracy and fluency in reading the presented words and pseudo-words (good readers: T value 48–60; dyslexics: T value below 32).

Cognitive abilities were tested using the Hungarian adaptation of the WISC-IV and the Rey Complex Figure Test (RCFT) (Ogino et al. 2008; Ohtuska 2008).

The data analysis was performed using the 3.4.0 version of the R statistical program (R Core Team 2017).

The performance scores were analyzed by general linear mixed models (GLMM), using the “nlme” package (Pinheiro et al. 2015).

The group was the fix factor in the models, and the school was the random factor to take the similarities within the schools into account. In case of significant results (p < 0.05), the estimated scores of the groups were compared by linear contrasts using Tukey adjustments of p values and 95% confidence limits (Tukey 1949).

The cognitive background of the reading and spelling abilities was tested applying Random Forests and Classification and Regression Trees (Hothorn et al. 2006a, b).

According to the GLMM the groups had significant effects on all the 3 measures.

In order to reveal the cognitive background of the reading and spelling abilities a 3-step analyses were performed. Considering the few measurement points and the relatively great number of independent variables, in order to select the meaningful explanatory variables first Random Forests (Hothorn et al. 2006a) were applied. RF is an ensemble of Classification and Regression Trees (see in the next paragraph). The trees are built using random subsets of the data and random subsets of variables chosen at branching point of the trees. The importance of the variables can be estimated by this method (Hothorn et al. 2006b; Strobl et al. 2008).

In the second step a Classification and Regression Tree (CART) was fitted using the previously selected important variables to refine the selection of the explanatory variables and discover the potential interaction; CARTs are built using a nonparametric regression approach. Both numerical and categorical variables can be used to build a tree. The general rule is to split the observations into two parts based on a predictor variable (root) and then to split the subset further based on another or the same variable on a recursive way (Hothorn et al. 2006b).

Finally, general linear mixed models (GLMM) were fitted with the previously selected explanatory variables to prove their significance. The linear mixed model made it feasible to take into consideration the dependency of the data (the children attending the same school cannot be considered independent).

For reading accuracy the following explanatory variables were selected by the Random Forest: Rey MS, FSIQ, Rey CT, and WMI. On the ground of the intercorrelations, the CART eliminated the FSIQ and the Rey CT. The most powerful predictor was the recall accuracy of the Rey figures. The accurate readers had Rey MS > 16, but subjects with Rey MS ≤ 16 had chances to be accurate readers if scored > 106 on WMI. The WMI moderated the contribution of the Rey MS to reading accuracy (Fig. 1).

For reading fluency the Rey MS, PrSI, and WMI came out as important variables at the first step which was corroborated by the CART technique. The most powerful predictor of fluent reading was the recall accuracy of the Rey figures (Rey MS). Working memory and processing speed had mediator roles: Children with lower Rey MS (≤ 14) but having relatively high WMI (> 106) and/or PrSI (> 97) could as well be fluent readers (Fig. 2).

For spelling accuracy RF identified by FSIQ, Rey MS, and VCI as explanatory variables. According to the CART IQs > 105 provide suitable bases for accurate spelling. The role of the FSIQ is moderated by the recall accuracy of the Rey figures and the working memory (VCI). The ideal background structure for spelling accuracy is Rey MS, if FSIQ > 105 and VCI > 121, or if FSIQ < 105 and Rey MS > 21.5 (Fig. 3).

At the first step for spelling speed from among the heavily intercorrelated independent variables the PrSI was eliminated by the Random Forest; then, CART selected FSIQ and VCI as significant predictors. FSIQ was the main explanatory variable and VMI was a mediator: In order to react quickly, children either needed high FSIQ (> 111) or at least relatively high VCI (> 93) (Fig. 4).

As it was noted earlier, the aim of the study did not require the groups to be matched on IQ.

Nevertheless, we checked a potential bias in the data resulted by the higher IQs in the good reader group and lower IQs in the other two groups. We made subsets from the groups by filtering out children with high and low IQs. According to the repeated analysis using this restricted data set (preterms n = 13, dyslexics n = 17, good readers n = 20), the mean values hardly changed after the removal of children with high and low IQs. The results of the comparisons of the cognitive measures in the three groups, using the same general linear mixed models as in case of the total sample, were as follows: No significant difference was found in FSIQ (p = 0.8454), VCI (p = 0.4849), PRI (p = 0.2673), WMI (p = 0.0954), PSI (p = 0.3002), Rey CT (p = 0.0675), Rey_MT (p = 0.0846). However, the differences remained significant in reading accuracy (p < 0.0001), reading fluency (p < 0.0001), spelling accuracy (p = 0.0066), and spelling speed (p = 0.0042), as well as in Rey CS (p = 0.0023) and Rey MS (p = 0.046). These results suggested that the group differences in IQ were very unlikely to influence the results of the analyses (which would not have been feasible with the small data set).