Abstract

Introduction

Cognitive abilities comprise information-processing functions such as selection, storage, manipulation and organization of information (Deary et al., 2010). Individual differences in cognitive abilities contribute to differences in important outcomes including educational attainment, job performance, economic preferences, health behaviors, and mortality (Burks et al., 2009; Calvin et al., 2011; Gottfredson, 1997). Verbal ability— understanding, processing and producing meanings linguistically—is an important cognitive function which correlates strongly with general cognitive ability (Deary et al., 2010).

Because language is crucially involved in cognitive abilities, differences in language development predict later cognitive outcomes. For example, earlier speaking age predicts better cognitive ability at age 8, better school achievement in childhood and adolescence, and higher educational level and verbal ability in adulthood (Hohm et al., 2007; Murray et al., 2007; Taanila et al., 2005), and language development among preschoolers predicts reading skills in school age (Hayiou-Thomas et al., 2010).

In the quest for predictors of alcohol problems, researchers have also assessed the role of cognitive abilities. Large-scale longitudinal studies have found poorer cognitive abilities in adolescence and early adulthood to predict increased risk for alcohol-related morbidity and mortality among drinkers several years later (Gale et al., 2008; 2010, Sjölund et al., 2012). Individual differences in specific cognitive functions are also risk factors for drinking behaviors. Notably, variation in executive functions, or cognitive control, has strong associations with alcohol and drug use behaviors (Clark et al., 2008).

Other longitudinal studies have investigated whether cognitive ability predicts initiation of alcohol use and variation in drinking patterns in adolescence and adulthood. Contrary to findings on alcohol-related problems, cohort studies have found better cognitive test performance in childhood to predict earlier experimentation with alcohol and higher levels of drinking in adolescence and adulthood, whereas poorer ability has been linked to a higher likelihood of remaining abstinent (Batty et al,. 2008; Hatch et al., 2007; Jefferis et al., 2008; Kanazawa and Hellberg, 2010; Wilmoth, 2012). The origins of these associations are poorly understood. Specifically, while previous studies have assessed school-aged children with standard tests of cognitive ability, it is not known whether individual differences in early childhood cognitive development predict future drinking behaviors. For example, large variation in language development exists (Fenson et al., 1994). As reviewed above, there is continuity between childhood verbal development and cognitive ability up to adulthood, but associations between language development and later substance use behaviors are unknown. To better understand the role of cognitive abilities in the development of alcohol use, investigating such early developmental indicators would be important.

Cognitive and verbal development could be predictive of drinking behaviors due to several factors. Peer influences and the tendency to seek novel experiences could link better verbal development with drinking behaviors. Good language skills reduce the likelihood of peer rejection (Braza et al., 2009; Menting et al., 2011) whereas higher social activity predicts more frequent drinking in adolescence (Pitkänen et al., 2008). The personality trait novelty seeking is positively associated with alcohol use (Zuckerman and Kuhlman, 2000), and cognitive performance and reading abilities in childhood are related to higher stimulation seeking tendencies (Raine et al., 2002).

Associations between better cognitive development and experimentation with alcohol might merely reflect confounding by familial background. Higher parental education and income are positively associated with early cognitive and verbal development (Kelly et al., 2011; Schjølberg et al., 2011), and alcohol use may be more common among adolescents from higher-income households (Melotti et al., 2011). As familial factors are difficult to measure and fully control using statistical adjustments, their contribution to the relation of cognitive development and drinking behavior is uncertain.

In the present study, we hypothesize that individual differences in childhood verbal development are associated with drinking behaviors from early adolescence to young adulthood. In two longitudinal population-based samples of twin families, we use the twins’ frequency of drinking and intoxicating and self-reported problems related to alcohol use as indicators of drinking behaviors. As indicators of verbal development, we use parental reports of age at speaking words and learning to read, and expressive language skills during school age. We know of no previous studies that used such early information to investigate associations between cognitive development and the development of alcohol use behaviors.

We use the discordant co-twin design. By examining drinking behaviors within like-sexed twin pairs whose members are discordant for verbal development, we aim to minimize the possible confounding effects of sex, age, and birth cohort, as well as familial confounding factors shared by the co-twins. The within-pair comparison also adjusts for all genetic influences among genetically identical monozygotic (MZ) twins and part of the genetic influences among dizygotic (DZ) twins who share, on average, 50% of their segregating genes. Thus, the aim of the present analysis was to investigate whether childhood verbal development is associated with drinking behaviors in adolescence independently of familial background.

Materials and Methods

Results

Discussion

In two independent, population-based samples of twins, associations between co-twin order in verbal development in childhood and differences in alcohol use behaviors across adolescence and young adulthood were studied. Jointly, the findings are suggestive of a within-pair relationship between better verbal development and engaging in drinking behaviors, although that association was not found for all verbal developmental and drinking variables. In both samples, the earlier speaking co-twin tended to report more frequent drinking than the later speaking co-twin at some point in adolescence. In addition, the twin who learned to read before his/her co-twin did, was more likely to report more frequent drinking or intoxication in young adulthood in both samples. Finally, the co-twin whom the parents considered to be more advanced in expressive language skills during elementary school was less likely to be an abstainer at age 16, and was more likely to report frequent drinking in young adulthood, as well as intoxication across adolescence and in young adulthood. Within twin pairs, verbal development was also positively related to having consumed a larger amount of alcohol within 24 hours in both samples, and to reporting more alcohol-related problems in one of the samples, although these associations were rather weak.

To our knowledge, no previous study has assessed the association between verbal development in early childhood and later development of alcohol use behaviors. Our findings thus extend evidence from earlier studies suggesting that better cognitive abilities in school-aged children and adolescents may relate to a higher likelihood of reporting drinking and some alcohol-related problems (Batty et al., 2008; Hatch et al., 2007; Jefferis et al., 2008; Kanazawa and Hellberg, 2010; Wilmoth, 2012; Windle and Blane, 1989).

The present analysis extends earlier research in another important way. The associations between verbal development and drinking behaviors were observed within twin pairs, indicating that these associations were not confounded by familial influences shared by co-twins, such as factors related to parents or the rearing environment. Nor were the associations influenced by gender differences in verbal development and drinking behaviors, as only like-sexed twin pairs were studied. However, the findings were not unequivocal regarding the role of genetic influences, because the relatively small samples informative for conditional logistic regression analyses did not permit robust testing within MZ and DZ pairs.

Complementary analyses revealed that having been verbally more advanced in childhood was also related to an increased likelihood in adolescence to have friends who drink and to report higher levels of experience seeking than the verbally less advanced co-twin. Seeking novel experiences and having drinking peers are associated with higher levels of alcohol use, and these factors might link verbal development with drinking behaviors. But most adolescents report having drinking friends, so these results indicate that youth with better verbal development were more likely to belong to this majority.

For interpretation of the results, it should be recalled that in most twin pairs the co-twins did not differ in verbal development, and of the pairs that did differ, in less than half the co-twins were discordant for alcohol use. Second, none of the associations were strictly replicated in both studied samples. We can only speculate the reasons behind this inconsistency. Data collection for FT16 took place a decade earlier than that for FT12, and important differences between these periods in Finland do exist. FT16 twins were adolescents in the middle of a serious economic depression, whereas FT12 twins grew up during an economic boom. Such macro-environmental contextual factors may have influenced the results in a complex manner (Rose et al., 1999).

Further, the current findings could potentially be due to some unknown, pre-existing differences between co-twins, present prior to verbal development. Finally, cooperative sibling interaction effects on substance use could have been affected the results (Penninkilampi-Kerola et al., 2005; Rhee et al., 2003; Rose et al., 1999). Such effects may have reduced differences in drinking between co-twins and thus weakened the association between verbal development and alcohol use.

This study has some important limitations. First, only retrospective parental reports of verbal development were available. Such reports may be biased by several factors, most notably by parental perceptions of the child’s later development and behavior. Second, with one exception, information was available only on the within-pair order of verbal development in co-twins, which limited the present analyses. Parents are likely to note and recall clear differences between co-twins in the timing of reaching important developmental milestones, such as speaking words, but without information on the magnitude of co-twin differences, the meaningfulness of these data can be questioned. We were able to investigate this issue with the item on expressive language skills during elementary school. This item originally had the response options “Slightly more advanced” and “Clearly more advanced”, which were collapsed for the analyses (see Methods). Using only pairs with a “clear difference” in expressive language skills produced fairly similar results as the whole sample. Some evidence for a stronger association in more discordant pairs was also found (results available from the corresponding author).

Validity of the parental reports of developmental differences was supported by the observation, in both samples, that better verbal development as reported by parents predicted better verbal ability, assessed with a standard neuropsychological test, and higher educational level, reported by the twins, in DZ but not in MZ twin pairs, compatible with genetic influences on the continuity of verbal and cognitive abilities (Deary et al., 2000; Hayiou-Thomas et al., 2012). However, we cannot fully exclude the possible effect of rater bias in parental reports (Bartels et al., 2007) of verbal developmental differences.

A third potential limitation is that data on complex verbal developmental phenomena were collected with single items, ambiguous for their exact meaning. For example, “speaking words” could denote several different stages of early expressive language development (Fenson et al., 1994).

Strengths of the present study include the use of two large and independent population-based samples with extensive information on the development of alcohol use spanning from early adolescence to emerging adulthood. The within-pair design had the notable strength of controlling for all potential confounding factors shared by co-twins.

In conclusion, the present study found that within like-sexed twin pairs, the co-twin that parents reported as more advanced in language development in childhood was more likely to report more frequent drinking and intoxication in adolescence and young adulthood. Results mostly related verbal development to engaging in normative drinking behaviors. Longitudinal studies with representative samples and informative study designs are required to chart the development of the complex connections between cognitive factors and drinking behaviors.

Supplementary Material

Acknowledgments

References