| Last Updated:
Creating DOI. Please wait...
Authors: Elizabeth McNeilly*, Natalie Saragosa-Harris*, Kathryn Mills, Ronald Dahl, Lucía Magis-Weinberg
*Joint first authorship.
Research has demonstrated a link between pubertal development and risk for internalizing symptoms in adolescence (Patton et al., 2008; Crockett et al., 2013; Mendle et al., 2014; Wang et al., 2016; Boivin et al., 2017; Byrne et al., 2017; McGuire et al., 2019). The relationship between puberty and internalizing symptomology may in part stem from pubertal changes in neural reward circuitry (Davey et al., 2008; Forbes et al., 2010; Braams et al., 2015; Luking et al., 2016; Ladouceur et al., 2019). A notable portion of work examining pubertal associations with internalizing symptoms has focused on adolescents ages eleven and older (Galvao et al., 2014; Ullsperger & Nikolas, 2017) and a similar focus on older adolescents is evident in research on pubertal-related changes in brain development (Vijayakumar et al., 2018). Research focused on individuals in this age range has demonstrated important links between various markers of pubertal development and risk for psychopathology. Without the incorporation of younger individuals, however, research focused on this older range of adolescents likely overlooks early stages of pubertal development, particularly the transition to puberty. Given the heterogeneity of timing in which individuals begin puberty (Anderson et al., 2007; Marceau et al., 2011; Lee & Styne, 2013), it is important to incorporate younger samples and capture development of those who begin the process of puberty at an earlier age. Moreover, research has demonstrated that several biological markers (e.g., pubertal hormones) of puberty are evident earlier than the external markers that are often assessed in self-report or medical evaluations of secondary sex characteristics (Mendle et al., 2019). Together, these findings suggest that in order to evaluate pubertal development comprehensively, it is important to consider pubertal development in younger samples and to account for both self-reported and biological assays of pubertal development.
In the current project, we plan to investigate the relationship between puberty, reward circuitry, and internalizing symptoms in a large, nationally representative sample of nine- and ten-year-olds. We will use publicly available data from the Adolescent Brain Cognitive Development (ABCD) Study (Jernigan, Brown, & Dowling, 2018). Many (but not all) studies in peri-pubertal adolescent girls have found an association with puberty but not with age in the emergence of internalizing symptoms and disorders (Angold, Costello and Worthman, 1998; Killen et al., 1992; Patton et al.,1996), although there is some evidence that this association may differ by race (Hayward et al., 1999). Although prior work has demonstrated that risk for internalizing disorders is heightened in adolescent girls compared to boys, some research suggests that these gender differences do not emerge until age thirteen or older (Hankin et al., 1998; Salk et al., 2016). Moreover, it is unclear whether there are gender differences in the association between puberty and internalizing behaviors. A recent meta analysis suggests that gender does not moderate the effects of puberty on psychopathology, but most of the studies included in the analysis did not include participants as young as nine years old and the focus was more broadly on psychopathology rather than internalizing disorders (Ullsperger & Nikolas, 2017). To our knowledge, the specific relationship between pubertal stage and internalizing symptoms has yet to be examined in a large, nationally representative sample of nine-and ten-year-olds. One particular strength of the ABCD Study is the possibility to robustly investigate the associations between gender, pubertal stage, and internalizing in a large, diverse sample of youth within a narrow age range. Moreover, examining differences in pubertal development in such a sample will afford us the opportunity to investigate variability in reward-related processes and internalizing symptoms across a range of pubertal stages. We hope that by studying a population of younger individuals than those more frequently examined in similar studies, we will further our understanding of ways in which pubertal development may present risk for later internalizing symptomology. Importantly, identifying predictors of mental illness in a younger age range could also serve to inform future early interventions.
Although puberty is a normative and expected process for adolescents, there are notable individual differences in pubertal development (Anderson et al., 2007; Marceau et al., 2011; Lee & Styne, 2013). In particular, although individuals likely progress through pubertal stages in a similar and consistent order, the ages at which they progress through these stages can vastly differ. We acknowledge that, without examining stage-matched peers (i.e., individuals at the same pubertal stage but potentially different ages or “timing”), it will be difficult to distinguish between pubertal stage and timing in the current project. However, given the young age of participants in the sample, advanced pubertal stage in these individuals is potentially indicative of early pubertal timing. That said, we will not categorize individuals by “early onset” or “late onset” in our analyses but will instead measure both pubertal stage and testosterone levels continuously. In later waves of the ABCD Study, it will be possible to further distinguish pubertal stage from pubertal timing, as more of the sample begins to reach puberty at different ages.
By analyzing data from ABCD Study, we have the opportunity to employ split-half analyses while maintaining a large sample size. The nature of these data thus allow for examination of exploratory analyses in one half of the dataset (hereinafter referred to as “Sample 1”) and confirmatory analyses in the other half of the dataset (hereinafter referred to as “Sample 2”). Notably, the literature has conflicting evidence regarding the role of reward anticipation versus reward feedback in predicting internalizing symptoms (Luking et al., 2016). Without conclusive evidence in support of either system as the best predictor of internalizing behaviors, it is important to consider both processes, particularly in a large dataset where both processes have been measured within subjects. Fortunately, our planned analysis structure will allow for a transparent and data-driven examination of these two potentially distinct but important processes.
Therefore, the purpose of this preregistration is to outline our plan for an analysis that will include two phases. The first phase (hereinafter referred to as “Phase 1”) will include the analyses on Sample 1, which are described in detail in this preregistration. For this stage, there are both exploratory and confirmatory hypotheses, all of which are outlined here. Results from Sample 1 will guide our subsequent analyses of Sample 2, so that we can confirm any hypotheses supported by Sample 1 in Sample 2. In this way, Phase 1 will assess both questions for which we have specified, directional hypotheses (i.e., the initial confirmatory hypotheses listed in this preregistration) and questions that are exploratory in nature (i.e., the exploratory hypotheses listed in this preregistration). Phase 2, on the other hand, will consist of analyses to replicate and confirm findings from Phase 1. Following our analyses of Sample 1, we will add an amendment to this preregistration with our plan for confirmatory testing. That is, we will preregister the confirmatory hypotheses derived from the results of Phase 1 of the study. It is possible that we will discover mistakes in logic or missing details while analyzing Phase 1. We may amend or clarify approaches in the second preregistration before moving on to apply the analyses in Sample 2. Any models that are changed during Phase 1 (i.e., are different from what is written in this initial preregistration) will be detailed in our second preregistration. While all results from both phases will be reported, only effects that replicate will be interpreted.
CC-By Attribution 4.0 International