Assessing stress reactivity indexed via salivary cortisol in preschool-aged children

Summary

Identifying a stressor paradigm that elicits mean increases in salivary cortisol in young children has proven elusive, possibly due to characteristics of the paradigms used and how and when cortisol is sampled. We therefore examined the validity of a standardized task (adapted from Lewis and Ramsay, 2002) and procedures developed to assess cortisol reactivity in 215 preschool-aged children. Children participated in a standardized stress task during a home visit, which was videorecorded for future coding. Salivary cortisol samples were obtained at baseline and 10, 20, 30, 40, and 50 min post-stress. In support of the validity of the task, significant increases in cortisol levels from baseline were found, followed by a significant decline, and a quadratic function provided a good fit to the data. Children also showed a significant increase in negative emotions and a decrease in positive emotions over the course of the stress task. Results indicate that the task successfully elicited the hypothesized cortisol response in 3-year-old children.

Introduction

Individual differences in stress sensitivity have been strongly implicated in the etiology of major depression and anxiety disorders (Heim and Nemeroff, 2001, McFarlane et al., 2005), suggesting that early-emerging indicators of heightened stress sensitivity could be used to develop targeted preventative and early intervention strategies. As befitting a topic with such important implications, numerous studies have aimed to measure childhood stress reactivity, operationalizing sensitivity to stress via activity of the hypothalamic–pituitary–adrenocortical (HPA) axis in response to standardized stress paradigms. Cortisol, the hormonal end-product of the HPA system, can be readily indexed noninvasively through salivary assays and is a potentially useful tool in developmental psychopathology research aimed at characterizing stress responsivity in childhood. However, the methods used to elicit cortisol reactivity in children vary widely, and the resulting literature is, accordingly, mixed and inconclusive.

Gunnar et al. (2009) recently provided an excellent review of this literature, highlighting the substantial variability in methodologies used to elicit cortisol in early childhood. In this review, the authors note that only a small minority of studies of young children (i.e., ages 2–5 years old) report a mean increase in child cortisol in response to standardized stressor paradigms (Gunnar et al., 2009). Gunnar and colleagues concluded that without “concerted effort to develop reliable stress tasks for this age range, we may be left with a broad span of development during which we cannot really assess reactivity of the axis in typically developing children” (p. 964).

Despite the absence of a consistent literature to draw upon, Gunnar et al. (2009) made recommendations regarding task characteristics that may elicit a cortisol increase in young children. For example, Gunnar et al. observed that leading children to believe that they have failed on a task that younger children are capable of doing may elicit a cortisol increase. This suggestion is remarkably consistent with results from a recent meta-analysis of cortisol reactivity in adults (Dickerson and Kemeny, 2004), which showed that tasks that entailed a socially evaluative component, along with perceived uncontrollability and motivated performance, produced the greatest and most prolonged cortisol response. While this meta-analysis included studies of adults only, designing developmentally appropriate downward extensions of tasks that elicit cortisol responses in older samples would represent a valuable contribution to the field, as research aimed at examining stability and change in cortisol reactivity across development will be facilitated by tasks that are similar in terms of the nature of the stress manipulation.

An array of methodological considerations may have influenced the extent to which previous research has been able to accurately characterize early-emerging cortisol reactivity in young children. The vast majority of studies of cortisol reactivity to laboratory stressors have assessed cortisol at two time points only (Earle et al., 1999, Matthews et al., 2001, Roy et al., 2001, Smeekens et al., 2007; for exceptions, see Mills et al., 2008, Zoccola et al., 2008). This practice may have limited the extent to which studies can accurately capture peaks in cortisol response, as individuals, including children, vary in terms of how rapidly a maximum cortisol response is expressed post-stressor (Goldberg et al., 2003, Gunnar and Talge, 2008, Lewis and Ramsay, 2002, Lopez-Duran et al., 2009). Additionally, obtaining a minimal number of samples hinders the ability to characterize post-stress downregulation/recovery, which may also have implications for psychopathology risk.

Other methodological considerations include where and when assessments of cortisol reactivity are conducted. Many previous studies have assessed cortisol reactivity in children in laboratory settings (Lewis and Ramsay, 2002, Lopez-Duran et al., 2009, Smeekens et al., 2007). While laboratory settings have advantages, this approach raises the concern that the novelty of the setting itself might influence children's cortisol levels. Research by Tottenham et al. (2001) indicates that this is so, by showing that assays of children yield higher levels of baseline cortisol when samples are collected in the laboratory versus the home (as cited in Gunnar and Talge, 2008, p. 351). This indicates that the baseline cortisol levels reported in many studies reflect a psychophysiological reaction to coming to a laboratory, rather than a true baseline measure of cortisol activity, and suggest that testing done in the home may better identify true increases and recoveries in cortisol levels in response to stressors of interest. Also, cortisol varies throughout the day, with peak levels being reached upon awakening, and lowest levels found after sleep onset (deWeerth et al., 2003, Gunnar and Talge, 2008). This natural variability can introduce additional, nonsubstantive variability into cortisol assays, which is problematic given that most studies of children's cortisol responses to stress have tested participants at different times of day.

With these issues in mind, we examined whether a significant cortisol increase could be elicited from preschool-aged children using a task adapted from Lewis and Ramsay (2002). This task, which already incorporated many features advocated by Gunnar et al. (2009) for use in tasks to elicit cortisol from children, was further adapted to be suitable for use with younger children, and to incorporate additional elements implicated by Dickerson and Kemeny (2004). Additionally, we attempted to control or eliminate nonsubstantive influences (e.g., time of day) on children's cortisol.