Cumulative exposure to socioeconomic and psychosocial adversity and hair cortisol concentration: A longitudinal study from 5 months to 17 years of age

Highlights

• Youth exposed to lower and higher cumulative adversity had higher HCC levels;

• This finding emerges for males and females;

• The adversity-cortisol association ought to be tested linearly and non-linearly.

Abstract

Background

Exposure to early adversity has been associated with long-lasting risks for poor health and functioning. Prior research suggests that the hypothalamic-pituitary-adrenal (HPA) axis, and its main end-product glucocorticoid hormone cortisol, may be at play. This study tested whether an index of cumulative socioeconomic and psychosocial adversity assessed prospectively, from infancy to adolescence, was associated with hair cortisol concentration (HCC), and if this association differed by sex.

Methods

The sample comprised 556 adolescents (42.0% males) who provided hair for cortisol measurement at 17 years of age. Adversity indicators (young and single motherhood, low socioeconomic status (SES), maternal alcohol use, hostile-reactive parenting, and depressive symptoms, as well as peer victimization and neighborhood dangerousness) were repeatedly reported by mothers or youths between the ages of 5 months and 15 years.

Results

Chronic adversity was non-linearly associated with HCC; youth exposed to lower and higher levels of adversity had moderate-to-higher HCC compared to lower HCC noted in participants with moderate levels of adversity, for both males and females. None of the indicators taken separately, except the perception of neighborhood dangerousness, were significantly associated with HCC.

Conclusion

Our findings support the hypothesis that HPA axis activity varies according to cumulative adversity, albeit non-linearly, which may bear consequences for later health and functioning.

Introduction

Children growing up in environments characterized by socioeconomic (e.g., low family income and parents’ educational attainment) and psychosocial adversity (e.g., maternal depression, coercive parenting practices) are more likely to experience health and behavioral problems (Hunt et al., 2017, Oh et al., 2018). While the causal chain of events underlying this vulnerability is yet to be specified, many physiological systems are thought to be implicated, including inflammation, cellular aging, epigenetic mechanisms, and brain functioning and connectivity (Danese and McEwen, 2012, McCrory et al., 2017). The hypothalamic-pituitary-adrenal (HPA) axis and end-product glucocorticoid stress hormone cortisol have also been proposed to represent a key pathway by which early adversity induces health and behavioral disparities, namely because of their role in mobilizing energy required to cope with stressors and maintain homeostasis (Koss and Gunnar, 2018). Specifically, it has been argued that disrupted salivary cortisol responses to stress (i.e. higher, prolonged, repeated or blunted responses) and diurnal salivary cortisol secretion (flattened circadian rhythm, higher or lower levels at specific times of the day) may signal the wear and tear of the HPA axis following adversity, which could jeopardize other physiological systems, health and behavior over time, a phenomenon known as allostatic load (McEwen et al., 2015).

The association between childhood adversity and HPA axis functioning has, to date, mainly been investigated through the lenses of salivary cortisol. While experiences of abuse and parental loss were not associated with disrupted patterns of salivary cortisol secretion in basal or stressful contexts in a meta-analysis (Fogelman and Canli, 2018), participants maltreated as children were shown in another meta-analysis to be more likely to have blunted awakening cortisol levels, especially among agency-referred study samples (Bernard et al., 2017). Child maltreatment has also been associated with blunted cortisol responses to social stress in adulthood, albeit weakly so (Bunea et al., 2017). The evidence for a robust association thus still awaits, although blunted secretion across diverse indicators is not uncommon. The large heterogeneity noted between the studies included in these meta-analyses also raises the possibility that other factors may obscure this association, including the assessment of cortisol and adversity. For example, indicators of salivary cortisol are known to be sensitive to time-varying confounders (e.g., allergies, unexpected acute stressors, and time of saliva sampling) and poor adherence to saliva collection protocols. Differences across studies in the measurement of salivary cortisol may thus affect, to some degree, the accuracy of the cortisol indicators used, and thus the statistical power to detect significant associations with early adversity.

In contrast to salivary cortisol, which captures circulating cortisol secretion according to specific times and contexts, hair cortisol concentration (HCC) is a cumulative index of cortisol secretion that gradually incorporates into the growing hair over extended periods of time, typically 2–3 months (Russell et al., 2012, Stalder et al., 2017). Accordingly, HCC offers a complementary strategy to capture stable patterns of cortisol secreted across a variety of contexts, including in response to acute or repeated stress and throughout the day (Bryson et al., 2018). HCC is also less likely to be affected by time-varying confounders. Despite these potential advantages, however, recent meta-analyses and systematic reviews conducted exclusively on HCC also reported mixed findings with adversity (Gray et al., 2018, Khoury et al., 2019, Stalder et al., 2017), echoing in part what has been noted for salivary cortisol indicators.

It is noteworthy that the association between childhood adversity and HCC have mostly been examined using a cross-sectional research design (Gray et al., 2018). These studies have, for instances, reported higher HCC in young children growing up with less educated parents, whereas no differences could be detected for family income (Ursache et al., 2017, Tarullo et al., 2020, although see Bryson et al., 2018). Neighborhood disadvantage was also positively associated with HCC in participants of various ages (Tarullo et al., 2020, Zilioli et al., 2017, Vliegenthart et al., 2016). Alternatively, no HCC differences could be detected between school-aged children who grew up with a mother who experienced child maltreatment (Fuchs et al., 2018), or according to parental stress (Gerber et al., 2017), in comparison to controls. Lifetime exposure to trauma was nonetheless associated with higher HCC in children (Simmons et al., 2016). In their meta-analysis derived from 66 independent studies, Stalder et al. (2017) showed that individuals exposed to chronic stress (e.g., unemployment, natural disasters) had higher HCC. Notably, this association was greater in magnitude when the stressors were still present at the time of hair collection, whereas a non-significant trend for lower HCC levels was noted in conditions of past/absent stress (Stalder et al., 2017). An overall positive, but small association between HCC and adversity was also reported in the meta-analysis of 28 study samples aggregating a total of 3397 participants (Khoury et al., 2019). Building on prior knowledge of hyper- and hypo-secretion of cortisol in the literature, Khoury et al. (2019) showed that studies who reported lower HCC included participants who were maltreated as children, while those reporting higher HCC comprised participants that had either been victims of child maltreatment or who had experienced trauma as adults.

The findings emerging from these systematic reviews and meta-analyses – and the independent studies comprised in them – raise the possibility that the timing of adversity and the time elapsed from its onset at the time of cortisol measurement could affect the adversity-cortisol association and, thus, partly explain the mixed findings. This falls in line with prior findings indicating that the strength of the association between chronic stressors and multiple indicators of cortisol secretion (i.e. morning cortisol, daily output, and the adrenocorticotropin hormone or ACTH) decreases as time passes by between the stressors onset and cortisol measurement (Miller et al., 2007). In other words, while recent (or acute) exposure to adversity may co-occur with heightened HPA axis activity, lower levels of HPA axis activity is anticipated as these experiences become more distant (Miller et al., 2007). It has also been argued that adversity may have differential impact on HPA axis activity depending on when it was first encountered (Lupien et al., 2009). That is, variability in the timing of exposure and time elapsed between the onset of adversity and hair collection between and within the studies may obscure the association between childhood adversity and HCC. Studies conducted in large age-homogenous cohorts for which these factors are held constant or systematically set to vary are needed to test more specifically these hypotheses and thus clarify the childhood adversity-HCC association.

Second, the association between adversity and HCC has been investigated across a variety of adversity indicators, from parental education, housing problems, and maltreatment to neighborhood insecurity, which may bring about inconsistency if these experiences or contexts differentially relate to HCC (Tarullo et al., 2020). Also, importantly, these distinct forms of adversity often congregate within the same families. For example, children growing up in families headed by less educated parents or those earning a lower family income are disproportionally exposed to harsh parenting practices (Shaw and Shelleby, 2014) and peer victimization (Arseneault, 2018). Beyond the investigation of the unique associations between HCC and each indicator, we must test whether the accumulation of adversity across multiple socioeconomic and psychosocial adversity indicators affects chronic cortisol secretion. Preliminary evidence suggests that cumulative stress indexes are positively associated with HCC (Karlen et al., 2011, Simmons et al., 2016, O’Brien et al., 2013), although Bryson et al. (2018) did not detect a significant association between a cumulative index of 17 indicators of socioeconomic and psychosocial adversity and HCC among two-year-old toddlers (Bryson et al., 2018). To the best of our knowledge, no studies have yet tested whether the cumulative experience of multiple forms of socioeconomic and psychosocial adversity, from infancy to adolescence, is associated with HCC.

Third, accumulating evidence suggests that the association between childhood adversity and HPA axis activity may not always follow a linear function. That is, individuals who grew up in increasingly severe adverse environments may not necessarily show incrementally higher HPA axis activity, such as it is expected in a dose-response association. Attention in capturing severity in adversity is thus warranted. Indeed, while exposure to moderate stressors has been linked to lower HPA axis activity, more severe experiences of childhood adversity have been associated with heightened secretion (Gunnar, 2020, Boyce and Ellis, 2005). For example, 10–12-year-old children adopted internationally early in life from foster care systems had lower salivary cortisol responses to psychosocial stress than children who were adopted later on, suggesting a blunted HPA axis responsiveness at moderate levels of adversity (Gunnar et al., 2009). Notably, however, children who remained in institutions for a longer period did not differ from children reared with their biological parents, thus pointing to a U-shaped non-linear association between severity of adversity and salivary cortisol response to stress. Our team also reported a shift from moderate to lower to higher salivary cortisol responses to a psychosocial stress test in young adults who reported increasingly severe child maltreatment, indicating a J-shaped non-linear association (Ouellet-Morin et al., 2018).

Growing evidence also suggests that non-linear associations are also detected between childhood adversity and HCC. We have previously shown that persistent peer victimization was non-linearly associated with HCC among 17-year-old boys. While lower HCC was noted for participants who experienced moderate levels of peer victimization, moderate-to-higher HCC were noted for both youths who had been more victimized and those who had not been victimized by their peers (Ouellet‐Morin et al., 2020). Non-linear associations were also reported in adults between HCC and SES, as derived from income and levels of education, pointing to inverted and non-inverted U-shaped functions according to the minority group of the participants (O’Brien et al., 2013). Interestingly, the Adaptive Calibration Model (Del Giudice et al., 2011) proposes four prototypical patterns of responsiveness according to increasing levels of adversity/trauma, which captures and expands on these findings. While moderate-to-high responsiveness is anticipated in lower stress or safe environments, a buffered pattern of responsiveness is expected in the context of moderate stress. Notably, a point of inflection toward an enhanced pattern of responsiveness is proposed for individuals confronted with dangerous and unpredictable environments, once more suggesting a U-shaped function. However, females exposed to even more severe and traumatic stress are expected to show even higher responsiveness to stress (i.e. J-shaped function), whereas males are expected to revert to unresponsive patterns of response under the same conditions (Del Giudice et al., 2011). Altogether, theorization and preliminary findings suggest that non-linear functions (i.e., U-shaped, inverted U-shaped, J-shaped) may help to capture the variation in HCC across increasingly severe adverse context. Although most evidence points to a U-shaped association between adversity and HCC, the possibility that an inverted U-shaped or J-shaped functions may also be detected cannot yet be discarded.

Finally, several studies suggested that males display higher HCC than females (Gray et al., 2018, Rippe et al., 2016), although not consistently (Prado-Gascó et al., 2019). For instance, higher HCC was associated with person-related adverse life events (e.g., loss of a relationship) for males (12–18 years), but not female adolescents (van Dammen et al., 2020). Higher daily salivary cortisol levels were also reported for boys exposed to more pervasive maltreatment in comparison to girls (Doom et al., 2013), although it is uncertain whether these differences would be detected in hair. Several lines of evidence support a sex-moderation hypothesis, including sex or gender differences in sensitivity to stress, coping mechanisms, timing of puberty or hormonal changes occurring during puberty (Del Giudice et al., 2011, Negriff et al., 2015, Shirtcliff et al., 2012). It is also hypothesized that males and females differentially respond to stress, starting in adolescence, with males being more sensitive to stressors of a competitive nature and females to social exclusion (Stroud et al., 2009). Sexual dimorphic associations may, however, not be present across the distribution of adversity, but uniquely arise in severely adverse and traumatic contexts, as argued in the Adaptive Calibration Model (Del Giudice et al., 2011). For the most part, however, HCC studies have not tested (or reported) sex-moderation analyses and sex was not detected as a significant moderator in Khoury et al.'s (2019) recent meta-analysis. Systematic tests of this hypothesis are thus needed.

The present study aimed to test whether a cumulative index of persistent socioeconomic (e.g., low family income, young and single motherhood) and psychosocial (e.g., maternal depression, hostile-reactive parenting, peer victimization) adversity, measured between 5 months and 15 years of age, was associated with HCC measured two years later, at 17 years of age. To do so, we estimated group-based trajectories to capture distinct-but-homogenous levels of persistent adversity for each indicator and derived a cumulative score of adversity. We expected that the association between cumulative adversity and HCC would be best described non-linearly, as previously suggested in the Adaptive Calibration Model (Del Giudice et al., 2011) and reported in this cohort in regards to peer victimization (Ouellet‐Morin et al., 2020). Finally, we explored whether these patterns of findings differed for boys and girls, but did not articulate specific hypotheses as for whom, males or females, stronger associations with HCC are expected to arise considering previous mixed findings.