Innovations in biological assessments of chronic stress through hair and nail cortisol: Conceptual, developmental, and methodological issues

Abstract

Much of the existing research on biological mechanisms underlying the stress experience has focused largely on moment-to-moment stress, rather than on chronic stress, an arguably more powerful predictor of long-term outcomes. Recent methodological innovations have paved the way for new lines of research on chronic stress, with promising implications for developmental researchers and for those who study health and adversity. In particular, there are increasing studies that have focused on chronic stress assessments by relying on cortisol derived from hair and nails as a biomarker for chronic stress. In this paper, we provide an overview of their use, describe how hair and nail cortisol ought to be conceptualized differently across the lifespan, how developmental factors may impact its interpretation, and the circumstances under which its use may be more methodologically sensible. The purpose of this review is to provoke further discussion and encourage careful research designs that utilize hair and nail cortisol for understanding the effects of chronic stress exposure from the early developmental period, across adverse contexts, and in association with psychological and physical health outcomes.

1 |. INTRODUCTION

Understanding the basic processes that underlie the effects of stress across development has major implications for addressing real world problems such as poverty and trauma. It is understood that childhood is a vulnerable period within development, by which the effects of stress may take hold, producing adverse outcomes later in life (National Scientific Council on the Developing Child, 2014). Yet, among researchers, there lacks a consensus regarding the definition and operationalization of stress. The distinction made between stress reactivity and chronic stress is a promising direction in assessing and understanding the effects of stress across development.

Stress reactivity focuses on moment-to-moment responses to the exposure of acute stressors, characterized by uncontrollability and social-evaluative threat (Dickerson & Kemeny, 2004). On the other hand, chronic stress refers to the exposure to a stimulus or threat that takes place for an extended period of time. Methodological approaches to measuring chronic stress include documenting the presence of stressors, including major life events, daily events, and chronic strains; the psychological subjective reports of individuals’ own affective responses; and related biological responses (Vanaelst, De Vriendt, Huybrechts, Rinaldi, & De Henauw, 2012). It is important to note that the distinction between chronic and acute stressors is not always clear (Gottlieb, 1997). For example, a situation where the acute stressor is associated with the loss of a spouse can in turn be both foreshadowed and followed by a series of stressful events that altogether are prolonged in nature.

As such, an ongoing challenge is determining whether chronic stress can be measured. Substantial efforts have been made to obtain measurements within and across systems. Much of the psychobiological research on physiological dysregulation in both children and adults have focused extensively on the hypothalamic–pituitary–adrenal (HPA) axis and the autonomic nervous system (Chida & Hamer, 2008), with cortisol, the end product of the HPA axis, being perhaps the most extensively studied biological stress variable in developmental science. Cortisol concentrations from saliva or urine inform a relatively short time frame (at most a few days), can be invasive (when collected via blood), and require repeated measurements across 24 hr over several days to obtain average chronic concentrations (Wosu, Valdimarsdóttir, Shields, Williams, & Williams, 2013). With respect to capturing chronic stress, diurnal cortisol rhythms and the total amount of cortisol released during a stressor (e.g., AUC, area under the curve) has been used to make assumptions about cumulative cortisol exposure. Specifically, chronic stress has been associated with a flattening of the diurnal slope (Suglia et al., 2010) and a lower cortisol awakening response (CAR; Heim, Ehlert, & Hellhammer, 2000). Research with children who have faced adversity, including children living in Romanian orphanages or foster care, have shown such stressful circumstances to be associated with higher levels of cortisol over the daytime hours (Gunnar, Morison, Chisholm, & Schuder, 2001), or blunted early morning cortisol levels and no decrease over the day (Bruce, Fisher, Pears, & Levine, 2009). Overall, much has been written regarding salivary cortisol (including the diurnal slope and CAR) as a biological marker of chronic stress and we refer the reader to recent reviews on this area (An et al., 2016; Smyth, Hucklebridge, Thorn, Evans, & Clow, 2013; Stalder et al., 2016). Vanaelst, De Vriendt, et al. (2012) also provide an overview on epidemiological approaches in the measurement of stress among children between the ages of 6–12 years, covering the use of questionnaires and interviews as well as the cortisol from serum, urine, saliva, and hair.

This paper critically evaluates the use of hair and nail cortisol as biological markers of chronic stress (Dettenborn, Muhtz et al., 2012; Russell, Koren, Rieder, & Uum, 2012; Yamada et al., 2007). We do so for the following reasons. First, hair and nail cortisol are believed to represent an accumulation of cortisol output over a time frame of months and thus may be a potential means for characterizing chronic stress. Second, there is ease in the sampling and storage of hair and nail cortisol, which may facilitate its use in children and other populations. For example, accumulated cortisol as an outcome among children may be less vulnerable to fluctuations (e.g., naptime, breastfeeding) relative to salivary cortisol values (Russell et al., 2012); therefore, hair and nail cortisol may be an attractive potential biomarker for chronic stress in younger samples (Egliston, Mcmahon, & Austin, 2007; Gunnar & White, 2001; de Weerth & van Geert, 2002). Third, there is a rising interest in understanding the biological mechanisms of chronic stress, since capturing perceived or specific life stress through subjective report is not always feasible in very young children (Miller, Chen, & Parker, 2011). Overall, while stress responses in the moment can promote adaptation, prolonged stress over time can lead to biological dysregulation, producing vulnerabilities for illness and disease (McEwen, 2015). The accumulation of cortisol captured in early development may be a key mechanism underlying this link.

There are several reviews on the use of hair cortisol (Gow et al., 2010; Meyer & Novak, 2012; Mustonen et al., 2018; Russell et al., 2012; Stalder & Kirschbaum, 2012; Stalder et al., 2017; Staufenbiel, Penninx, Spijker, Elzinga, & van Rossum, 2013; Wosu et al., 2013), and our paper is by no means meant to be an exhaustive review. Instead we provide a brief overview of what is understood about hair cortisol as a purported measure of chronic stress and provide a review of the small number of studies that have been conducted with nail cortisol, a similar but newer potential measure of chronic stress. In our paper, we also cover conceptual, developmental, and methodological considerations for the use of these measures with children. Our purpose is to provoke further discussion and encourage thoughtful research designs that utilize hair and nail cortisol within developmental research.

1.1 |. Hair cortisol

With the rapid evaluation and popularity of hair cortisol in human research over the course of the past decade, much more is known about hair cortisol concentration in contrast to nail cortisol. Cortisol production from the hair has been reliably measured. Assuming a growth rate of approximately 1-cm length of hair per month, a 6-cm length of hair represents the retrospective cortisol concentration over the previous 6 months (Meyer & Novak, 2012). Low to high intraindividual stability (within a 12-month interval) has been documented in adults (van Ockenburg et al., 2016; Short et al., 2016; Stalder, Steudte, Alexander et al., 2012), in infants with samples at or before 12 months (Liu, Fink, Brentani, & Brentani, 2017; Liu, Snidman, Leonard, Meyer, & Tronick, 2016), and in children at age 4 years (Groeneveld et al., 2013).

An important question for developmental researchers is the validity of hair cortisol in humans. Studies that have directly examined the relationship between hair cortisol and cortisol obtained from saliva, blood, or urine (e.g., D’Anna-Hernandez, Ross, Natvig, & Laudenslager, 2011; Davenport, Tiefenbacher, Lutz, Novak, & Meyer, 2006; Short et al., 2016; Stalder, Steudte, Alexander et al., 2012) show weak to moderate relationships. One recent validation study by Short et al. (2016) addressed the limitation of previous studies by assessing cortisol levels from adult hair, saliva, and urine over the same time span of 1-month, which included daily assessments of waking, postwakening, and bedtime samples of saliva. While the sample size was small with 17 participants, their findings showed that hair cortisol concentration was most strongly associated with salivary cortisol area under the curve (AUC) over the month (r = 0.61, p = 0.01) relative to calculations based on averages of salivary AUC in 1-, 2-, 3-, and 4-week durations (r = 0.38–0.56). This study also revealed hair cortisol concentration to be a potentially more accurate reflection of overall cortisol output relative to other metrics of salivary cortisol (CAR or levels based on the diurnal rhythm), which showed no association with hair cortisol concentration. No association was obtained with cortisol concentrations from urine, with the authors proposing that cortisol conversion, which can affect the levels in the urine, as a reason for the lack of an association.

With regard to hair cortisol concentration and salivary cortisol in children, hair cortisol obtained among 7-year-old girls was associated with AUCg calculated from salivary cortisol collected post stressor (i.e., matching game) at 10, 20, 30, 40, and 50 min when the children were 3 years old (Ouellette et al., 2015). In a recent study by Kao, Doan, St. John, Meyer, and Tarullo (2018), hair cortisol was positively correlated with AUCg and AUCi, calculated from salivary cortisol responses at baseline, 20, and 40 min post stressor in preschool children. Hair cortisol was also associated with salivary cortisol AUC calculated from samples obtained at over the day in 5–11-year-old girls (Vanaelst, Huybrechts et al., 2012). Although a wider range of hair concentration values has been observed in younger children (1 year vs. 3, 5, and 8 years) perhaps due to the development of the HPA axis (e.g., Karlén, Frostell, Theodorsson, Faresjö, & Ludvigsson, 2013), these studies on children, to date, still suggest that there may be some association between hair cortisol and salivary cortisol, with chronic stress reflecting repeated measures of stress reactivity.

Aside from the validity of newer measures such as hair cortisol against previously established biological measures such as salivary cortisol, a critical question is whether the measures reflect psychosocial stress constructs, that is, whether hair cortisol is associated with either demographic or individual characteristics that reflect stress experiences, or self-reported stress. In focusing on the former, studies have examined hair cortisol in relation to categories of individuals exposed to chronic stress, either through stress-related conditions or contexts. These have included athletes (Skoluda, Dettenborn, Stalder, & Kirschbaum, 2012), those experiencing trauma from a natural disaster (Gao et al., 2014), and hospitalized infants (Yamada et al., 2007), to name a few. Despite the large variations in these conditions and contexts, and different analytical techniques for cortisol extraction, hair cortisol has generally been associated with those experiencing ongoing stress (for a review, see Herane-Vives, DeAngel et al., 2015). Indeed, a recent meta-analysis determined that groups exposed to chronic stress as a whole were estimated to have 22% elevated hair cortisol concentration, with those still currently experiencing the stress at the time of the study showing 43% increased hair cortisol concentration.

However, research examining hair cortisol and subjective reports of stress tend to report no association (Stalder et al., 2017). Despite the inconsistency in findings, it is hardly different from those found among salivary cortisol; in a review of the literature, Hjortskov, Garde Ørbæk, and Hansen (2004) reported that there was insufficient evidence for an association between subjective reports of psychological stress and the cortisol response in field studies (Hjortskov et al., 2004). Whether these findings generalize specifically in the hair cortisol from children cannot be concluded given the need to better define the experience of stress in children, the duration and timing of stress, and the developmental stages in which the exposures take place (Gray et al., 2018).

Relatedly, much work has focused on the evaluation of hair cortisol among individuals characterized by clinical concerns, including those experiencing high levels of chronic stress, such as those with major depressive disorder, bipolar disorder, generalized anxiety disorder, and post-traumatic stress disorder (PTSD). The findings involving such comparisons have been mixed and have been attributed to the complexities of participant characteristics such as psychiatric and physical comorbidities (Herane-Vives, Papadopoulos et al., 2015). Notably, meta-analysis by Stalder et al. (2017) covering an overwhelming number of adult studies showed a 17% lower hair cortisol concentration in those with anxiety disorders including PTSD. These findings align with the hypocortisolism documented in people who experience repeated traumatic events, such as those with PTSD. We discuss later on why hypocortisolism may also be observed in such chronically stressed groups.

A handful of studies have examined the relation between hair cortisol and clinical characteristics in children showing no association between hair cortisol concentration with anxiety and depression (Milam, Slaughter, Verma, & Mcconnell, 2014; Ouellette et al., 2015; Simmons et al., 2016; Ursache, Merz, Melvin, Meyer, & Noble, 2017). Higher hair cortisol levels were observed after school entry among children first entering elementary school at a mean age of 4.2 years, especially for fearful children (Groeneveld et al., 2013). Furthermore, higher hair cortisol levels were also associated with hospitalized infants compared to healthy term infants (Yamada et al., 2007). Notably, among children aged 5–11 years, AUC calculated from salivary cortisol obtained across the morning and evening was associated with child-reported stressful events in the recent past (3 months). On the other hand, hair cortisol from the past 6 months was associated with stressful events from beyond 3 months, altogether suggesting some link between reported stress and hair cortisol from periods further in history (Vanaelst, Huybrechts et al., 2012).

1.2 |. Nail cortisol

Cortisol obtained through fingernails is an emerging chronic stress biomarker with features that may overcome some of the limitations involved in using hair cortisol. For example, nails may be more easily obtained from the majority of individuals, including infants and children who often lack sufficient hair and arguably more resistant to decomposition and disintegration. As with hair, nail clippings have been used previously to detect toxin and substance levels (Jenkins, 1979). The quantification of cortisol through nails has been demonstrated through liquid chromatography/electrospray ionization-tandem mass spectrometry (Higashi et al., 2016; Khelil et al., 2011) and enzyme immunoassay (Izawa et al., 2015). With the compound incorporated into the nail matrix, cortisol does not become metabolized given that nails consist of dead keratinous cells. As with hair, determining the average growth rate for nails is imperative for determining the duration of cortisol exposure. Fingernails grow at a rate of 1.0 mm/10 days on average. Unlike hair, it may take several months for the nails to extend from the nail matrix to the point where sampling is possible (Gupta et al., 2005; Warnock et al., 2010).

The validation of nail is less developed than in hair, although nail cortisol has been shown to have similar concentration levels and variance with hair cortisol concentration (Hubmann et al., 2016), with moderate associations in hair cortisol concentration from facial hair (Nejad et al., 2016). Nail cortisol in humans has also been associated with salivary cortisol across the day, as calculated by AUC (Izawa et al., 2015). Interestingly, nail cortisol was also not found to be associated with hair cortisol in a recent study (Binz et al., 2018).

As with hair cortisol, the major question is whether the concentration of cortisol obtained from nail reflects psychosocial experience and whether it can be used as a reliable marker for chronic stress. Only a handful of studies have begun to examine this. Recently, Wu et al. (2018) showed an association between present perceived stress and increased nail cortisol among 51 medical students from China. In a small study of 19 university students, nail cortisol was higher during exam periods (Nejad et al., 2016). Nail cortisol was also positively correlated with stressful life events but not job stress or perceived stress in middle aged men and women (Izawa, Matsudaira, Miki, Arisaka, & Tsuchiya, 2017). Nail cortisol concentrations did not differ between healthy controls and individuals with bipolar disorder, nor was it associated with cognitive impairment (Herane-Vives, Papadopoulos et al., 2015). However, nail cortisol was higher in a sample of 26 depressed individuals than in 45 healthy controls matched by age and gender, and concentrations were positively associated with depression severity and particular symptomatology, although negatively associated with fatigue (Herane-Vives et al., 2018).

To our knowledge, a few studies have been conducted to examine nail cortisol in children. In a study we conducted with 46 mostly low-income minority adolescents, nail cortisol was not associated with perceived stress but instead was positively associated with selfregulation as assessed by effortful control (Doan, Deyoung, Fuller-Rowell, Liu, & Meyer, 2018).

An unpublished study presented in 2016 assessed nail cortisol in 124 at-risk children at 24 and 36 months but showed stability across time points in only a third of the children (Nejad et al., 2016). In a recent study, nail cortisol was also not found to be associated with diurnal measures of salivary cortisol among 324 children between 2 and 6 years of age (Messerli-Bürgy et al., 2018).

Compared to cortisol, dehydroepiandrosterone (DHEA) is a less researched adrenal steroid in its response to stress, although it is considered to be a cortisol antagonist that may serve to protect against the effects of cortisol. Although they did not assess nail cortisol, Tegethoff et al. (2011) observed that nail DHEA was higher for infants of mothers with greater life stress during pregnancy in a study of 80 infants (<3 weeks old). Among infants, nail DHEA was positively associated with prenatal stress in the first trimester, with nail DHEA positively associated with infant irritability (Mikoteit et al., 2018). Nail DHEA in the previously mentioned study on low-income minority adolescents was also shown to be positively associated with stressful events, sleep disturbance, and waking (Doan et al., 2018). In another study with 33 university students, an increase in the ratio between nail cortisol and nail DHEA was observed during a stressful period in the semester. This increase in ratio was attributed to a decrease in nail DHEA (Warnock et al., 2010), suggesting that chronic low levels of DHEA may be implicated as a result of repeated stress exposure. In light of these initial and intriguing findings, future work involving the analysis of cortisol and DHEA from nail may be promising for developmental research. Although steroid hormone extraction from nail has been established, the reliance of steroids from nail as a biomarker of chronic stress should be considered with caution at this point in time, given the correlational nature of findings and the limited number of studies involving the steroid concentrations within nail.

2 |. CONCEPTUAL, DEVELOPMENTAL, AND METHODOLOGICAL ISSUES

While both hair and nail cortisol offer exciting and new directions for understanding the biological role of chronic stress in children’s development, several important conceptual, developmental, and methodological limitations must be raised when implementing these approaches as part of a study design.

2.1 |. Conceptual issues

Do hair and nail cortisol represent stress? We put forth that the cumulative cortisol exposure from hair and nail cannot be readily assumed to represent “stress” of the psychosocial nature that is often the interest of developmental researchers. This is due in part to the different methods for defining and operationalizing stress, as discussed. Furthermore, given the months-long time frame of the sampling of hair and nail cortisol, disentangling the source for cortisol concentration levels is a challenge, relative to the salivary sampling obtained from acute paradigms.

Adding to the difficulty in definition is that the relations between cortisol and negative outcomes are not linear (Miller, Chen, & Zhou, 2007). Higher levels of cortisol are not always associated with negative outcomes, and alternative models explaining the mechanism by which stress affects health have argued that stress may negatively impact cortisol signaling and lead to stress-induced declines in cortisol (Heim et al., 2000). Treatment may involve the increase in cortisol (Papadopoulos & Cleare, 2012). Cortisol deviations—either too low or too high—can be detrimental, suggesting the importance of understanding the specific context or condition in order to determine the extent to which these levels are pathogenic (Marklund, Peltonent, Nilsson, & Olsson, 2004; Miller et al., 2007). At the same time, there is not a sufficiently large enough body of research to determine definitively the specific contexts in which high or low levels signal dysfunction. Thus, findings related to risk are framed post hoc, as a function of the results.

With respect to hair and nail cortisol, regulatory responses including increases in cortisol output from within positive or negative conditions altogether are embedded in the total concentration (McEwen & Seeman, 1999; Smith & Cidlowski, 2010). However, increases due to acute challenges, with some adaptive in the moment, likely produce wear-and-tear on the system over time. The extent to which hair and nail cortisol capture wear-and-tear make them ideal chronic stress indices that predict negative health outcomes. Interestingly, the onset of chronic stressors may initially lead to hypercortisolism, but with advancing chronicity result in hypocortisolism (Miller et al., 2007). That is, and analogously to subjective reports, in the context of chronic or repeated stress exposures, the HPA axis will habituate (Schommer, Hellhammer, & Kirschbaum, 2003). Moreover, the extent to which the stressor is still ongoing or in the past is likely to matter (Stalder et al., 2017). On balance, it is important to consider that hair and nail cortisol do not necessarily indicate stress but reflect the accumulation of repeated and ongoing HPA axis responses from a variety of sources that altogether could reflect psychosocial stress. Furthermore, hair and nail cortisol occur within a dynamic regulatory system that may or may not have habituated.

One approach to determining the extent to which hair and nail cortisol represent psychosocial stress is by examining its association with additional measures. Studies that have examined the association of hair cortisol and perceived stress have been largely mixed, with positive (Kalra, Einarson, Karaskov, Uum, & Koren, 2007), negative (Gerber et al., 2013; Karlén, Ludvigsson, Frostell, Theodorsson, & Faresjö, 2011), or no associations documented. One explanation is that the studies including perceived stress measures have primarily been of populations experiencing lower intensity of stress (Stalder et al., 2017). Another possibility may be the discrepancies in time frame; for example, the commonly used Perceived Stress Scale indexes stress from the last month, in contrast to most studies of hair cortisol, which capture concentrations from over the past 3 months (Frodl & Keane, 2013). Beyond this, individuals may not be very good at reporting stress experiences over a long period of time either due to memory failures or social desirability (Stalder, Steudte, Alexander et al., 2012), or even simple habituation. Herane-Vives, DeAngel et al. (2015) argue that perceived stress may actually be less sensitive of a measure, given differences in hair cortisol concentration but no differences in perceived stress scores in comparison and control subjects across different populations of participants categorized based on exposures to chronic stress conditions (Herane-Vives, DeAngel et al., 2015). Indeed, greater associations between hair cortisol concentration and negative life events are observed compared to hair cortisol concentration and perceived stress (Karlén et al., 2011).

2.2 |. Developmental issues

Various developmental factors can have an effect on cortisol concentrations in either hair or nail, including age-related changes of the HPA axis, developmental periods of sensitivity to stress (e.g., instability, rapid change), and genetic and contextual influences.

First, the HPA axis is not a static system but one that develops and changes over time (Gunnar, Wewerka, Frenn, Long, & Griggs, 2009). Specific time periods in development may be tied to normative fluctuations in cortisol, thus caution is required in the interpretation of hair cortisol across age groups in a cross-sectional design. For example, the HPA axis is relatively quiescent during childhood (Netherton, Goodyer, Tamplin, & Herbert, 2004) but reactivates during adolescence (Hankin, Badanes, Abela, & Watamura, 2010; Marceau, Ruttle, Shirtcliff, Essex, & Susman, 2015). Basal levels of HPA activity increase due to developmental changes, which may lead to increased responsivity of the stress response (Romeo, 2013). In addition, sex hormones influence the regulation of the HPA axis beginning in mid-adolescence (McCormick & Mathews, 2007), and chronic HPA activation can suppress gonadal function (Stratakis & Chrousos, 2007).

Second, the HPA axis may be also be more vulnerable during specific developmental periods. In animal models, there is evidence that glucocorticoids cross the placenta affect fetal HPA development, potentially resulting in long term and widespread changes to the organism (Kapoor, Petropoulos, & Matthews, 2008). Relatedly, prenatal stress is thought to influence the developing HPA axis; however, the effects are likely to vary based on a host of factors including the nature of stress, stage of gestation, and sex (Glover, Connor, & O’Donnell, 2010). In humans, prenatal anxiety and stress have been shown to predict variability in children (Gutteling, Weerth, & Buitelaar, 2004; O’Connor et al., 2005). At the same time, the direction of the association between maternal stress and child cortisol is not consistent. Prenatal anxiety has been correlated with infant cortisol in the first year of life, but the association varied as a function of infant age and the nature of the stressor (Tollenaar, Beijers, Jansen, Riksen-Walraven, & De Weerth, 2011).

Cortisol is also influenced by particular developmental time periods. As such, standard interpretations in using cortisol as a measure of a stress response may require re-interpretation. For instance, in pregnancy, a multitude of altered physiological parameters affect adrenocortical activity leading to an increase in total cortisol levels, with varying changes across each trimester. Based on salivary cortisol, unstimulated cortisol levels increase at around the 25th week of gestation, returning to normal levels around 5–7 days after parturition (Allolio et al., 1990; Meulenberg & Hofman, 1990; de Weerth & Buitelaar, 2005). Hair cortisol levels across pregnancy appear to similarly reflect the rise of cortisol across trimesters until birth and its return to normal levels by the postpartum (D’Anna-Hernandez et al., 2011). It is possible that cortisol, like allostatic load, may not be an appropriate measure of stress during pregnancy because normal physiological changes related to pregnancy may affect these measures (Morrison, Shenassa, Mendola, Wu, & Schoendorf, 2013). Relatedly, biological and social changes associated with puberty, including higher overall hormonal activity of the adrenal gland during puberty (Kiess et al., 1995), may also have an effect on early life stress that is reflected through HPA activity (Netherton et al., 2004). However, puberty was not associated with hair cortisol levels in a recent study examining children up to age 14 years (Noppe et al., 2014).

For these reasons, the relationships between hair cortisol concentration and age require greater clarification (Gray et al., 2018). The hair cortisol concentrations of very young children have been reported to be particularly high, with decreases across early development (1–9 years of age; Dettenborn, Tietze, & Kirschbaum, Tietze, Kirschbaum & Stalder, 2012) and decreases by adolescence (Binz et al., 2018). On the other hand, others have not observed associations between age and hair cortisol (Manenschijn et al., 2011; Rippe et al., 2016). Overall, hair cortisol may be appropriate for capturing change in cortisol output when considering a time frame based on developmental phases that are conceptualized across months. This may not necessarily generalize to nail cortisol given its sampling timeframe.

Finally, beyond normative developmental changes associated with cortisol, contextual changes that take place across the lifespan must be taken into account. According to Bronfenbrenner’s bioecological theory of development, the chronosystem is a fundamental level by which an individual and the larger context interact across development (Bronfenbrenner & Ceci, 1994). As such, adverse events or prolonged stress experiences following an exposure may vary in its impact based on both the child’s developmental stage and the particular context at that given time (Charmandari, Kino, Souvatzoglou, & Chrousos, 2003). A characterization of the context and its change would provide meaningful information that would inform the interpretation of hair or nail cortisol.

Contextual considerations across development have been investigated in salivary cortisol studies, demonstrating its sensitivity to new experiences and encounters within the environment involving school, family, and the community (Gunnar, Kryzer, Van Ryzin, & Phillips, 2010). In contrast, there has been very little direct assessment of the enduring conditions in hair cortisol studies, and even more, in nail cortisol studies involving children. Animal studies have relied on experimental conditions (e.g., relocation of Rhesus monkeys) to detect change in hair cortisol (Dettmer, Novak, Suomi, & Meyer, 2012). As we have previously described, chronic stress studies have often relied on natural experiment approaches, comparing groups of individuals in exposed and unexposed conditions, which may be valuable for looking at different populations of children.

Parent education and income have been examined in relation to child hair cortisol, which shed some light on context, although the findings remain mixed. The association between parent education or income with hair cortisol concentration has shown a negative relationship (Rippe et al., 2016; Ursache et al., 2017; Vaghri et al., 2013), positive relationship (White et al., 2017), or no relationship at all (Gerber et al., 2017; Groeneveld et al., 2013; Karlén et al., 2013, 2015; Liu et al., 2016; Maurer et al., 2016; Ursache et al., 2017). In one study of 26 children between the ages of 5–7 years, hair cortisol concentration was higher with lower parent education, but no association was observed between hair cortisol concentration and income (Ursache et al., 2017). Based on a review of the literature focusing on associations between socioeconomic status and hair cortisol concentration in children, most of the existing studies have been conducted on populations with little variation in wealth. It is possible that the lack of association with adversity may be due to methodological constraints. Beyond simple measures of parent education and income, however, lifetime history of trauma in mothers obtained during pregnancy was recently found to be associated hair cortisol in their children at age 3 and 4 years (Slopen et al., 2018). A crucial next step for research is to determine whether one’s status in society and/or concurrent impoverished conditions can be biologically captured through chronic measures such as hair or nail cortisol within early development.

Given this, there is ample opportunity to carefully assess the specific stress experiences that occur, the experiences that appear to be elicited by “low level” exposures, and the extent to which consequences to exposures are psychologically, emotionally, and physically enduring. Doing so may help to determine what is involved in the change in cortisol output as assessed by hair or nail cortisol. While self-reported measures of stress are not ideal for use with children, other approaches such as comprehensive behavioral (e.g., observational) and physiological monitoring (e.g., cardiac response) over the target period may further contribute to this assessment.

2.3 |. Methodological issues

Hair and nail assessments of cortisol may be more sensible for some groups than others. A question often asked by developmental researchers is whether hair can be obtained in children. Hair sampling may not be permissible for certain populations, who object to providing hair samples for cultural or religious reasons, or as a personal preference. Furthermore, some children, including infants and children with short hairstyles (often boys), will not have sufficient hair. Even when hair is available, a certain amount is required. The minimum weight of a required hair sample depends on the type of analysis; immunoassay analyses require approximately 5–50 mg (Karlén et al., 2011; Kirschbaum, Tietze, Skoluda, & Dettenborn, 2009). For high-performance liquid chromatography tandem mass spectrometry (LC–MS/MS), 20 mg of hair is required (Xie et al., 2011). In general, LC-MS methods are more difficult and costly as compared to immunoassays; however, they show high sensitivity, reproducibility, and little cross reactivity (Gao et al., 2013).

Assumptions regarding hair and nail growth rates are made when using hair and nail cortisol, introducing some probability of error in the time frame that is being studied. As of now, there is no easy method to assess individual hair growth rate (however, see Hayashi, Miyamoto, & Takeda, 1991 for a method using optical microscopy and image analysis). To date, the minimum length of hair reported in the hair cortisol concentration literature has been 1 cm of hair taken from the scalp, indexing an approximate 1-month retrospective period (Kirschbaum et al., 2009). There are individual variations in hair growth rate (LeBeau, Montgomery, & Brewer, 2011), seasonal effects (Randall & Ebling, 1991), and demographic differences (Loussouarn, 2001) to be considered.

Additionally, although analysis of cortisol in less than 1 cm is theoretically possible, the exact details and timing of cortisol incorporation into the scalp and hair root, and time for the hair to grow out is not fully understood (Pragst & Balikova, 2006), thus limiting interpretation of the timing of stress exposure. The activity of hair follicles consists of an active phase (anagen), transitional (catagen), and resting phase (telogen), and it is unclear the extent to which these stages affect cortisol levels (Sauvé et al., 2007). Relatedly, once grown, cortisol levels in hair are likely not static. Currently, the idea that a 1-cm length of hair reflects a 1-month period may not be generalizable across individuals, although improved time estimation is possible by relying on anagen hair, the individual growth rate, and methodologically consistent sampling (Pragst, Rothe, Spioegel, & Sporkert, 1998). There is some evidence to suggest the occurrence of degradation in cortisol obtained from the hair, which may render it unreliable for a 6-month period (Kirschbaum et al., 2009), although Manenschijn and colleagues found stable mean cortisol levels in women across 18 months through the analysis of six 3-cm segments (Manenschijn, Koper, Lamberts, & van Rossum, 2011).

Nails may be a promising alternative to hair given its availability, although issues remain in obtaining a sufficient amount, especially in very young children (e.g., newborns). Methods for surveying nail growth rate, including the marking of nails close to the proximal nail fold with a nail file, followed by measuring the distance of the mark from the nail fold at a later date have been reported in the literature (Yaemsiri, Hou, Slining, & He, 2010). With this approach, there seems to be variation in nail growth across fingers with the little finger growing more slowly (Yaemsiri et al., 2010). The growth rate of nails is also variable, which may affect the assessment of cumulative exposure. While fingernails grow from the nail matrix about 0.1 mm per day, the rate of growth is faster during pregnancy and adolescence and slower among older adults (Shearn, 1978). As such, it appears that the growth rate slows down across the lifespan (Orentreich, Markofsky, & Vogelman, 1979). Race, disease status (Geyer, Onumah, Uyttendaele, & Scher, 2004), handedness, geographical location (Geoghegan, Roberts, & Sampford, 1958), but not seasons (Bean, 1974), influence fingernail growth rates (Griffiths & Reshad, 1983). Toenails seem to grow slower and may potentially reflect a longer window of time (Bean, 1980).

An understanding of nail growth rate is needed given nail sampling requirements. While hair cut from the root indexes the recent past (depending on how much hair is cut), nails that are available to be cut are actually indexing several months past the time the nail emerges from the matrix. Thus, it may make more sense to collect data on potential causes of cortisol (e.g., stressful events) and wait a few months before collecting nail samples (Doan et al., 2018). Overall, there may be some imprecision in the time frame of the cortisol output being assessed when using hair and nail cortisol; accordingly, the measures would not be well suited for studies that aiming to detect change by the day.

Finally, important confounds affecting cortisol in hair need to be considered (Greff et al., 2018). In addition to age and sex (Dettenborn, Tietze et al., 2012), physical activity, adiposity, substance abuse (Wosu et al., 2013), and light exposure have been shown to influence cortisol levels in hair (Grass et al., 2016). In pregnant women, BMI, smoking, single motherhood, season, and mode of delivery has been associated with maternal hair cortisol in the last trimester (Braig et al., 2015). Certain medical conditions including diabetes mellitus (Staufenbiel, Penninx, de Rijke, van den Akker, & van Rossum, 2015) and use of cardiovascular medications (Abell et al., 2016) also influence hair cortisol levels. In a sample of 3,507 participants aged 59–60 years, seasonal variation, length of sample storage, and demographics (higher in Black as compared to other ethnic groups) had an impact on hair cortisol levels (Abell et al., 2016). Some of these relations (e.g., ethnic differences) may be mediated by stress exposure, but other factors (e.g., physical characteristics of hair) still need to be ruled out. For example, hair color, which overlaps with ethnicity, has been shown to be associated with hair cortisol concentration (Binz et al., 2018; Rippe et al., 2016). Because sweat on the scalp affects hair cortisol concentrations, frequency of hair washes should be considered (Greff et al., 2018). However, the effect of chemical treatment of hairs (e.g., dyes, shampoos) on hair cortisol has been inconclusive and thus should be documented (Hoffman, Karban, Benitez, Goodteacher, & Laudenslager, 2014; Kristensen, Larsen, Olsen, Fahrenkrug, & Heitmann, 2017). Relatedly, nail polish on nails seems to lead to implausibly high levels of cortisol and thus should be considered as a confound (Frugé et al., 2018).

3 |. CONCLUSION

Advances in methodology have improved the ease by which cortisol is assessed, leading to a substantial increase in studies using cortisol as a proxy for biological assessments of stress. For developmental researchers there is tremendous opportunity at the moment in evaluating and using hair and nail cortisol in the study of chronic stress across development, and in contributing to our understanding of concerns, such as poverty and trauma.

The ease of collection, storage, and analysis of hair cortisol, in particular, has led to its increased use. In light of this, we raise several important considerations for its use in future developmental research. Major conceptual considerations are needed to understand whether hair and nail cortisol are useful in relation to other constructs being studied and the particular study context being used before assuming its use as a measure of stress. We urge researchers to avoid simply assuming cortisol to be “the stress hormone,” and to consider the complexity of human behavior and physiology, which account for the cortisol concentration. Measures such as hair and nail cortisol reflect a cumulative response to a series of nonspecific exposures. There is a need to better capture the duration and timing of stress exposures on the developing HPA system before determining the extent to which it confers long-term outcomes. As it stands, our ability to capture these exposures alongside the collection of hair and nail cortisol has been limited, with much reliance on demographic or clinical characteristics and other self-reported stress.

The HPA axis is impacted by particular developmental periods and related contextual changes; as such, the interpretation of hair and nail cortisol must take this into account. As noted, there has been very little characterization of the contexts among studies using hair or nail cortisol, with studies mostly relying on simple measures of socioeconomic status. For instance, only a few hair cortisol studies have made minority and low-income children and families a focus of study (Doan et al., 2018; Liu et al., 2017; Rippe et al., 2016; Vaghri et al., 2013). Given the very little variation in context among most studies, there remains a limited understanding regarding the influence of context on hair and nail cortisol across development. Hair and nail cortisol may be particularly promising for the study of stress among vulnerable populations for children and families, and an important key construct toward addressing health disparities within underserved communities.

Finally, methodological considerations such as hair and nail growth rates, and confounds and factors related to both growth rates and the cortisol output (e.g., timing) from these samples should be considered in the use of hair and nail cortisol. Methodological studies that contribute to resolving these concerns would help to improve the utility of these biomarkers across different populations.

While the overall literature is definitive in suggesting that numerous social and demographic factors shape the HPA axis and hence cortisol levels, their relative contributions, the direction of association between the factors and cortisol levels, and importantly, the consequences of high or low cortisol levels remain to be seen. Future studies that capitalize on the existing literature while carefully considering these conceptual, developmental, and methodological concerns may provide meaningful information that addresses stress-based health outcomes, especially for vulnerable populations of children.