Hair cortisol, perceived stress, and social support in mother–child dyads living in an urban neighborhood

Abstract

Women and children belonging to a racial/ethnic minority bear a disproportionate burden of psychosocial stress that increases their vulnerability to adverse health outcomes. Hair cortisol has been rapidly advanced as a biomarker of the intensity and course of the stress response over time and may provide an opportunity to increase our understanding of the role of psychological stress in health. However, research on the link between hair cortisol levels and subjective measures of maternal and child stress among low-income and minority individuals is limited. The goal of this study was to examine the association between stress and hair cortisol in low income, minority women and children who experience disproportionate exposure to chronic stress. A convenience sample of 54 minority mother/child dyads from a busy primary care clinic in the second poorest medium-sized city in the US participated in the study. Mothers self-reported perceived stress, social support, household characteristics and other demographic factors, and their children (ages 7–14 years) reported on the perceived level of safety in their neighborhood and exposure to violence as markers of child stress. Three-centimeter hair samples were collected from both mothers and children during the clinic visit, and hair cortisol levels were assessed via enzyme-linked immunosorbent assays. Linear regression models examined associations between maternal and child hair cortisol, and between hair cortisol and perceived stress level in women, and moderation by social support. Maternal hair cortisol was not significantly associated with mother’s perceived stress. Maternal and child hair cortisol levels were positively associated (p = .007) but this association was not moderated by maternal perceived social support. These findings suggest that hair cortisol is strongly associated among this sample of minority mother–child dyads and is not moderated by social support.

Lay summary

The majority of hair cortisol research has not focused on racial/ethnic minorities who bear a disproportionate burden of psychosocial stress. Here, we demonstrate that hair cortisol is strongly associated among a sample of minority mother/child dyads. This study fills a gap by demonstrating maternal/child HCC association in an understudied Hispanic/Latina, low-income population that is disproportionately at-risk of adverse health outcomes.

Keywords:

• Stress
• hair cortisol
• social support
• minority
• urban

Introduction

Chronic environmental stressors related to poverty, neighborhood crime, discrimination, or exposure to violence can lead to maternal psychosocial stress (Dunkel Schetter, 2011). Studies have demonstrated an association between maternal psychosocial stress and their child’s socioemotional development (Palmer et al., 2013) and negative health outcomes, such as preterm birth, low birth weight, depression, and childhood asthma (Dunkel Schetter, 2011). It is well documented that stress alters regulation of the hypothalamic-pituitary-adrenal (HPA) axis (Kirschbaum, Tietze, Skoluda, & Dettenborn, 2009). Maternal depression, poverty, and related stressors are often associated with a dysregulated HPA axis in offspring (Dunkel Schetter, 2011). Yet, few studies have examined these associations among racial/ethnic minorities, particularly Hispanic/Latina women and their children, who bear a disproportionate burden of these environmental stressors.

The HPA axis coordinates many basic physiological functions and the body’s response to stress (Wosu, Valdimarsdottir, Shields, Williams, & Williams, 2013). Cortisol is the final effector of the HPA axis, and historically salivary and plasma cortisol levels have been employed as indicators of psychosocial stress. These measures, however, only provide point estimates of HPA axis activity making them impractical for assessing chronic stress responses. In addition, these measures are affected by normal daily fluctuations, are highly sensitive to context, and thus, should ideally be sampled repeatedly over several days. Assessing cortisol levels in hair provides an index of cortisol response to stress over time- (Davenport, Tiefenbacher, Lutz, Novak, & Meyer, 2006; Russell, Koren, Rieder, & Van Uum, 2012) circumventing both the point estimates of salivary and plasma cortisol levels and the intra-individual day-to-day variability. Thus, hair cortisol concentration (HCC) serves as a biomarker of long-term, retrospective chronic stress exposure.

Hair cortisol research has grown rapidly in recent years. A meta-analysis primarily of research on adults found that stress-exposed groups in general exhibited a 22% increase in HCC (Stalder et al., 2017). Similarly, a systematic review of hair cortisol correlates among children concluded that sex and anthropometry are potential confounders, while age and socioeconomic status (SES) need further investigation (Gray et al., 2018). Though many studies contributed to the meta-analysis and systematic review, few focused on individuals belonging to a racial/ethnic minority in the United States. Stressful life experiences, perceptions, and responses are not evenly distributed across races/ethnicities, with individuals from lower SES neighborhoods and members of racial/ethnic minorities exposed to higher rates. Ethnic minorities and low-income women tend to report more chronic stress and less social support from family and friends, which may influence coping mechanisms (Bandoli et al., 2016). The etiology of racial/ethnic differences in stress-induced HPA alterations among pregnant women has been the subject of recent investigation. After stratifying by race/ethnicity, greater lifetime exposure to traumatic life events was associated with greater HCC during pregnancy among Black women only (Schreier et al., 2016). Similarly, pregnant Black women experiencing physical and/or sexual abuse in childhood had higher levels of HCC than women who had not experienced abuse (Schreier, Enlow, Ritz, Gennings, & Wright, 2015). Cumulative psychological stress was associated with lower morning salivary cortisol levels among Black, urban pregnant women but not among Hispanic women (Suglia et al., 2010). Increased social support may explain the lack of association between stress and salivary cortisol rhythms among Hispanic/Latina women. Though much has been described in pregnant Black and Hispanic/Latina women, there is a lack of knowledge on the relation between stress and hair cortisol in Black and Hispanic/Latina maternal-child dyads.

We chose to assess HCC among urban mother–child dyads because of the evidence suggesting that, in close pairs, such as a mother and child, stressors experienced by one member may influence the other. In one of the first hair cortisol studies among mother–daughter dyads, chronic stress was associated with lower HCC in mothers while the mother–daughter HCC association became stronger as quality of parenting decreased (Ouellette et al., 2015). Other groups have examined the association between a mother’s and child’s HCC demonstrating conflicting results in whether maternal-child HCC was correlated (Boeckel, Viola, Daruy-Filho, Martinez, & Grassi-Oliveira, 2017; Olstad et al., 2016; Ursache, Merz, Melvin, Meyer, & Noble, 2017). For example, hair cortisol levels were strongly positively associated with each other among an Australian sample of Caucasian maternal-adolescent pairs (Olstad et al., 2016). In a socioeconomically diverse sample of parents and children from the Northeastern United States, parent and child HCC were not associated (Ursache et al., 2017). Among a Brazilian sample of females exposed to intimate partner violence (IPV) and their children, women reporting IPV had elevated HCC, while children who witnessed IPV did not have HCC different than controls, and there were no associations between HCC in mothers and children (Boeckel et al., 2017). Similarly, there are inconsistencies in maternal-infant or maternal-preschool HCC associations. While three studies demonstrated positive correlations among maternal-infant or maternal-preschooler HCC association, one study observed no correlation (Flom, St John, Meyer, & Tarullo, 2017; Kao, Doan, St John, Meyer, & Tarullo, 2018; Liu, Fink, Brentani, & Brentani, 2017; Liu, Snidman, Leonard, Meyer, & Tronick, 2016). With a handful of studies examining the association of HCC in primarily non-Hispanic, White parent-youth dyads, more work is needed to understand the parent-youth HCC association among Hispanic/Latinos.

A major question is whether HCC may be a useful measure of chronic stress experiences in mothers and their youth. A significant proportion of studies have examined the relationship between a subjective report of perceived stress and HCC with inconsistent results. For example, Kalra et al. demonstrated a positive association between perceived stress and hair cortisol among pregnant women while Duffy et al. did not (Duffy, Schminkey, Groer, Shelton, & Dutra, 2018; Kalra, Einarson, Karaskov, Van Uum, & Koren, 2007). In a non-pregnant Caucasian sample perceived stress was not associated with a mothers’ HCC, while a curvilinear association between perceived stress and HCC was observed among a pooled sample of Canadian adults, where HCC increased with higher perceived stress but decreased at the highest level of stress (Olstad et al., 2016; Wells et al., 2014). Therefore, we sought to further examine the relationship between perceived stress and HCC among a sample of mothers-youth dyads as they waited during an appointment with their child’s pediatrician.

Social support is a crucial factor in maintaining health and is an essential component of human wellbeing. According to the stress-buffering hypothesis, social support can serve as a buffer against stress and as a coping resource against the negative effects of adverse life events (Cohen & Wills, 1985). Several studies have observed positive associations between salivary cortisol and measures of perceived social support (Iob, Kirschbaum, & Steptoe, 2018). Furthermore, studies have documented strong social support among Latino communities (Campos et al., 2008). Research with the Puerto Rican community indicates they have greater vulnerability in relation to physical and mental health as compared to other Hispanic sub-groups. A qualitative study of Puerto Rican women describes lack of social support during pregnancy and raising children as single mothers are significant life stressors (Bermudez-Millan et al., 2011). Nevertheless, there is a dearth of research examining social support among Puerto Rican women and how its psychobiological effects can manifest via the HPA axis and HCC. Understanding whether perceived social support can modify the association between maternal perceived stress and HCC among Puerto Rican women is, therefore, a gap in the literature that we will address in this study.

Hair cortisol is a relatively new biological measure, with few reporting the association of youth and mothers’ cortisol levels among racial/ethnic minorities in the United States (Ursache et al., 2017). Therefore, examining the associations between psychosocial stress and HCC among Hispanic/Latina maternal/child dyads residing in an inner city in the United States who experience health disparities addresses a major gap. The Latina population in Hartford, CT is primarily Puerto Rican (78%, 2010 census); Hispanic/Puerto Ricans experience the greatest health disparities, lowest education level, and lowest incomes, and exhibit more adverse behaviors compared to other Hispanics (Hajat, Lucas, & Kington, 2000). In light of this limited literature, more work is needed to examine links between hair cortisol levels and psychosocial stress among women and children belonging to racial/ethnic minority groups. Therefore, this study has three main goals: 1) to examine the association between subjective measures of mother’s psychosocial stress and maternal hair cortisol, 2) to assess the association between child and maternal hair cortisol levels, and 3) to examine whether a mother’s perceived level of social support can modify either association mentioned above.

Materials and methods

Participants

A total of 176 parents/caregivers and their children were approached from a busy pediatric primary care center in Hartford, CT. Inclusion criteria included 1) parent/caregiver ≥18 years of age, 2) child between 7 and 14 years of age, 3) neither mother nor child on oral corticosteroids within 4 weeks prior to enrollment, 4) both mother and child hair length greater than 3 cm, and 5) both English-speaking. Seventy-seven dyads were not eligible to participate because they did not speak English (n = 34), child was <7 years or >14 years (n = 25), adult was not the parent/guardian (n = 7), had taken oral corticosteroids within 4 weeks (n = 4), hair <3 cm (n = 1), not enough time (n = 1) or for other unknown reasons (n = 2), and three had previously participated; one was eligible but chose not to participate, and 40 were not interested in participating. Fifty-nine dyads agreed to participate. All participants provided informed consent and children provided assent. This study was approved by the Institutional Review Board at Connecticut Children’s Medical Center, and data collection was conducted between August 2015 and February 2017. Because 97% of the caregivers were the child’s biological mother, the primary caregivers are subsequently referred to as “mothers.”

Procedure

Demographic data were collected on all dyads, including race/ethnicity, age, sex, maternal education, income, city of residence, and BMI (kg/m²). Maternal height and weight were collected via self-report, while child’s height and weight were obtained during the clinic visit using a stadiometer and digital scale. Anthropometry may be associated with HCC and, therefore, was included as a covariate (Gray et al., 2018; Stalder et al., 2017).

Measures

Mothers completed questionnaires on characteristics of the household and sociodemographic characteristics that were slightly modified from the one used in the Collaborative Study of the Genetics of Asthma (Blumenthal, Banks-Schlegel, Bleecker, Marsh, & Ober, 1995; Rosas-Salazar et al., 2016). They also completed the Perceived Stress Scale (PSS 4-item) (Cohen & Williamson, 1988), and the Multidimensional Scale of Perceived Social Support (MSPSS) (Dahlem, Zimet, & Walker, 1991). Children 7–14 years old completed a questionnaire on exposure to violence and on perceived neighborhood safety which was adapted from the Survey of Children’s Exposure to Community Violence (ETV) (Ramratnam et al., 2015).

Parental perceived stress

The PSS (Cohen & Williamson, 1988) is a 14-item instrument widely used for assessing the perception of chronic stress (Liu et al., 2016). We utilized the Short Form PSS-4 (Warttig, Forshaw, South, & White, 2013) which measures the degree to which respondents perceive their lives over the past 4-weeks as unpredictable, uncontrollable, and overburdened. The PSS-4 was developed for community samples with at least an eighth-grade reading level. We administered the 4-item short version during the single study visit. Items were answered on a 5-point Likert scale (0 = never to 4 = very often) based on how often a participant felt, e.g. “In the last month, how often have you felt that you were unable to control the important things in your life?” and “In the last month, how often have you felt that things were going your way?” Responses were summed, with two positively worded items reverse-coded, for a total score from 0 to 16 (Cohen & Williamson, 1988).

Maternal social support

The MSPSS (Zimet, Powell, Farley, Werkman, & Berkoff, 1990) measures a participant’s perceived level of support from three sources: Family, Friends, and a Significant Other. The scale is comprised 12 items with 4 items for each subscale, e.g. “There is a special person who is around when I am in need.” The MSPSS was completed by the mother. Items were answered on a 7-point Likert scale (1 = very strongly disagree to 7 = very strongly agree). Mean scores were calculated by summing across all 12 items then dividing by 12, for a total score from 1 to 7.

Neighborhood safety and exposure to violence

Exposure to violence was ascertained using a modified version of the Survey of Children’s ETV, validated for use in children 7 years and up. Children completed this interviewer-led survey and another questionnaire on their perceived neighborhood safety. The scale is comprised four questions that assess a child’s exposure to violence, e.g. “Have any of your family members been hurt by a violent act?” A Yes to any one of the four questions on exposure to violence was dichotomized into Yes – exposed to violence or No, not exposed. The neighborhood safety variable was comprised 10 items which assessed how safe a child feels in their community, e.g. “Have you ever been worried about your safety in your neighborhood?” Items were answered either Yes (1) or No (0) and summed, with a minimum of 0 or maximum score of 10.

Hair cortisol

Hair samples consisting of ∼100 hairs (minimum 15 mg) were taken from the posterior vertex of the scalp. The hair sample was tied off near the scalp end before cutting to indicate the scalp end and then stored in an envelope at room temperature until analysis. The proximal 3 cm of hair, reflecting the most recent 3 months of hair growth, was cut and weighed on an electronic scale and transferred to disposable screw-top glass vials. Hair samples were washed three times with isopropanol and allowed to dry completely. One mL of HPLC-grade methanol (MeOH) was added to each vial and cortisol was extracted overnight at room temperature. After the overnight extraction, MeOH was decanted into disposable glass tubes and extracts were dried at 37 °C in a vacuum centrifuge (Sorvall, Thermo Fisher Scientific, Waltham, MA) for at least 1 h. Extracts were reconstituted in 250 μL of phosphate buffered saline (PBS), pH 8.0 and vortexed. Cortisol quantification was conducted in duplicate along with standards and quality controls using the ELISA salivary cortisol kit according to the manufacturer’s instructions (Salimetrics, Carlsbad, CA). The interpolated concentrations were converted to picograms (pg) cortisol and then corrected to the mass (mg) of hair analyzed. As is typical for cortisol data (Staufenbiel, Penninx, de Rijke, van den Akker, & van Rossum, 2015), raw HCCs were not normally distributed and were log10-transformed to reduce skewness and kurtosis. These log-transformed values were used in all analyses. Intra- and inter-assay coefficients of variation for this assay are <10%. Fifty-nine mothers provided hair cortisol data while 58 children provided hair cortisol data. Two dyads were excluded from further analyses because the mother was not biological.

Analysis plan

A power analysis was conducted to determine a sample size (n = 50–67) that would provide 85% power to detect statistical significance of regression coefficients at the p < .05 level between maternal hair cortisol levels and perceived stress based on the literature (Kalra et al., 2007). Descriptive analyses were performed to summarize the demographics of the participants. The associations of interest included association between maternal HCC and mother’s perceived stress, association between child’s HCC (outcome) and mother’s HCC. Additionally, we were interested in the moderation effect of mother’s perceived social support on the association between maternal perceived stress and maternal HCC. Both unadjusted and adjusted association analyses were performed. In adjusted regression analyses, potential confounders that were associated with the outcome (p < .1) were included. Mother’s age, BMI, and education level, family income, child’s neighborhood safety, and child’s exposure to violence were considered as potential confounders for maternal HCC. Child’s age and BMI, mother’s education, family income, neighborhood safety, and exposure to violence were considered as potential confounders for child’s HCC. Neighborhood safety (p = .029) and exposure to violence (p = .084) were associated with mother’s HCC at the significance level of 10% and were adjusted for associations with mother’s HCC in adjusted analyses. Mother’s education (p = .069) and neighborhood safety (p = .037) were associated with child’s HCC at the significance level of 10% and were adjusted for associations with child’s HCC in adjusted analyses.

To evaluate the effect of mother’s perceived social support, a model that predicted the children’s hair cortisol levels from the mothers’ hair cortisol level, and the interaction of the mothers’ hair cortisol level and their social support was tested. The outcome was log-transformed to correct for non-normality. An association with a p value smaller than 5% was considered statistically significant. All the statistical analyses were performed in R version 3.5.0 (R Development Core Team, Vienna, Austria).

Results

Participating mothers were on average 36 (SD = 6) years of age, their children were 11 (SD = 2.0) years of age, and 75% of the children were female (Table 1). More than half of the dyads were Hispanic/Puerto Rican (59%) and resided in an urban city, Hartford, CT (66%). Thirty-six percent of mothers reported completing less than a high school education and 72% reported a family income of less than $30,000/year. The mean HCC for mothers was 27.8 pg/mg (SD = 30.8 pg/mg), while PSS scores averaged 4.7 (SD = 3.1) and social support scores averaged 5.5 (SD = 1.6). On average, mothers were obese (mean BMI = 32.2 ± 9.8) with 32% reporting a BMI > 30. The mean HCC for children was 28.5 pg/mg (SD = 31.7) with 22% reporting exposure to violence. On average, children were borderline overweight (mean BMI = 21.0 ± 6.0).

Table 1. Demographic characteristics of parental respondent and child.

	N = 59 Frequency (%) or mean ± SD^a
Parental respondent characteristic
Age (years)	36 ± 6
Relation to child
Biologic mother	57 (97%)
Legal guardian	2 (3%)
Race/ethnicity
Puerto Rican	35 (59%)
Non-Hispanic White	3 (5%)
African American	5 (9%)
Hispanic, non-Puerto Rican	13 (22%)
Other	3 (5%)
Maternal education level
Less than high school	21 (36%)
High school/vocational school	12 (20%)
Some college	14 (24%)
College graduate	12 (20%)
Income
<$15,000	19 (36%)
$15,000–$29,999	19 (36%)
$30,000–$49,999	8 (15%)
$50,000	7 (13%)
City of residence
Hartford	39 (66%)
Greater Hartford	20 (34%)
Body mass index (BMI)	32.2 ± 9.8
Perceived stress score (range 0–16)	4.7 ± 3.1
Multidimensional Scale of Perceived Social Support (range 1–7)	5.5 ± 1.6
Child characteristics
Age (years)	11 ± 2
Sex
Female	44 (75%)
Race/ethnicity
Puerto Rican	35 (59%)
Non-Hispanic White	3 (5%)
African American	7 (12%)
Hispanic, non-Puerto Rican	10 (17%)
Other	4 (7%)
BMI	21.0 ± 6.0
Birth order	2 ± 1

^aSD: standard deviation.

Mother’s perceived stress score was not significantly associated with mother’s HCC in unadjusted analysis (p = .124) nor under the model adjusted for neighborhood safety and exposure to violence (p = .241) (Table 2). We examined whether maternal perceived social support moderated the association between maternal perceived stress and HCC by including an interaction term between maternal perceived stress and social support. Under the model, mother’s perceived stress was not associated with mother’s HCC (p = .403) and the interaction with mother perceived stress on mother’s HCC was not significant (p = .469).

Table 2. Models to evaluate the association between mother’s HCC and mother’s perceived stress.

Model	Outcome	Predictor	β	95% CI	p
Unadjusted	Maternal HCC	PSS score	0.03	(−0.01, 0.07)	.095
Adjusted	Maternal HCC	PSS score	0.02	(−0.01, 0.06)	.206
		Neighborhood safety score	0.04	(−0.03, 0.12)	.279
		Violence (=Yes)	0.10	(−0.24, 0.44)	.555
Interaction	Maternal HCC	PSS score		(−0.09, 0.29)	.308
		Neighborhood safety score		(−0.03, 0.12)	.245
		Violence (=Yes)		(−0.24, 0.45)	.558
		MSPSS score			.485
		MSPSS score		(−0.11, 0.35)	.320
		PSS score × MSPSS score		(−0.04, 0.02)	.516

Child’s HCC was positively associated with mother’s HCC in unadjusted analysis (p = .002) and in an adjusted model accounting for mother’s education and neighborhood safety (p = .007) (Table 3). Mother’s perceived social support was not a significant moderator for the association between mother’s HCC and child’s HCC (p = .457) and mother’s perceived social support was not significantly associated with child’s HCC (p = .483). In the model with the interaction term, mother’s HCC became statistically insignificant (p = .133) likely due to a lack of power to fit the model. Lastly, maternal perceived stress was associated with child’s hair cortisol (p = .035), but not maternal hair cortisol.

Table 3. Models to evaluate the association between child’s HCC and maternal HCC.

Model	Outcome	Predictor	β	95% CI	p
Unadjusted	Child’s HCC	Maternal HCC	0.47	(0.19, 0.76)	.002
Adjusted	Child’s HCC	Maternal HCC	0.42	(0.14, 0.70)	.004
		Mother’s education			.022
		High school/vocational school	−0.12	(−0.45, 0.19)	.428
		Some college	0.34	(0.03, 0.66)	.032
		College graduate	−0.11	(−0.44, 0.20)	.463
		Neighborhood safety score	0.08	(0.01, 0.14)	.019
Interaction	Child’s HCC	Maternal HCC	0.60	(−0.00, 1.21)	.051
		Mother’s education			.075
		High school/vocational school	−0.11	(−0.44, 0.22)	.507
		Some college	0.32	(0.00, 0.65)	.048
		College graduate	−0.08	(−0.42, 0.25)	.622
		Neighborhood safety score	0.07	(0.01, 0.14)	.033
		MSPSS score			.691
		MSPSS score	0.04	(−0.06, 0.14)	.399
		Maternal HCC × MSPSS score	−0.02	(−0.08, 0.04)	.476

Effect sizes in bold are significantly associated with the outcome at the level p 0.05

Discussion

In this study of primarily Hispanic/Latina mother–child dyads, we observed a positive association between maternal and child HCC. We assessed mother–child dyads because recent findings suggest that stressors experienced by one member of a close pair may also influence the other member’s effect and reaction to physiological stress(Waters, West, & Mendes, 2014). Our data suggest that the significant association between maternal and child HCC observed in previous studies also exists among a sample of minority, primarily Hispanic mother/child dyads who participated in this study. The explanation for this observation may be due to genetic heritability or to maternal calibration of the child’s HPA axis in utero (Karlen, Frostell, Theodorsson, Faresjo, & Ludvigsson, 2013). Recent observations from twin studies demonstrated a strongly heritable component of HCC (72%) but no correlation between HCC and perceived stress (Rietschel et al., 2017). In this study, maternal and child HCC levels were strongly and positively associated reflecting shared genetic influences, environmental exposures, or both. Altogether, our study fills a gap by demonstrating maternal/child HCC association in an understudied Hispanic/Latina, low-income population that is disproportionately at-risk of adverse health outcomes.

In this study, maternal HCC was not associated with self-reported perceived stress. Others have found similar results (Olstad et al., 2016; Ursache et al., 2017). This may be because HCC and the PSS-4 were measuring two distinct constructs – the PSS-4 is an acute measure of proximal stress within the last four weeks while HCC, in this study, is a measure of the last 3 months of stress. The stress response is characterized by physiological and psychological changes, with two of the most prominent responses being activation of the HPA axis and negative subjective psychological states. In this study, we attempted to measure both of these response systems (via the PSS and hair cortisol) to gauge stress levels. Theoretically, physiological and psychological measures of stress should correlate as they represent the same construct; however, our study and several others demonstrate this is not the case (Rietschel et al., 2017). For example, Ursache et al. did not observe an association between parental HCC and parental perceived stress, even when using a composite measure of stress comprised the PSS-14 (Cohen, Kamarck, & Mermelstein, 1983), negative life events (Life Experiences Survey [LES] (Sarason, Johnson, & Siegel, 1978), and material deprivation. This lack of association between the physiologic stress-response and the psychological stress-response may be an example of “lack of pscyhoendocrine-psychological covariance,” which postulates that subjective psychological states mediate associations between stressors and the HPA axis (Schlotz et al., 2008). In support of this, HCC demonstrates considerable intra-individual stability that may not be responsive to proximate or less severe stress changes and likely represents a trait measure (Stalder et al., 2012). It is unclear whether and to what extent findings from a cross-sectional, short-term study explain the longer-term, integrated stress responses assessed by hair cortisol in this study. A recent meta-analysis suggests an unclear association between HCC and stress-related self-report data which may signal that significant HCC elevation, or suppression, may require chronic stress exposure above a certain threshold (Stalder et al., 2017). Finally, it is conceivable that the 4-item PSS is limited in its ability to capture the experience of psychosocial stress.

We also observed an association between maternal HCC and her child’s perceived level of neighborhood safety. This may reflect the ability of a mother to sense her child’s stress from the adversity that her child is experiencing. The mother–child bond might influence or regulate physiological responses to stress (Hibel, Granger, Blair, Cox, & Family Life Project Key, 2009). For example, low SES parents tend to be less sensitive and more authoritarian, resulting in children who may be more likely to develop insecure attachment orientations as compared to children in high SES environments (Anton, Jones, & Youngstrom, 2015). Though we did not measure levels of maternal-child attachment nor reflective parenting, it is conceivable that the degree of attachment and/or of reflective functioning could explain our observed association between a mother’s HCC and her child’s perception of neighborhood safety. It is also possible that both mother and child are aware of the dangers of the neighborhood and violence, but perhaps mothers are reacting to these dangers more physiologically. This may be because mothers are more acutely aware of potential consequences of residing in an unsafe neighborhood. Additionally, it is possible that the mothers’ physiological reactions to stress are a reflection of her concern for her child’s perception of neighborhood safety. Gibbons et al. demonstrated that African American children’s early reports of discrimination were associated with mothers’ depression and anxiety (Gibbons, Gerrard, Cleveland, Wills, & Brody, 2004). Future studies examining the association between maternal and child HCC should consider measuring the degree of attachment, reflective parenting, or parenting quality that may explain or account for the observed effects.

We observed an association between child’s perceived level of neighborhood safety, but not exposure to violence, with hair cortisol. This observation contrasts with other studies that found no significant relationship between HCC and childhood anxiety, depression, or internalizing symptoms (Gray et al., 2018) and between a child’s exposure to IPV and HCC (Boeckel et al., 2017). Nevertheless, our results support a study demonstrating increased HCC among fearful young children after entering elementary school (Groeneveld et al., 2013). Overall, study results between HCC and “stress” and/or trauma exposure in children are heterogeneous and may point to difficulties in defining and measuring stress. Children entering school for the first time displayed elevated HCC (Groeneveld et al., 2013), and lower SES has been associated with elevated HCC among Dutch children (Vliegenthart et al., 2016). Similarly, elevated HCC was positively associated with the experience of lifetime trauma in community-based children, adding to the growing body of literature suggesting that HCC is a biomarker of stress, in particular, chronic stress exposure (Simmons et al., 2016).

This study has a number of strengths, including its focus on a racial/ethnic group that bears a disproportionate burden of psychosocial stress, and the use of an integrated measure of long-term cortisol release. Limitations include the relatively small sample size, although our sample size should have provided 85% power to detect statistical significance of regression coefficients at the p < .05 level between maternal hair cortisol levels and perceived stress. Second, we did not assess categorical psychiatric diagnoses (i.e. depression, anxiety), and thus we cannot exclude their effects in our analyses. However, even after controlling for both post-traumatic stress disorder (PTSD) and depression, Schreier et al. found that exposure to stressful and traumatic life events was associated with significantly greater hair cortisol in pregnant women (Schreier et al., 2016). Third, this was a cross-sectional study and neither directionality nor causality can be inferred. Fourth, though hair cortisol is associated with cumulative exposure to stressors (Schreier et al., 2016), we did not investigate the history of stressful life events.

In conclusion, this study demonstrates that HCC is strongly and positively associated with an urban-dwelling, minority sample of mother/child dyads. In support of other studies, we observed maternal/child correlation in HCC which may be explained by (epi)genetic heritability and maternal calibration of the HPA axis or by environmental influences or, more likely, through a combination of both. Further, a child’s perception of an unsafe neighborhood was associated with both the mother’s HCC and child’s HCC, consistent with the theory of reflective parenting. Lastly, the observed association between maternal/child HCC was not modified by maternal social support. This lack of association and moderation may be explained by the low level of perceived stress reported by our sample, which would not require buffering by a social support system. Perhaps future studies should include a measure of resilience and its association with HCC. The perception of stress has been shown to vary by race/ethnicity, with levels of perceived stress, negative life events and pregnancy stress differing by race/ethnicity within income categories (Dunkel Schetter et al., 2013). Our observation of low perceived stress even when poor is consistent with the Hispanic/Latino paradox, which involves better health among low-income Latino individuals (Franzini, Ribble, & Keddie, 2001). Future studies should explore acculturation, immigration status, and place of birth, beyond simply self-identification as Hispanic or non-Hispanic, as variations in both may influence patterns of stress and poverty.

Disclosure statement

The authors report no conflicts of interest.

Additional information

Notes on contributors

Jessica P. Hollenbach

Jessica P. Hollenbach: Conceptualization, Methodology, Validation, Formal Analysis, Investigation, Resources, Data Curation, Writing, Visualization, Supervision.

Chia-Ling Kuo

Nicole Gherlone: Project Administration, Data Curation, Writing – Review & Editing, Investigation.

Jinjian Mu

Franklin Sylvester: Data Curation, Writing – Review & Editing, Investigation.

Meg Gerrard

Mary Ojukwu: Data Curation, Writing – Review & Editing, Investigation.

Nicole Gherlone

Chia-Ling Kuo: Formal Analysis, Software, Methodology, Writing – Original Draft, Writing – Review & Editing.

Franklin Sylvester

Jinjian Mu: Formal Analysis, Software, Methodology

Mary Ojukwu

Michelle M. Cloutier: Conceptualization, Methodology, Resources, Writing – Original Draft, Writing – Review & Editing, Supervision.

Michelle M. Cloutier

Meg Gerrard: Conceptualization, Methodology, Writing – Original Draft, Writing – Review & Editing, Supervision.