Feasibility and Methodologic Considerations to Hair Collection for Cortisol among Youth Experiencing Homelessness

Abstract

Background:

Investigations of chronic physiologic stress measured by hair cortisol are rapidly expanding among community samples of adolescents and adults. However, research examining physiologic stress among youth experiencing homelessness is nascent despite the youths’ increased risk for adverse exposures and subsequent impaired mental health.

Objective:

This paper aimed to examine the feasibility of collecting hair for measuring cortisol among diverse youth experiencing homelessness and gain an understanding of variation in participation.

Methods:

Analysis of survey and hair participation data from three pilot studies among youth experiencing homelessness was conducted. Survey measures included sociodemographic characteristics (age, race and ethnicity, sex assigned at birth, and sexual orientation) and reasons for nonparticipation. Descriptive analysis examined participation rates in hair collection for cortisol measurement, including sociodemographic differences in participation.

Results:

Participation in the hair sampling for cortisol was high for the combined sample (88.4%), with some variation across the three pilot studies. Insufficient hair for cutting was the most common reason for not participating; Black and multiracial youth, as well as male youth, had a higher prevalence of nonparticipation.

Discussion:

The collection of hair for cortisol research among youth experiencing homelessness is feasible, and integration of physiologic measures of stress into research with this vulnerable population should be considered, given their high risk for adversity and death by suicide and drug overdose. Methodologic considerations and avenues for potential research are discussed.

The collection of hair for the stress hormone, cortisol, has become increasingly common in studies examining long-term physiologic stress and its relationships to youth mental health. (Ford et al., 2019; Kische et al., 2021; Shapero et al., 2019) As hair grows approximately 1 cm a month, the 1 cm of hair growth closest to the scalp approximates the cumulative cortisol output for the month preceding the collection (Meyer et al., 2014). Researchers typically assay up to 3 cm of hair as cortisol is thought to leach out of the hair over time due to environmental damage (e.g., frequent hair washing); thus, 3 cm of hair approximates 3 months of cumulative cortisol output, providing a longer-term assessment of physiologic stress.

Prior research found that the collection of hair for cortisol assessment was feasible in a large community sample of racially and socioeconomically diverse youth (Ford et al., 2016). However, the collection of physiologic stress measures among historically marginalized youth, including youth experiencing homelessness (YEH), is limited. YEH are at high risk for impaired mental health, death by suicide, and overdose due to the significant stressors and violence they face on a daily basis, and for many, the child abuse and neglect they experienced prior to leaving the home setting (Bender et al., 2015; Gewirtz O’Brien et al., 2020; Whitbeck et al., 2007). These adverse experiences place YEH at high risk for “toxic stress,” resulting in frequent or chronic activation of physiologic stress response systems, including irregularities in cortisol output (Shonkoff & Garner, 2012). In extant research, both low and high hair cortisol concentration (HCC) has been found to be associated with depression (Ford et al., 2019; Malisiova et al., 2021; Shapero et al., 2019; Staufenbiel et al., 2013) and low HCC associated with suicidal behavior (Melhem et al., 2017). Thus, in the future, hair cortisol may be a clinically useful biomarker to predict depression and suicide trajectories aiding in the treatment and prevention of impaired mental health.

However, YEH may be reluctant to participate in the collection of stress biomarkers, including hair for cortisol. Trust is often difficult to establish, given the youths’ significant history of trauma and violence. In addition, YEH report stigma, marginalization, and feeling devalued by larger society. In one recent study, nearly 50% of YEH reported experiencing discrimination due to their homelessness status, and discrimination overall was associated with increased psychological distress (Narendorf et al., 2022). In reviews of prior research examining research participation among people experiencing homelessness, mistrust and fears of stigma and negative interactions with research staff contributed to a reluctance to participate in research as well as engagement in interventions (Magwood et al., 2019; Parkes et al., 2022).

The collection of hair as a medium to examine physiologic stress may also lead to differential participation among YEH. For example, in our community sample of housed youth, Black adolescents were less likely to participate in the collection of hair than White adolescents (83.5% vs. 95.1%) due in part to concerns over privacy and the collection of biological material, not wanting to be touched, insufficient hair length (< 1 cm) to cut, or not wanting their hair cut because of braids or extensions (Ford et al., 2016). Research on physiologic stress with sexual and gender minority (SGM) youth and adults is nascent (Christian et al., 2021), particularly investigations of hair for cortisol. As Black and SGM youth are overrepresented among the YEH population `(Morton et al., 2018), an understanding of potential racial and SGM differences in participation in research collecting physiologic stress measures among YEH is necessary to inform future sampling strategies.

Due to the limited research on chronic physiologic stress among YEH, the current analysis aimed to examine the feasibility of collecting hair for cortisol among YEH. We focused primarily on participation rates in the collection of hair for cortisol among YEH, including sociodemographic differences in participation and reasons for nonparticipation. The findings of this study can inform the development of high-quality and representative protocols for collecting hair for cortisol assessment among youth experiencing homelessness.

Methods

Study Design and Sampling

In the current analysis, we examined data from three pilot studies that included the collection of hair for cortisol measurement among YEH aged 18–24 years. Study 1, conducted in 2018, utilized a cross-sectional design to examine the feasibility of collecting stress biomarkers and to explore the relationships between stress and mental health outcomes (N = 39). In Study 2, conducted in 2019, we investigated the feasibility of implementing a mindfulness virtual reality intervention and the effects of the intervention in reducing symptoms of anxiety and salivary cortisol concentration (N = 30; Clinical Trials Registration NCT04004520; Chavez et al., 2020). Study 3, conducted in 2020, was a feasibility study examining a modified nonrandomized Housing First intervention for youth experiencing homelessness to prevent opioid use disorder (N = 21; Kelleher et al., 2021). Four youths from Study 2 were excluded as they had also participated in Study 1. The final combined sample size for the current analysis was N = 86.

The three studies recruited participants from a drop-in center for YEH located in a metropolitan city in the Midwest. The drop-in center is the only center in the city serving youth ages 14–24 experiencing homelessness. The center provides food, laundry, shower facilities, counseling/case management, recreational activities, and interactions with other youth and staff. Convenience sampling was employed, and eligibility criteria for all three studies included: youth (a) aged 18 to 24 years; (b) met the criteria for homelessness as defined by the federal McKinney–Vento Act as “lacking a fixed, regular, stable, and adequate nighttime residence” and includes “living in a publicly or privately operated shelter designed to provide temporary living accommodations, or a public or private place not designed for, or ordinarily used as, regular sleeping accommodations for human beings” (National Center for Homeless Education, n.d.); and (c) spoke and understood the English language. In addition, the housing intervention feasibility study included only youth who did not meet diagnostic criteria for an opioid use disorder at the time of the study. For each study, research personnel verbally recruited and screened the youth at the drop-in center to determine eligibility and interest in study participation. Informed consent was obtained prior to study participation, and all studies were approved by the university institutional review board.

Data Collection

Data collection for the first two studies took place at the drop-in shelter in a private room, and collection for the third study took place at either the drop-in shelter or a mutually agreed upon location. For all studies, the research personnel conducted face-to-face surveys with the youth and cut approximately 10–75 mg of hair (~50 hairs) from the posterior vertex region as close to the scalp as possible with professional shears.

Measures

Outcome Measures

Participation in Hair Collection.

Participation was a dichotomous measure (yes/no) of providing a hair sample for examination of HCC.

Reasons for Nonparticipation in Hair Collection.

Reasons for nonparticipation were a multicategorical measure. Youth who chose not to participate were asked why they did not want to participate with the following response options: (a) I don’t feel I have enough hair to give; (b) I’m trying to grow my hair out; (c) I am concerned what you will test my hair for; (d) I am concerned with what you will do with my hair afterward; (e) Current hairstyle (weave or extensions) does not allow sample to be taken; (f) Hair is braided and prefer not to undo; (g) I just don’t want to; and (h) other, with an option to qualitatively provide a reason. Youth could select more than one reason for nonparticipation. In addition, youths who had less than 1 cm of hair were not able to participate in the hair collection due to the difficulty of cutting hair so close to the scalp.

Sociodemographic Characteristics

Age.

Age was a continuous measure ranging from 18 to 24 years.

Race and Ethnicity.

Race and ethnicity were presented as a multicategorical measure of youths’ racial and ethnic self-identity. Response options included African; Black or African American; Asian, Asian American, Pacific Islander; American Indian or Alaskan Native (American Indian category); Hispanic; White; unsure; or “other.” Youths who identified as African or Black/African American were grouped into a Black racial/ethnic category due to the small number of youths who identified as African. In addition, youth were able to identify with more than one racial and ethnic group, and a multiracial category was created. No youth identified only as American Indian or Alaskan Native nor unsure. The final categories for this analysis included Black (African/Black or African American), Asian (Asian, Asian American, Pacific Islander), Hispanic, White, multiracial, and “other.”

Sex at Birth.

Sex was a multicategorical measure of the youths’ sex assigned at birth. The three studies asked about sex differently. In Study 1, youth were first asked about the gender they identified with (male, female, nonbinary, unsure, and “other”; all youth reported either male or female), followed by a question that asked the youths if they identified with their sex or gender at birth (yes, no, or unsure; all youth reported yes). In Studies 2 and 3, youths were asked directly about their sex assigned at birth (male, female, intersex, or none of these describe me). Based on youth responses, three categories were created: male, female, and intersex. We explored a gender identity measure as all three studies asked the youths the gender they identified with; however, the frequency distribution between sex assigned at birth and gender identity for the two studies was nearly identical due to the small number of gender-diverse youth in the three studies.

LQB(T)Q Orientation.

LQB(T)Q orientation was a dichotomous measure (yes/no). The question was asked differently in the three studies. In two studies, the youths were asked to describe their sexual orientation with a list of response options (did not include gender identity). In the other study, the youths were asked if they identified as lesbian, gay, bisexual, transgender, or queer with a yes or no response option. Thus, for this analysis, youth who responded to any category other than heterosexual or who responded yes to identifying as LGBTQ were coded as yes to LGB(T)Q orientation.

Analytic Strategy

A series of univariate analyses were conducted on the combined sample of youths as well as on each individual study to identify participation rates in the hair sampling for cortisol concentration and reasons for nonparticipation. Sociodemographic characteristics were examined for the total sample of youth and stratified by participation versus nonparticipation in hair sampling.

Results

The sociodemographic characteristics of the combined sample and each individual study sample are presented in Table 1. The average age of the youths in the combined sample was 21.7 years—a majority of youths across the three studies identified as Black or multiracial. In the combined sample, 52.3% of youths reported male as their sex assigned at birth, 44.2% reported female, and 2.3% reported intersex. In the combined sample, 25.6% of the youths identified as LGB(T)Q, with variation found across the three individual studies (20.5% vs. 19.2% vs. 42.9%).

Table 1.

Sociodemographic Characteristics and Participation in the Collection of Hair for Cortisol, Combined and Individual Study Samples

	Total (N = 86)		Study One (N = 39)		Study Two (N = 26)		Study Three (N = 21)
	n	%	n	%	n	%	n	%
Race/ethnicity
Black	40	46.5	14	35.9	15	57.7	11	52.4
White	13	15.1	8	20.5	5	19.2	0	0
Multiracial	31	36.0	15	38.4	6	23.1	10	47.6
Hispanic	1	1.2	1	2.6	0	0	0	0
Other	1	1.2	1	2.6	0	0	0	0
Sex at birth
Male	46	53.5	20	51.3	15	57.7	11	52.4
Female	38	44.2	19	48.7	11	42.3	8	38.1
Intersex	2	2.3	0	0	0	0	2	9.5
Identifies as LGB(T)Q
Yes	22	25.6	8	20.5	5	19.2	9	42.9
No	64	74.4	31	79.5	21	80.8	12	57.1
Participated in hair collection
Yes	76	88.4	36	94.9	20	76.9	20	95.2
No	10	11.6	3	5.1	6	23.1	1	4.8
Reason for non-participation
Hair too short (< 1 cm)	7	70.0	3	100	4	66.7	0	0
Hair extensions	2	20.0	0	0	2	33.3	0	0
Scheduling issue	1	10.0	0	0	0	0	1	100
Youth concerned not enough hair to give	0	0	0	0	0	0	0	0
Youth trying to grow hair out	0	0	0	0	0	0	0	0
Youth concerned what hair will be tested for	0	0	0	0	0	0	0	0
Youth hair braided and does not want to undo	0	0	0	0	0	0	0	0
Youth does not want to	0	0	0	0	0	0	0	0
Other reason	0	0	0	0	0	0	0	0
	N	Mean (sd)	N	Mean (sd)	N	Mean (sd)	N	Mean (sd)
Age (minimum-maximum) 18–24 years)	86	21.7 (1.9)	39	21.9 (1.9)	26	21.4 (1.9)	21	21.9 (1.9)

Note. Other=youth selected only the “other” category for their racial/ethnic identity; LGB(T)Q=lesbian, gay, bisexual, transgender or queer; cm=centimeter

Participation in the hair sampling for cortisol assay was high for the combined sample (88.4%); however, variation was found between the three individual studies (see Table 1). Study 2 had the lowest participation at 76.9% (n = 20/26) compared to Study 1 at 94.9% (n = 36/39) and Study 3 at 95.2% (n = 20/21). Of the 10 youths who did not participate in hair sampling, insufficient hair length (< 1.0 cm) was the most common reason (n = 7/10), followed by hair extensions (n = 2/10), and a scheduling issue (n = 1/10).

Using the combined sample, sociodemographic differences between those who participated in the hair sampling and those who did not were examined (see Table 2). Statistical differences were not calculated due to the small sample sizes within the sociodemographic groups for those who did not participate. For race and ethnicity, 80% of Black youth (n = 32/38), 100% of White youth (n = 13/13), 93.5% of multiracial youth (n = 29/31), 100% of Hispanic youth (n = 1/1), and 100% of youths who identified as “other” (n = 1/1) participated in the hair sampling. For sex at birth, 82.6% of youths who reported male sex assignment (n = 38/46), 97.4% of youths who reported female sex assignment (n = 37/38), and 50% of youths who reported intersex assignment (n = 1/2) participated in the hair sampling. Last, 86.3% of youth who identified as LGB(T)Q (n = 19/21) and 89.1% who identified as heterosexual (n = 57/64) participated in the hair sampling.

Table 2.

Sociodemographic Characteristics of Participants versus Non-Participants in the Collection of Hair for Cortisol

	Participants (N=76)		Nonparticipants (N=10)
	n	Mean (SD)	n	Mean (SD)
Age (range 18–24 years)	76	21.9 (1.9)	10	21.4 (1.9)
	n	%	n	%
Race/ethnicity
Black	32	80.0	8	20.0
White	13	100.0	0	0
Multiracial	29	93.5	2	6.5
Hispanic	1	100.0	0	0
Other	1	100.0	0	0
Sex at birth
Male	38	82.6	8	17.4
Female	37	97.4	1	2.6
Intersex	1	50.0	1	50.0
Identifies as LGB(T)Q
Yes	19	86.3	3	13.6
No	57	89.1	7	10.9

Note. Other=youth selected only the “other” category for their racial/ethnic identity; LGB(T)Q=lesbian, gay, bisexual, transgender or queer

Discussion

The findings from the three pilot studies support the feasibility of collecting hair samples for cortisol among YEH. The most frequent reason for nonparticipation was insufficient hair length for collection. Although we could not reliably conduct significance tests for differences in participation by sociodemographic characteristics due to the small number of nonparticipating youth, our descriptive analysis found a higher proportion of Black, multiracial, and male youths did not provide a hair sample compared to White and female youths, respectively. These findings are consistent with prior research examining HCC in a community sample of housed adolescents (Ford et al., 2016) and reinforce the need for researchers to monitor the sociodemographic composition of their participant samples to ensure adequate representation and inclusivity. In addition, more research examining variation in HCC collected at different areas of the scalp is needed; in one small study of participants with extremely short-length Afro-textured hair, researchers found no significant differences in HCC in a pooled sample of hair collected at multiple sites across the scalp in comparison to hair cut from the posterior vertex region of the scalp (Doyle et al., 2019). The latter is the current recommendation and most standard location for cutting hair due to more even growth rates of hair and lower intra-individual variability in HCC (Sauvé et al., 2007). If HCC could be reliably collected at different sites on the scalp, then exclusion of participants due to short hair at the posterior vertex region of the scalp could be minimized.

While no youth across the three studies refused participation in the hair collection due to concerns about the purpose of the hair collection, prior research has reported YEH and Black or African American individuals expressed concerns related to trusting researchers, the research process, and institutions (George et al., 2014; Magwood et al., 2019). Strategies to promote research with persons experiencing homelessness and among Black or African American persons have been described in the extant literature; several aspects of the three pilot studies may have enhanced research engagement (George et al., 2014; Magwood et al., 2019). For example, the research team had an established collaborative relationship with the community partners at the drop-in shelter, a history of intervention and observational research conducted at the drop-in shelter promoting the health and well-being of YEH, training in motivational or psychosocial interviewing with vulnerable and marginalized populations, assurances that youth could withdraw from the study at any time without penalty, and face-to-face interviews that facilitated rapport with the youth. Additionally, our informed consent process reinforced that only cortisol would be assayed from the hair. Researchers conducting investigations with YEH should consider incorporating these recommendations along with concepts of trauma-informed care into all aspects of the research process to promote youth safety, empowerment, trust, and prevent retraumatization (Campbell et al., 2019; Isobel, 2021; Voith et al., 2020).

Conclusion

While our analysis is limited by the smaller sample size and convenience sampling, our findings support the feasibility of collecting hair for cortisol analysis among YEH. Our analysis also highlights methodologic considerations for study design and data collection to enhance the validity and generalizability of future research examining physiologic stress among YEH. Our lack of understanding of the biological effect of the extensive adversity that YEH face while living on the street as well as earlier in their childhood and adolescence precludes our identification of youth most vulnerable to poor outcomes. Changes in physiologic stress as mediators of intervention and as predictors of health and behavior are potential avenues for future research that may advance our understanding and, ultimately, improve the health and well-being of YEH.