Assessing Systemic Stress in Otolaryngology: Methodology and Feasibility of Hair and Salivary Cortisol Testing

Abstract

Objective

Elevated systemic stress is a predictor of adverse health outcomes, and stress can be objectively quantified by cortisol concentration. Despite its utility, such testing is rarely performed in otolaryngology. This manuscript provides details on the principles, methodology, and feasibility of performing laboratory assessments of hair and salivary cortisol to inform researchers wishing to incorporate these novel tests in future otolaryngologic studies.

Methods

Participants were older adults with hearing impairment. One hair sample and eight saliva samples were collected. Feasibility of study design was assessed through rates of participation in hair and saliva sampling and protocol adherence for saliva collection. Area under the curve (AUC) was used to evaluate overall secretion, and cortisol awakening response (CAR) was used to evaluate the dynamic secretion response.

Results

From 9/1/2013 to 12/31/2013, 26/30 (86.7%) eligible participants agreed to hair sampling. All 30 subjects agreed to collect saliva, with 29 (96.7%) adhering to the collection protocol. Mean AUC was 401.2 nmol/L per hour, and CAR was 4.5 nmol/L.

Conclusions

Evaluating systemic stress in an otolaryngologic population using hair and saliva is feasible with acceptable participation and adherence. Repeat measurements over time will allow for evaluation of changes in systemic stress in relation to treatment.

Introduction

Systemic stress refers to the body’s nonspecific responses to attempt to adapt to a perturbation, and elevated stress can lead to an increase in allostatic load, which is the sum of all physiologic effort to compensate for perturbations caused by a stressor [1]. Chronic stress and allostatic load have been associated with the exacerbation or development of various disease states, including poor mental health, impaired cognitive function, cardiovascular disease, obesity, diabetes, chronic obstructive pulmonary disease, and poor pregnancy outcomes [1-3]. Various otolaryngologic populations with chronic disease, such as hearing loss, prolonged vestibular dysfunction, chronic sinusitis, recalcitrant tracheal or glottic stenosis, or advanced cancer, would likely benefit from research examining how these conditions contribute to chronic stress and broader health outcomes.

Allostatic load is often determined through measurement of systemic cortisol levels. Cortisol is a well-recognized, reliable biomarker of stress and allostatic load [1]. Current standard laboratory tests to determine cortisol levels use urine or blood samples; however, these media have significant disadvantages. A useful urine sample typically requires a 24-hour collection period, mandating that the collecting individual remain home or transport a large collection container throughout the day, and collection of blood is invasive and requires specially trained personnel. Fortunately, additional media for testing that address the disadvantages of conventional media have recently been identified. Testing of hair, the most novel medium, allows for determination of a person’s average cortisol level over the previous 3-month period, and saliva testing allows for determination of an individual’s real-time cortisol level. Both of these media offer the advantages of being non-invasive, easily collected with minimal instruction, and biochemically stable in a variety of environments.

These newer media have been used as research tools to determine stress levels in various medical specialties. However, to date, there are no reports of the use of hair cortisol testing in an otolaryngologic population, and there are no detailed descriptions of the protocols related to hair and salivary cortisol testing in the otolaryngology literature. Therefore, to inform researchers wishing to incorporate these novel tests in future otolaryngologic studies, we herein provide details on the principles, methodology, data analysis techniques, and feasibility of performing laboratory assessments of hair and saliva for cortisol analysis. Preliminary data were gathered as part of a larger prospective study (Studying Multiple Outcomes after Aural Rehabilitative Treatment [SMART] study) that is investigating the impact of hearing aids and cochlear implants on cognitive and social functioning in older adults.

Materials and Methods

Study Population

Participants were adults aged 50 years and older who were enrolled in the Studying Multiple Outcomes after Aural Rehabilitative Treatment (SMART) study, a prospective observational study aimed at evaluating the cognitive, social, and physical functioning of older adults with post-lingual hearing impairment before and after treatment with a hearing aid or cochlear implant. To qualify for the study, participants were required to be English-speaking, use verbal language as their primary means of communication, and to be receiving a hearing aid for the first time or with minimal prior use (<1 hour/day) or a first cochlear implant. Here were present feasibility data from our experience with cortisol testing using hair and saliva in 30 participants receiving hearing aids who enrolled in the SMART study. The institutional review board of Johns Hopkins University approved this study.

Cortisol Testing

Sample Collection

Both hair and saliva samples were collected from each participant. The details of the collection procedures, analysis methods, and useful documents for study administration can be found on a website for dissemination to interested otolaryngology researchers [4].

Hair Collection

A single sample was collected from each participant. Each sample consisted of approximately 50 strands of hair, and samples were taken from the posterior vertex of the scalp. Hair from the posterior vertex has been shown to have the lowest coefficient of variation (CV) (15.6%) compared to other areas of the scalp [5]. All samples were stored in a dry environment at room temperature until being shipped to the analysis laboratory.

Saliva Collection

Participants collected saliva samples at home using specialized swabs. Eight samples were collected over two days. To determine the diurnal pattern of secretion, samples were collected at specific times: 1) immediately after waking, 2) 30 minutes after waking, 3) before lunch, and 4) before dinner. Timing of the first two samples was vital to observe the change in cortisol secretion that occurs after waking, known as the cortisol awakening response (CAR). After collection, the samples were mailed to the coordinating center and stored in a dedicated refrigerator.

Sample Laboratory Analysis

Hair Analysis

A summary of the necessary components for hair analysis is shown in Table 1. At the laboratory, hair samples were processed and analyzed using a commercial high-sensitivity enzyme immunoassay (HS-EIA) kit (Salimetrics LLC, State College, PA, USA). For this laboratory, inter-assay CV for the control hair pool is 11.6%, and intra-assay CV is 1.9% for duplicates.

Table 1.

Specialized supplies and laboratory equipment required for salivary and hair cortisol testing

Supply type	Description	Product Name	Vendor
Salivary Testing
Saliva sample collector	Synthetic swab in a dual test tube (allows for centrifugation)	Sardstedt, Inc., Salivette Cotton Swab	Fisher Scientific (www.fishersci.com); Catalog number: NC0506065
Saliva analysis kit	High sensitivity salivary cortisol enzyme immunoassay kit	Salimetrics, Inc, HS Salivary Cortisol EIA Kit	Salimetrics (www.salimetrics.com); Item number: 1-3002
Hair Testing
Hair analysis kit	High sensitivity salivary cortisol enzyme immunoassay kit	Salimetrics, Inc, HS Salivary Cortisol EIA Kit	Salimetrics (www.salimetrics.com); Item number: 1-3002
Analysis Laboratory	Laboratory Location	Pre-analysis Preparation Required	Laboratory Equipment Needed
Salivary Testing	Johns Hopkins Bayview Medical Center Core Laboratory	Centrifugation to extract saliva from swab	Centrifuge; ELISA microplate reader, capable of reading 96-well plate
Hair Testing	Behavioral Immunology and Endocrinology Laboratory – University of Colorado	Isolation of the proximal 3 cm of hair sample; cortisol extraction in solvent	Ball mill; microcentrifuge; ELISA microplate reader, capable of reading 96-well plate

Abbreviations: HS, high-sensitivity; EIA, enzyme immunoassay; ELISA, enzyme-linked immunosorbent assay

Saliva Analysis

A summary of the necessary components for saliva analysis is shown in Table 1. At the laboratory, each cotton swab was centrifuged at 3500 RPM for 10 minutes to obtain a clear supernatant. Cortisol concentration in this supernatant was determined using a commercial HS-EIA kit (Salimetrics LLC) according to the manufacturer’s instructions. For this laboratory, inter-assay CV for the control pool is 8.2% and intra-assay CV is 5.0% for duplicates.

Data Analysis

For demographic and clinical characteristics of the study population, continuous variables were compared using Student’s t-test or one-way ANOVA, and categorical variables were compared using chi-square or Fisher’s exact test. A threshold of p <.05 was used to evaluate statistical significance. Data were analyzed using Stata, version 13.1 (Stata Corp., College Station, Texas).

Study participation was defined as agreement to take part in testing. Adherence to study protocol for saliva was defined as collecting all samples at home, recording all collection times, and returning the samples. Participation and adherence were calculated as proportions. Only participants with available values from cortisol analysis were included in calculations involving salivary cortisol data.

Area under the curve (AUC) was calculated using all four daily saliva samples to determine total cortisol secretory activity, in units of nmol/L per hour, according to the formula:

$$AUC=\sum _{i=1}^{\pi -1}(t_i\cdot \frac{(m_{(i+1)}+m_i)}{2})$$

with t_i denoting time distance between the ith sample and (i+1)th sample in minutes, m_i denoting concentration of the ith sample in nmol/L, and n denoting total number of samples [6]. Because total collection time per day varied between participants, ranging from 12-18 hours, raw AUC values were normalized to one hour by dividing by the number of hours elapsed between collection of the first and last samples to make the values directly comparable. The average of the two days was reported as AUC.

Two measures were used to determine the dynamic of the cortisol response, which is a measure of the sensitivity of the hypothalamic-pituitary-adrenal (HPA) axis. Cortisol awakening response (CAR) was calculated as the difference in concentration (nmol/L) between the first and second samples. Time between collection of the first and second samples was set at 30 minutes per the study protocol. However, due to slight variation in participants’ performance, actual collection time varied, ranging from 20-40 minutes; therefore, raw CAR values were normalized to 30 minutes to make the values directly comparable. Normalization was accomplished by multiplying the initial CAR value by the ratio of 30 minutes to actual collection time, as shown in the formula:

$$CAR=(m_2-m_1)\cdot \frac{30}{t_2-t_1}$$

with m₁ and m₂ denoting concentrations of the 1^st and 2^nd samples and t₁ (zero by definition) and t₂ denoting concentrations of the 1^st and 2^nd samples, respectively. The average of the two days was reported as CAR. If the time difference is <20 or >40 minutes for either day, the values for that participant were considered invalid and not used in data analysis. Responder status was determined by the direction and magnitude of the CAR. A responder was defined as a participant with a CAR at least +2.5 nmol/L on both days, as an increase of 2.5 nmol/L has been shown to be equivalent to one secretory episode [7]. A responder was a person considered to have adequate HPA axis sensitivity, while a mixed or non-responder may have an impaired HPA axis [7].

Results

From September 1, 2012 through December 31, 2013, 30 eligible participants were enrolled. Of those 30, 26 (86.7%) agreed to participate in hair testing. For saliva testing, all 30 eligible subjects agreed to participate, and 29 (96.7%) of those adhered to the home collection protocol by collecting and returning all samples.

Demographic and clinical characteristics for the 30 baseline participants are shown in Table 2. Mean age was 71.2 +/− 7.6 years, and men (n=17) were more highly represented than women (n=13). Overall, study participants were mostly White, college-educated, and high earners. Calculated cortisol parameters are shown in Table 3. The mean AUC level was 401.2 +/− 150.3 nmol/L per hour. The mean CAR level was 4.5 +/− 11.3 nmol/L. Eleven (50.0%) of the 22 participants were responders.

Table 2.

Demographic and clinical characteristics of current participants (N=30)^a

Characteristic	Number of Participants (%)b
Age, mean (SD), years	71.2 (7.6)
Sex
Male	17 (56.7)
Female	13 (43.3)
Race
White	28 (93.3)
Black	1 (3.3)
Asian or Pacific Islander	1 (3.3)
Education
Less than high school	2 (6.7)
High school grad or equivalent	0
Some college or more	28 (93.3)
Income, USD
<25,000	2 (7.4)
25,000- 74,999	5 (18.5)
≥75,000	20 (74.1)
Hypertension
Yes	17 (60.7)
No	11 (39.3)
Diabetes mellitus
Yes	4 (14.3)
No	24 (85.7)
Smoking Status
Past smoker	17 (58.6)
Non-smoker	12 (41.4)
Hearing Impairmentc
Normal (≤25 dB HL)	1 (3.3)
Mild (>25 to ≤40 dB HL)	15 (50.0)
Moderate (>40 to ≤60 dB HL)	14 (46.7)
PTA, mean (SD), dB HL	39.9 (9.2)

Abbreviations: SD, standard deviation; USD, United States dollars; PTA, pure tone average; dB HL, decibels re: Hearing Level

^aAll values are expressed as No. (%) of participants unless otherwise indicated. Hearing is defined by speech-frequency pure tone average (PTA) of thresholds at 0.5, 1, 2, and 4 kHz in the better hearing ear.

^bSome categories include a number of participants less than that total (N=30) if some participants did not answer or refused to answer, in which case (%) is based upon number of participants who answered the question.

^cNo patients with severe-to-profound hearing impairment (>60 dB HL).

Table 3.

Cortisol awakening response (CAR) and area under the curve (AUC) using salivary cortisol testing

Demographic/Clinical Characteristic	Area Under the Curvea nmol/L per hr, mean (SD) (N=28)d	P-value	Cortisol Awakening Responseb nmol/L, mean (SD) (N=22)e	P-value	Consistent Respondersc no. (%) (N=22)e,f	P-value
All Patients	401.2 (150.3)	-	4.5 (11.3)	-	11 (50.0)	-
Sex		.95				.079
Male	402.7 (151.8)		1.7 (11.5)	.15	4 (30.8)
Female	399.2 (155.0)		8.7 (10.2)		7 (77.8)
Hearing Impairmentg		.044		.12		.74
Normal (≤25 dB HL)	333.3 (-)h		-i		-i
Mild (>25 to ≤40 dB HL)	367.0 (142.4)		1.2 (11.6)		5 (41.7)
Moderate (>40 to ≤60 dB HL)	449.6 (158.3)		8.6 (9.9)		6 (60.0)

Abbreviations: SD, standard deviation; dB HL, decibels re: Hearing Level

^aNormalized to per hour (in order to make the values directly comparable).

^bNormalized to 30 minutes (in order to make the values directly comparable), excluding those individuals with especially short (<20 min) or long (>40 min) time spans between collection of 1st and 2nd daily sample on at least one day.

^cDefined as at least 2.5 nmol/L increase, which is equivalent to one secretory episode, at 30 minutes post-awakening on both sample collection days. Number (percentage) is based upon the number of consistent responders to non-responders within that subpopulation.

^dTwenty-eight participants qualified for area under the curve analysis, with 2 being excluded for invalid collection times (did not collect all samples on both days).

^eTwenty-two participants qualified for cortisol awakening response and responder status analysis, with 6 being excluded due to invalid collection times for the first two samples (less than 20 minutes or greater than 40 minutes) on at least one day.

^fNumber of patients with positive response on both days in the noted subgroup. Percentage given is the proportion of members within that subgroup who were consistent positive responders.

^gNo patients with severe-to-profound hearing impairment (>60 dB HL)

^hOnly 1 participant in this subgroup.

ⁱZero participants in this subgroup.

Discussion

Overall, 86.7% of eligible persons agreed to participate in hair testing, and 100% agreed to participate in saliva testing. Of those agreeing to collect saliva, 96.7% adhered to study protocol. Preliminary results demonstrate that cortisol analysis using hair and saliva in an outpatient otolaryngologic patient population is feasible and that reliable data can be collected. Additional information regarding study protocols and study participant handouts can be found on a website setup to disseminate information to researchers [4].

In the present study, hair and saliva were used to measure free cortisol, which is the physiologically active component [8]. Hair and saliva were chosen for multiple reasons. Salivary cortisol correlates closely with serum free cortisol (correlation coefficient: 0.91) [9] and gives real-time measurements, allowing for determination of the secretory pattern over the day (diurnal variation). Additional advantages of saliva include ease of collection, minimal invasiveness, nominal expense (approximately 3-5 US Dollars per sample), temperature stability, and widespread availability of standardized assay kits [8]. However, disadvantages include the effect of acute stressors on real-time cortisol levels, requirement of patient compliance, and inability to determine chronic levels. A saliva cortisol level is analogous to a snapshot of the activity of the HPA axis, which is inherently volatile.

The inclusion of hair analysis to the study affords additional advantages: 1) it provides data on HPA axis activity prior to consent; 2) it provides an approximation of the total activity of the HPA axis for the past three months; 3) sample collection is performed by the study team; 4) no special storage requirements are necessary; and 5) the shelf life of hair for cortisol determination is essentially indefinite. Drawbacks include hair’s sensitivity to hair color processing, an inability to collect hair from individuals with shaved hair, concerns regarding a cosmetic defect, and unknown influences of differential growth rates. Despite these difficulties, initial research on hair cortisol has found values to be highly correlated with AUC (r=.50, n = 58, p <.001) [10]. Additionally, both saliva and hair can be shipped without specialized protocols, as these substances are considered non-infectious and not required to meet dangerous-goods regulations [11]. A summary of advantages and disadvantages of saliva and hair for cortisol analysis is shown in Table 4. It is important to remember that any medical condition that affects the HPA axis can confound results found on cortisol testing. Additionally, a number of oral medications can affect cortisol levels and should be considered when creating a research protocol, and, while the nature and effects of these medications is beyond the scope of this report, excellent reviews are available [12].

Table 4.

Pros and cons of saliva and hair cortisol analysis

Issue	Saliva		Hair
	Pros	Cons	Pros	Cons
Collection timing		Time of collection must be tightly controlled and accurately recorded	Time of collection, regarding time of day, not important
Potential for self-collection by subject	Easily accomplished			Not easily accomplished, but family member/friend can be instructed to collect
Storage temperature prior to shipping to laboratory		Ideally frozen at −20°C; storage at room temperature not recommended for >7 days	Can be maintained indefinitely at room temperature
Duration of storage and implication for biorepository		Up to a year of more, but requires freezing at −20°C or colder	Indefinitely (many years) at room temperature
Biohazard risk		Low to moderate	Low
Adequate sample “volume” collection	Generally easy to collect			Can be difficult to collect because of concerns regarding cosmetic impact or short hair
Subjects’ willingness to participate	Generally good		Generally good
Shipping temperature		Room temperature, but shipping time should be limited to <7 days	Room temperature regardless of shipping time
Miscellaneous sources of error	Level not influenced by flow rate			Differences in growth rates between individuals and ethnic groups
	Excellent quantification (<5%)			Inability to accurately collect the same length of hair affects quantification (+/− 10%)
		Subject adherence to collection times varies		Hair processing requires meticulous handing and protocol
	Availability of sample not affected by sex			Short hair length, typically in males, can preclude collection
		Steroid medications impact results		Steroid medications impact results
Validation of technique	Numerous			Few
Lab-to-lab reliability	Relatively good			Presently, relatively poor
Sampling frequency		Multiple times a day; varies with protocol	Single specimen provides data on previous three months
Hygiene effects		Tooth brushing or recent fluid intake can affect results		Hair color and shampooing may affect results
Response to chronic stressors	Yes, but generally with caveats		Yes, direction moderately reliable
Response to acute stressors	Excellent			Poor
Extant literature	>3,500 papers to date			<400 papers to date, but increasing
Cost		Expensive when numerous samples must be collected and processed, but not prohibitive	Relatively inexpensive since a single sample gives data on previous three months
Availability of laboratories to process samples	Many			Few
Potential marker of allostatic load		Yes, but requires frequent sampling	Yes, using single sample

Other samples used to measure cortisol include serum and urine; however, these were not chosen due to their disadvantages. Serum collection requires invasive blood draws, and handling and shipping require specialized protocols due to blood’s classification as a Category B Biological Substance (infectious agent) [11]. Regarding urine, measurement typically involves 24-hour collection, requiring all-day carriage of a large container. Furthermore, a 24-hour urine sample does not allow for determination of diurnal variation.

Cortisol data analysis requires determination of both total hormonal secretory output and the dynamic response, which is a measure of HPA axis sensitivity [13]. These two parameters have been differentially associated with various medical conditions, and their values can have entirely different meanings [14]. A comprehensive discussion of cortisol analysis is beyond the scope of this article, but a review on the subject is available by Fekedulegn et al [13]. To measure total secretion, we calculated area under the curve (AUC). Area under the curve has been widely used in previous studies, has acceptable day-to-day stability for an individual, and is easily calculated [6,13,14]. To assess the dynamic of the CAR, we calculated two parameters. First, we calculated the absolute change in cortisol level at 30 minutes post-awakening (approximate time to peak in morning cortisol [15]) compared to immediately after waking [7,14,16]. A previous study by Clow et al. [14] found the change in cortisol concentration from waking to 30 minutes in healthy adults was relatively consistent at 9.3 +/− 3.1 nmol/L. Wust et al. [7] defined an adequate CAR as at least +2.5 nmol/L, which is equivalent to one secretory episode [14]. Based upon this information, we additionally classified individuals as responders if they had a CAR of at least +2.5 nmol/L on both days.

A necessary aspect of salivary cortisol research is participant adherence to study protocol. A well-designed protocol should include clear instruction with a practice session, a simple-to-use collection method, clearly labeled collection materials, a protocol reminder card, and a simple method of returning the samples [17]. Overall, the process should be as clear, simple, and as burden free as possible to maximize adherence.

Specific issues in the present study were few and mostly minor. Most participants are willing to participate in saliva and hair sampling. Some subjects take extra time to return their samples; however, nearly all packets are returned after contacting the participants. Additionally, variability in waking and collection times presents some difficulty for data analysis; to compensate for this, we normalized our parameters (per hour for AUC and to 30 minutes for CAR) to make them directly comparable. Limitations of our methodology include an inability to determine the exact times of saliva collection and to determine if pre-collection protocol (e.g., not smoking or drinking prior to collection) was followed. Furthermore, some samples were collected outside of the time window specified in the protocol, resulting in disqualification of those data from analysis. To correct this, the importance of following the timing protocol is being emphasized further with the participants. Furthermore, collection devices that record the exact time of sample collection and even offer reminder alarms are commercially available and would help to alleviate this limitation.

Conclusion

This is the first study to demonstrate the broad feasibility of using hair and saliva cortisol testing in an otolaryngologic population. Importantly, we provide comprehensive details on collection, preparation, and analysis of hair and saliva samples on a website for use by other researchers and clinicians interested in utilizing cortisol testing.