Clinical Biochemistry 44 (2011) 665–668

Contents lists available at ScienceDirect

Clinical Biochemistry
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / c l i n b i o c h e m

Day-to-day variation of late-night salivary cortisol in healthy voluntaries

Gregori Casals ⁎, Laura Foj, María Jesús Martínez de Osaba
Service of Biochemistry and Molecular Genetics, Hospital Clinic of Barcelona, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Objectives: The study examines the components of biological variation of nocturnal salivary cortisol in
Received 7 November 2010 healthy subjects.
Received in revised form 3 February 2011 Design and methods: Eight repetitive measurements were performed in seven subjects during a 25-day
Accepted 5 February 2011 time period (study A), and then, for comparison, two salivary specimens were taken during two consecutive
Available online 21 February 2011
days from 20 subjects (study B). Salivary cortisol was measured with the Salimetrics HS-Cortisol assay.
Results: Mean salivary cortisol (1.27 nmol/L), analytical variation (CVa = 15.4%), within-subject variation
Keywords:
Salivary cortisol
(CVi = 34.1%), between-subject variation (CVg = 35.3%), index of individuality (II = 1.06) and reference
Cushing's syndrome change value (RCV = 104%) were obtained for study A. Similar results were obtained from the set of samples
Biological variation of study B.
Conclusion: The study results show a medium degree of individuality for salivary cortisol. Both
conventional reference values and comparison of serial results may be equally used for clinical interpretation.
A change greater of 104% between two successive measurements should be considered significant.
© 2011 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

Introduction Materials and methods

Salivary cortisol is an excellent indicator of plasma-free cortisol Subjects and samples

[1,2] increasingly used to assess hypothalamic–pituitary–adrenal axis
secretory activity and rhythm [3–6]. At present it is widely accepted Seven healthy volunteers (four females/three males; age 26–
that late-night salivary cortisol measurement is a simple and reliable 65 years) with no clinical features of Cushing's syndrome were
way to screen patients for Cushing's syndrome [7–11]. In fact, the requested to enroll in a serial testing study (study A). A sample of
Clinical Guideline Committee of the Endocrine Society recommends saliva collected between 23:30 h and 0:00 h should be provided two
the use of nocturnal salivary cortisol as a first step procedure in the times per week during four consecutive weeks. In a second study
diagnosis of Cushing's syndrome [12]. Two salivary cortisol measure- (study B), 20 additional healthy volunteers (13 females/7 males; age
ments per subject of different salivary collections are recommended 25–65 years) with no clinical features of Cushing's syndrome were
[12]. However, magnitude of day-to-day intra-individual and be- requested to provide two successive late-night saliva samples (two
tween-subject variations of late-night salivary cortisol concentrations consecutive evenings), an approach commonly used in clinical
in an outpatient setting is unknown. Data on the biological variations practice for screening of Cushing's disease. Volunteers were provided
of analyte concentrations are necessary for judging the usefulness of with a sheet with collection directions and cotton swab devices
conventional population-based reference intervals, correctly inter- (Salivette, Sarstedt, Nümbrecht, Germany) for saliva collection. The
preting serial laboratory results from a single patient in clinical participants were directed to collect saliva at least 2 h after eating,
practice and objectively determining the analytical goals required to smoking or brushing their teeth. Samples were stored at 4 °C
facilitate optimal patient care [13–15]. overnight and frozen next morning at − 20 °C upon arrival at
The aim of the present study was to determine the within- and laboratory. During the period of study any change in lifestyle habit
between-subject biological variation of late-night salivary cortisol by (smoking, drugs, diet, stress) was kept to minimum. Studies were
using individual means obtained from day-to-day data in healthy conducted in accordance with the guidelines of the declaration of
subjects. Helsinki and informed consent was obtained from all participants.

Salivary cortisol measurements

⁎ Corresponding author at: Service of Biochemistry and Molecular Genetics, Hospital

At the end of the collection period, all frozen samples were
Clinic Universitari, Villarroel 170, Barcelona 08036, Spain. Fax: +34 93 22775697. thawed, mixed, centrifuged at 3000 g for 10 min at room temperature
E-mail address: [email protected] (G. Casals). and analyzed in duplicate with a competitive immunoassay [16]

0009-9120/$ – see front matter © 2011 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.clinbiochem.2011.02.003
666 G. Casals et al. / Clinical Biochemistry 44 (2011) 665–668

specifically validated for the quantitative measurement of salivary

cortisol (Salimetrics LLC, State College, Pennsylvania). The analytical
sensitivity was 0.08 nmol/L and the calibration range was 0.33–
82.8 nmol/L. Inter-assay coefficient of variation was 18% (2.8 nmol/L)
and 11% (28 nmol/L). The analytical sensitivity was determined by the
manufacturer by interpolating the mean minus two standard
deviations for 10 sets of duplicates at 0 nmol/L standard. Determina-
tion of inter-assay coefficient of variation was performed in our
laboratory by analyzing two levels of internal quality control material
in 15 runs.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 5 (Graph-

Pad Software Inc.) and Excel 2008 (Microsoft Corp., USA). Investigation Fig. 1. Range of salivary cortisol concentrations in seven healthy subjects, measured by
for outlier results for each individual's set of samples was performed by serial analysis with salivary collection between 23:30 and 00:00 over a 4-week period.
n = 8 values for each subject. The empty boxes correspond to female individuals (n = 4;
comparing each subject standard deviation to the mean of all standard
subjects 1, 4, 6 and 7). The filled boxes correspond to male individuals (n = 3; subjects
deviations± 3 SD. Cochran's variance test was used to investigate if any 2, 3, 5) (Study A).
participant was an outlier from the group [13,17]. Based on these tests,
one sample from a male subject was excluded. Then, different sets of
data were taken to different calculations that are used in the clinical between mean salivary cortisol for women (1.44 ± 0.39 nmol/L)
laboratory to estimate the biological variation and to facilitate the compared with men (1.32 ± 0.14 nmol/L).
interpretation of the results. This included the calculation of the The means and the components of biological variation of salivary
analytical coefficient of variation for cortisol in saliva (CVa), the cortisol found in studies A and B, expressed in terms of CV, are shown
within-subject coefficient of biological variation (CVi), the between- in Table 1. Also shown in Table 1 is the index of individuality (II) and
subject coefficient of variation (CVg), the index of individuality (II) and the reference change value (RCV). Table 2 displays the analytical goals
the reference change value (bilateral, RCV 95%) according to the for salivary cortisol based on the components of the biological
approach of Fraser and Harris [13,14]. Analytical variance (SDa) was variation obtained from studies A and B.
calculated from the differences between the duplicates according to the
formula: Sa2 = Σd2/2N, where d is the difference between duplicates Discussion
and N is the number of duplicates. One-way analysis of variance was
used to distinguish the total variance into between-subject variance Biochemical screening studies for Cushing's syndrome have
(S2g) and total within-subject variance (S2i + a). Biological within- traditionally included low-dose dexamethasone suppression testing,
subject (intra-individual) variation (S2i) was estimated from the total 24-h urine free cortisol measurement and evaluation of diurnal
within-subject variance (S2i + a) minus analytical variance, according to rhythmicity [8]. These studies are often cumbersome and may require
the formula: S2i = S2i + a – S2a. The analytical, within- and between- hospitalization. At present, late-night salivary cortisol measurement is
subject biological variations were expressed as coefficients of variation also recommended as a first step procedure in the diagnosis of
(CVa [%], CVi [%] and CVg [%], respectively). The reference change value Cushing's syndrome [12]. It is a much more practical way of screening
(RCV) or critical difference is the minimal significant difference for Cushing's syndrome since it can be easily performed on an
(p b 0.05) between two consecutive measurements in the same subject. outpatient basis without disrupting a normal routine. In addition, the
It was calculated according to the following formula: RCV(%) =
2.77 × (CV2a + CV2i)1/2. The index of individuality (II), CVi + a/CVg,
describes the relationship between within-subject and between-subject
variation and it was used to evaluate the usefulness of population-based
reference values [15].
We also calculated the desirable quality specifications for
imprecision (I), bias (B), and total error (TE), which were calculated
using the formulas [18]: I b 0.5CVi; B b 0.25(CV2i + CV2g)1/2 and
TE b 1.65 × I + B (α b 0.05).

Results

Salivary cortisol means and absolute ranges (minimum and

maximum values) of the seven subjects that provided eight salivary
samples (study A) are shown in Fig. 1. The median age (range) of
subjects was 29 years (26–65). Mean salivary cortisol concentrations
ranged from 0.69 to 2.01 nmol/L for subjects, with an overall mean of
1.27 nmol/L. There was no significant differences between mean
salivary cortisol for women (1.21 ± 0.28 nmol/L) compared with men
(1.46 ± 0.30 nmol/L).
Salivary cortisol means and absolute ranges (minimum and
maximum values) of the 20 subjects that provided two consecutive
salivary samples (study B) are shown in Fig. 2. The median age (range)
Fig. 2. Range of salivary cortisol concentrations in 20 healthy subjects measured in two
of subjects was 31 years (25–65 years). Mean salivary cortisol consecutive salivary collections obtained between 23:30 and 00:00. Female subjects
concentrations ranged from 0.36 to 3.42 nmol/L for subjects, with (n = 13) correspond to subjects 3, 4, 6–13, 17, 18, 20. Male subjects (n = 7) correspond
an overall mean of 1.35 nmol/L. There was no significant differences to subjects 1, 2, 5, 14–16, 19 (Study B).
 G. Casals et al. / Clinical Biochemistry 44 (2011) 665–668 667

Table 1 Table 3
Mean values, estimated mean analytical (CVa), intra-individual (CVi), and interindi- Within-subject variation (CVi), between-subject variation (CVg) and index of
vidual (CVg) variation, and derived indices for salivary cortisol in healthy volunteers. individuality (II) values obtained for cortisol in previous studies.

Mean (nmol/L) CVa (%) CVi (%) CVg (%) II RCV (%) CVi (%) CVg (%) II Reference

Study A 1.27 15.4 34.1 35.3 1.06 103.6 Serum 22 46 0.47 [21]
Study B 1.35 14.6 32.1 38.1 0.93 98.6 Serum 27 53 0.50 [22]
Serum 21 45 0.45 [14]
Study A: in seven subjects, determined eight times from samples collected between
Serum 37 21 1.91 [20]
23:30 and 00:00 during 4 weeks.
Saliva 6 21 0.36 [19]
Study B: in 20 subjects, determined twice from samples collected between 23:30 and
Serum (evening) 103 36 2.89 [20]
00:00 on two consecutive days.
Saliva (late-night) 34 35 1.06 This study
CVa: analytical coefficient of variation; CVi: within-subject coefficient of variation; CVg:
between-subject coefficient of variation; II: index of individuality; RCV: reference
change value.
study for late-night salivary cortisol (CVi = 34.1%, CVg = 35.3%,
II = 1.06). Beyond the higher concentrations of morning salivary
saliva collection is a non-invasive sampling procedure that avoids the cortisol compared to late-night salivary cortisol, these results may
stress-induced rise in adrenal secretion associated with blood indicate a different degree of within- and between- biological
sampling. Therefore, salivary cortisol measurements are increasingly variation in the cortisol acrophase or nadir. In fact, a similar behavior
used on a routine basis. However, there is a lack of knowledge has been observed for serum cortisol in a study performed in 11
regarding significant data required for correctly interpreting late- healthy females. In this study, within-subject biological variation of
night salivary cortisol laboratory results, such as the degree of day-to- serum cortisol in the same subjects was much greater in the evening
day intra-individual variation or the degree of inter-individual than in the morning [20].
variation. The reference change value (RCV) is a parameter useful to
The index of individuality (II) is a means to assess the usefulness of determine how much is the significant difference between two
conventional population-based reference intervals. When this index is sequential laboratory results in the same subject. In our study, RCV for
lower than 0.6, the analyte displays high individuality and a single result salivary cortisol was close to 100%. This means that if a salivary
compared with the reference interval is considered to have low cortisol concentration more than twice than the previous result is
diagnostic value. In contrast, when the II is higher than 1.4 the analyte obtained in the follow-up of a patient, this change is greater than the
shows low individuality and reference intervals are considered to be variance explained by analytical and biological variance and,
clinically useful [14,15]. In the present study (study A), intra-individual therefore, should be considered a true change. However, it is
variation (CVi), inter-individual variation (CVg) and the index of important to remark that day-to-day within-subject variance is also
individuality (II) for late-night salivary cortisol were 34.1%, 35.3% and affected by preanalytical factors. In our both studies (A and B), the
1.06, respectively. In comparison, serum cortisol is reported to have a healthy volunteers received the same instructions given to the
lower CVi but a higher CVg (Table 3), resulting in a lower index of patients in a routine clinical setting. However, given the character-
individuality (b0.6). Thus, although for serum cortisol serial results may istics of salivary collection, patients acquire a central role in
be considered more useful than population-based reference intervals on preanalytics aspects (compliance with time of sampling and storage
the basis of an II b 0.6, for late-night salivary cortisol (II = 1.06) there conditions) that are difficult to check for the laboratory and that may
could not be assumed a better utility of the evaluation of serial results in increase day-to-day variation in the clinical setting if appropriate
the same patient over population-based reference ranges (or vice mechanisms to assure the patient's comprehension of the instructions
versa). are not provided.
Because the hypercortisolism of Cushing's syndrome can be Within-subject variation is useful to establish analytical goals for
variable and in order to increase confidence in the test results, most methods used for monitoring patients in clinical practice. For
clinicians using the late-night salivary cortisol test request patients imprecision (disagreement between replicates), it has been suggested
to collect a saliva sample on two separate evenings [12]. For this that a desirable performance can be defined as CVa b 0.5 × CVi [18]. In
reason we performed a second study (study B) in 20 patients who our study, CVa/CVi was b0.5, indicating that the analytical variation in
were asked to provide two consecutive late-night saliva samples. the method employed was adequate for use in measurements of
Similar results than those obtained in study A were obtained in healthy subjects. Bias (systematic measurement error with respect to
study B, thus supporting that within-subject and between-subject a reference quantity value) and total error for late-night salivary
variations for late-night salivary cortisol were of a similar degree cortisol measurements were also obtained (Table 2).
and confirming the existence of a medium degree of individuality The relatively low number of subjects included may be considered
for salivary cortisol (II between 0.6 and 1.4). as a limitation of the study. However, results obtained in the set of
A study of the biological variation of first morning salivary cortisol subjects of study A were similar than those obtained in the set of
reported a within-subject variation (CVi) of 6.3%, between-subject subjects of study B. The upper limit of the reference interval reported
variation (CVg) of 20.5%, and an index of individuality (II) of 0.36 [19]. for midnight levels of salivary cortisol for the assay used in our study
It represents a higher degree of individuality than the obtained in our is 4.3 nmol/L [16]. With other assays, upper limits of reference interval
in adults and cutoffs based on receiving operator curves are variable,
Table 2
ranging from 2.2 nmol/L to 15.2 nmol/L with sensitivities and
Desirable quality specifications for late-night salivary cortisol. specificities for the diagnosis of Cushing's syndrome mostly higher
than 90% [3]. However, our results suggest that increments greater
Imprecision (%) Bias (%) Total error (%)
than 100% between two measurements within a short follow up
Study A 17.0 12.3 40.4 (4 weeks) may indicate a true increase of cortisol levels regardless of
Study B 16.0 12.5 38.9
cortisol concentrations are above or below a fixed cut-point.
Study A: in seven subjects, determined eight times from samples collected between Therefore, these increments may be also taken into consideration
23:30 and 00:00 during 4 weeks. when considering the diagnosis of Cushing's disease and should be
Study B: in 20 subjects, determined twice from samples collected between 23:30 and
00:00 on two consecutive days.
especially helpful in the biochemical assessment of the recurrence of
Imprecision (I), bias (B) and total error (TE) were calculated according to the formulas: Cushing's disease as well as in the detection of cyclic Cushing's
I b 0.5CVi; B b 0.25(CV2i + CV2g)1/2 and TE b 1.65 × I + B. disease. However, it is important to remark that intra-individual
668 G. Casals et al. / Clinical Biochemistry 44 (2011) 665–668

variability of salivary cortisol might be different in Cushing and/or [8] Papanicolaou A, Mullen N, Kyrou I, Nieman LK. Nighttime salivary cortisol: a useful
test for the diagnosis of Cushing's syndrome. J Clin Endocrinol Metab 2002;87:
pseudo-Cushing's patients compared to healthy people. Assessment 4515–21.
of biological variation in these patient populations remains to be [9] Elamin MB, Murad MH, Mullan R, Erickson D, Harris K, Nadeem S, Ennis R, Erwin
established. PJ, Montori VM. Accuracy of diagnostic tests for Cushing's syndrome: a systematic
review and metaanalyses. J Clin Endocrinol Metab 2008;93:1553–62.
In summary, we obtained biological variation data for late-night [10] Viardot A, Huber P, Puder JJ, Zulewski H, Keller U, Müller B. Reproducibility of
salivary cortisol. The degree of individuality measured in late-night nighttime salivary cortisol and its use in the diagnosis of hypercortisolism
salivary cortisol is such that both conventional population-based compared with urinary free cortisol and overnight dexamethasone suppression
test. J Clin Endocrinol Metab 2005;90:5730–6.
reference ranges and comparison of serial results by means of the [11] Trilck M, Flitsch J, Ludecke DK, Jung R, Petersenn S. Salivary cortisol measurement
reference change values may equally be used to identify major —a reliable method for the diagnosis of Cushing's syndrome. Exp Clin Endocrinol
alterations in individual subjects. In the later case, changes greater Diabet 2005;113:225–30.
[12] Nieman LK, Biller BM, Findling JW, Newell-Price J, Savage MO, Stewart PM,
than 104% between two consecutive measurements may be consid-
Montori VM. The diagnosis of Cushing's syndrome: an Endocrine Society Clinical
ered significant. Practice Guideline. J Clin Endocrinol Metab 2008;93:1526–40.
[13] Fraser GC, Harris EK. Generation and application of data on biological variation in
clinical chemistry. Crit Rev Clin Lab Sci 1989;27:409–37.
[14] Fraser CG. Biological Variation: From Principles to Practice. 2101 L Street, NW,
References Suite 202, Washington DE 20037–1558: AACC Press; 2001.
[15] Fraser CG. Inherent biological variation and reference values. Clin Chem Lab Med
[1] Arafah BM, Nishiyama FJ, Tlaygeh H, Hejal R. Measurement of salivary cortisol 2004;42:758–64.
concentration in the assessment of adrenal function in critically ill subjects: a [16] Raff H, Homar PJ, Skoner DP. New enzyme immunoassay for salivary cortisol. Clin
surrogate marker of the circulating free cortisol. J Clin Endocrinol Metab 2007;92: Chem 2003;49:203–4.
2965–71. [17] Cochran WS. The distribution of the largest of a set of estimated variances as a
[2] Dorn LD, Lucke JF, Loucks TL, Berga SL. Salivary cortisol reflects serum cortisol: fraction of their total. Ann Eugen 1941;11:47–52.
analysis of circadian profiles. Ann Clin Biochem 2007;44:281–4. [18] Fraser CG, Petersen PH. Analytical performance characteristics should be judged
[3] Raff H. Utility of salivary cortisol measurements in Cushing's syndrome and against objective quality specifications. Clin Chem 1999;45:321–3.
adrenal insufficiency. J Clin Endocrinol Metab 2009;94:3647–55. [19] Chiu SK, Collier CP, Clark AF, Wynn-Edwards KE. Salivary cortisol on ROCHE
[4] Putignano P, Toja P, Dubini A, Pecori Giraldi F, Corsello SM, Cavagnini F. Midnight Elecsys immunoassay system: pilot biological variation studies. Clin Biochem
salivary cortisol versus urinary free and midnight serum cortisol as screening tests 2003;36:211–4.
for Cushing's syndrome. J Clin Endocrinol Metab 2003;88:4153–7. [20] Hansen AM, Garde AH, Skovgaard LT, Christensen JM. Seasonal and biological
[5] Martinelli Jr CE, Sader SL, Oliveira EB, Daneluzzi JC, Moreira AC. Salivary cortisol for variation of urinary epinephrine, norepinephrine, and cortisol in healthy women.
screening of Cushing's syndrome in children. Clin Endocrinol (Oxf) 1999;51: Clin Chim Acta 2001;309:25–35.
67–71. [21] Maes M, Mommen K, Hendrickx D, Peeters D, D'Hondt P, Ranjan R, De Meyer F,
[6] Yaneva M, Mosnier-Pudar H, Dugue MA, Grabar S, Fulla Y, Bertagna X. Midnight Scharpé S. Components of biological variation, including seasonality, in blood
salivary cortisol for the initial diagnosis of Cushing's syndrome of various causes. J concentrations of TSH, TT3, FT4, PRL, cortisol and testosterone in healthy
Clin Endocrinol Metab 2004;89:3345–51. volunteers. Clin Endocrinol (Oxf) 1997;46:587–98.
[7] Nunes ML, Vattaut S, Corcuff JB, Rault A, Loiseau H, Gatta B, Valli BN, Letenneur L, [22] Statland BE, Winkel P, Killingsworth M. Factors contributing to intra-individual
Tabarin A. Late-night salivary cortisol for diagnosis of overt and subclinical variation of serum constituents: 6. Physiological day-to-day variation in
Cushing's syndrome in hospitalized and ambulatory patients. J Clin Endocrinol concentrations of 10 specific proteins in sera of healthy subjects. Clin Chem
Metab 2009;94:456–62. 1976;22:1635–8.