Abstract
The cortisol awakening response (CAR), the marked increase in cortisol secretion over the first 30–45 min after morning awakening, has been related to a wide range of psychosocial, physical and mental health parameters, making it a key variable for psychoneuroendocrinological research. The CAR is typically assessed from self-collection of saliva samples within the domestic setting. While this confers ecological validity, it lacks direct researcher oversight which can be problematic as the validity of CAR measurement critically relies on participants closely following a timed sampling schedule, beginning with the moment of awakening. Researchers assessing the CAR thus need to take important steps to maximize and monitor saliva sampling accuracy as well as consider a range of other relevant methodological factors. To promote best practice of future research in this field, the International Society of Psychoneuroendocrinology initiated an expert panel charged with (i) summarizing relevant evidence and collective experience on methodological factors affecting CAR assessment and (ii) formulating clear consensus guidelines for future research. The present report summarizes the results of this undertaking. Consensus guidelines are presented on central aspects of CAR assessment, including objective control of sampling accuracy/adherence, participant instructions, covariate accounting, sampling protocols, quantification strategies as well as reporting and interpreting of CAR data. Meeting these methodological standards in future research will create more powerful research designs, thus yielding more reliable and reproducible results and helping to further advance understanding in this evolving field of research.
Keywords
CAR; Measurement; Saliva; Adherence; Covariates; Guidelines
Introduction
Abnormal secretion of the glucocorticoid hormone cortisol as the final product of the hypothalamus-pituitary-adrenal (HPA) axis is considered a crucial factor in linking the experience of chronic psychosocial stress to adverse effects on health (Chrousos, 2009). Besides reactivity to acute stressors, changes to the circadian regulation of cortisol secretion are considered important in this context (Kondratova and Kondratov, 2012, Menet and Rosbash, 2011, Nader et al., 2010). An aspect of cortisol regulation that is of special interest to psychoneuroendocrinological (PNE) inquiry is the cortisol awakening response (CAR), which describes the marked increase in cortisol levels across the first 30–45 min following morning awakening (Clow et al., 2004, Clow et al., 2010, Elder et al., 2014, Kudielka and Wüst, 2010). The CAR was first systematically described in the mid-1990s (Pruessner et al., 1997) and soon gained attention as a favorable biomarker in PNE research due to several methodological advantages over previously employed cortisol assessment strategies (see Section 2). These advantages together with evidence showing unique associations of the CAR with psychosocial, psychiatric and health-related parameters have resulted in a rapid increase in publications over the past 15 years.
The CAR combines features of a reactivity index (response to awakening) with aspects tied to circadian regulation (occurring roughly at the same time every 24 h) making it a fascinating research topic. However, precisely these features also make accurate assessment of the CAR a challenging task. When relying on CAR data acquired by participants themselves (usually through saliva sampling), validity critically relies on participants closely following a timed sampling schedule, beginning with the moment of awakening. Inaccurate sample timing can occur easily and can substantially bias CAR estimates. Furthermore, a number of other methodological factors, such as accounting for covariates, the number and nature of study days and the timing of sampling, can markedly affect CAR data. While not all questions regarding the role and regulation of the CAR have been adequately solved until now, several careful investigations have examined the impact of methodological factors on accurate CAR assessment and have recommended strategies for dealing with them (described below). Unfortunately, such recommendations have not been widely implemented in published CAR research. Fig. 1b provides an overview of methodological characteristics of such studies, published between 2013 and 2014. It can be seen that the employed methodological standards varied widely between investigations, with a high number of studies falling short of previous recommendations for best practice in CAR research (e.g., objective control of sampling times: Broderick et al., 2004, Kudielka et al., 2003).
To address this, the International Society of Psychoneuroendocrinology (ISPNE) has initiated an expert panel to summarize relevant evidence and collective experience on methodological factors affecting CAR assessment. The goal of this initiative was to formulate clear consensus guidelines based on current knowledge for future studies in this evolving field of research. The present report summarizes the results of this undertaking. As a large proportion of CAR research uses salivary cortisol assessments in participants' domestic setting, a particular focus is put on methodological challenges in this research context. Given the importance of sample time accuracy, the first three sections are devoted to an in-depth discussion of this topic, including strategies to increase sampling accuracy by maximizing participant adherence. In the subsequent sections a range of further methodological factors are covered. In the final section, the derived consensus guidelines are outlined and explained.
Section snippets
Cortisol awakening response
The CAR is expressed as part of normal, healthy human circadian physiology. Deviations from a typical CAR pattern are assumed to mark maladaptive neuroendocrine processes. A general review of psychosocial, psychiatric and health-related correlates of the CAR is beyond the scope of this article (reviews: Chida and Steptoe, 2009, Clow et al., 2004, Fries et al., 2009, Kudielka and Wüst, 2010). However, as a prerequisite for interpreting such data, some distinct features of the CAR need to be...
Inaccurate sampling: prevalence and impact
The validity of CAR data critically relies on the temporal accuracy of saliva sampling across the post-awakening period. A typical sampling schedule involves taking a first sample immediately after awakening followed by repeated assessments at specified times, e.g., at 10 or 15 min intervals over the subsequent 30–60 min. Fig. 3a illustrates an exemplary CAR sampling schedule. Failure to comply with such a schedule can occur in multiple ways. In the following, we distinguish between participants...
Strategies for dealing with inaccurate sampling
The following sections describe available objective monitoring strategies for ambulatory CAR research, discuss ways for dealing with identified inaccurate data and look into potential strategies in lieu of objective measures...
Maximizing adherence
Irrespective of objective monitoring, it is expedient to work towards maximizing participant adherence. Such strategies are cost-efficient as they prevent data loss through the exclusion of inaccurately sampled data and increase data quality (i.e., fully adherent data are superior to statistically inferred/corrected data). Table 2 lists strategies for maximizing adherence in CAR research. Several of these strategies are derived from the authors' collective research experience, without formal...
Dealing with covariates
Researchers have to deal with the fact that hormone secretion is related to a large number of state and trait factors (Schlotz, 2011). Depending on the research context, these covarying factors may be considered confounders, mediators, moderators or direct variables of interest (Adam and Kumari, 2009, Kudielka et al., 2012, Schlotz, 2011). If a covariate is not of main interest, the most critical question is whether it confounds observed associations (Schlotz, 2011). Confounding is given when a...
Sampling times
Fig. 1b shows that the number of post-awakening samples has varied widely between CAR studies. While basic CAR research, which is usually conducted on smaller samples, has tended to employ protocols with 4–5 post-awakening samples (e.g., Edwards et al., 2001b, Pruessner et al., 1997, Wüst et al., 2000b), large-scale epidemiological research often only uses two sampling times (typically on awakening and 30–45 min post-awakening; Adam and Kumari, 2009). The choice about the number of sampling...
Summary and guidelines
The present review shows that in order to derive meaningful data from CAR research, attention needs to be paid to methodological detail to prevent the danger of obtaining biased results. The variant methodological standards employed in past research (see Fig. 1b) are thus likely to have contributed to the inconsistency observed in this field. Hence, it is strongly recommended that researchers follow the guidelines described here for obtaining valid CAR data in future research.