Medication effects on salivary cortisol: Tactics and strategy to minimize impact in behavioral and developmental science

Summary

The non-invasive measurement of cortisol in saliva has enabled behavioral scientists to explore the correlates and concomitants of the interaction between the activity of the hypothalamic–pituitary–adrenal (HPA) axis, intrinsic factors, and social forces as they occur naturally in everyday life. The widespread integration of salivary cortisol into behavioral science has also revealed that omnipresent features of everyday life such as, over-the-counter and prescription medications, have the capacity to influence measurement validity. We identify several pathways by which pharmacologic agents could influence salivary cortisol, including (a) direct agonistic and antagonistic effects on the HPA axis, (b) indirect effects on physiological systems networked with the HPA axis, (c) moderation or mediation effects on cortisol secretion via pharmacologically induced change in subjective experience, (d) iatrogenic effects on the availability or composition of saliva, or the diffusion of serum constituents into oral fluid, and (e) cross-reactivities with antibodies used to detect cortisol by immunoassay. Specific medications with the capacity to influence salivary cortisol via these pathways are documented in an effort to procedurally and statistically minimize this potential source of error variance in the next generation of studies.

Introduction

The widespread integration of measures of biological processes into contemporary behavioral and developmental science has been driven, at least in part, by technological advances that enable a variety of analytes to be measured in oral fluid (Riad-Fahmy et al., 1982, Read, 1989, Hellhammer et al., 2009). Saliva specimens can be collected non-invasively in everyday social contexts and special circumstances with minimal interruption to the natural flow of activities and experience (e.g., Adam, 2006). Thus, in contrast to more traditional biological specimens (e.g., urine or plasma), analytic measurements in oral fluid seem to afford researchers an opportunity to gather measures of the confluence of biobehavioral processes during, and in response to, naturalistic, and contextually meaningful events with direct relevance to variation in individuals’ everyday interpersonal demands and experiences. Theorists speculate this embedded bio-social-ecological approach is a conceptual advance that will extend our understanding of why some individuals are placed at risk by, while others are resilient to, similar circumstances, adversities, and experience (e.g., Curtis and Cicchetti, 2003).

The behavioral and developmental literature directs empirical attention to socially oriented characteristics of the person–environment fit (e.g., childhood temperament, maternal caregiving, interpartner violence, poverty) as a primary source of individual differences in biological sensitivity to context (e.g., Davies et al., 2007, Albers et al., 2008, Hill-Soderlund et al., 2008, Lisonbee et al., 2008). If one imagines concentric circles representing the theoretical sources of influence on individual differences and intra-individual change in the activity of the psychobiology of the stress response the literature clearly places the person–environment fit near the core (e.g., Gottlieb, 1992, Boyce and Ellis, 2005, Belsky et al., 2007). Even in the simplest model, however, few would argue that this inner core should be encompassed by a wider circle representing the influence that physical and other non-social features of everyday life have on the psychobiology of the stress response (e.g., Lupien et al., 2000, Evans and English, 2002, Watamura et al., 2002, Gump et al., 2008, Schuetze et al., 2009, Eiden et al., 2009). Without discounting the value and importance of the mainstream focus, the present purpose is to broaden our perspective by drawing attention to ubiquitous components of life in the 21st century that are capable of influencing our exploration of biosocial models as they operate in the context of our social worlds. Specifically, despite the high usage rate of prescription and over-the-counter medications, the effects of consuming these substances on individual differences in salivary cortisol remain largely unknown (but see Lowe and Dixon, 1983, Masharani et al., 2005, Hibel et al., 2006, Hibel et al., 2007).

Our literature search suggests that few behaviorally oriented studies involving salivary cortisol have comprehensively documented medication usage. When studies do so, small sample sizes combined with variation in the frequency in the usage of specific medications, restrict statistical power and increase the probability that we fail to detect medication–cortisol relationships that exist. Moreover, the lack of normative data and wide-ranging intra- and inter-individual differences in salivary cortisol levels make case-by-case exclusion of samples from individuals taking any particular medication questionable–unless the cortisol deviation is so large in magnitude to not be physiologically plausible, or meets established criteria (e.g., ±3 STDEV) as a statistical outlier.

Of course, the simplest approach to resolving this issue would be to exclude anyone from participation in research involving salivary cortisol who is taking any medication. Although appropriate from the perspective of rigorous experimental design, this conservative approach yields no information to help us develop our knowledge about which medications are, or are not, problematic. Also, in studies of some specialized populations (e.g., physical or mental illness; premature infants; frail elderly), medication use is the norm; therefore, having participants withhold their medication treatment raises ethical questions and excluding those taking medications increases the potential that findings generated would be confounded by selection bias.

Whether obtained over-the-counter, by prescription, or illegally, many pharmacotherapeutic agents are capable of directly or indirectly affecting the function of the hypothalamic–pituitary–adrenal (HPA) axis, or the measurement of cortisol in saliva, or both. The list of potential candidates is daunting. Over four thousand medications are listed in the Physicians Desk Reference (PDR, 2009). With billions invested annually in research, the list of pharmacotherapeutic candidates will undoubtedly expand as active ingredients are discovered, reformulated, and combined. Given this expansive and ever changing list, the reality is that it is not cost effective to generate, or maintain, a knowledge base that characterizes the effects of every pharmaceutical ingredient on salivary cortisol.

Thus, in this report, we consider a more strategic and tactical approach. We develop a logical structure based on the potential mechanisms of action by which any particular medication could influence salivary cortisol. This includes agonistic and antagonistic effects on the activity of the HPA axis, changes in the biochemical pathways by which cortisol is formed from its steroid precursors, effects on the percentage of cortisol bound to glucocorticoid binding globulin, moderation (or mediation) effects induced by change in subjective experience, direct cross-reactivity or non-specific interference with antibody binding in the assay of salivary cortisol, iatrogenic effects that impact the availability and composition of oral fluid samples, and changes in peripheral physiology that may influence the transport of serum constituents into oral fluids by ultrafiltration. From this set of pathways or mechanisms of influence, examples of medications with potential to affect salivary cortisol are presented to procedurally and statistically minimize this potential source of error variance. The overarching aim is to provide a framework from which researchers can make appropriate decisions about which medications should be used to screen or exclude research participants or as covariates to be sure the relationships they detect involving salivary cortisol are not confounded or spurious.