Physical activity and cortisol regulation : A meta-analysis

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC-ND 4.0 https://creativecommons.org/licenses/by-nc-nd/4.0/ Physical activity and cortisol regulation : A meta-analysis © 2023 Elsevier Accepted version (Final draft) Moyers, Susette A.; Hagger, Martin S. Moyers, S. A., & Hagger, M. S. (2023). Physical activity and cortisol regulation : A meta-analysis. Biological Psychology, 179, Article 108548. https://doi.org/10.1016/j.biopsycho.2023.108548 2023 PHYSICAL ACTIVITY AND CORTISOL REGULATION 1 Physical Activity and Cortisol Regulation: A Meta-Analysis Susette A. Moyers1,2 & Martin S. Hagger1,3,4,5 1Department of Psychological Sciences, University of California, Merced 2Center for Rural Health, Oklahoma State University Center for Health Sciences, Tulsa, OK, United States 3Health Sciences Research Institute, University of California, Merced 4Faculty of Sport and Health Sciences, University of Jyväskylä 5School of Applied Psychology, Griffith University Susette A. Moyers ORCiD: 0000-0001-6867-2741 Martin S. Hagger ORCiD: 0000-0002-2685-1546 Article Citation: Moyers, S. A., & Hagger, M. S. (2023). Physical activity and cortisol regulation: A meta-analysis. Biological Psychology, 179, 108548. https://doi.org/10.1016/j.biopsycho.2023.108548 Correspondence concerning this article should be addressed to Susette A. Moyers, Center for Rural Health, Oklahoma State University Center for Health Sciences, 1013 East 66th Place Tulsa, OK, 74136, United States. This study was preregistered on the Prospero registry of systematic reviews (https://www.crd.york.ac.uk/prospero/), Prospero ID: CRD42021247847 PHYSICAL ACTIVITY AND CORTISOL REGULATION 2 Abstract Physical activity participation is associated with effective stress coping, indicated by decreases in both physiological stress reactivity and perceived stress. Quantifying the effect of physical activity on the diurnal regulation of one key physiological stress indicator, the stress hormone, cortisol, across studies may demonstrate the extent to which physical activity participation is associated with diurnal HPA axis regulation. We meta-analyzed studies examining relations between physical activity participation and indices of HPA axis regulation: the diurnal cortisol slope and the cortisol awakening response. We also examined moderators of the relation. The analysis revealed a small, non-zero negative averaged correlation between physical activity and the diurnal cortisol slope (r = -0.043, 95% CI [-0.080, -0.004]). Examination of sample sociodemographic differences, study design characteristics, cortisol measurement methods, and physical activity variables as moderators revealed few effects on the relation between physical activity and diurnal cortisol slope. We did not observe lower levels of variability in the mean cortisol awakening response at higher levels of physical activity participation, and moderator analyses showed little evidence of reductions in heterogeneity for this effect. We found some evidence of systematic publication bias. Findings suggest higher physical activity is associated with a steeper diurnal cortisol slope. However, the cortisol awakening response did not differ by physical activity level. Future studies testing the physical activity and cortisol regulation association should use standardized physical activity measures, follow guidelines for better quality cortisol sampling collection and analysis, and test relations in large-scale empirical studies to confirm the direction and causality of the effect. Keywords: Diurnal cortisol slope; Cortisol awakening response; Physical activity; Exercise; HPA axis regulation; Health behavior PHYSICAL ACTIVITY AND CORTISOL REGULATION 3 Physical Activity and Cortisol Regulation: A Meta-Analysis Introduction Chronic psychological stress 1 is associated with the development or exacerbation of multiple physical and psychological health conditions including cardiovascular disease, type II diabetes, stroke (Rosmond & Bjorntorp, 2000), cancer (Abercrombie et al., 2004), premature cellular aging (Parks et al., 2009), autoimmune disorders and inflammation (Silverman & Sternberg, 2012), systemic hypertension (Wirtz et al., 2007), depression (Stetler & Miller, 2005), and anxiety (Vedhara et al., 2003). Exposure to psychological stress is ubiquitous, impacting individuals worldwide. For example, in 2018 about one third of the worldwide population, and more than half of Americans, reported experiencing elevated stress during the day (Gallup, 2019). Therefore, identifying the correlates of psychological stress and its development, as well as the mechanisms involved, is considered a priority area of research that may inform the development of interventions to minimize the deleterious effects of stress on health (O’Connor et al., 2020). One mechanism that has been implicated in the association between stress and stressrelated health outcomes is cortisol, a steroid hormone that is reactive to stressful stimuli. Physical (e.g., physical activity) and psychological (e.g., work stress) forms of stress can activate stressrelated responses, which, in turn, activate cortisol secretion to provide the energy and biological substrate necessary to cope with stress-provoking stimuli or escape threats (Jankord & Herman, 2008). Although cortisol secretion is an adaptive short-term response to stress, chronic stressrelated cortisol secretion can cause the system that produces cortisol, the hypothalamic-pituitary- 1 Psychological stress is defined as the perception that the demands of a task or situation exceed available personal and social resources (Lazarus & Folkman, 1984) and is associated with characteristic cognitive and somatic responses (Ulrich-Lai & Herman, 2009). PHYSICAL ACTIVITY AND CORTISOL REGULATION 4 adrenal axis (HPA axis), to become dysregulated. This means that cortisol secretion does not follow the expected secretion pattern. There are two main patterns of dysregulated cortisol secretion: tonic and phasic. Tonic cortisol actions generally refer to effects from the long-term presence, ranging from a few hours to days, of basal circulating cortisol, and produce a somewhat constant modulation of cortisol secretion (Sapolsky et al., 2000). In contrast, phasic cortisol actions can be attributed to the effects that stem from a rapid and transient increase in cortisol elicited by exposure to stress inthe-moment (Sapolsky et al., 2000). Both phasic and tonic cortisol secretion patterns contribute to the regulation of basal and stimulated HPA axis activity (Dallman et al., 1987; Sapolsky et al., 2000). However, for the purpose of the current review, where we refer to HPA axis regulation we mean tonic HPA axis regulation patterns. Tonic HPA axis dysregulation is typically measured through parameters that capture the circadian diurnal rhythm of cortisol secretion, such as the diurnal cortisol slope and the cortisol awakening response (Chida and Steptoe, 2009; Saxbe, 2008). The diurnal cortisol slope refers to the naturally occurring decline in cortisol throughout the waking day from awakening until bedtime, and the cortisol awakening response is a measure of the increase in cortisol levels during the first 30 minutes to 1 hour after awakening. Deviations from the expected rhythm of the diurnal cortisol slope and the cortisol awakening response are indicative of HPA axis dysregulation. Research has demonstrated associations between HPA axis dysregulation and increased risk of many chronic physical and mental health conditions such as cardiovascular disease (Whitworth et al., 2005), obesity-related metabolic disorders (Baudrand & Vaidya, 2015), insomnia (Rodenbeck & Hajak, 2001), hypertension (Whitworth et al., 1995), and uni- and bi-polar depression (McIsaac & Young, 2009). PHYSICAL ACTIVITY AND CORTISOL REGULATION 5 In addition, meta-analytic evidence has shown that deviations from the expected steep daily decline in the diurnal cortisol slope (i.e., a ‘flatter’ slope) are associated with many stressrelated mental and physical health outcomes such as depression, fatigue, immune/inflammatory outcomes, obesity and adiposity, cancer, and all-cause mortality (Adam et al., 2017). Stressrelated deviations in HPA axis regulation may, therefore, play a role in mediating associations between stress exposure and health outcomes (Chrousos & Gold, 1992; Davis & Sandman, 2010; Lupien et al., 2009), including both the onset and progression of mental and physical health disorders (Heim et al., 2008). Based on these data, it is imperative to identify modifiable targets for interventions that promote HPA axis regulation, and one target may be physical activity behavior. Physical activity participation is associated with improvements in, and prevention of, many health conditions that are also associated with a flatter diurnal cortisol slope such as depression (Dunn et al, 2005), fatigue (Puetz, 2006), elevated inflammation levels (Abramson & Vaccarino, 2002), cancer (McTiernan et al., 2019), and all-cause mortality (Lear et al., 2017). Therefore, examining the overall relations between physical activity and diurnal cortisol indices are an important first step in exploring the potential of promotion of physical activity as an efficacious means to promote HPA axis regulation. In the current study, we aimed to examine the association between physical activity participation and HPA axis regulation, as well as factors that moderate this association across the extant literature. Physical activity participation, defined as participation in exercise, sport, or physical activity as part of daily living, occupation, leisure, and active transportation (Garber et al., 2011), encompasses activities at light, moderate, and vigorous intensities and of any duration. We expected our study to make an important contribution to knowledge by examining the strength of the evidence across studies of physical activity participation as a potential correlate of PHYSICAL ACTIVITY AND CORTISOL REGULATION 6 cortisol regulation, a variable that provides a good indicator of effective stress coping. We also aimed to synthesize evidence on the conditions that affect the association between physical activity participation and HPA axis regulation. Physical Activity and HPA axis Regulation HPA axis regulation is one potential mechanism by which physical activity participation may impact stress-related outcomes (Tsatsoulis & Fountoulakis, 2006). Previous research supports the psychological stress-reducing effects of regular physical activity participation (Nguyen-Michel et al., 2006). Physical activity participation 2 is associated with lower levels of physiological stress reactivity after exposure to psychosocial stress (heart rate, blood pressure; Forcier et al., 2006; Hamer et al., 2006; Wipfli & Ramirez, 2013). These findings brought about the development of the cross-stressor adaptation hypothesis, which suggests that regular physical activity leads to adaptations in the stress response systems that induce decreased physiological responses to psychological stressors (Sothmann, 2006). While these data provide robust evidence that physical activity participation is related to effective in-the-moment physiological stress coping to stressful psychosocial stimuli, the underlying mechanisms that mediate this effect have yet to be fully elucidated. Developing an understanding of how physical activity interventions produce these changes and the mechanisms involved may provide targets for future interventions that will reliably produce adaptive changes in stress-related health parameters. HPA axis regulation has been identified as a proposed mechanism by which physical activity relates to better stress coping (Chen et al., 2017). Chronic physical activity is proposed to affect the medial prefrontal cortex and the hippocampus, two brain regions that are key to 2 For the purposes of this review, physical activity is defined as participation in exercise, sport, or physical activity as part of daily living, occupation, leisure, and active transportation (Garber et al., 2011) PHYSICAL ACTIVITY AND CORTISOL REGULATION 7 regulating the HPA axis negative feedback loop (Diorio et al., 1993; Herman et al., 2003; Mizoguchi et al., 2003), which can be thought of as the ‘off-switch’ that stops cortisol output after a stressor subsides (Zschucke et al., 2015; for a detailed explanation, see the review Chen et al., 2017). The diurnal cortisol slope has been proposed to represent an intact HPA axis negative feedback loop as a steep diurnal cortisol slope models an individual’s physiological ability to recover and disengage from stressful events over the course of a day (Heim et al., 2000; Miller et al., 2007). Physical activity participation may facilitate optimal HPA axis recovery to disengage from daily stress (Kim et al., 2008; van Hooff et al., 2019; Zschucke et al., 2015). Confirming that physical activity participation is associated with indices of diurnal HPA axis regulation may provide important evidence of this proposed mechanism. While many studies have found an association between physical activity and parameters related to HPA axis regulation such as the diurnal cortisol slope (Gubelmann et al., 2018; Ho et al., 2020; Vreeburg et al., 2009) and the cortisol awakening response (Calogiuri et al., 2016; Gubelmann et al., 2018; Tortosa-Martìnez et al., 2015), findings across studies have been inconclusive. For example, some studies have reported statistically significant or non-zero effects (Tortosa-Martìnez et al., 2015; Vreeburg et al., 2009), while others have reported effects that do not differ from the null (Foss et al., 2014; Lederbogen et al., 2010; McHale et al., 2014; Menke et al., 2014). Reviews of research examining relations between physical activity on parameters and HPA axis regulation have suggested an association (Anderson & Wideman, 2017; Pascoe et al., 2017), but, to date, no research synthesis has provided an estimate of the size of the relation between physical activity participation, broadly defined, and indices of HPA axis regulation. Quantifying the relation between physical activity participation and these indices across studies may provide important evidence to support the size of the effect and its degree of PHYSICAL ACTIVITY AND CORTISOL REGULATION 8 heterogeneity across studies, and, in so doing, provide a reliable estimate of the true variability in the effect. It may also assist in identifying possible conditions or factors that may affect it through an analysis of moderators. The research is expected to provide valuable information for those developing stress coping interventions to reduce stress and improve health outcomes. The Present Study Study Overview In the present study, we aimed to synthesize research testing relations between physical activity participation and indices of effective HPA axis regulation and estimate the variability and heterogeneity in the effect. Specifically, we aimed to estimate the average size and heterogeneity of the relation between physical activity participation and two independent indices of HPA axis regulation, the diurnal cortisol slope and the cortisol awakening response, across studies in existing literature using multi-level meta-analysis. These indices were selected because they are among the most commonly used means to indicate HPA axis dysregulation in the extant literature (Adam & Kumari, 2009), enable accurate estimation of the variability in cortisol across key diurnal timepoints of interest, and represent independent aspects of HPA axis dysregulation (Wilhelm et al., 2007). We aimed to evaluate the extent to which the observed variability in the correlation between physical activity and the diurnal cortisol slope, and the mean and standard deviation for salivary cortisol awakening responses at different levels of physical activity participation, across studies was attributable to methodological artifacts corrected for in metaanalysis (i.e., sampling error), or to true variability in the effect across studies and the extent of that variability. PHYSICAL ACTIVITY AND CORTISOL REGULATION 15 Mean variability of CAR in high and moderate PA levels vs. low PA levels in experimental and intervention studies smaller in longer vs. shorter periods of physical activity. Time lag Greater heterogeneity in PA-DCS in larger time lag in longitudinal studies vs. shorter time gaps. Greater heterogeneity in mean cortisol awakening response in the physical activity-CAR in studies with longer time lag vs. studies with shorter time lag. Longitudinal studies that measure constructs in close proximity may inflate relations Intervention components Single (physical activity only) vs. multiple intervention components. Exploratory Season Greater heterogeneity in PA-DCS in fall/winter months vs. spring/summer Greater heterogeneity in mean cortisol awakening response in the physical activity-CAR in studies that collected data in fall/winter months vs. spring/summer months. Miller et al., 2016 Race/Ethnicity Larger PA-DCS in white vs. non-white participants. Mean variability of CAR in high and moderate PA levels vs. low PA levels larger in non-white participants vs. white participants. Adam et al., 2015; DeSantis et al., 2007 Physical activity measurement Measure type Greater precision and less variability in the PA-DCS effect in non-self-report measures vs. self-report measures. Mean variability of CAR in high and moderate PA levels vs. low PA levels smaller in non-self-reported physical activity measures vs. self-reported physical activity measures Greater precision and less variability in the PA-DCS effect in validated scales vs. bespoke scales. Mean variability of CAR in high and moderate PA levels vs. low PA levels smaller validated scales vs. bespoke physical activity scales. Adams et al., 2005; Sallis & Saelens, 2005 Milton et al., 2011 Physical activity PA-DCS larger when bouts of physical activity exceeded an individuals’ 50% VO2 max threshold. Duclos et al., 2003; Hackney & Viru, 1999; Hill et al., 2008; PHYSICAL ACTIVITY AND CORTISOL REGULATION 16 intensity PA intensity was not assessed in the PA-CAR effect because intensity is inherent in the classification of study effects into PA subgroups. Viru, 1992 Physical activity intensity derivation Greater averaged PA-DCS in studies that measured intensity as relative PA intensity vs. those based on absolute values Mean variability of CAR in high and moderate PA levels vs. low PA levels smaller in in relative exercise intensity measures vs. absolute values. Freedson et al., 1998 Ozemek et al., 2013 Physical activity duration per bout Larger PA-DCS when bouts of PA were 30 minutes or greater in duration vs. less time per bout. PA duration was not assessed in the PA-CAR effect because duration is inherent in the classification of study effects into physical activity subgroups. Duclos et al., 2003; Hackney & Viru, 1999; Hill et al., 2008; Viru, 1992 Level of physical fitness Larger PA-DCS in athletes and individuals with high levels of physical fitness vs. lower levels of fitness. Mean variability of CAR in high and moderate PA levels vs. low PA levels smaller in high fitness vs. lower fitness. Duclos et al., 2003 Tortosa-Martínez et al., 2015 Fitness assessment type Studies utilizing non-self-reported measures of physical fitness are less heterogeneous than studies that utilized self-reported fitness assessment. Non-self-reported measures may be less prone to error self-report. Physical activity time of day Larger PA-DCS in morning PA vs. afternoon or evening PA. Mean variability of CAR in high and moderate PA levels vs. low PA levels smaller in morning PA compared to afternoon or evening PA. Kanaley et al., 2001 Physical activity type Larger PA-DCS in aerobic PA vs. anaerobic PA. Mean variability of CAR in high and moderate PA levels vs. low PA levels smaller in aerobic PA vs. anaerobic PA. Beserra et al., 2018; Vrinceanu et al., 2019 Physical activity frequency Larger PA-DCS in three or more PA sessions a week compared to fewer than three sessions per week. PA frequency was not assessed in the PA-CAR effect because is inherent in the Beserra et al., 2018 PHYSICAL ACTIVITY AND CORTISOL REGULATION 17 classification of study effects into PA subgroups. Cortisol Measurement Diurnal cortisol slope measurement Larger PA-DCS in studies using amplitude measures and wake-bed slopes vs. other methods. Larger PA-DCS in waking vs. fixed time point cortisol measurements. Adam et al,., 2017 Wilhelm et al., 2007 Cortisol awakening response measurement Mean variability of CAR in high and moderate PA levels vs. low PA levels smaller in waking vs. fixed time point cortisol measurement. Wilhelm et al., 2007 Number of cortisol sample taken over relative time period Larger PA-DCS in studies with more than two samples in one day vs. studies that took only two samples. Mean variability of CAR in high and moderate PA levels vs. low PA levels greater in studies with two samples of cortisol vs. more than two. Adam et al., 2017 ; Chida & Steptoe, 2009 Number of days cortisol was sampled Greater variability in the PA-DCS in one day vs. more than one day of cortisol samples. Mean variability of CAR in high and moderate PA levels vs. low PA levels larger in one day vs more than one day of cortisol samples. Adam & Kumari, 2009 Cortisol sampled on resting day Less variability in the PA-DCS in studies with a rest period prior to diurnal cortisol measurement vs. those that do not instruct participants to rest prior to cortisol measurement. We assumed that participants would not be likely to engage in PA between awakening and 30-45 minutes after awakening samples. Viru, 1992 Methodological quality of cortisol sampling Lower quality studies were likely to exhibit greater error variance in the PADCS vs. studies of acceptable quality. Greater error variance in the variability of the mean CAR between high and moderate versus low levels of PA. Johnson et al., 2014 Note. PA = physical activity participation; DCS-PA = the physical activity-diurnal cortisol slope association; CAR= the cortisol awakening response. PHYSICAL ACTIVITY AND CORTISOL REGULATION 18 before or after competition, these measurements were included. Studies that measured the diurnal cortisol slope on a day when physical activity participation was performed were excluded as such measures are not likely to effectively capture HPA axis regulation but, instead, may reflect diurnal cortisol reactivity to physical activity participation. Studies focusing on daily average cortisol, area under the curve measures of salivary cortisol, or other integrated cortisol measures such as overnight urinary cortisol or hair cortisol were excluded, as these measures do not provide information on daily cortisol variability (Badrick et al., 2017; Fekedulegn et al., 2007). Studies on participants with endocrine disorders, genetic disorders or polymorphisms, or impaired capacity to provide responses on self-report questionnaires were excluded. Studies with proxy measures (e.g., parent, caregivers) of study variables were also excluded. We located ‘fugitive’ data (Rosenthal, 1994) by sending email requests to authors of eligible studies that did not report sufficient data to compute an effect size, and gave them a fourweek time frame to respond. This included studies that reported taking measures of physical activity participation and two time-dependent cortisol secretion measurements in the same day, but did not calculate the cortisol awakening response and/or the diurnal cortisol slope (disaggregate values reported only); did not calculate the cortisol awakening response as the raw mean difference (e.g., the peak cortisol value minus the awakening cortisol value); did not report the physical activity participation-HPA axis regulation association; or reported the association within a multivariate analysis from which a unique physical activity participation-HPA axis regulation effect size could not be isolated. Furthermore, authors were contacted if the study only included a statistic reporting the association between physical activity participation and either the diurnal cortisol slope or the cortisol awakening response, and sufficient data was gathered to assess the other outcome, but it was not reported. Authors were also contacted if physical activity PHYSICAL ACTIVITY AND CORTISOL REGULATION 19 data were collected as part of a global ‘health behavior’ measure or as a covariate, but isolated effects of physical activity participation were not reported. In cases where interventions or experimental studies employed a manipulation of physical activity participation and measured changes in HPA axis regulation parameters, then the effect of the manipulation on the outcome was taken as the estimate of the effect. If the experimental manipulation did not consist of physical activity participation (e.g., pharmacological administration, psychological stress exposure), targeted an outcome other than HPA axis regulation, or did not specify the target construct, then data at baseline or in the control group were used to estimate the effect where possible, otherwise the authors were contacted to provide the relevant data. Screening After removal of duplicate studies, 16,558 studies were identified for inclusion. Titles and abstracts of the studies retrieved in search were screened against inclusion/exclusion criteria. This comprised an initial title screen, followed by abstract and full text screening, conducted by the lead researcher and two trained research assistants. During title and abstract screening, studies were divided into retained, excluded, or potentially eligible categories. Queries raised during screening were discussed between the research team and resolved through mutual agreement, with 250 of the studies screened by all researchers to validate the screening procedure. Of these, 10,270 were excluded after title screening and a further 6,269 studies excluded after abstract screening. Agreement between the three reviewers was calculated for each screening phase using the AC1 coefficient (Gwet, 2008). While Cohen’s kappa statistic is used more widely in meta-analyses to determine inter-rater reliability, Gwet’s AC1 coefficient has been shown to provide a more stable inter-rater reliability estimate than Cohen’s kappa, and it is also less affected by high agreement prevalence and marginal probability than Cohen’s PHYSICAL ACTIVITY AND CORTISOL REGULATION 20 kappa (Wongpakaran et al., 2013). Further, Gwet's AC1 is the statistic of choice for the case of two raters because it does not depend upon the assumption of independence between raters (Gwet, 2008). Therefore, we opted to utilize the AC1 coefficient over Cohen’s kappa in this meta-analysis. During title screening there was 90.03% average agreement (AC1 = .844, [0.819, 0.909], p < .001), and with 92.71% average agreement (AC1 = .897, [0.874, 0.941], p < .001) during abstract screening. Main reasons for exclusion included: studies were theoretical or conceptual reviews, systematic reviews, and off-topic (not pertaining to any content related to physical activity participation or HPA axis regulation). The remaining studies (k = 1,559) were subjected to full-text analysis for inclusion in the final sample, of these, 1,462 did not meet criteria for inclusion. Full text screening was conducted by the lead and senior researchers, with 25% of the studies screened by both researchers to validate the screening procedure. Disagreements were also resolved through discussion and inclusion/exclusion criteria were modified and reapplied if necessary. Average inter-rater reliability across raters for inclusionexclusion decisions during full-text screening was acceptable (AC1 = .750, [0.471, 1.00], p < .001). Studies were excluded at this stage for the following reasons: cortisol was only measured one time in a day, cortisol was measured twice, but data were insufficient to calculate either the diurnal cortisol slope or the cortisol awakening response; cortisol was measured twice, but physical activity was performed in between sampling times; no measure of physical activity participation was included; study design included physical activity with other stressful components (e.g., hiking in hypoxic conditions, combat military training, overtrained athletic competitors) that may influence the cortisol awakening response and/or the diurnal cortisol slope; or insufficient data were available in the article to compute effect sizes and authors could not be contacted or were unable to supply the required data. Study selection procedures are PHYSICAL ACTIVITY AND CORTISOL REGULATION 21 summarized in the PRISMA (Moher et al., 2009) flow diagram presented in Appendix C (supplemental materials). Data Extraction Study characteristics, effect size data, and data for moderator variable coding were extracted from all eligible studies by the lead and senior researchers. The following study characteristics were extracted from each study: author names, publication year, sample size, HPA axis regulation measure to be included (diurnal cortisol slope/cortisol awakening response). Extracted data for moderator coding were: sociodemographic variables (sex, age, race/ethnicity, BMI, clinical sample); type of calculation for the diurnal cortisol slope (levels: amplitude, wakebed slope, fixed timepoint slope, and peak and late decline slopes); number of cortisol samples taken to assess the diurnal cortisol slope and/or the cortisol awakening response (levels: 2 samples and >2 samples); number of days cortisol was sampled (levels: 1 day and >1 day); cortisol sampling quality rating (score range: 0-9); report of physical activity performed on cortisol sampling day (levels: cortisol sampled on resting day and not reported); physical activity measure (levels: previously-validated physical activity scales, bespoke physical activity related questions, non-self-report physical activity measure, longitudinal grouped physical activity, and experimentally assigned physical activity); physical activity intensity (levels: low intensity and moderate-to-vigorous intensity); physical activity intensity assessment type (levels: absolute measure of intensity based on metabolic equivalents and relative measure of intensity based on maximal oxygen consumption); duration (levels: more than 30 min per activity bout and less than 30 min per activity bout); frequency of activity (levels: <3 times a week and 3 or more times a week); time of day physical activity was performed (levels: morning and afternoon or evening); physical activity type (levels: aerobic and anaerobic) and level of fitness (levels: athlete, high fit, PHYSICAL ACTIVITY AND CORTISOL REGULATION 22 and low fit); fitness assessment type (levels: self-report and non-self-report); study design (levels: correlational, longitudinal, and experimental); intervention components (level: physical activity intervention only and physical activity intervention plus one or more additional intervention component/s); season of data collection; and time lag. The zero-order correlation coefficient (r) was selected as the effect size metric for the diurnal cortisol slope outcome as the majority of the studies were correlational in design and the correlation coefficient was expected to be the most frequently adopted effect size. Where effect sizes were not expressed as a correlation, available effect size data including computed effect sizes (Cohen’s d or f, eta-squared) and tests of difference (e.g., t and F-ratios, chi-square values) were converted to r using published formulae. All formulae utilized are reported in Appendix D (supplemental materials; Borenstein et al., 2009; Digby, 1983). To assess the relation between physical activity participation and the cortisol awakening response, the variability in the mean cortisol awakening response was compared in groups of studies reporting low, moderate, or high levels of physical activity participation. Physical activity level classification was based on the International Physical Activity Questionnaire (IPAQ) categories, representing low, moderate, and high levels of physical activity participation (Sjöström et al., 2002). Activity level was considered moderate if the level of physical activity participation met at least one of the following criteria: (a) 3 or more days of vigorous-intensity physical activity of at least 20 minutes per day; (b) 5 or more days of moderate-intensity and/or walking of at least 30 minutes per day; or (c) 5 or more days of any combination of walking, moderate-intensity, or vigorousintensity activities achieving a minimum of at least 600 MET-min/week of physical activity. Activity level was considered high if the level of physical activity participation met at least one of the following criteria: (a) vigorous-intensity physical activity participation on at least 3 days PHYSICAL ACTIVITY AND CORTISOL REGULATION 23 and accumulating at least 1500 MET-min/week; or (b) 7 or more days of any combination of walking, moderate- or vigorous-intensity physical activity accumulating at least 3000 MET- min/week. Activity was considered low if neither of these criteria were met. Data for relations between physical activity and HPA axis regulation were extracted from included studies by two researchers. In cases where data for one of the target variables was expressed as a categorical variable (e.g., regular vs. irregular slope), with means and standard deviations for a measure of physical activity participation for each category, we computed a standardized mean difference using the available data and converted it to r. For experimental or intervention studies with a manipulation of physical activity participation, we computed an effect size using baseline and follow-up data for the dependent variable for either the experimental or control group (controlled designs), or baseline to follow-up (pre-post designs) manipulations of the independent variable (Borenstein et al., 2009). Multiple measures and comparisons between multiple levels of physical activity participation within the same study were utilized, where available, and effects were treated as dependent effects. Where studies reported multiple effect sizes for the physical activity participation-HPA axis regulation relation within each study, we treated each effect size according to a pre-defined protocol. Studies reporting separate effect sizes estimated in two or more independent samples were treated as separate studies (e.g., male/female, low/moderate/high physical activity subgroups). Where multiple effect sizes were reported in the same study, such as when studies reported correlations between a measure of HPA axis regulation and two or more levels of physical activity intensity, these data were treated as multiple effects from the same study. However, when studies reported statistics between a measure of physical activity participation and both outcomes of interest, both were included in the separate diurnal cortisol slope and PHYSICAL ACTIVITY AND CORTISOL REGULATION 24 cortisol awakening response analyses. We coded the data according to whether data were from independent samples or multiple effect sizes within a single study, and this coding was used as input for subsequent data analysis. Some studies that met inclusion criteria utilized the same data set (e.g., multiple published studies from the Midlife in the United States data set (MIDUS)), and the published studies had varying degrees of sample overlap. Therefore, all articles which met inclusion criteria from the same data set were added, but we treated them as a single data point, to avoid inflating relations. Also, in many of these instances, we directly accessed the publicly available data to perform the analyses to include the data point for these studies, and then coded moderators and study quality from the available data across the published articles. Our protocol outlining how data were treated for each outcome is summarized in Appendix E (supplemental materials), with coding of effect size for each study for the diurnal cortisol slope and the cortisol awakening response provided in Appendixes F and G, respectively (supplemental materials). The moderator coding protocol is reported in Appendix H (supplemental materials). Assessment of Quality of Cortisol Sampling Methods We assessed the quality of the cortisol sampling methods in each study based on criteria from previous meta-analyses and systematic reviews of HPA axis regulation (Adam et al., 2017; Chida & Steptoe, 2009). The assessment was based on whether studies reported accounting for the following participant characteristics and conditions during cortisol sampling: age; gender; smoking status; use of steroid-based medications; wake time; sampling day (weekday or weekend); self-reported adherence with sampling times; objective adherence to sampling times based on electronic monitoring; and clear sampling instructions provided to participants (e.g., to refrain from brushing their teeth, drinking, or eating 15 minutes prior to sampling). Scores were summed and both dichotomous and continuous quality of cortisol sampling methods rating PHYSICAL ACTIVITY AND CORTISOL REGULATION 31 Figure 3 Forest plot of the physical activity and diurnal cortisol slope metaanalysis PHYSICAL ACTIVITY AND CORTISOL REGULATION 32 Figure 4 Forest plot of the mean cortisol awakening response in the low physical activity subgroup PHYSICAL ACTIVITY AND CORTISOL REGULATION 33 Figure 5 Forest plot of the mean cortisol awakening response in the moderate physical activity subgroup PHYSICAL ACTIVITY AND CORTISOL REGULATION 34 Figure 6 Forest plot of the mean cortisol awakening response in the high physical activity subgroup PHYSICAL ACTIVITY AND CORTISOL REGULATION 35 Table 2 Results of Multi-Level Meta-Analysis of the Association Between Physical Activity Participation and the Diurnal Cortisol Slope Model r 95% CI 2 within 2 between Q df AIC Model comparisons I2 95% CI LL UL χ2 LRT p LL UL Overall -.06** -0.09 -0.02 0.01 0.01 298.84*** 97 -62.34 34.17 – – 70.75 60.72 82.35 Betweenstudy variance (2) only -.07*** -0.10 -0.04 – 0.01 298.84*** 97 -62.20 33.10 2.14 0.14 – – – Withinstudy variance (2) only -.04* -0.08 -0.004 0.01 – 298.84*** 97 0.96 1.52 65.29 <.001 – – – Note. Number of studies is 98, and total sample size 19,744 in all analyses. N = Number of participants; r = Average sample-weighted correlation; 2 between = Between-study variance; 2 within = Within-study variance; Q = Cochrane’s Q statistic; df = Degrees of freedom; AIC = Akaike’s Information Criterion; χ2 = Chi-square (log likelihood); LRT = Likelihood ratio test; p = Significance level of the LRT; I2 = Index of the dispersion of effect sizes. *p < .05 **p < .01 ***p < .001 PHYSICAL ACTIVITY AND CORTISOL REGULATION 36 Table 3 Results of Multi-Level Meta-Analysis of the Association Between Physical Activity Participation and the Cortisol Awakening Response Moderator k N M 95% CI 2 within 2 within 95% CI 2 between 2 between 95% CI 2 absolute differenc e Q df AIC Model comparisons I2 95% CI LL UL LL UL LL UL χ2 LRT p LL UL Low Physical Activity Subgroup Overall 57 5313 4.76*** 3.55 5.97 1.17 0.00 5.89 11.63 5.13 20.65 4.89a 3382.48*** 56 300.88 -147.44 – – 99.11 98.61 99.41 Betweenstudy variance (2) only 57 5313 4.76*** 3.54 5.98 – – – 12.87 8.08 21.72 2.34b 3382.48*** 56 300.45 -148.22 1.57 .21 – – – Withinstudy variance (2) only 57 5313 4.62*** 3.67 5.58 10.72 7.12 16.75 – – – 2.55c 3382.48*** 56 306.44 -151.22 7.56 .01 – – – Moderate Physical Activity Subgroup Overall 44 2794 4.38*** 3.45 5.32 0.32 0.00 3.40 0.56 0.00 1.84 1045.99*** 43 205.95 -99.98 – – 98.90 98.17 99.35 Betweenstudy variance (2) only 44 2794 4.38*** 3.44 5.31 – – – 6.11 3.57 11.06 1045.99*** 43 204.66 -100.33 0.71 .40 – – – Withinstudy variance (2) only 44 2794 4.36*** 3.54 5.18 5.83 3.59 9.89 – – – 1045.99*** 43 211.03 -103.52 7.08 .01 – – – High Physical Activity Subgroup Overall 52 3416 4.23*** 2.98 5.48 .000 0.00 0.05 10.61 0.00 19.38 1587.66*** 51 257.90 -125.95 – – 99.70 99.56 99.86 Betweenstudy variance 52 3416 4.23*** 2.98 5.48 – – – 10.61 6.27 19.38 1587.66*** 51 255.90 -125.95 0.000 1.00 – – – PHYSICAL ACTIVITY AND CORTISOL REGULATION 37 (2) only Withinstudy variance (2) only 52 3416 4.16*** 3.24 5.08 8.38 5.38 13.63 – – – 1587.66*** 51 281.27 -138.64 25.37 <.01 – – – Note. aAbsolute difference of the variance of the mean cortisol awakening response between low and moderate physical activity subgroups; bAbsolute difference of the variance of the mean cortisol awakening response between low and high physical activity subgroups; cAbsolute difference of the variance of the mean cortisol awakening response between moderate and high physical activity subgroups; k = Number of studies; N = Number of participants; M = Average sample-weighted mean cortisol awakening response; 2 between = Between-study variance; 2 within = Within-study variance; Q = Cochrane’s Q statistic; df = Degrees of freedom; AIC = Akaike’s Information Criterion; χ2 = Chi-square (log likelihood); LRT = Likelihood ratio test; p = Significance level of the LRT; I2 = Index of the dispersion of effect sizes. *p < .05 **p < .01 ***p < .001 PHYSICAL ACTIVITY AND CORTISOL REGULATION 38 analyses did not resolve the observed variability or lead to a narrowing of the confidence intervals about the effect size. However, there were a few noteworthy exceptions. Difference tests in the effect sizes across moderator categories revealed that the physical activity-diurnal cortisol slope size was larger in mixed age groups relative to older age groups (t(68) = -2.04, p < .05); larger in studies that grouped participants based on physical activity level when compared to a validated physical activity measure (t(32) = 3.57, p < .001); larger in studies that experimentally assigned physical activity when compared to a validated physical activity measure (t(53) = 2.38, p < .05); and larger in studies that utilized a non-self-report physical activity measure when compared to a bespoke physical activity measure (t(34) = 2.80, p < .01). Importantly, the effect was non-zero in experimental studies, providing evidence that changing physical activity participation behavior produced changes in the diurnal cortisol slope. However, in each of the cases above, heterogeneity statistics remained substantive and confidence intervals about the mean differences were wide. Taken together, these findings suggest that although there was some evidence of moderation, the differences did not lead to homogenous cases or a narrowing of confidence intervals about the correlation in each moderator group. We found homogenous cases for the effect sizes in studies with mixed age groups, studies utilizing bespoke physical activity measures, and studies measuring the diurnal cortisol slope as a peak-to-bed or late decline slope, although effect sizes and comparisons with other moderator groups were no different from zero. We also found homogenous effect sizes for studies with mostly male participants, studies utilizing physical activity only as the intervention design, studies on vigorous-intensity physical activity only, and studies that took two samples of cortisol to determine the diurnal cortisol slope, with the correlation in studies using physical activity only as the intervention the only non- PHYSICAL ACTIVITY AND CORTISOL REGULATION 39 zero effect size. It must be stressed, however, that the number of studies in each of these cases was small and are unlikely to have provided a precise estimate of effect size heterogeneity in each group. 6 Results of the moderator analyses for the cortisol awakening response for each physical activity subgroup are reported in Table 5. We found large absolute differences in the variability about the mean cortisol awakening response between physical activity subgroups in some analyses. For example, the comparison between low and moderate, and between low and high, physical activity subgroups in studies with samples with low and high BMI. However, in all instances, the confidence intervals about the variance estimates of the effect size at each level of the moderator were wide and overlapped substantially, so differences across moderator groups were not supported. Numbers of effect sizes at each level of the moderator groups comparing averaged cortisol awakening response values across the physical activity subgroups numbered few (typically k < 10) in most cases precluding a meaningful comparison of the effect at levels of each moderator for many of the planned moderator variables: clinical status, intervention components, study design, physical activity type, cortisol awakening response measurement type, and our exploratory moderator variables: race/ethnicity and season of data collection. It should also be noted that moderator analyses for the mean cortisol awakening response significantly reduced the heterogeneity of the effect in some instances. For example, the effect was homogenous in the analysis of male participants in the low and moderate physical activity subgroups. This was also the case in some other moderator analyses, but the number of studies in the level of each of these moderators were small, and the confidence intervals largely overlapped across subgroups (see Appendix P, supplemental materials, for a complete summary of methods 6 Meta-regressions were conducted to examine the effects of categorical and continuous moderators simultaneously and mirrored the above results (see Appendix O, supplemental materials). PHYSICAL ACTIVITY AND CORTISOL REGULATION 40 used to compute effect sizes in moderator analyses for the physical activity-cortisol awakening response relation that correspond with table subscripts). Analysis of Bias Regression tests for the diurnal cortisol slope analysis revealed non-zero effects of the precision estimate on the effect size for the PET (B0 = 0.01, 95% CI [-.01, .04], p = 0.27) and PEESE (B1 = -.02, 95% CI [-.04, -.01], p = 0.007) tests. As the PET estimate for the PHYSICAL ACTIVITY AND CORTISOL REGULATION 47 Methodological Quality of Cortisol Sampling Questionable Quality (0-6) 25 3.11*** 1.84 4.38 663.19*** 0.05 0.00 7.93 5.52 0.00 12.83 6.84x 27 3.81*** 2.60 5.03 601.98*** 0.21 0.00 2.77 5.06 1.86 11.92 0.43y 29 3.10*** 1.72 4.49 339.23*** 0.00 0.00 0.07 6.20 2.81 15.44 – Acceptable Quality (7-9) 32 6.13*** 4.31 7.95 1708.09*** 1.79 0.00 10.13 13.67 3.35 30.61 – 17 5.13*** 3.63 6.63 437.68** 6.37 0.00 17.05 0.73 0.00 15.42 – 23 5.54*** 3.46 7.63 1186.21*** 0.00 0.00 0.98 13.41 6.38 33.90 – Note. aValues reported on the upper, middle and lower lines for each moderator level are for the low, moderate, and high physical activity subgroups, respectively; k = Number of studies; M = Average sample-weighted mean; 95% CI = 95% confidence intervals of the average sample-weighted mean; LL = Lower limit of the 95% confidence interval; UL = Upper limit of the 95% confidence interval; Q = Cochrane’s Q statistic; 2 between = Between-study variance; 2 between 95% CI = 95% confidence intervals of the between-study variance; LL = Lower limit of the 95% confidence interval of the between-study variance; UL = Upper limit of the 95% confidence interval of the between-study variance; 2 within = Within-study variance; 2 within 95% CI = 95% confidence intervals of the within-study variance; LL = Lower limit of the 95% confidence interval of the within-study variance; UL = Upper limit of the 95% confidence interval of the within-study variance. * p < .05 ** p < .01 ***p < .001 PHYSICAL ACTIVITY AND CORTISOL REGULATION 48 effect size was no different from zero, the PET estimate was taken as the corrected estimate for the physical activity participation-diurnal cortisol slope effect size. These findings provided some evidence of small study bias for this effect in the sample of studies. Although there was an observed difference in the corrected estimate from the PET analysis from the uncorrected estimate, the difference was modest, although the corrected estimate did not differ from zero. The tests revealed non-zero effects for the precision estimate on the cortisol awakening response mean for the PET and PEESE versions in all physical activity subgroups. These analyses suggested the presence of small study bias in each subgroup, however, observed differences in the corrected effect size estimates from the tests did not differ substantially from the original estimates in each subgroup and did not lead us to alter our overall conclusions on the size and variability of mean cortisol awakening response in each physical activity subgroup. ‘Funnel’ plots and full details of the bias analyses are presented in Appendix Q (supplemental materials). Sensitivity Analyses The leave-one-out analysis for the physical activity-diurnal cortisol slope analysis revealed two effect sizes that may have affected the overall effect size. Omitting each study from the analysis did not substantively change the overall estimate of the effect and its variability (see Appendix R, supplemental materials). The leave-one-out analysis for the mean cortisol awakening response in each physical activity subgroup identified a few influential studies in each physical activity subgroup, and in all cases, omission of those studies did not change the overall estimate of the mean effect. Full details of the analysis are presented in Appendix R. PHYSICAL ACTIVITY AND CORTISOL REGULATION 49 Discussion The aim of the present study was to estimate the average size and variability of the relation between physical activity participation and two independent indices of HPA axis regulation, the diurnal cortisol slope and the cortisol awakening response, across studies using multi-level meta-analysis. In line with our hypotheses, findings revealed a small, negative, nonzero averaged correlation between physical activity and the diurnal cortisol slope across studies, suggesting that the diurnal cortisol slope is steeper at higher levels of physical activity participation. However, contrary to our hypotheses, confidence intervals about the average mean variance estimates for the cortisol awakening response overlapped considerably across physical activity subgroups and did not provide definitive evidence supporting lower levels of variability in the mean cortisol awakening response at higher levels of physical activity participation. Significant heterogeneity in the averaged correlation between physical activity and diurnal cortisol slope suggested the presence of moderators. Tests of moderators revealed some differences in the effect in some age, clinical status, and physical activity measure moderator subgroups, but no clear pattern emerged and significant heterogeneity in the effect sizes in each moderator group remained. Significant heterogeneity in the averaged mean estimate of the cortisol awakening response in each physical activity subgroup also suggested the presence of moderators. However, the small number of studies in each subgroup precluded a meaningful comparison for many planned moderators. Where there were sufficient studies at each level of the moderator across subgroups, we found large absolute differences in the variability about the mean cortisol awakening response between physical activity subgroups in the comparison between low and moderate, and between low and high, physical activity subgroups in studies with samples with low and high BMI; the comparison between low and moderate physical PHYSICAL ACTIVITY AND CORTISOL REGULATION 50 activity subgroups, comparing validated and non-self-report physical activity measures; the comparison between low and moderate physical activity subgroups, comparing bespoke and nonself-report physical activity measures; the comparison between low and moderate physical activity subgroups comparing non-self-report and experimental assignment or longitudinal grouped physical activity measures; and the comparison between low and moderate physical activity subgroups comparing studies that sampled cortisol on one day only and studies that sampled cortisol on more than one day. Further, the effect was homogenous in the analysis of male participants in the low and moderate physical activity subgroups, experimental studies in the moderate physical activity subgroup, experimental studies that assigned physical activity as the only intervention component and experimental studies that assigned physical activity plus other intervention components in the moderate intensity subgroup, studies that measured physical activity as longitudinal grouped or experimentally assigned in the moderate intensity subgroup, and studies that based the cortisol awakening response measurement on a fixed awakening time in the high physical activity subgroup. Physical Activity and the Diurnal Cortisol Slope Current findings provide evidence of a small overall association between physical activity participation and the diurnal cortisol slope, indicative of better overall HPA axis regulation and cortisol secretion throughout the day among those participating in physical activity corroborating findings elsewhere (e.g., Gubelmann et al., 2018; Ho et al., 2020; Vreeburg et al., 2009). However, it must be stressed that applying bias-correction analyses yielded a corrected estimate that did not differ from zero, which indicates that the effect may be inflated due to selective reporting in studies. However, this finding should not be considered definitive given that the bias correction analyses are less precise in small samples and under PHYSICAL ACTIVITY AND CORTISOL REGULATION 51 conditions of heterogeneity (Stanley, 2017). So, results from the bias analyses should be interpreted with this caveat in mind. Analysis of Moderators Although the candidate moderators of the physical activity-diurnal cortisol slope association tested in the current analysis did not confirm the presence of moderators of the effect, some of the correlations for specific moderator groups were non-zero and worth noting. Of particular note is the non-zero the averaged correlation observed in studies adopting experimental or intervention designs. This suggests studies using these designs are associated with changes in the diurnal cortisol slope such that it becomes steeper, an indication of HPA axis regulation. Furthermore, we also observed a non-zero correlation when comparing intervention studies that assigned physical activity as the sole intervention component. Taken together, these findings provide evidence that systematically and intentionally changing physical activity behavior leads to changes in a key index of effective stress coping. This finding is important because, unlike interventions, other physical activity measurement methods such as self-report questionnaires inflate levels of physical activity due to the phrasing of the measurement items, compromising their accuracy. If intentionally and systematically changing one’s physical activity behavior produces a steeper diurnal cortisol slope, this may lead to an inverse of the expected effect, that is a flatter slope, if individuals who engage in regular physical activity were to intentionally limit their physical activity levels. Further, since follow-up periods of the included intervention studies were short, the greatest change in cortisol indices may be observed during the initial stages of physical activity intervention prescriptions but may later return to baseline levels. We therefore encourage researchers to adopt longitudinal designs in future studies, including multiple sampling periods over longer timeframes to monitor diurnal cortisol indices. PHYSICAL ACTIVITY AND CORTISOL REGULATION 52 However, it should be noted that these findings are associated with considerable heterogeneity, indicating a highly variable effect, and we could not identify moderators that accounted for this variance. This points to a need for large-sample experimental or intervention studies that not only provide further robust evidence for the causal effect of physical activity change on the diurnal cortisol slope, but also systematically tests the effects of salient moderators. For example, researchers should consider testing whether effects of intervention or experimental manipulation of physical activity levels on cortisol regulation changes according to types of measurement methods (e.g., type of diurnal cortisol slope used, number of cortisol samples taken over the relative time period, number of days cortisol was measured, cortisol sampling quality, biochemical analysis procedure), type of physical activity assessment method (physical activity intensity, intensity assessment type, physical activity duration, level of physical fitness, fitness assessment type, time of day physical activity was performed, physical activity type, physical activity measurement, and physical activity frequency), and general study design (time lag and study design). Assessment of these moderators my assist in resolution of the heterogeneity observed in the effect size observed in studies adopting these designs. Physical Activity and the Cortisol Awakening Response Findings from the current analysis did not support the hypothesized relations between physical activity participation and adequate regulation of the cortisol awakening response. Although there were some observed differences in the variability estimates for the mean cortisol awakening response in the different physical activity subgroups, they were associated with substantial variability and our analysis did not enable us to verify any significant differences. For example, the moderate physical activity subgroup had the lowest absolute degree of variability in the mean cortisol awakening response relative to the low physical activity subgroup, which was PHYSICAL ACTIVITY AND CORTISOL REGULATION 53 expected. However, the high physical activity subgroup exhibited higher absolute variability than the moderate physical activity subgroup. The findings from this review corroborate the proposed ‘inverted-U’ response as described by Anderson and Wideman (2017) – individuals participating in moderate physical activity have less variability about the mean cortisol awakening response while those participating in low or high levels of physical activity participation have higher variability about the mean (Anderson & Wideman, 2017). Although the high variability in the observed mean in each physical activity subgroup suggested the presence of moderators, none of the moderator tests yielded differences in the effects, some non-significant observed differences notwithstanding. One explanation for the higher degree of variability about the mean cortisol awakening response in high versus moderate physical activity subgroups may be that the high physical activity subgroup contains highly trained athletes and individuals participating in particularly high levels of physical activity. Previous research has shown that athletes exhibit blunted cortisol awakening responses after periods of intense physical training and competition (Filaire et al., 2013). Therefore, those engaging in very high levels of physical activity participation likely have a higher percentage of intense training days, whether they are an athlete or not, and this may be why much higher variability was observed in the mean cortisol awakening response in this subgroup. Although we excluded studies that measured the cortisol awakening response within two days before or after competition, highly trained individuals may exhibit more variability on intense training days, possibly contributing to the high degree of heterogeneity and more variability about the mean cortisol awakening response in the high physical activity subgroup. Importantly, the mean estimate for the cortisol awakening response at each level of physical activity participation was outside of the range of cortisol awakening response values PHYSICAL ACTIVITY AND CORTISOL REGULATION 54 that are considered adequately regulated based on population norms (i.e., a rise in cortisol from awakening to peak secretion of 9.3 nmol/l +/- 3.1). Findings suggest that, on average, each subgroup displayed a somewhat blunted cortisol awakening response (Clow et al., 2004; Wüst et al., 2000), which has been associated with fatigue, burnout, and exhaustion (Chida & Steptoe, 2009). However, our analysis adopted normative values for healthy adults presented in previous research – but many studies have reported cortisol awakening response values that lie well outside of the ranges suggested in this literature (e.g., Calogiuri et al, 2016; Gubelmann et al., 2018). Therefore, as moderators of the cortisol awakening response are revealed, accounting for extraneous variables may explain more of the variance in the cortisol awakening response and as the literature proliferates, updating population norms for an adequately regulated cortisol awakening response for use in future research will be necessary. Due to the variability in reported cortisol awakening response population norms, many researchers choose to examine the overall association between physical activity participation and the cortisol awakening response, however this negates the conceptual understanding that a midlevel cortisol awakening response is considered regulated whereas both a heightened and blunted cortisol awakening response have been associated with maladaptive psychological states (Chida & Steptoe, 2009). For example, previous findings have shown that higher levels of physical activity participation are associated with a higher cortisol awakening response values (Gubelmann et al., 2018), but without comparing values to a population norm, it adds little in the overall understanding of the meaning of this association for clinically relevant health purposes. It is, therefore, imperative to account for known moderating factors and re-establish population norms. Further, establishing population norms for samples that may not be classified as healthy PHYSICAL ACTIVITY AND CORTISOL REGULATION 55 adults will also contribute to an understanding of the cortisol awakening response and associated health outcomes. A number of factors may have contributed to the high degree of heterogeneity in both the physical activity-diurnal cortisol slope effect and the physical activity-cortisol awakening response effect in the present analysis that were not accounted for in moderator analyses. For example, various environmental and contextual factors have been shown to influence cortisol awakening response, such as the time of awakening, ambient light exposure, prior day experiences, anticipation of the day ahead, ovulation, jet lag, and alcohol consumption (Adam et al., 2006; Doane et al., 2010; Edwards et al., 2001; Stalder et al., 2009, 2010; Wolfram et al., 2011). We were unable to account for these factors in the current analysis because these data were not reported or unavailable in the pool of studies. These factors may have contributed to high heterogeneity in the mean estimates of the cortisol awakening response in each subgroup. Furthermore, recent evidence has shown that sleep moderates the association between prior-day physical activity and the cortisol awakening response the next morning (Anderson et al., 2021). Sleep may, therefore, have been a further factor contributing to the heterogeneity in the effect. In addition, a wide range of immunoassay methods were utilized to determine cortisol values for the cortisol awakening response, and each has different sensitivities and cortisol estimates from different methods do not compare favorably, which is likely to have contributed to the observed variability in effects (Kirschbaum & Hellhammer, 1989; Miller et al., 2013). Taken together, environmental extraneous variables and the methods used to measure cortisol may have introduced additional error variance to the mean cortisol awakening response in each physical activity subgroup. Results should be interpreted with these potential confounding variables in mind. Future studies need to broaden knowledge of conditions likely to affect the PHYSICAL ACTIVITY AND CORTISOL REGULATION 56 physical activity-cortisol regulation association by systematically examining the effect of these factors in adequately powered studies of the effects using strong experimental design and measures. Key Contributions and Considerations for Future Research This is the first meta-analysis to examine the association between physical activity participation and indices of diurnal cortisol regulation. A key contribution of the current research is that it provides initial empirical support of the link between physical activity participation and the diurnal cortisol slope across multiple studies, an association which is implicated in the mechanism by which physical activity assists in stress coping (Adam et al., 2017; Heim et al., 2008). Since physical activity is associated with a reduction of stress indices and regulation of the diurnal cortisol slope, it may mitigate risk of long-term chronic illness (Pedersen & Saltin, 2015; Warburton & Bredin, 2017; Warburton et al., 2006) and mental health problems (Biddle et al., 2019; Dunn, Trivedi, & O’Neal, 2001) that have been associated with chronic stress. Clinical Implications Although we found a non-zero effect size between physical activity participation and the diurnal cortisol slope, particularly for intervention and experimental studies, the effects were small in size. This has implications for whether changes in the diurnal cortisol slope are sufficient to produce clinically meaningful changes in health outcomes that have been associated with ‘flatter’ diurnal cortisol slopes. Future studies should test whether the effect of physical activity participation on the diurnal cortisol slope also translates to concomitant, clinically meaningful change in health outcomes that are associated with the diurnal cortisol slope. While meta-analytic evidence supports an association between the diurnal cortisol slope and health outcomes (Adam et al., 2017), evidence is needed to evaluate the mediating role of the diurnal PHYSICAL ACTIVITY AND CORTISOL REGULATION 63 Similarly, most studies did not report the time of year that their data were collected. In those that did, a large majority collected data over a span of a year or several years, so all seasons were included. While there is evidence that season of data collection impacts diurnal cortisol patterns (Miller et al., 2016), future studies should report which season(s) data were collected in to control for possible influences. With respect to the cortisol awakening response, we divided effect sizes from the included studies into physical activity subgroups based on physical activity level in order to provide a fit-for-purpose test of the physical activity-cortisol awakening response relations. However, this classification limited the numbers of studies in each activity subgroup, reducing the scope of the moderator analyses due to small numbers of studies in moderator groups. Like the moderator analyses for the physical activity-diurnal cortisol slope association, there were also studies with participants on the boundaries of the cut-off for certain moderators (e.g., age) and participants who changed physical activity subgroup classification over the course of the study. Second, while physical activity subgroup classification was based on the IPAQ classification scheme, physical activity measurements were highly variable, and some studies did not provide sufficient information to classify the sample into the defined subgroups with high precision. Considering the above limitations, caution should be exercised in interpreting current results. Future studies examining the physical activity-cortisol awakening response relations should evaluate the cortisol awakening response using standardized measures that adequately capture the response (Stalder et al., 2016) and measure physical activity using measures with appropriate precision, and adequately report the contextual factors that likely moderate the effect. Conclusion PHYSICAL ACTIVITY AND CORTISOL REGULATION 64 We set out to determine the average size and degree of heterogeneity of the relations between physical activity participation and indices of cortisol regulation, the diurnal cortisol slope and the cortisol awakening response, across available studies using meta-analysis. Consistent with theory and research, findings revealed a non-zero negative correlation between physical activity and the diurnal cortisol slope across studies, supporting the hypothesis that the diurnal cortisol slope is correlated with physical activity participation. For the cortisol awakening response, we did not find strong support that lower levels of variability in the mean cortisol awakening response would be observed at higher levels of physical activity participation. Moderator analyses did not provide clear evidence of moderation and did not resolve the substantive observed heterogeneity in the effect for both indices. There were some noteworthy findings, foremost among them was the non-zero, negative effect of experimental or intervention manipulations of physical activity on the diurnal cortisol slope, providing initial evidence that physical activity interventions may have efficacy in promoting diurnal cortisol regulation. Results also serve to highlight some important gaps in the current literature, such as the need for more precision in physical activity assessment (e.g., including the time of day, current fitness level of participants, intensity, duration, and frequency of participation); more precise cortisol data collection and analysis (e.g., based on consensus guidelines, Stalder et al., 2016); and reporting of sufficient data among all included variables including covariates and cortisol regulation parameters to compute effect sizes for synthesis. Findings also suggest the importance of more research examining the mediating effect of the diurnal cortisol slope between physical activity participation and the subsequent impacts on health outcomes. There is also a need for review studies to examine how other health behaviors relate to cortisol regulation. Studies with factorial designs are also advocated to determine how multiple health behaviors may relate to the PHYSICAL ACTIVITY AND CORTISOL REGULATION 65 diurnal cortisol slope. Further, it is important to note that research in this field is receiving much attention and is evolving (e.g., Anderson, 2021; Ogasawara et al., 2022), and as research in this field expands, future syntheses may directly address some of the knowledge gaps highlighted here, particularly the analysis of moderators of the effects. This research is expected to provide valuable data on the relative contribution of health behaviors in cortisol regulation, which may be utilized to intervene to improve cortisol regulation, and, indirectly, health conditions associated with cortisol dysregulation. Declaration of Interest The authors have no conflicts of interest to disclose. Data Availability Study materials, data files, data analysis scripts, and analysis output presented in this manuscript are available online: https://osf.io/ebpy2/. PHYSICAL ACTIVITY AND CORTISOL REGULATION 66 References Abercrombie, H. C., Giese-Davis, J., Sephton, S., Epel, E. S., Turner-Cobb, J. M., & Spiegel, D. (2004). Flattened cortisol rhythms in metastatic breast cancer patients. Psychoneuroendocrinology, 29(8), 1082–1092. https://doi.org/10.1016/J.PSYNEUEN.2003.11.003 Abramson, J. L., & Vaccarino, V. (2002). Relationship between physical activity and inflammation among apparently healthy middle-aged and older US adults. Archives of Internal Medicine, 162(11), 1286-1292. https://www.doi.org/10.1001/archinte.162.11.1286 Adam, E. K., Hawkley, L. C., Kudielka, B. M., & Cacioppo, J. T. (2006). Day-to-day dynamics of experience–cortisol associations in a population-based sample of older adults. Proceedings of the National Academy of Sciences, 103(45), 17058-17063. https://doi.org/10.1073/pnas.0605053103 Adam, E. K., Heissel, J. A., Zeiders, K. H., Richeson, J. A., Ross, E. C., Ehrlich, K. B., ... & Eccles, J. S. (2015). Developmental histories of perceived racial discrimination and diurnal cortisol profiles in adulthood: A 20-year prospective study. Psychoneuroendocrinology, 62, 279-291. https://doi.org/10.1016/j.psyneuen.2015.08.018 Adam, E. K., & Kumari, M. (2009). Assessing salivary cortisol in large-scale, epidemiological research. Psychoneuroendocrinology, 34(10), 1423-1436. https://doi.org/10.1016/j.psyneuen.2009.06.011 Adam, E. K., Quinn, M. E., Tavernier, R., McQuillan, M. T., Dahlke, K. A., & Gilbert, K. E. (2017). Diurnal cortisol slopes and mental and physical health outcomes: A systematic PHYSICAL ACTIVITY AND CORTISOL REGULATION 67 review and meta-analysis. Psychoneuroendocrinology, 83, 25-41. https://doi.org/10.1016/j.psyneuen.2017.05.018 Allen, J. O., Watkins, D. C., Chatters, L., & Johnson-Lawrence, V. (2019). Mechanisms of racial health disparities: Evidence on coping and cortisol from MIDUS II. Journal of Racial and Ethnic Health Disparities, 7, 207-216. https://doi.org/10.1007/s40615-019-00648-y Anderson, T. (2021). The Impact of Prior Evening Exercise on Endocrine Awakening Responses (Doctoral dissertation, The University of North Carolina at Greensboro). Anderson, T., Corneau, G., Wideman, L., Eddington, K., & Vrshek-Schallhorn, S. (2021). The impact of prior day sleep and physical activity on the cortisol awakening response. Psychoneuroendocrinology, 126, 105131. https://doi.org/10.1016/j.psyneuen.2021.105131 Anderson, T., Lane, A. R., & Hackney, A. C. (2018). The cortisol awakening response: Association with training load in endurance runners. International Journal of Sports Physiology and Performance, 13(9), 1158-1163. https://doi.org/10.1123/ijspp.2017-0740 Anderson, T., & Wideman, L. (2017). Exercise and the cortisol awakening response: A systematic review. Sports Medicine-Open, 3(1), 37. https://doi.org/10.1186/s40798-017- 0102-3 Assink, M., & Wibbelink, C. J. M. (2016). Fitting three-level meta-analytic models in R: A stepby-step tutorial. The Quantitative Methods for Psychology, 12(3), 154-174. https://doi.org/10.20982/tqmp.12.3.p154 Badrick, E., Kirschbaum, C., & Kumari, M. (2007). The relationship between smoking status and cortisol secretion. The Journal of Clinical Endocrinology & Metabolism, 92(3), 819-824. https://doi.org/10.1210/jc.2006-2155 PHYSICAL ACTIVITY AND CORTISOL REGULATION 68 Banasik, J., Williams, H., Haberman, M., Blank, S. E., & Bendel, R. (2009). Effect of Iyengar yoga practice on fatigue and diurnal salivary cortisol concentration in breast cancer survivors. Journal of the American Academy of Nurse Practitioners, 23(3), 135-142. https://www.doi.org/10.1111/j.1745-7599.2010.00573.x Baudrand, R., & Vaidya, A. (2015). Cortisol dysregulation in obesity-related metabolic disorders. Current Opinion in Endocrinology, Diabetes, and Obesity, 22(3), 143. https://www.doi.org/10.1097/MED.0000000000000152 Bei, B., Seeman, T. E., Carroll, J. E., & Wiley, J. F. (2017). Sleep and physiological dysregulation: A closer look at sleep intraindividual variability. Sleep, 40(9), 1-10. http://dx.doi.org/10.1093/sleep/zsx109 Biddle, S. J., Ciaccioni, S., Thomas, G., & Vergeer, I. (2019). Physical activity and mental health in children and adolescents: An updated review of reviews and an analysis of causality. Psychology of Sport and Exercise, 42, 146-155. https://doi.org/10.1016/j.psychsport.2018.08.011 Bogg, T. & Slatcher, R. B. (2015). Activity mediates conscientiousness’ relationship to diurnal cortisol slope in a national sample. Health Psychology, 34(12), 1195-1199. https://www.doi.org/10.1037/hea0000243 Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction to metaanalysis. Wiley. https://doi.org/10.1002/9780470743386 Booth, M.L. (2000). Assessment of Physical Activity: An International Perspective. Research Quarterly for Exercise and Sport, 71 (2): s114-20. Bower, J. E., Greendale, G., Crosswell, A. D., Garet, D., Sternlieb, B., Ganz, P. A., Irwin, M. R., Olmstead, R., Arevalo, J., & Cole, S. W. (2014). Yoga reduces inflammatory signaling in PHYSICAL ACTIVITY AND CORTISOL REGULATION 69 fatigued breast cancer survivors: A randomized controlled trial. Psychoneuroendocrinology, 43, 20-29. https://doi.org/10.1016/j.psyneuen.2014.01.019 Brendgen, M., Ouellet-Morin, I., Girard, A., Lupien, S., Vitaro, F., Dionne, G., & Boivin, M. (2017). Environmental influence of problematic social relationships on adolescents’ daily cortisol secretion: A monozygotic (MZ) twin difference study. Psychological Medicine, 47(3), 460-470. https://doi.org/10.1017/S003329171600252X Bruce, J., Davis, E. P., & Gunnar, M. R. (2002). Individual differences in children’s cortisol response to the beginning of a new school year. Psychoneuroendocrinology, 27(6), 635- 650. https://doi.org/10.1016/S0306-4530(01)00031-2 Bullock, N., Martin, D. T., Ross, A., Rosemond, D., & Marino, F. E. (2007). Effect of long haul travel on maximal sprint performance and diurnal variations in elite skeleton athletes. British Journal of Sports Medicine, 41(9), 569-573. https://www.doi.org/10.1136/bjsm.2006.033233 Cadegiani, F. A., Kater, C. E., & Gazola, M. (2019). Clinical and biochemical characteristics of high-intensity functional training (HIFT) and overtraining syndrome: Findings from the EROS study (The EROS-HIFT). Journal of Sports Sciences, 37(11), 1296-1307. https://doi.org/10.1080/02640414.2018.1555912 Calogiuri, G., Evensen, K., Weydahl, A., Andersson, K., Patil, G., Ihlebæk, C., & Raanaas, R. K. (2016). Green exercise as a workplace intervention to reduce job stress. Results from a pilot study. Work, 53(1), 99-111. https://www.doi.org/10.3233/WOR-152219 Carter, E. C., Schönbrodt, F. D., Gervais, W. M., & Hilgard, J. (2019). Correcting for bias in psychology: A comparison of meta-analytic methods. Advances in Methods and Practices in Psychological Science, 2(2), 115-144. PHYSICAL ACTIVITY AND CORTISOL REGULATION 70 https://doi.org/10.1177/2515245919847196 Chandwani, K. D., Perkins, G., Nagendra, H. R., Raghuram, N. V., Spelman, A., Nagarathna, R., Johnson, K., Fortier, A., Arun, B., Wei, Q., Kirschbaum, C., Haddad, R., Morris, G. S., Scheetz, J., Chaoul, A., & Cohen, L. (2014). Randomized, controlled trial of yoga in women with breast cancer undergoing radiotherapy. Journal of Clinical Oncology, 32(10), 1058-1065. https://www.doi.org/10.1200/JCO.2012.48.2752 Chen, C., Nakagawa, S., An, Y., Ito, K., Kitaichi, Y., & Kusumi, I. (2017). The exerciseglucocorticoid paradox: How exercise is beneficial to cognition, mood, and the brain while increasing glucocorticoid levels. Frontiers in Neuroendocrinology, 44, 83-102. https://doi.org/10.1016/j.yfrne.2016.12.001 Cheung, M. W. L. (2014). Modeling dependent effect sizes with three-level meta-analyses: A structural equation modeling approach. Psychological Methods, 19(2), 211-229. https://doi.org/10.1037/a0032968 Chin, B., Murphy, M. L. M., & Cohen, S. (2018). Age moderates the association between social integration and diurnal cortisol measures. Psychoneuroendocrinology, 90, 102-109. https://www.doi.org/10.1016/j.psyneuen.2018.02.008 Christopher, M. S., Goerling, R. J., Rogers, B. S., Hunsinger, M., Baron, G., Bergman, A. L., & Zava, D. T. (2016). A pilot study evaluating the effectiveness of a mindfulness-based intervention on cortisol awakening response and health outcomes among law enforcement officers. Journal of Police and Criminal Psychology, 31(1), 15-28. https://doi.org/10.1007/s11896-015-9161-x Chrousos, G. P., & Gold, P. W. (1992). The concepts of stress and stress system disorders: Overview of physical and behavioral homeostasis. Jama, 267(9), 1244-1252. PHYSICAL ACTIVITY AND CORTISOL REGULATION 71 https://www.doi.org/10.1001/jama.1992.03480090092034 Chida, Y., & Steptoe, A. (2009). Cortisol awakening response and psychosocial factors: A systematic review and meta-analysis. Biological Psychology, 80(3), 265-278. https://doi.org/10.1016/j.biopsycho.2008.10.004 Clow, A., Hucklebridge, F., Stalder, T., Evans, P., & Thorn, L. (2010a). The cortisol awakening response: More than a measure of HPA axis function. Neuroscience & Biobehavioral Reviews, 35(1), 97-103. https://doi.org/10.1016/j.neubiorev.2009.12.011 Clow, A., Thorn, L., Evans, P., & Hucklebridge, F. (2004). The awakening cortisol response: Methodological issues and significance. Stress, 7(1), 29-37. https://doi.org/10.1080/10253890410001667205 Cohen, S., Schwartz, J. E., Epel, E., Kirschbaum, C., Sidney, S., & Seeman, T. (2006). Socioeconomic status, race, and diurnal cortisol decline in the Coronary Artery Risk Development in Young Adults (CARDIA) Study. Psychosomatic Medicine, 68(1), 41-50. https://www.doi.org/10.1097/01.psy.0000195967.51768.ea Corey, S. M., Epel, E., Schembri, M., Pawlowsky, S. B., Cole, R. J., Araneta, M. R. G., Barrett- Connor, E., & Kanaya, A. M. (2014). Effect of restorative yoga vs. stretching on diurnal cortisol dynamics and psychosocial outcomes in individuals with the metabolic syndrome: The PRYSMS randomized controlled trial. Psychoneuroendocrinology, 49, 260-271. https://doi.org/10.1016/j.psyneuen.2014.07.012 Dallman, M. F., Akana, S. F., Cascio, C. S., Darlington, D. N., Jacobson, L., & Levin, N. (1987). Regulation of ACTH secretion: Variations on a theme of B. Recent Progress in Hormone Research, 43, 113-173. https://doi.org/10.1016/B978-0-12-571143-2.50010-1 PHYSICAL ACTIVITY AND CORTISOL REGULATION 72 Daly, M., Baumeister, R. F., Delaney, L., & MacLachlan, M. (2014). Self-control and its relation to emotions and psychobiology: Evidence from a Day Reconstruction Method study. Journal of Behavioral Medicine, 37(1), 81-93. https://doi.org/10.1007/s10865-012-9470- 9 Davis, E. P., & Sandman, C. A. (2010). The timing of prenatal exposure to maternal cortisol and psychosocial stress is associated with human infant cognitive development. Child Development, 81(1), 131-148. https://doi.org/10.1111/j.1467-8624.2009.01385.x DeSantis, A. S., Adam, E. K., Doane, L. D., Mineka, S., Zinbarg, R. E., & Craske, M. G. (2007). Racial/ethnic differences in cortisol diurnal rhythms in a community sample of adolescents. Journal of Adolescent Health, 41(1), 3-13. https://doi.org/10.1016/j.jadohealth.2007.03.006 Di Corrado, D., Agostini, T., Bonifazi, M., & Perciavalle, V. (2014). Changes in mood states and salivary cortisol levels following two months of training in elite female water polo players. Molecular Medicine Reports, 9(6), 2441-2446. https://doi.org/10.3892/mmr.2014.2115 Digby, P. G. N. (1983). Approximating the tetrachoric correlation coefficient. Biometrics, 39(3), 753-757. https://doi.org/10.2307/2531104 Diorio, D., Viau, V., & Meaney, M. J. (1993). The role of the medial prefrontal cortex (cingulate gyrus) in the regulation of hypothalamic-pituitary-adrenal responses to stress. Journal of Neuroscience, 13(9), 3839-3847. https://doi.org/10.1523/JNEUROSCI.13-09-03839.1993 Doane, L. D., Kremen, W. S., Eaves, L. J., Eisen, S. A., Hauger, R., Hellhammer, D., Levine, S., Lupien, S., Lyons, M. J., Mendoza, S., Prom-Wormley, E., Xian, H., York, T. P., Franz, C. E., & Jacobson, K. C. (2010). Associations between jet lag and cortisol diurnal PHYSICAL ACTIVITY AND CORTISOL REGULATION 79 patients. Psychoneuroendocrinology, 87, 119-126. https://doi.org/10.1016/j.psyneuen.2017.10.012 Holleman, M., Vreeburg, S. A., Dekker, J. J. M., & Penninx, B. W. J. H. (2012). The relationships of working conditions, recent stressors and childhood trauma with salivary cortisol levels. Psychoneuroendocrinology, 37(6), 801-809. http://dx.doi.org/10.1016/j.psyneuen.2011.09.012 Holtermann, A., Hansen, J. V., Burr, H., Søgaard, K., & Sjøgaard, G. (2012). The health paradox of occupational and leisure-time physical activity. British Journal of Sports Medicine, 46(4), 291-295. https://www.doi.org/10.1136/bjsm.2010.079582 Holtermann, A., Krause, N., Van Der Beek, A. J., & Straker, L. (2018). The physical activity paradox: Six reasons why occupational physical activity (OPA) does not confer the cardiovascular health benefits that leisure time physical activity does. British Journal of Sports Medicine, 52(3), 149-150. http://dx.doi.org/10.1136/bjsports-2017-097965 Holtermann, A., Mortensen, O. S., Burr, H., Søgaard, K., Gyntelberg, F., & Suadicani, P. (2010). Physical demands at work, physical fitness, and 30-year ischaemic heart disease and allcause mortality in the Copenhagen Male Study. Scandinavian Journal of Work, Environment & Health, 357-365. https://www.doi.org/10.5271/sjweh.2913 Hsiao, F.-H., Jow, G.-M., Kuo, W.-H., Chang, K.-H., Liu, Y.-F., Ho, R. T. H., Ng, S.-M., Chan, C. L. W., Lai, Y.-M., & Chen, Y.-T. (2012). The effects of psychotherapy on psychological well-being and diurnal cortisol patterns in breast cancer survivors. Psychotherapy and Psychosomatics, 81(3), 173-182. http://dx.doi.org/10.1159%2F000329178 PHYSICAL ACTIVITY AND CORTISOL REGULATION 80 Hucklebridge, F. H., Clow, A., Abeyguneratne, T., Huezo-Diaz, P., & Evans, P. (1999). The awakening cortisol response and blood glucose levels. Life Sciences, 64(11), 931-937. https://doi.org/10.1016/S0024-3205(99)00019-3 Ice, G. H. (2005). Factors influencing cortisol level and slope among community dwelling older adults in Minnesota. Journal of Cross-Cultural Gerontology, 20(2), 91-108. https://doi.org/10.1007/s10823-005-9085-5 Ice, G. H., Katz-Stein, A., Himes, J., & Kane, R. L. (2004). Diurnal cycles of salivary cortisol in older adults. Psychoneuroendocrinology, 29(3), 355-370. https://doi.org/10.1016/S0306- 4530(03)00034-9 Imboden, C., Gerber, M., Beck, J., Eckert, A., Lejri, I., Pühse, U., Holsboer-Trachsler, E., & Hatzinger, M. (2021). Aerobic exercise and stretching as add-on to inpatient treatment for depression have no differential effects on stress-axis activity, serum-BDNF, TNF-alpha and objective sleep measures. Brain Sciences, 11(4), 411. https://doi.org/10.3390/brainsci11040411 Iyengar, S., & Greenhouse, J. B. (2009). Sensitivity analysis and diagnostics. In H. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (pp. 417-433), Russell Sage Foundation. Jankord, R., & Herman, J. P. (2008). Limbic regulation of hypothalamo‐pituitary‐adrenocortical function during acute and chronic stress. Annals of the New York Academy of Sciences, 1148(1), 64-73. https://doi.org/10.1196/annals.1410.012 Jaremka, L. M., Sunami, N., & Nadzan, M. A. (2018, April). Chronic loneliness amplifies agerelated differences in diurnal cortisol trajectories. Psychosomatic Medicine (80)3, A40- A40). Conference meeting abstract. PHYSICAL ACTIVITY AND CORTISOL REGULATION 81 Kallen, V. L., Stubbe, J. H., Zwolle, H. J., & Valk, P. (2017). Capturing effort and recovery: Reactive and recuperative cortisol responses to competition in well-trained rowers. BMJ Open Sport & Exercise Medicine, 3(1), Article e000235. http://dx.doi.org/10.1136/bmjsem-2017-000235 Keiver, K., Bertram, C. P., Orr, A. P., & Clarren, S. (2015). Salivary cortisol levels are elevated in the afternoon and at bedtime in children with prenatal alcohol exposure. Alcohol, 49(1), 79-87. http://dx.doi.org/10.1016/j.alcohol.2014.11.004 Kern, W., Perras, B., Wodick, R., Fehm, H. L., & Born, J. (1995). Hormonal secretion during nighttime sleep indicating stress of daytime exercise. Journal of Applied Physiology, 79(5), 1461-1468. https://doi.org/10.1152/jappl.1995.79.5.1461 Kim, H. G., Lim, E. Y., Jung, W. R., Shin, M. K., Ann, E. S., & Kim, K. L. (2008). Effects of treadmill exercise on hypoactivity of the hypothalamo-pituitary-adrenal axis induced by chronic administration of corticosterone in rats. Neuroscience Letters, 434(1), 46-49. https://doi.org/10.1016/j.neulet.2008.01.025 Kirschbaum, C., & Hellhammer, D. H. (1989). Salivary cortisol in psychobiological research: An overview. Neuropsychobiology, 22(3), 150-169. https://doi.org/10.1159/000118611 Küüsmaa, M., Schumann, M., Sedliak, M., Kraemer, W. J., Newton, R. U., Malinen, J. P., ... & Häkkinen, K. (2016). Effects of morning versus evening combined strength and endurance training on physical performance, muscle hypertrophy, and serum hormone concentrations. Applied Physiology, Nutrition, and Metabolism, 41(12), 1285-1294. https://doi.org/10.1139/apnm-2016-0271 PHYSICAL ACTIVITY AND CORTISOL REGULATION 82 Law, R., Hucklebridge, F., Thorn, L., Evans, P., & Clow, A. (2013). State variation in the cortisol awakening response. Stress, 16(5), 483-492. https://www.doi.org/10.3109/10253890.2013.817552 Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. Springer publishing company. Lear, S. A., Hu, W., Rangarajan, S., Gasevic, D., Leong, D., Iqbal, R., ... & Yusuf, S. (2017). The effect of physical activity on mortality and cardiovascular disease in 130 000 people from 17 high-income, middle-income, and low-income countries: The PURE study. The Lancet, 390(10113), 2643-2654. https://doi.org/10.1016/S0140-6736(17)31634-3 Lederbogen, F., Kühner, C., Kirschbaum, C., Meisinger, C., Lammich, J., Holle, R., & Krumm, B. (2010). Salivary cortisol in a community sample, 50–69 years of age. Results from 990 men and women of the KORA-F3 Augsburg study. European Journal of Endocrinology, 163(3), 443-451. https://www.doi.org/10.1530/eje-10-0491 Logel, C., Hall, W., Page-Gould, E., & Cohen, G. L. (2019). Why is it so hard to change? The role of self-integrity threat and affirmation in weight loss. European Journal of Social Psychology, 49(4), 748-759. https://doi.org/10.1002/ejsp.2536 Lucertini, F., Ponzio, E., Di Palma, M., Galati, C., Federici, A., Barbadoro, P., ... & Minelli, A. (2015). High cardiorespiratory fitness is negatively associated with daily cortisol output in healthy aging men. PloS one, 10(11), e0141970. https://doi.org/10.1371/journal.pone.0141970 Lupien, S. J., McEwen, B. S., Gunnar, M. R., & Heim, C. (2009). Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nature Reviews Neuroscience, 10(6), 434- 445. https://doi.org/10.1038/nrn2639 PHYSICAL ACTIVITY AND CORTISOL REGULATION 83 MacDonald, D. & Wetherell, M. (2019). Competition stress leads to a blunting of the cortisol awakening response in elite rowers. Frontiers in Psychology, 10(1684). https://doi.org/10.3389/fpsyg.2019.01684 McHale, S. M., Blocklin, M. K., Walter, K. N., Davis, K. D., Almeida, D. M., & Klein, L. C. (2012). The role of daily activities in youths' stress physiology. Journal of Adolescent Health, 51(6), 623-628. https://doi.org/10.1016/j.jadohealth.2012.03.016 McIsaac, S. A., & Young, A. H. (2009). The role of hypothalamic pituitary-adrenal axis dysfunction in the etiology of depressive disorders. Drugs of Today, 45(2), 127. https://www.doi.org/10.1358/dot.2009.45.2.1336104 McTiernan, A. N. N. E., Friedenreich, C. M., Katzmarzyk, P. T., Powell, K. E., Macko, R., Buchner, D., ... & Piercy, K. L. (2019). Physical activity in cancer prevention and survival: A systematic review. Medicine and Science in Sports and Exercise, 51(6), 1252. https://www.doi.org/10.1249/MSS.0000000000001937 Meland, A., Ishimatsu, K., Pensgaard, A. M., Wagstaff, A., Fonne, V., Garde, A. H., & Harris, A. (2015). Impact of mindfulness training on physiological measures of stress and objective measures of attention control in a military helicopter unit. The International Journal of Aviation Psychology, 25(3-4), 191-208. https://www.doi.org/10.1080/10508414.2015.1162639 Menke, A., Arloth, J., Gerber, M., Rex-Haffner, M., Uhr, M., Holsboer, F., ... & Beck, J. (2014). Dexamethasone stimulated gene expression in peripheral blood indicates glucocorticoidreceptor hypersensitivity in job-related exhaustion. Psychoneuroendocrinology, 44, 35- 46. https://doi.org/10.1016/j.psyneuen.2014.02.013 PHYSICAL ACTIVITY AND CORTISOL REGULATION 84 Michels, N., Clays, E., De Buyzere, M., Huybrechts, I., Vanaelst, B., De Henauw, S., & Sioen, I. (2012). Stress biomarkers in children: heart rate variability versus salivary cortisol. In 42nd Annual conference of the International Society of Psychoneuroendocrinology (ISPNE): Effects of traumatic stress: molecular and hormonal mechanisms, 3(1), 39-40. Conference meeting abstract. Michels, N., Sioen, I., Boone, L., Clays, E., Vanaelst, B., Huybrechts, I., & De Henauw, S. (2015). Cross-lagged associations between children’s stress and adiposity. Psychosomatic Medicine, 77(1), 50-58. https://doi.org/10.1097/PSY.0000000000000122 Michels, N., Sioen, I., Clays, E., De Buyzere, M., Ahrens, W., Huybrechts, I., Vanaelst, B., & De Henauw, S. (2013). Children’s heartrate variability as stress indicator: Association with reported stress and cortisol. Biological Psychology, 94(2), 433-440. http://dx.doi.org/10.1016/j.biopsycho.2013.08.005 Mikkelsen, S., Forman, J. L., Fink, S., Vammen, M. A., Thomas, J. F., Grynderup, M. B., Hansen, A. M., Kaerlev, L., Kolstad, H. A., Rugulies, R., & Bonde, J. P. (2017). Prolonged perceived stress and saliva cortisol in a large cohort of Danish public service employees: Cross-sectional and longitudinal associations. International Archives of Occupational and Environmental Health, 90(8), 835-848. http://dx.doi.org/10.1007/s00420-017-1241-z Miller, G. E., Chen, E., & Zhou, E. S. (2007). If it goes up, must it come down? Chronic stress and the hypothalamic-pituitary-adrenocortical axis in humans. Psychological Bulletin, 133(1), 25–45. https://doi.org/10.1037/0033-2909.133.1.25 Miller, R., & Plessow, F. (2013). Transformation techniques for cross-sectional and longitudinal endocrine data: application to salivary cortisol concentrations. PHYSICAL ACTIVITY AND CORTISOL REGULATION 85 Psychoneuroendocrinology, 38(6), 941-946. https://doi.org/10.1016/j.psyneuen.2012.09.013 Miller, R., Plessow, F., Rauh, M., Gröschl, M., & Kirschbaum, C. (2013). Comparison of salivary cortisol as measured by different immunoassays and tandem mass spectrometry. Psychoneuroendocrinology, 38(1), 50-57. https://doi.org/10.1016/j.psyneuen.2012.04.019 Miller, R., Stalder, T., Jarczok, M., Almeida, D. M., Badrick, E., Bartels, M., ... & Kirschbaum, C. (2016). The CIRCORT database: Reference ranges and seasonal changes in diurnal salivary cortisol derived from a meta-dataset comprised of 15 field studies. Psychoneuroendocrinology, 73, 16-23. https://doi.org/10.1016/j.psyneuen.2016.07.201 Mizoguchi, K., Ishige, A., Aburada, M., & Tabira, T. (2003). Chronic stress attenuates glucocorticoid negative feedback: Involvement of the prefrontal cortex and hippocampus. Neuroscience, 119(3), 887-897. https://doi.org/10.1016/S0306-4522(03)00105-2 Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., & Prisma Group. (2009). Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. PLoS Med, 6(7), e1000097. https://doi.org/10.1371/journal.pmed.1000097 Nagy, T., Salavecz, G., Simor, P., Purebl, G., Bódizs, R., Dockray, S., & Steptoe, A. (2015). Frequent nightmares are associated with blunted cortisol awakening response in women. Physiology & Behavior, 147, 233-237. https://doi.org/10.1016/j.physbeh.2015.05.001 Nordahl-Hansen, A., Øien, R. A., Volkmar, F., Shic, F., & Cicchetti, D. V. (2018). Enhancing the understanding of clinically meaningful results: A clinical research perspective. Psychiatry Research, 270, 801-806. https://doi.org/10.1016/j.psychres.2018.10.069 PHYSICAL ACTIVITY AND CORTISOL REGULATION 86 O'Connor, D. B., Thayer, J. F., & Vedhara, K. (2020). Stress and health: A review of psychobiological processes. Annual Review of Psychology, 72. https://doi.org/10.1146/annurev-psych-062520-122331 O’Flynn, J., Dinan, T. G., & Kelly, J. R. (2018). Examining stress: An investigation of stress, mood and exercise in medical students. Irish Journal of Psychological Medicine, 20(25). https://doi.org/10.1017/ipm.2017.54 Ogasawara, Y., Kadooka, S., Tsuchiya, H., & Sugo, T. (2022). High cortisol awakening response measured on day following high-intensity exercise. The Journal of Physical Fitness and Sports Medicine, 11(2), 59-66. https://doi.org/10.7600/jpfsm.11.59 Parks, C. G., Miller, D. B., McCanlies, E. C., Cawthon, R. M., Andrew, M. E., DeRoo, L. A., & Sandler, D. P. (2009). Telomere length, current perceived stress, and urinary stress hormones in women. Cancer Epidemiology and Prevention Biomarkers, 18(2), 551-560. https://www.doi.org/10.1158/1055-9965.EPI-08-0614 Pascoe, M. C., Thompson, D. R., Ski, C. F. (2017) Yoga, mindfulness-based stress reduction and stress-related physiological measures: A meta-analysis. Psychoneuroendocrinology, 86, 152-168. https://doi.org/10.1016/j.psyneuen.2017.08.008 Pauly, T., Michalowski, V. I., Nater, U. M., Gerstorf, D., Ashe, M. C., Madden, K. M., & Hoppmann, C. A. (2019). Everyday associations between older adults’ physical activity, negative affect, and cortisol. Health Psychology, 38(6), 494-501. https://psycnet.apa.org/doi/10.1037/hea0000743 PHYSICAL ACTIVITY AND CORTISOL REGULATION 87 Pedersen, B. K., & Saltin, B. (2015). Exercise as medicine–evidence for prescribing exercise as therapy in 26 different chronic diseases. Scandinavian Journal of Medicine & Science in Sports, 25(S3), 1-72. https://doi.org/10.1111/sms.12581 Peterson, J. C., Charlson, M. E., Wells, M. T., & Altemus, M. (2014). Depression, coronary artery disease, and physical activity: How much exercise is enough? Clinical Therapeutics, 36(11), 1518-1530. https://doi.org/10.1016/j.clinthera.2014.10.003 Prick, A. E., de Lange, J., Twisk, J., & Pot, A. M. (2015). The effects of a multi-component dyadic intervention on the psychological distress of family caregivers providing care to people with dementia: A randomized controlled trial. International Psychogeriatrics, 27(12), 2031-2044. http://dx.doi.org/10.1017/S104161021500071X Puetz, T.W. (2006). Physical activity and feelings of energy and fatigue. Sports Medicine, 36, 767–780. https://doi.org/10.2165/00007256-200636090-00004 Pulopulos, M. M., Hidalgo, V., Puig-Perez, S., Montoliu, T., & Salvador, A. (2020). Relationship between cortisol changes during the night and subjective and objective sleep quality in healthy older people. International Journal of Environmental Research and Public Health, 17(4), 1264. http://dx.doi.org/10.3390/ijerph17041264 Pulopulos, M. M., Puig-Perez, S., Hidalgo, V., Villada, C., & Salvador, A. (2016). Cortisol awakening response and walking speed in older people. PLoS ONE, 11(5), Article e0152071. https://doi.org/10.1371/journal.pone.0152071 Rahman, M. S., Zhao, X., Liu, J. J., Torres, E. Q., Tibert, B., Kumar, P., Kaldo, V., Lindefors, N., Forsell, Y., & Lavebratt, C. (2018). Exercise reduces salivary morning cortisol levels in patients with depression. Molecular Neuropsychiatry, 4(4), 196-203. https://www.doi.org/10.1159/000494699 PHYSICAL ACTIVITY AND CORTISOL REGULATION 88 Ratcliff, C. G., Milbury, K., Chandwani, K. D., Chaoul, A., Perkins, G., Nagarathna, R., Haddad, R., Nagendra, H. R., Raghuram, N. V., Spelman, A., Arun, B., Wei, Q., & Cohen, L. (2016). Examining mediators and moderators of yoga for women with breast cancer undergoing radiotherapy. Integrative Cancer Therapies, 15(3), 250-262. https://www.doi.org/10.1177/1534735415624141 Rodenbeck, A., & Hajak, G. (2001). Neuroendocrine dysregulation in primary insomnia. Revue Neurologique, 157(11 Pt 2), S57-61. No DOI. Rosenthal, M. C. (1994). The fugitive literature. In H. Cooper & L. V. Hedges (Eds.), The handbook of research synthesis (p. 85–94). Russell Sage Foundation. Rosmond, R., & Bjorntorp, P. (2000). The hypothalamic-pituitary-adrenal axis activity as a predictor of cardiovascular disease, type 2 diabetes and stroke. Journal of Internal Medicine, 247(2), 188–197. https://doi.org/10.1046/j.1365-2796.2000.00603.x Sabiston, C. M., Wrosch, C., Fong, A. J., Brunet, J., Gaudreau, P., O’Laughlin, J., & Meterissian, S. (2018). Life after breast cancer: Moving on, sitting down or standing still? A prospective study of Canadian breast cancer survivors. BMJ Open, 8(7), Article e021770. http://dx.doi.org/10.1136/bmjopen-2018-021770 Sapolsky, R. M., Romero, L. M., & Munck, A. U. (2000). How do glucocorticoids influence stress responses? Integrating permissive, suppressive, stimulatory, and preparative actions. Endocrine Reviews, 21(1), 55-89. https://doi.org/10.1210/edrv.21.1.0389 Saxbe, D. E. (2008). A field (researcher's) guide to cortisol: Tracking HPA axis functioning in everyday life. Health Psychology Review, 2(2), 163-190. https://doi.org/10.1080/17437190802530812 Schenker, N., & Gentleman, J. F. (2001). On judging the significance of differences by PHYSICAL ACTIVITY AND CORTISOL REGULATION 95 study conducted with personality disorder samples. BMC Medical Research Methodology, 13(1), 1-7. https://www.doi.org/10.1186/1471-2288-13-61 Wüst, S., Federenko, I., Hellhammer, D. H., & Kirschbaum, C. (2000). Genetic factors, perceived chronic stress, and the free cortisol response to awakening. Psychoneuroendocrinology, 25(7), 707-720. https://doi.org/10.1016/S0306- 4530(00)00021-4 Yu, T., Zhou, W., Wu, S., Liu, Q., & Li, X. (2020). Evidence for disruption of diurnal salivary cortisol rhythm in childhood obesity: Relationships with anthropometry, puberty and physical activity. BMC Pediatrics, 20(1), 1-10. https://doi.org/10.1186/s12887-020- 02274-8 Zeiders, K. H., Causadias, J. M., & White, R. M. (2018). The health correlates of culture: Examining the association between ethnic-racial identity and diurnal cortisol slopes. Journal of Adolescent Health, 62(3), 349-351. https://doi.org/10.1016/j.jadohealth.2017.09.020 Zschucke, E., Renneberg, B., Dimeo, F., Wüstenberg, T., & Ströhle, A. (2015). The stressbuffering effect of acute exercise: Evidence for HPA axis negative feedback. Psychoneuroendocrinology, 51, 414-425. https://doi.org/10.1016/j.psyneuen.2014.10.019