Physical activity and cortisol regulation: A meta-analysis

Dysregulated diurnal secretion patterns of the hormone cortisol are implicated in mediating associations between stress exposure and health outcomes (Chrousos & Gold, 1992; Davis & Sandman, 2010; Lupien et al., 2009), including the onset and progression of mental and physical health disorders (Heim et al., 2008). Physical activity participation has been associated with lower levels of cortisol secretion during times of stress and may contribute to effective regulation of cortisol secretion over time. Adequate HPA-axis regulation is one potential mechanism by which physical activity participation may impact stress-related outcomes and quantifying the effect across studies will demonstrate the degree to which physical activity participation is associated with effective HPA-axis regulation and help to identify the potential conditions or factors that may affect it. We synthesized research examining the association between physical activity participation and two distinct indices of effective HPA-axis regulation; the diurnal cortisol slope and the cortisol awakening response, as well as factors that moderate this relationship across existent literature. The three-level meta-analysis revealed a small, non-zero negative correlation between physical activity and the diurnal cortisol slope with high heterogeneity. Moderator analyses examining effects of sample sociodemographic differences, study design characteristics, cortisol measurement methods, and physical activity-related differences indicated few differences in the averaged physical activity-diurnal cortisol slope relationship across studies at different levels of each moderator. There were observed differences effect in some isolated cases, for example, the correlation was slightly larger and non-zero in experimental studies compared to observational studies in which the correlation was smaller and no different from zero. However, overall, evidence for the effect of moderators on the physical activity-diurnal cortisol slope relationship was weak. For the cortisol awakening response, while the variability estimates about the averaged mean cortisol awakening response were lowest in the moderate physical activity subgroup when compared with the high and low physical activity subgroups, confidence intervals about the variance estimates overlapped considerably across subgroups and did not provide definitive evidence to support lower levels of variability in the mean cortisol awakening response at higher levels of physical activity participation. Moderator analyses did not reduce heterogeneity in the effect in most cases, and confidence intervals indicated no differences in variability estimates for the averaged mean at most levels of the moderator. Overall, findings suggest physical activity participation is associated with an adequately regulated or steeper diurnal cortisol slope. However, physical activity level was not associated with an adequately regulated cortisol awakening response as the variability about the mean cortisol awakening response across high, moderate, and low physical activity subgroups did not differ. Based on current findings, studies that test the relationship between physical activity and cortisol regulation relationship should consider using standardized measures of physical activity including frequency, intensity, and duration of physical activity participation, follow consensus guidelines for cortisol sample collection and analysis, and test these relationships in large-scale empirical studies to confirm the direction and causality of the effect. Supplemental materials submitted with this dissertation include the following: Table 1: Results of multi-level meta-analysis of physical activity participation and the diurnal cortisol slope Table 2: Results of multi-level meta-analysis of physical activity participation and the cortisol awakening response Table 3: Moderator analyses of the association between physical activity participation and the diurnal cortisol slope Table 4: Moderator analyses of the mean cortisol awakening response in each physical activity subgroup Appendix A: PRISMA Flow Diagram Figure A1: Flow diagram illustrating literature search and study selection and retention in meta Appendix B: Multiple and overlapping studies Table B1: Studies included in diurnal cortisol slope meta-analysis with multiple samples or measures of constructs Table B2: Studies included in cortisol awakening response meta-analysis with multiple samples or measures of constructs Appendix C: Study characteristics for the diurnal cortisol slope analysis Table C1: Summary characteristics and moderator coding for study design and sample characteristics of studies included in the diurnal cortisol slope meta-analysis Table C2: Summary characteristics and moderator coding of physical activity measurement in studies included in the diurnal cortisol slope meta-analysis Table C3: Summary characteristics and moderator coding of diurnal cortisol measurement in studies included in the diurnal cortisol slope meta-analysis Appendix D: Study characteristics for the cortisol awakening response analysis Table D1: Summary characteristics and moderator coding for study design and sample characteristics of studies included in the cortisol awakening response analysis meta-analysis Table D2: Summary characteristics and moderator coding of physical activity measurement in studies included in cortisol awakening response analysis meta-analysis Table D3: Summary characteristics and moderator coding of diurnal cortisol measurement in studies included in the cortisol awakening response analysis meta-analysis Appendix E: Moderator coding protocol Table E1: Moderator protocol applied to all studies Appendix F: Coding for comparison analyses of sampling dependencies in the cortisol awakening response meta-analysis Table F1: Coding of tested sampling dependencies in the multi-level meta-analysis of physical activity participation and the cortisol awakening response comparisons Appendix G: Model comparison results comparing sample dependencies coding for cortisol awakening response meta-analysis Table G1: Model comparisons of multi-level meta-analysis of physical activity participation and the cortisol awakening response Appendix H: Studies included in the diurnal cortisol slope meta-analysis Appendix I: Studies included in the cortisol awakening response meta-analysis Appendix J: PRISMA Checklist Appendix K: Meta-regression analyses of moderators Table K1: Meta-regression analysis of categorical moderators in the physical activity-diurnal cortisol slope meta-analysis Table K2: Meta-regression analysis of categorical variables plus time lag as a continuous moderator in the physical activity-diurnal cortisol slope meta-analysis Table K3: Meta-regression analysis of categorical variables plus methodological quality of cortisol sampling as a continuous moderator in the physical activity-diurnal cortisol slope meta-analysis Table K4: Meta-regression analysis of categorical variables plus physical activity duration per bout as a continuous moderator in the physical activity-diurnal cortisol slope meta-analysis Table K5: Meta-regression analysis of categorical moderators in the physical activity-cortisol awakening response meta-analysis Table K6: Meta-regression analysis of categorical variables plus time lag as a continuous moderator in the physical activity-cortisol awakening response meta-analysis Table K7: Meta-regression analysis of categorical variables plus methodological quality of cortisol sampling as a continuous moderator in the physical activity-cortisol awakening response meta-analysis Appendix L: Funnel plot for examination of publication bias Figure L1: Funnel plot of diurnal cortisol slope effect size on physical activity engagement (standardized difference in means) against study precision (standard error) for visual inspection of publication bias Figure L2: Funnel plot of cortisol awakening response mean on physical activity engagement (standardized difference in means) against study precision (standard error) for visual inspection of publication bias Figure L2a: Low physical activity subgroup Figure L2b: Moderate physical activity subgroup Figure L2c: High physical activity subgroup Appendix M: ‘Leave-one-out’ sensitivity analysis for outliers Table M1: Results of the leave-one-out sensitivity meta-analysis for inspection of outliers in the physical activity-diurnal cortisol slope meta-analysis Table M2: Results of the leave-one-out sensitivity meta-analysis for inspection of outliers in the physical activity-cortisol awakening response meta-analysis in the low physical activity subgroup only Table M3: Results of the leave-one-out sensitivity meta-analysis for inspection of outliers in the physical activity-cortisol awakening response meta-analysis in the high physical activity subgroup only Table M4: Results of the leave-one-out sensitivity meta-analysis for inspection of outliers in the physical activity-cortisol awakening response meta-analysis in the moderate physical activity subgroup onl