Design Decisions to Optimize Reliability of Daytime Cortisol Slopes in an Older Population

The daytime log-cortisol slope appears to be of growing importance in studying the relationship between stress and health. How best to estimate that slope with minimal burden to the participants and the cost of the study is a decision often made without empiric foundation. In 50 older participants, the authors examined cortisol assay comparability across laboratories, assay reliability, test–retest reliability of slopes, and comparability of slope estimates for two, three, and four samples per day. The authors demonstrate in an older sample that 1) assay reliability is a relatively minor issue, that one assay per saliva sample suffices; 2) the use of a sample obtained at wake time for each participant appears to be a preferred anchor for the slope estimate in comparison to a sample 30 minutes postwake time; 3) self-reported times appear preferable to automatic time recording; and 4) test–retest reliability of slopes, however, is not sufficiently high to base a slope estimate on one day; minimally two days and preferably three should be required. Whether these conclusions apply to other populations, or using other protocols, is not assured, but the study itself provides a model that can be used to check research decisions. Unnecessarily imposing a burdensome protocol has both ethical and scientific ramifications and should be carefully avoided.