Modeling neuroendocrine stress reactivity in salivary cortisol: adjusting for peak latency variability

Figures & data

Figure 1. (A) Presents two responses to the TSST reflecting significant phase variability (timing) but limited amplitude variability. The observed differences at 15 min and 35 min are likely due primarily to differences in the timing of the response. (B) Shows the same data after adjusting the curves to control for timing differences. Analysis of Figure 1(B) data would correctly show that the groups do not differ in reactivity or regulation once timing differences are accounted for.

Table 1. Example of database setup in wide format prior to long-transformation.

Participant	Cort_0	Cort_10	Cort_15	Cort_25	Cort_35	Cort_45	Cort_60	Cort_75	PkTime	Sex
001	0.0209	0.001	0.1126	0.1759	0.2196	0.1186	0.0859	0.0399	35	F
002	0.1241	0.0855	0.3011	0.3799	0.2725	0.2187	0.1597	0.0933	25	M
003	0.0845	0.1245	0.1318	0.1052	0.0832	0.0739	0.702	0.0645	15	M

Identified peak is in bold.

Table 2. Example of database setup after long-format transformation and creation of adjusted time variables.

Participant	Cortisol	Time	PkTime	Time-to-Pk	TimeBeforePeak	TimeAfterPeak	Sex	TimeBeforeGroupPeak	TimeAfterGroupPeak
001	0.0209	0	35	−35	−35	0	F	−35	0
001	0.0010	10	35	−25	−25	0	F	−25	0
001	0.1126	15	35	−20	−20	0	F	−20	0
001	0.1759	25	35	−10	−10	0	F	−10	0
001	0.2196	35	35	0	0	0	F	0	0
001	0.1186	45	35	10	0	10	F	0	10
001	0.0859	60	35	25	0	25	F	0	25
001	0.0399	75	35	40	0	40	F	0	40
002	0.1241	0	25	−25	−25	0	M	−35	0
002	0.0855	10	25	−15	−15	0	M	−25	0
002	0.3011	15	25	−10	−10	0	M	−20	0
002	0.3799	25	25	0	0	0	M	−10	0
002	0.2725	35	25	10	0	10	M	0	0
002	0.2187	45	25	20	0	20	M	0	10
002	0.1597	60	25	35	0	35	M	0	25
002	0.0933	75	25	50	0	50	M	0	40
003	0.0845	0	15	−15	−15	0	M	−35	0
003	0.1245	10	15	−05	−05	0	M	−25	0
003	0.1318	15	15	0	0	0	M	−20	0
003	0.1052	25	15	10	0	10	M	−10	0
003	0.0832	35	15	20	0	20	M	0	0
003	0.0739	45	15	30	0	30	M	0	10
003	0.702	60	15	45	0	45	M	0	25
003	0.0645	75	15	60	0	60	M	0	40

Figure 2. ACTH changes in response to TSST for males and females modeled via repeated measures (A) and growth curve modeling (B). Repeated measures results show means and standard errors. Growth curve model shows estimated cubic curves. *Significant comparisons after Bonferroni adjustment.

Figure 3. Saliva cortisol changes in response to TSST for males and females modeled via repeated measures (A) and traditional growth curve modeling (B). Repeated measures results show means and standard errors. Growth curve model shows estimated quadratic curves.

Figure 4. Estimated saliva cortisol slopes before and after post-TSST peak for males and females modeled via growth curve modeling with landmark registered data (common peak). *Significant sex effect on peak and regulation slope, but not on activation slope.