Assessment of salivary cortisol dynamics in an infantry training exercise: a pilot study

Introduction Measuring cortisol during military training offers insights into physiological responses to stress. We attempted precisely timed, cortisol awakening response (CAR) and pre-sleep cortisol (PSC), and diurnal slope (peak morning minus evening cortisol), during a British Army exercise. We aimed to understand cortisol dynamics and evaluate the feasibility of CAR and PSC in this environment. Method Setting: high-intensity, 10-day infantry exercise. Participants: regular infantry soldiers exercising (EX, n=25) or headquarters-based (HQ, n=6). Participants undertook PSC and WAKE and WAKE+30 min samples after 1-2 days, 5-6 days and 9-10 days. Wrist-worn GENEActiv accelerometers were used to assess sleep duration in EX only. Samples taken +/- 15 min from prespecified time points were deemed adherent. Validated questionnaires were used to measure resilience and perceived stress. Cortisol and cortisone were measured simultaneously by liquid chromatography tandem mass spectrometry. Results From adherent participants' samples, CAR was positive and tended to decrease as the exercise progressed. From all available data, HQ demonstrated greater diurnal slope than EX (F=7.68, p=0.02), reflecting higher morning cortisol (F=4.72, p=0.038) and lower PSC (p=0.04). No differences were seen in cortisol:cortisone ratio. 26.1% of CAR samples were adherent, with moderately strong associations between adherence and stress (r=0.41, p=0.009) but no association between adherence and day of exercise (chi(2)=0.27, p=0.8), sleep duration (r=-0.112, p=0.43) or resilience (r=-0.79, p=0.75). Test-retest reliability ratings for CAR were Cronbach's alpha of 0.48, -11.7 and 0.34 for the beginning, middle and end of the exercise, respectively. Conclusions We observed a reduction in morning cortisol and decreased diurnal slope during a high-intensity military exercise, compared with the HQ comparator cohort in whom diurnal slope was preserved. A carefully timed CAR was not feasible in this setting.