Assessing mid-trimester salivary cortisol levels across three consecutive days in pregnant women using an at-home collection protocol


Dr Claudia B. Holzman, Michigan State University, B601 West Fee, East Lansing, MI 48824, USA. E-mail:


The Pregnancy Outcomes and Community Health (POUCH) Study enrolled women in their 15th to 27th week of pregnancy from 52 prenatal clinics located in five communities. After the study began, an at-home protocol to measure maternal stress was added, which included collection of urine and saliva twice a day (waking = AM, bedtime = PM) for three consecutive days, and completion of a daily diary. The at-home protocol was in place for 2852 POUCH participants at enrolment, and 81.3% (n = 2318) consented and returned samples and diary. For these analyses, salivary cortisol was analysed in a subset of 846 women who delivered at term.

Day-to-day correlation coefficients for AM cortisol (0.45–0.55) exceeded those for PM cortisol (0.31–0.43). Study diaries indicated that there was variation in the time interval between waking and sample collection. Analyses of discrete intervals showed the same awakening response pattern in cortisol levels that has been reported in studies with serial AM sampling. The adjusted mean AM cortisol (µg/dL) was 0.506 at 0–15 min post-waking, 0.544 at 16–30 min (P < 0.05), 0.582 at 31–60 min (P < 0.01), and 0.515 at >60 min post-waking. In addition, adjusted mean AM cortisol in samples collected at or before 9 a.m. was higher than that in samples collected after 9 a.m. (0.564 vs. 0.510 µg/dL, P < 0.01). Among working women, adjusted mean AM cortisol was higher on work days than non-work days (0.564 vs. 0.489 µg/dL, P < 0.01), and in multiparae compared with primiparae (0.551 vs. 0.502 µg/dL, P = 0.07). The parity effect was not evident in non-working women. The adjusted mean PM cortisol significantly increased as week of pregnancy at sampling increased, but was not significantly related to time of collection, work day and parity.

Factors influencing AM and PM cortisol levels in pregnancy appear to differ. In studies of AM cortisol levels, it is important to gather data on time of sample collection, interval from waking to sample collection, parity and work.