Changes of serum cortisol during pregnancy and labor initiation: an onsite cross-sectional study

Abstract

Background: Increased maternal cortisol secretion has been observed during pregnancy and labor. However, due to the limitations in diagnostic methods, the dynamic change of cortisol during the short period between threatened labor and labor is unknown. In this study, we aim to evaluate the changes in serum cortisol during late pregnancy and full-term labor initiation, verifying if cortisol could serve as a biomarker for the diagnosis of labor initiation from threatened labor.

Methods: This cross-sectional onsite study involved 564 participants of 6 different gestational stages (C: Control; T1: Trimester 1; T3: Trimester 3; E: expectant; TL: threatened labor; L: labor), all patients in the E, TL, and L groups were at full term. The serum cortisol concentration was quantified with a point-of-care test (POCT), and the gestation, age, parity, and BMI of participants were documented. Morning serum cortisol was collected between 8:00 and 10:00 a.m., except for the TL and L group women who were tested upon arrival or during latent labor. With cortisol levels or all five variables, L was distinguished from TL using machine learning algorithms.

Results: Significant elevation of cortisol concentration was observed between T1 and T3, or TL and L group (P< 0.001). Women belonging to the E and TL group showed similar gestation week and cortisol levels. Diagnosis of labor initiation using cortisol levels (cutoff = 21.46 μg/dL) yielded sensitivity, specificity, and AUC of 86.50%, 88.60%, and 0.934. With additional variables, a higher specificity (89.29%) was achieved. The diagnostic accuracy of all methods ranged from 85.93% to 87.90%.

Conclusion: Serum cortisol could serve as a potential biomarker for diagnosis of L form TL. The rapid onsite detection of serum cortisol with POCT could facilitate medical decision-making for admission and special treatments, either as an additional parameter or when other technical platforms are not available.

Keywords: diagnosis of labor; labor initiation; point-of-care testing; serum cortisol; threatened labor.

Figure 1

Difference of variables between groups at each significant level (A) and cortisol level on different gestation weeks (B). ***: p<0.001; C, Control; T1, Trimester 1; T3, Trimester 3; E, expectant; TL, threatened labor; L, labor; BMI, body mass index before pregnancy. ns, not significant.

Figure 2

ROC curves for the diagnosis of labor initiation with the two datasets based on serum cortisol levels. E, expectant; TL, threatened labor; L, labor; AUC, Area Under Curve.

Figure 3

Decision curves of net benefit constructed with dataset L and TL (A) and the alternative expression of decision curves as interventions avoided per 100 patients (B). Decision curves of net benefit constructed with dataset L and E+TL (C) and the alternative expression of decision curves as interventions avoided per 100 patients (D).

Figure 4

Calibration curves for the performance evaluation of four models using dataset L and TL (A), and dataset L and TL+E (B). Cor, cortisol; LR, Logistic Regression; RF, Random Forest; SVM, Support Vector Machine.

Figure 5

Workflow proposed for labor/threatened labor diagnosis with the inclusion of serum cortisol level.