Table 3.

Nonlinearity addressing of progesterone.

Exposure	Progesterone (OR, 95% CI, P value)
Fitting model using standard binary logistic regression model	0.99 (0.98, 0.99) <0.01
Fitting model using two-piecewise regression model
Inflection point	90.62
<inflection point	0.97 (0.95, 0.98) <0.01
≥inflection point	1.01 (0.99, 1.02) 0.27
P for log-likelihood ratio test	<0.01

We used basal progesterone and adjusted for maternal age, BMI, smoking, drinking, gestational week, marital status, basal β-HCG (log10), gestity, parity, abortion, embryo visualization, cardiac movement visualization, medical history, and use of progesterone.