Table 5.

Multiple logistic regression analysis for the androgen prediction of PCOS including age, body mass index, salivary testosterone and serum testosterone.

Parameters (units)	Adjusted
	Odds Ratio	95% CI	P-value
Age (years)	0.951	0.888–1.017	0.144
Body mass index (kg/m2)	1.201	1.111–1.299	<0.001
salT (nmol/L)	1.075	1.018–1.135	0.010
Serum Testosterone (nmol/L)	1.568	0.701–3.507	0.274

Logistic regression taking into account age and body mass index (BMI) showing that in this model salivary testosterone (salT) is significantly more predictive of PCOS than the free androgen index (FAI) or serum T (Tables 5 and 6, respectively), and that FAI was more predictive than serum T (C).