Objective: To investigate, using logistic regression analysis and receiver operator characteristic analysis, the biochemical diagnosis of polycystic ovary syndrome (PCOS) and if it could be improved by using an array of hormone measurements chosen to yield optimal and cost-effective discrimination between women with PCOS and healthy women.
Design: Retrospective clinical study.
Setting: Outpatient clinic of reproductive endocrinology at Turku University Central Hospital, Turku, Finland.
Patients: Fifty-four oligomenorrheic women with PCOS diagnosed by ovarian ultrasonography, and 29 healthy regularly menstruating women with normal ovarian morphology.
Main outcome measures: Concentrations of LH, FSH, androstenedione (A), T, and sex hormone-binding globulin in serum.
Results: Luteinizing hormone, FSH and A, when used in combination, were the hormonal analytes of highest clinical utility. Diagnostic sensitivity, specificity, and overall concordance of 98%, 93%, and 96%, respectively, were attained. Each of the analytes used alone yielded lower degree of discrimination.
Conclusions: Simultaneous use of the levels of LH, FSH, and A in serum can be used effectively for classification between women with PCOS and healthy women.