Defining thresholds for abnormal premature progesterone levels during ovarian stimulation for assisted reproduction technologies.

OBJECTIVE To evaluate methodologies to establish abnormal progesterone (P) levels on the day of trigger for recommending freeze only cycles. DESIGN Threshold analysis and cost analysis. SETTING Private ART practice. PATIENT(S) Fresh autologous ART. INTERVENTIONS(S) None. MAIN OUTCOME MEASURE(S) Live birth. RESULT(S) Fourteen established statistical methodologies for generating clinical thresholds were evaluated. These methods were applied to 7,608 fresh ART transfer cycles to generate various P thresholds which ranged widely from 0.4 to 3.0 ng/mL. Lower thresholds ranged from 0.4 to 1 ng/mL and classified the majority of cycles as abnormal as well as required very large number needed to treat (NNT) to increase one live birth. Frozen embryo transfer was cost-effective when P was ≥1.5 ng/mL, with 12% of the population having an abnormal test result and an NNT of 13. Statistical and cost-effective thresholds clustered between 1.5 and 2.0 ng/mL. CONCLUSION(S) Statistically significant thresholds for P were demonstrated as low as 0.4 ng/mL but resulted in a very large NNT to increase one live birth. A clinical benefit to a freeze-only approach was demonstrated above P thresholds ranging from 1.5 to 2.0 ng/dL. At these thresholds, elevated P has a demonstrable and clinically significant negative effect and captures a smaller percentage of the patient population at higher risk for fresh transfer failure, thus making freeze-only a cost-effective option.