Serial progesterone levels more accurately predict the time of ovulation in subfertile women: a prospective cohort study

Aydın T., Koroglu N., Albayrak N., İnsel M. A.

Journal of Assisted Reproduction and Genetics, vol.40, no.8, pp.1897-1903, 2023 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 40 Issue: 8
  • Publication Date: 2023
  • Doi Number: 10.1007/s10815-023-02864-2
  • Journal Name: Journal of Assisted Reproduction and Genetics
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, ATLA Religion Database, BIOSIS, Biotechnology Research Abstracts, CAB Abstracts, EMBASE, MEDLINE, Veterinary Science Database
  • Page Numbers: pp.1897-1903
  • Keywords: Estradiol, Luteinizing hormone, Ovulation time, Progesterone, Subfertile women
  • Yıldız Technical University Affiliated: Yes


Purpose: To predict ovulation in subfertile women using serial follicular growth (FG) and serum hormone measures (estradiol (E2), luteinizing hormone (LH), and progesterone (P) levels) in mathematical models. Methods: This was a prospective observational study of 116 subfertile women aged between 18 and 40 years. FG was assessed by serial transvaginal ultrasonography starting from cycle days 8–12, depending on cycle length. Once the dominant follicle reached 15–16 mm, hormone levels were assessed daily. The primary outcome measure was ovulation (Ov), with a serum LH level ≥15 IU/l defining the start of the LH surge (the day prior to ovulation) and a serum P level >1 μg/ml concurrent with a drop in serum E2 levels indicating Ov. To determine Ov, mathematical models were generated using FG, LH, E2, and P measurements. Results: A mathematical model was constructed using exponential regression to relate days until and after ovulation with P levels. The Ov(P) model was found to be superior to the Ov(LH) model in the prediction of Ov, with high R2 and low RMSE values of 0.9983 and 0.2454, respectively. In the range of [−2, 2] days, the net accuracy of the Ov(P) model was 63.0%, while with an allowed one-day error, the accuracy was 99.6%. Conclusion: Serum P levels display a highly predictable linear curve in natural cycles, which enables the prediction of ovulation. The Ov(P) model can be independently used to schedule embryo transfer in natural frozen-thaw cycles and could therefore replace the Ov(LH) model in clinical practice.