Assessing the accuracy, reliability, and validity of menstrual cycle phase tracking equations in an applied sport setting

Purpose
To assess the accuracy, reliability, and validity of five mathematical equations across multiple cycles to predict menstrual cycle onset and date of ovulation in the applied environment.

Methods
11 professional female athletes who were either naturally menstruating or intrauterine device (IUD) users tracked their menstrual cycles throughout the course of an entire national women’s soccer league (NWSL) season. Participants recorded the first and last dates of each menstrual cycle, which were used in mathematical equations to predict next bleed and ovulation date, which were validated by retrospective analysis and confirmed by ovulation tests from two consecutive cycles.

Results
When predicting next bleed date, there was no difference in accuracy (MAE: 6 days), reliability (ICC3 = 0.60, 95% CI [0.53–0.68], p < 0.001) (CV: 24.3 to 25.1%), or prediction error (β = −0.20, SE = 0.16, t(335) = − 1.22, 95% CI [− 0.65 − 0.24]) between the equations, yet there was a significant, positive association between cycle length and prediction error (β = 1.10, SE = 0.03, t(335) = 31.54, 95% CI [0.94–1.13]). When predicting ovulation date, there was no significant difference in accuracy (MAE: 4), or reliability (ICC3 = 0.45, 95% CI [0.18–0.75], p < 0.001) (8.5–22.1%).

Conclusion
No equation emerged as significantly better when predicting next bleed or next ovulation date, and all equations had relatively high variability and error rate. It is imperative that practitioners monitor menstrual characteristics and consider cycle variability when utilizing equations to predict menstrual cycle bleed and ovulation date.