Abstract
BACKGROUND: A need exists to incorporate evidence-based tracking methods that measure menstrual cycle (MC) variability to describe the data quality provided by an app. The study purpose was to assess the agreement between an app's cycle phase identifications and a modified version of the three-step method (m3stepMC) of hormone verification. MATERIALS AND METHODS: Participants across Canada were recruited to track their MC over 3 months by entering data into a female-health MC tracking app (the app) while collecting measures of ovulation and salivary hormones around the late-follicular (FP) and mid-luteal (MLP) phases, respectively. Bland-Altman plots assessed the limits of agreement (LoA) between the identified days within each of the app's predetermined phases and the m3stepMC-identified days when MC dates aligned between the methods. Pearson's correlations (r) were used to examine the effect size of relationships between variables. RESULTS: Participants' (n = 25) mean age was 29.3 ± 4.24 with self-reported mean cycle lengths of 27.3 ± 2.38 days. The agreement between the app's estimated (1) end of phase one and the estimated start of the mid-FP was 0.6 ± 1.66 days (95% LoA: 2.65-3.85; r = 0.66), (2) end of phase two and the identified luteinizing hormone (LH) surge day and midpoint of phase three and the estimated 48-hour ovulatory window post-LH surge day were -0.6 ± 1.71 days (95% LoA: -3.95 to 2.75; r = 0.64), and (3) phase four and the estimated MLP day verified by salivary hormones and the start of the app's phase five and the estimated late-luteal midpoint day were -2.2 ± 0.97 (95% LoA: -4.13 to -0.32; r = 0.94). CONCLUSION: This study describes the agreement between a m3stepMC tracking method and hormone measures and an app's predetermined MC phase system in eumenorrheic cycles when MC dates aligned between methods.