Standardized protocols for characterizing women's fertility: A data-driven approach

Experts are divided on whether women's cognition and behavior differs between fertile and non-fertile phases of the menstrual cycle. One of the biggest criticisms of this literature concerns the use of indirect, imprecise, and flexible methodologies between studies to characterize women's fertility. To resolve this problem, we provide a data-driven method of best practices for characterizing women's fertile phase. We compared the accuracy of self-reported methods and counting procedures (i.e., the forward- and backward-counting methods) in estimating ovulation using data from 140 women whose fertility was verified with luteinizing hormone tests. Results revealed that no counting method was associated with ovulation with >30% accuracy. A minimum of 39.5% of the days in the six-day fertile window predicted by the counting methods were non-fertile, and correlations between counting method conception probabilities and actual conception probability were weak to moderate, rs=0.11-0.30. Poor results persisted when using a lenient window for predicting ovulation, across alternative estimators of the onset of the next cycle, and when removing outliers to increase the homogeneity of the sample. By contrast, combining counting methods with a relatively inexpensive test of luteinizing hormone predicted fertility with accuracy >95%, but only when specific guidelines were followed. To this end, herein we provide a cost-effective, pragmatic, and standardized protocol that will allow researchers to test whether fertility effects exist or not.