Louis Faust1, Dani Bradley2, Erin Landau2, Katie Noddin2, Leslie V Farland3, Alex Baron2, Adam Wolfberg4. 1. Department of Computer Science & Engineering, University of Notre Dame, Notre Dame, Indiana; Ovia Health, Boston, Massachusetts. Electronic address: lfaust@nd.edu. 2. Ovia Health, Boston, Massachusetts. 3. Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson, Arizona. 4. Ovia Health, Boston, Massachusetts; Newton Wellesley Hospital, Newton, Massachusetts.
Abstract
OBJECTIVE: To investigate the validity of self-reported fertility data generated by a mobile application-based cohort in comparison with data collected by traditional clinical methodologies. DESIGN: Data were collected from July 2013 to July 2018 through a mobile application designed to track fertility. Bayesian hierarchical models were used to assess day-specific pregnancy probabilities. Descriptive statistics were used to estimate differences in day of ovulation and lengths of menstrual phases and to assess changes in the cervix and ovulation-related symptoms drawing closer to the day of ovulation. SETTING: Not applicable. PATIENT(S): Data consisted of 225,596 menstrual cycles from 98,903 women. INTERVENTION(S): None. MAIN OUTCOME MEASURE(S): Day-specific probabilities of pregnancy, variability in lengths of the follicular and luteal phases, trends in prevalence of symptoms and cervix changes across the fertile window. RESULT(S): Analyses were consistent with established clinical knowledge. Probability of conception was highest during the 5 days before and day of ovulation, with the highest probability occurring the day before ovulation. The average cycle length was 29.6 days, and average lengths of the follicular and luteal phases were 15.8 and 13.7 days, respectively. Closer to day of ovulation, women were more likely to report changes in the cervix corresponding to fluid consistency, feel, position, and openness and symptoms associated with ovulation, including pelvic pain, tender breasts, increased sex drive, and cramps. CONCLUSION(S): Components of the menstrual cycle and fertile window, when re-evaluated with a mobile application-based cohort, were found to be consistent with established clinical knowledge, suggesting an agreement between traditional and modern data collection methodologies.
OBJECTIVE: To investigate the validity of self-reported fertility data generated by a mobile application-based cohort in comparison with data collected by traditional clinical methodologies. DESIGN: Data were collected from July 2013 to July 2018 through a mobile application designed to track fertility. Bayesian hierarchical models were used to assess day-specific pregnancy probabilities. Descriptive statistics were used to estimate differences in day of ovulation and lengths of menstrual phases and to assess changes in the cervix and ovulation-related symptoms drawing closer to the day of ovulation. SETTING: Not applicable. PATIENT(S): Data consisted of 225,596 menstrual cycles from 98,903 women. INTERVENTION(S): None. MAIN OUTCOME MEASURE(S): Day-specific probabilities of pregnancy, variability in lengths of the follicular and luteal phases, trends in prevalence of symptoms and cervix changes across the fertile window. RESULT(S): Analyses were consistent with established clinical knowledge. Probability of conception was highest during the 5 days before and day of ovulation, with the highest probability occurring the day before ovulation. The average cycle length was 29.6 days, and average lengths of the follicular and luteal phases were 15.8 and 13.7 days, respectively. Closer to day of ovulation, women were more likely to report changes in the cervix corresponding to fluid consistency, feel, position, and openness and symptoms associated with ovulation, including pelvic pain, tender breasts, increased sex drive, and cramps. CONCLUSION(S): Components of the menstrual cycle and fertile window, when re-evaluated with a mobile application-based cohort, were found to be consistent with established clinical knowledge, suggesting an agreement between traditional and modern data collection methodologies.
Authors: Joseph B Stanford; Sydney K Willis; Elizabeth E Hatch; Kenneth J Rothman; Lauren A Wise Journal: Hum Reprod Date: 2020-10-01 Impact factor: 6.918