O-105 High prevalence of obesity and polycystic ovary syndrome in an online community seeking assistance with fertility

Abstract

Abstract Study question What are the characteristics of a large online community of women trying to conceive and what are the factors that predict infertility in this group? Summary answer The cohort was not characterized by increased age, though age, obesity and PCOS strongly correlated with fertility. PCOS may be under-recognized within some ethnic groups. What is known already Obesity, polycystic ovary syndrome (PCOS), and age are among the primary predictors of infertility in women. However, most studies assessing reproductive health employ populations who have sought medical assistance for infertility or associated disorders, potentially leading to populations more likely to exhibit disorders such as PCOS or to be of increased age. Characterization of factors affecting fertility in a general population may highlight epidemiological influences that should be better addressed to aid women trying to conceive. Study design, size, duration This study employed users of the mobile application, FertilityAnswers, for people searching for answers to fertility problems. Users answered a short survey describing themselves, their fertility history and goals. Recruitment for this study began in March 2017 and ended in January 2021. 61814 participants downloaded the application during this period, and 56878 at least provided their age. The primary inclusion criteria were that study participants be US females over eighteen years of age. Participants/materials, setting, methods Regression models estimated beta coefficients and corresponding confidence intervals. Multivariable models determined independence of variables. To model PCOS, the available data in January 2020 was split randomly into discovery and validation subsets. Missing value imputation using random forests was performed in the discovery dataset. Minimal feature selection used a linear regression model penalized with a lasso and elastic net. The model was then validated on samples collected after the model was trained and tested. Main results and the role of chance Age was a significant predictor of fertility in this study (p < 1x10-10). However, the distribution of age in the cohort was very similar to that of women at first birth in the United States, therefore we did not observe the majority of study participants to be of an age where in a typical clinical setting age-related concerns would be addressed (i.e., approaching 35 years). Using National Center for Health Statistics data, the mean age of 1,433,604 women at first birth in 2018 in the US was younger than the study population (restricted to those without children) by only one month. Obesity was of increased prevalence in this cohort, with 55% being obese, compared to 37% in an age-matched US population. Participants reported a variety of fertility-related disorders, with polycystic ovary syndrome (PCOS) being most prevalent (19.0%), followed by endometriosis (6.0%). Prediction of PCOS, performed by modeling on training and test sets (10476 and 5312 participants) and then validating with an additional 21097 participants collected after the model creation, found that African-Americans and Latino members of this cohort had a lower self-reported rate of the disorder than was anticipated by the model, in contrast to those of Asian or European descent. Limitations, reasons for caution All health data was self-reported. Additionally, as this is the initial survey of this population, no a priori hypotheses were made as to the expected relationships to be observed. Instead, all associations were examined, and measurements of false discovery rate were estimated. Wider implications of the findings We found that women were seeking answers about infertility at ages coincident with that of their peers achieving first pregnancies. ART is often not a first-line treatment for women of this age, but there may be a disconnect between traditional clinical response to this group and their desire for assistance. Trial registration number not applicable