Utilizing a digital cohort to understand the health burden and lifestyle characteristics across the life course in individuals with polycystic ovary syndrome (PCOS) and possible PCOS

Elizabeth Peebles^1*Zifan Wang¹Erin Dracup¹Carrie Sarcione¹Christine L Curry²Kayley Abrams²Jukka-Pekka Onnela³Michelle A Williams⁴Russ Hauser^1,5Brent Coull³Shruthi Mahalingaiah^1,5

• ¹Department of Environmental Health, School of Public Health, Harvard University, Boston, United States
• ²Health, Apple (United States), Cupertino, California, United States
• ³Department of Biostatistics, School of Public Health, Harvard University, Boston, Massachusetts, United States
• ⁴Epidemiology and Population Health, Stanford University, Stanford, California, United States
• ⁵Department of Epidemiology, School of Public Health, Harvard University, Boston, Massachusetts, United States

Introduction Polycystic ovary syndrome (PCOS) is an ovulation disorder associated with multiple health conditions. This study analyzed health and lifestyle characteristics of those with diagnosed and possible PCOS in a large, digital cohort. Methods We analyzed data from female participants who enrolled in the Apple Women’s Health Study and provided informed consent—a mobile-application-based cohort in the United States—from 11/14/2019–12/14/2024. Specific analyses were further restricted to those who responded to relevant survey questions. Self-reported sociodemographic, health (conditions and age at diagnosis), and lifestyle characteristics were evaluated, stratified by PCOS status: PCOS (self-reported physician diagnosed PCOS), possible PCOS (self-reported irregular menses and androgen excess), and no PCOS. Among those with PCOS/possible PCOS, we further evaluated potential predictors of not reporting a PCOS diagnosis using multivariable logistic regression. Results Of participants providing medical history at enrollment, 12.6% (n=11,022) reported PCOS, and among the subset without a PCOS diagnosis and with relevant survey data, 17.4% (n=7,152) were assigned possible PCOS. The median baseline age was 35 years. Most participants self-identified as non-Hispanic White (74.2%). The possible PCOS group was slightly less educated (≤high school: possible PCOS 17.3%, PCOS 14.5%, no PCOS 14.0%). The PCOS/possible PCOS groups reported lower socioeconomic status (SES) than the non-PCOS group (low SES: PCOS 32.7%, possible PCOS 31.6%, no PCOS 23.5%). The PCOS and possible PCOS groups displayed a high burden of disease (cardiometabolic, endometrial hyperplasia/cancer, pregnancy complications, mental health conditions). Compared to those without PCOS, those with PCOS reported less healthy lifestyle behaviors relevant to physical activity/sleep/stress/smoking and more healthy lifestyle behaviors relevant to alcohol intake/diet. The age at diagnosis for multiple health conditions was earlier for participants with PCOS compared to those without PCOS. Young/old age (18-29/40-50 years), lower educational attainment, lower SES, and lower BMI were positive predictors of not reporting a PCOS diagnosis. Conclusions This study demonstrated significant differences in health and lifestyle characteristics across PCOS status (PCOS, possible PCOS, no PCOS), identifying populations that could benefit from early risk reduction counseling. Our results may inform discussions around clinical care models through improving awareness of health predictors and lifestyle interventions.