NEW EVIDENCE ON CONCEPTION AND PREGNANCY LOSS IN THE U.S. ABSTRACT Conception failure and pregnancy loss impede the pathway to parenthood for millions of US couples each year. Despite being modifiable and common—pregnancy loss occurs fifty times as often as infant mortality— miscarriage is among the least well-studied aspects of population health or population dynamics. We know little about what causes these losses, the generalizability of estimates generated from the study of small recruited samples, including how the risk of miscarriage differs across time, place, or subpopulation. These enormous gaps are driven by how difficult it is to study the pathway to live birth in humans, requiring a sufficiently large and diverse “preconception cohort.” To date small, nonrepresentative cohorts contribute most of what we know about early pregnancy. But limited sample size, variation, and coverage preclude the study of social and ecological factors—poverty, partner health, environmental exposures—that are now widely understood to shape mortality at later ages: e.g., stillbirth, infant, and child mortality. Such omissions are consequential. Conception failures and pregnancy loss have significant emotional and economic costs. Though pregnancy failures are likely distributed unevenly across populations, the size of this disparate burden is unknown. We do not have evidence about the potential impact of policies on the prevention of these outcomes. Our research combines the tools of data science and social science to scale the preconception design to a large, diverse cohort in the US. Period and pregnancy “tracking” on digital devices is now common among reproductive-age women. We use data on 4 million women’s day-to-day recording of menstrual cycles, ovulation, and pregnancy, as well as reported social and economic characteristics. The data allow us to measure both reported and detected conception and pregnancy loss. We combine the data with georeferenced information about the physical, social, economic, and policy environments in which women and their partners live. The sample is diverse: users come from over 99% of U.S. counties; over 1 million users are on Medicaid. We use these extraordinary data to provide the first estimates of large-scale population variability in the pathway to live birth and to document factors that affect this pathway. Aim 1 develops new estimates of conception and pregnancy loss, including disparities associated with education, income, racial identification, and neighborhood poverty. Aim 2 estimates the contribution of male partner characteristics to conception and pregnancy loss, drawing on detailed data reported about partners. Aim 3 uses quasi-experimental designs to provide the first estimates of understudied exposures from multiple environmental, social, economic domains on conception and pregnancy loss. The research sheds new light on a pervasive and massively understudied public health concern, with direct implications for the improvement of early pregnancy outcomes. In so doing, the research also generates estimates that are much-needed inputs to support advancement in multiple, connected fields of study in the social, health, and biomedical sciences.

NARRATIVE Conception failure and miscarriage affect millions of U.S. couples each year. This research uses new georeferenced data on pre-pregnancy and early pregnancy for 4 million U.S. women to provide the first evidence on individual, family, ecological, and policy factors that reduce the risk of early pregnancy failure.