Infertility is more than a quality of life issue and is recognized as a disease of the reproductive system by the World Health Organization along with the American Society for Reproductive Medicine, and is is a disability under the Americans with Disabilities Act. The right to procreate is considered a fundamental human right, yet large racial, ethnic, and geographic disparities persist in risk factors for infertility and for individuals who seek access to diagnostic and treatment services. Public health programs rely on accurate estimates to not only target services to vulnerable subpopulations, but also to identify strategies for prevention and for the provision of quality care and services for infertility. Thus, methodologies that enable researchers and public health programs to more accurately estimate the magnitude of the problem and differentiate risk factors related to prevention versus access to care are needed. The current duration approach is a novel statistical methodology for estimating the distribution of time-to-pregnancy (TTP), which can be used to assess fecundity (the biologic capacity for reproduction) and infertility (a TTP of 12+ months). It uses cross-sectional data, which has advantages to other data sources including the ability to be nationally representative, and quality public data sources like the National Survey of Family Growth (NSFG) are available. These advantages are tempered by the inability of current biostatistical methods to appropriately handle women that had undergone infertility treatment before sampling. Current methods rely on ad hoc fixes, which we show can bias results and lead to inconsistent inferences, particularly with the increased availability of infertility treatments. The objective of this application is to develop statistical methods that can be used to estimate the distribution of TTP and time-to-fertility-treatment (TTFT), then apply those methods to determine trends in infertility and fertility treatment, and identify subgroups with higher rates of infertility and low access to fertility treatment. Overall, the Specific Aims of the proposal are (A1) develop semi-competing risk current duration methodology for estimation of the distribution of TTP and TTFT from cross-sectional data, including the incorporation of covariate data, (A2) obtain nationally-representative prevalence estimates of infertility over time in the United States by applying the methodology from Aim 1 to NSFG cycles from 2002, 2006-2010, and 2011-2015, and (A3) examine subgroups and risk factors associated with prolonged TTP or TTFT using the most recent NSFG cycle (2011-2015). The significance of this contribution lies in the findings on the current trends in infertility, and risk-factors associated with infertility and low access to fertility treatment. These findings can be used by clinicians and public health programs to target interventions and address barriers in access to care. We will further the field of length-biased cross-sectional survival analysis data by developing theory, statistical models and software for incorporating competing risks outcomes.
This proposal is relevant to public health due to the commonly preventable and treatable nature of infertility, the large racial, ethnic, and geographic disparities in access to fertility treatments and the limitations of current methods to produce valid and representative inference on this condition. This proposal develops a semi-competing risk current duration statistical methodology for nationally representative cross-sectional data, which correctly accounts for fertility treatment as a semi-competing risk. This proposal is relevant to the NIH's mission by ensuring that the magnitude and trends in infertility are well defined, identifying high-risk subgroups, and monitoring the availability of fertility treatment.