Joint Modeling Of Hierarchical Data With Application To Prospective Pregnancy Studies

Backward recurrence time;Bayesian hierarchical model;Fecundity;Intercourse timing;Length-bias;Menstrual cycle length;Mixed effects accelerated failure time model;Perfluoroalkyl surfactants;Pregnancy probability.;Biostatistics

Lum, Kirsten J.Lees, Peter S. ;

Johns Hopkins University

关键词: Backward recurrence time;    Bayesian hierarchical model;    Fecundity;    Intercourse timing;    Length-bias;    Menstrual cycle length;    Mixed effects accelerated failure time model;    Perfluoroalkyl surfactants;    Pregnancy probability.;    Biostatistics;   

Others  :  https://jscholarship.library.jhu.edu/bitstream/handle/1774.2/39335/Thesis.tex?sequence=11&isAllowed=y

瑞士|英语

来源: JOHNS HOPKINS DSpace Repository

【 摘 要 】

Although fecundity has been studied for decades, the heterogeneity among couples, both biologically and behaviorally, is not well understood. The length of the woman;;s menstrual cycle has been shown to play an important role; however, a complete assessment of this role requires a model that accounts for both male and female risk factors and the couple;;s intercourse pattern.We develop and implement a Bayesian joint modeling approach to estimate the woman;;s underlying distribution of cycle length and assess its relation with couple fecundity while accounting for risk factors of both partners and intercourse frequency and timing relative to ovulation. We apply our approach to prospective pregnancy studies in which couples may enroll when they learn of the study as opposed to waiting for the start of a new menstrual cycle.Due to length-bias, the enrollment cycle will be stochastically larger than the general run of cycles, a typical property of prevalent cohort studies.We develop and evaluate an approach for unbiased estimation of the cycle length distribution for the study population that accounts for length-bias and selection effects, where the probability of enrollment may depend on the time since the last menstrual period.We find that shorter and longer cycle lengths are negatively associated with fecundity even with adjustment for semen quality, age, smoking, and intercourse pattern.This finding motivates investigation of environmental chemicals, in particular perfluoroalkyl surfactants (PFASs), and their potential role in cycle length and fecundity.We extend the joint model to include exposure to PFASs and find that 2-(N-methyl-perfluorooctane sulfonamido) acetate (MeFOSAA) and perfluorooctanoate (PFOA) are associated with shorter cycles while perfluorodecanoate (PFDA) is associated with longer cycles. Further, we find perfluorononanoate (PFNA) and perfluorooctane sulfonamide (PFOSA) are adversely associated with fecundity.