METHODS
We conducted a 10-year retrospective chart review of all Black American, Haitian, African, and White American women seeking infertility care at Boston Medical Center (BMC) between January 1, 2005 and July 15, 2015. Patients with infertility ICD-9 diagnoses seen by a reproductive endocrinologist were included in the cohort; these patients were identified by analyzing the BMC Clinical Data Warehouse database (Appendix S1). The study was approved by the Boston University School of Medicine Institutional Review Board (IRB# H-34265). No funding was received for this study.
Charts were reviewed to determine infertility diagnosis with information obtained from physician notes, clinical history, and fertility testing. Data was first stratified by place of birth, and then subdivided by self-identified race (White or Black) and among Black women ethnicity (defined as Haitian, African, or Black American) as determined by place of birth and primary language. Women were included if they had a confirmed infertility diagnosis, identified as either Black or White, and were born in either Haiti, Africa, or the United States (US). Women were excluded if race and place of birth were unavailable, they identified with an ethnicity different from those of interest regardless of place of birth, or the infertility diagnosis could not be corroborated from the medical record. White American women were used as a comparison group. Demographic and infertility testing results including day 3 follicle-stimulating hormone (FSH) levels were compared between groups. Infertility diagnoses were compared between White and Black women. Subgroup analyses were then performed comparing White women to Black Haitian, African, and American women seeking infertility treatment.
Statistical analyses using unpaired t-test or one-way ANOVA were used for analysis of continuous variables. For ANOVA, a significant omnibus F-test was followed by Fisher’s PLSD post hoc comparisons. Discrete data were analyzed by chi squared tests followed by comparison of cell chi-squared contributions. Multivariate multinomial logistic regression was then used to evaluate associations of independent variables with the dichotomous outcome variable, infertility diagnosis. Univariate and multivariable regression models were used to identify pertinent risk factors. Medical insurance type was used as a proxy for socioeconomic status, with uninsured status and Medicaid insurance as an indicator of low socioeconomic status (SES). SAS (version 9.3) and StatView (version 5.0.1) statistical software were used to perform the analyses. Statistical significance was defined as a p-value < 0.05.