This is a cross-sectional analysis of all OB-GYN residency program websites in the United States. We systematically examined each residency program website from November 1st to November 22nd, 2021, searching for specific LGBTQI keywords. We excluded programs with no existing website from the analysis.
We identified all OB-GYN residency programs in the United States using the American Medical Association’s Residency & Fellowship Programs Database (FREIDA). Two researchers then examined each program website separately to search for LGBTQI content. A third reviewer resolved any conflicts or discrepancies between reviewers. We searched the website home page and all additional sub-pages using the following LGBTQI-specific keywords: LGBT/LGTBQ/LGBTQI, transgender, trans health, queer, lesbian/gay/bisexual. We selected similar keywords to those used in existing research assessing LGBTQI healthcare content online .
Using our keyword searches, we determined whether programs websites had any mention of LGBTQI health in any portion of the website; mention of LGBTQI-specific rotations or didactics, mention of LGBTQI health in the program’s diversity, equity, and inclusion (DEI) statement, or other mentions of LGBTQI health. We also collected quotes from the websites that reflected this content.
We collected various program characteristics such as location (city, state, region) and program type (community, university based, etc.) from FRIEDA . We also looked at the extent of program websites, and defined them as simple, moderate, or complex. We defined a simple website as having no subpages, a moderate website as having less than or equal to 5 subpages, and a complex website as having greater than 5 subpages. We used the program website to determine number of residents, number of fellowships available, sex of department chair, and sex of program director. Of note, we based the sex of the chair and program director on external characteristics and/or the individual’s name as we were unable to collect data on the individuals’ gender identities. To determine state political party, we looked at the last five presidential elections and assigned party based on election results. We labeled states in which no one party won more than three of the last five presidential elections as swing states.
We created a dichotomous variable for any mention of LGBTQI health (whether didactics, rotations, or mention in DEI statement, or “other” mention of LGBTQI content) versus none (primary outcome). An example of the “other” category is mention of LGBTQI research. We also created a dichotomous secondary outcome that only included mention of LGBTQI health didactics or rotations, to ensure that we had specific information about training.
We conducted descriptive statistical analyses of variables reflecting program characteristics. To compare the characteristics of programs that did and did not include LGBTQI content, we initially used the Chi Square or Fisher’s Exact tests. To examine the predictors of having any LGBTQI content on the program website (primary outcome), and of having LGBTQI training (rotation or didactics, secondary outcome), we constructed two multivariable logistic regression models. To do so, we initially included all variables with p < 0.1 on univariable analyses. We then build multivariable regression models using a stepwise, backward selection approach. We first included all covariates with p < 0.25 on bivariate analysis and then sequentially removed covariates with the highest p value until all variables were significant.
We conducted all analyses using STATA Version 16.0 and set the significance level at 5%.
This research study does not involve human subjects. As such, the New York Medical College Institutional Review Board deemed the study exempt from review.