Secondary analyses were conducted using data from the ESSENCE Study, a retrospective cohort study that evaluated the impact of neighborhood-level characteristics of the built and social environment on FS, elucidating how FS and physiological factors influence behavioral mechanisms that increase risk for HIV acquisition. Data were collected between 2015 and 2018 among Black women (n = 305) accessing services at public health clinics in Baltimore, MD, USA [16]. Interested individuals who provided informed consent had to meet the following eligibility criteria: (a) assigned female at birth, (b) be between ages 18–44 years, (c) identify as Black or African American, (d) have a negative HIV test result at the time of enrollment, (e) report sex with a man in the past 6 months, and (f) report one or more of the following risk behaviors in the past year: two or more male sexual partners; or sex with a high-risk male partner (i.e., one who has injected drugs, is HIV seropositive, has sex with men, or has been incarcerated). Participants completed a one-time, audio computer-assisted self-interview (ACASI) [17] administered survey and received $10 for completing the screener, and $25 for completing the survey.

Reproductive Coercion (RC, dependent variable; victimization and perpetration) was measured using items from the reproductive coercion scale (RCS) [18]. The 9-item measure assesses participants’ experiences with RCV across subdomains of pregnancy coercion and condom manipulation, using dichotomous (Yes/No) response options (Cronbach’s alpha = 0.89). Example items included, “Has any partner ever told you not to use contraception?” and “Has any partner ever told you he would leave you if you didn’t get pregnant?” RCP was assessed by asking 11 questions adapted from the RCS that focused on RC and abusive behaviors enacted by the participant (Cronbach’s alpha = 0.87). Example items included, “Have you ever told your partner you would leave him if he didn’t get you pregnant?” and “Have you ever told your partner you would have a baby with someone else if he didn’t get you pregnant?" A dichotomous variable was created such that if participants answered ‘yes’ to any one of the scale questions, they were coded as ‘yes’ in the analyses. Lifetime and past 12-month RCV and RCP experiences were assessed.

Forced sex (FS, primary independent variable) was defined as physical force (e.g., hit, held down, use of a weapon) or threats thereof by a male perpetrator since the age 18 years. Two questions were asked upon screening due to the study design, in which two groups of women were recruited: exposed (FS history) and unexposed. Unexposed participants reported no experiences of physical and/or sexual violence since the age of 18 years.

Psychological violence (independent variable) was measured using the Women’s Experience with Battering (WEB) scale [19]. The 10-item scale assesses perceived vulnerability to psychological danger by a current or recent physically or sexually abusive partner (Cronbach’s alpha = 0.99). Example items included, “He makes me feel like I have no control over my life, no power, no protection” and “He makes me feel unsafe even in my own home.” Likert responses were scored on a scale of 1–6, from agree strongly to disagree strongly, and ranged from 10 to 60. Participant responses were summed and a cutoff value of 20 was used to establish increased vulnerability to psychological violence [20]. Scores were then dichotomized such that a score of 20 or higher was coded as ‘yes’ for experiencing psychological violence.

Physical violence and respondent-perpetrated physical violence (independent variables) were measured using the physical assault scale and injury scale of the CTS-2 (Cronbach’s alpha = 0.99 and 0.99, respectively) [21]. First, frequency was coded into a scale variable ranging from 0 to 25, using the median value of each range. For example, “Once in the past year” was coded to 1, “Twice in the past year” was coded to 2, “3–5 times in the past year” was coded to 4, “6–10 times in the past year” was coded to 8, “11–20 times in the past year” was coded to 15, “More than 20 times in the past year” was coded to 25. Some of the items used simplified options of “never, once, a few times, many times, not in less than 12 months, but it did happen.” In such cases, “never” was coded to 0, “once” was coded to 1, “a few times” was coded to 3, “many times” was coded to 25, “not in less than 12 months, but it did happen” was coded to 0. Thus, physical violence victimization score (7 items) ranged from 0 to 175 and respondent perpetration (18 items) ranged from 0 to 450.

Sociodemographic covariates included age (years), sexual orientation (straight/gay, lesbian or bisexual), current employment status (unemployed/employed), current relationship status (in a relationship/not in a relationship), and lifetime pregnancy history (have ever been pregnant/never been pregnant).

Statistical analyses were performed using R Studio version 1.3.1093–1 [22]. Participants were excluded by listwise deletion to only include those with complete data (n = 298) [23]. We reported frequencies and percentages for categorical variables, and medians and interquartile ranges for continuous variables. Analyses included descriptive statistics and binomial logistic regression.

The prevalence of each RC subscale (lifetime and 12-month) was calculated and stratified by FS history. Binomial logistic regression was conducted to examine the association between the primary independent variable, FS history, other independent variables psychological violence, physical violence victimization, and respondent-perpetrated physical violence, and dependent variables of RCV (including subscales pregnancy coercion and condom manipulation) and RCP. Unadjusted models included a single independent variable in each model. Adjusted logistic regression models were conducted to examine the association between FS history (primary independent variable) and lifetime and past 12-month RC experiences (victimization and perpetration; dependent variables). We accounted for psychological violence, physical violence victimization, and respondent-perpetrated physical violence, and sociodemographic characteristics of age, sexual orientation, current employment status, current relationship status, and lifetime pregnancy history. Both unadjusted and adjusted models were also constructed to examine the association between FS history and the subscales of RCV (pregnancy coercion and condom manipulation).