Background
Experiences of discrimination in healthcare settings contribute to inequities in healthcare access and outcomes. Black, Latina, and Indigenous communities have lower levels of trust in reproductive healthcare due to a longstanding history of abuses. In this context, discriminatory behaviors by providers may be particularly problematic. This study aimed to test the psychometric properties of a nine-item measure of lifetime experiences of discrimination in family planning settings developed by Thorburn Bird and Bogart (2003).
Methods
Data came from the nationally representative Person-Centered Contraceptive Access Metrics Survey (unweighted N = 3,059). We performed exploratory factor analysis using principal axis extraction and Promax rotation, assessed reliability using coefficient omega, and employed weighted linear regression to test known-groups validity.
Results
We identified two factors and retained seven scale items. The two factors, general healthcare discrimination and stereotype-based discrimination, were consistent with the scale’s original 2005 validation with African American women. The measure demonstrated strong reliability (coefficient omega 0.97) and goodness-of-fit (TLI = 0.988, RMSEA = 0.066). Known-groups validity testing indicated that people of color experienced higher levels of discrimination in family planning settings compared to white respondents.
Conclusions
This updated validation with a nationally representative sample confirms the scale’s utility for assessing discrimination in family planning settings and identifying the relationship with healthcare outcomes. The measure provides a tool to support research on the prevalence of discrimination and its impact on reproductive autonomy, including preferred method use and access to care. The scale may aid the assessment of clinic procedures and interventions promoting equitable access to high-quality reproductive healthcare.
Keywords: Discrimination, Family planning, Reproductive healthcare, Psychometric analysis, Scale validation, Health equity, Patient experiences
Plain language summary
Many people of color get treated badly when they visit doctors for birth control. Doctors might be rude to them or make wrong guesses about their personal lives.
We tested a survey about how people were treated during birth control visits. The survey was first made in 2003 for Black women, but we wanted to see if it works for more types of people across the United States.
We studied answers from people who had talked to doctors about birth control. The questions ask about two things: being treated badly (like being rude or not listening) and unfair assumptions (like wrongly guessing how many kids or sexual partners someone has).
People of color faced more unfair treatment than white people. Black people faced the most, especially unfair assumptions about their lives.
This survey can help researchers and hospitals spot discrimination in family planning care. It can be used to find problems, test if efforts to improve care are working, and make sure hospitals treat all patients fairly. While this tool cannot stop all unfair treatment by itself, it gives us a way to measure problems and see if things get better over time. The goal is fairer birth control care for everyone.
Background
The sexual and reproductive health (SRH) field has a bleak history and contemporary practice of racism-related harms targeting marginalized communities. Historical SRH harms include the forced sterilization of Black, Indigenous, Latina, impoverished, and disabled women, unethical contraceptive trials, and coercive marketing of high-risk birth control methods to Black and Indigenous women [1–3]. These harmful practices continue today, as evidenced by non-consensual hysterectomies on detained immigrants in Georgia [4] and coercive family planning practices such as providers pressuring patients to choose specific contraceptive methods or refusing to remove long-acting reversible contraception (LARC) [5, 6]. Black and Indigenous women often face bidirectional contraceptive coercion, with some being coerced into using LARC, while others face barriers to using their preferred method of contraception [7].
Scientific literature widely documents that communities of color exhibit lower levels of trust in healthcare providers compared to white populations and that personal experience with racism at the hands of providers is high in these communities [3, 8–11], affecting satisfaction with contraceptive care and utilization of preferred contraceptive method [12–14].
One pathway through which inequities in contraceptive outcomes manifest is experiences of racial discrimination within the healthcare setting [15–18]. Racial discrimination may include provider bias, stereotyping, and differential treatment. Widespread racial and ethnic disparities in healthcare quality, treatments offered, and health outcomes in the U.S. [19] can be attributed to several factors, including inequitable healthcare access, health status, patient preferences, and provider bias or discrimination [20]. When controlling for access to health care, patient health status, and patient preferences, these health-related inequities persist, suggesting that discrimination plays a major role in contributing to disparities in healthcare access and outcomes [20].
Black and Latina women disproportionately report negative contraceptive counseling interactions with providers, including discouraging the patient’s preferred method, prescribing unwanted contraception, or failing to share information or solicit patient preferences [21, 22]. In a recent study with a nationally representative sample, more than half of current contraceptive users reported experiencing discrimination in family planning settings, and these experiences were significantly associated with decreased odds of using their preferred contraceptive method [14]. Given these documented patterns of discrimination in family planning settings, accurately measuring these experiences is critical for research and intervention.
Measurement & gaps
Despite acknowledged provider discrimination in contraceptive care, the population health field lacks validated measures to assess patient-reported racial discrimination in the family planning healthcare setting. Most existing studies have used general discrimination measures that assess experiences in settings such as workplaces or schools, rather than specifically in healthcare settings [20, 23, 24]. Many studies assessing discrimination in healthcare settings are qualitative and do not use specific measures of discrimination, and few existing measures have been evaluated psychometrically [25]. Furthermore, Loder et al. [26] emphasized the need for better measures of discrimination specifically in reproductive healthcare settings, particularly given the history of coercive practices that have affected low-income women and women of color. Some researchers have adapted general healthcare discrimination scales, such as the Perceived Discrimination in Healthcare Scale and the Experiences of Discrimination measure, to assess self-reported discrimination in family planning settings. However, these adapted measures fail to capture experiences specific to family planning settings, such as contraceptive coercion or assumptions about sexual or reproductive health [27–30]. Therefore, there remains a need for measures of discrimination that have been developed and validated for use in family planning settings.
To our knowledge, Thorburn Bird1 and Bogart are the only researchers to have developed a measure of racial discrimination for use in family planning settings, which they published in 2003 [31–33]. Their measure asked, “When getting family planning or birth control services, how often have the following things happened to you because of your race or color?” and was followed by nine items, which captured experiences like poor treatment, stereotyping, and coercive behaviors. In their 2005 study of 205 African American women, Thorburn Bird and Bogart conducted an exploratory factor analysis (EFA) of this measure, yielding two factors: general healthcare discrimination experiences and stereotypes about African Americans [31]. This means that the scale measures two distinct but related constructs within the category of racial discrimination: the general healthcare discrimination factor, which includes items about interpersonal treatment (such as being treated with less courtesy or respect or receiving poorer service), and the stereotype-based discrimination factor, which includes items about assumptions providers made based on racial stereotypes (such as assumptions about number of children, sexual partners, or STI status). In their analysis, the authors retained eight of the nine original items, dropping an item about providers encouraging patients to use a method of birth control they did not prefer.
While this measure holds promise as a tool to investigate experiences of racial discrimination in family planning settings, additional evaluation is needed for three reasons. First, the original validation occurred 20 years ago in a substantially different healthcare context. Secondly, the systematic sidelining of equity initiatives in the U.S. has created distinct cultural and healthcare conditions, and thirdly, validation with a larger, more diverse sample is needed to assess the generalizability and appropriateness of the measure beyond the original study population. The original validation was conducted in a sample of African American women; because racial discrimination experiences may manifest differently across racial and ethnic groups, understanding how this measure performs in a racially and ethnically diverse sample was important. In the 20 years since the initial study, despite increased recognition of systemic racism as a public health issue, health disparities and discrimination in healthcare settings have persisted [17, 34, 35]. The COVID-19 pandemic exacerbated already dramatic inequities in healthcare access, further reducing access to care in some populations [36–40].
Additionally, increasing state and federal restrictions on reproductive healthcare access are likely to widen disparities in access to care and subsequent health outcomes [41–43]. Moreover, recent federal policy changes have systematically dismantled health equity initiatives, including the elimination of diversity, equity, and inclusion programs, narrowing of healthcare non-discrimination protections, and dissolution of health equity advisory committees [44, 45]. These changes not only set the stage for increased discrimination in healthcare settings but also systematically reduce protections against these experiences. These converging shifts in the national landscape highlight the need for an updated assessment of the Bird-Bogart measure to test its validity and generalizability to a contemporary and more racially and ethnically diverse population.
Assessing experiences of discrimination enables the identification of patterns and trends that can inform the development and evaluation of targeted interventions and accountability measures to reduce harmful discriminatory practices. Reducing experiences of discrimination based on race, ancestry, or national origin holds the potential to improve one piece of the structural puzzle that contributes to poor health outcomes and access among people of color and others who may experience discrimination in healthcare settings. Given the salience of discrimination in family planning settings and the limitations of existing measures, evaluating the psychometric properties of the only existing scale developed for use in the family planning context is an important step toward elucidating and addressing the role of discrimination in the continuation and exacerbation of reproductive health inequities. We aimed to evaluate the psychometric properties of the Bird-Bogart scale using a large, nationally representative sample of individuals in the U.S. aged 15 to 44 and assigned female sex at birth. We conducted an EFA to identify the underlying factor structure and determine whether the measure captures one or more distinct facets of discrimination experiences. Furthermore, we aimed to assess the reliability of the scale and any subscales and to evaluate the known-groups validity of the measure by examining differences in scale scores across racial and ethnic groups.
Methods
Study sample and data collection
Data for this analysis come from the Person-Centered Contraceptive Access Metrics project conducted by the Sexual Health and Reproductive Equity (SHARE) program at the University of California, Berkeley School of Social Welfare [46]. The project developed novel population-level metrics of person-centered contraceptive access in the U.S. to improve national surveillance efforts and ultimately improve access to contraceptive care. The nationally representative survey data were collected by NORC (National Opinion Research Center) at the University of Chicago, an independent research organization, using their AmeriSpeak panel [47]. The sampling frame for the panel covers an estimated 97% of U.S. households and includes approximately 54,000 members aged 13 and older [47].
The survey was conducted from January to March 2022. Panel members aged 15–44 years, assigned female at birth, not known to be sterile, and able to complete the survey in English or Spanish were eligible. Panelists were invited to participate via email and the AmeriSpeak platform. The survey was self-administered and accessed via a web browser. Participants provided informed consent, or if minors, parental consent and participant assent, followed by eligibility screening. If eligible, participants completed the survey in an average of 21 min. As an incentive, participants were given AmeriPoints equivalent to $8. Over three thousand (3,059) eligible respondents completed the survey.
The study protocol was approved by the Committee for Protection of Human Subjects at the University of California, Berkeley (#2021-02-14025) on June 24, 2021, and by the NORC Institutional Review Board (#21-09-468) on November 4, 2021.
Survey measures
The survey included a nine-item measure of discrimination in family planning settings. This measure was adapted by Thorburn Bird and Bogart from the Everyday Discrimination Scale developed by Williams, Yu, Jackson, and Anderson [28, 32, 33]. Thorburn Bird and Bogart first adapted this scale in 2001 to capture experiences of discrimination in healthcare settings, and further modified it in 2003 to reflect discrimination specifically in family planning settings [31–33]. In the present study, the research team expanded the measure to include experiences of discrimination based on ancestry or national origin in addition to race since the sample included people from all racial and ethnic groups. We conceptualize these as interconnected racialized experiences of discrimination. Due to panel survey length constraints, we were unable to ask about each attribute individually. Responses were not disaggregated by the specific attribute, and these additions were not formally pilot tested. The scale items are presented below.
When getting family planning or birth control services, how often have the following things happened to you because of your race, ancestry, or national origin?
You were treated with less courtesy than other people.
You were treated with less respect than other people.
You received poorer service than other people.
You felt like a doctor or nurse was not listening to what you were saying.
A doctor or nurse assumed you were on welfare.
A doctor or nurse made assumptions about the number of children you have.
A doctor or nurse assumed you had multiple sexual partners.
A doctor or nurse strongly encouraged you to use one method of birth control when you preferred another.
A doctor or nurse assumed you had a sexually transmitted disease such as chlamydia, gonorrhea, genital warts, herpes, or HIV.
Missing data
Of the total unweighted sample of 3,059 respondents, 2,754 (90.0%) had ever discussed contraception with a healthcare provider and were therefore eligible to answer the discrimination questions. Of these, 2,674 (97.1%) responded to all nine discrimination questions. Given that only 80 respondents (2.9%) were missing data on one or more discrimination items, we performed complete case analysis using listwise deletion, resulting in an analytic sample of 2,674 respondents.
Demographic variables
NORC constructed survey weights to create a nationally representative sample based on age, education, race/ethnicity, marital status, number of children in the household, and age by race/ethnicity. We generated descriptive statistics for sociodemographic characteristics including age, age categories, race/ethnicity, and geographic region. We used the survey weights to calculate the weighted proportion and population estimate for each characteristic.
Exploratory factor analysis
We conducted an EFA to assess whether the overall construct of racial discrimination in family planning settings consists of one or multiple distinct constructs. In preparation for the EFA, we assessed multivariate normality using a Mardia test. Because the data were ordinal, we used polychoric correlations for the analysis. We assessed sampling adequacy using the Kaiser-Meyer-Olkin (KMO) measure and conducted Bartlett’s test of sphericity to confirm that the variables were adequately related for factor analysis. Given the multivariate nonnormality, we used principal axis extraction, as principal axis methods can handle this nonnormality. We utilized scree plots, parallel analysis, and network graphs to determine the number of factors to retain. Because the underlying factors are assumed to be related, we selected an oblique rotation. We used a Promax rotation, which is an oblique version of the varimax rotation. We used model fit statistics to aid in model selection. We considered root mean square error of approximation (RMSEA) values > 0.10 as a poor fit, 0.05–0.08 as an adequate fit, and < 0.05 as a good fit [48]. We used a threshold of 0.4 for cross-loadings, consistent with established guidelines.
Additional reliability and validity assessments
We assessed scale reliability using the coefficient omega rather than the coefficient alpha. Coefficient omega provides a more robust reliability estimate when factor loadings vary across items and is less likely to be biased by multidimensionality [49]. To assess known-groups validity, we used the svyglm function from the survey package in R to fit survey-weighted linear models accounting for the complex survey design, and we conducted Wald F-tests to assess the overall significance of race/ethnicity as a categorical predictor of discrimination experiences. Our primary model used a summative count variable of discrimination experiences with respondent race as the predictor. The summative count variable was created by summing the number of different types of discrimination ever experienced, with each experience counted if the respondent selected any response option other than “never happened.” We also ran parallel models using the subscales identified in the EFA. As a sensitivity analysis, we conducted analyses of models using the sum score of the discrimination measure as the outcome variable.
We hypothesized that there would be differential levels of experiences with racial discrimination among survey respondents based on race, with white respondents reporting lower levels of racial discrimination compared to non-white respondents. Additionally, we hypothesized that the subscale related to racial stereotypes, which was originally designed to be about Black women, would be most relevant to Black respondents, although other non-white respondents may also encounter discrimination based on these stereotypes. This stereotype subscale consists of three items from the stereotype factor in Thorburn Bird and Bogart’s original analysis, capturing provider assumptions about the number of children, sexual partners, and sexually transmitted infections (STIs). Recent research has found that these stereotypes remain prevalent [50, 51]. Because these stereotypes may be more salient for Black patients, overall, Black respondents may report more of these experiences in the full scale, as well as the stereotype subscale.
Statistical analysis
We performed all analyses using RStudio Version 2023.06.0 + 421 [52]. We conducted the EFA using the psych package and created visualizations using qgraph and EGAnet. We used survey weights to create survey-weighted estimates for the known-groups validity test using the survey package. All reported proportions are weighted. We considered p-values < 0.05 statistically significant.
Results
Sociodemographic characteristics
Table 1 presents the sociodemographic characteristics of the analytic sample (unweighted n = 2,674; population estimate = 39,652,279). The mean age was 29.0 years (standard deviation (SD) = 7.4), with respondents ranging from 15 to 44 years. The largest age group was 18–24 years (26.0%), followed by 25–29 years (22.8%). The majority of the sample was non-Hispanic white (55.3%). Latina/Hispanic respondents made up 20.0% of the sample, followed by non-Hispanic Black respondents (14.7%). The sample included respondents from the South (35.9%), West (26.7%), Midwest (23.0%), and Northeast (14.4%).
Table 1.
Sociodemographic characteristics
| Categories | Weighted % | 95% CI (%) | Unweighted N |
|---|---|---|---|
| Age | |||
| 15–17 | 6.1 | 4.7–7.5 | 85 |
| 18–24 | 26.0 | 23.0–29.0 | 253 |
| 25–29 | 22.8 | 20.7–24.9 | 586 |
| 30–34 | 20.6 | 18.8–22.4 | 692 |
| 35–39 | 13.8 | 12.4–15.1 | 611 |
| 40–44 | 10.7 | 9.5–11.9 | 447 |
| Race/Ethnicity | |||
| White only | 55.3 | 52.6–57.9 | 1520 |
| Black only | 14.7 | 12.9–16.5 | 370 |
| Latina/Hispanic | 20.0 | 17.8–22.3 | 460 |
| Asian/Pacific Islander only | 5.7 | 4.6–6.8 | 152 |
| Another race/ethnicity only | 0.8 | 0.4–1.1 | 37 |
| Multiracial, not incl. Latina/Hispanic | 3.5 | 2.7–4.2 | 135 |
| Region | |||
| Northeast | 14.4 | 12.5–16.3 | 358 |
| Midwest | 23.0 | 20.8–25.3 | 724 |
| South | 35.9 | 33.3–38.4 | 932 |
| West | 26.7 | 24.3–29.1 | 660 |
Scale summary statistics
Over half (51.9%, 95% CI 49.2–54.5%) of respondents reported any discrimination experiences, while 48.1% reported no discrimination experiences. Table 2 presents the prevalence of discrimination for each item. Prevalence of individual discrimination experiences ranged from 17.9% (provider assumed you had an STI) to 41.9% (provider didn’t listen). Conversely, the percentage reporting no discrimination for each item ranged from 58.1% to 82.1%. Among respondents who reported experiencing discrimination, most indicated that they had experienced it only rarely or sometimes rather than often or always (Table 6 in Appendix).
Table 2.
Respondents reporting discrimination, by type
| Reported Discrimination | Reported No Discrimination | |||
|---|---|---|---|---|
| Unweighted N | Weighted % (95% CI) |
Unweighted N | Weighted % (95% CI) |
|
| Treated with less courtesy (Courtesy) | 729 |
27.8 (25.4, 30.2) |
1945 |
72.2 (69.8, 74.6) |
| Treated with less respect (Respect) | 737 |
28.9 (26.5, 31.3) |
1937 |
71.1 (68.7, 73.5) |
| Received poorer service (Service) | 703 |
27.3 (24.9, 29.7) |
1971 |
72.7 (70.3, 75.1) |
| Provider didn’t listen (Listen) | 1095 |
41.9 (39.2, 44.5) |
1579 |
58.1 (55.5, 60.8) |
| Provider assumed you were on welfare (Welfare) | 516 |
19.8 (17.7, 21.9) |
2158 |
80.2 (78.1, 82.3) |
| Provider made assumptions about number of children (Children) | 526 |
19.3 (17.3, 21.3) |
2148 |
80.7 (78.7, 82.7) |
| Provider made assumptions about number of sexual partners (Partners) | 638 |
25.0 (22.7, 27.3) |
2036 |
75.0 (72.7, 77.3) |
| Provider encouraged non-preferred method (Method) | 780 |
30.4 (27.9, 32.9) |
1894 |
69.6 (67.1, 72.1) |
| Provider assumed you had STI (STI) | 444 |
17.9 (15.8, 20.0) |
2230 |
82.1 (80.0, 84.2) |
Analysis Preparation
A Mardia test showed that the data had a multivariate skew of 23,843.02 (p < 0.001) and kurtosis of 394.05 (p < 0.001), indicating that the ordinal data were not multivariate normal.
The KMO measure of sampling adequacy [53] was excellent, with an overall MSA of 0.95 (individual scores 0.94–0.97). Bartlett’s test of sphericity [54] indicated that the variables were sufficiently related for factor analysis (chi-square = 27,704.80; df = 36; p < 0.001). These measures indicated that the polychoric correlation matrix was appropriate for conducting an EFA.
Parallel analysis [55, 56], a Minimum Average Partial (MAP) test [57], and a scree plot [58] all indicated that a one-factor solution was the best fit for the data (see Fig. 1) [59].
Fig. 1.
Parallel analysis and scree plot for factor retention. Legend: Parallel analysis and scree plot used to determine the optimal number of factors for the nine-item discrimination scale (n = 2,674). The blue line shows the actual values from the data, while the dotted red line displays simulated random data. The analysis suggests retaining only factors where actual eigenvalues exceed simulated values
Correlation matrix and factor retention
The polychoric correlation matrix (Fig. 2) shows correlations between each item, ranging from 0.65 to 0.90.
Fig. 2.
Polychoric correlation matrix for discrimination scale items. Legend: Polychoric correlations between the nine items in the discrimination scale. Coefficients range from 0.65 to 0.90, with higher numbers and darker blues indicating stronger correlations. All items show high correlations, indicating strong internal consistency
Although initial tests (parallel analysis, MAP test, scree plot) suggested a one-factor solution, we tested multiple factor structures to determine which structure best fit the data. We evaluated model fit, using root mean square residual (RMSR), Tucker-Lewis Index (TLI), RMSEA, and Bayesian Information Criterion (BIC) to compare one- and two-factor solutions. We tested four models, beginning with a one-factor solution with nine items (Table 3). This model showed high loadings ranging from 0.82 to 0.93 and explained 76% of the variance. The one-factor model showed good fit for RMSR (0.032) and TLI (0.929), but poor fit for RMSEA (0.1433).
Table 3.
One- and two-factor, nine-item solution factor loadings
| Variable | One-Factor Solution | Two-Factor Solution | |||||
|---|---|---|---|---|---|---|---|
| PA1 | h2 | u2 | PA1 | PA2 | h2 | u2 | |
| Courtesy | 0.919 | 0.844 | 0.156 | 0.833 | 0.122 | 0.879 | 0.121 |
| Respect | 0.928 | 0.862 | 0.138 | 0.836 | 0.129 | 0.896 | 0.104 |
| Service | 0.913 | 0.834 | 0.166 | 0.839 | 0.110 | 0.870 | 0.130 |
| Listen | 0.822 | 0.677 | 0.323 | 0.723 | 0.131 | 0.698 | 0.302 |
| Welfare | 0.893 | 0.798 | 0.202 | 0.416 | 0.514 | 0.796 | 0.204 |
| Children | 0.832 | 0.692 | 0.308 | 0.250 | 0.622 | 0.710 | 0.290 |
| Partners | 0.863 | 0.745 | 0.255 | 0.085 | 0.829 | 0.812 | 0.188 |
| Method | 0.822 | 0.675 | 0.325 | 0.438 | 0.417 | 0.673 | 0.327 |
| STI | 0.846 | 0.716 | 0.284 | 0.032 | 0.867 | 0.799 | 0.201 |
PA1 and PA2 indicate factor loadings on the first and second factors. h2 indicates communality (proportion of item variance explained by the factor in the one-factor solution or both factors in the two-factor solution) and u2 indicates uniqueness (proportion of item variance not explained by the factor in the one-factor solution or both factors in the two-factor solution). Bolded loadings indicate substantive factor loadings (≥ 0.40) in the two-factor solution
A two-factor solution (Table 3) resulted in four items loading cleanly onto the first factor and three loading cleanly onto the second factor, with two items (“A doctor or nurse assumed you were on welfare” and “A doctor or nurse strongly encouraged you to use one method of birth control when you preferred another”) cross-loading onto both factors (threshold = 0.4). RMSR and TLI scores improved in comparison to the one-factor model (RMSR = 0.016; TLI = 0.974). The RMSEA (0.0858) was slightly improved from the one-factor model but still indicated only a mediocre fit. Based on these results, we dropped the two cross-loading items and reran the models, retaining the remaining seven items.
After dropping these two items, we reran the parallel analysis, MAP test, and scree plot. All tests again indicated that a one-factor solution best fit the data. Next, we conducted a one-factor analysis with seven items (Table 4). Factor loadings ranged from 0.819 to 0.935, indicating a strong association with the single factor. The RMSR (0.036) and TLI (0.9203) indicated good fit, whereas the RMSEA (0.1716) indicated poor fit.
Table 4.
One- and two-factor, seven-item solution factor loadings
| Variable | One-Factor Solution | Two-Factor Solution | |||||
|---|---|---|---|---|---|---|---|
| PA1 | h2 | u2 | PA1 | PA2 | h2 | u2 | |
| Courtesy | 0.923 | 0.852 | 0.148 | 0.855 | 0.100 | 0.882 | 0.118 |
| Respect | 0.935 | 0.875 | 0.125 | 0.861 | 0.107 | 0.905 | 0.096 |
| Service | 0.916 | 0.840 | 0.160 | 0.853 | 0.095 | 0.870 | 0.130 |
| Listen | 0.823 | 0.677 | 0.323 | 0.728 | 0.122 | 0.691 | 0.309 |
| Children | 0.819 | 0.671 | 0.329 | 0.332 | 0.530 | 0.681 | 0.319 |
| Partners | 0.858 | 0.736 | 0.264 | 0.140 | 0.783 | 0.814 | 0.187 |
| STI | 0.845 | 0.714 | 0.286 | 0.041 | 0.879 | 0.833 | 0.167 |
PA1 and PA2 indicate factor loadings on the first and second factors. h2 indicates communality (proportion of item variance explained by the factor in the one-factor solution or both factors in the two-factor solution) and u2 indicates uniqueness (proportion of item variance not explained by the factor in the one-factor solution or both factors in the two-factor solution). Bolded loadings indicate substantive factor loadings (≥ 0.40) in the two-factor solution
Because the RMSEA for the single-factor solution was poor, we ran a two-factor model (Table 4). Four items loaded highly on the first factor (0.728 to 0.861), and the other three loaded highly on the second factor (0.530 to 0.879). Given a threshold of 0.4, there were no cross-loadings, although item 6 did have a loading of 0.332 on the first factor.
Communalities were high (0.681–0.905), indicating that the two factors explained a substantial amount of the variance, and uniqueness was low. The cumulative variance explained was 81.1%, with the first factor accounting for 48.8%, and the second accounting for 32.3%. The correlation between the two factors was 0.822. All three fit indices were improved over the other three models: the RMSR and TLI indicated good model fit (0.009 and 0.9882, respectively), while the RMSEA was improved with an acceptable model fit of 0.0659 (95% CI: 0.0548–0.0777).
We selected the two-factor model with seven items as the model that best represents the underlying factor structure. This model had the only acceptable RMSEA of the four models (0.0659), good fit indices (RMSR = 0.009 and TLI = 0.9882), and no cross-loadings above 0.4, indicating good factor structure. The model also had high communalities and low uniqueness, suggesting that a large amount of the variance in each item was explained by the two factors. The two factors explain 81.1% of the total variance, and the BIC was lowest for this model, indicating the best model fit. Overall, the two-factor, seven-item model addresses issues with model fit while explaining a high amount of variance and maintaining a clear factor structure. The two-factor model is also theoretically appropriate and consistent with the findings from the 2005 factor analysis [31].
The two subscales identified by this model encompass (1) general healthcare discrimination, including “You were treated with less courtesy than other people,” “You were treated with less respect than other people,” “You received poorer service than other people,” “You felt like a doctor or nurse was not listening to what you were saying,” and (2) stereotype-based discrimination, including “A doctor or nurse made assumptions about the number of children you have,” “A doctor or nurse assumed you had multiple sexual partners,” and “A doctor or nurse assumed you had a sexually transmitted disease such as chlamydia, gonorrhea, genital warts, herpes, or HIV.”
Reliability
McDonald’s coefficient omega [60] was used to assess the reliability of the measure and its subscales. The overall omega value was 0.97, indicating excellent reliability across all sources of common variance in the model. The omega hierarchical values for the two factors were 0.84 and 0.82, indicating that they also had good reliability and accounted for high amounts of variance.
Known-groups validity
The results of the known-groups validity analyses are displayed in Table 5. Race/ethnicity was significantly associated with experiences of discrimination overall. For the seven-item measure, the overall effect of race/ethnicity was statistically significant (F(5, 2668) = 10.85, p < 0.001). The healthcare discrimination (F(5, 2668) = 11.71, p < 0.001) and stereotype subscale (F(5, 2668) = 6.66, p < 0.001) were both also associated with race/ethnicity.
Table 5.
Known-Groups validity: number of discrimination types experienced
| Full Scale | |||||
|---|---|---|---|---|---|
| Predicted Score | Estimate | Std. Error | 95% CI | P-value | |
| White, non-Hispanic | 1.47 | Ref | |||
| Another race/ethnicity only | 2.55 | 1.08 | 0.48 | 0.13, 2.02 | 0.026 |
| Asian/Pacific Islander only | 1.90 | 0.43 | 0.26 | -0.08, 0.95 | 0.100 |
| Black, non-Hispanic | 2.74 | 1.27 | 0.21 | 0.86, 1.69 | < 0.001 |
| Latina/Hispanic | 2.27 | 0.80 | 0.18 | 0.46, 1.15 | < 0.001 |
| Multiracial, not inc. Latina/Hispanic | 2.36 | 0.89 | 0.31 | 0.29, 1.50 | 0.004 |
| Healthcare Discrimination Scale | |||||
|---|---|---|---|---|---|
| Predicted Score | Estimate | Std. Error | 95% CI | P-value | |
| White, non-Hispanic | 0.98 | Ref | |||
| Another race/ethnicity only | 1.85 | 0.87 | 0.36 | 0.17, 1.56 | 0.015 |
| Asian/Pacific Islander only | 1.31 | 0.33 | 0.17 | -0.01, 0.67 | 0.057 |
| Black, non-Hispanic | 1.79 | 0.81 | 0.13 | 0.55, 1.06 | < 0.001 |
| Latina/Hispanic | 1.55 | 0.57 | 0.12 | 0.34, 0.80 | < 0.001 |
| Multiracial, not inc. Latina/Hispanic | 1.60 | 0.62 | 0.20 | 0.22, 1.02 | 0.002 |
| Stereotype Scale | |||||
|---|---|---|---|---|---|
| Predicted Score | Estimate | Std. Error | 95% CI | P-value | |
| White, non-Hispanic | 0.49 | Ref | |||
| Another race/ethnicity only | 0.70 | 0.21 | 0.23 | -0.24, 0.67 | 0.363 |
| Asian/Pacific Islander only | 0.59 | 0.10 | 0.10 | -0.10, 0.31 | 0.330 |
| Black, non-Hispanic | 0.96 | 0.47 | 0.09 | 0.29, 0.64 | < 0.001 |
| Latina/Hispanic | 0.72 | 0.23 | 0.07 | 0.08, 0.38 | 0.002 |
| Multiracial, not inc. Latina/Hispanic | 0.77 | 0.28 | 0.13 | 0.02, 0.53 | 0.036 |
Bolded p-values indicate statistical significance at p < 0.05.
The linear model using the seven-item measure (where possible counts range from zero to seven types of discrimination experienced) revealed significant racial/ethnic differences, as hypothesized. Black, Latina/Hispanic, multiracial, and respondents of other races all reported having experienced more types of discrimination compared with white respondents (predicted score 1.47). Black respondents experienced the most discrimination types (predicted score 2.74; 95% confidence interval (CI) 0.86 to 1.69), followed by people classified in the “other” race/ethnicity category (predicted score 2.55; 95% CI 0.13 to 2.02).
The healthcare discrimination subscale yielded results similar to those of the seven-item measure. In alignment with our hypothesis, the stereotype scale showed that Black (0.96 types; difference: 0.47; 95% CI 0.29, 0.64), Latina/Hispanic (0.72 types; difference: 0.23; 95% CI 0.08, 0.38), and multiracial (0.77; difference: 0.28; 95% CI 0.02 to 0.53) respondents had significantly more discrimination experiences than white respondents (predicted score 0.49).
Sensitivity analyses using the continuous scores yielded comparable results to the primary analyses (see Table 2 in Appendix).
Discussion
This study aimed to assess the psychometric properties of a scale designed to measure experiences of discrimination in family planning settings, originally developed by Thorburn Bird and Bogart [32], in a diverse, nationally representative sample. Based on the model fit statistics and factor loadings, we selected a two-factor solution with seven items as the factor structure that best fit the data. These two factors are (1) general healthcare discrimination and (2) racial stereotype-based discrimination. We dropped two items from the original scale: “A doctor or nurse assumed you were on welfare,” and “A doctor or nurse strongly encouraged you to use one method of birth control when you preferred another.” Our decision to drop these items was empirically grounded and conceptually informed. Model fit statistics supported this decision, with all fit statistics improving after removing both items. The welfare item cross-loaded onto both factors, suggesting it does not map cleanly onto either subscale in this sample. While assumptions about welfare status are rooted in racial discrimination and harmful stereotypes, particularly for Black and Latina women [50, 51], this type of discrimination is less explicitly specific to the family planning setting relative to the other items in the subscale. The general discrimination subscale contains items that relate to treatment by providers, without attribution to a particular reason for that treatment. The stereotype subscale contains items relating to racist assumptions about reproductive health-related topics. While the item about a provider assuming someone is on welfare is about assumptions, it is not explicitly related to reproductive health behaviors or outcomes in the way that number of children, STI status, or number of sexual partners is. The item’s cross-loading pattern suggests it captures a broader sociopolitical construct that extends beyond reproductive health-specific stereotypes. By removing the welfare item from the subscale, we are better able to capture discrimination specific to family planning settings. The welfare item captures a distinct construct not fully captured by the current scale; nevertheless, this item remains relevant and important in its own right and may be used independently of the full scale.
The item about encouragement to use a non-preferred contraceptive method differs from the other items in the scale in that it refers to a specific clinical recommendation behavior rather than general discriminatory treatment or stereotype-based treatment. Additionally, this captures one type of contraceptive coercion, missing other types, such as encouraging a particular method when patients had no preference or preferred no method. This narrow focus means the item does not capture the range of coercive experiences. Given these measurement limitations and poor factor loading, contraceptive coercion may be better assessed through separate, more comprehensive measures.
By retaining only the items that load cleanly onto one of the two subscales, the two-factor solution provides a more statistically robust measure. Beyond statistical considerations, this two-factor structure has important conceptual value. Two distinct but related types of racial discrimination experiences that patients have in family planning settings are measured. The general discrimination factor captures the quality of interpersonal treatment, including whether patients were treated with respect or listened to. These items reflect discriminatory behavior that is fundamental to the quality of care received. The stereotype factor captures discriminatory experiences that reflect providers’ racialized assumptions about their patients’ reproductive lives, including assumptions about number of sexual partners, number of children, and STI status. These stereotypes are particularly salient in the family planning setting, where patients’ reproductive lives are the focus of care and where harmful stereotypes about Black women’s sexuality have historically been used to justify reproductive coercion [1–3, 50, 61, 62]. This conceptual distinction can allow quality improvement efforts to assess whether interventions differentially affect more general interpersonal treatment quality versus the activation of racialized stereotypes, and act to more directly address either or both impediments to more patient-centered care.
Our results are broadly consistent with the conclusions of Thorburn Bird and Bogart’s 2005 EFA, which also found two factors along the same patterns but retained eight rather than seven items [31]. The 2005 analysis dropped the item about contraceptive method but retained the item about welfare status.
Comparison with other discrimination measures
This measure addresses critical gaps in discrimination measurement in the reproductive healthcare context. While general measures of discrimination and those measuring discrimination in healthcare settings exist and provide valuable insights into discriminatory experiences [27–30, 63], we are not aware of other validated tools specifically designed to measure racial discrimination in reproductive healthcare settings. Past research into discrimination in childbirth settings has measured how often respondents were treated poorly because of their race, ethnicity, cultural background, or language [64], but it is unknown whether and how such items might capture the specific racialized assumptions women of color face in these settings. The present measure complements existing tools by explicitly measuring both interpersonal treatment and stereotype-based discrimination specific to reproductive health.
Measurement approaches and theoretical considerations
This updated psychometric evaluation of the Bird-Bogart scale confirms its utility for assessing experiences of racial discrimination in family planning settings. The revised seven-item version maintains strong reliability and demonstrates expected patterns of discrimination experiences across racial/ethnic groups. The two-factor structure offers important advantages over a single-factor solution, offering researchers and practitioners more nuanced information about the related but distinct types of discrimination patients may encounter in a family planning setting. The general healthcare discrimination factor captures information about interpersonal treatment quality, while the stereotype factor reflects SRH-specific stigmatization. Researchers can use these subscales independently or as separate predictors in statistical models to examine whether each type of discrimination has a differential association with reproductive health outcomes. This approach enables the prioritization of interventions intended to address these experiences.
This paper tested two approaches to scoring the scale. The first is a count-based approach (range: 0–7 for the full scale; 0–4 for the healthcare subscale; and 0–3 for the stereotype subscale), where any response other than “never happened” indicates experience of that type of discrimination. Count-based approaches to scale scoring, which describe the number of different types of discrimination experienced, may reflect a relationship to outcomes based on the breadth of discrimination types rather than their frequency. The second (Table 2 in Appendix) is a continuous, frequency-based measure (range: 7–35 for full scale), where lower scores indicate poorer treatment/more discrimination experiences. This approach implies a dose-response relationship where the frequency of discriminatory experiences is the driving factor affecting outcomes.
In our known-groups validity testing, we tested the two approaches and found that both methods demonstrated similar patterns of racial/ethnic differences. The consistency of these findings suggests that either approach may be used to detect meaningful group differences, providing researchers with flexibility in choosing the approach that best aligns with their theoretical framework. However, count-based scoring offers easier interpretability compared with continuous scoring. While the continuous sum score analysis showed that Black respondents scored 2.38 points lower than white respondents (30.28 vs. 32.66 on the full scale), this difference is difficult to interpret meaningfully. Contrastingly, using the count-based approach, Black respondents experienced an average of 1.27 additional types of discrimination compared with white respondents (2.74 vs. 1.47 types). This clarity is helpful for practical applications of the scale, such as intervention design and quality improvement efforts.
Recent research has used the scale as a three-level ordinal variable, capturing the number of types of discrimination experienced (0, 1–4, or 5–9 types) [14, 65], and has demonstrated meaningful relationships with contraceptive method satisfaction and healthcare engagement. Based on our psychometric evaluation, comparison between the two approaches, and recent successful applications of the count approach, we recommend the count-based approach given its greater interpretability, while noting that continuous scoring preserves response variability and may be preferable for some modeling approaches.
Limitations
While this study has several strengths, including a large, diverse, and nationally representative sample and a robust psychometric evaluation, it also has limitations. The discrimination measure asks respondents how often discrimination occurred over their lifetimes rather than at a discrete point in time. Because of this, responses may be subject to recall bias, as respondents may struggle to remember specific experiences or over- or under-estimate the frequency of discriminatory experiences, especially if long periods of time have passed. There may also be differences in how respondents interpret and respond to the question, depending on factors such as age or how often they have sought family planning care in the past.
The survey instrument may also be subject to cultural or linguistic biases that affect how respondents understand and respond to the questions. Respondents from diverse linguistic and cultural backgrounds may conceptualize discrimination differently, depending on cultural norms around mistreatment, discrimination, or expectations for healthcare interactions. Additionally, while the survey was provided in English and Spanish, the item was developed in English, and translation may not fully address underlying differences in how discrimination is conceptualized and discussed.
Some limitations related to the measurement of race, ancestry, and national origin should be noted. The addition of ancestry and national origin to the measure was not pilot tested. While we conceptualize race, ancestry, and national origin as interconnected racialized experiences, future work should validate this combined approach and assess whether it captures the intended construct as effectively as a race-based question alone. This is an especially salient question given the current U.S. sociopolitical context in which immigration enforcement policies and executive actions disproportionally target immigrant communities, particularly those of Latin American origin.
Secondly, respondents may have answered differently if they had been asked about race, ancestry, and national origin individually. While we were unable to disaggregate responses, this approach prevents us from understanding whether respondents attribute their mistreatment to one identity over another, or to multiple intersecting racialized or otherwise marginalized identities.
Implications and future directions
By building on the original 2005 validation of the Bird-Bogart scale with African American women, our study extends validation to a large, contemporary, nationally representative sample. Our findings confirm the two-factor structure originally identified by Thorburn Bird and Bogart, while suggesting minor modifications to enhance the scale’s measurement properties.
Future psychometric research should examine whether the factor structure identified here replicates across study populations with differing racial/ethnic compositions and across other sociodemographic characteristics. In the present sample, the welfare item cross-loaded onto both factors and was dropped, whereas in Thorburn Bird and Bogart’s 2005 psychometric analysis with Black women, the item loaded cleanly onto the stereotype-based discrimination factor [31]. This suggests some variation in how the measure behaves across differing samples. Future scale development work might explore additional items that could hang together with the welfare item to more robustly measure this important construct and help distinguish between general discrimination, reproductive health-specific stereotypes, and broader sociopolitical assumptions, all of which may affect patient experiences and outcomes in family planning settings. Additionally, researchers may consider including all nine items in the original scale in their surveys and assessing the welfare and method items independently, as they capture important experiences even while not fitting cleanly onto the current factor structure.
Recent research demonstrates the value of the scale in understanding reproductive health inequities. This work has found that lifetime experience of more types of discrimination in family planning settings is associated with women not using their preferred contraceptive method and lacking adequate information for contraceptive decision-making [14]. These experiences, including having concerns dismissed, facing stereotyping, and receiving inadequate care, can lead to self-protective behaviors among Black women, including delaying care, overpreparing for appointments, or seeking care only from same-race providers [66, 67]. Experiences of racial discrimination contribute to lower satisfaction with contraceptive methods and reduced autonomy, with some patients receiving a non-preferred method or avoiding contraceptive care entirely to avoid experiences of discrimination [13, 29]. This work could expand beyond family planning and into other reproductive healthcare settings, and future work could assess the utility of this scale as written or with content changes specific to fertility, prenatal care, or abortion care settings.
Beyond research, this measure has promising implications for healthcare settings. The scale assesses discrimination experiences across the respondents’ lifetimes. Researchers might consider developing and validating a version that assesses discriminatory experiences at the most recent encounter to enhance the scale’s utility for quality assessment in clinical settings. Clinics may consider incorporating this measure into post-visit surveys within electronic health records to facilitate tracking and intervention, such as by sending post-visit follow-up surveys via the patient portal or text message. This measure has potential value in using the stereotype subscale in settings serving predominantly Black and Latina populations. The scale may also be used to evaluate programs and interventions intended to improve quality of care through provider education. It can serve as a tool for pre- and post-assessments of the impact of such programs at the organizational, community, or state levels. For example, statewide contraceptive access initiatives could implement this measure to promote equity-centered principles that center patient experiences [68]. Clinical quality improvement initiatives could also incorporate this measure as a patient-reported outcome to monitor and address discriminatory experiences in family planning care.
Conclusions
This validated measure serves as one tool in the efforts to address racism in reproductive healthcare. By quantifying patient experiences of discrimination, this measure can be used to identify patterns of discrimination and provide support for needed systemic changes. While no single measure can solve deeply rooted systemic issues, as part of larger systemic changes and reforms at the organizational, state, or national level, this measure offers a concrete, evidence-based mechanism to assess the patient experience. Used in conjunction with equity-oriented, patient-centered initiatives, this measure can provide a standardized baseline and a means to ensure that efforts to reduce discrimination in family planning settings are evidence-based and accountable to the communities they aim to serve.
Acknowledgements
Not applicable.
Abbreviations
- BIC
Bayesian Information Criterion
- CI
Confidence Interval
- EFA
Exploratory Factor Analysis
- HIV
Human Immunodeficiency Virus
- KMO
Kaiser-Mayer Olkin
- LARC
Long-Acting Reversible Contraception
- MAP
Minimum Average Partial
- NORC
National Opinion Research Center
- RMSEA
Root Mean Square Error of Approximation
- RMSR
Root Mean Square Residual
- SD
Standard Deviation
- SHARE
Sexual Health and Reproductive Equity
- SRH
Sexual and Reproductive Health
- STI
Sexually Transmitted Infection
- TLI
Tucker-Lewis Index
Appendix
Table 6 Frequency of types of racial discrimination experiences in family planning settings
| Always | Most of the Time | Sometimes | Rarely | Never | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Unweighted N | Weighted % (95% CI) | Unweighted N | Weighted % (95% CI) | Unweighted N | Weighted % (95% CI) | Unweighted N | Weighted % (95% CI) | Unweighted N | Weighted % (95% CI) |
|
|
Treated with less courtesy (Courtesy) |
23 |
0.93 (0.49–1.37) |
48 |
2.28 (1.45–3.12) |
249 |
9.6 (8.07–11.07) |
409 |
14.98 (13.10-16.92) |
1945 |
72.22 (69.84–74.58) |
|
Treated with less respect (Respect) |
22 |
0.90 (0.42–1.38) |
56 |
2.73 (1.85–3.62) |
237 |
9.24 (7.70-10.73) |
422 |
16.03 (14.05–18.07) |
1937 |
71.11 (68.65–73.53) |
|
Received poorer service (Service) |
21 |
0.83 (0.36–1.30) |
46 |
2.15 (1.28–3.01) |
235 |
8.58 (7.20–9.98) |
401 |
15.69 (13.66–17.72) |
1971 |
72.75 (70.34–75.14) |
|
Provider didn’t listen (Listen) |
40 |
1.40 (0.87–1.93) |
110 |
4.49 (3.39–5.57) |
468 |
17.54 (15.50–19.60) |
477 |
18.43 (16.33–20.60) |
1579 |
58.14 (55.46–60.75) |
|
Provider assumed you were on welfare (Welfare) |
29 |
1.34 (0.68–2.01) |
44 |
1.73 (1.08–2.38) |
189 |
6.71 (5.53–7.89) |
254 |
10.05 (8.37–11.73) |
2158 |
80.17 (78.06–82.28) |
|
Provider made assumptions about number of children (Children) |
30 |
1.24 (0.64–1.84) |
42 |
1.58 (0.94–2.23) |
188 |
6.88 (5.66–8.14) |
266 |
9.59 (8.04–11.10) |
2148 |
80.71 (78.69–82.74) |
|
Provider made assumptions about number of sexual partners (Partners) |
31 |
1.09 (0.63–1.55) |
62 |
3.23 (2.15–4.33) |
230 |
8.93 (7.43–10.44) |
315 |
11.72 (10.01–13.41) |
2036 |
75.03 (72.70-77.35) |
|
Provider encouraged non-preferred method (Method) |
48 |
1.71 (1.12–2.30) |
96 |
4.46 (3.33–5.60) |
307 |
11.94 (10.22–13.70) |
329 |
12.26 (10.42–14.10) |
1894 |
69.64 (67.14–72.10) |
|
Provider assumed you had STI (STI) |
16 |
0.61 (0.18-1.00) |
40 |
2.13 (1.24–3.02) |
135 |
5.31 (4.20–6.44) |
253 |
9.86 (8.23–11.47) |
2230 |
82.1 (80.04–84.17) |
Table 7 Known-Groups Validity: Continuous Scale Score
| Full Scale | |||||
|---|---|---|---|---|---|
| Predicted Score | Estimate | Std. Error | 95% CI | P-value | |
| White, non-Hispanic | 32.66 | Ref | |||
| Another race/ethnicity only | 30.06 | -2.60 | 1.17 | -4.89, -0.32 | 0.026 |
| Asian/Pacific Islander only | 31.95 | -0.72 | 0.47 | -1.64, 0.21 | 0.130 |
| Black, non-Hispanic | 30.28 | -2.38 | 0.40 | -3.16, -1.59 | < 0.001 |
| Latina/Hispanic | 31.08 | -1.58 | 0.36 | -2.28, -0.87 | < 0.001 |
| Multiracial, not inc. Latina/Hispanic | 30.91 | -1.75 | 0.66 | -3.04, -0.45 | 0.008 |
| Healthcare Discrimination Scale | |||||
|---|---|---|---|---|---|
| Predicted Score | Estimate | Std. Error | 95% CI | P-value | |
| White, non-Hispanic | 18.47 | Ref | |||
| Another race/ethnicity only | 16.33 | -2.14 | 0.96 | -4.02, -0.26 | 0.026 |
| Asian/Pacific Islander only | 17.92 | -0.55 | 0.30 | -1.14, 0.04 | 0.066 |
| Black, non-Hispanic | 16.94 | -1.53 | 0.25 | -2.01, -1.04 | < 0.001 |
| Latina/Hispanic | 17.39 | -1.08 | 0.22 | -1.52, -0.64 | < 0.001 |
| Multiracial, not inc. Latina/Hispanic | 17.17 | -1.30 | 0.45 | -2.18, -0.43 | 0.004 |
| Stereotype Scale | |||||
|---|---|---|---|---|---|
| Predicted Score | Estimate | Std. Error | 95% CI | P-value | |
| White, non-Hispanic | 14.19 | Ref | |||
| Another race/ethnicity only | 13.73 | -0.47 | 0.49 | -1.42, 0.49 | 0.338 |
| Asian/Pacific Islander only | 14.03 | -0.17 | 0.20 | -0.56, 0.23 | 0.412 |
| Black, non-Hispanic | 13.34 | -0.85 | 0.18 | -1.20, -0.50 | < 0.001 |
| Latina/Hispanic | 13.69 | -0.50 | 0.16 | -0.81, -0.19 | 0.002 |
| Multiracial, not inc. Latina/Hispanic | 13.75 | -0.44 | 0.26 | -0.95, 0.06 | 0.085 |
Authors’ contributions
AAL co-conceptualized the study, acquired funding, designed the research questions and methods, led the data analysis, created the visualizations, and drafted the manuscript. She also ensured adherence to ethical standards and project timelines. AMG co-conceptualized the study, contributed to data collection, and provided critical revisions to the manuscript. AHB contributed to data collection and manuscript revisions. MDL provided statistical guidance and contributed to the interpretation of results. JMS reviewed and revised manuscript drafts for intellectual content. SKR acquired funding, provided oversight, and contributed to manuscript revisions. WSR co-conceptualized the study, acquired funding, provided senior guidance, and contributed to the interpretation of findings and manuscript development.All authors read and approved the final manuscript.
Funding
Data analysis and this publication were supported by The Center for Reproductive Health Research in the Southeast at Emory University through grants from an anonymous foundation. SK Redd was supported in part by the Building Interdisciplinary Research Careers in Women’s Health of the National Institutes of Health (NIH) under Award Number K12AR084234. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Data availability
The datasets analyzed during the current study are available on the Open Science Framework (OSF) at https://osf.io/wvb5m/.
Declarations
Ethics approval and consent to participate
The study protocol was approved by the Committee for Protection of Human Subjects at the University of California, Berkeley (#2021-02-14025) on June 24, 2021, and by the NORC Institutional Review Board (#21-09-468) on November 4, 2021. Informed consent to participate in the study was obtained from all participants. For participants who were minors, informed consent was obtained from their guardian, and assent was attained from the participants.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Thorburn Bird has previously been published under the names Bird, Thorburn Bird, and Thorburn. For the sake of clarity, we will refer to her as Thorburn Bird throughout, while referring to the scale as the Bird-Bogart Scale. Citations are as published.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Thompson BY. Centering reproductive justice: transitioning from abortion rights to social justice. In: Ross LJ, Roberts L, Derkas E, Peoples W, Bridgewater Toure P, editors. Radical reproductive justice: Foundation, theory, practice, critique. New York, NY: Feminist; 2017. [Google Scholar]
- 2.Roberts DE. Killing the black body: race, reproduction, and the meaning of liberty. 2017.
- 3.Washington HA. Medical apartheid: the dark history of medical experimentation on black Americans from colonial times to the present. Doubleday Books; 2006.
- 4.Ghandakly EC, Fabi R. Sterilization in US immigration and customs enforcement’s (ICE’s) detention: ethical failures and systemic injustice. American Public Health Association; 2021. pp. 832–4. [DOI] [PMC free article] [PubMed]
- 5.Cohen RE, Wilkinson TA, Staples-Horne M. The need for reproductive justice in pediatrics. JAMA Pediatr. 2021;175(12):1207–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Brandi K, Woodhams E, White KO, Mehta PK. An exploration of perceived contraceptive coercion at the time of abortion. Contraception. 2018;97(4):329–34. [DOI] [PubMed] [Google Scholar]
- 7.Bryson A, Koyama A, Hassan A. Addressing long-acting reversible contraception access, bias, and coercion: supporting adolescent and young adult reproductive autonomy. Curr Opin Pediatr. 2021;33(4):345–53. [DOI] [PubMed] [Google Scholar]
- 8.Thom DH, Campbell B. Physician-Patient trust: an empirical study. J FAM PRAC. 1997;44:169. [PubMed] [Google Scholar]
- 9.Boulware LE, Cooper LA, Ratner LE, LaVeist TA, Powe NR. Race and trust in the health care system. Public Health Rep. 2003;118(4):358–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sullivan LS. Trust, Risk, and race in American medicine. Hastings Cent Rep. 2020;50(1):18–26. [DOI] [PubMed] [Google Scholar]
- 11.Armstrong K, Putt M, Halbert CH, Grande D, Schwartz JS, Liao K, et al. Prior experiences of Racial discrimination and Racial differences in health care system distrust. Med Care. 2013;51(2):144–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rocca CH, Harper CC. Do Racial and ethnic differences in contraceptive attitudes and knowledge explain disparities in method use? Perspect Sex Reprod Health. 2012;44(3):150–8. [DOI] [PubMed] [Google Scholar]
- 13.Oakley LP, Harvey SM, Lopez-Cevallos DF, Racial, Discrimination E. Medical Mistrust, and satisfaction with birth control services among young adult Latinas. Womens Health Issues. 2018;28(4):313–20. [DOI] [PubMed] [Google Scholar]
- 14.Gomez AM, Bennett AH, Arcara J, Stern L, Bardwell J, Cadena D, et al. Estimates of use of preferred contraceptive method in the united states: a population-based study. Lancet Reg Health Am. 2024;30:100662. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Alhusen JL, Bower KM, Epstein E, Sharps P. Racial discrimination and adverse birth outcomes: an integrative review. J Midwifery Womens Health. 2016;61(6):707–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Dominguez TP, Dunkel-Schetter C, Glynn LM, Hobel C, Sandman CA. Racial differences in birth outcomes: the role of general, pregnancy, and racism stress. Health Psychol. 2008;27(2):194–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Davis DA. Obstetric racism: the Racial politics of Pregnancy, Labor, and birthing. Med Anthropol. 2019;38(7):560–73. [DOI] [PubMed] [Google Scholar]
- 18.Giurgescu C, McFarlin BL, Lomax J, Craddock C, Albrecht A. Racial discrimination and the black-white gap in adverse birth outcomes: a review. J Midwifery Womens Health. 2011;56(4):362–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nelson A. Unequal treatment: confronting Racial and ethnic disparities in health care. J Natl Med Assoc. 2002;94(8):666–8. [PMC free article] [PubMed] [Google Scholar]
- 20.Kressin NR, Raymond KL, Manze M. Perceptions of race/ethnicity-based discrimination: a review of measures and evaluation of their usefulness for the health care setting. J Health Care Poor Underserved. 2008;19(3):697–730. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Yee LM, Simon MA. Perceptions of coercion, discrimination and other negative experiences in postpartum contraceptive counseling for low-income minority women. J Health Care Poor Underserved. 2011;22(4):1387–400. [DOI] [PubMed] [Google Scholar]
- 22.Logan RG, Daley EM, Vamos CA, Louis-Jacques A, Marhefka SL. When is health care actually going to be care? The lived experience of family planning care among young black women. Qual Health Res. 2021;31(6):1169–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sanford K, Pizzuto AE. The healthcare discrimination experience scale: assessing a variable crucial for explaining Racial/Ethnic inequities in patient activation and health. J Racial Ethn Health Disparities. 2023;10(4):1642–52. [DOI] [PubMed] [Google Scholar]
- 24.Shavers VL, Fagan P, Jones D, Klein WM, Boyington J, Moten C, et al. The state of research on racial/ethnic discrimination in the receipt of health care. Am J Public Health. 2012;102(5):953–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Akinade T, Kheyfets A, Piverger N, Layne TM, Howell EA, Janevic T. The influence of racial-ethnic discrimination on women’s health care outcomes: A mixed methods systematic review. Soc Sci Med. 2023;316:114983. [DOI] [PubMed] [Google Scholar]
- 26.Loder C, Hall K, Kusunoki Y, Harris LH, Dalton V. Associations between perceived discrimination and contraceptive method use: why we need better measures of discrimination in reproductive healthcare. Women Health. 2021;61(5):461–9. [DOI] [PubMed] [Google Scholar]
- 27.MacDonald S, Hausmann LRM, Sileanu FE, Zhao X, Mor MK, Borrero S. Associations between perceived Race-based discrimination and contraceptive use among women veterans in the ECUUN study. Med Care. 2017;55(9 Suppl 2):S43–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Williams DR, Yan Y, Jackson JS, Anderson NB. Racial differences in physical and mental health: Socio-economic Status, stress and discrimination. J Health Psychol. 1997;2(3):335–51. [DOI] [PubMed] [Google Scholar]
- 29.Kossler K, Kuroki LM, Allsworth JE, Secura GM, Roehl KA, Peipert JF. Perceived racial, socioeconomic and gender discrimination and its impact on contraceptive choice. Contraception. 2011;84(3):273–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Krieger N, Smith K, Naishadham D, Hartman C, Barbeau EM. Experiences of discrimination: validity and reliability of a self-report measure for population health research on racism and health. Soc Sci Med. 2005;61(7):1576–96. [DOI] [PubMed] [Google Scholar]
- 31.Thorburn S, Bogart LM. African American women and family planning services: perceptions of discrimination. Women Health. 2005;42(1):23–39. [DOI] [PubMed] [Google Scholar]
- 32.Thorburn Bird S, Bogart LM. Birth control conspiracy Beliefs, perceived Discrimination, and contraception among African americans: an exploratory study. J Health Psychol. 2003;8(2):263–76. [DOI] [PubMed] [Google Scholar]
- 33.Bird ST, Bogart LM. Perceived race-based and socioeconomic status(SES)-based discrimination in interactions with health care providers. Ethn Dis. 2001;11(3):554–63. [PubMed] [Google Scholar]
- 34.Williams DR, Lawrence JA, Davis BA. Racism and health: evidence and needed research. Annu Rev Public Health. 2019;40:105–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Paradies Y, Ben J, Denson N, Elias A, Priest N, Pieterse A, et al. Racism as a determinant of health: A systematic review and Meta-Analysis. PLoS ONE. 2015;10(9):e0138511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Maier M, Samari G, Ostrowski J, Bencomo C, McGovern T. Scrambling to figure out what to do’: a mixed method analysis of COVID-19’s impact on sexual and reproductive health and rights in the united States. BMJ Sex Reprod Health. 2021;47(4):e16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Kavanaugh ML, Pleasure ZH, Pliskin E, Zolna M, MacFarlane K. Financial instability and delays in access to sexual and reproductive health care due to COVID-19. J Womens Health (Larchmt). 2022;31(4):469–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Shaver J. The state of telehealth before and after the COVID-19 pandemic. Prim Care. 2022;49(4):517–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Doraiswamy S, Abraham A, Mamtani R, Cheema S. Use of telehealth during the COVID-19 pandemic: scoping review. J Med Internet Res. 2020;22(12):e24087. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Lopez L 3rd, Hart LH 3rd, Katz MH. Racial and ethnic health disparities related to COVID-19. JAMA. 2021;325(8):719–20. [DOI] [PubMed] [Google Scholar]
- 41.Bossick AS, Brown J, Hanna A, Parrish C, Williams EC, Katon JG. Impact of State-Level reproductive health legislation on access to and use of reproductive health services and reproductive health outcomes: A systematic scoping review in the affordable care act era. Womens Health Issues. 2021;31(2):114–21. [DOI] [PubMed] [Google Scholar]
- 42.Reingold RB, Gostin LO. State abortion restrictions and the new supreme court: women’s access to reproductive health services. JAMA. 2019;322(1):21–2. [DOI] [PubMed] [Google Scholar]
- 43.Blake VK, McGowan ML. Filling a federal void: promises and perils of state law in addressing women’s health disparities. J Law Med Ethics. 2020;48(3):485–90. [DOI] [PubMed] [Google Scholar]
- 44.Hill L, Artiga S, Pillai A, Rao A. Elimination of Federal Diversity Initiatives: Implications for Racial Health Equity: Kaiser Family Foundation; 2025. Available from: https://www.kff.org/racial-equity-and-health-policy/issue-brief/elimination-of-federal-diversity-initiatives-implications-for-racial-health-equity/. Updated 21 Mar 2025.
- 45.The White House. Ending Radical and Wasteful Government DEI Programs and Preferencing. 2025.
- 46.Gomez AM, Bennett JA, Marshall AH. CJ. Person-Centered contraceptive access metrics project. Berkeley, CA: Sexual Health and Reproductive Equity Program, University of California, Berkeley; 2024. [Google Scholar]
- 47.NORC at the University of Chicago. Technical overview of the AmeriSpeak panel, NORC’s probability-based household panel. University of Chicago. 2022. 8 Feb 2022.
- 48.Browne MW, Cudeck R. Alternative ways of assessing model fit. Sociol Methods Res. 1992;21(2):230–58. [Google Scholar]
- 49.Flora DB. Your coefficient alpha is probably wrong, but which coefficient Omega is right? A tutorial on using R to obtain better reliability estimates. Adv Methods Practices Psychol Sci. 2020;3(4):484–501. [Google Scholar]
- 50.Rosenthal L, Lobel M. Stereotypes of black American women related to sexuality and motherhood. Psychol Women Q. 2016;40(3):414–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Bond KT, Leblanc NM, Williams P, Gabriel CA, Amutah-Onukagha NN. Race-Based sexual Stereotypes, gendered Racism, and sexual decision making among young black cisgender women. Health Educ Behav. 2021;48(3):295–305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Posit team. RStudio: integrated development environment for R. Boston, MA: Posit Software, PBC; 2023. [Google Scholar]
- 53.Kaiser HF, Rice J. Little jiffy, mark IV. Educ Psychol Meas. 1974;34(1):111–7. [Google Scholar]
- 54.Bartlett MS. Tests of significance in factor analysis. Br J Stat Psychol. 2011;3(2):77–85. [Google Scholar]
- 55.Lim S, Jahng S. Determining the number of factors using parallel analysis and its recent variants. Psychol Methods. 2019;24(4):452–67. [DOI] [PubMed] [Google Scholar]
- 56.Horn JL. A rationale and test for the number of factors in factor analysis. Psychometrika. 1965;30:179–85. [DOI] [PubMed] [Google Scholar]
- 57.Velicer WF, Jackson DN. Component analysis versus common factor analysis: some issues in selecting an appropriate procedure. Multivar Behav Res. 1990;25(1):1–28. [DOI] [PubMed] [Google Scholar]
- 58.Cattell RB. The scree test for the number of factors. Multivar Behav Res. 1966;1(2):245–76. [DOI] [PubMed] [Google Scholar]
- 59.Watkins MW. A step-by-step guide to exploratory factor analysis with R and Rstudio. New York, NY: Routledge New York, NY; 2021. Available from: https://www.taylorfrancis.com/books/9781003120001. Accessed March 8, 2024.
- 60.McDonald RP. Test theory: A unified treatment. psychology; 2013.
- 61.Rosenthal L, Lobel M. Gendered racism and the sexual and reproductive health of black and Latina women. Ethn Health. 2020;25(3):367–92. [DOI] [PubMed] [Google Scholar]
- 62.Harris LH, Wolfe T. Stratified reproduction, family planning care and the double edge of history. Curr Opin Obstet Gynecol. 2014;26(6):539–44. [DOI] [PubMed] [Google Scholar]
- 63.Prater C, Cohen L, Chau E, Carter EB, Kuebee B, Tepe M, et al. Perceived discrimination during prenatal care at a community health center. J Racial Ethn Health Disparities. 2023;10(3):1304–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.Attanasio LB, Hardeman RR. Declined care and discrimination during the childbirth hospitalization. Soc Sci Med. 2019;232:270–7. [DOI] [PubMed] [Google Scholar]
- 65.Gomez AM, Bennett AH, Schulte A, Arcara J, Stern L, Aina AD, et al. Use of preferred source of contraception among users of the Pill, Patch, and ring in the US. JAMA Netw Open. 2024;7(10):e2439191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Treder K, White KO, Woodhams E, Pancholi R, Yinusa-Nyahkoon L. Racism and the reproductive health experiences of U.S.-Born black women. Obstet Gynecol. 2022;139(3):407–16. [DOI] [PubMed] [Google Scholar]
- 67.Thompson TM, Young YY, Bass TM, Baker S, Njoku O, Norwood J, et al. Racism runs through it: examining the sexual and reproductive health experience of black women in the South. health affairs. (Project Hope). 2022;41(2):195–202. [DOI] [PubMed] [Google Scholar]
- 68.Danaux J, Malcolm N, Wilson J, Parikh A, Redd SK, Rice WS, et al. Statewide contraceptive access initiatives: recommendations report. Coalition to Expand Contraceptive Access; 2023.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets analyzed during the current study are available on the Open Science Framework (OSF) at https://osf.io/wvb5m/.