ÌÇÐÄvlog

Key Points

QuestionÌý How representative are obstetrics and gynecology (OB-GYN) trials of the US population and how does representation vary by subspecialty and funder?

FindingsÌý In this cross-sectional study of 1300 US OB-GYN clinical trials with results from ClinicalTrials.gov and 1147 clinical trial publications (2007-2020), 51% of trials and 75% of publications reported race and ethnicity data. American Indian or Alaskan Native, Asian, Black, and Latinx groups were underrepresented in trials and publications, but underrepresentation varied by race and ethnicity, subspecialty, and funder; obstetric and family planning trials had the most diversity of the subspecialties.

MeaningÌý Racial and ethnic representation in OB-GYN trials is suboptimal; all subspecialties should strive for more equitably representative research.

ImportanceÌý Clinical trials guide evidence-based obstetrics and gynecology (OB-GYN) but often enroll nonrepresentative participants.

ObjectiveÌý To characterize race and ethnicity reporting and representation in US OB-GYN clinical trials and their subsequent publications and to analyze the association of subspecialty and funding with diverse representation.

Design and SettingÌý Cross-sectional analysis of all OB-GYN studies registered on ClinicalTrials.gov (2007-2020) and publications from PubMed and Google Scholar (2007-2021). Analyses included logistic regression controlling for year, subspecialty, phase, funding, and site number. Data from 332 417 studies were downloaded. Studies with a noninterventional design, with a registration date before October 1, 2007, without relevance to OB-GYN, with no reported results, and with no US-based study site were excluded.

ExposuresÌý OB-GYN subspecialty and funder.

Main Outcomes and MeasuresÌý Reporting of race and ethnicity data and racial and ethnic representation (the proportion of enrollees of American Indian or Alaskan Native, Asian, Black, Latinx, or White identity and odds of representation above US Census estimates by race and ethnicity).

ResultsÌý Among trials with ClinicalTrials.gov results (1287 trials with 591 196 participants) and publications (1147 trials with 821 111 participants), 662 (50.9%) and 856 (74.6%) reported race and ethnicity data, respectively. Among publications, gynecology studies were significantly less likely to report race and ethnicity than obstetrics (adjusted odds ratio [aOR], 0.54; 95% CI, 0.38-0.75). Reproductive endocrinology and infertility trials had the lowest odds of reporting race and ethnicity (aOR, 0.14; 95% CI, 0.07-0.27; reference category, obstetrics). Obstetrics and family planning demonstrated the most diverse clinical trial cohorts. Compared with obstetric trials, gynecologic oncology had the lowest odds of Black representation (ClinicalTrials.gov: aOR, 0.04; 95% CI, 0.02-0.09; publications: aOR, 0.06; 95% CI, 0.03-0.11) and Latinx representation (ClinicalTrials.gov: aOR, 0.05; 95% CI, 0.02-0.14; publications: aOR, 0.23; 95% CI, 0.10-0.48), followed by urogynecology and reproductive endocrinology and infertility. Urogynecology (ClinicalTrials.gov: aOR, 0.15; 95% CI, 0.05-0.39; publications: aOR, 0.24; 95% CI, 0.09-0.58) had the lowest odds of Asian representation.

Conclusions and RelevanceÌý Race and ethnicity reporting and representation in OB-GYN trials are suboptimal. Obstetrics and family planning trials demonstrate improved representation is achievable. Nonetheless, all subspecialties should strive for more equitably representative research.

Clinical trials have transformed the field of obstetrics and gynecology (OB-GYN) and underlie the highest-quality evidence behind interventions.^1,2 Although trial results inform policies affecting health for the entire US population, trials often feature nonrepresentative cohorts.^3,4 Skewed participant demographics create a racial data gap; the individuals most affected by health disparities may be inadequately considered in solutions.⁵ When research systematically overlooks marginalized populations, it fails to account for the impact of social determinants of health that often fall along racial lines in the US due to the role of structural and interpersonal racism.⁶ Their omission from clinical trials excludes them from the individual benefits realized through clinical trial participation.^5,6 Furthermore, research disparities advance a biased framework that informs therapies and becomes systemized in national practice recommendations with unproven efficacy across the population.^7,8 The resulting disparities raise particular concern in OB-GYN where increased morbidity and mortality among Black and Latinx individuals are well established.⁹ Obstetric and gynecologic conditions often affect other aspects of female health (eg, cardiovascular) throughout their life course. The differential health outcomes underscore the importance of diversity in clinical trials and the potential harm in neglecting representative populations.

In response to racial health inequities in OB-GYN and across medical fields, physicians, government officials, and health advocates have collaborated to pass legislation and energize inclusivity efforts in clinical trials.^10,11 Despite these initiatives, foundational steps, such as fully quantifying disparities across OB-GYN clinical trials, have yet to be completed. Previous studies on racial and ethnic reporting and representation in OB-GYN clinical trials include only small numbers of trials and publications, specific disease foci, or narrow time frames.^2,8,12-18

Given the importance of characterizing the current state of diversity in OB-GYN trials, we aimed to characterize race and ethnicity reporting and representation among all OB-GYN trials and to elucidate the association of OB-GYN subspecialty and funding with race and ethnicity reporting and representation.

Data sources included the ClinicalTrials.gov registry,¹⁹ PubMed,²⁰ and Google Scholar²¹ publication databases in addition to the 2010 Census.²² The 2010 Census was selected because 2010 falls near the midpoint of the study. Additionally, 2010 provides a more conservative estimate of the outcomes compared with the 2020 census given the documented increases in diversity between 2010 and 2020.²² The study included only publicly available data and was exempt from oversight by the Stanford Institutional Review Board. The study follows Strengthening the Reporting of Observational Studies in Epidemiology () reporting guideline.

Our previous analyses demonstrated most clinical trials that disseminate findings either report results to ClinicalTrials.gov or publish in a peer-reviewed journal.²³ Thus, we included data from both sources in separate parallel analyses.

We downloaded all studies registered on ClinicalTrials.gov prior to March 9, 2020, from the Aggregate Analysis of ClinicalTrials.gov database.¹⁹ Studies registered prior to October 1, 2007, without interventional study design or without relevance to OB-GYN were excluded. The date was selected based on the passage of the US Food and Drug Administration Amendments Act of 2007.²⁴ ClinicalTrials.gov has evolved several iterations of race and ethnicity reporting requirements (eAppendix 1 in the Supplement). We included trials from before and after the implementation of the Final Rule for ClinicalTrials.gov after our previous sensitivity analysis accounting for the Final Rule demonstrated similar results with respect to race and ethnicity reporting and representation in the years proceeding and after the implementation of the Final Rule (eAppendix 2 in the Supplement).⁴ The methods to select and categorize OB-GYN trials by subspecialty were described previously.^23,25 All trials were linked to PubMed’s MEDLINE bibliographic database with unique trial identifying numbers from ClinicalTrials.gov to extract publications published before October 2021.²⁰ Completed trials without publications were manually reviewed. Publications were only included if they were original studies with primary results. Qualitative studies, secondary analyses, meta-analyses, and protocols were excluded.

We applied US Census and Department of Health and Human Services guidelines for racial and ethnic categories (eAppendix 3 in the Supplement).⁴ We focused on the most common racial and ethnic groups in the US population: American Indian or Alaskan Native, Asian (including Pacific Islander and Native Hawaiian), Black, Hispanic/Latina (hereafter, Latinx), and White. Clinical trials presented Latinx as both a race and an ethnicity. Our previous sensitivity analysis found no difference in outcomes between analyzing it as a race or an ethnicity.⁴ Most Latinx individuals in the US identify their Latinx heritage as part of their racial identity.²⁶ Therefore, we included Latinx as a race and ethnicity category.

We extracted all race and ethnicity participant data from the Aggregate Analysis of ClinicalTrials.gov database and from publications through both manual and programmatic searching. Race and ethnicity data were then recategorized (eg, Mexican American recategorized as Latinx) to align with US Census and Department of Health and Human Services guidelines.²²

Our primary exposure variables were specialty including the broad categories obstetrics and gynecology (including all gynecologic subspecialties) as well as mutually exclusive subspecialties family planning, gynecologic-oncology, obstetrics, other benign gynecology, reproductive endocrinology and infertility, and urogynecology.

Our other primary exposure was funding including academic/government funding and industry funding. We classified funding in accordance with previous studies (eAppendix 4 in the Supplement).²⁷ Due to the overlapping nature of government and academic funding in OB-GYN, we combined academic and government funding. As secondary exposures, we included 10 other trial features (eAppendix 5 in the Supplement).

We included 2 primary outcomes: (1) reporting: the reporting of race and ethnicity data and (2) representation: the proportion of participants in a clinical trial belonging to a specific racial or ethnic group. For logistic regression analysis, representation was calculated as the odds of median trial representation exceeding representation of that racial and ethnic group in the 2010 US Census.

Trial characteristics for reporting and representation were described and compared using Pearson χ² tests. We conducted multivariable logistic regression analysis. In analyses of representation, we only included trials with reported race and ethnicity information. All logistic regression analyses were adjusted for year, trial phase, and number of study sites as well as funding or subspecialty when not the exposure variable. Multivariable logistic regression analysis could not be computed for American Indian or Alaskan Native representation due to a small sample size. We conducted inductive hypothesis-generating analysis of the 10 other trial features. The analyses were 2 sided with statistical significance at α = .05 level. We used R version 3.5.2 (R Foundation) for all analysis. In adjusted regression analysis, we managed missing data with multiple imputation by chained equations (eAppendix 6 in the Supplement). Our sample did not contain missing continuous data.

We extracted 332 417 studies registered before March 9, 2020. Only 11 032 studies were clinical trials focused on OB-GYN registered after October 1, 2007. Nearly 1300 of these were US-based trials (1287 trials, approximately 591 196 participants) with reported results on ClinicalTrials.gov and 1147 (approximately 821 111 participants) were publications of clinical trials (Figure 1).

A majority of trials and publications (ClinicalTrials.gov: 655 [50.9%]; publications: 856 [74.6%]) reported race and ethnicity data. ClinicalTrials.gov race and ethnicity reporting increased over time (129 of 426 trials [30.2%] in the first 3 years of the study [2007-2009] vs and 88 of 98 trials [89.8%] in the last 3 complete years of the study [2017-2019]; Figure 2 and eTable 1 in the Supplement). A similar pattern was not observed in publications.

Among ClinicalTrials.gov trials with race and ethnicity reporting, Black and White race were reported the most frequently (607 of 655 [92.7%] and 626 of 655 [95.6%], respectively), and American Indian or Alaskan Native (455 of 655 [69.5%]), Asian (543 of 655 [82.9%]), and Latinx (410 of 655 [62.6%]) were reported less frequently (Figure 2 and eTable 2 in the Supplement).

Among subspecialties, reproductive endocrinology and infertility reported race and ethnicity the least (ClinicalTrials.gov: 18 of 43 [41.9%]; publications: 19 of 48 [39.6%]) and obstetrics the most (ClinicalTrials.gov: 143 of 264 [54.2%]; publications: 289 of 363 [79.6%]; eTable 3 in the Supplement).

In adjusted multivariable analysis, findings differed by source (Table). ClinicalTrials.gov results demonstrated no significant association between subspecialty and race and ethnicity reporting. However, among publications, gynecology studies were significantly less likely to report race and ethnicity than obstetrics (adjusted odds ratio [aOR], 0.54; 95% CI, 0.38-0.75). Variation was seen within subspecialties; reproductive endocrinology and infertility had the lowest odds of reporting race and ethnicity in publications followed by urogynecology, gynecologic oncology, and general benign gynecology (reference category, obstetrics).

No significant difference was seen between funders in reporting on ClinicalTrials.gov or in publications in multivariable analysis (Table and eTable 3 in the Supplement).

In both ClinicalTrials.gov results and publications, multi-arm trials, randomized trials, and placebo-controlled trials reported race and ethnicity more frequently than their respective reference categories (Table).

Among OB-GYN trials with reported race and ethnicity data, American Indian or Alaskan Native, Asian, Black, and Latinx populations were underrepresented compared with US Census estimates in both ClinicalTrials.gov and publication results except American Indian or Alaskan Native and Black participants in publications (eTable 4 in the Supplement and Figure 3). White individuals comprised the majority of participants followed by Black, Latinx, Asian, and American Indian or Alaskan Native participants.

In both ClinicalTrials.gov results and publications, more American Indian or Alaskan Native, Asian, Black, and Latinx participants enrolled in obstetrics trials than any other subspecialty (Figure 3 and eTable 4 in the Supplement). In obstetrics, the median enrollment percent for Black individuals was 10 percentage points higher than census estimates in both ClinicalTrials.gov and publication results, but all other race and ethnicity enrollment percentages were below census estimates. After obstetrics, family planning demonstrated the most diverse representation (median [IQR]: American Indian or Alaskan Native, 0.3% [0%-1.9%]; Asian, 4.0% [0%-12.5%]; Black, 21.3% [0%-55.9%]; Latinx, 18.2% [0%-43.7%]; White, 54.6% [15.0%-94.2%]). Gynecologic oncology had the least Black and Latinx representation of all subspecialties. Urogynecology and other benign gynecology enrolled the smallest proportion of Asian participants.

In multivariable analysis, all gynecology trials were more likely to have above census-level representation of White participants than obstetrics (Figure 4; eTables 5 and 6 in the Supplement).

Gynecologic oncology demonstrated the lowest odds of representation of Black individuals meeting census population levels (ClinicalTrials.gov: aOR, 0.04; 95% CI, 0.02-0.09; publications: aOR, 0.06; 95% CI, 0.00-0.13), followed by urogynecology and reproductive endocrinology and infertility trials. Representation of Latinx individuals followed a similar pattern; gynecologic oncology had the lowest odds of Latinx representation (ClinicalTrials.gov: aOR, 0.05; 95% CI, 0.02-0.14; publications: aOR, 0.23; 95% CI, 0.10-0.48). Urogynecology, reproductive endocrinology and infertility, and other benign gynecology also demonstrated lower odds of Latinx representation compared with obstetrics. Urogynecology had the lowest odds of representation of Asian individuals (Clinicaltrials.gov: aOR, 0.15; 95% CI, 0.05-0.39; publications: aOR, 0.24; 95% CI, 0.09-0.58).

Academic/government trials (median [IQR]: ClinicalTrials.gov, 29.4% [0%-89.8%]; publications, 40.9% [0%-89.8%]) enrolled more American Indian or Alaskan Native, Asian, Black, and Latinx participants than industry trials (ClinicalTrials.gov: median [IQR], 15.1% [0%-36.4%]; publications: median [IQR], 23.7% [0%-54%]; eTable 4 in the Supplement). In both cohorts, academic/government trials enrolled more Black participants, but no other race and ethnicity trends were consistent across ClinicalTrials.gov and publications.

In multivariable analysis, industry demonstrated lower odds of above census levels of American Indian or Alaskan Native, Asian, Black, and Latinx representation than academic/government trials in ClinicalTrials.gov data but not in publications (ClinicalTrials.gov: aOR, 0.44; 95% CI, 0.26-0.74; publications: aOR, 0.76; 95% CI, 0.46-1.26; eTables 5 and 6 in the Supplement).

In multivariable analysis of other trial features, year was not associated with representation in ClinicalTrials.gov data. In contrast, for publications, later years were associated with lower odds of American Indian or Alaskan Native, Asian, Black, and Latinx representation (2013-2019: aOR, 0.68; 95% CI, 0.48-0.98; reference category, 2007-2012).

In this analysis of all US-based OB-GYN clinical trials and publications between 2007 and 2020, we identify suboptimal reporting of race and ethnicity in ClinicalTrials.gov results and publications as well as deficient representation of American Indian or Alaskan Native, Asian, Black, and Latinx populations in trials. Patterns of underrepresentation varied by race and ethnicity, subspecialty, and funder. Gynecologic oncology enrolled the least diverse participants, with the lowest odds of representation of Black and Latinx individuals of all gynecologic subspecialties. Urogynecology featured the lowest proportion of Asian participants. All gynecology subspecialities except family planning lagged behind obstetrics in representative enrollment. These findings demonstrate the need for improved resources and inclusive methods to recruit and support racialized minorities in clinical trial research.

This examination of US-based OB-GYN clinical trials uncovers new information about race and ethnicity reporting in research. Among over 1250 ClinicalTrials.gov entries and 1100 publications, less than 55% and 75% of trials report race and ethnicity data on ClinicalTrials.gov and in publications, respectively. Not all subspecialties report race and ethnicity equally. Among publications, gynecology trials, and in particular reproductive endocrinology and infertility trials, had the lowest odds reporting race and ethnicity compared with obstetrics trials. Our findings corroborate and expand previous results by covering a larger cohort and including comparisons among subspecialties, funders, and other trial features.^2,8,12-18 We found similar^12,13 or higher rates of race and ethnicity reporting than prior work in OB-GYN with more limited samples.¹⁴ Compared with ClinicalTrials.gov database entries including all medical specialties, OB-GYN trials reported race and ethnicity more frequently (50.9% vs 43.0%).⁴ The methods from this study could be applied to investigate race and ethnicity reporting in other fields. Although OB-GYN trials perform better than clinical trials in other subspecialties, our findings indicate significant work remains toward systematically documenting the demographic characteristics of OB-GYN research participants.

Federal law requires the majority of US-based clinical trials to register with ClinicalTrials.gov,²⁸ and academicians rely on PubMed as a premier source for peer-reviewed publications. Together, these sources provide a comprehensive picture of the US clinical trial ecosystem. When nearly half of studies registered on ClinicalTrials.gov and a quarter of publications omit race and ethnicity information, we cannot reliably track clinical trial diversity over time in the setting of incomplete data. Notably, ClinicalTrials.gov and publications results failed to include data on American Indian Alaska Native, Asian, and Latinx participants more than Black and White participation data. The erasure of race and ethnicity from clinical trials data typifies the colorblind approach previously lauded in research and medicine, an approach that further stigmatizes racialized minorities by discounting the impact of bias in society and medicine.²⁹ We propose OB-GYN leadership should require a standardized, antiracist approach to race and ethnicity reporting for all studies. This approach should account for individuals of multiple races and for the nuances of varied ethnic groups within racial categories. Journals and academic societies could collaborate to instruct and regulate trialists as they implement a specialty-wide system for measuring, recording, and studying demographics. The consistent reporting of race and ethnicity represents an achievable goal and a prerequisite to improve equity in research in OB-GYN and across specialties.

Our study establishes that American Indian or Alaskan Native, Asian, Black, and Latinx racial and ethnic groups remain underrepresented in OB-GYN without demonstrated improvement over time.^10,11 Previous OB-GYN studies also found underenrollment of American Indian or Alaskan Native, Asian, Black, and Latinx groups compared with their representation in the population.^2,8,12-18 Although we demonstrate that racial and ethnic disparities remain in OB-GYN, OB-GYN trials perform better with respect to equitable representation than the overall ClinicalTrials.gov database and other medical specialties.⁴ For example, in our publication cohort, Black participants were enrolled in greater proportion than their representation in the US population, largely driven by obstetrics and family planning trials. Previous studies detailed the lack and worsening of diversity in gynecologic oncology trials.^6,16,17,30 We add that gynecologic oncology trials perform the worst of all subspecialties with respect to Black and Latinx representation. Several studies have investigated the etiology of poor trial diversity in gynecologic oncology and attribute the homogeneity to historic, socioeconomic, structural, and cultural factors.³¹ The poor performance of oncology compared with other subspecialties exacerbates the disproportionate cancer morbidity and mortality experienced by Black and Latinx communities and exemplifies the harm of inadequate representation.⁶

The contrast between OB-GYN trials provides evidence for an important reframing of barriers to diverse enrollment. Previous studies across specialties ascribe low enrollment of American Indian or Alaskan Native, Asian, Black, and Latinx participants to complex root causes.^31,32 Yet, both obstetrics and family planning demonstrate that more diverse enrollment, while still far from optimal, is achievable. These differences are likely multifactorial. Research infrastructure may differ between these subspecialties due to differences in insurance coverage, academic hospital–community collaborations, local demographics of study institutions, and cultural exclusivity.³¹ For example in the US, all pregnant patients have access to insurance through Medicaid. This likely improves access to obstetric clinical trials for individuals from marginalized communities. In contrast, reproductive endocrinology and infertility services accept limited insurance and clinical trial participants may not have access to any reproductive endocrinology and infertility care or may have to pay for their involvement. Each of these barriers merit investigation and strategic mobilization of resources. Researchers often cite mistrust of the medical system as a contributor to underenrollment of American Indian or Alaskan Native, Asian, Black, and Latinx individuals, which inherently lays responsibility on participants.³¹ Our findings, in alignment with recent research disputing this theory,³² suggest it is feasible to overcome this barrier and that it is essential to redistribute the onus of improved research equity off of participants and onto investigators. Future investigations should evaluate the impact of study location, institutional characteristics, and researcher demographics on trial representation as well as account for intersectional participant demographic characteristics like socioeconomic status, age, and sex identity. With increased resources, collaboration, a race-conscious approach, and obstetrics and family planning to serve as models, all subspecialties in OB-GYN could endeavor to redesign research systems to reach and benefit all US communities.

Our research has limitations. It is beyond the scope of the study to investigate the problematic nature of singular racial and ethnic categories that represent many heterogenous groups (eg, the Asian category representing several distinct communities). Reporting of Latinx was inconsistent across ClinicalTrials.gov and publications results with certain trials reporting Latinx as a race and others as an ethnicity. We previously conducted a sensitivity analysis and found similar results across models.⁴ In addition, our findings cannot be generalized to trials that did not report race and ethnicity data or to trials that did not register with ClinicalTrials.gov. Previous studies have documented several limitations of ClinicalTrials.gov that apply to the ClinicalTrials.gov cohort in our study.³³

In the setting of escalating racial and ethnic health inequities, the OB-GYN community must develop mechanisms for monitoring and improving inclusion of underrepresented groups in clinical trials. Our biomedical apparatus cannot improve what we do not measure. Not only do American Indian or Alaskan Native, Asian, Black, and Latinx individuals comprise a greater proportion of those eligible to participate in trials due to their disproportionate disease burden, but they also stand to benefit the most from the improved health outcomes and cutting edge treatments accessed by study participants. Obstetrics and family planning demonstrate the potential for improved representation and should be considered as models for the field. Stakeholders must prioritize transparency, accountability, diversity, and equity to transform research infrastructure and adequately serve marginalized communities. Until we mend the racial disparities of clinical trials, research will continue contributing to unjust health outcomes borne by these communities.

Accepted for Publication: September 4, 2022.

Published Online: December 21, 2022. doi:10.1001/jamasurg.2022.6600

Corresponding Author: Jecca R. Steinberg, MD, MSc, Department of Obstetrics and Gynecology, Northwestern Feinberg School of Medicine, 250 E Superior St, Chicago, IL 60611 (jecca.steinberg@northwestern.edu).

Correction: This article was corrected on February 8, 2023, to fix errors in the author affiliations.

Author Contributions: Drs Steinberg and Yee had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Steinberg, Turner, DiTosto, Weeks, Magnani, Anderson, Roque, Yee.

Acquisition, analysis, or interpretation of data: Steinberg, Turner, DiTosto, Weeks, Young, Lu, Wolgemuth, Holder, Laasiri, Squires, Zhang, Richardson, Magnani, Anderson, Yee.

Drafting of the manuscript: Steinberg, DiTosto, Weeks, Young, Laasiri, Roque.

Critical revision of the manuscript for important intellectual content: Steinberg, Turner, DiTosto, Weeks, Young, Lu, Wolgemuth, Holder, Squires, Zhang, Richardson, Magnani, Anderson, Roque, Yee.

Statistical analysis: Steinberg, Turner, DiTosto, Wolgemuth, Holder, Laasiri, Magnani.

Administrative, technical, or material support: Steinberg, Turner, Weeks, Young, Lu, Wolgemuth, Laasiri, Squires, Zhang, Richardson.

Supervision: Weeks, Young, Richardson, Roque, Yee.

Conflict of Interest Disclosures: Dr Roque reported grants from Bristol Myers Squibb Foundation Robert A. Winn Diversity in Clinical Trials Award Program and the National Cancer Institute (grant P20 CA233304-01A1) during the conduct of the study and personal fees from GSK, Myriad Genetics, and ImmunoGen outside the submitted work. Dr Yee reported grants from the National Institutes of Health outside the submitted work. No other disclosures were reported.