Reporting of Participant Race and Ethnicity in Published US Pediatric Clinical Trials From 2011 to 2020

Key Points

Question  How did participant race and ethnicity in pediatric clinical trials published from 2011 to 2020 compare with the corresponding US populations?

Findings  In this cross-sectional study of 612 articles published from 2011 to 2020 in leading pediatric and general medical journals with 565 618 total participants, Black/African American children were enrolled proportionally more than the US population of Black/African American children; Hispanic/Latino children were enrolled commensurately with the population of Hispanic/Latino children; and American Indian/Alaska Native, Asian, and Native American/Pacific Islander children were enrolled less than the respective US populations of these groups. White children were enrolled less than expected but represented 46.0% of participants.

Meaning  The findings suggest that disparities exist in pediatric clinical trial enrollment of Black/African American, American Indian/Alaska Native, Asian, and Native American/Pacific Islander pediatric populations in the US.

Importance  Equitable representation of participants who are members of racial and ethnic minority groups in clinical trials enhances inclusivity in the scientific process and generalizability of results.

Objective  To assess participant race and ethnicity in pediatric clinical trials published from 2011 to 2020.

Design, Setting, and Participants  This cross-sectional study examined articles reporting pediatric clinical trials conducted in the US published in 5 leading general pediatric and 5 leading general medical journals from January 1, 2011, to December 31, 2020.

Main Outcomes and Measures  Reporting of participant race and ethnicity and comparison of enrolled participants vs US census populations of pediatric racial and ethnic groups in published clinical trials.

Results  The study included 612 articles reporting pediatric clinical trials during the study period, with 565 618 total participants (median per trial, 200 participants [IQR, 90-571 participants]). Of the 612 articles, 486 (79.4%) reported participant race and 338 (55.2%) reported participant ethnicity. From 2011 to 2020, relative rates of reporting of participant race increased by 7.9% per year (95% CI, 0.2%-16.3% per year) and reporting of ethnicity increased by 11.4% per year (95% CI, 4.8%-18.4% per year). Among articles reporting race and ethnicity, the method of assignment was not reported in 261 of 511 articles (51.1%) and 207 of 359 articles (57.7%), respectively. Black/African American children were enrolled proportionally more than the US population of Black/African American children (odds ratio [OR], 1.88; 95% CI, 1.87-1.89). Hispanic/Latino children were enrolled commensurately with the US population of Hispanic/Latino children (OR, 1.02; 95% CI, 1.01-1.03). American Indian/Alaska Native (OR, 0.82; 95% CI, 0.79-0.85), Asian (OR, 0.56; 95% CI, 0.55-0.57), and Native Hawaiian/Pacific Islander (OR, 0.66; 95% CI, 0.61-0.72) children were enrolled significantly less compared with the respective US populations of these groups. White children were enrolled less than expected (OR, 0.84; 95% CI, 0.84-0.85) but represented 188 156 (46.0%) of participants in trials reporting race or ethnicity.

Conclusions and Relevance  This cross-sectional study revealed that the proportion of published pediatric clinical trials that reported participant race and ethnicity increased from 2011 to 2020, but participant race and ethnicity were still underreported. Disparities existed in pediatric clinical trial enrollment of American Indian/Alaska Native, Asian, and Native Hawaiian/Pacific Islander children. The greater representation of Black/African American children compared with the US population suggests inclusive research practices that could be extended to other historically disenfranchised racial and ethnic groups.

Clinical trials provide the highest level of evidence for assessment of the safety, comparative effectiveness, and efficacy of medications and interventions that guide clinical care.¹ However, results from clinical trials that lack racial and ethnic minority participants may not be generalizable to all populations that may benefit from trial results.^2,3 Race is a social construct reflecting the “impact of unequal social experiences on health,”^4,p872 and ethnicity refers to social groupings based on culture and language.^4,5 The National Institutes of Health and the US Food and Drug Administration mandate the planned inclusion of participants from multiple racial and ethnic groups when applying for support.^6,7

Nevertheless, studies have shown that participants’ race and ethnicity are often not reported in clinical trial results.^8-11 Studies evaluating enrollment in pediatric clinical trials for hematologic malignancies have demonstrated that Black/African American participants^12,13 and Hispanic/Latino participants^14,15 have been underrepresented. In a 2003 study, Walsh and Ross¹⁶ reviewed articles published in 3 general pediatric journals during a single year and demonstrated that Black participants were included more than expected and White participants and Hispanic/Latino participants were included less than expected in clinical trials; Black/African American and Hispanic children were underrepresented in therapeutic research.¹⁶

Understanding trial enrollment patterns by race and ethnicity is necessary to ensure that pediatric clinical trials address, rather than exacerbate, health care inequities. Characterization of participant race and ethnicity in trial enrollment may inform future recruitment strategies and requirements. Our objective was to assess participant race and ethnicity reported in published pediatric clinical trials and to evaluate rates and trends in race and ethnicity published in general pediatric and medical journals. We hypothesized that children in racial and ethnic minority groups would be underrepresented in published clinical trials.

We conducted a cross-sectional study of articles reporting results of pediatric clinical trials published in 5 leading general pediatric journals and 5 leading general medical journals from January 1, 2011, to December 31, 2020. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. The Boston Children’s Hospital institutional review board exempted this study from review because all data were publicly available. No participant consent was needed for this analysis because all data were obtained from published articles.

We reviewed articles reporting pediatric clinical trials in JAMA Pediatrics, The Lancet Child & Adolescent Health, Pediatrics, the Journal of Adolescent Health, and the Journal of Pediatrics. We also reviewed articles reporting pediatric clinical trials in the New England Journal of Medicine, The Lancet, JAMA, BMJ, and PLoS Medicine. These were the leading general pediatric and general medical journals by impact factor.¹⁷

To compare trial enrollment by race and ethnicity with the US population, we used the 2019 annual estimate of population sizes from the US census at the national, state, and county levels for children and adolescents aged 0 to 18 years (aged 0 to 19 years at the county level).¹⁸ We defined clinical trials according to the National Institutes of Health definition.¹⁹

We included articles reporting results of clinical trials that exclusively enrolled participants 0 to 18 years of age or those in which the median or mean age of participants was less than or equal to 18 years. We excluded articles that did not report trial results, trials conducted outside the US, and secondary analyses of trials if the original trial was included.

We queried PubMed to identify all published articles in the selected journals. Then using key search terms, we extracted articles likely to report pediatric clinical trials (eAppendix in the Supplement). After an initial training period, 1 of us (C.A.R., A.M.S., S.M., E.A., J.J., J.M., E.N.P., or E.W.F.) reviewed the abstract and full text of each article. We extracted the number of enrolled participants, participant age group or groups, race, ethnicity, how race and ethnicity were ascertained, whether participants’ primary language was reported, community stakeholders’ involvement, and participants’ reported socioeconomic data.²⁰ We reviewed the instructions for authors of each journal to ascertain whether policies regarding the reporting of participant race and ethnicity existed. Data were collected and managed using the secure REDCap data capture platform.²¹ Any question about inclusion of an article or specific variables was discussed among us to achieve consensus.

Additional trial characteristics extracted included randomization, masking, trial phase, intervention type, disease or diseases studied, listed funding sources, and trial locations. All variables were extracted from the primary article, supplement information, or ClinicalTrials.gov. For articles without participants’ race and ethnicity, we sent a standardized email to corresponding authors to solicit this information up to 2 times.

Our primary outcome was the proportion of children enrolled in pediatric clinical trials by race and ethnicity compared with the US population. Race was classified according to the Standards for the Classification of Federal Data on Race as follows: American Indian/Alaska Native, Asian, Black/African American, Native Hawaiian/Pacific Islander, and White.²² We recorded the number of participants who were documented as “other” race but excluded “other” from our analyses given the heterogeneity of this categorization. Ethnicity was categorized as Hispanic/Latino and treated as a separate category because of inconsistent reporting of participant race and ethnicity (ie, some trials listed Hispanic/Latino as an ethnicity [n = 217], and others listed it as a race [n = 168]). We recorded how participant race and ethnicity were ascertained in each trial.

We calculated descriptive statistics of trial characteristics and participant race and ethnicity. We calculated odds ratios (ORs) directly (not from models) using the cross-product ratio of the proportion of trial participants in each racial or ethnic group to the expected proportion (ie, children aged 0 to 18 years in the US census in each racial or ethnic group) with 95% CIs. Because most pediatric clinical trials were conducted in academic medical centers in urban areas, we additionally conducted a sensitivity analysis using counties in which trials were conducted as the census comparator. We conducted a subanalysis excluding trials that targeted specific racial or ethnic groups. To understand how reporting changed over time, we created a logistic regression model at the level of individual studies, with race reporting as the dependent variable and year as a linear term (ie, any value between 2011 and 2020) as the independent variable. We calculated the proportion of enrolled children by each racial and ethnic group by year and the mean diversity index of all trials by year using Poisson regression, with an offset for each trial’s log total sample size. Diversity was measured using the diversity index, ranging from 0 (no diversity) to 1 (equal representation of all possible groups).^23-25 The diversity index was calculated using the formula D = 1 − [(∑ n × [n − 1])/(N × [N − 1])], where n is the number of children for each racial or ethnic group and N is the total number of children in all groups.

For trials conducted in a single state, we conducted an exploratory analysis using logistic regression to test bivariable associations between the outcome of high diversity (defined as trial diversity greater than or equal to that of the state) and covariates theoretically associated with high diversity. All associations with an overall 2-sided P < .20 were included in a multivariable model to assess characteristics associated with participant enrollment with high diversity. We mapped the diversity index of pediatric clinical trials by state and compared them with the respective state’s diversity index. All analyses were conducted using R, version 4.0.3 (R Foundation for Statistical Computing).

Among 99 866 articles, 3782 were potentially related to pediatric clinical trials, and 612 reported results of pediatric clinical trials conducted in the US (eFigure 1 in the Supplement). Of these 612 articles, 574 (93.8%) were randomized, 159 (26.0%) were nonblinded, and 153 (25.0%) were open-label (Table 1). There were 565 618 total participants (median per trial, 200 participants [IQR, 90-571 participants]). Behavioral interventions were most common (244 [39.9%]) (Table 1). A total of 193 197 participants (34.2%) were adolescents (Table 2).

Of the 612 articles, 486 (79.4%) reported participant race and 338 (55.2%) reported participant ethnicity. Of the 126 and 274 articles that did not report race or ethnicity, respectively, corresponding authors provided race data for 25 (19.8%) and ethnicity data for 21 (7.7%). Of the 511 articles with participant race data available, 261 (51.1%) did not report how it was measured, 231 (45.2%) measured race through participant or caregiver report, 16 (3.1%) obtained race from electronic health records, and 3 (0.6%) reported that trial staff assigned participants’ race. Of the 359 articles with participant ethnicity data available, 207 (57.7%) did not report how it was measured, 139 (38.7%) measured ethnicity by participant or caregiver report, 9 (2.5%) obtained participant ethnicity from electronic health records, and 4 (1.1%) reported that trial staff assigned participant ethnicity. A total of 156 898 participants (27.7%) had no race or ethnicity reported (Table 2).

There was a relative increase in reporting of participant race (7.9% per year; 95% CI, 0.2%-16.3% per year) and ethnicity (11.4% per year; 95% CI, 4.8%-18.4% per year) from 2011 to 2020 (Figure 1A). In 2020, there were 59 published pediatric trials, of which 49 (83.1%) reported participant race and 43 (72.9%) reported participant ethnicity. There was a decrease over time in the proportion of enrolled American Indian/Alaska Native (incidence rate ratio [IRR], 0.84; 95% CI, 0.83-0.85), Black/African American (IRR, 0.89; 95% CI, 0.89-0.89), Native Hawaiian/Pacific Islander (IRR, 0.81; 95% CI, 0.79-0.83), and White children (IRR, 0.97; 95% CI, 0.97-0.97). The proportion of enrolled Asian children increased (IRR, 1.10; 95% CI, 1.01-1.11), and the proportion of Hispanic/Latino children was similar over time (IRR, 1.0; 95% CI, 1.00-1.01). The mean annual diversity index of pediatric clinical trials did not change during the study period (Figure 1B).

In comparison with national US census data, Black/African American children were enrolled 88% more than expected (OR, 1.88; 95% CI, 1.87-1.89), and Hispanic/Latino participants were enrolled at a rate commensurate with the US population of Hispanic/Latino children (OR, 1.02; 95% CI, 1.01-1.03). American Indian/Alaska Native (OR, 0.82; 95% CI, 0.79-0.85), Asian (OR, 0.56; 95% CI, 0.55-0.57), and Native Hawaiian/Pacific Islander (OR, 0.66; 95% CI, 0.61-0.72) children were enrolled significantly less compared with the respective US populations of these groups (Table 3). In comparison with US census data in counties where trials were conducted, Black/African American children were enrolled proportionally more than expected (OR, 1.50; 95% CI, 1.49-1.51), as were American Indian/Alaska Native (OR, 1.16; 95% CI, 1.11-1.20) and White children (OR, 1.24; 95% CI, 1.23-1.25) (eTable 1 in the Supplement).

Trials that enrolled fewer than 100 participants and trials that enrolled 100 to 499 participants demonstrated lower Hispanic/Latino enrollment than expected (eTable 2 in the Supplement). Our subanalysis excluding trials that targeted specific racial or ethnic groups demonstrated comparisons between enrolled and expected participants that were similar to those in all trials (eTable 3 in the Supplement). Some differences in enrollment were observed in comparison with the US population by intervention type, but enrollment by intervention type did not differ from overall enrollment patterns (eTable 4 in the Supplement).

The mean diversity index of included pediatric clinical trials was 0.39 (95% CI, 0.38-0.41; range, 0-0.75). In bivariable analysis, trials with 500 or more participants had a diversity index of at least the level of the state where the research was conducted compared with trials that had fewer than 100 participants (OR, 2.30; 95% CI, 1.11-4.70). Stakeholder involvement in the trial design was associated with lower odds of enrolling participants at least as diverse as in the state where the trial was conducted (adjusted OR, 0.40; 95% CI, 0.10-0.90). Journal policy statements on reporting of race and ethnicity were not associated with a diversity index of at least the level of the state where the research was conducted (eTable 5 in the Supplement).

Diversity indexes of pediatric clinical trials conducted in single states are shown in Figure 2A. The diversity indexes of trials conducted in Arizona, New Mexico, Georgia, Tennessee, Virginia, and Maryland were lower than the respective states’ pediatric diversity indexes (Figure 2B). Trials conducted in Maine were most diverse in comparison with the state’s diversity index. At the state level, Black/African American participants were included more than expected, and White participants were least represented compared with individual states’ respective populations (eFigure 2 in the Supplement).

Equitable clinical trial enrollment is an actionable step that investigators can take to address health inequities in the US. In this cross-sectional study of 612 pediatric clinical trials with 565 618 children that were published from 2011 to 2020, the proportion of published pediatric clinical trials that reported participant race and ethnicity increased over time, but such data were still underreported. Black/African American children were enrolled 88% more than expected based on the US pediatric population, and Hispanic/Latino children were enrolled commensurately with the population of Hispanic/Latino children; however, children in other racial and ethnic minority groups were enrolled less than expected based on the respective US populations of these groups.

Studies in adult populations have reported that 27% to 83% of published articles did not report participant race or ethnicity.^8-11 In the present study, 83.1% of trial articles published in 2020 reported participant race and 72.9% reported ethnicity, which is higher than the percentage reported in a study in 2003 (pediatric participant race and ethnicity reported in 67% of articles).¹⁶ The increased proportion of published pediatric clinical trials that reported participant race and ethnicity in the present study may be associated with the increased recognition of the importance of inclusive trial enrollment and journals’ increased requirement of reporting of participant race and ethnicity. There was significant heterogeneity in how race and ethnicity were measured, with fewer than half of articles reporting that race or ethnicity data were obtained from caregivers or participants. Accurate recording of participant race and ethnicity may be best achieved by asking guardians to report their child’s race and ethnicity according to standardized categories.^26,27 Standardization of how race and ethnicity are measured and reported may reduce discrepancies.

Contrary to our hypothesis, Black/African American children were enrolled more than expected and Hispanic/Latino children were enrolled proportionately based on the US population. These findings differ from work in adult populations demonstrating lower enrollment of Black/African American and Hispanic/Latino adults in clinical trials^28-33 and may reflect greater focus on enrollment of Black/African American and Hispanic/Latino participants by pediatric clinical trialists. Trials that report race and ethnicity may have a greater focus on enrolling these populations. This higher rate of enrollment occurred despite prior evidence suggesting that caregivers of Black/African American children may have distrust of medical research^34-36 because of the historical and current racism exhibited in medicine in the US.^37-39 This finding aligns with those of the 2003 analysis of pediatric studies¹⁶ but differs from disease-specific studies assessing pediatric participant enrollment.^12,13 The majority of pediatric clinical trials likely occurred in academic centers, which are often clustered in urban centers that have higher populations of children in racial and ethnic minority groups. Critical race theory allows examination of elements of structural racism that have led to the over- or underrepresentation of different racial and ethnic groups in these studies. It is possible that the overrepresentation of Black/African American children is associated with the history of redlining and other policies that led to the segregation of Black children in low-income neighborhoods, which frequently exist near academic medical centers.⁴⁰ Structural racism has also been associated with high poverty rates⁴¹ and subsequent high Medicaid enrollment rates among Black/African American families,⁴² which may lead to the utilization of academic medical centers at higher rates among these families than among privately insured families.⁴³ Our findings may also have been influenced by potential undercounting of children in racial and ethnic minority groups in the US census.⁴⁴

Our study adds to the existing body of literature by demonstrating that, at the national level, American Indian/Alaska Native, Asian, and Native Hawaiian/Pacific Islander children were enrolled less than expected. However, when the proportion of enrolled American Indian/Alaska Native and Native Hawaiian/Pacific Islander children was compared with the population of each state, representation was approximately as expected. Thus, equitable representation may be associated not only with local recruitment but also with the geographic distribution of trial locations.

A recent analysis of trial enrollment among trials funded by the Best Pharmaceuticals for Children Act similarly demonstrated lower representation of Asian pediatric participants but otherwise commensurate enrollment for children of other races and ethnicities.²⁵ Although the enrollment of diverse participant populations likely relies on multiple factors, there was no association between articles in journals that required reporting of participant race and ethnicity and the enrollment of diverse pediatric participants in single-state trials in our analysis. However, journals’ websites did not explicitly state when these policies were put into place. Articles reporting stakeholder involvement in trial design were less likely to enroll a diverse population in single-state studies; this finding may have been associated with the encouragement of greater representation of participants of racial and ethnic minority groups,⁴⁵ which in turn may have led to lower diversity indexes.

Given the relatively low representation of some groups in pediatric clinical trials, modifications to current enrollment practices may be needed. Previous studies have suggested that Asian and Native Hawaiian/Pacific Islander participants may be more likely to enroll if translation is included in recruitment materials.⁴⁶ Providing recruitment materials in participants’ preferred language and employing multilingual staff have been associated with improved enrollment rates among Hispanic/Latino populations.^47-50 The creation of culturally sensitive materials to explain trial details may be associated with increased recruitment and retention of pediatric participants in trials.⁵¹ Beginning in 2022, Medicaid will provide coverage for “the routine costs associated with clinical trial participation”⁵²; this coverage may be associated with greater enrollment among groups that otherwise would not participate because of additional costs.

This study has limitations. The sample of published articles may not be representative of all pediatric clinical trials. We reviewed published articles to assess the trials with highest impact and to obtain as much detailed information about the trials as possible.³⁴ The lack of standardized reporting of race and ethnicity among published articles may have led to over- or underestimation of enrollment of some groups. We excluded children categorized as “other” in both trials and the US census because this term had no clear meaning within or among studies. This exclusion may have led to underrepresentation of some groups that might have been categorized as “other” in some trial results. Some trials included populations with known higher rates of diseases, such as sickle cell disease and cystic fibrosis, which likely affected representation of racial and ethnic groups in those trials. However, our subanalysis excluding these trials demonstrated results consistent with those for the overall population. We did not record the reported efficacy of individual interventions by participant race and ethnicity. It is difficult to define catchment areas in clinical trials, particularly because they are reported in published articles and many clinical trials are conducted at academic medical centers in urban areas. We were not able to determine when journals enacted policies regarding the reporting of race and ethnicity.

This cross-sectional study revealed that the proportion of published pediatric clinical trials that reported participant race and ethnicity from 2011 to 2020 increased, but participant race and ethnicity were still underreported. Disparities existed in pediatric clinical trial enrollment among American Indian/Alaska Native, Asian, and Native Hawaiian/Pacific Islander children. The higher representation of Black/African American children compared with the US census population suggests that inclusive research practices could be extended to other historically and currently disenfranchised racial and ethnic groups.

Accepted for Publication: December 1, 2021.

Published Online: March 21, 2022. doi:10.1001/jamapediatrics.2022.0142

Corresponding Author: Chris A. Rees, MD, MPH, Division of Pediatric Emergency Medicine, Emory University School of Medicine, 1405 Clifton Rd NE, Atlanta, GA 30322 (chris.rees@emory.edu).

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

Concept and design: Rees, Stewart, Mehta, Portillo, Duggan, Fleegler.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Rees, Stewart, Mehta, McKay, Portillo, Duggan, Fleegler.

Critical revision of the manuscript for important intellectual content: Rees, Stewart, Mehta, Avakame, Jackson, Portillo, Michelson, Duggan, Fleegler.

Statistical analysis: Rees, Michelson.

Administrative, technical, or material support: Stewart, Mehta, McKay, Duggan, Fleegler.

Supervision: Duggan, Fleegler.

Conflict of Interest Disclosures: Dr Duggan reported receiving royalties from PMPH USA and UpToDate, Inc, and income for editorial duties from the American Society for Nutrition outside the submitted work. No other disclosures were reported.

Funding/Support: This study was supported in part by grants K24DK104676 and 2P30 DK040561 from the National Institutes of Health (NIH) (Dr Duggan) and grant K08HS026503 from the Agency for Healthcare Research and Quality (AHRQ) (Dr Michelson).

Role of the Funder/Sponsor: The NIH and the AHRQ had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: Chloe Rotman, MSLIS (Boston Children’s Hospital), provided assistance in developing the PubMed query and acquiring the full-text articles reviewed in this study. Ms Rotman did not receive compensation for her contributions.