General Rheumatology · Baptist Medical Center Jacksonville
Methods
The data that support the findings of this study are available from the corresponding author upon reasonable request. A study protocol for this study was reviewed by the institutional review boards and approved as a nonhuman subjects study. As the study did not use individual patient‐level data, but publicly available trial‐level data from the ClinicalTrials.gov website, informed consent was not applicable.The clinical trials search tool21 on the NIH website (ClinicalTrials.gov) was used to obtain a list of trials related to stroke (search date: May 12, 2022) using the following search algorithm: (1) the condition/disease specified as Stroke in the search tool; (2) conducted in the United States and completed between January 1, 2010 and December 31, 2020; (3) studies that enrolled adults (18–64 years) or older adults (65+ years); (4) study phase listed as phase 1 to 4 or not applicable. An initial review was conducted by dividing the studies between all authors to assess the eligibility and for data collection by 7 reviewers in the 3 months following the search for trials.22 (having incident or recurrent stroke as an outcome of interest, or assessment of poststroke outcomes) and were conducted in adult populations by reviewing the trial summary and outcomes. The reviewers also assessed trial locations and those with 1 or more trial sites outside of the United States were excluded. Eligible trials were further assessed to see if reports of race and ethnicity were available on the ClinicalTrials.gov website or if there were publications of the main trial results associated with the registered National Clinical Trial number. Collected data underwent adjudication by 3 reviewers (H.D.N., M.A., C.L.) to ensure accuracy. Any conflicts between the reviewers related to the eligibility clinical trials and data were resolved through discussion.The results of each clinical trial included in this study were examined to gather data for the following outcomes: (1) if race or ethnicity were reported or not, and (2) the frequencies and proportions of each race or ethnicity group. Race was categorized as White, Black, Asian or Pacific Islander (PI), Native American, multiracial, and other/unknown. The studies varied in reporting of Hispanic ethnicity. Most studies included in this review categorized the Hispanic group as ethnicity; however, in some studies (trials N=27; 13.2%) Hispanic ethnicity was reported as a race. In the present study, Hispanic identity was reported as an ethnicity in accordance with NIH guidelines.22Information about trial characteristics such as funding source, trial allocation (randomization, blinding, trial design), type of intervention, and year of trial completion (2010–2020) was abstracted using the ClinicalTrials.gov search tool. Funding agencies were vastly diverse and recoded based on an algorithm used by similar prior studies.23, 24, 25 Briefly, 3 types of funding sources were identified: federal/NIH, industry, and other. Trials with NIH or other US federal agencies listed as lead sponsor or collaborator were recorded as NIH/federal. Trials with an industry‐based organization listed as the lead sponsor or collaborator, and without collaboration from NIH/federal agencies were recorded as “industry” funded. Trials with any other sponsor or collaborator and without collaboration from NIH/federal agencies or an industry‐based organization were coded as “other” funding sources. Categories for trial design (crossover, factorial, parallel, sequential, single‐group) and blinding (none, single, double, triple, quadruple) were presented as designated by the ClinicalTrials.gov database. Two types of intervention categories were used. The first classification used categories designated by the NIH as follows: behavioral, biological, device, drug, and other interventions. The second classification specific to stroke was based on a literature review of National Institute of Neurological Disorders and Stroke stroke trials by Marler et al26 and included 5 categories of stroke‐specific interventions including acute poststroke intervention, primary/secondary stroke prevention, poststroke recovery and rehabilitation, stroke education, and others.Reporting of race and ethnicity by key trial characteristics was summarized as frequencies and percentages. Pearson's chi‐square test or Fisher's exact test were used to assess the differences in proportions of race or ethnicity reporting by key trial characteristics as appropriate. The distribution of race and ethnicity groups were calculated as proportions and were further described by the source of funding, type of intervention, type of stroke trial, and year of completion. The trend of changes in proportions of trials reporting race or ethnicity (yes/no) by year of trial completion was tested using the Cochran–Armitage test. Results were considered statistically significant under the α=0.05. All analysis was carried out using SAS, Version 9.4 (Cary, NC).
Results
A total of 588 stroke‐related clinical trials completed between 2010 and 2020 were identified from the ClinicalTrials.gov website (Figure 1). After exclusion, 488 trials were retained for further review, 336 (68.9%) of which had published results available on either the ClinicalTrials.gov website or in journal articles. Race or ethnicity distribution was reported in 202 (60.1%) trials out of all published trials. Although not statistically significant, trials with federal/NIH funding had a higher proportion of race or ethnicity reporting (Table 1; 68.6%; P=0.06) compared with industry (55.7%) and others (55.4%). Race or ethnicity reporting differed significantly by intervention type (P<0.01) with lower proportions of reporting in device trials (47.0%) compared with all other interventions. Race or ethnicity reporting also increased with an increase in sample size (P<0.01). Trials with a sample size of 1000 or more participants had the highest proportion of race and ethnicity (83.3%), whereas trials with small sample sizes, <10 (34.5%) and <50 participants (58.2%), were less likely to report race or ethnicity. Race or ethnicity reporting improved significantly over time (Z=−4.38; P<0.01) by 2.6‐fold (Figure 2), with 30% reporting race or ethnicity in 2010 compared with 78% in 2019 and 61% in 2020.Trial Characteristics by Race and Ethnicity ReportingIndicates statistical significance under α=0.05.Fisher's exact test used for assessment of statistical significance.Out of 37 589 participants enrolled across trials with race and ethnicity reports, 65.0% were White, 24.8% Black, 2.4% Asian or PI, 0.4% Native American, 0.4% multiracial, and 7.0% were reported as other or unknown. Hispanic ethnicity made up 9.0% of all participants. Overall, the race distribution was found to be consistent over time with the highest proportion of White participants followed by Black participants (Figure 3A). An exception was noted in 2018 when a higher proportion of Black participants (53.1%) were recruited compared with White participants (35.8%). This difference disappeared upon the exclusion of trials that enrolled only participants from racial and ethnic minority groups (Figure 4A). Racial minority groups such as Asian or PI, Native American, and multiracial participants remained consistently low over the study period.A, Black vs White. B, Asian or PI (Pacific Islander), Native American, multiracial, other/unknown vs White. C, Hispanic vs non‐Hispanic White.A, Black vs White. B, Asian or PI (Pacific Islander), Native American, multiracial, other/unknown vs White. C, Hispanic vs non‐Hispanic White.Enrollment of Black participants was higher in trials with federal/NIH funding (Figure 5A; 28.1%) and other/academic funding (26.6%) compared with industry‐funded trials (11.3%). Trials funded through federal agencies/NIH had a higher proportion of Hispanic participants (13.8%) compared with industry (6.9%) and other (5.4%) trials.A, By funding source B, By intervention type C, By stroke‐specific trial type. NIH indicates National Institutes of Health; and PI, Pacific Islander.The enrollment of minority groups, including Black (7.1%) and Hispanic (7.9%) participants, was notably lower than the White population (84.2%) in biological intervention trials (Figure 5B). The highest proportions of participants with Asian/PI backgrounds (5.0%) were enrolled in pharmacological studies. Behavioral intervention trials had the most diverse enrollment with equal proportions of White (41.1%) and Black participants (41.4%) and 14.5% Hispanic participants. The highest enrollment of Hispanic participants (16.1%) and Asian/PI participants (5.0%) was reported in drug trials. Enrollment of Native American and multiracial groups remained <1% across all intervention types. Stroke prevention trials had the most racially diverse enrollment, with 34.6% Black participants and 9.7% Hispanic participants (Figure 5C). The highest proportions of Asian/PI (4.2%) and Hispanic enrollees (11.5%) were reported in stroke rehabilitation and recovery trials.
Discussion
The representation of racial and ethnic minority groups remained low in stroke trials from 2010 to 2020. This is especially true for minority groups such as Asian/PI individuals, Native American individuals, and individuals with multiple races, which represented <4% of the trial participants. Representation of individuals from Black and Hispanic backgrounds was more favorable compared with Asian/PI, Native American, and multiracial individuals. This trend remained consistent throughout the study period across all clinical trials included in this study, regardless of the funding source or intervention type.The lower recruitment of individuals from racial and ethnic minority groups such as Asian and Native American populations compared with non‐Hispanic White participants in these clinical trials is discordant with the higher risk of stroke mortality seen among these groups. Studies have reported significantly higher odds of hemorrhagic stroke10, 27 and stroke mortality28 among Asian participants as compared with the non‐Hispanic White population. Similarly, multiple studies have reported higher stroke mortality29, 30 in the Native American population compared with the non‐Hispanic White population. However, these disparities are not adequately reflected in recruitment for stroke trials. Conversely, the higher risk of stroke mortality may in fact affect the enrollment and participation due to fewer participants from these underrepresented racial and ethnic groups available for enrollment.It is well documented that risk of stroke is disproportionately higher in Black population compared with non‐Hispanic White. Of the 795 000 stroke cases in the United States annually, 13.2% occur among the Black population.2 The prevalence of stroke is higher among Black men (4.8%) and women (5.4%) than any other race and sex strata.2 This is reflected in the relatively better enrollment statistics for Black participants than other racial and ethnic minority groups. In fact, in 2018, we observed a higher proportion of Black (53.1%) enrollees compared with White (35.8%) enrollees. Upon further assessment, we identified 3 studies completed in 2018,31, 32, 33 that targeted racial minority groups and, hence, specifically enrolled participants from Black (N=2161) population, which was the driving force behind the seemingly overrepresented Black participants. This finding indicates that trials focused on specific racial groups can reach sufficient and diverse enrollment targets. Such practices have also successfully increased diversity in observational stroke and cardiovascular cohort enrollment.9, 34, 35, 36 Similar strategies for stroke trials with a commitment toward reaching these underrepresented racial or ethnic groups should be encouraged by funding agencies.Our findings indicated a variation in racial and ethnic minority enrollment based on funding source. Specifically, trials funded by federal agencies, including NIH, had higher enrollment of individuals from racial and ethnic minority groups compared with industry funded trials. It was noted that recruitment of Black participants in other/academically funded trials was comparable to federally/NIH‐funded trials. A similar finding was reported by a previous review of clinical trials (not limited to stroke trials) conducted and registered on ClinicalTrials.gov between 2010 and 2020.25 It is difficult to determine the reason for this finding given the descriptive nature of the present study; however, agencies like NIH use specific strategies to encourage diversification of trials. For example, specific procedures and policies around providing a detailed plan to recruit underrepresented population groups such as women and racial or ethnic minority groups implemented in NIH grant applications likely affects the rigor in recruiting a diverse trial population.19 NIH also offers funding mechanisms designed to target underserved racial and ethnic minority groups. Another possible reason could be differences in the monitoring procedures and oversight practices for the trials funded through federal agencies like NIH. Implementing universal reporting guidelines, streamlining monitoring procedures, and requiring a detailed recruitment plan from the investigators across funding agencies would likely have a positive impact on these observed disparities by funding source.Despite the lack of heterogeneity in race and ethnicity distribution among stroke trial participants, we observed a 2.6‐fold increase in reporting between 2010 and 2020. A previous review examining race and ethnicity reporting in phase III stroke trials funded by the National Institute of Neurological Disorders and Stroke and conducted within an earlier time frame (1985–2008) indicates a higher percentage of race reporting (86.4%).18 The difference in our results from this study may be due to differences in inclusion criteria, given that our study included trials regardless of their funding sources, along with our inclusion of early phase trials. Another recent review conducted on ischemic stroke‐related randomized controlled trials for the same time frame as that of our study (2010–2020) observed an underwhelming proportion of race and ethnicity reporting (42%) among these trials.37 This study also reported a slightly lower proportion of Black participants (19%) and slightly higher proportion of Hispanic participants (11%) compared with our results. The proportion of Asian participants was similar (2%) in both studies.37 The differences in our results from this study could be explained by our inclusion of nonrandomized trials along with randomized controlled trials, trials with sample size <100, and trials related to all stroke subtypes. Our study also expands the literature further by assessing the differences in race and ethnicity distribution by trial characteristics such as funding source and type of intervention.Multiple barriers to representation in clinical trial enrollment have been identified previously. Research on barriers to trial participation in cancer trials is particularly robust. On a systemic level, institutional environment, leadership support, availability of resources, staff allocation, and community stakeholder involvement are some factors that influence the trial enrollment process.17 On an individual level, patient knowledge about the disease, benefits of participation, access to clinical trials, participation logistics such as time or transportation, trust in care providers, and language proficiency are some of the well‐known barriers to trial participation.16, 17 Some of the proposed strategies from qualitative assessment of trial recruitment in other clinical disciplines may be applicable to recruitment in stroke trials. For example, one of the more recently proposed solutions to logistical barriers has been using telehealth services in conducting clinical trials. Research has indicated high patient interest and acceptability for trials conducted remotely using telehealth services.38, 39Cultural barriers often contribute to difficulties in recruiting a diverse trial cohort. Within the area of stroke research, a survey of investigators reported that the lack of trust, awareness about trials, and communication issues were the most frequently encountered barriers to minority recruitment in stroke‐related trials.40 Over 51% of stroke researchers stated that implementing a proactive recruitment goal for minority inclusion and requiring cultural competency training for recruitment personnel were effective strategies in improving minority recruitment.40 A mixed methods study of different recruitment strategies for Black communities reported that involvement of community stakeholders, newspaper advertisements, and initiating a snowball effect through word‐of‐mouth communication were largely effective strategies in reaching recruitment goals.41Our study is unique in including early phase stroke‐related trials, which have been excluded in previous similar studies. Early phase or pilot trials are very good tools for defining recruitment goals and identifying strategies for optimal enrollment in larger, later‐phase trials.42 We also assessed various trial characteristics and their association with race and ethnicity reporting and representation for stroke‐related trials. Although descriptive, our study holds value in its evaluation of the current racial landscape of the clinical trial practices in stroke research. It identifies gaps in the trial reporting guidelines and oversight. Particularly, as the racial distribution of the American population is likely to evolve significantly by 2060,1 it is imperative to retain our knowledge and understanding of the differences in various health outcomes across all racial and ethnic groups and implement necessary changes reflecting these differences in research practice.This study should be interpreted within the scope of its limitations. First, this is a descriptive study, including trials conducted in the United States. Therefore, the results may not be generalizable to other countries. Second, 39.9% of eligible trials had no published results available despite the NIH requirement to submit results within a year after the primary completion of trials.43 Therefore, these results do not represent studies that have not reported their recruitment rates on ClincalTrials.gov or in publications. Third, we recognized some discrepancies in the categorization of race and ethnicity among trials. For example, 7% of participants in trials reporting race and ethnicity were categorized as other, unknown, or not reported, which comprised the highest proportion of participants after White (65%) and Black (25%) participants. It is unclear whether the race encoding other indicates lower numbers of participants from racial and ethnic minority groups or if there is some other reason for it. However, it warrants standardization and transparency in race and ethnicity reporting. Fourth, we did not have information on how race and ethnicity was assigned for each trial that reported these characteristics. Lastly, we also included trials assessing primary stroke prevention as an outcome, which indicates that some trials would have recruited individuals who were stroke‐free. Therefore, we were not able to make a direct comparison between the observed proportions of race or ethnicity groups in clinical trials and the incidence or prevalence of stroke among these groups. Additionally, the tracking of stroke cases by race and ethnicity is not as straightforward similar to the Surveillance, Epidemiology, and End Results Program for cancer cases,44 which allows investigators to assess disease burden by most race and ethnicity groups, including the Asian and the Native American populations. Similar surveillance strategies for other diseases like stroke would make a direct comparison between clinical trial representation and disease burden by race much more plausible.
Conclusions
Although race and ethnicity reporting among stroke trials has improved within the past decade, the gaps in representation of racial and ethnic minority groups remain prevalent. To improve consistency and transparency in clinical trial reporting, it is essential to enforce and monitor standardized reporting of race and ethnicity for all trials regardless of their sample size, trial phase, or source of funding. Trials with an enrollment goal for racial and ethnic minority groups embedded into their design and research question are successful in reaching their enrollment targets.31, 32, 33 Such trials should be encouraged by funding mechanisms to improve diversity, equity, and inclusion in clinical research for stroke. Given the shifting racial distribution of the American population, it is essential to extend our efforts beyond examining the reasons for racial disparities in stroke research and move towards affirmative action.
Sources of Funding
This study was funded through Veterans Affairs grants received by Dr Lin (VA IK2 CX00201 and VA I21 RX003612).
Disclosures
None.
JS
Dr. Jeffrey Shen
Rheumatology · General RheumatologyBaptist Medical Center Jacksonvillejeffreyshen