Exclusion of Families who Speak Languages Other than English from Federally Funded Pediatric Trials

Abstract

Objective:

To determine whether individuals in the U.S. who speak languages other than English (LOE) are excluded from federally funded, pediatric clinical trials; and whether such trials adhere to NIH policy regarding inclusion of members of minority groups.

Study Design:

Using ClinicalTrials.gov, we identified all completed, federally funded, U.S.-based trials inclusive of children ≤17 and focused on one of 4 common chronic childhood conditions (asthma, mental health, obesity, dental caries) as of June 18, 2019. We reviewed ClinicalTrials.gov online content, as well as published manuscripts linked to ClinicalTrials.gov entries, to abstract information about language-related exclusion criteria. Trials were deemed to exclude LOE participants/caregivers if explicit statements regarding exclusion were identified in the study protocol or published manuscript.

Results:

189 trials met inclusion criteria. Two-thirds (67%) did not address multilingual enrollment. Of 62 trials that did, 82% excluded LOE individuals. No trials addressed enrollment of non-English, non-Spanish-speaking individuals. In 93 trials with non-missing data on ethnicity, Latino individuals comprised 31% of participants in trials that included LOE individuals and 14% of participants in trials that excluded LOE individuals.

Conclusions:

Federally funded pediatric trials in the U.S. do not adequately address multilingual enrollment, a seeming violation of federal and contractual requirements for accommodation of language barriers by entities receiving federal funding.

INTRODUCTION

In the United States (U.S.), more than 25 million people ages five years and older speak English less than “very well”—referred to as “limited English proficiency” or LEP in U.S. census data.¹ Additionally, there are approximately 11.4 million children who live with at least one parent who speaks English less than very well, and of those children 4.4% are themselves LEP.²

Language barriers are particularly salient for Latino and Asian American children: Approximately 40% of Latino and Asian American children in the US have at least one parent who speaks English less than very well.² Nationally, these families are strikingly diverse. While 64% speak Spanish, the remaining third speak over 100 different languages, including Chinese (6%), Vietnamese (3%), Korean (2%), and Tagolog (2%).¹ Within some Asian American subgroups (e.g., Chinese, Vietnamese, and Korean communities) over 40% of individuals have limited English proficiency.¹

Because nearly 16% of U.S. children have at least one parent who speaks English less than very well and these children are predominantly from racial/ethnic minority groups, there are pressing questions about whether families who speak languages other than English (LOE) are equitably enrolled in and benefit from pediatric research.^2,3-5 Equitable enrollment and benefit from research are described in the National Institutes of Health (NIH) policies pertaining to the enrollment of individuals from minority groups and by the ethical principles for human subjects research described in the Belmont Report. Further, all NIH requests for proposals include language reminding applicants of their Title VI obligations and require funding recipients to be inclusive of LOE individuals.⁶ Finally, there are federal, state and contractual legal requirements for accommodation of language barriers by entities receiving federal and/or state funds, e.g., Medicare, Medicaid, and National Institutes of Health (NIH) grants.^{7, 8}

However, a published review of predominantly adult emergency medicine research found that 42% of articles did not address multilingual enrollment. Among those that did, 19% excluded LOE individuals.⁹ Similarly, among Australian studies that described language exclusion and/or inclusion criteria, 21% excluded LOE participants,¹⁰ and exclusion of LOE participants was three times higher in pain and mental health trials as opposed to cancer treatment trials. Even among cancer treatment trials, however, exclusion of LOE participants appears to be prevalent: For example, 10% of protocols at one large cancer research center were found to explicitly exclude LOE individuals.¹¹ Further, a review of over 10,000 registered clinical trials found that English-language requirements were more common after 2010 than in the prior decade.¹² The same analysis found that English-language requirements were particularly common among trials registered in ZIP codes with a high proportion of Asian residents.¹² These studies did not specifically examine trends in pediatric research. However, a series of recent manuscripts raise troubling questions about the exclusion of LOE children and families, as well as the exclusion of children from racial/ethnic minority groups where LOE is prevalent. For example, Chen et al found that individuals who speak languages other than English were included in only 469 of 5008 studies published in three pediatric journals from 2012-2021.¹³ Among studies inclusive of languages other than English, the most common language was Spanish (75%).

Such exclusion compromises efforts to measure disease prevalence and health outcomes among Latino and Asian American children. It can also undermine the generalizability of prevention, diagnosis, and treatment evidence to these populations.^13-16 Finally, it may lead to exclusion of members of minority racial/ethnic groups.

To augment the growing literature on inclusion or exclusion of LOE children and families in child health research, we examined language-related exclusion criteria for federally funded pediatric clinical trials, as such trials should ideally follow NIH and DHHS policies regarding inclusion. We also compared enrollment of racial/ethnic minority participants in studies that did and did not exclude LOE participants and/or caregivers.

METHODS

This is a retrospective, observational study that examines the inclusion and exclusion of LOE individuals from pediatric research in the U.S. Data were collected from ClinicalTrials.gov, a publicly available clinical trial registry and results database developed and maintained by NIH through the National Library of Medicine (NLM) in collaboration with the Food and Drug Administration (FDA).¹⁷ It includes clinical studies funded and/or sponsored by the NIH, other federal agencies, private industry, and foundations, as well as unfunded and investigator-initiated trials. A web-based system is used to facilitate clinical trial registration by any sponsor, principal investigator or other entity with primary responsibility for a trial/study.

Using the “advanced search” function on ClinicalTrials.gov, we limited our search to four common chronic conditions of childhood: mental health disorder, obesity, asthma, and dental caries. We used the broadest search term for each condition and limited results to trials that met all the following conditions: interventional, completed recruitment, with results, Age Group = Child (birth-17 years old), NIH or other U.S. federal funding, Location = US. We included all trials that met these criteria as of June 18, 2019. Detailed definitions describing these variables can be found on ClinicalTrials.gov Protocol Registration Data Element Definitions for Intervention Studies and ClinicalTrials.gov Results Data Element Definitions for Intervention Studies issued by the National Library of Medicine.^18,19 To ensure consistency in results, two study team members independently ran the agreed upon search criteria on June 18, 2019 and confirmed that ClinicalTrials.gov output was identical. Data were abstracted into REDCap.²⁰

We then categorized each study as either (1) excluding LOE participants, (2) inclusive of LOE participants, or (3) not specifying. To make each determination, we first reviewed inclusion and exclusion criteria published on ClinicalTrials.gov (Table A; online). If criteria pertaining to English proficiency were not documented, we reviewed the attached study protocol, if available, followed by linked peer-reviewed publications. Consistent with prior research, trials were deemed to exclude LOE individuals if explicit statements regarding exclusion were identified, e.g., “Participants do not speak and read in English” as an exclusion criterion or “Parent and child must be English speaking” as an inclusion criterion. Trials with clearly stated criteria for the inclusion of LOE individuals, e.g., trials that listed languages other than English, were categorized as such. Trials with no information regarding LOE inclusion or exclusion were categorized as not specifying their language-related eligibility criteria.

For enrollment data pertaining to race and ethnicity, we followed a similar 3-step protocol: First, we examined the ClinicalTrials.gov demographic table. Then, we reviewed linked publications in order of citation until the desired data were extracted. However, there were several cases where demographic data in publications were either missing or the reported sample size on ClinicalTrials.gov did not match the reported sample size in the cited publication. In these instances, race and/or ethnicity data were marked as “not reported.” Whenever possible, we used federal/NIH categories for race and ethnicity.

STATISTICAL ANALYSES

We conducted a descriptive analysis, describing the completeness for each data element and summarizing the characteristics of our sample and our primary outcome (LOE study eligibility) by condition/disease. Because there were relatively few trials focused on obesity and dental caries, these were combined into a single category. To examine the relationship between LOE eligibility and enrollment of racial/ethnic minorities, we created 5 subgroups comprised of trials with non-missing data for the 5 federal/NIH categories for race and ethnicity. For example, one subgroup comprised trials with non-missing data for Latino enrollment. For each subgroup, we then compared the proportion of racial/ethnic minority participants in studies that excluded LOE children and/or caregivers, studies that included LOE children and/or caregivers, and studies that did not describe language-related eligibility criteria. All analyses were done using REDCap and MS Excel.

RESULTS

As described in Table 1, there were 189 clinical trials that met inclusion criteria. Two-thirds (67.2%) of the trials did not address multilingual enrollment, 27.0% explicitly excluded LOE individuals, and only 5.8% explicitly included LOE individuals (Table 2). Of the 11 trials that allowed LOE participation, eight were asthma studies and three focused on mental health. All 11 focused only on Spanish-speakers. In other words, no trials had eligibility criteria inclusive of non-English, non-Spanish-speaking individuals. Sample language pertaining to LOE exclusion and inclusion are shown in Table 3.

Table 1:

Characteristics of U.S.-based, federally funded, pediatric protocols registered with ClinicalTrials.gov¹

	All Protocols	Mental Health Protocols	Asthma Protocols	Other Protocols2
N	189	134	31	24
Study Enrollment, median (IQR)	86 (44, 208)	71 (40, 143)	294 (104, 395)	97 (51, 360)
NIH-funded, % (N)a	89.9 % (170)	89.6% (120)	93.5% (29)	87.5% (21)
Intervention Type, % (N)b
Drug Alone	45.5% (86)	48.5% (65)	48.4% (15)	25% (6)
Behavioral Alone	31.2% (59)	29.9% (40)	22.6% (7)	50% (12)
Drug and Behavioral	6.9% (13)	9.7% (13)	0% (0)	0% (0)
Device Alone	4.8% (9)	4.5% (6)	3.2% (1)	8.3% (2)
Drug + Other	3.2% (6)	0.7% (1)	16.1% (5)	0% (0)
Behavioral + Other	2.6% (5)	2.9% (4)	0% (0)	4.2% (1)
Other	5.8% (11)	3.7% (5)	9.7% (3)	12.5% (3)
Inclusion of Children/Caregivers who Speak Language Other than English (LOE), % (N)c
Not Reported	67.2% (127)	70.9% (95)	45.2% (14)	75% (18)
Excluded LOE	27.0% (51)	29.1% (36)	29.0% (9)	25% (6)
Included LOE	5.8% (11)	2.2% (3)	25.8% (8)	0% (0)
Study Reported Ethnicity of Participants, % (N)d	49.2% (93)	41.0% (55)	77.4% (24)	58.3% (14)
Study Reported Race of Participants, % (N)d	65.6% (124)	59.7% (80)	90.3% (28)	66.7% (16)

¹All studies identified on 06/18/2019 using the following ClinicialTrials.gov search criteria: condition=mental health OR asthma OR obesity OR dental caries, study type=interventional, status=completed, age group=child (0-17 years), location=United States, AND funder type=NIH OR Other U.S. Federal agency. Note that studies were included if enrollment was open children (ages 0-17 years), even if they also included adults of any age. Half of the studies identified were open to both adults and children. Primary Completion dates ranged from 08/01/2004 to 07/01/2018

²Clinicial trials for obesity (N=X) and dental caries (N=Y) were combined due to small sample sizes.

^aStudies funded in whole or in part by the NIH, as reported to ClinicalTrials.gov. Other federal funders included the Department of Defense (DOD), Centers for Disease Control and Prevention (CDC), Food and Drug Administration (FDA), Veterans Administration (VA), and Health Resources and Services Administration (FDA).

^bOther interventions included vaccines/biologics, procedures/surgeries, dietary supplements, educational programs, and service delivery innovations, e.g., web portals or telemedicine.

^cAs reported to ClinicalTrials.gov under “Eligibility Criteria” and/or “Study Documents” or, if not available on ClinicalTrials.gov, in a manuscript listed under “Publication of Results.” Among all protocols reporting whether or not LOE children/caregivers were excluded, the majority (51 of 62) excluded LOE individuals. Among mental health, asthma, and other protocols reporting whether LOE children/caregivers were excluded, 92.3%, 45.2%, and 100% of protocols excluded LOE individuals, respectively.

^dAs reported to ClinicalTrials.gov under “Study Results” or, if not available on ClinicalTrials.gov, in a manuscript listed under “Publication of Results.”

Table 2:

Inclusion of minority participants in U.S.-based, federally funded, pediatric protocols registered with ClinicalTrials.gov, by language inclusion/exclusion status

	All Protocols	LOE Individuals: Included1	LOE Individuals: Excluded	LOE Individuals: Unknown
Studies that report Latino enrollment data a
Cumulative enrollment	17,540	2,882	5,725	8,933
Latino, % (N)	17% (2,989)	31% (906)	14% (786)	15% (1,297)
Studies that report Asian enrollment data b
Cumulative enrollment	16,832	1,331	6,314	9,187
Asian, % (N)	3% (505)	0.2% (3)	4% (275)	2% (227)
Studies that report Hawaiian or Pacific Islander enrollment data b
Cumulative enrollment	15,560	1,331	6,020	8,209
Hawaiian or Pacific Islander, % (N)	0.2% (31)	0% (0)	0.3% (17)	0.2% (14)
Studies that report American Indian or Alaska Native enrollment data b
Cumulative enrollment	15,976	1,331	6,020	8625
American Indian or Alaska Native, % (N)	1.1% (181)	0.7% (9)	1.7% (105)	0.8% (67)
Studies that report Black/African American enrollment data b
Cumulative enrollment	23,363	2,889	6,946	13,528
Black/African American, % (N)	31.3% (7,315)	52% (1,504)	17.1% (1,188)	34.2% (4,623)

¹Trials were deemed to exclude participants/caregivers who speak language other than English (LOE) if explicit statements regarding exclusion were identified in the study protocol (“excluded”). Otherwise, studies were categorized as including LOE individuals (“included”)—in which case included languages were recorded—or as not specifying language-related eligibility criteria (“unknown”). Those few studies that included LOE individuals were open only to participants who spoke English and/or Spanish.

^aIncludes any study that reported ethnicity for enrolled or evaluable participants, even if the number of enrolled Latino participants was 0. Studies that reported ethnicity only for a sub-population of enrollees were not included.

^bIncludes any study that reported the specified race for enrolled or evaluable participants, even if the number of enrolled participants from the specified group was 0. Studies that reported race only for a sub-population of enrollees were not included.

Table 3.

Selected text pertaining to exclusion of participants/caregivers with emerging English proficiency from U.S.-based, federally funded, pediatric protocols registered with ClinicalTrials.gov¹

Exclusion Criteria:
Do not primarily speak English
Inability to read or understand English
Participants do not speak and read in English
Does not primarily speak English, including caretakers of participants when the participant is a child
Does not speak English primarily
Inability to speak and understand English
Caregiver does not speak English
Non-English speaking
Non-native English speaker
Unable to read English. Unable to speak English.
Inability to communicate satisfactorily and directly (without a translator) in English
Inclusion Criteria:
Parent and child must be English speaking
Able to read and write in English
Able to speak and understand English
At least parent/caregiver fluent in English
English-speaking
Be fluent in English
Parent and participants able to compete study rating scales and other measures in English
Speaks, reads, and writes English
Sufficient proficiency in English to complete assessments
The child and parent(s) have a working command of English
Can read, write, and speak English

¹Language regarding exclusion of participants/caregivers who speak languages other than English (LOE) was abstracted from studies identified on 06/18/2019 using the following ClinicialTrials.gov search criteria: condition=mental health OR asthma OR obesity OR dental caries, study type=interventional, status=completed, age group=child (0-17 years), location=United States, AND funder type=NIH OR Other U.S. Federal agency.

Overall, 49.2% of trials reported the ethnicity of study participants and 65.6% reported data for at least one racial group. Asthma trials were more likely to report ethnicity (77.4%; data not shown) and race (90.3%; data not shown) than were mental health trials (ethnicity 41.0%, race 59.7%; data not shown). As shown in Table 2, there were 93 trials with non-missing data on ethnicity: Latino individuals comprised 31.0% of participants in trials that explicitly included LOE individuals (n=8) and 14.0% of participants in trials that explicitly excluded LOE individuals (n=33). There were 117 trials with non-missing data on Black/African American race: Black/African American individuals comprised 52.0% of participants in trials that explicitly included LOE individuals (n=9) and 17.1% of participants in trials that explicitly excluded LOE individuals (n=40). Other racial groups comprised less than 5.0% of participants, regardless of LOE inclusion or exclusion policies. Notably, there were fewer Asian participants in trials inclusive of LOE individuals than in trials that excluded LOE individuals.

DISCUSSION

In this subset of federally funded, pediatric clinical trials in the U.S., most either exclude individuals with LOE or did not address language in their eligibility criteria. Because multilingual study enrollment requires prior preparation of dedicated resources (e.g., translation of study documents) this suggests that most pediatric trials in this sample were not open to LOE children and/or caregivers even if these individuals were not formally excluded.

Our findings suggest that protocols that include LOE individuals better adhere to NIH policy for enrollment of individuals with ethnic minority backgrounds. Specifically, studies inclusive of Spanish-speaking participants had significantly higher Latino enrollment that more closely resembled the proportion of the Latino children in the US general population (~25% of U.S. children).² Overall the enrollment of Asian Americans remains poor even though they represent ~5% of the U.S. child population. We suspect that Asian American subgroups with the highest proportion of LOE individuals (e.g., the Vietnamese, Chinese, and Korean communities, among others) are particularly poorly represented in pediatric clinical trials given the lack of accommodation for Asian languages. In this sample—which was smaller than that examined by Chen et al and excluded observational studies and multinational research—we did not identify any trials that explicitly included individuals who speak languages other than Spanish and English.¹³

This study has key limitations. First, we did not examine inclusion and exclusion criteria for all federally funded pediatric clinical trials in the U.S. Instead, we restricted our sample to studies examining the four most common chronic conditions of childhood. Clinical trials for other conditions—particularly conditions where clinical trials may be less dependent upon surveys and interviews, e.g., pediatric oncology—may have more inclusive multilingual enrollment. Second, we were limited to data available in ClinicalTrials.gov and linked protocols and manuscripts. While we have made every effort to be thorough in our review, data on the race and/or ethnicity of enrolled participants were often incomplete. However, we expect that studies attentive to minority enrollment would also be more likely to publish these enrollment data, and so our findings may represent the best-case scenario. Third, we are unable to provide insight into enrollment of racial/ethnic subgroups who do not self-identify with the categories used most often in the US, e.g., Arabic-speaking individuals of Middle Eastern/North African heritage. Fourth, we are unable to describe the racial and ethnic composition of the communities residing in the catchment area of each of the included protocols. However, prior studies have examined this question. Finally, this study was conducted in 2019. However, more recent research suggests that these findings remain relevant.¹³

CONCLUSION

Concerted federal and institutional efforts are necessary to ensure that child health researchers are better prepared to adhere to federal requirements for language access and minority inclusion.¹² Failing to create the necessary conditions to allow LOE individuals to participate in research is discriminatory and, given the disparate impact on Latino and other racial/ethnic minority communities, contributes to structural racism in child health research.²¹ Exclusion also undermines trust between academic institutions and LOE communities, and trust is another important determinant of research participation.

At present, the U.S. offers few if any formal training programs in multilingual child health research. Such programs should include interdisciplinary training of professional interpreters alongside research teams. Institutions should make better use of technology to address regulatory barriers to LOE inclusion, e.g., electronic signatures to ensure remote interpreters can witness short-form informed consent. They should also negotiate with language services vendors to ensure appropriate attention to research-related requirements (such as witnessing consent) and prioritize recruitment and training of multilingual research staff.^{22-25, 26} Institutional Review Boards should advocate for the creation of plain language consent and assent forms, as these are more readily translated and benefit both LOE and English-proficient communities. We also recommend that NIH and other major funding entities better enforce existing regulations pertaining to multilingual enrollment and offer internal and extramural research supplements to cover costs associated with staff training, interpreting, and translation. Finally, the NIH should prioritize the translation and cultural validation of commonly-used standardized instruments and make these readily available to researchers.

Without such efforts, pediatric research in the U.S. is at risk of becoming less and less generalizable and less and less just. The goal in clinical research must be to provide families and children who speak language other than English the opportunity to participate to ensure the generalizability of results and improve the health of all children.

ABREVIATIONS:

LEP

Limited English Proficiency

LOE

Language(s) other than English

NIH

National Institutes of Health