Is the Enrollment of Racial and Ethnic Minorities in Research in the Emergency Setting Equitable?

Abstract

Background

Concerns have been raised about the enrollment of racial and ethnic minorities in research in the emergency setting when it is not possible to obtain informed consent. However, there is a paucity of data related to the validity of such claims.

Methods

Retrospective comparison of registry enrollment (4/1/06–3/31/07) and trial enrollment (4/1/07–3/31/08) from 3 sites in the Resuscitation Outcomes Consortium. Subjects compared met the following criteria: 1) Shock, defined by blunt or penetrating force to the body with either systolic blood pressure (SBP) ≤ 70 mmHg or SBP 71–90 mmHg and heart rate ≥ 108 beats/min; and/or 2) Traumatic Brain Injury (TBI), defined by blunt force to the head with out-of- hospital Glasgow Coma Score ≤ 8.

Results

Overall, compared to a registry there were no differences in the percent of racial or ethnic groups enrolled in the clinical trial [Odds Ratio (OR) for Blacks versus Whites: 0.87, 95% Confidence Interval (CI) 0.65–1.16, p=.34; OR for Hispanics versus Whites 1.04; CI 0.72–1.49, p=.85]. However, Blacks were less likely than Whites to be enrolled in the TBI cohort [OR 0.58 (0.34–0.97); p=.04].

Conclusions

Despite some discordance in subgroups, there was no overall difference in the racial and ethnic distribution of subjects enrolled in a multi-center clinical trial of severe trauma compared to a registry accounting for study entry criteria. These findings help address justice concerns about enrollment of racial and ethnic minorities in trauma research performed using an exception from informed consent under emergency circumstances.

INTRODUCTION

Across a wide spectrum of medical research, concerns have been raised about the appropriate representation of those with a variety of socio-demographic characteristics including gender, age, and race.(1) As a matter of justice, it is essential to ensure that enrollment in research is fair and representative of those who experience a given disease or condition for a variety of reasons. First, if research is not conducted within a particular group, it may be unclear whether a particular intervention is safe and effective in that group. Conversely, members of one group ought not to disproportionately bear the risks of research so that those in another group may benefit.

With respect to race, low participation rates in clinical research trials among African-Americans in the United States are often attributed to a lack of trust in the medical and research communities stemming from historical abuses. (2–9) There are also suggestions that low participation rates may be due to the fact that persons of color may be approached for enrollment less frequently. (10, 11)

In a study of hypothermia treatment in trauma that used prospective informed consent for the first nine months of the study, minorities were under represented by 30% in that time period.(12) Concerns have also been raised about racial enrollment in research in the emergency setting, where it may not be feasible to obtain informed consent due to the nature of the conditions being studied, such as cardiopulmonary arrest or severe trauma.(13–17) Here, however, the matter relates to possible over-enrollment of racial and ethnic minorities, rather than low participation rates that are decried in other types of clinical research. The underlying issue in this sort of research seems to be the suspicion that the research will be conducted among racial and ethnic minorities without long-term benefits accruing to them for an experimental intervention that proves to be effective. Alternatively, a group might assume the risks of research when the intervention turns out not to be ineffective. Yet, given the prevalence of these conditions and the poor outcomes associated with them among persons of all races and ethnicities(17–20), it is essential to ensure equitable enrollment in ethically sound research in the emergency setting. Only in this way can potentially effective interventions be properly tested and eventually introduced appropriately into clinical practice.

While such concerns about disproportionate enrollment have been raised for research in the emergency setting, the actual extent to which racial and ethnic minorities are enrolled in this research is unclear. In this study, we sought to determine whether racial and ethnic minority enrollment is disproportionate in ongoing research using an exception from informed consent under emergency circumstances.(14) Such data are essential for helping to inform policy and practice for research in the emergency setting.

METHODS

Background

The Resuscitation Outcomes Consortium (ROC) is a research network established to evaluate the treatment of people with life-threatening injury and out-of- hospital cardiac arrest and to conduct clinical trials of promising scientific and clinical advances so as to improve survival. This network consists of 11 sites and one central coordinating center. See https://roc.uwctc.org/tiki/tiki-index.php?page=roc-public-home. (21) ROC created a registry (Epistry) that includes all cases of cardiopulmonary arrest and severe traumatic injuries assessed or treated by emergency medical services (EMS) personnel in participating geographic regions.(22) In addition, a randomized trial (HS) of normal saline versus hypertonic saline versus hypertonic saline with dextran is enrolling patients with shock after traumatic injury or traumatic brain injury (TBI).(23) The Epistry enrolls subjects with a waiver of the requirement to obtain informed consent.(24) The HS trial enrolls subjects using an exception from the requirement to obtain informed consent under emergency circumstances.(14) Use of this exception requires that the research meet a series of provisions which include community consultation and public disclosure.(25) We sought to compare the race and ethnicity of subjects enrolled in the Epistry to the HS trial to assess whether racial and ethnic enrollment is disproportionate. The primary null hypothesis was that the proportion of each race or ethnicity enrolled did not differ when comparing subjects meeting criteria for enrollment in the HS trial who were enrolled in Epistry versus subjects enrolled in the HS trial.

Design

This is a retrospective comparison of observational (Epistry) versus clinical trial (HS) data from EMS agencies and receiving institutions in eight US sites and three Canadian sites that are participants in the ROC. In both Epistry and HS, EMS personnel classify race and ethnicity based on visual assessment, supplemented by additional information such as name or information provided by family members if available. Three sites with sufficient information describing the race and ethnicity of those enrolled in both Epistry and HS were included in the primary analyses.

Population

The study population was all traumatic injury cases that occurred within the catchment area of a participating EMS agency who met HS study entry criteria before the trial began and those patients with traumatic injury who were enrolled in the trial following study implementation. The census tract of the location of each case was identified and recorded in order to assess the catchment population served by each agency.

Inclusion and Exclusion Criteria

The ROC Trauma Epistry is a consortium-wide epidemiologic registry of trauma cases that are assessed or treated by EMS agencies participating in ROC the details of which are described elsewhere. (22)

To be enrolled in the HS trial a person must be at least 15 years of age and must experience: 1) blunt or penetrating force to the body with either out-of-hospital systolic blood pressure (SBP) ≤ 70 mmHg or out-of- hospital SBP 71–90 mmHG and heart rate ≥ 108 beats/min, and/or 2) blunt force to the head with out-of-hospital Glasgow Coma Score ≤ 8. Patients who experienced both criteria are considered to be part of the first (shock) cohort. Patients are enrolled by the out-of-hospital provider and receive a blinded 250 cc bag of study fluid (either normal saline, hypertonic saline, or hypertonic saline with dextran).

Epistry cases occurring between April 1, 2006 and March 31, 2007 that were assessed by EMS and met the HS physiologic entry criteria, and HS cases entered in the clinical trial (occurring between April 1, 2007 and March 31, 2008) were included in all analyses. These time periods were chosen to permit a comparison of one full year of data from the registry to the subsequent year of data from the trial. Trauma Epistry data were not fully collected after March 31, 2007, and all sites had not yet begun HS at this time, so insufficient data were available to perform concurrent analyses.

Enrollment data from all sites are displayed in Table 1. Sites were assigned a random code to preserve confidentiality. To minimize potential bias, only sites with out-of-hospital race and ethnicity data recorded for >75% of eligible patients in both Epistry and HS were included in subsequent analyses. Further analyses were limited to subjects who were identified to be White, Black, or Hispanic, representing >98% of those with available data about race and ethnicity in these three sites. The number of subjects identified as Asian was extremely small, precluding analysis in this regard.

Table 1.

Completeness of Data by Site^*.

	TRAUMA EPISTRY RACE/ETHNICITY DATA (4/2006–3/2007)			HS TRIAL RACE/ETHNICITY DATA(4/2007–3/2008)
Random Site Code	Missing	Available	% Available	Missing	Available	% Available
216	119	2	2%	127	0	0%
307	306	0	0%	72	0	0%
315	18	254	93%	2	113	98%
389	556	0	0%	105	0	0%
434	67	93	58%	56	50	47%
477	30	56	65%	4	1	20%
663	114	58	34%	0	18	100%
671	309	211	41%	103	129	56%
791	258	0	0%	73	1	1%
864	10	209	95%	3	92	97%
925	45	315	88%	34	112	77%
Total in sites with sufficiently complete race/ethnicity data in both Epistry and HS [315, 864, 925]	72	780	92%	39	317	89%

^*Note: Agencies not enrolling in HS were excluded from Epistry analysis, and vice versa. Sites in bold (315, 864, and 925) had sufficient race and ethnicity data available for analysis.

The distribution of these three classifications in the catchment population was estimated via data from the 2000 US census. Analysis was restricted to those 18 years of age and over with only one designated race or ethnicity and who lived in a census tract covered by one or more EMS agencies participating in Epistry. However, incidence of trauma may not reflect this racial and ethnic distribution, so the distribution of race and ethnicity among all trauma cases recorded in Epistry was also provided.

While all ROC sites participated in Epistry, a few ROC EMS agencies within these sites did not. Therefore these analyses are restricted to subjects enrolled by agencies that participated in both Epistry and HS.

Human Subjects

The Epistry and the HS trial were approved by the University of Washington Institutional Review Board, as well as by all participating local institutional review boards or research ethics boards. Epistry was approved with a waiver of the requirement for informed consent. HS was approved with an exemption from the requirement to obtain consent in emergency setting. The Institutional Review Board at Johns Hopkins Medical Institutions deemed this analysis of previously collected data to be exempt from review.

Data Analysis

The primary analysis involved chi-square tests to assess whether or not there were associations between the races (Black vs. White) or ethnicities (Hispanic vs. White) and enrollment status. Odds ratios and 95% confidence intervals (CI), comparing the odds of being in HS among Blacks (Hispanics) to the odds of being in HS among Whites were also calculated.

RESULTS

From April 1, 2006 to March 31, 2007 there were 8299 subjects enrolled in the Epistry database at 11 sites. Of these 3276 met entry criteria for the HS study. From April 1, 2007 to March 31, 2008 there were 1064 subjects enrolled in the HS study at the 11 sites. The primary analyses were restricted to the three sites with sufficiently complete out-of-hospital data on race and ethnicity (Table 1).

A comparison of matched US Census tract data from 2000 to overall enrollment in Epistry in the sites with sufficient data for analysis is displayed in Table 2. While the catchment population as defined by census data are 63% White, 20% Black, and 17% Hispanic, enrollment in the trauma registry is 39% White, 41% Black, and 20% Hispanic, estimating the prevalence of trauma in this population.

Table 2.

Overall US Census Data and Enrollment in Epistry by Race and Ethnicity ^*

Race/Ethnicity	US Census 2000 (tracts covered by Epistry)	# enrolled in Epistry (Epistry 4/2006–3/2007)
White	63%	1231 (39%)
Black	20%	1285 (41%)
Hispanic	17%	627 (20%)

^*Note: Analyses are restricted to sites with sufficient race/ethnicity data. Census data include those 18 years and older who were designated as being of only one race or ethnicity. Agencies not enrolling in HS were excluded from Epistry analysis.

For the three sites with sufficiently complete data on race and ethnicity, there where 855 subjects enrolled in Epistry who met HS enrollment criteria, while 351 subjects were enrolled in HS. These observed differences in enrollment in the registry and the trial are likely related to our data sources. For example, registry data do not include information regarding all the exclusion criteria for the trial (such as having received more than 2 liters of fluids). In addition, data in the registry are less complete than corresponding data in the trial, which represent only those actually enrolled and not those who would simply be eligible. For example, a patient might be eligible, but not enrolled because EMS personnel were too busy to do so. We are unaware of any intervening factors that would account for differential enrollment over time.

Of the Epistry subjects, 782 had available race and ethnicity data and 771 were White, Black, or Hispanic. Of the HS subjects, 317 had available race and ethnicity data and 315 were White, Black, or Hispanic. Subject characteristics are described in Table 3. Consistently, a greater proportion of men were enrolled in HS compared to Epistry.

Table 3.

Subject Characteristics by Study and Race/Ethnicity^*

Race/Ethnicity	Eligible for HS (Epistry 4/2006–3/2007)		Enrolled in HS (HS 4/2007–3/2008)		p-value
	N		N
Shock Cohort
Age – mean (SD)
White	175	40 (23)	52	37 (17)	0.25
Black	247	27 (18)	94	31 (11)	0.02
Hispanic	87	23 (17)	27	30 (11)	0.01
Gender - % male
White	175	67%	52	73%	< 0.001
Black	247	79%	94	93%	< 0.001
Hispanic	87	67%	27	93%	< 0.001
Injury Type - % pen.
White	175	22%	52	31%	0.28
Black	247	60%	94	81%	< 0.001
Hispanic	87	28%	27	59%	0.01
TBI Cohort
Age – mean (SD)
White	137	40 (20)	82	38 (19)	0.54
Black	75	36 (17)	26	35 (13)	0.57
Hispanic	50	28 (11)	34	29 (9)	0.61
Gender - % male
White	137	71%	82	71%	< 0.001
Black	75	73%	26	77%	< 0.001
Hispanic	50	78%	34	88%	< 0.001
Injury Type - % pen.
White	137	0%	82	1%	-
Black	75	0%	26	4%	-
Hispanic	50	0%	34	3%	-
Entire HS Sample
Age – mean (SD)
White	312	40 (22)	134	38 (18)	0.24
Black	322	29 (19)	120	32 (11)	0.09
Hispanic	137	25 (16)	61	30 (10)	0.01
Gender - % male
White	312	69%	134	72%	< 0.001
Black	322	78%	120	89%	< 0.001
Hispanic	137	71%	61	90%	< 0.001
Injury Type - % pen.
White	312	13%	134	13%	0.92
Black	322	46%	120	64%	< 0.001
Hispanic	137	18%	61	28%	0.14

^*Note: Analyses were restricted to those with sufficient race/ethnicity data.

By definition, no injuries were penetrating in HS-TBI-eligible Epistry, whereas some subjects with penetrating injuries were enrolled in HS-TBI. P-values for age were computed using a t-test. P-values for gender and injury type were computed using Pearson’s Chi-squared Test. Agencies not enrolling in HS were excluded from Epistry analysis, and vice versa.

Overall enrollment in Epistry compared to HS, as well as enrollment by cohort in HS, is shown in Table 4. Blacks were not differentially enrolled compared to Whites (OR for enrollment 0.87; 95% CI, 0.65–1.16, p=.34); and Hispanics were not differentially enrolled in the trial compared to Whites (OR for enrollment 1.04; 95% CI 0.72–1.49, p=.85).

Table 4.

Enrollment in Epistry versus HS by Race/Ethnicity

Race/Ethnicity	# eligible for HS (Epistry 4/2006–3/2007)	# enrolled in HS (HS 4/2007–3/2008)	Odds Ratio (95% CI)	p-value
Shock Cohort
White	175	52	-	-
Black	247	94	1.28 (0.87–1.90)	0.21
Hispanic	87	27	1.05 (0.61–1.77)	0.87
TBI Cohort
White	137	82	-	-
Black	75	26	0.58 (0.34–0.97)	0.04
Hispanic	50	34	1.14 (0.67–1.90)	0.63
Entire HS Cohort
White	312	134	-	-
Black	322	120	0.87 (0.65–1.16)	0.34
Hispanic	137	61	1.04 (0.72–1.49)	0.85

Note: Analyses restricted to sites with sufficient race/ethnicity data. P-values were computed using Pearson’s Chi-squared Test. Agencies not enrolling in HS were excluded from Epistry analysis, and vice versa.

However, when looking at the cohorts separately, Blacks were less likely to be enrolled in the TBI cohort [OR 0.58 (0.34–0.97); p=.04]; there was a non-significant trend for Blacks to be more likely to be enrolled in the shock cohort [OR 1.28 (0.87–1.90); p=.21].

DISCUSSION

Some critics of research in the emergency setting have claimed that racial and ethnic minorities are or will be enrolled disproportionately in this research.(13, 15) Examining such subjective claims with objective data is essential. We found that the enrollment of racial and ethnic minorities in an ongoing randomized trial was proportionate to the prevalence of the condition being studied. However, our analyses of census tract data compared to a trauma based registry also confirm that blunt or penetrating trauma and traumatic brain injury occur disproportionately among racial and ethnic minorities in the catchment areas where the research is being conducted. As has been suggested by others, this may be due to the location of many trauma centers and the prevalence of trauma among racial and ethnic minorities.(13, 15)

Regardless, while it may be true that more persons in a group may be enrolled in research directed at addressing a particular disease or condition that is more prevalent in that group, prima facie it would be incorrect to conclude that this alone is an ethical problem. In fact, one could argue that it is appropriate. From the perspective of justice, such an approach comports with the notion that resources, should be directed at addressing conditions that are highly prevalent and that are associated with considerable morbidity and mortality. On this view, one might expect communities who are disproportionately affected by such conditions, not only to endorse research efforts, but to facilitate them. Of course, one might reasonably contend, at least in the case of traumatic injury, that resources ought to be directed at prevention as well as treatment.

Nonetheless, despite the substantial morbidity and mortality associated with traumatic injury among all people, deliberations about the appropriate enrollment of racial and ethnic minorities for research in the emergency setting are influenced by the disturbing historical legacy of research and medical care among African -Americans.(13, 26, 27) Such concerns mirror earlier concerns regarding medical research and care in general.(28) Of issue here is the assumption that this legacy is associated with a reluctance to enroll in research and that many racial and ethnic minorities would refuse to consent if it was feasible to obtain consent in this setting. While it is impossible to know for sure whether a particular individual would give consent for a specific trial, this study was not conducted to address this question. Although understanding community attitudes towards research in the emergency setting is of critical importance, deciding not to conduct scientifically and ethically sound research in otherwise appropriate settings simply because of the suspicion that some people in the community would not be likely to give consent if it was possible to do so comes at the price of conducting biased research, which may have profound consequences for public health.(29) Biased research may lead to a further erosion of health disparities by resulting in a failure to apply effective interventions or continuing to apply ineffective ones.

Similarly, our secondary analyses revealed a statistically significant under-enrollment of women compared to men in the trial compared to the registry. While the reasons for this finding are unclear, it warrants subsequent primary examination since it potentially raises similar issues with respect to ensuring appropriate representation in clinical trials. (1) Such issues are recognized in NIH policy. (30)

Limitations

Our findings must be interpreted with some limitations in mind. First, we used historical controls which may not accurately reflect the true incidence of the condition over time. Nevertheless, control data were gathered only one year prior to the time of enrollment and were obtained using the same data collection techniques, thereby minimizing the risk of bias. Furthermore, more accurate comparator data are unavailable.

Second, identification of race and ethnicity was made by EMS personnel and may be inaccurate. However, the same method for classifying race and ethnicity was used for both cases and historical controls, and we have no reason to suspect that these classifications of race and ethnicity would have been made differently for the two datasets or over time, minimizing the likelihood of bias.

Third, our analysis involves three geographic sites within the context of a single clinical trial. Therefore, findings may differ in other geographic regions or in other conditions. Regardless, the multi-center nature of this trial and using sites in the analysis that had sufficient data minimize such concerns.

Fourth, while we did not identify a difference in enrollment according to race and ethnicity, it is conceivable that with a larger sample we may have identified a difference with respect to ethnicity given the width of the confidence intervals associated with these data. Similarly, the sample size precludes further examination of subgroups of those experiencing trauma.

Fifth, while we made every effort to accurately identify those cases in the registry that would have been eligible for the trial, it is possible that some misclassification occurred, as described earlier. Nevertheless, they are unlikely to affect our findings regarding racial and ethnic enrollment.

Sixth, our analysis did not assess whether there were disparities among those enrolled under the exemption from the requirement for informed consent versus those enrolled by a legally authorized representative. While this approach makes the comparison of enrolment in the trial to those in the registry cleaner, it does not provide information about whether there is disparate enrollment in these subgroups.

Finally, information about those not enrolled in the trial was not routinely collected. Nevertheless, such data would likely have enhanced the rigor of our findings.

Concluding Comments

Research in the emergency setting is vital to improving survival after major trauma. While such research raises important ethical issues, especially when it is impossible to conduct it with informed consent, it is essential to bring data into the debate. Such an approach promises to enhance the design and conduct of this research, as well as direct attention to those aspects of the research that most demand it. Moving forward, future research should consider prospectively gathering data about those patients who are not enrolled. Regardless, to meet the broader obligations of the ethical principle of justice, efforts need to be made to ensure that the results of this research be applied in the settings in which it was conducted in hopes of decreasing morbidity and mortality.