Social and demographic predictors of willingness to participate in an exception from informed consent trial

Consent is a cornerstone of ethical research, but when patients have serious injuries, they – and their Legally Authorized Representatives (LAR) – are often unable to participate in the informed consent process. In 1996, the US Department of Health and Human Services (DHHS) and the Food and Drug Administration (FDA) issued guidelines for the conduct of clinical trials in patients with emergent medical conditions [1], including processes for community consultation and public disclosure (CC/PD). Community Consultation is a process that primarily involves consultation between the investigative team and representatives of the communities, overseen and informed by feedback from the institutional review board (IRB), where the relevant “community” includes people residing in the geographical area where the research will be conducted and communities likely to have or at-risk for the condition [2, 3].

Many previous studies of CC, including a systematic review, have demonstrated that respondents support emergency care research [4,5,6,7,8]. The literature on the respondent characteristics associated with specific attitudes and responses, in contrast, is much more limited. Biros et al. showed that those with higher education or incomes were favorably inclined or tended to be favorable towards research in general, and EFIC research in particular, whereas those with less education and older individuals were not supportive. They also observed that women tended to be less supportive, although this finding was not statistically significant. [5] Ross et al. studied willingness to participate in an EFIC trial in pediatric intensive care found that willingness rates varied by race. [6] Both of these studies were small, which may have obscured differences.

The Center for Injury Science at the University of Alabama at Birmingham (UAB) uses an interactive media based approach to CC, which consists of social media communications on Facebook and Instagram, linked websites, online community forums, and targeted community surveys [9,10,11,12]. The community surveys solicit community members’ opinions on whether they would want to be enrolled in a proposed study, whether they would want a family member to be enrolled, whether they regard emergency care research as important, and whether they think the proposed study should be conducted in their community. Along with these results, the surveys collect demographic information, including age, gender, race/ethnicity, education level, household size, and household income.

This interactive media based approach was used to conduct CC activities for the Trauma And Prothrombin Complex Concentrate (TAP) trial, a randomized clinical trial that compares the early administration of 4-Factor Prothrombin Complex Concentrate with placebo, in trauma patients predicted to require a large volume blood transfusion [13]. The trial, which is sponsored by CSL Behring, was being conducted in approximately 120 trauma centers. Approximately 95 of the sites were located in the United States and therefore required to conduct CC/PD, making this one of the largest CC/PD campaigns conducted to date.

We have reported the overall results of the TAP trial CC/PD in a previous publication, with emphasis on reach, as well as similarities and differences between sites in terms of website views, survey responses, community forum attendance, and opt-out requests [4, 14]. The scope of the TAP trial CC/PD permits a much more comprehensive analysis of the characteristics of respondents who agree, or disagree, to participate in EFIC trials. The purpose of this present analysis was, therefore, to evaluate the social and demographic factors associated with survey respondents’ willingness to participate in the trial, in order to facilitate the further development of CC/PD methods for multicenter exception-from-informed-consent (EFIC) trials.

Community surveys

The survey consisted of six questions (supplementary file 1), designed to evaluate respondents’ willingness to be enrolled in the TAP trial without consent if incapacitated (question 1); respondents’ willingness for family members to be enrolled in the TAP trial without consent if incapacitated (question 3); whether they believe that emergency medical research is necessary (question 5), and whether they believe that the TAP trial should be done in their community (question 6). The responses to questions 2 and 4, which relate to reasons why they would not want to be included, and were conditional of “no” answers to questions 1 and 3 respectively, were not included in this analysis. Supplementary file 1 also lists the answer choices.

The surveys were administered using Qualtrics (Qualtrics, Seattle, WA), using the company’s cloud-based survey software platform. The surveys were accessible to anyone, from the TAP trial websites, from a QR code on a study handout that sites placed in areas where they felt they could reach at-risk or underserved community members, or directly, using a browser. In addition, invitations to complete the surveys were distributed, by Qualtrics, to individuals who reside in counties in the catchment area of participating trauma centers. We aimed to have at least 300 completed surveys from each participating trial site. The surveys were completed between October 20th, 2022 and November 14th, 2023.

Outcomes

Questions 1, 3 and 5 had four answer choices – “Yes”, “No”, “I don’t know”, and “I don’t want to answer”. We excluded the “I don’t want to answer”(1.97%) responses from the analysis. Questions 6 had two answer choices, “Yes” and “No”.

Analysis

We used a multinomial logistic regression model for questions 1, 3, and 5 and a binomial logistic model for question 6 to estimate odds ratios (ORs) and associated 95% confidence intervals (95% CIs) to examine whether the demographic characteristics associated with answering “Yes”, “No”, or “I don’t know” differed. We modeled answer choices using a random utilities model in which individuals faced three choices: Agreement (I would want to be enrolled; I would want family members to be enrolled; I believe that emergency medical research is necessary; I believe that the study should be done in my community), disagreement (I would not want to be enrolled; I would not want family members to be enrolled; I do not believe that emergency medical research is necessary; I do not believe that the study should be done in my community), and uncertainty (I don’t know if I would want to be enrolled; I don’t know if I would want family members to be enrolled; I don’t know if emergency medical research is necessary). The utilities associated with each of these choices were modelled as a function of individual specific demographic characteristics. The analysis was performed in SAS v9.4.

The responses to each of the questions are shown in Table 1. In total, there were 22,441 responses to question 1; 21,617 to question 3; 22,329 to question 5; and 22,324 to question 6. (Please note that these numbers differ from our previous publication, which only included 52 trial sites, whereas the present analysis includes data from 77 centers.)

The majority of those surveyed responded that they would want to be entered into the TAP trial, even if they could not give consent (13,201; 58.8%), that they would want family members to be entered (12,174; 56.3%), and that emergency medical research was necessary (19,430; 87.0%). Respectively, 3,625 (16.2%), 3,885 (18.0%), and 906 (4.1%) replied that they would not want be entered; and 4,903 (21.9%), 4,814 (22.3%), and 1,700 (7.6%) replied that they did not know if they would want to be entered. With regards to question 6 (“Do you believe that this study should be done in your community?”), which did not include an “I don’t know” answer choice, 19,555 (87.6%) responded “yes”, and 2,769 (12.4%) “no”. Table 1 also lists the number of “I don’t want to answer” responses, which were excluded from the subsequent analyses.

Tables 2– 5 present the weighted sample means by outcome (“Yes”, “No”, “I don’t know”), and the results of the multinomial logit model. In each of the tables, the first column shows the characteristics of the “Yes” respondents, as number and percentage, for categorical and ordinal outcomes; or mean and standard deviation, for continuous outcomes. Subsequent pairs of columns show the results for the “No” and “I don’t know” respondents, compared to the “Yes” respondents. In each case, the first column shows the raw data, and the second the odds ratios (OR) and associated 95% confidence intervals. Values > 1 indicates that higher values of the explanatory variable increase the predicted probability of the first relative to the second outcome. For the categorical and ordinal characteristics, a referent was chosen. For example, the odds of being female (rather than male) for those who responded that they would not want to be entered into the trial, compared with those that did, were 1.13 (95% CI 1.05–1.22).

For question 1 (“If you were severely injured and needed blood transfusions, would you want to be entered into this research study, even though you couldn’t give consent?”), those who responded “No”, compared to those who responded “Yes”, were more likely to be younger, female, non-white and non-Hispanic, have an education level lower than high-school, and a household income of less than $80,000). Those who responded “I don’t know”, compared to those who responded “Yes”, were more likely to be older, non-male, multi-racial and non-Hispanic, have an education level lower than high-school, and have an income of less than $35,000. (Table 2).

For question 3 (“If one of your family members was severely injured and needed blood transfusions, would you want them to be entered into this research study, even if they or you couldn’t give consent?”), those who responded “No”, compared to those who responded “Yes”, were more likely to be younger, non-male, non-White, have an education level lower than high-school, and have a household income of less than $100,000. Those who responded “I don’t know”, compared to those who responded “Yes”, were more likely to be older; non-male; Asian alone or multiracial, non-Hispanic; have a less-than-high-school education or some college education, and have an income of less than $50,000. (Table 3).

For question 5 (“Do you believe that emergency medical research is necessary?”), those who responded “No”, compared to those who responded “Yes”, were more likely to be younger, non-male, non-white, have an education level lower than high-school, and have an income of less than $80,000. Those who responded “I don’t know”, compared to those who responded “Yes”, were more likely to be non-male, non-white, have a less-than-high-school education, and have an income of less than $65,000. (Table 4).

For question 6 (“Do you believe that this study should be done in your community?”), those who responded “No”, compared to those who responded “Yes”, were more likely to be younger, non-binary/gender-nonconforming, none-white, have an education level lower than high-school, and have an income of less than $80,000. (Table 5).

The relationship between individuals’ age, income, and education level, and their willingness to participate, is shown in a series of stacked contour graphs. These graphs illustrate how different characteristics interact and impact individuals’ opinions. Supplementary Fig. 1 shows the probability of a “yes”, “no”, or “I don’t know” response, with reference to age and household income, to the four questions. Supplementary Fig. 2 shows the probability with reference to age and educational attainment. Darker colors indicate a lower probability.

We have previously shown that our interactive, media-based approach to community consultation and public disclosure reaches large numbers of individuals, through the use of social media [9]. However, exception from informed consent regulations emphasize consultation with the community, which requires the solicitation of opinions.

In a socially just world, individuals should have equal opportunities to share their views. Encouragingly, our present analysis shows that community members who completed our surveys are racially, ethnically, educationally, and economically diverse. It is conceivable that the online distribution of the community surveys provides opportunities for otherwise underrepresented populations who might not be reached through traditional means of community outreach to learn about complex healthcare issues affecting their communities, and to offer their opinions.

Our results show that the majority of individuals would want to be enrolled in an EFIC clinical trial or have their family member enrolled in such a trial when they are in a life-threatening emergency situation. As well, the majority of the survey participants agreed that emergency research with an exception from informed consent is important to conduct. These findings are in keeping with previous research [5,6,7,8, 15, 16]. Our results, however, also show that not everyone is willing to participate in the TAP trial. There is little previous research on how demographics impact individuals’ opinions of such studies, and their willingness to participate, or for their family members to participate. Our study, which is much larger than the two previous studies [5, 6], identifies clear patterns in a large and diverse patient population. We found that the patient group which tended to select “no”, or were unsure about participation, tended to be younger, non-male, non-white, have a lower education level, and an unknown or lower household income. These results are in keeping with the studies by Biros and Ross [5, 6], and affirm that social and demographic factors play an important role in community healthcare initiatives. However, like previous work, our study cannot shed light on the reasons for these differences.

Interestingly, when we asked broader questions about the community as a whole (“Do you believe that emergency medical research is necessary?” and “Do you believe that this study should be done in your community?”), responses were more favorable than those received in response to questions about individual willingness to participate. Again, these findings are in keeping with previous work [8].

It is possible that the results reflect genuine differences in opinions, which individuals are entitled to, and should be respected. Anyone could request to opt out of the TAP trial, should they be injured. However, there is also a possibility that those who selected no, I don’t know or I don’t want to answer, might benefit from more detailed or better explanations of why a particular study is being done, and more information about EFIC regulations and overall study risks and benefits.

There are many types of social activities that are good for society, or the community at large, but that inconvenience the individual or puts them at risk. The dichotomy between agentic (advancing one’s self) and communal motives (serving others) is widely known [17]. It is plausible that people might both want the research to be done, but would prefer that they not be the ones participating in the trial.

Other forms of community outreach during the community consultation and public disclosure did not see similar feedback from this specific demographic. Online forums are relatively easily directed towards certain groups. However, we have previously shown that engagement with these types of community meetings is low. Furthermore, those who are opposed to EFIC research may also not be prepared to participate in community meetings, making this approach problematic. Targeting social media communications towards certain geographic areas and age groups is also possible and straightforward. Targeting certain racial or ethnic groups, in contrast, is no longer possible, following the inappropriate use of such data [18,19,20].

Our findings have practical and ethical implications and will require examination by ethicists, regulators, investigators, sponsors and regulators. The investigators’ role is to “consult with the community”, which is defined as a bidirectional process that requires the provision of information and the solicitation of opinions. This definition would appear to support the idea of increased engagement with communities that exhibit increased hesitancy. However, it should be noted that the EFIC regulations do not stipulate a certain level of agreement within a community. Therefore, such activities should be less about changing opinions and more about increased engagement and outreach with better communication of relevant information without being coercive. Beyond community consultation, our study should also prompt investigators to examine how discussions with patients – once they have been enrolled, under EFIC regulations – are conducted, with reference to the demographic characteristics identified. This knowledge may help to guide such conversations, and better address individuals’ concerns.

This study is, to our knowledge, the largest analysis of the social and demographic characteristics of individuals taking part in a community consultation for an EFIC study. It included 22,441 responses – more than then ten times the number in the study by Biros et al [5] – and therefore provides a highly granular picture, with tight confidence intervals. However, it also has limitations. Not all EFIC studies are the same, and respondents’ opinions may vary depending on the type of treatment under investigation. These differences are worthy of further study. Furthermore, our study did not include geography as a possible factor. For example, the opinions of people who live in San Francisco may differ from those in Houston. This will require further investigation. Lastly, the method of survey distribution is important. Obtaining large numbers of “organic” responses (those submitted by individuals who found their own way to the survey website, from the trial website) is difficult, given the large number of requests for opinions that individuals receive every day by social media, email, or otherwise. This type of “consultation fatigue” is increasingly recognized. We therefore use a survey company that maintains extensive lists of individuals, in known locations, who complete surveys for a fee. It is not known how representative this group of survey-takers is or if this population may be biased in some way and this is a further area of investigation.

Our data shows that it is worth continuing to explore the reasons for certain demographics of people to resist or hesitate to be a part of exception from informed consent trials, so that appropriate remedies can be implemented to make community consultation more informative and effective for such groups. Engaging communities with sensitivity and humility can help investigators build trust with hesitant groups by showing respect and willingness to listen, which may foster long-term trust.

Exception from Informed Consent trials are, rightly, highly scrutinized. Community consultation is a key component of establishing trust with the community in which a given study will be conducted. Understanding which individuals, or groups of individuals, are reluctant to engage is important. We recognize that disadvantaged communities often experience a greater burden of injustice; therefore, researchers should make concerted efforts to ensure adequate representation of these groups are met. Lastly, more work is required to better understand how digital inequities could possibly contribute to underrepresented community groups and how better to address these structural biases. Everyone deserves an equal opportunity to learn about research that is occurring in their communities.