Barriers to Clinical Trial Participation: Comparing Perceptions and Knowledge of African American and White South Carolinians

Abstract

Analyzing data from a survey of African American and White residents in South Carolina, this study attempts to understand how to better promote clinical trial participation specifically within the African American population. To explore why participation is lower in the African American population, the authors examined two sets of potential barriers: structural/procedural (limited accessibility, lack of awareness, doctors not discussing clinical trial options, lack of health insurance) and cognitive/psychological (lack of subjective and factual knowledge, misperceptions, distrust, fear, perceived risk). Findings revealed that African Americans were significantly less willing than Whites to participate in a clinical trial. African Americans also had lower subjective and factual knowledge about clinical trials and perceived greater risk involved in participating in a clinical trial. The authors found that lack of subjective knowledge and perceived risk were significant predictors of African Americans’ willingness to participate in a clinical trial. Implications of the findings are discussed in detail.

Notes

¹This response rate was calculated as follows (American Association for Public Opinion Research Response Rate 3):

(Completed Interviews + Partially Completed Interviews)/[(Completed Interviews + Partially Completed Interviews) + (Refusals + Language Barrier + Ill/Senile + Unable to Complete During Fielding Period) + e(Never Answered Numbers + Consistent Answering Machines)].

²In this article, we used the term misperceptions interchangeably with misconceptions in a sense that most of our misperception measures were largely cognitive. We used the term misperceptions because some of previous studies have considered similar measures as perceptual barriers (e.g., Frank, 2004; Harris Interactive, 2001; Quinn et al., 2007).

³To minimize potential multicollinearity between the components of an interaction term and the interaction term itself, ethnicity and nine barrier measures were all z-standardized before entering the regression (Dunlap & Kemery, 1987).

⁴In our study, the two items for willingness to participate (in general and if diagnosed) were combined into a single additive measure as the dependent variable for the regression model, even though the two items were only moderately correlated (r = .427). In a series of additional analyses (not shown in Table 3), these two items were analyzed as two separate dependent variables and regressed separately onto the independent variables to see whether the findings are comparable to the findings from the additive measure reported here. Fear, which was not statistically significant among White respondents (see Table 3), became significant (β = .179, p < .05) when the same Whites were asked their willingness to participate in general. This variable, however, was significant only once in all eight analyses (two measures of willingness × four separate regressions); this exception would not affect our overall conclusion that fear may not be an important barrier to clinical trial participation. Misperceptions, which were not statistically significant among White respondents (see Table 3), also became significant (β = .156, p < .05) when the same Whites were asked their willingness to participate if diagnosed. Again, this variable was significant only once in all eight analyses, and this finding would not change our overall conclusion that misperceptions may not be an important barrier to clinical trial participation. Perceived risk, which was significant among African Americans, was no longer significant (β = .117, p = ns.) when the same African Americans were asked their willingness to participate if diagnosed. Again, this variable was not significant only once in all eight analyses. The interaction between distrust and ethnicity, which was significant earlier, became nonsignificant (β = .065) when the respondents were asked their willingness to participate in general. This finding may indicate that distrust plays a more significant role among Whites, than among African Americans, only when they are diagnosed with a serious disease and have to decide whether to participate. Besides these four cases, all the significant barriers from the additive measure were also significant from the separate dependent measures. At the same time, all nonsignificant barriers from the additive measure were still nonsignificant from the separate dependent measures. We can conclude that the findings from these additional analyses are largely consistent with current findings.

⁵Some of the most recent studies indicate that participation in clinical trials may have been evening out across different races over the recent years. From a telephone survey of cancer patients, Byrne and colleagues (2014) found that African American and Hispanic patients were equally likely as White patients were to be willing to participate in cancer trials. Another recent study (Cottler et al., 2013) found that African Americans were more likely than other ethnic groups were to be interested in participating in medical research. Although no definitive conclusion can be drawn from these two studies, there seems to be reason to further investigate the possibility that the long-lasting racial gap in research participation might have narrowed over the recent years. Nevertheless, it is important to point out that a recent survey of South Carolinian medical researchers (Tanner et al., 2015) has reported that it is difficult, overall, to recruit participants in clinical trials, but particularly difficult to find African American participants compared with White participants.

⁶According to the American Association for Public Opinion Research (Steeh & Piekarski, 2008), cell phone response rates (RR3) are typically in the neighborhood of 10% to 20%, while landline rates are about 5% to 10% higher than cell phone rates.