Wheeler, Stephanie B.

PURPOSE Lack of race- and age-diversity in clinical trials adversely affects generalizability of trial results and equity of trial access. Research on barriers to representative accrual has focused on patient-level factors. We aimed to define site-level factors affecting the diversity of accrual to a multicenter clinical trial. METHODS Alliance 191901 (GETSET) is a national randomized controlled trial testing interventions to promote breast cancer endocrine therapy adherence, which oversamples Black women and those under age 50 years at 30% thresholds. A secondary study objective is to examine site-level factors associated with accrual of Black and younger participants. At 50% accrual to the parent study, we performed prespecified analyses of association between site characteristics (region, neighborhood racial composition, Alliance membership type, number of recent accruals to cooperative research group trials) and the proportion of Black patients and younger patients accrued. We also examined trajectories of accrual over time for race- and age-specified subgroups. RESULTS The analysis included 124 sites. Among 590 participants, 9.7% were Black, and 22.4% were age <50 years. Neither site type nor historical accrual volume was associated with Black participant recruitment. Southern sites recruited higher proportions of Black participants (19.0% v 7.3% for West, P < .001). Neighborhood racial composition was positively associated with recruitment of Black participants (16.2% at sites from highest Black composition v 1.4% in lowest, P < .01). Recruitment trajectories of Black participants were slower than target rates, whereas those among non-Black participants were faster. CONCLUSION Neighborhood composition and geographic region were predictors of proportions of Black participants accrued to A191901 sites, but site type and historical accrual volume were not. When trials must limit site selection, identifying sites based on neighborhood composition and geography may be an appropriate tool for increasing trial diversity.

PURPOSE Patients with cancer take different types of medications with varying schedules and settings. They are also sometimes instructed to stop medications due to toxicity. To measure self-reported nonadherence in this heterogeneous population, we modified and evaluated a measure originally developed to assess nonadherence to daily oral antihypertensives, the Domains of Subjective Extent of Nonadherence (DOSE-Nonadherence). METHODS The measure was refined in an iterative process incorporating feedback from patient investigators and participant interviews in English and Spanish. Branching logic was added for participant selection of medication administration setting: (1) only home, (2) only clinic, or (3) partly home/partly clinic. Participants reported their adherence to medications taken over a setting-specific reference period (1 week for home, 1 month for clinic medications). Participants who missed medications then reported on reasons for nonadherence. Adherence was dichotomized for analysis (complete adherence v any nonadherence). For participants who received clinic-administered medications, concordance between chart-abstracted and self-reported adherence was evaluated. RESULTS Seventy-three participants completed the measure (68% English; 32% Spanish; 86% female; 44% age ≥60 years). The majority had breast cancer; 64% had metastatic disease. Twenty-six percent (15/58) of participants who received medication in clinic and 24% (11/46) of those who took medication at home reported nonadherence. Participants felt able to respond accurately to both reference periods and perceived all reasons for nonadherence to be relevant. Among participants who completed the final version of the measure for clinic-administered medication, 96% (26/27) accurately reported their adherence compared with chart-abstracted data. CONCLUSION Our results support the validity of the DOSE-Nonadherence-Cancer for assessing cancer treatment nonadherence. This measure can be used to assess nonadherence in patients with cancer receiving a broad array of systemic therapies.

68 Background: Accrual of racial and ethnic minority patients (pts) to clinical trials has been an ongoing challenge in research. Barriers exist at many levels of engagement including trial availability, screening biases, strict enrollment criteria, and structural racism. Lack of trial diversity has led to less generalizable medical evidence, suboptimal data for toxicity and efficacy of cancer treatments, poorer outcomes and continued mistrust by communities of color in research processes that exclude or marginalize them. Site factors and study design strategies associated with successful recruitment of diverse study populations are understudied. Methods: Alliance A191901 is an RCT testing combinations of text and telephone support to promote endocrine therapy adherence among breast cancer survivors. It oversamples Black participants and those < 50 years at thresholds of 30% each. Administrative enrollment data was used to track and describe monthly accrual trajectory by race among the initial 50% of participants (n = 590) and to examine associations between % Black participants accrued and accruing site characteristics, including site type, geographic region, % Black population in the site’s zip code and volume of pts by race accrued to recent (2018-2019) Alliance trials, using χ2 tests. We also examined patterns of Black participant accrual before and after closure of the A191901 study to non-Black participants. Results: At 50% accrual, 124 sites had enrolled at least 1 participant. Among 590 participants, 9.7% were Black and 22.5% were < 50 years. Neither site type nor volume was associated with % Black participants recruited. Sites in the South recruited higher proportions of Blacks (19.0% vs 5.6% for Midwest, the lowest region, p < 0.001). Neighborhood racial composition was positively associated with accrual of Black participants (17.2% at sites from highest Black composition vs. 2.3% in lowest, p < 0.001). Recruitment trajectories of Black participants consistently lagged those of non-Black participants in the first half of the study; however, Black participants recruited per month numerically increased after the study reached target accrual of non-Blacks and closed to non-Black participants. Conclusions: Sites in neighborhoods and geographic regions with larger Black populations recruited higher proportions of Black participants, but site type and volume were not associated with recruitment of Black participants. Closure of non-Black study strata based on predetermined accrual targets may positively impact recruitment trajectories of Black participants, but further study is needed. Site selection based on neighborhood composition and geography may be an appropriate tool for increasing trial diversity. Support: UG1CA189823. Clinical trial information: NCT04379570.