Abstract
Background Race influences medical decision making, but its impact on advanced heart failure therapy allocation is unknown. We sought to determine whether patient race influences allocation of advanced heart failure therapies. Methods and Results Members of a national heart failure organization were randomized to clinical vignettes that varied by patient race (black or white man) and were blinded to study objectives. Participants (N=422) completed Likert scale surveys rating factors for advanced therapy allocation and think-aloud interviews (n=44). Survey results were analyzed by least absolute shrinkage and selection operator and multivariable regression to identify factors influencing advanced therapy allocation, including interactions with vignette race and participant demographics. Interviews were analyzed using grounded theory. Surveys revealed no differences in overall racial ratings for advanced therapies. Least absolute shrinkage and selection operator regression selected no interactions between vignette race and clinical factors as important in allocation. However, interactions between participants aged ≥40 years and black vignette negatively influenced heart transplant allocation modestly (-0.58; 95% CI, -1.15 to -0.0002), with adherence and social history the most influential factors. Interviews revealed sequential decision making: forming overall impression, identifying urgency, evaluating prior care appropriateness, anticipating challenges, and evaluating trust while making recommendations. Race influenced each step: avoiding discussing race, believing photographs may contribute to racial bias, believing the black man was sicker compared with the white man, developing greater concern for trust and adherence with the black man, and ultimately offering the white man transplantation and the black man ventricular assist device implantation. Conclusions Black race modestly influenced decision making for heart transplant, particularly during conversations. Because advanced therapy selection meetings are conversations rather than surveys, allocation may be vulnerable to racial bias.