DO DISPARITIES IN SURGICAL REFERRALS INDICATE PREJUDICE? EVIDENCE FROM A PHYSICIAN SITUATIONAL CHOICE EXPERIMENT
Author(s)
Milla Frieman, BS1, Reed Johnson, PhD2, Jessie Sutphin, MA3, Rachel Kelz, MD, MSCE4.
1Center for Surgery and Health Economics, University of Pennsylvania, Philadelphia, PA, USA, 2Professor, Duke School of Medicine, Duke Clinical Research Institute, Durham, NC, USA, 3Duke Clinical Research Institute, Mooresville, NC, USA, 4University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA.
1Center for Surgery and Health Economics, University of Pennsylvania, Philadelphia, PA, USA, 2Professor, Duke School of Medicine, Duke Clinical Research Institute, Durham, NC, USA, 3Duke Clinical Research Institute, Mooresville, NC, USA, 4University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA.
OBJECTIVES: Racial disparities in healthcare have been well-documented, but correlation with socioeconomic status (SES) make determining causality difficult. This study aimed to isolate potential racial and SES disparities in physicians’ stated current surgical-referral priorities.
METHODS: Physician judgments were elicited in the context of training a hypothetical large-language model to incorporate current perspectives on surgical-referral criteria. The study design employed a situational discrete-choice experiment, where the choice “situations” incorporated a 2x2 patient-profile design (White/High SES, White/Low SES, Black/High SES, Black/Low SES). Referral options varied by surgeon and hospital quality ratings as well as availability of hospital-based patient-navigation or on-call-nurse support programs and community-based childcare or transportation programs.
RESULTS: Latent-class analysis identified 2 preference classes. A dummy-coded model omitted White/High SES profile. Relative to that profile, both class preferences weighted surgical and hospital ratings higher for the two Black patient profiles, although the Class-2 ratings were smaller for the low-SES profile. Class-1 preferences also weighted the hospital programs for the Black profiles more heavily, while Class-2 preferences weighted the programs no differently. However, Class-2 preferences favored the community program for both Black profiles more, while Class 1 did not favor the program differently for Black/high SES and only slightly favored the program more for the Black/low SES profile. Within the 3 included profiles, t-tests for both classes indicated no statistically significant differences in the quality rating parameters between White/Low SES and either of the Black SES profiles.
CONCLUSIONS: While we identified heterogeneity in physicians’ referral judgments controlled for race and SES differences, estimated importance-weight parameters generally indicated statistical support for Black referrals to surgeons and hospitals with better quality ratings compared to White/High SES referrals. However, the results identified different perceptions on whether and which hospital and community programs would benefit Black referrals with different SES levels.
METHODS: Physician judgments were elicited in the context of training a hypothetical large-language model to incorporate current perspectives on surgical-referral criteria. The study design employed a situational discrete-choice experiment, where the choice “situations” incorporated a 2x2 patient-profile design (White/High SES, White/Low SES, Black/High SES, Black/Low SES). Referral options varied by surgeon and hospital quality ratings as well as availability of hospital-based patient-navigation or on-call-nurse support programs and community-based childcare or transportation programs.
RESULTS: Latent-class analysis identified 2 preference classes. A dummy-coded model omitted White/High SES profile. Relative to that profile, both class preferences weighted surgical and hospital ratings higher for the two Black patient profiles, although the Class-2 ratings were smaller for the low-SES profile. Class-1 preferences also weighted the hospital programs for the Black profiles more heavily, while Class-2 preferences weighted the programs no differently. However, Class-2 preferences favored the community program for both Black profiles more, while Class 1 did not favor the program differently for Black/high SES and only slightly favored the program more for the Black/low SES profile. Within the 3 included profiles, t-tests for both classes indicated no statistically significant differences in the quality rating parameters between White/Low SES and either of the Black SES profiles.
CONCLUSIONS: While we identified heterogeneity in physicians’ referral judgments controlled for race and SES differences, estimated importance-weight parameters generally indicated statistical support for Black referrals to surgeons and hospitals with better quality ratings compared to White/High SES referrals. However, the results identified different perceptions on whether and which hospital and community programs would benefit Black referrals with different SES levels.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
HSD72
Topic
Health Service Delivery & Process of Care
Disease
No Additional Disease & Conditions/Specialized Treatment Areas