The Effect of Other-cause Mortality Adjustment on Access to Alternative Treatment Modalities for Localized Prostate Cancer Among African American Patients

Background: African American (AA) men might be less likely to benefit from certain treatment types for localized prostate cancer (PCa). Objective: To test treatment rate differences between AA and Caucasian patients with clinically localized PCa, with and without adjustment for other-cause mortality (OCM). Design, setting, and participants: Within the Surveillance Epidemiology and End Results (SEER) database (2004–2014), we identified 260 309 (94.0%) Caucasian and 15 534 (6.0%) AA patients with PCa. Intervention: Radical prostatectomy (RP), external beam radiotherapy (EBRT), brachytherapy (BT), combination of BT and EBRT (BT + EBRT), or nonlocal treatment (NLT). Outcome measurements and statistical analysis: We used multivariable logistic regression to assess treatment rates according to race, with or without adjustment for OCM risk according to D'Amico risk classification. OCM was defined using a multivariable Cox regression model, developed using a 50% random sample and validated using the remaining 50%. Results and limitations: Before OCM adjustment, AA patients were less likely to receive RP regardless of D'Amico risk (odds ratio [OR] 0.54 for low risk [LR], 0.45 for intermediate risk [IR], and 0.43 for high risk [HR]) and were less likely to receive BT if D'Amico intermediate risk (OR 0.84) or high risk (OR 0.89). After OCM risk adjustment, AA men were still less likely to receive BT (OR 0.53 for LR, 0.32 for IR, 0.22 for HR) and EBRT (OR 0.74 for LR, 0.69 for IR, 0.83 for HR), but were no longer less likely to receive RP (OR 2.58 for LR, 3.07 for IR, 2.67 for HR) regardless of their D'Amico risk classification. The Cox model of OCM risk was 74.9% accurate in the validation cohort. Conclusions: For AA men, rates of treatment for localized PCa depend on OCM risk. Lack of OCM risk adjustment may incorrectly suggest that some treatments are delivered at a lower rate than for Caucasians, and vice versa. Patient summary: Our study critically appraised the validity of reported prostate cancer treatment rates for African American men when adjustment for other-cause mortality was not performed. African Americans might be less likely to receive certain treatment types such as radical prostatectomy, external beam radiotherapy, brachytherapy, or a combination of brachytherapy and external beam radiotherapy for localized prostate cancer. Treatment rates that were unadjusted for other-cause mortality (OCM) may provide a biased result. Application of a model predicting OCM allowed us to examine treatment rates after OCM adjustment. Using this approach, African Americans were no longer less likely to benefit from radical prostatectomy. The differences between OCM-adjusted and unadjusted treatment rates recorded for African Americans and Caucasians strongly suggest an important effect by OCM that requires consideration.