Abstract
BACKGROUND: Ultrasound-based evaluation of endometrial thickness to rule out endometrial cancer in patients with postmenopausal bleeding (PMB) is not sufficiently validated in the non-Hispanic Black (NHB) population. Differences in ultrasound performance between NHB and non-Hispanic White (NHW) patients may contribute to documented disparities in cancer-associated survival. METHODS: We developed a decision-tree model, informed by literature and institutional data, comparing an immediate biopsy-based strategy (BBS) to an ultrasound-based strategy (UBS) to evaluate PMB in NHB and NHW patients. Primary outcomes were the probability of a missed cancer diagnosis and incremental biopsies per additional cancer detected. Fibroid prevalence, endometrial visibility, and race were modeled using institutional patient-level data. Prompt endometrial sampling following abnormal ultrasound (perfect follow-up) was assumed; institutionally observed follow-up rates were alternatively modeled. RESULTS: In a simulated cohort of 10,000 patients with PMB, UBS missed 109 (95 % prediction interval (PI) 85-159) cancer diagnoses compared to 70 missed for BBS. Compared to UBS, BBS resulted in 123 (95 % PI 45-265) biopsies/additional cancer detected for NHB and 155 (95 % PI 47-398) for NHW. Under observed follow-up rates, BBS resulted in 33 (95 % PI 26-38) biopsies/additional cancer detected for NHB and 44 (95 % PI 30-54) for NHW. NHB derived higher value (fewer biopsies/additional cancer detected) from BBS than NHW, particularly with fibroids present. CONCLUSION: BBS has higher value compared to UBS under observed/imperfect levels of follow-up and enhanced value when fibroids are present and among NHB patients. These results suggest that endometrial sampling should be offered at time of initial evaluation for PMB.