Variability and Hospital-Level Predictors of Magnetic Resonance Imaging for Elevated Prostate Specific Antigen: Results of a Medicare and American Hospital Association Linkage.

INTRODUCTION

Despite level 1 evidence supporting prostate MRI for suspected prostate cancer, its use remains limited in the United States. The impact of patient-level and hospital-level factors on MRI use after a new elevated PSA among Medicare beneficiaries is unclear. We evaluated these factors to explain variation in MRI utilization.

METHODS

We conducted a retrospective study of Medicare beneficiaries with newly elevated PSA (2019-2023), linked to hospital-level data from the American Hospital Association Annual Survey. The primary outcome was receipt of MRI for elevated PSA, modeled using a generalized linear mixed model with a binomial distribution and logit link function. Patient-level and hospital-level covariates were included, and their relative contributions assessed using Cox and Snell pseudo-R2 statistics.

RESULTS

Of 304,161 patients across 2657 hospitals, the top quartile of hospitals achieved a mean prebiopsy MRI utilization rate of 8.5% (SD: 2.9) and the bottom quartile reported rates below 1.75%. Black patients were more likely to undergo MRI than White patients (OR: 1.15, 95% CI: 1.06-1.26; P < .001), as were those residing in metropolitan areas. Major teaching hospitals were associated with increased MRI use (OR: 1.19; 95% CI: 1.07-1.32; P = .001). Patient-level variables explained 37.2% of the variance in MRI utilization, while hospital-level factors accounted for less than 1%.

CONCLUSIONS

Patient-level factors represent the primary drivers of variation in prebiopsy MRI use, with minimal contribution from hospital characteristics. Efforts to improve equity in prostate cancer care should prioritize reducing patient-level barriers and standardizing referral practices across institutions.