Determinants of variation in radical local treatment for men with high-risk localised or locally advanced prostate cancer in England

Background Many factors are implicated in the potential 'under-treatment' of prostate cancer but little is known about the between-hospital variation. Methods The National Prostate Cancer Audit (NPCA) database was used to identify high-risk localised or locally advanced prostate cancer patients in England, between January 2014 and December 2017, and the treatments received. Hospital-level variation in radical local treatment was explored visually using funnel plots. The intra-class correlation coefficient (ICC) quantified the between-hospital variation in a random-intercept multivariable logistic regression model. Results 53,888 men, from 128 hospitals, were included and 35,034 (65.0%) received radical local treatment. The likelihood of receiving radical local treatment was increased in men who were younger (the strongest predictor), more affluent, those with fewer comorbidities, and in those with a non-Black ethnic background. There was more between-hospital variation (P < 0.001) for patients aged >= 80 years (ICC: 0.235) compared to patients aged 75-79 years (ICC: 0.070), 70-74 years (ICC: 0.041), and <70 years (ICC: 0.048). Comorbidity and socioeconomic deprivation did not influence the between-hospital variation. Conclusions Radical local treatment of high-risk localised or locally advanced prostate cancer depended strongly on age and comorbidity, but also on socioeconomic deprivation and ethnicity, with the between-hospital variation being highest in older patients.