Background: A considerable number of prostate cancer (PCa) patients eligible for expectant management receive definitive treatment. We aimed to investigate the hospital-level contribution to overtreatment in the United States.
Methods: Using the National Cancer Database we identified two nonoverlapping cohorts: (1) men with a life expectancy <10 years harbouring low or intermediate risk PCa (2) men with life expectancy ≥10 years with low-risk PCa. Multivariable mixed models with patient characteristics as fixed and hospital-level intercept as random effect were used to assess the hospital-level risk-adjusted probability of definitive treatment in both groups. Pearson's correlation coefficient was calculated to investigate the correlation between the hospitals probabilities of treating patients of both cohorts.
Results: We found 33,431 men with a life expectancy <10 years and 122,514 men with a life expectancy ≥10 years and low-risk PCa. In the latter, the probability of treatment ranged from 29.0% in the bottom to 90.0% in the top decile and from 35.0% to 88.0% for men with a life expectancy <10 years. Age and race were independent predictors of low-value treatment in both cohorts. The correlation between 1,225 hospitals treating men of both cohorts was strong (Pearson's r = 0.66, P < 0.001).
Conclusion: There is wide hospital-level variability in low-value treatment of men with limited life expectancies and low-risk PCa. Hospitals more likely to treat men with limited life expectancies were more likely to treat men with low-risk PCa and vice versa. Identifying drivers and modifying practice at these hospitals may represent an effective tool for reducing overtreatment.
Keywords: Active surveillance; Overtreatment; Prostate cancer; Public health; Quality of health care.
Copyright © 2019 Elsevier Inc. All rights reserved.