Background: Practice guidelines for nonmuscle invasive (ie, early stage) bladder cancer are ambiguous, resulting in substantial practice variation without a clear patient benefit.
Objectives: To profile urologist practice styles and empirically derive better patterns of use for common bladder cancer services.
Research design: Retrospective cohort.
Subjects: Elderly patients diagnosed with early-stage bladder cancer between January 1, 1992 and December 31, 2005 in Surveillance, Epidemiology, and End Results-Medicare linked data.
Measures: After identifying each patient's treating urologist, we fit multilevel models to obtain reliability-adjusted measures of the urologist's use of surveillance-associated (cytoscopy and urine cytology) and treatment-associated (intravesical therapy) services during the 2 years after diagnosis. We then used the Cox proportional hazards regression to evaluate the association between a patient's risk of bladder cancer death and his urologist's frequency of service use.
Results: Regardless of disease severity, no measurable patient benefit was associated with care delivery by a urologist residing in the highest quartile for cystoscopy or intravescial therapy use. However, maximal intensity of cytology use was associated with a lower risk of bladder cancer death for patients with high-grade stage Ta/Tis (highest vs. lowest intensity quartiles: hazard ratio, 0.73; 95% confidence interval, 0.56-0.95) and stage T1 disease (hazard ratio, 0.59; 95% confidence interval, 0.49-0.72).
Conclusions: Our analysis supports a more tailored approach to patients with early-stage bladder cancer. Further, it serves as an example for applying observational data to characterize better clinical practices in the absence of experimental studies.