Establishing a Population-Based Cohort of Nonmuscle Invasive Bladder Cancer Cases to Improve Care by Enhancing Surveillance and Risk Stratification

Background: Among the ~75% of bladder cancer patients who present with superficial (stage \u3c T2) disease, there is substantial variability in the progression to muscle-invasive (stage ≥ T2) disease. An expert panel representing the American Urological Association has expressed the need for population-based data and better predictive tools for bladder cancer surveillance. However, an existing risk-calculator has faced limited adoption, possibly because of reliance on aggregate data from clinical trial populations and failure to design the risk-calculator for use in specific clinical contexts (“use cases”). Methods: We are developing a risk-assessment tool using a population-based cohort from Kaiser Permanente Northwest (KPNW). The cohort includes all superficial bladder cancer cases diagnosed and treated at KPNW from 1990 to 2014. Data from tumor registry, pathology and health plan membership files comprise our database. We conducted in-depth interviews with 8 urologists in a variety of practice settings to identify use cases for improving clinical care through personalized risk-assessment. Our analysis describes key characteristics of our cohort as well as stakeholder suggestions for how a risk-calculator could enhance care. Results: We identified 2,131 cases of superficial bladder cancer. The population had a median of 2.0 pathology records per case, ranging up to 24 records, with 95% of cases having 6 or fewer records. Half of the population did not have a recurrence or progression of bladder cancer during follow-up. Overall, 86% of pathology reports per case reported positive bladder cancer findings. Stakeholder interviews with urologists identified numerous use cases for personalized risk calculations of recurrence and progression. These included: enhancing patient communication and documentation about recurrence/progression; providing guidance about when to discontinue or reduce surveillance for very low-risk patients; identifying and creating scheduling and call-back supports for patients who have high priority for follow-up cystoscopy; and contextualizing bladder cancer mortality risk in the context of competing comorbid conditions. Urologists in general practice reported different use preferences than urologic oncologists. Conclusion: By interviewing a variety of clinicians, we have identified a range of clinical decision support and patient education uses for a personalized risk-calculator for bladder cancer surveillance based on a large population-based cohort reflecting “real world” practice