Understanding Active Surveillance for Prostate Cancer

Abstract

Purpose: To assess how active surveillance for prostate cancer is apportioned across specialties and how testing patterns and transition to treatment vary by specialty.

Methods: We used a 20% national sample of Medicare claims to identify men diagnosed with prostate cancer from 2010 through 2016 initiating surveillance (N = 13,048). Patients were assigned to the physician responsible for the bulk of surveillance care based on billing patterns. Freedom from treatment was assessed by specialty of the responsible physician (urology, radiation oncology, medical oncology, and primary care). Multinomial logistic regression models were used to examine associations between specialty and treatment patterns.

Results: Urologists were responsible for surveillance in 93.7% of patients in 2010 and 96.2% of patients in 2016 (P for trend = .01). Testing patterns varied by specialty. For example, patients of medical oncologists had more frequent prostate-specific antigen testing compared with patients of urologists (1.85 v 2.39 tests per year, respectively; P < .01). Three years after diagnosis, a significantly smaller proportion of patients managed by radiation oncologists (64.3%) remained on surveillance compared with patients managed by other physicians (75.8%-79.5%; P < .01). Although radiation was the most common treatment among all men who transitioned to treatment, a disproportionate percentage of patients followed by radiation oncologists (28.9%) ultimately underwent radiation compared with patients followed by other physicians (15.1%-15.4%; P < .01).

Conclusion: Nontrivial percentages of patients on active surveillance are managed by physicians outside of urology. Given the interspecialty variations observed, efforts to strengthen the evidence underlying surveillance pathways and to engage other specialties in guideline development are needed.

FIG 1.

Kaplan-Meier plot of freedom from treatment on surveillance according to the specialty of the responsible physician. Source: Authors' analysis of 20% sample of national Medicare data of men diagnosed with prostate cancer between 2010 and 2016 who initiated active surveillance. Notes: Freedom from treatment differed markedly based on the specialty of the physician performing surveillance (P = .0003, log-rank). Three years after diagnosis, a significantly smaller percentage of patients followed by radiation oncologists (64.3%) remained on surveillance compared with those followed by other specialties (range, 75.8%-79.5%; P = .0003).

FIG 2.

Percentage of patients receiving treatment and treatment type according to the specialty of the responsible physician, adjusted for age, race, socioeconomic class, education, place of residence, and comorbidity. Source: Authors' analysis of 20% sample of national Medicare data of men diagnosed with prostate cancer between 2010 and 2016 who initiated active surveillance. Notes: A significantly larger percentage of patients under surveillance by radiation oncologists received radiation therapy compared with those under surveillance by nonradiation oncologists (28.9% v 15.1% to 15.4%; P < .01 for all specialties compared with radiation oncology).

FIG A1.

Flowchart for the analytic sample. Source: Authors' analysis of 20% sample of national Medicare data of men diagnosed with prostate cancer between 2010 and 2016, and followed through 2017. Notes: A total of 13,048 patients are included in analytic sample. E&M, evaluation and management; HMO, health maintenance organization.

FIG A2.

Percentage of patients on surveillance according to the specialty of the responsible physician from diagnosis date until 3 months before censorship. Source: Authors' analysis of 20% sample of national Medicare data of men diagnosed with prostate cancer between 2010 and 2016 who initiated active surveillance, and followed through 2017. Notes: Urologists performed surveillance for 93.7%-96.2% of patients in surveillance, with an increasing involvement over time (P for trend = .01, Cochran-Mantel-Haenszel test). ^aPY, patient-year (estimate of the actual time at risk in years that all patients contributed to this study).