60 VALIDATING A CLAIMS-BASED METHOD FOR ASSESSING LONG TERM URINARY ADVERSE EFFECTS OF PELVIC RADIOTHERAPY; A PILOT STUDY

INTRODUCTION AND OBJECTIVES: Severe (Grade 3 and 4) urinary adverse effects (AEs) of pelvic radiotherapy (RT) can occur years after treatment for cancers of the prostate, bladder, uterus, cervix, or rectum. These can include such complications as urethral stricture requiring dilation, ureteral stricture requiring stenting or hemorrhagic cystitis requiring clot evacuation and fulguration. Such RT AEs may be under-reported in single institution series as they can occur late after RT and are often not cared for by the radiation oncologist. We sought to validate a claims-based algorithm for detecting such RT AEs in order to inform future research using administrative datasets which could take advantage of a larger population and extended follow-up. METHODS: An institutional billing analysis was performed in order to identify patients managed with RT (brachytherapy and/or external beam radiotherapy) for prostate or cervical cancer at the University of Minnesota, between 2000-2006. Prostate (n 225) and cervical cancer (n 181) were chosen to represent common pelvic cancers treated with high doses of RT in men and women, respectively. A priori we identified CPT procedural codes consistent with treatment for severe urinary RT AEs (such as those described above). A retrospective chart review and a billing (i.e. “claims”) analysis were both performed in order to detect procedures used to treat urinary RT AEs. The accuracy of the claims-based algorithm was compared to the gold standard, chart review. RESULTS: On chart review, 20 severe urinary RT AEs were detected among 406 patients with non-metastatic cancer at diagnosis (4.9%; 3.6% in prostate cancer and 6.6% in cervical cancer). Mean follow-up was 3.3 years. The most common AEs were urethral stricture (24% of all AEs, all after prostate RT) and ureteral stenosis (26%, all after cervical RT). The sensitivity and specificity of the claims-based analysis were 90% (95% CI 68%-99%) and 99% (98%-99%), respectively. The positive and negative predictive values were 90% (68%-98%) and 99% (98%-99%), respectively. CONCLUSIONS: We have demonstrated high validity of our claims-based algorithm for detecting treatment of severe urinary RT AEs. We confirm that with institutional chart review, follow-up after RT is limited and the frequency of detected RT AEs is low, hampering the power of single institution research studies. Future studies will use claims-based methods in large administrative datasets to assess the long-term risk of severe urinary RT AEs and compare risk factors.