Back to Annual Meeting

Back to Annual Meeting

Abstract #148777

• Audio (mp3) recording
• Slides (pdf) or Handout
• Multimedia recording

George J. Stukenborg, PhD¹, Douglas P. Wagner, PhD¹, Kerry Kilbridge, MD², M. Norman Oliver, MD³, Michael J. Fisch, MD⁴, Scott M. Strayer, MD³, and Dan Theodorescu, MD, PhD⁵. (1) Public Health Sciences, University of Virginia School of Medicine, Blake Center Room 400B, 1224 West Main Street, Charlottesville, VA 22908-0821, 434-924-8649, gstukenborg@virginia.edu, (2) Massachusetts General Hospital, Medical Practices Evaluation Center, 50 Staniford St., 9th floor, St 955, Boston, MA 02114, (3) Department of Family Medicine, University of Virginia School of Medicine, Box 800729, Charlottesville, VA 22908, (4) M.D. Anderson Community Clinical Oncology Program, University of Texas, M.D. Anderson Cancer Center, 1100 Holcombe Blvd, Houston, TX 77030-4009, (5) Department of Urology, University of Virginia School of Medicine, Box 800422, Charlottesville, VA 22908

Purpose: Estimates of life expectancy are essential to prostate cancer treatment decision making, and comorbid diseases are important independent predictors of patient life expectancy after diagnosis. Adaptations of the Charlson index have been commonly used to measure differences in comorbid disease in studies that examine mortality risk. Prior studies suggest that further research is needed to develop measures of comorbid disease effects on mortality that are more valid and more complete. Methods: This study uses SEER-Medicare data and multivariable logistic regression analysis to identify which diagnoses and related conditions recorded up to 12 months prior to the date of prostate cancer diagnosis are the best predictors of five-year mortality among men age 66 and older with localized prostate cancer. The regression model's validated explanatory power and capacity to discriminate between survivors and decedents is compared to that obtained by a regression model using the Charlson index score method, along with the difference in the estimated effects of comorbid disease on mortality risk between these models. Results: Adjustments for the effects of comorbid disease measured by empirically selected ICD-9-CM diagnoses improve model statistical performance twice as much as adjustments made using the Charlson score method. When applied to individual patients, empirically selected ICD-9-CM diagnosis codes obtain substantially different estimates of five year survival than adjustments made using the Charlson method. Conclusion: Five year mortality risk at diagnosis for older men with localized prostate cancer is better estimated by using empirically selected ICD-9-CM diagnosis codes to measure the effects of comorbid disease.

Learning Objectives:

• Demonstrate methods for developing and validating mortality risk adjustment models using data from a large cancer registry database with merged records from Medicare administrative data

Keywords: Outcomes Research, Cancer

Presenting author's disclosure statement:

Any relevant financial relationships? No
Any institutionally-contracted trials related to this submission?

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.

• Audio (mp3) recording
• Slides (pdf) or Handout
• Multimedia recording