Comorbidity and survival in cancer patients receiving palliative care

Abstract

9066 Background: Comorbidity has received increasing attention in the assessment of patients with early stage cancer, or at diagnosis. We studied whether three indices of comorbidity, the Charlson Comorbidity Index (CMI), the Cumulative Illness Rating Scale (CIRS), and the Kaplan Feinstein Index (KFI) add prognostic information for cancer patients receiving palliative care. Methods: In an IRB approved protocol, 103 patients with advanced cancer were seen at the time they were starting palliative care. They had a Karnofsky Performance Status (KPS) determination, and were followed longitudinally. Comorbidity scores were coded from the medical record. At this time, all patients had died and survival analyses were performed. Results: The median age was 69 years (range 41–87), median Karnofsky Performance Status (KPS) was 70% (range 20–90); primary sites were lung 41 pts (40%), prostate 23pts (22%), colorectal 10 pts (10%), other cancers 29 pts (28%). Median survival was 111 days (range 4–1,145 days). Median CMI was 10 (range 4–14), CIRS15 4 (2–5), CIRS16 9 (4–12), CIRS17 2.3 (1.5–3.33), CIRS18 1 (0–3), KFI 2 (0–3). In univariate survival analyses, when bisected by median values, the KPS, age, CMI, and subscales of the CIRS (CIRS 16, CIRS 17, CIRS18) were significantly related to survival, but not the KFI. In multivariate Cox regression analyses that included KPS (p<0.0001) and age (p<0.003) and a comorbidity index, the CMI (p<0.0001), and certain subscales of the CIRS were independently predictive of survival, specifically the CIRS 15 (p<0.0001), CIRS16 (p<0.0001), CIRS 17 (p<0.0001), and CIRS18 (p<0.0001). The primary site was not an independent survival predictor. Conclusion: In patients with advanced cancer receiving palliative care, measures of comorbidity may contribute to refining estimates of prognosis and ultimately to health care resource utilization. The optimal comorbidity measure remains to be determined. These results will be confirmed in larger populations. Supported in part by the Soros Open Society Institute Project Death in America and VA HSRD IIR 02–103 No significant financial relationships to disclose.