A retrospective study of performance status in oncology patients at diagnosis and over the first year in routine clinical practice.

Abstract

e16521 Background: Measurements of performance status with the Eastern Cooperative Oncology Group Performance Status (ECOG-PS) score are widely used to stratify subjects at clinical trial entry are correlated with treatment outcome; however, ECOG-PS is rarely tracked over time. Longitudinal ECOG-PS measurement between tumor types may provide important data for patient management in community settings, therefore this study examined the distribution of ECOG-PS by tumor type from diagnosis through the subsequent 12 months in community oncology clinics. Methods: We retrieved ECOG-PS scores (0=best, 5=dead) from the electronic medical records (EMRs) of 47 general oncology/hematology clinics across the US. Patients included had ICD-9 diagnostic codes for breast, lymphoma, prostate, colorectal and lung cancers and ECOG-PS score recorded within 14 days of diagnosis date. All patients with at least one ECOG-PS measurement were included in the study. Descriptive statistics of ECOG-PS at diagnosis and by longitudinal month were analyzed as means with standard deviation. ECOG-PS at diagnosis was compared between tumor types using Wilcoxon rank sum test. Results: A total of 28,754 cancer patients were identified from the EMR: 10,489 breast, 4,072 lymphoma, 2,769 prostate, 4,636 colorectal, and 7,252 lung. At baseline, mean (SD) ECOG-PS scores were: breast, 0.56 (0.79); lymphoma, 0.67 (0.83); prostate, 0.75 (0.91); colorectal, 0.85 (0.88); and lung, 1.17 (0.95). Mean scores within tumor types remained stable over the 12 following months. Proportion of patients with ECOG-PS <2 at diagnosis was 89%, 87%, 82%, 81% and 68% for each tumor respectively. Pairwise comparisons of baseline ECOG-PS between tumor types were all statistically significant, p<0.05. Conclusions: This study provides a description of a national sampling of community oncology patients’ performance status. Although significant differences between tumor types were found, the persistence of relatively good ECOG-PS in all 5 tumor types over time contrasts with the poor status of patients in clinical trials. These findings offer oncology providers, data upon which resource planning can be based.