Contemporary real-world associations between performance status and clinical outcomes in patients with cancer: A retrospective cohort study.

Abstract

6578 Background: The ECOG-PS (Eastern Cooperative Oncology Group Performance Status) scale is often used to guide cancer care, but the degree to which it predicts contemporary real-world clinical outcomes, in general and within certain patient groups, is relatively unknown. This retrospective cohort study examined associations between ECOG-PS levels and adverse outcomes in cancer patients with diverse patient characteristics. Methods: Various patient characteristics and nurse-rated ECOG-PS scores (range: 0-4) were recorded for all 21,730 adult patients with cancer receiving intravenous systemic therapy between 01/01/2017 and 12/31/2019 at 18 Kaiser Permanente Northern California cancer centers. Differences in baseline characteristics by ECOG-PS scores were evaluated using chi-square tests for categorical variables and ANOVA for continuous variables. Univariable and multivariable Cox Proportional Hazard models were used to test the ability of ECOG-PS to predict the occurrence of adverse clinical outcomes, including 1-month emergency department (ED) visits and hospitalizations, and 6-month mortality. Results: Overall, 42.5% of patients had ECOG-PS = 0, 42.5% had ECOG-PS = 1, 10.5% had ECOG-PS = 2, 4% had ECOG-PS = 3 and 0.4% had ECOG-PS = 4. Most patients were women (58%), non-Hispanic White (61%), English speakers (93%) and married/domestic partners (63%). African Americans, men, older patients, and those with higher Charlson comorbidity index or Stage IV cancer were found to have higher ECOG-PS levels (all p < 0.001). In multivariable analysis, ECOG-PS of 3-4 were associated with higher ED visits (HR 3.85, 95% CI [3.47-4.26]), hospitalizations (HR 4.7, [4.12-5.36]) and mortality (HR 7.34, [6.64-8.11]), compared with ECOG-PS = 0. Upper gastrointestinal (GI) and Stage IV cancers were associated with a higher risk of ED (upper GI: HR 2.39, [2.12-2.68], (stage IV: HR 1.31, [1.21-1.42]), hospitalization (HR 2.67, [2.27-3.13], (HR 1.51, [1.35-1.68]), and mortality rates (HR 3.37, [2.97-3.81], (HR 1.82, [1.68-1.98]), compared to Breast and Stage I cancers; however, advanced age was not associated with these outcomes. Interactions between ECOG-PS and cancer type as well as ECOG-PS and age group were statistically significant (p < 0.001), such that ECOG-PS was more predictive of adverse outcomes in younger patients and those with breast cancer. Conclusions: In this contemporary real-world cohort, multivariable analysis showed that ECOG-PS, cancer type and stage were strong predictors of ED visits, hospitalizations and mortality; however, advanced age was not. These results also show that ECOG-PS is more predictive of clinical outcomes in certain patient groups. Our findings may have implications on the use of ECOG-PS for clinical decision making.