Abstract 3434: An evaluation of laboratory data at admission for predicting mortality among critically ill patients with cancer

Abstract

Abstract BACKGROUND: Accurately estimating the life expectancy of patients with metastatic or advanced cancer is a crucial step to planning appropriate palliative or supportive care. We analyzed laboratory test results within 2 days of hospital admission for their value in predicting the likelihood of mortality at 14-day hospital admission. METHODS: We retrospectively selected patients 18 years of age or older who had a diagnosis of cancer and were defined as critically ill by their admission to medical ICU, surgical ICU, and palliative or supportive care services in our hospital. We extracted the highest and lowest values of each laboratory test result within 2 days of admission, and the worst value was kept. We used the median and quartiles of results as cutoff points to create dummy variables for estimating odds ratio (OR) at the logistic regression model. We analyzed the following independent variables for their predictive value: age, gender, and the results of the 7 laboratory tests most commonly ordered, including serum creatinine, BUN, total serum albumin, serum LDH, hematocrit, WBC count, and platelet count. The endpoint is death within 14 days after admission, whether in hospital or out of hospital. RESULTS: Of 901 patients identified in development cohort, 573 had solid tumor (84% with metastatic); 328 had hematologic malignancy (37% with relapse); and 44% died within 14 days (ICU vs. supportive care: 49% vs. 40%, P<0.01). Of 424 patients in the validation cohort, 267 had solid tumor (83% with metastatic); 157 had hematologic malignancy (41% with relapse); and 48% died within 14 days (ICU vs. supportive care: 49% vs. 44%, P>0.1). There was no difference in mortality between patients with metastatic and/or relapsed disease compared to the rest of the group for both cohorts (Ps>0.5). The risk of death within 14 days of follow up after admission was associated with age, LDH, WBC count, albumin, and platelet count. In the patients with age ≥ 65 years had an OR of death 1.52 (P<0.01) compared with the patients younger than 55 years old. In the group of patients in the 2nd, 3rd, and 4th quartiles of LDH levels had an OR 2.69 (P<0.01), 2.69 (P<0.01), 4.86 (P<0.01)) compared with the lowest quartile. In the patients in the lowest 1st quartile of albumin had an OR 1.86 (P<0.01) compared with the highest quartile . In the patients in the lowest 1st quartile of platelet count had an OR of death of 1.90 (P<0.01) compared with the highest quartile. In the patients in the 3rd and 4th quartiles of WBC count had an OR of 1.67 (P<0.01), 2.03 (P<0.01) compared with the 2nd quartile. The c-statistics for development and validation cohorts were 0.74 (P<0.01) and 0.70 (P<0.01), respectively. DISCUSSION: Our results suggest that laboratory test results within the first 2 days from admission are valuable factors in predicting short-term mortality for patients with metastatic or advanced cancer and who are at significant risk of dying. Note: This abstract was not presented at the meeting. Citation Format: Lee Cheng, Yvette DeJesus, Alma M. Rodriguez. An evaluation of laboratory data at admission for predicting mortality among critically ill patients with cancer. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 3434. doi:10.1158/1538-7445.AM2015-3434