Serum albumin level is associated with mortality and hospital stays: A real-world data analysis

Background & aims: The value of serum albumin might be underestimated, especially in cancer patients. We thus aimed to evaluate the association between serum albumin level at hospital admission and clinical outcomes in hospitalized patients with cancer. Methods: This is a retrospective, cross-sectional, and real-world data analysis. Hospitalized adult patients with malignant cancer were recruited from two tertiary hospitals. Serum level of albumin, which was measured within 24 h after hospital admission, was the exposure. Length of hospital stays (LOS) was the primary and all-cause in-hospital mortality was the secondary outcomes. Other information, including age, sex, types of cancer, history of hypertension and diabetes, surgery, blood routine test, liver and renal function, and dietary intake, were also abstracted from medical records. Results: A total number of 5187 adult patients with cancer (2949 were men and 2238 women; average age 61.6 +/- 12.4 years and average albumin 40.3 +/- 5.2 g/L) were included. The prevalence of hypoalbuminemia was 12.2 % (634/5187). Older patients, patients with liver injury, anemia, and with high level of WBC were positively, while those with overweight and high level of total triglycerides, were negatively associated with hypoalbuminemia. After adjustment of covariates, hypoalbuminemia was significantly associated with longer LOS in the current study. The increase of 5 g/L in serum level of albumin could result in 1.09 days (95%CI: -1.38, -0.80 days) shorter in LOS. The increase of 5 g/L in serum level of albumin was also associated with 45 % lower in risk of mortality (OR = 0.55; 95% CI: 0.43, 0.7) after fully adjustment. Conclusions: Serum albumin level at admission was associated with both LOS and mortality in patients with malignant cancer. (c) 2024 European Society for Clinical Nutrition and Metabolism. Published by Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.