A reliable interpretation of albumin levels is essential when assessing nutrition in elderly people, but this is complex as it is affected by a number of parameters. The main objective of this study was to evaluate whether a correction formula proposed for prolonged inflammation could be extrapolated to acute biological inflammation situations: corrected albuminemia = measured albuminemia + CRP/25. This prospective, single-centre observation study included patients over 65 years of age who were the subject of an unscheduled hospitalisation in the geriatrics and internal medicine departments of Grenoble University Hospital, excluding carriers of active neoplasia, hepatic insufficiency, nephrotic syndrome and those who continued to present with an acute biological inflammation on the eighth day. Clinical and biological samples were taken on the first and eighth days. The primary objective was the comparison of albumin levels, corrected using the formula on day (d) 1, with albumin levels measured on d8. One hundred and seventy-five patients were analysed. Average CRP was 64 (3; 324) mg/L on d1 and 24 (3; 99) mg/L on d8. Between the corrected albumin levels on d1 and albumin levels measured on d8, the correlation was ρ = 0.58, 95% CI [0.47; 0.67], P < 0.001, with a mean difference of 2.9 (-13.5; 18) mg/L, 95% CI [-3.68-2.20], P < 0.001. Between the albumin levels measured on d1 and d8, the correlation was ρ = 0.74, 95%CI [0.66; 0.80], P < 0.001, with a mean difference of 0.4 (-14; 11) mg/L, 95%CI [-0.24; 1.02], P = 0.23. The proposed formula cannot be extrapolated to acute inflammatory situations. Albumin levels appear to be stable during acute biological inflammation. Albumin levels could be used as a reference for biological nutritional assessments and be integrated into the patient's clinical history.
Keywords: albumin levels; elderly; inflammation; malnutrition; nutritional status.