Ángel Candela-Toha1, María Carmen Pardo2, Teresa Pérez3, Alfonso Muriel4, Javier Zamora5. 1. Servicio de Anestesia y Reanimación, Hospital Universitario Ramón y Cajal (IRYCIS) Consorcio FRA (CIFRA) , Madrid, España. Electronic address: angel.candela@salud.madrid.org. 2. Departamento de Estadística e Investigación Operativa, Facultad de Matemáticas, Universidad Complutense de Madrid , Madrid, España. 3. Departamento de Estadística e Investigación Operativa III, Facultad de Estudios Estadísticos, Universidad Complutense de Madrid , Madrid, España. 4. Unidad de Bioestadística Clínica, Hospital Universitario Ramón y Cajal (IRYCIS). CIBER Epidemiología y Salud Pública (CIBERESP), Madrid, España. 5. Unidad de Bioestadística Clínica, Hospital Universitario Ramón y Cajal (IRYCIS). CIBER Epidemiología y Salud Pública (CIBERESP), Madrid, España; Barts and The London School of Medicine and Dentistry, Queen Mary University, Londres, Reino Unido.
Abstract
BACKGROUND: and objective Acute kidney injury (AKI) diagnosis is still based on serum creatinine and diuresis. However, increases in creatinine are typically delayed 48h or longer after injury. Our aim was to determine the utility of routine postoperative renal function blood tests, to predict AKI one or 2days in advance in a cohort of cardiac surgery patients. PATIENTS AND METHODS: Using a prospective database, we selected a sample of patients who had undergone major cardiac surgery between January 2002 and December 2013. The ability of the parameters to predict AKI was based on Acute Kidney Injury Network serum creatinine criteria. A cohort of 3,962 cases was divided into 2groups of similar size, one being exploratory and the other a validation sample. The exploratory group was used to show primary objectives and the validation group to confirm results. The ability to predict AKI of several kidney function parameters measured in routine postoperative blood tests, was measured with time-dependent ROC curves. The primary endpoint was time from measurement to AKI diagnosis. RESULTS: AKI developed in 610 (30.8%) and 623 (31.4%) patients in the exploratory and validation samples, respectively. Estimated glomerular filtration rate using the MDRD-4 equation showed the best AKI prediction capacity, with values for the AUC ROC curves between 0.700 and 0.946. We obtained different cut-off values for estimated glomerular filtration rate depending on the degree of AKI severity and on the time elapsed between surgery and parameter measurement. Results were confirmed in the validation sample. CONCLUSIONS: Postoperative estimated glomerular filtration rate using the MDRD-4 equation showed good ability to predict AKI following cardiac surgery one or 2days in advance.
BACKGROUND: and objective Acute kidney injury (AKI) diagnosis is still based on serum creatinine and diuresis. However, increases in creatinine are typically delayed 48h or longer after injury. Our aim was to determine the utility of routine postoperative renal function blood tests, to predict AKI one or 2days in advance in a cohort of cardiac surgery patients. PATIENTS AND METHODS: Using a prospective database, we selected a sample of patients who had undergone major cardiac surgery between January 2002 and December 2013. The ability of the parameters to predict AKI was based on Acute Kidney Injury Network serum creatinine criteria. A cohort of 3,962 cases was divided into 2groups of similar size, one being exploratory and the other a validation sample. The exploratory group was used to show primary objectives and the validation group to confirm results. The ability to predict AKI of several kidney function parameters measured in routine postoperative blood tests, was measured with time-dependent ROC curves. The primary endpoint was time from measurement to AKI diagnosis. RESULTS: AKI developed in 610 (30.8%) and 623 (31.4%) patients in the exploratory and validation samples, respectively. Estimated glomerular filtration rate using the MDRD-4 equation showed the best AKI prediction capacity, with values for the AUC ROC curves between 0.700 and 0.946. We obtained different cut-off values for estimated glomerular filtration rate depending on the degree of AKI severity and on the time elapsed between surgery and parameter measurement. Results were confirmed in the validation sample. CONCLUSIONS: Postoperative estimated glomerular filtration rate using the MDRD-4 equation showed good ability to predict AKI following cardiac surgery one or 2days in advance.