ESTIMATES OF GLOMERULAR FILTRATION RATE BASED ON CREATININE AND CYSTATIN C EQUATIONS IN CRITICALLY ILL PATIENTS

Abstract

Introduction and Aim: Accurate assessment of renal function in the critically ill is a complex task. Estimated glomerular filtration rate (eGFR) is the best indicator to help physicians estimate kidney function and determine different stages of kidney disease. eGFR can be calculated using serum Creatinine (S_Cr), or serum Cystatin C (S_CysC), age, body size, race and gender using different equations. We evaluated eGFR based on S_Cr and S_CysC equations in critically ill patients to find an accurate, precise, and less biased equation for GFR estimation.

Methods: This is a single cantered, cross-sectional observational study in critically ill patients older than 18 years staying for over 24-hours in intensive care units. Urinary Creatinine, S_Cr, and S_CysC were measured at 8, 24, and 72-hour intervals. Acute kidney injury (AKI) was defined at admission using the Creatinine definition of the Kidney Disease Improving Global Outcome (KDIGO) guideline. After estimating GFR from Creatinine (eGFR_Cr), Cystatin C (eGFR_CysC), and combining Creatinine and Cystatin C (eGFR_Cr-CysC), results were compared with measured Creatinine clearance (Cr_CL). Cr_CL was defined based on urinary Creatinine, S_Cr, and urinary volume in a 24-hour period.

Results: Forty-three patients were recruited, of which 6 died and 37 alive. eGFR_Cr-24hrs had the highest correlation to measured Cr_CL-24hrs, with correlation coefficient of 0.81 (p<0.001). eGFR_Cr-72hrs had the highest accuracy at 30% and 50% of measured Cr_CL between 57.69% and 69.23% respectively. However, the eGFR_CysC-24hrs equation was more precise with the lowest bias of 0.94. Logisticregression-analysis reveals that p values for eGFRCr-24hrs, CrCL-24hrs, eGFRCr-CysC-24hrs, eGFRCr-72hrs, eGFR_{Cr-CysC-72hrs} equations were ? 0.05, which helps diagnose AKI. Logistic regression was >0.05 for in- hospital-mortality prediction, addressing none of the equations led to death prediction.

Conclusions: eGFR_Cr-24hrs had the highest correlation to measured Cr_CL-24hrs, with correlation coefficient of 0.81 (p<0.001). The most accurate equation was eGFR_Cr-72hrs However, eGFR_CysC-24hrs equation had the lowest bias and was relatively more precise (60.0?ml/min). Using logistic regression, most of the equations contributed to diagnosing AKI. However, none of the equation predicted in-hospital-mortality.

Acknowledgement: I would first like to express my heartfelt thanks to my supervisor Associate Professor Dr Azrina Md Ralib. A more supportive and considerate supervisor I could not have asked for. Secondly, I want to express my immense gratitude to the Afghan Government, my sponsor (HEDP) for allocating 100% sponsorship for my Master degree, and many thanks to everyone who assisted directly (family members) and indirectly in this whole journey.