The ability to accurately predict the rate of decline of kidney function over time is essential in ensuring a timely and adequate intervention for individuals such as those with diabetes at high risk of developing chronic kidney disease or of progressing to end stage kidney disease. Predictions may depend on the model of the rate of progression and the source of the serum creatinine result.
The eGFR study comprised a cohort of over 600 Aboriginal and Torres Strait Islander participants recruited across five strata of health, diabetes and kidney function. This analysis explores the trend of decline in eGFR (calculated using the CKD-EPI equation) over a four year period for 385 participants with at least three creatinine measurements from routine local laboratories. Decline estimated with at least three local creatinine measures was compared with decline estimated using two creatinine measures from a single central laboratory.
Linear trends of eGR decline fitted using linear mixed models were compared with non-linear trends fitted using fractional polynomial equations. Analyses were stratified by eGFR, albuminuria categories and diabetes.
Mean age of the participants was 48 years, 64% were female, 49% had diabetes and the median follow-up was 3 years. Results suggested that a linear trajectory was not substantially different from non-linear. The annual eGFR decline in the macroalbuminuria group was 2.9 ml/min/1.73m2 (95% CI: 1.8 to 4.0) for eGFR < 60; 5.7 (95% CI: 3.6-7.7) for eGFR 60-89; 5.4 (95% CI: 2.8-8.0) for eGFR 90-119 and 2.4 (95% CI: 0.9-3.8) for eGFR ≥ 120 ml/min/1.73m2. Decline was similar for those with and without diabetes. Annual declines estimated using two central laboratory creatinine measures were markedly similar.
Prediction of kidney function decline can be reliably derived using linear regression modelling and locally-measured creatinine in Australia is a reliable and robust measure for monitoring kidney function.