Performance of creatinine- or cystatin C-based equations to estimate glomerular filtration rate in sub-Saharan African populations.

Abstract

Glomerular filtration rate (GFR) is the best index for kidney function; however, the applicability of GFR estimating equations in sub-Saharan African populations remains unclear. In a cross-sectional study of adults living in Kinshasa, Democratic Republic of Congo (n=210) and Abidjan, Ivory Coast (n=284), we evaluated the performance of creatinine and cystatin C-based equations using plasma clearance of iohexol as the reference standard. The race coefficient did not improve the performance of creatinine-based GFR estimates; in fact, both the Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology (CKD-EPI) equations performed better without the race coefficient in participants with GFR ≥60 mL/min/1.73m². The CKD-EPI and Full Age Spectrum (FAS) equations were unbiased and had similar precision (SD of 17.9 versus 19 mL/min/1.73 m²) and accuracy within 30% (P30, 86.7% versus 87.4%) in participants with GFR ≥60 mL/min/1.73m². Both equations performed poorly in the subgroup with measured GFR < 60 mL/min/1.73m² (n=80), but the FAS equation had smaller bias (-4.8 mL/min/1.73m² versus -7.7 mL/min/1.73m² for CKD-EPI) and higher P30 (56.3% versus 31.3% for CKD-EPI). The corresponding equations including cystatin C alone or in combination with creatinine had similar performance. In a sub-Saharan African population, adjustment for race did not improve the performance of GFR estimating equations. The creatinine-based FAS and CKD-EPI equations performed reasonably well and were comparable when GFR was ≥ 60 mL/min/1.73m². Cystatin C did not improve performance. The FAS equation may be preferable when GFR is < 60 mL/min/1.73m², but this should be confirmed in larger studies.