Prevalence, concordance and associations of chronic kidney disease by five estimators in South Africa

Abstract Background To determine the prevalence, distribution, concordance and associations of chronic kidney disease (CKD) determined by five glomerular filtration rate (GFR) formulae in urban black residents of Cape Town. Methods Data collection in this cross-sectional study included interviews, clinical measurements and biochemical analyses, including serum creatinine and cystatin C levels. GFR was based on the CKD Epidemiology Collaboration (CKD-EPI) equations (CKD-EPI creatinine (CKD-EPIcr), CKD-EPI cystatin C (CKD-EPIcys), CKD-EPI creatinine-cystatins (CKD-EPIcr-cys)), Modification of Diet in Renal Disease (MDRD) and Cockcroft-Gault formula (CGF). GFR < 60 mL/min/1.73 m2 defined CKD. Results Among 392 men and 700 women, mean GFR, was between 114.0 (CKD-EPIcr) and 135.4 mL/min/1.73 m2 (CGF) in men, and between 107.5 (CKD-EPIcr-cys) and 173.4 mL/min/1.73 m2 (CGF) in women. CKD prevalence ranged from 2.3% (CKD-EPIcr and MDRD) to 5.1% (CKD-EPIcys) in men and 1.6% (CGF) to 6.7% (CKD-EPIcr-cys) in women. The kappa statistic was high between CKD-EPIcr and MDRD (0.934), and CKD-EPIcys and CKD-EPIcr-cys (0.815), but fair-to-moderate between the other eqs. (0.353–0.565). In the basic regressions, older age and body mass index ≥30 kg/m2, but not gender, were significantly associated with CKD-EPIcr-defined CKD. In the presence of these three variables, hypertension, heart rate ≥ 90 beats/minute, diabetes and low-density lipoprotein cholesterol were significant predictors of prevalent CKD. Conclusions Varying CKD prevalence estimates, because of different GFR equations used, underscores the need to improve accuracy of CKD diagnoses. Furthermore, screening for CKD should be incorporated into the routine assessment of high-risk patients such as those with hypertension or diabetes.