Characterization of kidney disease in sub-Saharan Africa

Background: Chronic kidney disease (CKD) is an established cause of morbidity and mortality worldwide. While methods to determine kidney function are well established in Western populations, there are good reasons to believe these methods may not translate well to sub-Saharan Africa (sSA), meaning the overall burden of CKD is uncertain. The aim of this thesis was to determine the prevalence and mortality associated with renal impairment using conventional estimated glomerular filtration rate (eGFR) approaches, we also compared performance of measured GFR (mGFR) to eGFR and used mGFR based model to impute CKD prevalence in sSA. Methods: We determined the prevalence and all-cause mortality from impaired renal function [defined as eGFR) <60mls/min/1.73m2 within a population-based cohort among adults of 18 years and above in rural Uganda. Working with the African Research on Kidney Disease Network (ARK) a consortium of three community-based cohorts from Malawi, Uganda and South Africa, we stratified participants by level of renal function. We intravenously injected 5millilitres bolus of exogenous iohexol and drew venous samples from the contralateral arm at 5, 120, 180 and 240- minute time points to determine the mGFR. We compared the performance of existing equations to mGFR and used a model to impute kidney function based on mGFR. Results: In Uganda, among 5,979 participants, we found an overall prevalence of eGFR <60 ml/min per 1.73 m2 of 1.6% (95% CI 1.34–1.99) with up to 1,089 (18.2%) having an eGFR <90 ml/min per 1.73 m2 in a predominantly young population. Older age, hypertension and anemia were independently associated with impaired renal function. In adjusted analyses, participants with baseline eGFR ≤45mls/min/1.73m2 had six-fold higher mortality compared to those with eGFR ≥90mls/min/1.73m2 (HR 6.12 (95% CI 2.27-16.45)) with strong evidence of a linear trend for risk of mortality as renal function declined (p<0.001). Among the 2,578 participants with mGFR and 2433 with cystatin C eGFR from the ARK study, we found that that all eGFR equations overestimate GFR compared to mGFR or cystatin C across the three countries and this was worsened by use of ethnicity coefficient. Using a model to impute kidney function based on mGFR, we estimated CKD prevalence to be two to three-fold higher compared to creatinine-based estimates in populations across six countries in sSA. Conclusion: Based on existing creatinine-based methods to estimate GFR, we found a relatively low prevalence of impaired renal function in the general population. We also demonstrated that eGFR <45mls/min/1.73m2 are associated with an increased all-cause mortality. However, using iohexol clearance, we showed that these creatinine-based measures over-estimate GFR and under-estimate CKD. This means a substantial proportion of people with kidney disease are missed by current eGFR equations which may have adverse effects on the health and care of patients with CKD in sSA.