We conducted a 5-month cross-sectional study at the outpatient department in a nephrology referral center in Yaoundé, Cameroon. We consecutively enrolled 27 non-dialysed patients aged more than 21 years, with CKD stages 3 to 5 (K/DOQI classification). Patients who had undergone an amputation or had a prosthesis, those who had the following comorbidities: Cancer, severe infection, heart or liver failure, and those with incomplete 24-hour urine collection, were excluded.

Before inclusion, we obtained a written consent from each participant after an explanation of the objectives, procedure, advantages, and disadvantages of the study. The ethical clearance was obtained from the ethical committee of Yaoundé General Hospital.

After informed consent, a standardized questionnaire was administered to collect data on age, gender, and comorbidities. Blood pressure (in mmHg) was measured in a sitting position on the left arm after at least 10 min of rest, using an electronic device (OMRON HEM-7222-Z intellisense). Two measures were taken 5 min apart, and the mean value was considered. We used a stadiometer (SECA®) to measure the height to the nearest 0.5 cm in the standing position. Weight was obtained at the nearest 0.1 kg with an electronic device (SECA®), the patient standing without shoes. BMI was calculated using the formula $\frac{\text{Weight}}{{\left(\text{Height}\right)}^{\text{2}}}$ in kg/m².

The procedure of 24-hour urine sample collection was explained to participants, and they received the urine collection form and written instructions. The sample was obtained after an initial voiding in the morning upon awaking; thereafter, all urines were collected until the next morning upon awaking. This sample was immediately transported to the laboratory at ambient temperature for volume (to the nearest 10 ml), urinary urea, and creatinine measurements. A fasting venous blood sample was collected into a 5 ml dry tube and was immediately transported to the laboratory, where it was centrifuged and serum analysed for serum urea and creatinine. Serum and urine creatinine and urea were measured using an automated instrument according to standards using the enzymatic method (Roche/Hitachi/Cobas C 311).

Estimated GFR in ml/min/1.73m² was calculated using the following equations:

$\text{eGFR}=\frac{\left(140-\text{age}\right)\times \text{weight}\times 0.85\left[\text{if}\text{female}\right]}{72\times \text{SCr}}$

where SCr represents serum creatinine in mg/dl, age is measured in years, and weight in kg.

$\text{GFR}=175\times {\left(\text{SCr}\right)}^{-1.154}\times {\left(\text{age}\right)}^{-0.203}\times 1.212\times 0.742\left[\text{if female}\right]$

where SCr represents serum creatinine in mg/dl, and age is measured in years.

$\begin{array}{c}\text{eGFR}=141\times \text{min}{\left(\text{SCr}/\text{k},1\right)}^{\alpha }\times \text{max}{\left(\text{SCr}/\text{k},1\right)}^{-1.209}\times {0.993}^{\text{Age}}\\ \times 1.018\left[\text{if female}\right]\times 1.159\end{array}$

where SCr is serum creatinine in mg/dl, k is 0.7 for females and 0.9 for males, α is −0.329 for females and −0.411 for males, min indicates the minimum of SCr/k or 1, and max indicates the maximum of SCr/k or 1. Race or ethnicity coefficient for “Black” individuals in the MDRD and CKD-EPI equations was not applied.

Urea clearance (UCl) was calculated using the following formula:

$\text{UCl}=\left(\text{urinary Urea}\times \text{urine flow}\right)/\text{plasma Urea}$

Creatinine clearance (CrCl) was calculated using the standard formula:

$\text{CrCl}=\left(\text{urinary creatinine}\times \text{urine flow}\right)/\text{plasma creatinine}$

The result was adjusted for body surface area (BSA) using Dubois’ formula [15] .

$\text{BSA}=0.007184\times \text{Weight}{\left(\text{kg}\right)}^{0.425}\times \text{Height}{\left(\text{cm}\right)}^{0.725}$

Measured glomerular filtration rate (mGFR) was obtained using this formula:

$\left(\text{Creatinine Clearance}+\text{Urea Clearance}\right)/2$

As creatinine clearance (CrCl) significantly overestimates the measured glomerular filtration rate and urea clearance (UCl) significantly underestimates mGFR, some authors have proposed calculating the average of CrCl and UCl [16] . The 2005 European Best Practices Guidelines for stages 4 - 5 CKD also propose this method when GFR is less than 15 ml/min/1.73m². In this study, we used the average urinary clearances of creatinine and urea as the gold standard measurement of GFR.

Characteristics of the population were described using mean (standard deviation) and median (interquartile range) for quantitative variables, and number (percentage) for qualitative variables. Agreement between eGFR and mGFR was assessed using Bland-Altman’s analysis. Bias was calculated as the mean difference between mGFR and eGFR. Precision was considered as one standard deviation (SD) of bias. Agreement limits were calculated (bias ± 1.96 SD) and presented on the graph with bias and corresponding 95% confidence intervals. We used c-statistics to assess the mGFR-based ESKD (mGFR < 15 ml/min/1.73m²) diagnostic accuracy of eGFR from each of the equations. The Receiver Operating Curve (ROC) was used to determine the best eGFR cut-off from each of the equations for the diagnosis of mGFR-based ESKD. The agreement of each of the eGFR equations (using the best cut-off) with mGFR for assessing ESKD (GFR < 15 ml/min/1.73m²) was tested using Cohen’s kappa statistics. Data analysis was done using IBM SPSS Statistics 23.0 software, and graphs were represented using Microsoft Excel 2013 and GraphPad Prism 6.0 for Windows.

Administrative authorization to carry out the study was obtained from the authorities of the Yaoundé General Hospital.

A total of 27 (16 Males) out of the 91 eligible patients were enrolled in the study ( Figure 1 ). The mean (standard deviation) age was 57 (12.17) years. Hypertension and diabetes were the most prevalent comorbidities, concerning 92.6% (n = 25) and 37% (n = 10) of patients, respectively. The median (interquartile range) evolution time of CKD was 5 (1 - 13) years. The main aetiologies of CKD were diabetes, hypertension, chronic glomerulonephritis, and chronic interstitial nephritis, affecting respectively 25.9% (n = 7), 25.9% (n = 7), 14.8% (n = 4), and 14.8% (n = 4) of patients. Other clinical characteristics are shown in Table 1 .

Overall, the three equations had an almost excellent accuracy in the diagnosis of end-stage kidney disease; the c-statistics were 0.97 for CKD-EPI and CG, and 0.98 for MDRD ( Table 2 ).

Bland-Altman plots for CKD-EPI, MDRD, and CG vs mGFR are shown in Figure 2 . Bias (95% confidence interval) was significantly different from zero for all three equations; it was lower for CKD-EPI [3.24 (1.5 - 4.9) ml/min/1.73m²] than for MDRD and CG [respectively 5.17 (2.9 - 7.4) and 5.76 (3.4 - 8.2) ml/min/1.73m²]. The limits of agreement were wide for all three equations, although these limits were narrower for the CKD-EPI equation. The ROC curves of equations to assess the best cut-off in the diagnosis of eGFR-based ESKD are shown in Figure 3 . CKD-EPI equation had the lowest cut-off value (16.25 ml/min/1.73m²), with sensitivity (Se) = 0.93 and specificity (Sp) = 0.92 ( Figure 3(a) ). For MDRD and CG equations, cut-offs (sensitivity and specificity) were respectively 18 ml/min/1.73m² (Se = 0.93; Sp = 0.92) ( Figure 3(b) ) and 16.8 ml/min/1.73m² (Se = 0.93; Sp = 0.85) ( Figure 3(c) ). Using these cut-offs, the agreement with mGFR was almost perfect for CKD-EPI (kappa = 0.852) and MDRD (kappa = 0.852), and moderate for CG (kappa = 0.78). Other details on the performances of equations are shown in Table 2 .

IQR: Interquartile Range; SD: Standard Deviation; CKD: Chronic Kidney Disease; BMI: Body Mass Index; CrCl: Creatinine Clearance; UCl: Urea Clearance; GFR: Glomerular Filtration Rate; BP: Blood Pressure; HBC: Hepatitis B Virus; HCV: Hepatitis B Virus; HIV: Human Immunodeficiency Virus.

IQR: Interquartile Range; SD: Standard Deviation; mGFR: measured Glomerular Filtration Rate; CI: Confidence Interval; MDRD: Modification of Diet in Renal Disease; CKD-EPI: Chronic Kidney Disease Epidemiology Collaboration; CG: Cockcroft-Gault.