Selection of participants was done according to well-defined criteria as outline in our earlier publication on the prevalence of diabetes in homeless and slum dwellers in Accra, Ghana .
A total of 512 homeless subjects took part in the study comprising 232 males and 280 females with mean age of 38.3 ± 12.0 and 40.0 ± 10.4 respectively. Information on participants’ socio-demographic characteristics were recorded. Height was measured of participants not wearing shoes using a stadiometer to the nearest 0.5 cm with the study participants standing upright and heels put together and the head in the horizontal plane. Weight was measured in kilograms using the Bioimpedance analyzer (BIA) (BSD01, Pure Pleasure, a division of the Stingray Group, Cape Town, South Africa). Body mass index (BMI) was calculated using the following formula: BMI = body weight (kg) / [height (m)]2 .
Five milliliters of venous blood and 10 ml of urine samples were taken from all participants into serum separator tubes and urine containers respectively. The blood samples were then
centrifuged to obtain sera and both urine and serum samples were stored
in several aliquots at − 80 °C until sample analysis. Serum and urine creatinine were determined using the Vitros dry chemistry analyzer (OrthoClinical Diagnostics, Johnson & Johnson, High Wycombe,UK). Urine albumin was determined using the dip-stick qualitative/semi-quantitative method (Urit Medical Electronic Co., Ltd, Guangxi, People’s Republic of China) following manufacturer’s instructions.
Renal function was estimated using the Modification of Diet in Renal Disease (MDRD) equation  and the Cockcroft–Gault (C–G) equation, normalized for the body surface area (BSA)  while the classification of the stages of chronic kidney disease was based on markers of renal pathology (eGFR and the presence of albuminuria) [1, 14]. Normal GFR was regarded as stage 1, mildly decreased was stage 2 while moderately decreased, severely decreased and kidney failure were regarded as stage 3, stage 4 and stage 5 respectively 
The data was first entered into Microsoft Office Excel 2007. We calculated prevalence estimates with the SAS 9.3 program (SAS Institute Inc., Cary,North Carolina). Proportions of those with renal insufficiency and stages of CKD were also calculated. Frequencies were reported as unweighted counts. The level of statistical significance was set at P < 0.05 for all tests and at 95% confidence interval.