Study design and setting
A cross-sectional study was conducted following the protocol of the Disadvantaged Populations eGFR Epidemiology Study (DEGREE) [20]. This study was conducted in Tumbes, a region located in the North of Peru, on the border with Ecuador, and where approximately 90% of the population is urban. In addition, nearly 20% and 1% of the population respectively are considered poor and extremely poor, and 15% do not have basic sanitation services [21].
Tumbes was chosen for the following reasons (see Table 1 for comparison with other regions where CKDu has been documented): agriculture workers (up to 10% of the active economic population), contamination of the Rio Puyango, the main source of water (i.e. cadmium, arsenic), and warm climate with low precipitation rates. The temperature of the city has an annual average temperature of 27ºC, reaching almost 35 ºC during the summer [22].
Study participants
Adults aged ≥18 years, resident in Tumbes, were invited to participate using the most updated census in the area (2014). Pregnant women, those who declined to give blood and urine samples, and those with mobility disabilities preventing bioimpedance assessment, were excluded from the study. The sample was stratified by population group, so that, participants were from urban and rural areas in similar proportions (50%). In addition, we decided not to recruit more women after their number reached 60% for each of these two subgroups.
Outcome
We calculated the estimated Glomerular Filtration Rate (eGFR) using the Chronic Kidney Disease – Epidemiology Collaboration (CKD-EPI) equation. We defined three categories of kidney function based on eGFR results: moderate or established kidney dysfunction (<60 mL/min/1.7m2), mild kidney dysfunction (60-90 mL/min/1.7m2), and normal (≥ 90 mL/min/1.7m2). CKD was defined as the proportion of individuals with moderate or established kidney function (eGFR <60 mL/min/1.7m2) in the overall sample; whereas the prevalence of CKDu was estimated using the same indicator but excluding individuals with previous diagnosis of HT and/or T2DM, and those having a heavy proteinuria, defined as 3+ in urine dipstick (equivalent to ≥ 300 mg/dl of protein) according to the DEGREE protocol [20].
Questionnaires
During the first visit, questionnaires were used to gather information on socio-demographic factors (age, sex) and socioeconomic status (education, employment status, house-hold income and health insurance). Environmental conditions associated with CKDu (previous or current work in agriculture or sugarcane, water source, heat, and pesticide exposure) were addressed. Information on past medical history was focused on cardiovascular diseases (HT, T2DM, myocardial infarction, stroke, and hypercholesterolemia), and behavioural risk factors (alcohol, smoking, and physical activity). We also included questions on CKD and its associated causes, including congenital kidney malformation, diabetic nephropathy, polycystic kidney disease, urolithiasis, use of nephrotoxic medications (non-steroidal anti-inflammatory, parenteral use of antibiotics, and herbal medicine), tuberculosis, hepatitis B, and leptospirosis.
Clinical assessment
Participants were invited for another visit to obtain blood samples for fasting glucose and creatinine, urine samples (dipstick urinalysis) and clinical examination. We asked the participants to refrain from eating meat, caffeine, tobacco and paracetamol 8 hours prior to taking the sample. The clinical evaluation included three blood pressure measurements taken after 5 minutes of resting (seated) position and 5 minutes apart from each other (evaluated with an automatic digital calibrated sphygmomanometer OMRON HEM-780, Tokyo, Japan), stand height (cm), weight (kg) by standardized procedures, and body composition (i.e. body fat percentage), for which we used a supine bioimpedance analyzer balance (BodyStat 1500, BodyStat Limited, British Islands) calibrated at a single frequency (50Hz).
Analysis of biological samples
Urine dipstick analyses yielded semi-quantitative values of blood, proteins, leukocytes, glucose, pH and urine density. We measured serum creatinine in 5ml of blood sample using a mass spectrometry method calibrated according to isotope dilution (IDMS, English acronym). As in the DEGREE Study protocol, serum samples were stored at -20 °C for future Cystatin C determination.
Other variables of interest
We also gathered information on key reported risk factors for CKD and CKDu. Age was considered as both a continuous and categorical variable. Level of education was classified in three categories (<7, 7–11, ≥12 years of education). We also considered socioeconomic status as monthly household income dichotomised with a cut-off point of 850 PEN (250 USD, minimum salary wage at the moment of the study).
We also collected information on ever working in agriculture, sugarcane, and pesticide use; these variables were built as categorical (yes/no). Due to potential sources of contamination of water, we chose access to piped water as a potential confounder. Heat intolerance, a proxy of heat exposure, was considered to have occurred if the participant reported to have fainted when exposed to hot weather. We built a variable analgesic/antibiotic use due to the variety of potential nephrotoxic medications (including herbal medicine use) usually sold without prescriptions. For this, we focused on the ever use of analgesics and intramuscular antibiotics. In addition, history of some diseases because their impact on CKD was also included: tuberculosis, hepatitis B, and leptospirosis.
Heavy drinking (consuming at least 6 alcoholic standard beverages, monthly), and current smoking within the last 12 months. Physical activity was dichotomised in low vs. moderate/high according to the International Physical Activity Questionnaire (IPAQ). Body mass index (BMI) was calculated according to the WHO recommendations, and then split into three categories (<25 kg/m2, between 25 and <30 kg/m2, and ≥30kg/m2). HT was defined as a systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg and/or previous diagnosis by a physician and current medication. To estimate the values for systolic and diastolic blood pressure, we used the mean of the last two (out of three) measurements conducted. T2DM was defined as a fasting (8-12 hours) glucose ≥126 mg/dl or self-report of previous diagnosis and treatment. The cut-off point for glycosuria was ≥250 mg/dl, urinary density ≥1020 (as a proxy for dehydration) and proteinuria if the urine level marker of protein was ≥30 mg/dl (equivalent to 1+ in urine dipstick).
Sample size
We planned a sample size of 750 participants in each population group (rural and urban). Assuming an expected CKDu prevalence of 5%, with 1500 individuals in total, we will have a precision of 1%. In addition, this precision would be of 1.6% if only 750 subjects are evaluated as in each study group.
Data analysis
We conducted descriptive analyses of socio-demographic, socioeconomic, cardiovascular risk profile and CKD risk profiles according to the study population group (rural and urban) and kidney function. These comparisons were conducted for the overall sample (compatible with CKD) and among those without known HT, T2DM or heavy proteinuria (as a proxy of CKDu). Chi-square tests for comparison of categorical variables, and Student's t test or Mann-Whitney U was used for comparison of numerical variables. In addition, the prevalence of CKD and CKDu were also estimated with their respective 95% confidence intervals (95%CI).
As very few cases with eGFR levels compatible with CKD and CKDu (i.e. eGFR <60 mL/min/1.7m2) were found, we analysed factors associated with impaired kidney function (eGFR <90 mL/min/1.7m2) rather than factors associated with eGFR<60. We used logistic regression models to determine associated factors for impaired kidney function. As a result, crude model and adjusted models are shown. The first adjusted regression model was controlled by age (as continuous variable) and sex; whereas the second regression model included also population group and education level as potential confounders.