High prevalence of chronic kidney disease in a community survey of urban Bangladeshis: a cross-sectional study

We obtained a detailed map of Mohammedpur, an urban residential area organized in the 1950s within Dhaka with a population of about 450,000 people living in 100,000 households [9]. Of the six blocks that comprise the district, we randomly selected three. We then assigned a study identification number to all of the buildings in the three blocks, and used random sample generator software (SPSS, Chicago, IL) to create a list of 500 buildings. We selected extra buildings to ensure enrollment of one person per building. Most buildings in Mohammedpur are apartment complexes. On arrival at a selected building, the survey team—made up of a research assistant, study nurse and study supervisor—counted the total number of households, and utilized a random number generator to select a single household. If the household members declined to participate or there was non-response, the survey team randomly selected another household within the same building.

Eligible participants were at least 30 years old; if there were more than one eligible participant in the selected household, we randomly selected one from all eligible participants using the same random number generator. We targeted equal enrollment of men and women in the 30 to 50 and > 50 year age category (i.e., 100 individuals in each category). We enrolled participants until the quota for any given combination was reached, after which participants were enrolled only into the remaining strata.

After obtaining informed consent, research assistants administered a patient questionnaire in Bangla. Using a structured questionnaire, we obtained data on: socioeconomic status including education and occupation, tobacco use, self-reported personal and family history of chronic diseases, diet and physical activity, and family planning. We administered the Medical Outcomes Study Short-Form-12 version two questionnaire (SF-12) to assess health related quality of life (HRQoL). At the completion of the patient questionnaire, the research assistant made an appointment for a second visit (within one week) to obtain a urine specimen, fasting serum sample, and anthropometric measurements.

A study nurse measured height, weight, and waist and hip circumference on all participants using standardized procedures [10]. The nurse also obtained two blood pressure readings using a manual sphygmomanometer after the participant had remained in a seated position for five minutes with his or her arm resting on the chair. The nurse used standard phlebotomy procedures to obtain a fasting serum sample and a urine specimen.

We transported serum and urine samples to the ICDDR,B central laboratory within one hour of collection. We processed all samples for serum creatinine, hemoglobin A1c, and fasting glucose. Due to budgetary limitations, we performed more expensive tests—urine albumin and creatinine, and serum lipids and triglycerides—in a select group of participants. This group (n = 341) fit at least one criteria for metabolic syndrome, had urine dipstick positive for proteinuria, or had risk factors for CKD (i.e., self-reported personal or family history of diabetes, hypertension, or CKD, or self-reported personal history of auto-immune disease, kidney stones, or birth weight below 2^.5 kg) [11].

Urine albumin was measured by immunoturbidimetric assay using Olympus AU 640, and plasma and urinary creatinine were measured by enzymatic colorimetric assay using Olympus AU 640 (Olympus corporation, Tokyo, Japan). Assay variation was checked daily with an internal quality assessment scheme, and the coefficient of variation for each test was below 5%.

We defined a participant as having T2DM if he or she had fasting blood glucose ≥ 7^.0 mmol/L or hemoglobin A1c ≥ 6^.5% or medication use for self-reported diabetes. We defined metabolic syndrome according to International Diabetes Federation criteria [12] as centripetal obesity (waist circumference ≥ 90 cm for men and ≥ 80 cm for women) plus any two of the followings: (1) triglycerides > 1^.7 mmol/L or treatment for elevated triglycerides, (2) HDL cholesterol < 1^.03 mmol/L in men or < 1^.29 mmol/L in women, or treatment for low HDL, (3) systolic blood pressure > 130 mmHg, diastolic blood pressure > 85 mmHg, or treatment for hypertension, and (4) fasting plasma glucose > 5^.6 mmol/L. We defined a participant as having insulin resistance if he or she met criteria for T2DM or metabolic syndrome. We used the following categories for Quételet’s (body mass) index (BMI): < 25 (lean or normal), 25 to < 30 (overweight), and ≥ 30 kg/m² (obese). We captured leisure time physical activity by the following question: 'How often do you walk outside the home for more than 10 minutes without stopping?’ We categorized the responses as 'Rarely,’ 'Weekly’, or 'Daily.’

We defined a participant as having cardiovascular disease if he or she reported using medications for blood pressure or heart disease, or had systolic blood pressure ≥ 140 mmHg and diastolic blood pressure ≥ 90 mmHg.

We used the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation to estimate glomerular filtration rate (eGFR), and the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (NKF/KDOQI) guidelines to define and stage CKD [11]. We used a sex- specific cut off for albuminuria: ≥ 1^.9 mg/mmol (17 mg/g) creatinine for men, and ≥ 2^.8 mg/mmol (25 mg/g) creatinine for women. Thus we defined a participant as having CKD if he or she had eGFR below 60 ml/min/1.73 m² or had albuminuria. We also present prevalence results for albuminuria and CKD using the ≥ 3^.4 mg/mmol (30 mg/g) creatinine cutoff.

We present continuous data as mean (± standard deviation) and categorical data as proportions, with comparisons via Student’s t-test or Pearson’s χ² test, as appropriate. We stratified by sex and performed logistic regression to test the associations among potential covariates including education, wealth status, physical activity, tobacco use, measures of body size (BMI, waist circumference, and waist-to-hip ratio), and presence of insulin resistance or cardiovascular disease. We used generalized linear models to evaluate the association between CKD and the Physical and Mental Health Composite t-scores on the SF-12. We used the New England Medical Center (NEMC) guidelines to score the SF-12 [13]. The primary analysis evaluated these associations using NKF/KDOQI stages of CKD, calculated based on eGFR and/or albuminuria. In multivariable analyses, we adjusted for the residual effects of age and sex, as well as for insulin resistance. Since age and sex are embedded in the eGFR calculation, we performed companion analyses in which we used the reciprocal of serum creatinine as a marker for kidney function. We considered inference tests as significant if the two-tailed p-value was <0^.05. We conducted all analyses using SAS, v9.3 (SAS Institute Inc., Cary, NC).

Ninety-four (26%) participants met the definition of CKD (Figure 1a and b), 45 men and 49 women. A majority (n = 58 (62%)) had stage one or two CKD (i.e., albuminuria only). Among the 36 participants with stage three or more advanced CKD, 21 (58%) had associated albuminuria. Overall, there were 79 (22%) participants with albuminuria, 65 (83%) of whom had microalbuminuria (Figure 2). Using an albuminuria cutoff of ≥ 3.4 mg/mmol creatinine, 56 and 72 (20%) participants were classified as having albuminuria and CKD respectively.

On unadjusted analyses, education level was not associated with the presence of CKD among men, although among women, those who had received fewer than 5 years of education experienced higher odds of CKD compared with women educated at the University level or beyond (Odds ratio [OR] 3^.6, 95% Confidence Interval [CI] 1^.4 to 9^.4). Wealth status was not associated with CKD among women or men. Since smokers in our sample were predominantly (96%) men, we tested the association between smoking and CKD among men only and found no association. Use of smokeless (chewable) tobacco was associated with higher odds of CKD among women (OR 2^.8, 95% CI 1^.4 to 5^.7).

Larger waist circumference and waist-to-hip ratio (using International Diabetes Foundation criteria) [12], BMI, and leisure-time physical activity were not significantly associated with CKD.

A majority of participants with CKD also had features of insulin resistance (Figure 3). The unadjusted odds of CKD among participants with insulin resistance were 3^.6 fold higher (95% CI 2^.1 to 6^.4) than among participants without insulin resistance. Upon stratification by sex, the odds of CKD among men with insulin resistance were 3^.1 (95% CI 1^.5 to 6^.4) and among women were 4^.7 (95% CI 1^.9 to 11^.9). Cardiovascular disease was also evident in more than two-thirds (n = 64, 68%) of participants with CKD. The measured average blood pressure of a participant with CKD was 125/78 mmHg, approximately 10/5 points higher than in participants without CKD (p <0^.001 and 0^.005, for systolic and diastolic pressure, respectively).

Mean t-scores on the SF-12 NEMC Physical Health Composite t-Score (PCS) were lower among participants with CKD (Table 2). There was no significant difference in mean t-scores on the SF-12 Mental Health Composite Score (MCS), although the mean MCS for participants with stage five CKD was significantly lower than for all other stages. Multivariable analyses adjusting for the residual effects of age and gender as well as for insulin resistance showed that mean scores on PCS were significantly lower among participants with stage three or more advanced CKD.

On companion analyses using the reciprocal of serum creatinine as a main exposure of interest (as opposed to eGFR), we found that the PCS was inversely correlated with serum creatinine and age, and was lower among women compared with men.