Cross-sectional study of association between socioeconomic indicators and chronic kidney disease in rural–urban Ghana: the RODAM study

Objectives Studies from high-income countries suggest higher prevalence of chronic kidney disease (CKD) among individuals in low socioeconomic groups. However, some studies from low/middle-income countries show the reverse pattern among those in high socioeconomic groups. It is unknown which pattern applies to individuals living in rural and urban Ghana. We assessed the association between socioeconomic status (SES) indicators and CKD in rural and urban Ghana and to what extent the higher SES of people in urban areas of Ghana could account for differences in CKD between rural and urban populations. Setting The study was conducted in Ghana (Ashanti region). We used baseline data from a multicentre Research on Obesity and Diabetes among African Migrants (RODAM) study. Participants The sample consisted of 2492 adults (Rural Ghana, 1043, Urban Ghana, 1449) aged 25–70 years living in Ghana. Exposure Educational level, occupational level and wealth index. Outcome Three CKD outcomes were considered using the 2012 Kidney Disease: Improving Global Outcomes severity of CKD classification: albuminuria, reduced glomerular filtration rate and high to very high CKD risk based on the combination of these two. Results All three SES indicators were not associated with CKD in both rural and urban Ghana after age and sex adjustment except for rural Ghana where high wealth index was significantly associated with higher odds of reduced estimated glomerular filtration rate (eGFR) (adjusted OR, 2.38; 95% CI 1.03 to 5.47). The higher rate of CKD observed in urban Ghana was not explained by the higher SES of that population. Conclusion SES indicators were not associated with prevalence of CKD except for wealth index and reduced eGFR in rural Ghana. Consequently, the higher SES of urban Ghana did not account for the increased rate of CKD among urban dwellers suggesting the need to identify other factors that may be driving this.

• Our study is also the first in Africa to use all three categories of CKD definition (albuminuria, reduced eGFR and CKD risk) by KDIGO 2012 in assessing association of SES with CKD in rural and urban setting, this provides a more detailed information on CKD outcomes.
• The limitation of intra laboratory variability in earlier studies was eliminated using the same standard operating procedures in the same laboratory for running all samples for both rural and urban Ghana.
• The use of three constructs of SES (educational level, occupational level and wealth index) in this study also provides a much better holistic approach to assessing SES association with CKD. Also, the distribution of SES in our study reflect on the national data allowing for generalization of our findings.
• Our study was limited by the use of cross sectional design which prevented us from determining causality between predictors and CKD progression. and associated with inadequate dialysis treatment, reduced access to kidney transplantation and poor health outcomes 1 . Recent studies have consistently found low SES to be associated with higher risk of CKD among people of African origin 2-5 . However, in some settings the well-known inverse association between SES and CKD seems to be absent, or even reversed. For example, Bryne et al. did not find an association between SES and End Stage Renal Disease 6 . Invariably, others studies have consistently found a positive association between SES and CKD 7 8 . Specifically, as SES improved, unhealthful lifestyle (unhealthy diet, physical inactivity, smoking and alcohol consumption) increased in China while that of the United States decreased with improved SES 9 . People with higher incomes, in these contexts, can afford a western lifestyle, which is more readily available in the urban areas than in the rural areas. There is therefore an interaction between individual SES and environmental factors, such as food and sedentary life style in such populations [10][11][12] .
Consequently, in those settings, people with a higher SES might have higher CKD risk.
In urban areas, the population in general has a higher SES than in rural areas 13 . For example, individuals with higher educational level migrate from rural areas to find higher occupations matching their higher education to improve on their wealth. If indeed a positive association between SES and CKD is observed in LMICs, this might underlie the well-known health differences between urban and rural areas, with urban areas having an increased risk of CKD 14 . So far, it is unknown whether the reversed SES gradient (higher risk in high SES group) might explain the higher burden of CKD in urban areas as compared to rural areas in Africa.
In view of this, we assessed the association of SES with CKD in rural and urban Ghana and studied what extent the higher SES of people in urban areas could account for differences in CKD between rural and urban populations.

Study population and study design
In the present analyses, data used were from the RODAM (Research on Obesity & Diabetes among African Migrants) study, a multi-centre cross-sectional study, were used. The rationale, conceptual framework, design and methodology of the RODAM study have been described in detail elsewhere 15 16 . As the Healthy Life in an Urban Setting (HELIUS) study conducted among Ghanaian migrants living in Amsterdam did not find any associations between SES and CKD 17  Data collection for the study was standardized across the sites. Written informed consent was obtained from each participant prior to enrolment in the study. The respective ethics committees in Ghana and the three European countries approved the study protocols before data collection began. The response rate was 76% in rural Ghana and 74% in urban Ghana. In Ghana, participants were randomly drawn from a list of 30 enumeration areas in the Ashanti region based on the 2010 population census using the multistage random sampling. These enumeration areas came from two purposively selected urban cities (Kumasi and Obuasi) and 15 randomly selected rural communities in the Ashanti region. Selected health and community authorities were first identified, notified of the study and letters were sent giving detailed explanation of the study. We sent team members to stay among the communities to familiarize with them and organize mini clinics in the field. This lasted between 1-2 weeks depending on the sampled population and responsiveness of respondents.
In Ghana, questionnaires administration and physical examination were done at the same day/time. The participants were instructed to fast from 10.00 pm the night before the physical examination. For the current study, 2566 participants with data available on both questionnaire data and physical measurements were used. We excluded (n=74) individuals outside the RODAM age range of 25-70 years resulting in a data set of 2492 for analysis. These  (BMI) was calculated as weight (kg) divided by height squared (m 2 ). Overweight was defined as BMI of ≥25 to <30 kg/m 2 and obesity as BMI ≥30 kg/m 2 18 . Waist circumference was measured in cm at the midpoint between the lower rib and the upper margin of the iliac crest. Per participant all anthropometrics were measured twice by the same assessor and the average of the two measurements were used for analyses.

Covariates
Socioeconomic indicators used in this study were educational level, occupational status and level of wealth index. Educational level was determined based on self-reported highest educational qualification accomplished based on the Ghanaian educational system. Occupational level was determined based on self-reported current occupation if still employed or/and last occupation before retirement or student. The reported occupations were further coded according to the International Standard Classification of Occupations scheme (ISCO-08). Wealth index was determined using the World Health Organization (WHO) standard of wealth index classification. Wealth index was based on data collected in the Household Questionnaire. The questionnaire comprised of questions on household's ownership of several consumer items such as television, car, flooring material, toilet facilities etc. Each household was assigned a standard score for each asset. Wealth index was then expressed in five quintiles. The five quintiles were further categorized into three quintiles by combining the second and third quintiles due to small numbers 19 . All three SES constructs were further classified as low, medium and high SES and their relationship to each other tested.

Outcome: CKD prevalence
Participants were asked to bring an early morning urine sample for the analyses of albuminuria and creatinine levels. Urinary albumin concentration (in mg/L) was measured by an immunochemical turbidimetric method (Roche Diagnostics). Urinary creatinine concentration (in umol/L) was measured by a kinetic spectrophotometric method (Roche Diagnostics). Estimated glomerular filtration rate (eGFR) F o r p e e r r e v i e w o n l y was calculated using the CKDEPI (CKD Epidemiology Collaboration) creatinine equation 20 . Urinary albumin-creatinine ratio (ACR; expressed in mg/g) was calculated by taking the ratio between urinary albumin and urinary creatinine. eGFR and albuminuria were categorized according to the 2012 KDIGO (Kidney Disease: Improving Global Outcomes) classification 21 . eGFR was categorized as follows: G1, ≥ 90 mL/min/1.73 m 2 (normal kidney function); G2, 60 to 89 mL/min/1.73 m 2 (mildly decreased); G3a, 45 to 59 mL/min/1.73 m 2 (mildly to moderately decreased); G3b, 30 to 44 mL/min/1.73 m 2 (moderately to severely decreased); G4, 15 to 29 mL/min/1.73 m 2 (severely decreased); and G5, < 15 mL/min/1.73 m 2 (kidney failure). Albuminuria categories were derived from ACR and were as follows: A1, < 3mg/mmol (normal to mildly increased); A2, 3 to 30 mg/mmol (moderately increased); and A3, > 30mg/mmol (severely increased). CKD status was categorized according to severity of kidney disease (green, low risk; yellow, moderately increased risk; orange, high risk; and red, very high risk) using the combination of eGFR (G1-G5) and albuminuria (A1-A3) levels defined by the 2012 KDIGO guideline 22 . Due to the small number of participants in the very high risk category of CKD, high and very high risk groups were combined. Reduced eGFR was defined as eGFR < 60 mL/min/1.73 m 2 . Because of the small number of participants in the severely increased albuminuria category, we defined albuminuria as ACR ≥3 mg/mmol by combining the moderately increased (A2) and severely increased (A3) categories.

Other variables
Blood pressure (BP) was measured three times using a validated semi-automated device (The Microlife WatchBP home) with appropriate cuffs in a sitting position after at least 5 min rest. The mean of the last two BP measurements was used in the analyses. Hypertension was defined as systolic BP ≥ 140 mmHg, and/or diastolic BP ≥ 90 mmHg, and/or being on antihypertensive medication treatment, and/or selfreported hypertension. Trained research assistants in the two sites collected fasting venous blood samples.
All the blood samples were processed and aliquoted immediately (within one hour to maximum three hours of the vena puncture) after collection per standard operation procedures, and then temporarily stored at the local research location at −20°C. The separated samples were then transported to the local research centres laboratories, where they were checked, registered and stored at −80°C. To avoid intralaboratory variability, the stored blood samples from the local research centres were transported to Berlin, Germany for biochemical analyses. Fasting plasma glucose concentration was measured using an enzymatic method (hexokinase). Type 2 diabetes was defined according to the WHO diagnostic criteria (fasting glucose ≥7.0 mmol/L, and/or current use of medication prescribed to treat diabetes, and/or selfreported diabetes) 23 . Concentration of total cholesterol was assessed using colorimetric test kits. All biochemical analyses were performed using an ABX Pentra 400 chemistry analyzer (ABX Pentra; Horiba ABX, Germany). Hypercholesterolemia was defined as total cholesterol level ≥ 6.22 mmol/L. Serum creatinine concentration (in umol/L) was determined by a kinetic colorimetric spectrophotometric isotope dilution mass spectrometry-calibrated method (Roche Diagnostics). Biochemical analyses were subject to extensive quality checks including blinded serial measurements.

Patient and Public Involvement
Community leaders were involved in the recruitment of patients. These comprised of religious communities (churches and mosques), endorsement from local key leaders and establishing relationships with healthcare organizations. We also provided information on the study by involving the local media (radio and television stations). We sent letters to all selected health and community authorities to notify participants of the study. Team members were sent to the various community to stay among the community and organize mini clinics for a period of 1-2 weeks. Results of the study were disseminated through seminars, durbars and via radio and television stations.

Statistical methods
Participants' characteristics were expressed as absolute numbers and percentages for categorical variables and as means and standard deviations (SD) for continuous variables. CKD prevalence with 5% error bars were presented as bar graphs for each SES construct across rural and urban Ghana. Spearman's rank correlation was used to determine correlations between the three SES constructs. Odds ratios (ORs) and their corresponding 95% confidence intervals (CIs) were estimated by means of logistic regression analyses to study the odds of albuminuria (ACR>3 mg/mmol, A2-A3, moderately to severely increased albuminuria), reduced kidney function (eGFR< 60 mL/min/1.73 m 2 , G3-G5 moderately to severely decreased kidney function) and increased CKD risk (high and very high CKD risk), with adjustments for potential covariates (age and sex). These covariates were adjusted for to account for their impact in the pathway of CKD incidence, prevalence and progression 24 . Model 1 was age and sex adjusted. The analyses were performed for the total population (using low educational level, low occupational status and low level of wealth index as reference categories). Further analysis was conducted using rural Ghana as reference. Model 1 was adjusted for age and sex while model 2 was adjusted for age, sex and educational level. Model 3 was adjusted for age, sex and occupational status while model 4 was adjusted for age, sex and level of wealth index (p<0.05). Tolerance test and variance inflation factor (VIF) showed very small degree of collinearity among SES predictors thus we adjusted for each of SES variables separately. Complete case analysis approach was used. All data available were included in the ageadjusted models. All analyses were performed using STATA, version 14.0 (StataCorp LP).  Table 1 shows characteristics of study participants. Participants in rural Ghana were slightly older than those in urban Ghana. Female preponderance was observed in both rural and urban Ghana, though higher proportions were observed in urban Ghana. Individuals living in rural Ghana were generally less educated compared with those living in urban Ghana. There were slightly more individuals with low occupational status in urban Ghana compared with their peers in rural Ghana. People in urban Ghana were wealthier than their rural counterparts. Rural Ghanaians were more physically active compared with their urban peers. Smoking was low among Ghanaians though rural Ghanaians were more likely to smoke compared with their urban peers. Hypercholesterolemia was more prevalent in urban Ghana than in rural Ghana.

Results
Hypertension and type 2 diabetes were more prevalent in urban Ghanaians compared with those living in rural Ghana. Urban Ghanaians were markedly more obese compared with their rural peers. Except for eGFR, albuminuria and CKD risk prevalence rates were higher in urban Ghana compared with rural Ghana.  In urban Ghana, high educational level was positively associated with high wealth index but inversely associated with occupation. In rural Ghana, high education was positively associated with high wealth index, but there was no significant association between education and occupation. High wealth index was inversely associated with high occupational status in both rural and urban Ghana (Table 2).  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47 o n l y Table 3 shows association between level of education, occupational status, level of wealth index and prevalence of CKD. After adjusting for age and sex, we observed no significant association between SES indicators (educational level, occupational status and wealth index) and CKD in urban Ghana. In rural Ghana, whereas educational level and occupational status were not associated with CKD prevalence, high wealth index was significantly associated with higher odds of reduced eGFR. o n l y Table 4 shows the contribution of all three SES constructs to rural and urban CKD prevalence differences. The odds of albuminuria and CKD risk was significantly higher in urban Ghana compared with rural Ghana. The higher rate of CKD observed in urban Ghana was not explained by the higher SES of that population as compared to their rural counterparts.

Key findings
Our study findings show no association between all three SES constructs and the prevalence of CKD in both rural and urban Ghana except for wealth index in rural Ghana, with the risk of CKD being higher in the wealthier populations. The higher rate of CKD observed in urban Ghana could not be attributed to the higher SES of that population compared to their rural counterparts.

Association of SES with CKD in rural and urban Ghana
Our study did not find any significant associations between all three SES constructs and CKD among rural and urban Ghana except for wealth index in rural Ghana. The positive association observed between wealth index in rural Ghana may be due to a number of reasons. A comparison of the three SES constructs showed higher educational level to be associated with wealth index in both rural and urban  6 26 which reported no association between SES and CKD in high-income countries and LMICs, but in contrast with other studies 2-4 27 that found positive associations between SES and CKD. The reasons for our current finding are unclear. However, it has been suggested that these inconsistent associations may be due to the varying pathways through which the effect of SES on health status is mediated. For example, at a given educational level marked ethnic differences have been reported. Additionally, similar differences were observed for wealth status at a given income level [28][29][30] .

Contribution of SES to observed CKD risk differences between rural and urban Ghana
We observed higher rates of CKD in urban Ghana compared with rural Ghana, as expected.

Strength and limitation
Our study presents several strengths. First, we used well-standardized study protocols across rural and urban Ghana. Our study is also the first in Africa to use all three categories of CKD definition (albuminuria, reduced eGFR and CKD risk) by KDIGO 2012 in assessing association of SES with CKD in rural and urban setting, this provides a more detailed information on CKD outcomes. The limitation of intra laboratory variability in earlier studies was eliminated using the same standard operating procedures in the same laboratory for running all samples for both rural and urban Ghana. The use of three constructs of SES in this study also provides a much better holistic approach to assessing SES. Also, the distribution of SES in our study reflect on the national data allowing for generalization of our findings. Our study was limited by the use of cross sectional design which prevented us from determining causality between predictors and CKD progression.

Conclusion
All three SES constructs appear not to be associated with prevalence of CKD in urban and rural Ghana except for wealth index in rural Ghana. The observed higher prevalence of CKD in urban Ghana was not explained by the higher SES in urban Ghana. Our study seems to suggest that other non-traditional factors such as nephrotoxins, herbal medications and misuse of over the counter drugs may play a role and underscores the need to further explore these factors.

Acknowledgement
The authors are very grateful to the research assistants, interviewers and other staff of the five research locations who took part in gathering the data and the Ghanaian volunteers in all the participating RODAM sites. We gratefully acknowledge the advisory board members for their valuable support in shaping the RODAM study methods and the Academic Medical Centre Biobank for their support in biobank management and high-quality storage of collected samples.

Contributors
My co-authors have all contributed substantially to this manuscript and approve of this             Disease: Improving Global Outcomes) severity of CKD classification: albuminuria (albumin-creatinine 52 ratio ≥ 3 mg/mmol (category ≥ A2)); reduced glomerular filtration rate (eGFR < 60 mL/min/1.73 m2 53 (category ≥ G3)) and high to very high CKD risk based on the combination of these two.

55
Results: All three SES indicators were not associated with CKD in both rural and urban Ghana

135
However, in some settings the well-known inverse association between SES and CKD seems to be absent,

144
In urban areas, the population in general has a higher SES than in rural areas 13

277
Participants' characteristics were expressed as absolute numbers and percentages for categorical variables 278 and as means and standard deviations (SD) for continuous variables. CKD prevalence with 5% error bars 279 were presented as bar graphs for each SES construct across rural and urban Ghana. Spearman's rank 280 correlation was used to determine correlations between the three SES constructs. Odds ratios (ORs) and 281 their corresponding 95% confidence intervals (CIs) were estimated by means of logistic regression 282 analyses to study the odds of albuminuria (ACR>3 mg/mmol, A2-A3, moderately to severely increased 283 albuminuria), reduced kidney function (eGFR< 60 mL/min/1.73 m 2 , G3-G5 moderately to severely 284 decreased kidney function) and increased CKD risk (high and very high CKD risk) by SES, with 285 adjustments for potential covariates (age and sex). These covariates were adjusted for to account for their 286 impact in the pathway of CKD incidence, prevalence and progression 24 . The analyses were performed 287 for the total population (using low educational level, low occupational status and low level of wealth 288 index as reference categories). Further analysis was conducted using rural Ghana as reference. Model 1 289

292
Tolerance test and variance inflation factor (VIF) showed very small degree of collinearity among SES 293 predictors thus we adjusted for each of SES variables separately. Complete case analysis approach was 294 used. All data available were included in the age-adjusted models. All analyses were performed using 295 STATA, version 14.0 (StataCorp LP).

296
Results 297 298 Table 1 shows characteristics of study participants. Participants in rural Ghana were slightly older than    1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 F o r p e e r r e v i e w o n l y 11 Among the whole group, educational level was positively associated with wealth index (p<0.01) and composite SES (P<0.01). Occupational level 329 was also inversely associated with educational level (p<0.01) and wealth index (p<0.01). In urban Ghana, high educational level was positively 330 associated with high wealth index but inversely associated with occupation (p<0.01). In rural Ghana, high education was positively associated with 331 high wealth index (p<0.01), but there was no significant association between education and occupation. High wealth index was inversely 332 associated with high occupational status in both rural and urban Ghana (p<0.01) ( Table 2). 333 o n l y 13 Table 3 shows association between level of education, occupational status, level of wealth index and prevalence of CKD. After adjusting for age 350 and sex for the whole group, albuminuria was associated with middle level education (p<0.01). After adjusting for age and sex, we observed no 351 significant association between SES indicators (educational level, occupational status and wealth index) and CKD in urban Ghana. However, 352 middle and higher level education was associated with reduced albuminuria in urban Ghana (p<0.01). Whereas educational level and occupational 353 status were not associated with CKD prevalence, high wealth index was significantly associated with higher odds of reduced eGFR (p<0.01).  Table 4 shows the contribution of all three SES constructs to rural and urban CKD prevalence differences. The odds of albuminuria and CKD 364 risk was significantly higher in urban Ghana compared with rural Ghana. The higher rate of CKD observed in urban Ghana was not 365 explained by the higher SES of that population as compared to their rural counterparts. 366 367   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  of SES in this study also provides a much better holistic approach to assessing SES. Also, the distribution 436 of SES in our study reflect on the national data allowing for generalization of our findings. Our study was 437 limited by the use of cross sectional design, which prevented us from determining causality between 438 predictors and CKD progression. Furthermore, there were more women than men in our study due to the 439 higher response rate in women compared with men. However, this applied to both rural and urban Ghana.
279x361mm (300 x 300 DPI)  CKD was defined as being in moderately increased risk, high-risk, or very high-risk groups.
279x361mm (300 x 300 DPI)           132 However, in some settings the well-known inverse association between SES and CKD 133 seems to be absent, or even reversed. 140 There is therefore an interaction between individual SES and environmental factors, 141 such as food and sedentary life style in such populations [10][11][12] . Consequently, in those 142 settings, people with a higher SES might have higher CKD risk.

368
279x361mm (300 x 300 DPI)  CKD was defined as being in moderately increased risk, high-risk, or very high-risk groups.

415
416     494 higher rates of CKD in our study were not explained by the higher SES of that population as compared to 495 their rural counterparts. Our results indicate that this is due to the lack of a clear difference in the SES 496 distribution of rural and urban Ghana observed in this study, as well as to the lack of associations between 497 SES and CKD. Consistent with our findings, in a study conducted in Northern Tanzania SES did not 498 explain increased risk of CKD in urban Tanzania 26 . The lack of associations between SES and CKD 499 could probably and partly be explained by the process of epidemiological transition in relation to the 500 "diffusion theory" of ischemic heart disease mortality. This theory attributes the commencement of 501 ischemic heart disease to individuals in the high SES group due to their ability to afford behaviours 502 (smoking, alcohol and sedentary lifestyles) which increased risk of ischemic heart disease. The lower SES 503 groups were later affected partially because of improved living standards, unhealthy life style imitation 504 and urbanization. The higher SES groups were the first to embrace behavioural changes required to 505 decrease the risk of ischemic heart disease and this resulted in reversing the gradient 31 . The rapid 506 urbanization of some rural communities in the Ashanti region of Ghana and the 507 imitation of urban lifestyle could account for our finding. Also, it could be that whereas 508 the high SES group in urban Ghana has already embraced favourable behavioural 509 changes, those in rural Ghana are yet to do so 32 . This explains the observed association of 510 wealth index with CKD in rural Ghana but not in urban Ghana. Also, the interplay of other less

515 Strength and limitation
516 517 Our study presents several strengths. First, we used well-standardized study protocols across rural and 518 urban Ghana. Our study is also the first in Africa to use all three categories of CKD definition 519 (albuminuria, reduced eGFR and CKD risk) by KDIGO 2012 in assessing association of SES with CKD 520 in rural and urban setting, this provided more detailed information on CKD outcomes. The limitation of 521 intra laboratory variability in earlier studies was eliminated using the same standard operating procedures 522 in the same laboratory for running all samples for both rural and urban Ghana. The use of three constructs 523 of SES in this study also provides a much better holistic approach to assessing SES. Also, the distribution 524 of SES in our study reflects on the national data allowing for generalization of our findings. Our study 525 was limited by the use of cross sectional design, which prevented us from determining causality between 526 predictors and CKD progression. Furthermore, there were more women than men in our study due to the 527 higher response rate in women compared with men. However, this applied to both rural and urban Ghana.
279x361mm (300 x 300 DPI)  CKD was defined as being in moderately increased risk, high-risk, or very high-risk groups.