Summary
Background and objectives According to the cost of health care utilization systems, there may be regional differences in the relative strength of association of income and education-based socioeconomic status measures with CKD. This study investigated the relative strength of the association of income and education with CKD in a United States and a Dutch population.
Design, setting, participants, & measurements This cross-sectional study examined individuals who participated in the 1999–2002 National Health and Nutritional Examination Survey (NHANES) and in Prevention of Renal and Vascular End-stage Disease (PREVEND 1997–1998), general population-based cohorts in the United States and The Netherlands, respectively. The main outcome was CKD, defined as estimated GFR <60 ml/min per 1.73 m2 (using creatinine) or albuminuria ≥30 mg/24 hours or albumin-to-creatinine ratio ≥30 mg/g.
Results In NHANES (n=6428), income was strongly associated with CKD (adjusted odds ratio, 2.34 [95% confidence interval (CI), 1.68 to 3.27]; P for trend<0.001) but education was not (adjusted odds ratio, 1.62 [95% CI, 0.87 to 2.25]; P for trend=0.05]. In contrast, in PREVEND (n=7983), low income was weakly associated with CKD whereas low education had a strong association. The fit of the logistic regression model estimating association of income and education with CKD was significantly improved only after income was added in NHANES (P<0.001) and education was added in PREVEND (P=0.01). Sensitivity analyses that used other CKD-defining variables and restricted analyses to participants <65 years of age resulted in similar findings.
Conclusion In the United States, where access to health care is traditionally income dependent, income appeared more strongly associated with CKD than in The Netherlands, where education showed a stronger association.
Introduction
The incidence and prevalence of CKD are increasing, and CKD has been recognized as a global public health problem (1). Many patients with CKD are not aware of having this disease and consequently are not treated. This has led to the recommendation that high-risk populations, such as people with known diabetes or hypertension, should be screened (2). Although some have argued against screening for CKD, the prevailing guidelines for CKD recommend that high-risk individuals should be screened (3). Lower socioeconomic status (SES) has been suggested to be associated with CKD, and some have additionally recommended screening this high-risk group (4). SES can be measured by several individual and neighborhood characteristics, but income and education are among the important defining components for SES. This brings forward the question of identifying a measure of SES that is the strongly linked with CKD.
The strength of income and education, in terms of determining health outcomes such as CKD, varies according to regional factors (5,6). We hypothesize that one of the regional factors can be the cost of accessing health care. Therefore, in a formal comparison of income and education in the United States, where access to health care is traditionally income dependent (7), income might have a stronger association with CKD than does education. Similarly, income might be less strongly associated with CKD in populations where access to health care is relatively income independent; for example, in The Netherlands, access to preventive and primary health care is not fundamentally different between the rich and poor (8–10). Education also exhibits a significant role in living a healthy lifestyle and using preventive health care services (11,12). In The Netherlands, compared with income, education might have a stronger association with CKD. To date, evidence testing these hypotheses is lacking.
Therefore, this study sought to formally investigate the relative strength of association of income and education with CKD prevalence in a United States and a Dutch population.
Materials and Methods
Study Design and Population
We used baseline data of the National Health and Nutritional Examination Survey (NHANES) and the Prevention of Renal and Vascular ENd-stage Disease (PREVEND) study participants.
Briefly, the NHANES surveys are cross-sectional, multistage, stratified, clustered probability samples of the noninstitutionalized civilian population in the United States (details can be found at http://www.cdc.gov/nchs/nhanes.htm). For the current study, data of participants aged 28–75 years from 1999–2000 and 2001–2002 NHANES surveys with measured serum creatinine were combined (n=7402) (13).
The PREVEND study is a prospective cohort study initiated in 1997 when 40,856 people aged 28–75 years were screened from the general population of Groningen, The Netherlands. From these participants, a cohort (n=8952) was formed that was screened in detail between 1997 and 1998 and followed up over time. PREVEND has been shown to be representative of the Dutch population (14,15); details have been published elsewhere (16,17).
Of the 7402 NHANES participants and 8592 PREVEND participants, we excluded those lacking information on education and income (n=93 in NHANES and n=527 in PREVEND), serum creatinine (n=882 in NHANES and n=67 in PREVEND), and albumin (n=597 in NHANES and n=18 in PREVEND), leaving 6428 and 7983 participants for analysis in NHANES and PREVEND, respectively. Characteristics of missing participants were similar to those of final study participants.
Measurements and Definitions
Income and Education.
For NHANES and PREVEND, income was recorded as poverty income ratio (PIR). The PIR was calculated by dividing the family income by the country-dependent poverty threshold specific for family size and year (18). The PIR was divided into quintiles. In PREVEND, income was not recorded as a continuously distributed variable but had a discrete distribution (classes); as a result, not all quintiles covered exactly 20%. For NHANES, quintiles 1 to 5 ranged from 4.12 to 5.00, from 2.39 to 4.11, from 1.38 to 3.38, from 0.80 to 1.37, and from 0 to 0.79, respectively. For PREVEND, quintiles 1–5 ranged from 1.62 to 3.70, from 1.88 to 2.21, from 1.24 to 1.59, from 0.78 to 1.23, and from 0 to 0.77, respectively. In NHANES and PREVEND, education levels were categorized according to the International Standard Classification of Education (ISCED) (19). Educational levels concern bachelor, master or doctorate graduate (level 1), postsecondary or nontertiary or short-cycle tertiary education (level 2), upper secondary education (level 3), lower secondary education (level 4), and primary or below primary education (level 5). With the ISCED, the association of education level with CKD can be compared between both regions.
Definition of CKD
For both PREVEND and NHANES, GFR was estimated with the CKD-Epidemiology Collaboration equation using serum creatinine as a marker of renal function (20). Albuminuria was expressed as the albumin-to-creatinine ratio in NHANES and as average urinary albumin excretion in PREVEND. CKD was defined as estimated GFR (eGFR) < 60 ml/min per 1.73 m2 or albuminuria ≥ 30 mg/24 hours (30 mg/g), or both, according to prevailing guidelines (21).
Serum Creatinine.
Serum creatinine was measured by the kinetic alkaline picrate method in NHANES and by dry chemistry (Eastman Kodak, Rochester, NY) in PREVEND.
Urinary Albumin and Creatinine.
In NHANES, urinary albumin and creatinine concentrations were measured from spot urine samples by the modified kinetic Jaffe method with a Synchron AS/Astra Analyzer (Beckman Coulter). In PREVEND, urinary albumin concentration was measured in two 24-hour urine collections by nephelometry (BNII; Dade Behring Diagnostic, Marburg, Germany).
Other Covariates.
During the physical examination, weight, height, and BP were measured. Body mass index was calculated as weight (kg) divided by the square of height (m2), and BP was measured in accordance with the recommendation of the American Heart Association (22). Information on cardiovascular disease history and smoking was obtained from questionnaires; information on diabetes was obtained from questionnaires as well as objective measurements (13,15). Current smokers were defined as those who had smoked at least 100 cigarettes during their lifetime and, at the time of the interview, reported current smoking or smoking in the year before the baseline measurement (23). Diabetes was defined on a single fasting plasma glucose level ≥126 mg/dl, a nonfasting glucose level ≥200 mg/dl, or the use of oral antidiabetic drugs/insulin or self-report (24).
Statistical Analyses
To compare the strength of association of income and education with CKD, we performed logistic regression analysis. We ran two parallel models with education and income separately. Model 1 adjusted for age, sex, and race. Income and education were added simultaneously in the subsequent model 2 to explore their relative strength of association with CKD. Interaction terms were also investigated. Further, we performed likelihood ratio tests to compare the fit of models with education and income.
For NHANES, to address population representativeness, we used recommended 4-year sampling weights from NHANES 1999–2002 (13). For PREVEND, a specific weighing factor was used to account for the overrepresentation of participants with higher levels of albuminuria.
We performed several sensitivity analyses. First, we redefined CKD for PREVEND by using the albumin-to-creatinine ratio instead of 24-hour urinary albumin excretion and repeated the analyses. Second, we conducted stratified analyses by age (in both cohorts) and race (in NHANES), accounting for higher prevalence of poverty and CKD in participants ≥65 years of age and the higher number of non-Hispanic blacks and Mexican whites in the NHANES compared with the PREVEND study (25). Third, we additionally tested the association between SES measures and CKD after adjusting for other covariates (smoking, diabetes, BP, body mass index [BMI], and cardiovascular disease). Fourth, we also tested associations for low renal function and albuminuria, separately. Fifth, we tested the observed differences in association between the two countries by introducing the interaction between SES and cohorts versus CKD prevalence in the combined data. Finally, we tested the association between income and CKD in NHANES and PREVEND using absolute income levels (expressed per family member) to explore the robustness of results against a differential classification of income levels. A P value <0.05 was considered to indicate statistical significance. All analyses were conducted using Stata software, version 11 (Stata Corp., College Station, TX).
Results
Mean age (47.3 years versus 48.9 years), percentage of men (48% versus 49%), and percentage of whites (69% versus 96%) were lower in NHANES than in PREVEND. The prevalence of CKD was 14% in NHANES and 9% in PREVEND. Mean serum creatinine ± SD was 0.83±0.37 mg/dl in NHANES and 0.95±0.18 mg/dl in PREVEND. Baseline characteristics of the NHANES and PREVEND cohorts are presented in Tables 1 and 2, respectively. Percentage distribution of income in education levels and education in income levels for NHANES and PREVEND are shown in Supplemental Figures 1 and 2, respectively.
Baseline characteristics of the National Health and Nutrition Examination Survey study population according to income quintiles and education levels
Baseline characteristics of the Prevention of Renal and Vascular End-stage Disease study population according to income quintiles and education levels
Income Levels
Participants in the highest income group were 7.1 times and 3.3 times richer than participants in the lowest income group in NHANES and in PREVEND, respectively. In NHANES, people with lower income levels had a higher prevalence of impaired eGFR, increased albuminuria, and CKD. A similar trend was observed for hypertension, smoking, high BMI, and diabetes in NHANES. In NHANES and PREVEND, prevalence rates of smoking were higher with lower income levels. The prevalence of hypertension was also highest in people with the lowest income level, but the pattern was less clear in PREVEND than in NHANES. Similarly, there was no clear pattern regarding increased albuminuria in PREVEND.
Education Levels
In NHANES and PREVEND, 16% of the study population had primary education. Participants with lower education levels were older in both cohorts; they had a higher BMI in PREVEND but not in NHANES. The lower education levels had higher proportions of patients with high albuminuria, impaired eGFR, and CKD in PREVEND. Similar trends were observed in NHANES. In NHANES, rates of hypertension and diabetes were higher in participants with lower education levels, but for smoking the rate was not highest in the lowest education level. In PREVEND, rates of hypertension, diabetes mellitus, and smoking were higher in participants with lower education levels.
Association of Income and Education Levels with CKD
The adjusted odds ratios for the income and education levels with CKD, for both NHANES and PREVEND, are presented in Tables 3 and 4, respectively. In model 1 (adjusted for age, sex, and race), income and education levels were independently associated with CKD. In NHANES, after simultaneous adjustment in model 2, only income level showed a graded and significant association with CKD. In contrast, in PREVEND, education showed a stronger significant association with CKD. Figure 1 shows odds ratios for income and education levels adjusted for age, sex, race, and education (for income)/income (for education) in NHANES (left panel) and in PREVEND (right panel). In NHANES, adding income to the model with education improved the fit of the model more strongly (P<0.001) than did adding education to the model with income (P=0.07). In contrast, in PREVEND, the addition of education to the model with income resulted in a statistically stronger improvement in the fit of the model (P=0.01) than did adding income to the model with education (P=0.05). The interactions of age and race with education and income for CKD prevalence were not statistically significant in either NHANES or PREVEND.
Association of income quintiles with CKD prevalence in the National Health and Nutrition Examination Survey and in the Prevention of Renal and Vascular End-Stage Disease study
Association of education levels with CKD in the National Health and Nutrition Examination Survey and in the Prevention of Renal and Vascular End-Stage Disease study
Association of income quintiles and education levels with CKD in the National Health and Nutrition Examination Survey (NHANES) (left panel) and in Prevention of Renal and Vascular End-stage Disease study (right panel). Presented are odds ratios (95% confidence intervals) for income quintiles (circles and lines) and education levels (squares and lines) adjusted for age, sex, race, and education (for income) or income (for education). Reference categories (category 1) were highest income quintile and highest education level, respectively.
Sensitivity Analyses
In PREVEND, the results from redefining CKD by using an albumin-to-creatinine ratio of ≥30 mg/g instead of 24-hour urinary albumin excretion were essentially similar to the earlier results. Stratified analyses by race in NHANES showed similar stronger associations of income levels with CKD prevalence than of education in all racial groups (Supplemental Table 1). Results for participants age <65 years were essentially similar to the overall results (Supplemental Table 2). In participants age ≥ 65 years, the association of income and education levels with CKD was not statistically significant in either NHANES or PREVEND. After adjustment for smoking, diabetes, BP, BMI, and cardiovascular disease along with age, sex, and race, the association of CKD with income was stronger in NHANES (Supplemental Table 3) while the association with education was stronger in PREVEND (Supplemental Table 4). Results for the association of low renal function and albuminuria with SES measures are presented in Supplemental Tables 5 and 6, respectively. Analysis of combined data from both studies showed a significant interaction of cohort with income (P=0.01) and education (P=0.01) versus CKD prevalence. The associations of CKD with absolute income values also showed stronger income-CKD associations in NHANES than in PREVEND (Supplemental Figure 3).
Discussion
This study examined income and education for the relative strength of their association with CKD prevalence in the United States and The Netherlands. In NHANES, low income was more strongly associated with the likelihood of having CKD than was low education. In contrast, in PREVEND, low education was more strongly associated with CKD prevalence. Furthermore, the association of income with CKD in the United States cohort was stronger than the association of education with CKD in the Dutch cohort.
The stronger association of income with CKD in the United States cohort compared with the Dutch cohort might be due to two factors. First, access to health care in the United States is income dependent. Therefore, along with affecting health lifestyle choices (e.g., access to healthy diet), low income additionally affects health care access in the United States and ultimately affects the prevalence of CKD (26). Second, in the United States income inequalities are bigger (27), as also reflected by the higher PIR between the fifth and first quintiles in our study. In the Dutch cohort, disparities in CKD prevalence are more strongly determined by education, but not as strongly as by income in the United States. This finding suggests that the overall role of SES in The Netherlands is less marked than in the United States. In this study, analysis of combined NHANES and PREVEND data also showed differential associations of SES measures with CKD in both cohorts.
We used PIR instead of absolute income levels. Poverty thresholds in the United States and The Netherlands are measured from similar parameters of an individual’s purchasing power and financial ability for social inclusion that are country specific (25). Hence, levels of poverty are likely to be different in both countries. We believe that using the PIR (calculated from regional poverty threshold) enables a more valid comparison between countries than using absolute income levels. We performed a sensitivity analysis with absolute income levels (expressed per family member), which showed that the association of income with CKD was still stronger in the United States than in The Netherlands. This illustrates that a differential classification of income levels in NHANES and PREVEND from PIR is not responsible for the country differences.
In a separate analysis for participants age ≥65 years, we found no significant associations between SES measures and CKD. This may be due to lower power but could also be due to the fact that in developed economies, elderly patients usually receive state financial benefits and aid in accessing health care.
Income can affect CKD incidence, prevalence, and progression via several biologic, environmental, or behavioral mechanisms. Strong links of low income with poor food and lifestyle choices can lead to diabetes mellitus, obesity, and hypertension, which are established risk factors for CKD (28). Along with poor food and lifestyle choices, in a gene-environmental effect, certain genes (e.g., APOLI) linked to kidney disease are known to have more adverse effect among individuals living in poor neighborhoods (29).
For education, the factors linking low education to CKD are complex. Education can influence health and health-related behavior in many ways, including an individual’s cognitive ability to resolve adverse health issues, self-control, uptake of information, and use of health care resources; effective patient-physician communication; and food and lifestyle choices (30–32).
Strengths of this study are the use of detailed baseline information for relevant covariates from two large community-based study populations. We used PIR as measure of income. Along with household income, PIR accounts not only for household size but also for family composition, which has been a limitation in several previous studies of health disparities (33,34).
This study also had limitations. We studied general population cohorts with participants especially in the earlier stages of CKD. In these stages, CKD is unlikely to affect earning potential. However, because data on exposure and outcome are measured at one point in time and reverse causation cannot be excluded, results should be interpreted with caution. Second, the studies used different laboratory methods to measure CKD markers. However, because conclusions are mainly drawn on within-study differences in associations of SES measures with CKD, this will not affect our results. Third, PREVEND predominantly comprises whites, whereas NHANES also comprises non-Hispanic blacks and Mexican Americans, groups known to have a higher CKD prevalence. Stratified analysis on race in NHANES, however, resulted in essentially similar findings in all racial subgroups (Supplemental Table 1). Fourth, some might consider the lack of adjustment for the associations of SES measures with CKD prevalence for health (behavior) factors to be a limitation. These risk factors are more prevalent in persons with lower SES but should not be considered confounders because they are probably situated in the causal pathway by which low SES leads to CKD (35). However, to test whether the associations change after accounting for these factors, we additionally adjusted for smoking, diabetes, obesity, BP, and cardiovascular disease along with age, sex, and race. Essentially similar results were obtained. Finally, although we tried to keep the time period in which the NHANES and PREVEND studies were performed as similar as possible, a relatively small difference exists (1999–2002 and 1997–1998, respectively), which should be taken into account in interpreting our results.
In wake of recent developments in health care organization in the United States and Western Europe, this study may have important public health implications. The results indicate that in the United States disparities in health care access dependent on income may have negative consequences for health and consequently favor ensuring affordable care irrespective of income. In The Netherlands, improvement of education attainment might be one of the correct policy responses if, for example, there were no alternative (or cheaper) methods for addressing CKD-related health inequalities among different socioeconomic classes. Furthermore, our findings indicate that the definition of low SES as high-risk group to screen for CKD may differ between societies dependent on the organization of reimbursement of care. Some experts suggest there is insufficient evidence of improved health outcomes in asymptomatic adults from early identification of CKD (36). However, prevailing guidelines for CKD suggest screening for target populations (3). Research consistently demonstrates that taking into account income and education is important in screening for chronic diseases (37). Similarly, suggesting either education- or income-based screening for CKD might improve rates of participation in CKD screening. It might also help in effective use of limited resources. In the current economic scenario, reorganization of health care from a socioeconomic perspective might benefit from future research comparing the role of SES measures in overall health disparities in the United States and Western Europe.
In conclusion, income has a stronger association with CKD in the United States population than in a Dutch population. Screening strategies addressing social inequalities in CKD prevalence should therefore be region specific and use income-based social stratification in the United States and education-based social stratification in The Netherlands.
Disclosure
None
Acknowledgments
We thank NHANES and PREVEND study participants and the staff who helped in data collection.
This study did not receive any specific funding. All authors received funding from their respective institutes. NHANES is funded by the National Center for Health Statistics and PREVEND is funded by the Dutch Kidney Foundation.
The funding bodies had no role in the design and conduct of the study, in the collection, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.
The abstract of this manuscript was published in the abstract supplement of the American Society of Nephrology Kidney Week 2012.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
This article contains supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.12521212/-/DCSupplemental.
- Received December 11, 2012.
- Accepted May 28, 2013.
- Copyright © 2013 by the American Society of Nephrology
References
In this issue
Jump to section
More in this TOC Section
Original Articles
Cited By...
- Long-term effects of socioeconomic status on the incidence of decreased glomerular filtration rate in a Southeast Asian cohort
- Associations between socioeconomic status and chronic kidney disease: a meta-analysis
- The interaction of socioeconomic position and type 2 diabetes mellitus family history: a cross-sectional analysis of the Lifelines Cohort and Biobank Study
- CKD Prevalence Varies across the European General Population
- Prevalence and Correlates of CKD in Hispanics/Latinos in the United States
- Added Value of Screening for CKD among the Elderly or Persons with Low Socioeconomic Status