The Association Between Household Income, Food Security, and Prevalence of Chronic Kidney Disease in Elderly Patients

This study aimed to clarify the association between food security and the prevalence of chronic disease. We analyzed the variables of The Korea National Health and Nutrition Examination Survey V (2010–2012), and VI (2013–2015) while merging data of the food security questionnaire of four years. We included 15,945 participants, performed propensity score matched analysis by quartile of household income (i.e., low, low-mid, high-mid, high) and sex, and presented the results by age group. Systolic blood pressure and proportion of current smokers were signicantly higher in the elderly group, compared with the middle-aged group. The prevalence of hypertension, diabetes mellitus (DM), metabolic syndrome, and chronic kidney disease (CKD) did not differ signicantly by income level in the elderly group. The food security questionnaire revealed that food security insurance was signicantly lower in the low-income level (1st quartile), compared with that in the high-income level (4th quartile). The logistic regression analysis for the association between the prevalence of chronic disease and food insecurity conrmed no signicant association with hypertension and DM. Food insecurity might be associated with CKD prevalence, especially in the elderly population.


Introduction
Several studies have shown a varying prevalence of chronic diseases and inequality in life expectancy by socioeconomic status in western societies. Many researchers have con rmed a higher prevalence and mortality of several metabolic diseases in socioeconomically vulnerable groups [1,2]. Strati cation of socioeconomic status re ects the differentiation in factors such as health behaviors, food security, health risk behaviors, and educational, occupational, and income levels. Among them, food security is de ned as the ability to consume a balanced diet of nutrients, fruits, and vegetables, while limiting salt, saturated fat, and carbohydrates. In recent times, there have been studies on the relationship between food insecurity and chronic diseases such as hypertension, diabetes, cardiovascular disease, and chronic kidney disease (CKD). Deidra [3] showed an association between food insecurity and CKD in diabetic and hypertensive patients in the United States, whereas Banerjee [4] revealed that food instability is an independent risk factor for the development of endstage renal disease (ESRD) in patients with CKD. The latter study was based on data from the American Health and Nutrition Examination Survey. Although we cannot conclude that food insecurity has a powerful effect like other known risk factors for chronic diseases such as smoking or alcohol consumption, it may affect vulnerable groups including patients with chronic disease and the elderly. In line with this, a study on the Japanese population suggests an association between lower income levels and higher risks of CKD [5].
However, few studies focus on the contribution of education and income level, and food security to the prevalence of chronic diseases in the Korean population. Therefore, this study con rmed the effect of income level and food safety on the health status of Koreans based on data from the Korean National Health and Nutrition Examination Survey (KNHANES).

Results
Baseline characteristics.
Participants' basic characteristics are shown in Table 1. The middle-aged and elderly groups were matched by the propensity score of 3,957 people. According to their income level, they were divided into quartiles comprising 921 people in the low group, 999 people in the low-mid group, 1012 people in the high-mid group, and 1025 people in the high group. In the middle age group, no differences in systolic and diastolic blood pressure according to income level were found.
Abdominal circumference and BMI tended to be higher among those with lower income levels. Further, the Modi cation of Diet in Renal Disease (MDRD) eGFR and proportion of smokers was higher among those with lower income levels.
Regarding educational level, higher income levels were signi cantly positively associated with university education, and negatively associated with elementary school education.
In the elderly age group, no signi cant differences in blood pressure, abdominal circumference, BMI, and total cholesterol level were found by income level. Lower income levels were associated with higher MDRD eGFR. Further, educational level and smoking rate showed similar trends by income level, as in the middle-aged group.
Prevalence of chronic disease and medical service needs.
In the middle age group, lower income levels were associated with a higher prevalence of hypertension, diabetes, and metabolic syndrome. The prevalence of hypertension was 17.5% and 24.0% in high-and low-income groups respectively. Additionally, 6.3% and 10.0% of patients in the high-and low-income group were diagnosed with diabetes, respectively.
However, the prevalence of CKD did not differ between groups. Regarding the lack of access to necessary medical services due to economic reasons, 0.5% and 7.6% of those in the high-and low-income groups, respectively did not receive essential medical services owing to economic reasons (Table 2). However, no statistically signi cant difference was observed for this variable. In the elderly age group, unlike the middle-aged group, no differences were observed across the groups in the prevalence of hypertension, diabetes, CKD, and metabolic syndrome. However, there were signi cant differences regarding the lack of access to necessary medical services due to economic reasons according to income level. Food safety scores in the survey showed signi cant differences between groups (Table 3).
All of the food security questions were recoded into two categories [(often/sometimes vs. never) or (yes vs. no) or (almost every month/some months but not every month vs. only one or two months)]. Each item was given a score of 1, if the answer pointed often/sometimes, yes or almost every month/some months but not every month, or 0 for the rest responses. An additive total score was created and was classi ed into four levels of food security: 1) food secure [score: 0-2 (households with children); 0-2 (households without children)]; 2) food insecure without hunger [score: 3-7 (households with children; 3-5 (households without children)]; 3) moderate food insecure with hunger (8-12; 6-8); and 4) severe food insecure with hunger (13-18; 9-10) In the middle age group, the relative risk of hypertension according to income level was 1.49 times (95% CI 1.19-1.86) and 1.55 times (95% CI 1.23-1.94) higher in the low-income group, compared with that in the high-income group in Models 1 and 2, respectively (Table 4A). However, since there was no signi cant increase in risk in Models 3 and 4, the effect of education level and food safety on the prevalence of hypertension could not be con rmed. The relative risk of diabetes according to income level was 1.63 times (95% CI 1.17-2.28) and 1.61 times (95% CI 1.15-2.27) higher in the low-income group, compared with that in the high-income group in Model 1 and 2, respectively; however, the effect of education level and food safety could not be con rmed (Table 4B). There was no increase in the relative risk of CKD in any of the Models.
In the elderly age group, no increase in the relative risk of hypertension or diabetes in Model 1-4 was observed. The effect of income level, education level, and food security on the prevalence of hypertension and diabetes could not be identi ed.

Discussion
The relationship between socioeconomic status and chronic diseases has been of great interest in western society for several years. Socioeconomic factors comprise factors such as education and income level, occupation, and health behavior. Health behaviors are related to physical activity, avoidance of drinking and smoking, maintaining food safety, and obesity. Our study showed that among other socioeconomic factors, income level affects the prevalence of hypertension and diabetes in the < 65 years population. These relationships were not altered after being adjusted for food safety. However, we found that the effect of income level on the prevalence of chronic disease was diminished after adjusting for factors related to education level and health behaviors such as blood pressure and abdominal circumference. This nding could re ect that health behaviors such as smoking, abdominal circumference, and BMI differ by income level. Moreover, this nding is similar to those of previous studies, which present differences in the prevalence of chronic diseases by socioeconomic status in western societies [6,7,8]. It is also consistent with studies that emphasize the importance of health behavior [9]. However, this nding contrasts those of other studies, which reveal that health behaviors do not affect the relationship between social class and the prevalence of chronic diseases in the Korean population [10].
In the elderly age group over 65 years, although the level of education varied by income, the effect on the prevalence of diabetes and hypertension could not be con rmed. Moreover, unlike that of the population under 65 years, no increase in health-risk behaviors was observed, which resulted in an increase in the risk of diabetes and hypertension such as BMI and waist circumference in the low-income group. Although food insecurity and smoking rates tended to be higher among those with lower income, other health behaviors did not increase in the elderly group. This may explain why income level does not affect the prevalence of diabetes and hypertension in the elderly.
It is well known that socioeconomic level in uences kidney health, as the prevalence of chronic metabolic diseases causing kidney disease, as well as drinking and physical activity is closely related to socioeconomic status [11,12,13].
Furthermore, food insecurity may be a risk factor for independent kidney disease progression as well as the occurrence of chronic diseases [14,15]. Insu cient intake of fruits and vegetables and consumption of high-energy processed foods leads to an increase in salt and saturated fatty acid intake, which is believed to aggravate kidney disease by increasing the dietary acid load [16,17]. Contrary to the results of previous studies, our study could not identify the effect of income and education levels on CKD. However, we found that food insecurity increases the risk of CKD in the elderly population (over age 65 years); moreover, the effect of food instability was attenuated when adjusted for height, abdominal circumference, and calorie intake.
Our study shows that the difference in chronic diseases by income level is similar to that found in western societies in the > 65 years Korean population. Therefore, we could infer that with increasing health risk behavior such as obesity and smoking in the lower-income level group, there is a greater need for social efforts to improve these behaviors. However, there was no difference in factors related to health-risk behaviors by income level in the elderly population. These characteristics would allow for socioeconomic factors like food instability to have a greater effect than preexisting traditional metabolic causes. This is supported by the fact that food insecurity could worsen glycemic control in diabetic patients [18]. Some studies have also shown a relationship between poverty and food insecurity [19,20], while others have revealed that poorer areas have a higher incidence of ESRD [21,22].
Our study has limitations in that occupational factors are not included among the socioeconomic status factors; moreover, cardiovascular diseases are not included among chronic diseases.
In conclusion, we found that food insecurity is associated with CKD in the Korean elderly population. Considering this nding in combination with disparities in access to essential medical services according to income level, we need to establish a different approach to tackle food security in the elderly population, compared with that of the middle age group.

Study population and baseline data
To study the association between household income, food security, and the prevalence of chronic disease, we used data based on the KNHANES, which has been conducted by the Division of Health and Nutritional Survey in the Korean Centers for Disease Control and Prevention from 2012-2015. The KNHANES is a population-based, cross-sectional survey with nationally representative samples of the civilian non-institutionalized Korean population. Among the 31,017 individuals surveyed from 2012-2015, 15,945 people were included, after eliminating those who did not complete the survey. We measured the propensity score matching by income level, which was categorized into four levels, and sex. Using the propensity score, participants were divided by age (< 65 years old: middle-aged adult, > 65 years old: elderly age adult) and gender. Finally, 3,957 participants were quali ed for statistical analysis (Table 1).

Socioeconomic and clinical measurements
Medical history and demographic data were collected through three component surveys: a health interview, health examination, and nutrition survey. Annual household income was divided into four quartiles using health interview data.
The educational status was strati ed into three categories including elementary, middle, high school, and college, based on the academic background. Body mass index (BMI) was calculated from anthropometric data, where weight was divided by height squared. Diabetes was de ned by self-report or measured hemoglobin A1c level 6.5%. Hypertension was de ned as self-report, measured average systolic blood pressure > 140 mm Hg, measured average diastolic blood pressure > 90 mm Hg, or reported using antihypertensive medications. CKD is de ned as estimated glomerular ltration rate (eGFR) of 15 to 59 mL/min/1.73 m 2 or urinary albumin creatinine ratio (ACR) > 30 mg/g.

Food security
To assess food security, we used the Food security questionnaire for Korean National Health and the Nutrition Examination Survey that addresses dietary behaviors, food frequency. and food intake. The dietary behavior questionnaire includes meal skipping, eating out, eating with family, taking dietary supplements, nutrition education, use of food labeling, and food security. The food frequency questionnaire comprises 63 food items that are key sources of energy and nutrients (Table 3).

Ethical considerations
This study was based on data collected during the KNHANES and the secondary analysis of a large open data set. The KNHANES methodology has been presented in detail previously and further details "The 5th KNHANES Sample Design" and accessible at https://knhanes.cdc.go.kr/knhanes/index.do

Statistical analysis
Regression analysis was performed with four models to identify the effects of income level, food security, and education level on chronic diseases as socioeconomic factors. In each age group, participants were matched by quartile of household income (low, low-mid, high-mid, high) and sex on base of propensity score. Baseline characteristics according to income level in each age group were compared by the Mann-Whitney U test for continuous variables and the 2 test for categorical variables. Multivariate logistic regression analysis was performed with four models to identify the effects of income level, food security, and education level on chronic diseases as socioeconomic factors yielding odds ratio (OR).