Socio-economic status and Cardiovascular Risk Factors in Rural and Urban Areas of Vellore, Tamilnadu, South India

Abstract

Background

We examined associations between socio-economic (SES) indicators and cardiovascular disease (CVD) risk factors among urban and rural South Indians.

Methods

Data from a population-based birth cohort of 2,218 men and women aged 26-32 years from Vellore, Tamilnadu were used. SES indicators included a household possessions score, attained education, and paternal education. CVD risk factors included body mass index, waist circumference, blood pressure, glucose tolerance, plasma cholesterol and triglyceride levels, and consumption of tobacco and alcohol. Multiple logistic regression analysis was used to assess associations between SES indicators and CVD risk factors.

Results

Most risk factors were positively associated with possessions score in urban and rural men and women, except for tobacco use, which was negatively associated. Trends were similar with the participants’ own education, and paternal education, though weaker and less consistent. In a concurrent analysis of all three SES indicators adjusted for gender and urban/rural residence, independent associations were observed only for the possessions score; compared with those in the lowest fifth of the possessions score, participants in the highest fifth had a higher risk of abdominal obesity (OR=6.4, 95%CI 3.4, 11.6), high total cholesterol to HDL ratio (OR=2.4, 95%CI 1.6, 3.5) and glucose intolerance (OR=2.8, 95%CI 1.9, 4.1). Their tobacco use (OR=0.4, 95%CI 0.2, 0.6) was lower. Except hypertension and glucose intolerance, risk factors were higher in urban than rural participants independently of SES.

Conclusion

In rural and urban populations, higher SES, as reflected by household possessions, was associated (apart from tobacco use) with a more adverse CVD risk factor profile.

Introduction

In recent years, cardiovascular disease (CVD) has emerged as a leading cause of death in developing countries¹. It is important to identify and target people who are at risk for CVD, given that a third of all deaths are expected to be due to CVD by 2020. Studies have shown socio-economic patterning in the prevalence of risk factors for CVD, including obesity^2,3, smoking⁴ and lipid profile⁵. In developed countries, the association between socio-economic status (SES) and CVD risk factors is negative, with a higher prevalence of CVD risk factors among people of lower SES^6,7. However, the evidence from developing countries, including India, has been inconsistent^4,8-11. In addition, there is scant information on differences in socio-economic patterning of CVD risk factors between urban and rural areas of India¹².

Our primary goal was to assess the prevalence of CVD risk factors, and their associations with SES, in urban and rural Indian settings. Most previous studies on this issue have used a single indicator of SES such as education, income or wealth index. However, studies have found different effects for individual SES indicators¹³. Education reflects degree of knowledge and skill, along with the ability to attract material wealth. On the other hand, income reflects current economic or materialistic welfare. Wealth index which is based on asset ownership could be considered as an indicator of long term economic status, as household assets are unlikely to change in response to short term economic shocks. Since SES indicators are interrelated to a certain extent, the effects of each of these indicators could be masked by the others. For instance, education could have a direct effect on CVD risk factors or reflect the effect of income or wealth. Further, differences in CVD risk factors between SES groups could arise in early life, and studying paternal education as a measure of childhood SES would provide an opportunity to compare the relative effects of current and childhood SES on CVD risk factors. Thus, examining the independent effect of these indicators could provide better understanding regarding the underlying mechanisms and help in identifying specific target groups and plan effective preventive programs for CVD. We have therefore used multiple indicators of SES (a score based on household possessions, educational status and paternal education) to assess the independent effect of each of these indicators on CVD risk factors among South Indian adults.

Methods

Participants and Settings

Original study

Data were from a cohort of infants born during 1969-73 in Vellore district, in Tamilnadu state. Twenty four wards in Vellore town representing different socio-economic strata and 25 of 42 villages from nearby rural settings were randomly selected. A total of 14,147 pregnancies were identified, which resulted in 10,691 singleton livebirths. Of the 10,691 singleton livebirths, 47% moved outside the study area as many mothers traditionally returned to their parents’ home for delivery and were not available for further examination. The remaining (n= 5,753) were measured (birthweight, length and head circumference) within 120h of delivery by trained personnel. These measurements were repeated during infancy (1-3 months), childhood (6-8 years) and adolescence (10-15 years). Further details regarding the establishment of this cohort are described elsewhere¹⁴.

Follow up study

In 1998-2002, when the cohort was aged 26-32 years, an attempt was made to retrace the members of the original cohort and study the relationship between early childhood factors and cardiovascular risk factors¹⁵. All subjects who were singleton births and whose parental and birth measurements were available (n=4,052) were considered eligible for the follow up study. Of the latter, 2,572 were traced by health workers and 2,218 agreed to participate in the study, 997 from urban areas (men: 544; women: 453) and 1,221 from rural areas (men: 617; women: 604). Information on SES indicators, anthropometric and blood pressure measurements were obtained during this period (1998-2002).

Variables

SES Indicators

We used the individual’s educational level, paternal education and a score based on household possessions as indicators of SES. Educational status and paternal education were each recorded as 1 of 4 categories from ‘no schooling’ (category 1, 0 years), primary and middle school (category 2, 1 to 8 years), high school and higher secondary (category 3, 9 to 12 years), graduates and professional (category 4, >12 years). Paternal education was used as a measure of childhood SES.

For the possessions score, participants were asked if they owned each of the following household items/amenities: electricity, fan, bicycle, radio, motorized 2-wheeled vehicle, gas stove, television, cable television, electric mixer, electric grinder, electric air cooler, washing machine, car, air conditioner, computer, television antenna, and telephone. One approach to such data involves summing the number of possessions, but this has the disadvantage of assigning equal weight to each item, regardless of its value or utility. We therefore created a composite score using weights obtained from Principal Component Analysis (PCA)¹⁶ and grouped the first principal component by quintiles.

Anthropometric measurements and biochemical analyses

Physical measurements included weight, height, blood pressure (using an automated device OMRON 711)¹⁷ and waist and hip circumferences. These were obtained according to standardised protocols by one of two measurers. Plasma glucose and lipid levels were measured fasting (12 hours overnight) and glucose was measured 30, 60 and 120 minutes following a 75 g oral glucose load. Plasma glucose levels were measured by glucose oxidase/peroxidase methods, and serum lipids using commercial enzymatic kits (Roche Diagnostics, Germany) on a Hitachi 911 autoanalyser (USA).

Risk factors for CVD

Body mass index (BMI) was calculated as weight (kg) divided by height (m) squared. BMI was categorized as underweight (<18.5), normal (18.5-22.9), overweight (23.0-24.9) and obese (25 and more)¹⁸. Abdominal obesity was defined as a waist circumference > 90 cm for men and > 85 cm for women¹⁹. Hypertension was defined as a systolic blood pressure ≥ 140 mmHg and/ or diastolic blood pressure ≥ 90 mmHg²⁰. Participants who had been diagnosed by a doctor as having hypertension and used antihypertensive drugs (N=1) were classified as hypertensive. A high total cholesterol to HDL ratio was defined as ≥ 4.5 and a high triglyceride concentration as ≥ 1.69 mmol/l²¹. Diabetes was defined as a fasting blood glucose ≥ 7 mmol/l or a 120 min value of ≥ 11.1 mmol/l. Impaired glucose tolerance (IGT) was defined as fasting blood glucose <7 mmol/l and a 120 min value of ≥ 7.8 mmol/l, but < 11.1 mmol/l and impaired fasting glucose (IFG) as a fasting blood glucose of ≥ 6.1 mmol/l and < 7 mmol/l²². Participants who had been diagnosed by a doctor as having diabetes and used medication for diabetes (N=8) were classified as diabetic.

Current alcohol consumption was assessed by questioning the participants about their frequency and volume of intake of spirits, beer, and wine and these were converted into units of alcohols per week (23 ml sprits, 1 unit =574 ml beer or 125 ml wine). They were categorised as current consumers or non-consumers of alcohol. Tobacco consumption was recorded as whether the subjects smoked (cigarettes, bidis, cigars, or hookah), chewed (raw tobacco or with pan), or inhaled (snuff). Subjects were categorized simply as current tobacco users or as nonusers.

Statistical Analysis

All descriptive analyses of the risk factors were performed separately for men and women by the place of residence (rural or urban). First we examined the distribution of risk factors according to the socio-economic indicators, by calculating prevalences, with 95% confidence intervals. P values for trends in CVD risk factors were obtained by treating the SES indicators as continuous variables in the logistic regression analyses. Second, stratified analyses of CVD risk factors by place of residence (urban or rural) were performed to consider potential interactions between place of residence and SES. These analyses revealed a similar pattern of associations of the SES variables with CVD risk factors in urban and rural residents. Finally, we used multiple logistic regression analyses to estimate the independent effect of different socio-economic indicators on risk factors. All analyses were done using STATA 10.0 (StataCorp, College Station, Texas, USA).

Results

Data from a total of 2,218 participants were available for the analysis, 997 urban participants (544 men and 453 women) and 1221 rural participants (617 men and 604 women). Their age ranged from 26 to 32 years with a mean of 28.3 years (SD=1.1). Table 1 shows the prevalence of CVD risk factors stratified by gender and place of residence. Urban men had the highest prevalence of abdominal obesity, high total cholesterol to HDL ratio and triglyceride levels, hypertension, diabetes, tobacco and alcohol use. The prevalence of obesity, overweight and IGT was highest among urban women. Rural women had the highest prevalence of underweight. Similar percentages of men in the urban and rural populations used tobacco, while urban men were more likely to consume alcohol. Very few women were tobacco users, and none consumed alcohol.

As shown in Table 2, rural residents were less likely than urban participants to be in the highest fifth of the possessions score, or in the highest education category. Household possessions score was positively correlated with education status (Kendall tau (τ) =0.44, P<0.001) and paternal education status (Kendall tau (τ) =0.42, p<0.001), as were individual and paternal education status (Kendall tau (τ) =0.36, p<0.001). Paternal education levels showed large differences by urban-rural residence. About 42.2% of the rural adults had fathers who received no formal education, compared with 22.7% of urban adults. However, the adults’ own educational levels did not vary greatly by urban-rural residence; 7.5% of urban adults had no formal education, compared with 21.5% of rural adults.

Body mass index and waist circumference

Tables 3-5 show the prevalence of risk factors according to the SES indicators, stratified by gender and place of residence. In general, there were increasing trends in the prevalence of obesity, overweight and abdominal obesity with the possessions score, in both sexes and in both urban and rural areas (Table 3). The prevalence of obesity was highest among urban women and abdominal obesity was most prevalent among urban men from the highest fifth of the possessions score. In contrast, the prevalence of underweight decreased with increasing possessions score and was highest among rural women in the lowest fifth of the possessions score. Although there were similar trends in the obesity measures with other SES indicators (Tables 4 and 5), these were weaker and less consistent, especially among rural participants.

Lipid profile, blood pressure and glucose tolerance

The prevalence of a high total cholesterol to HDL ratio, high triglycerides, hypertension and glucose intolerance (either IGT, IFG or diabetes) tended to increase with higher possessions score in both genders and in both urban and rural participants (Table 3). These trends were stronger in urban than in rural participants, and in men than women, with the exception of glucose intolerance. In contrast, associations of risk factors with educational status (Table 4) were present only for high total cholesterol to HDL ratio in urban men, and those with paternal education (Table 5) were present only for high total cholesterol to HDL ratio in rural men, hypertriglceridaemia in urban women, and glucose intolerance in rural women.

Tobacco and alcohol use

These analyses were carried out for men only. The prevalence of tobacco use was highest among urban men in the lowest fifth of the possessions score and fell with increasing score (Table 3) and with increasing educational status in urban and rural men (Table 4), and with increasing paternal educational status in rural men (Table 5). Alcohol use was unrelated to any of the SES indicators among urban men, but fell with increasing possessions score and educational status among rural men.

Independent effects of SES indicators on risk factors

Table 6 shows the odds ratios (OR) for risk factors according to fifths of the SES variables, adjusted for each other along with gender and place of residence. A higher household possessions score was associated with increased odds of obesity, overweight, abdominal obesity, high total cholesterol to HDL ratio, high triglycerides and glucose intolerance. Household possessions score was inversely associated with underweight and tobacco use. The individual’s education level did not show independent effects on most of the CVD risk factors in this analysis, however, higher educational status was associated with a reduction in tobacco use. Having a father who was a graduate was associated with higher odds of obesity, abdominal obesity and tobacco use. However, there were no trends across all categories of paternal educational status. Most risk factors were higher in urban than in rural participants independently of SES indicators; exceptions were hypertension and glucose intolerance.

Discussion

This study examined the prevalence of risk factors for CVD in rural and urban men and women living in South India, and their association with three measures of socio-economic status (household possessions score, adult educational status and paternal educational status). We found that cardiovascular risk factors were higher in the urban than in the rural population. All three indicators of socio-economic status were positively related to most of the CVD risk factors, including overweight and obesity, lipid profiles, hypertension and abnormal glucose tolerance. The exception was tobacco use, which tended to decrease with an increase in all three indicators of socio-economic status. The household possessions score (an indicator of wealth and ability to purchase consumer goods) showed stronger positive associations with CVD risk factors than the other SES indicators, and in a concurrent analysis of all three indicators together (Table 6), the household possessions score was the only indicator independently associated with risk factors.

The reasons for stronger associations of CVD risk factors with material wealth than with educational indicators are likely to be multi-factorial and to reflect differences between lower and higher SES groups in beliefs, perceptions about body size, and dietary practices². Greater material and financial status can enable the purchase of healthier food, and access to better quality health care, but it may also be associated with unhealthy lifestyle choices²³. Rapid economic development combined with modernization leads to an increase in the consumption of processed foods, animal fats and a shift to a more sedentary lifestyle²⁴. Other studies have observed these trends to occur first among wealthier subpopulations within developing countries¹. At the other end of the SES spectrum, as shown in our study, inadequate means to buy nutritious foods is associated with a high prevalence of underweight.

In contrast to overweight/obesity, blood pressure and the biochemical risk factors, higher tobacco use was associated with lower socio-economic status. There is strong evidence on social patterning of smoking or tobacco use in developing countries. In India, other studies have shown, like us, that higher levels of education and standard of living are associated, in a graded fashion, with a lower risk of smoking^25,26. Stress could be one explanation for a higher prevalence of tobacco use among people of lower socio-economic status; they may experience greater fear and economic stress. A tendency of individuals of lower socio-economic status to engage in behaviours detrimental to health under stressful environments has been demonstrated elsewhere^27-29. Cultural factors are also important, as indicated by the near zero prevalence of smoking among women. Higher educational status is likely to be associated with better knowledge of the adverse effects of tobacco, and perhaps with greater peer pressure against tobacco use. The independent association between higher paternal educational status and a lower risk of tobacco use may indicate an influence of attitudes acquired during childhood. Paternal educational status was used as marker for childhood SES. Our concurrent analyses of all three SES indicators suggest that the current SES to be an important determinant of CVD risk factors, regardless of childhood economic conditions.

Urban-rural patterns

Our findings indicate a clear rural-urban difference in the prevalence of most risk factors for CVD which persisted even after adjusting for gender and SES indicators. The prevalence of obesity, overweight, unfavourable lipid profile, and tobacco use was higher in the urban population. Urban rural differences in hypertension and glucose intolerance were attenuated after adjusting for SES indicators, suggesting that part of this difference is accounted for by rural-urban SES differences. Several studies have shown differences between urban and rural populations^12,30-34. For example, the prevalence of CVD risk factors differed between rural and urban populations in West Bengal¹². Possible reasons for this higher prevalence of risk factors among the urban population could relate to increasing sedentary lifestyles and westernized dietary preferences associated with urbanization or modernization. Our findings also indicate a similar socio-economic patterning in the prevalence of CVD risk factors in both urban and rural populations. Given the different levels of economic and environmental transition, we expected the association between CVD risk factors and SES to be inverse in the urban population and direct in the rural population. However, the observed similar direction of trends in rural and urban settings suggests that both populations are still at an early stage of transition. Another possible explanation for the greater similarity in socio-economic patterning in urban-rural settings could be due adoption of urbanized life styles among the high socio-economic groups in rural regions without migrating to the urban regions.

Comparison with other studies

Studies of socio-economic patterning in relation to CVD risk factors conducted in developed countries generally show inverse gradients^7,35,36. However, the patterning could vary with the current socio-economic context of the country and needs to be interpreted in that context. Our findings are consistent with a recent study of the socio-economic patterning of CVD risk factors among rural populations selected from four Indian cities (Lucknow, Nagpur, Hyderabad and Bangalore). This study reported a higher prevalence of risk factors among higher SES groups, with the exception of tobacco and alcohol use, which was found to be more common among lower SES groups¹¹. Likewise, a cross-sectional study of risk factors for CVD conducted in highly urban, urban and peri-urban regions of India, found that SES, as measured by level of education was positively associated with risk factors for CVD, but negatively associated with tobacco and alcohol use⁸. Similarly an inverse graded relationship of education with tobacco use, diabetes and hypertension has been observed. However, this trend was not consistent for other risk factors of CVD³⁷. Studies conducted in industrialized populations in Chennai (Southern India) revealed a higher prevalence of CVD risk factors compared to the general population³⁸. In North India, the prevalence of diabetes and hypertension was found to be positively associated with social class³³. In a study of women in five cities of India, social class was found to be directly associated with all risk factors for CVD and under-nutrition was negatively associated with social class³⁴. The prevalence of dyslipidaemia was found to be more common and severe among the middle income group compared to the low income group³⁹.

Double burden of under- and over-nutrition

We have confirmed the existence of a double burden of underweight and obesity in all layers of society. A recent cross-sectional study of a nationally representative sample has also demonstrated this coexistence of undernutrition with obesity, and showed a positive association between SES and obesity and a negative association of underweight with SES⁴⁰. Likewise, a review on SES and obesity in developing countries reported a direct association between SES and weight status and predicted that the burden of obesity will eventually shift to lower SES groups, as economies develop². Our findings are consistent with the previous work, indicating that the possessions score is positively associated with obesity and negatively associated with underweight.

Study limitations

The most important limitation of our analysis is that it is causally uninformative. Given that SES indicators and CVD risk factors were assessed at the same point in time, it is not possible to comment on the temporal relations between these factors. Further, like many other birth cohort studies there was considerable attrition due to migration and mortality and hence our study may lack adequate statistical power to detect associations for risk factors with a lower prevalence. Another limitation relates to the merging or collapsing of categories (smoking and tobacco use) due to small numbers in the specific sub-categories, which could have blurred important differences. Finally, we could not calculate NFHS Standard of Living (SLI), which came into widespread use after our study and has come to be accepted as the best composite SES scoring system for urban and rural populations in India⁴¹. The SLI questionnaire includes questions on education of the head of the household, family type, numbers of children and adults living in the household, type of house, number of rooms in the house and whether there is a separate kitchen, sources of drinking water and light, availability of toilet facilities, ownership of house, land and farm animals, and a list of 17 household possessions. Although there is considerable overlap with the data we collected, we are not able to calculate the exact SLI score for our cohort from the data we collected. Our study sample comprises a population-based sample of adults living in urban and rural regions of Vellore, Tamilnadu. It is not intended to be nationally representative and estimates of the prevalence of CVD risk factors are not generalizable to the whole Indian population. However, there is no a priori reason to believe that our estimates of associations between SES indicator and CVD risk factors are very different from studies based on nationally representative samples. Given these limitations, our study has the advantage of a population based birth cohort which is rare in developing countries like India. It also represents both urban and rural areas, with detailed measurements of various risk factors for CVD and social indicators.

Conclusion

Our study adds to scant existing information on social patterning of CVD risk factors in urban and rural residents from a developing country. We found that obesity and underweight coexist, but remain economically segregated, suggesting that the population is still at an early stage of nutrition transition. Our findings clearly indicate that most risk factors for CVD are associated with greater material wealth in both rural and urban settings. We conclude that public health strategies to control and prevent CVD should consider the rural-urban differentials and the presence of socio economic disparities.

KEY MESSAGES.

• Among the SES indicators, possessions score was most strongly associated with CVD risk factors in this Indian population

• People with higher possessions scores tended to have higher levels of CVD risk factors, with an exception of tobacco use

• The associations between possessions score and CVD risk factors were consistent in both urban and rural populations