Skip to main content

Analysing the socioeconomic determinants of hypertension in South Africa: a structural equation modelling approach



Epidemiological research has long observed a varying prevalence of hypertension across socioeconomic strata. However, patterns of association and underlying causal mechanisms are poorly understood in sub-Saharan Africa. Using education and income as indicators, we investigated the extent to which socioeconomic status is linked to blood pressure in the first wave of the National Income Dynamics Study — a South African longitudinal study of more than 15000 adults – and whether bio-behavioural risk factors mediate the association.


In a cross-sectional analysis, structural equation modelling was employed to estimate the effect of socioeconomic status on systolic and diastolic blood pressure and to assess the role of a set of bio-behavioural risk factors in explaining the observed relationships.


After adjustment for age, race and antihypertensive treatment, higher education and income were independently associated with higher diastolic blood pressure in men. In women higher education predicted lower values of both diastolic and systolic blood pressure while higher income predicted lower systolic blood pressure. In both genders, body mass index was a strong mediator of an adverse indirect effect of socioeconomic status on blood pressure. Together with physical exercise, alcohol use, smoking and resting heart rate, body mass index therefore contributed substantially to mediation of the observed relationships in men. By contrast, in women unmeasured factors played a greater role.


In countries undergoing epidemiological transition, effects of socioeconomic status on blood pressure may vary by gender. In women, factors other than those listed above may have substantial role in mediating the association and merit investigation.

Peer Review reports


Socioeconomic disparities in the prevalence of hypertension have long been observed in high income countries, where sound epidemiological evidence associates higher socioeconomic status (SES) with a lower prevalence of high blood pressure and cardiovascular disease, an association that is consistent across a variety of indicators of social position [1, 2].

By contrast, the pattern of association appears diverse in sub-Saharan Africa (sSA), where a mix of positive and negative gradients has been found across studies, in some distinct by gender [36].

Inconsistencies in SES measurements, sample heterogeneity and different degrees of economic development have been argued as possible explanations of these conflicting results [7]. However, the overall picture is far from complete and a better understanding of the reasons for this heterogeneity is needed in order to inform population based preventive interventions. Such understanding requires going beyond simply describing the association between SES and hypertension to identifying potentially modifiable mediating factors and causal pathways though which socioeconomic factors affect blood pressure.

Modifiable bio-behavioural factors affecting blood pressure levels and risk of hypertension include body mass index (BMI) and other measures of body shape such as waist circumference, as well as resting heart rate, alcohol consumption, exercise, and smoking. A positive association between BMI and blood pressure has been consistently observed in a large number of studies, including in sSA [3, 810]. Similarly, blood pressure tends to increase with alcohol consumption [11], and a positive relationship between resting heart rate and hypertension has been repeatedly found [12]. Conversely, physical activity is associated with lower blood pressure, and the relationship persists after adjustment for the body weight reduction associated with increased activity [13, 14]. Despite the fact that acute effects of smoking result in a transient rise in blood pressure, the evidence of increased risk of hypertension among smokers is scarce [15], and most observational studies show, conversely, that habitual smokers have lower blood pressures than non-smokers [16, 17].

These biological and behavioural factors are often unevenly distributed across socioeconomic strata, making them suitable candidates as mediators of the observed effects of socioeconomic variables on blood pressure [2]. Two recent studies have directly tested this hypothesis and analysed the extent to which the above factors mediate the association between SES and blood pressure [18, 19]. Their findings suggest that variations in BMI/waist circumference, heart rate, smoking and alcohol use account for a sizable proportion of the socioeconomic inequalities in blood pressure levels. However, both studies were carried out in western countries with high level of income per capita, relatively efficient health systems and morbidity profiles largely characterised by chronic diseases, while equivalent studies in developing countries, such as those in sSA, are lacking.

Recently, the National Income Dynamics Study (NIDS) [20] made available good quality anthropometric, sociodemographic and behavioural data for a large sample of the South African population. The analysis of these data offers an opportunity to improve the understanding of the relationships between SES and hypertension in a middle income country undergoing a rapid and complex epidemiological transition [21], and whose morbidity profile comprises coexisting infectious diseases (including a widespread epidemic of HIV/AIDS, the leading cause of worsened mortality between 1990 and 2005), increasing rates of non-communicable diseases and risk factors for cardiovascular disease, persisting child diarrhoea and malnutrition, and interpersonal violence and accidents [22].

The aim of the present study was (1) to assess the independent association of education and income, as SES indicators, with blood pressure in the adult population of South Africa, and (2) to examine the extent to which differences in body mass, resting heart rate, smoking and alcohol use explain these relationships.



This study analyses the adults subsample (15574 subjects 15 years and over, out of an estimated South African adult population of 34 million) of the first wave of the NIDS. The NIDS survey method is described in detail by Woolard et al. [20]. It was designed as a longitudinal panel survey of a nationally representative sample of households in South Africa. A two-stage cluster sample design was used to randomly select about 7300 households across 400 primary sampling units (areas), stratified by district council (a second level administrative division of South Africa’s territory in 53 areas). The first wave of the survey was conducted in 2008, and the target population was private households and residents in workers’ hostels, convents and monasteries, excluding other collective living quarters such as old age homes, hospitals, prisons and boarding schools. Trained fieldworkers were instructed to interview and collect anthropometric data on all available subjects belonging to the selected households. The household level response rate was 69% and the individual response rate within households was 93%. The NIDS study, the dataset of which is publicly available for research purposes [23], has been granted ethical approval by the Commerce Faculty Ethics Committee at the University of Cape Town.


Sociodemographic variables

Age in years was treated as a continuous variable, and race self-defined by participants according to the historical “population group” categorization used in South Africaa. Education was measured in years of completed schooling, and individual monthly income was calculated as the summation of a wide array of sources, which is considered a more reliable method than the use of single questions [24]. Missing data in specific sources of income were imputed according to the procedure described by Argent [25].

Blood pressure and resting heart rate

Supine blood pressure and heart rate were measured twice by trained field workers in the left arm after a 5 minute rest period, using an automated blood pressure monitor (Omron M7 BP, multi-size cuff, factory calibrated). Measurements were retained if systolic blood pressure (SBP) was between 80 and 240 mm Hg, diastolic blood pressure (DBP) ≥ 30 mm Hg, and their difference was ≥ 15 mm Hg. Heart rate measurements were retained if ≥ 30 bpm and < 150 bpm.

Antihypertensive medication

Use of antihypertensives was assessed by asking subjects if they were currently taking medication for high blood pressure.

Bio-behavioural risk factors

BMI, smoking, alcohol use, physical exercise and resting heart rate were considered as possible mediators of the association between SES and blood pressure.

Duplicate measures of weight and height were recorded, with a third measure taken if their difference was greater than 0.5 Kg and 0.5 cm, respectively. Excluding measures with implausible values (height < 100 cm or > 200 cm, weight < 20 Kg or > 200 Kg), the average of the available readings was used to calculate BMI. Current smoking status, alcohol use and physical exercise were represented by ordinal variables, as shown in Table 1. Measurement procedures are detailed in the fieldwork manual of the NIDS [26].

Table 1 Descriptive statistics for the adult sub-sample of the National Income Dynamics Study

Statistical analysis

Sample characteristics were described as median and interquartile range for continuous variables and frequency for categorical measures.

Based on epidemiological and biological evidence [13, 2731], we hypothesised that the variables considered in this study were causally linked as depicted in Figure 1. The model suggests a causal effect of education and income on SBP and DBP, partially mediated by BMI, heart rate, exercise frequency, alcohol use and smoking. It also assumes that BMI is affected by alcohol use, smoking and exercise, and that resting heart rate is influenced by smoking and exercise frequency.

Figure 1
figure 1

Hypothesised causal pathways between education, income and blood pressure. Squares and circles represent observed and latent variables, respectively. Arrows indicate hypothesised causal effects. Dashed squares indicate each of the multiple readings from which the values of the latent variables systolic blood pressure (SBP), diastolic blood pressure (DBP) and resting heart rate (HR) are inferred. Race, age and use of antihypertensive medication are omitted from the diagram, but taken into account as possible confounders in the model.

Structural equation modelling was used to evaluate the extent to which the hypothesised causal structure was able to explain the observed associations between variables, and to estimate the magnitude of the postulated effects. To minimise the bias due to measurement error, blood pressure and heart rate were introduced as latent variables, with the observed multiple readings as indicators [32].

Estimated model coefficients were used to decompose total effects of SES on blood pressure (i.e. the change in blood pressure per unit increase in each of the SES indicators) into mediated and unexplained effects. Mediated effects (i.e. effects statistically explained by variations in BMI, smoking, alcohol use, physical exercise and heart rate) are represented in Figure 1 by indirect paths connecting SES indicators to blood pressure levels through the different factors. Unexplained effects (i.e. effects unrelated to variation in the considered mediators) are represented by direct paths connecting SES indicators to blood pressure.

Model coefficients were estimated adjusting for age, race and hypertensive medication. In view of previous evidence that relationships might differ by gender, models were fitted separately for women and men.

To relax the assumption of multivariate normality underlying the estimation of standard errors with the usual formulae – rarely satisfied in multiple mediators models – 95% confidence intervals for mediated and unexplained effects were bootstrapped, and effects considered statistically significant at probability level α=0.05 when the confidence interval excluded the null value.

Analyses were carried out using Stata®; 12 and Mplus®; 6 [33, 34], taking into account the complex sampling scheme of the NIDS.

Further details on modelling assumption and estimation procedure are reported in Additional file 1.


Unweighted sample characteristics are described in Table 1. The great majority of participants were Black and Coloured. Whites were under-represented relative to the South African population, owing to their low response rate in the NIDS [20]. Using SBP ≥ 140 mm Hg and/or DBP ≥ 90 mm Hg as cut-offs [35], 28.4% of male participants and 30.6% of female participants would be classified as hypertensive. Current use of antihypertensives was reported by 7.4% of men and 16% of women.

Table 2 shows the estimated average blood pressure and hypertension prevalence in the South African adult population. The comparison of these estimates with those from the 1998 South African Demographic and Health Survey [36] (not shown), suggests that in the last 10 years the prevalence of hypertension among South African adults has increased considerably (proportionately by 22% in men and 28% in women) [37].

Table 2 Average blood pressure and prevalence of hypertension in the South African adult population

Association of education and income with blood pressure

The structural models showed an excellent fit with the data (see Table 3), supporting our hypothesis that the causal structure in Figure 1 is a plausible explanation of the observed associations between variables. The estimated values of the model coefficients (and corresponding confidence intervals) are listed in the additional material and qualitatively summarised in Table 4.

Table 3 Fit indices for the structural models
Table 4 Sign and statistical significance of the estimated path coefficients for the model in Figure 1

The estimated total effects indicated that, among women, each year of education was associated with 0.29 mm Hg drop in SBP and 0.12 mm Hg drop in DBP. A doubling of the monthly income was similarly related to a decrease of SBP by 0.15 mm Hg, while the association of income with DBP was trivial in magnitude and not statistically significant.

In men, by contrast, an increase in both education and income was associated with an increase in blood pressure levels, but only the relationships with DBP were statistically significant, indicating an increase of 0.11 mm Hg per year of education and 0.12 mm Hg for each income doubling.

Overall then, the findings show an inconsistent relationships between SES indicators and blood pressure across gender. Education and income have positive (or null) associations with blood pressure levels among men (harmful effect of increased SES on blood pressure), but inverse (or null) associations among women (protective effect). The relative size of the coefficients and the width of their confidence intervals suggest also that SES is more strongly associated with DBP than SBP in men, while the opposite holds for women.


Figures 2 and 3 summarize the results of the mediation analysis.

Figure 2
figure 2

Mediated, unexplained and total effects of education on blood pressure, and statistically significant specific pathways. Values represent the average increase in blood pressure (in mm Hg) per year of education.

Figure 3
figure 3

Mediated, unexplained and total effects of (log) income on blood pressure, and statistically significant specific pathways. Values represent the average increase in blood pressure (in mm Hg) when the income doubles.

Among men, BMI was, by far, the strongest mediator of the harmful effect of increasing SES on blood pressure. All statistically significant indirect paths included BMI, with the only exception being the one connecting education to DBP though exercise and resting heart rate. Overall, the sum of the effects mediated by BMI in men accounted for a 0.042 mm Hg increase in SBP per year of education (95% CI: 0.014 to 0.084) and a 0.019 mm Hg increase per income doubling (95% CI: 0.005 to 0.044). The corresponding values for DBP were 0.035 (95% CI: 0.014 to 0.063) and 0.015 (95% CI: 0.005 to 0.034).

Smoking, exercise frequency and heart rate were also involved in significant relationships in men, some of them (involving exercise and smoking through BMI and HR) representing protective effects, opposite to the total effect. However, their overall role was modest compared to that of BMI. The only statistically significant direct path (representative of an unexplained harmful effect) connected income to DBP.

As among men, BMI in women mediated statistically significant harmful effects, despite the overall protective role of higher SES on blood pressure. Effects mediated by BMI accounted for a 0.028 mm Hg increase in SBP per year of education (95% CI: 0.014 to 0.056) and a 0.019 mm Hg per income doubling (95% CI: 0.010 to 0.034) as well as a 0.028 mm Hg increase in DBP per year of education (95% CI: 0.014 to 0.056), and a 0.019 mm Hg per income doubling (95% CI: 0.005 to 0.034).

Independently of its effect on BMI, alcohol use in women accounted for a 0.024 mm Hg increase in DBP per income doubling (95% CI: 0.010 to 0.053). The mediating role of the remaining risk factors was similar to that of men, both in magnitude and direction. Except for the effect of income on DBP, all direct paths in women were statistically significant and accounted for a sizable share of the overall association between SES and blood pressure, which remained therefore largely unexplained by the mediators hypothesised in our model.


Relationships between SES and blood pressure with gender specific patterns consistent with our results (i.e. a protective effect of SES among women and a harmful effect among men) have been previously found in South Africa [5] and other middle-income countries [3840]. These findings are in partial contrast with those in high income countries, where an inverse gradient SESblood pressure is commonly found in both genders, even though often stronger and more consistent across SES indicators in women than in men [2]. In our study, total mediated effects (i.e. the sum of the effects through all indirect pathways) were similar in direction and magnitude across genders, and therefore the observed discrepancies cannot be explained by gender differences in the distribution of the hypothesised mediators. It is conversely the presence – in women but not in men – of sizable unexplained protective effects outmatching the overall mediated effects (harmful in both genders) which makes the difference and suggests that the inverse effect of SES on blood pressure observed in women is mediated mainly by factors not included in our analysis.

Increased awareness of hypertension, accessibility of and adherence to medical treatment, less chronic stress, and, recently, more favourable neighbourhood characteristics, have been indicated in the literature as possible mediators of a protective effect of higher SES on raised blood pressure, and may contribute to the share of effect among women which is unexplained in our model. The reason why a protective effect of the same magnitude is not observed in men might be related to the lower levels of awareness and control consistently observed in men than in women, and to the lower sensitivity to the adverse effect of unfavourable neighbourhood characteristics on blood pressure that recent studies suggest [2, 18, 4144].

Salt intake, whose causal relationship with blood pressure levels is supported by results of experimental studies in South Africa [41], has also been proposed as a possible mediator [2]. Nevertheless, results from two large scale surveys of the South African population suggests a positive association between SES and salt intake, making this factor a candidate mediator for the unexplained portion of the harmful effect of income on DBP in men, but not for the protective effects of SES in women [42, 45].

Beyond differences in relative magnitude and statistical significance, the overall patterns of association of physical exercise, alcohol use, smoking and heart rate with both SES and blood pressure are consistent with those found in high income countries [18, 19]. However, in contrast to those studies but coherently with other studies in sSA [6, 46], in our population BMI rises with increasing SES. It therefore mediates a harmful effect of increasing SES on blood pressure, accounting for a sizable proportion of the association in men, and contributing to reducing the overall protective effect found in women. These contrasting results may be partly explained by considering that the distribution of income in South Africa is extremely unequal, and, despite the mean income per capita being relatively high, a substantial proportion of the population lives near or below the poverty line [47]. It is likely that among people in this setting the increased knowledge of health risk and greater motivation to control weight associated with increasing SES – which have been argued as an explanation of the inverse SES/BMI relationship in high income countries – play a less significant role than the greater access to energy dense processed food among those with higher SES.

A separate analysis for the 5% of the total sample with the highest income offers some support for this hypothesis. In that subsample the associations between SES indicators and BMI become inverse also in men, albeit not statistically significant owing to the small sample size (see Additional file 1).

Finally, we found that socioeconomic variables affect DBP more strongly than SBP among men and vice versa among women. This different responsiveness may explain some incongruences between results of studies using only systolic blood pressure as the outcome variable and studies analysing hypertension prevalence (defined in terms of both SBP and DBP). This heterogeneity is not accounted for by the modest differences in the association between mediators and SBP or DBP, and calls for consideration of other variables. Among those, stress and dietary patterns (e.g. vegetable consumption) have been shown to be selectively associated with SBP and DBP, thus representing suitable targets for further mediation studies [48, 49].

Strengths of the present study include the use of a large sample and an analytical approach allowing for simultaneous testing of multiple mediation pathways (avoiding the potential bias arising from neglecting the correlation between mediators) and for the explicit consideration of measurement error in physiological variables. Moreover, this study is the first, to our knowledge, to perform mediation analysis in a large sample modelling simultaneously for both SBP and DBP.

The major limitations of this study are the intrinsic lack of temporal information in our cross-sectional dataset – which limits the interpretation of the temporal sequence of the relationships – and the possibility that important unmeasured confounding variables (e.g. undernutrition in infancy which could be associated with income and is a known risk factor for high blood pressure later in life) [50] have introduced bias into the study results.

Low reliability of self-report data, including those on physical exercise, alcohol and tobacco use is a well known problem in population-based surveys, which usually results in observed associations biased towards the null [51]. More precise measurements are therefore likely to strengthen the result of our analysis rather than invalidate them.

Other than for age – a strong predictor of blood pressure which is associated with many of the variables in our model – our analyses were adjusted for racial group and antihypertensive medication. Despite the fact that statistical control for race and medication is common in the literature (see for example [6, 18, 19]), we cannot exclude the possibility that these variables act instead as effect modifier and mediator respectively.

Besides income and education, racial group assignment may indirectly capture differences in household wealth, genetic ancestry, social stress (e.g. migration, discrimination) and dietary intake, possible confounders of the association between income and blood pressure not otherwise captured in our analyses [5257]. These considerations justify its introduction as a confounder in the models. However, the legacy and persistence of separate educational systems for the different racially defined groups [58] and the likely differential economic value of a given educational level [59] make it plausible that the value of education as a measure of SES differs by racial group. Race would thus be an effect modifier, as well as a confounder, of the hypothesised causal relationship between years of schooling and blood pressure [1].

Similarly, we cannot rule out that being on medication for high blood pressure (which is positively correlated with income and education in our sample, after adjustment for age and gender) lies on a causal pathway between SES and blood pressure (higher SES → increased access to health care → increased use of antihypertensive drugs → lower blood pressure). In this case, adjustment for a mediator would bias the observed association between SES and blood pressure towards the null.

However, restricting our analyses to the Black subsample or omitting adjustment for antihypertensive medication did not produce appreciable changes in the overall pattern of association between variables. All model coefficients maintained the same sign with negligible to moderate changes in magnitude (see Additional file 1), thus supporting the robustness of the results of this study to incorrect specification of the role of these variables.

Finally, the specific characteristics of the highly uneven socio-economic development of South Africa make it necessary to exercise a degree of caution in generalizing the results of this study to other areas of sSA or other developing countries. The findings need replication in other settings with rapid but complex epidemiologic transitions.


The findings of this study strengthen the case that SES is associated with hypertension and that physical exercise, alcohol use, smoking, resting heart rate and BMI play a role in explaining socioeconomic inequalities in blood pressure. However, in contrast to most studies in high income populations, our results suggest that in settings such as South Africa, characterised by rapid and complex epidemiological transition (1) effects of SES on blood pressure may vary by gender; and (2) factors other than those listed above may have a greater role in mediating the association in women.

The worsening in blood pressure with upward mobility observed in men can be viewed as a significant public health cost of socio-economic development, while understanding the causes of the opposite effect in women may inform action to reduce the growing cardiovascular health burden in developing countries.


a Under apartheid, South Africans were categorised into one of four socially defined groups: Asian (or Indian), Black (or African), Coloured (wide grouping of people of mixed ancestry) and White (or European). Race in this sense is closely and enduringly correlated with socioeconomic status in South Africa.


  1. Kaplan GA, Keil JE: Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation. 1993, 88 (4 Pt 1): 1973-1998.

    CAS  Article  PubMed  Google Scholar 

  2. Colhoun HM, Hemingway H, Poulter NR: Socio-economic status and blood pressure: an overview analysis. J Hum Hypertens. 1998, 12 (2): 91-110. 10.1038/sj.jhh.1000558.

    CAS  Article  PubMed  Google Scholar 

  3. Poulter NR, Khaw K, Hopwood BE, Mugambi M, Peart WS, Sever PS: Determinants of blood pressure changes due to urbanization: a longitudinal study. J Hypertens. 1985, 3 (3): S375-S377.

    CAS  Google Scholar 

  4. Lang T, Pariente P, Salem G, Tap D: Social, professional conditions and arterial hypertension: an epidemiological study in Dakar, Senegal. J Hypertens. 1988, 6 (4): 271-276.

    CAS  Article  PubMed  Google Scholar 

  5. Norman R, Bradshaw D, Steyn K: Chronic diseases, risk factors and lifestyle based on the, South Africa Adult Demographic and Health Survey. Poverty and Chronic Diseases in South Africa. Technical Report 2001. Edited by: Bradshaw D, Steyn K. 2001, Tygerberg: Medical Research Council of South Africa - Burden of Disease Research Unit, 53:103-53:103.

    Google Scholar 

  6. Ploubidis GB, Mathenge W, De Stavola B, Grundy E, Foster A, Kuper H: Socioeconomic position and later life prevalence of hypertension, diabetes and visual impairment in Nakuru, Kenya. Int J Public Health. 2012, 58: 133-141.

    Article  PubMed  Google Scholar 

  7. Fernald LCH, Adler NE: Blood pressure and socioeconomic status in low-income women in Mexico: a reverse gradient?. J Epidemiol Commun H. 2008, 62 (5): e8-10.1136/jech.2007.065219.

    CAS  Article  Google Scholar 

  8. Mollentze WF, Moore AJ, Steyn AF, Joubert G, Steyn K, Oosthuizen GM, Weich DJ: Coronary heart disease risk factors in a rural and urban Orange Free State black population. S Afr Med J. 1995, 85 (2): 90-96.

    CAS  PubMed  Google Scholar 

  9. Steyn K, Fourie J, Lombard C, Katzenellenbogen J, Bourne L, Jooste P: Hypertension in the black community of the Cape Peninsula, South Africa. E Afr Med J. 1996, 73 (11): 758-763.

    CAS  Google Scholar 

  10. Van Rooyen JM, Kruger HS, Huisman HW, Wissing MP, Margetts BM, Venter CS, Vorster HH: An epidemiological study of hypertension and its determinants in a population in transition: the THUSA study. J Hum Hypertens. 2000, 14 (12): 779-787. 10.1038/sj.jhh.1001098.

    CAS  Article  PubMed  Google Scholar 

  11. Xin X, He J, Frontini MG, Ogden LG, Motsamai OI, Whelton PK: Effects of alcohol reduction on blood pressure. Hypertension. 2001, 38 (5): 1112-1117. 10.1161/hy1101.093424.

    CAS  Article  PubMed  Google Scholar 

  12. Palatini P: Heart rate and the cardiometabolic risk. Curr Hypertens Rep. 2013, 15 (3): 253-259. 10.1007/s11906-013-0342-7.

    CAS  Article  PubMed  Google Scholar 

  13. Appel LJ: Lifestyle modification as a means to prevent and treat high blood pressure. J Am Soc Nephrol. 2003, 14 (suppl 2): S99-S102.

    Article  PubMed  Google Scholar 

  14. de Ramirez SS, Enquobahrie DA, Nyadzi G, Mjungu D, Magombo F, Ramirez M, Sachs SE, Willett W: Prevalence and correlates of hypertension: a cross-sectional study among rural populations in sub-Saharan Africa. J Hum Hypertens. 2010, 24 (12): 786-95. 10.1038/jhh.2010.14.

    Article  PubMed  Google Scholar 

  15. Narkiewicz K, Maraglino G, Biasion T, Rossi G, Sanzuol F, Palatini P: Interactive effect of cigarettes and coffee on daytime systolic blood pressure in patients with mild essential hypertension. HARVEST Study Group (Italy). Hypertension Ambulatory Recording VEnetia STudy. J Hypertens. 1995, 13 (9): 965-970. 10.1097/00004872-199509000-00005.

    CAS  Article  PubMed  Google Scholar 

  16. Green MS, Jucha E, Luz Y: Blood pressure in smokers and nonsmokers: epidemiologic findings. Am Heart J. 1986, 111 (5): 932-940. 10.1016/0002-8703(86)90645-9.

    CAS  Article  PubMed  Google Scholar 

  17. Mikkelsen KL, Wiinberg N, Høegholm A, Christensen HR, Bang LE, Nielsen PE, Svendsen TL, Kampmann JP, Madsen NH, Bentzon MW: Smoking related to 24-h ambulatory blood pressure and heart rate: a study in 352 normotensive Danish subjects. Am J Hypertens. 1997, 10 (5 Pt 1): 483-491.

    CAS  Article  PubMed  Google Scholar 

  18. Chaix B, Bean K, Leal C, Thomas F, Havard S, Evans D, Jégo B, Pannier B: Individual/neighborhood social factors and blood pressure in the RECORD Cohort Study: which risk factors explain the associations?. Hypertension. 2010, 55 (3): 769-775. 10.1161/HYPERTENSIONAHA.109.143206.

    CAS  Article  PubMed  Google Scholar 

  19. Brummett BH, Babyak MA, Siegler IC, Shanahan M, Harris KM, Elder GH, Williams RB: Systolic blood pressure, socioeconomic status, and biobehavioral risk factors in a nationally representative US young adult sample. Hypertension. 2011, 58 (2): 161-166. 10.1161/HYPERTENSIONAHA.111.171272.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  20. Woolard I, Leibbrandt M, De Villiers L: The South African national income dynamics study: design and methodological issues. J Stud Econ Econometrics. 2010, 34 (3): 7-24.

    Google Scholar 

  21. Omran ARD: The epidemiologic transition: a theory of the epidemiology of population change. 1971. Milbank Meml Fund Q. 1971, 41 (4): 509-538.

    Article  Google Scholar 

  22. Kahn K: Population health in South Africa: dynamics over the past two decades. J Public Health Pol. 2011, 32 (S1): S30-S36.

    Article  Google Scholar 

  23. Southern Africa Labour and Development Research Unit: National income dynamics study 2008, wave 1 [dataset]. Version 4.1. Cape Town. 2012, [],

    Google Scholar 

  24. Visagie J, Posel D: A Reconsideration of What and Who is Middle Class in South Africa. Working paper 249. 2011, Durban: University of KwaZulu-Natal

    Google Scholar 

  25. Argent J: Household Income: Report on NIDS Wave 3. Technical Paper no. 3. 2009, Cape Town: Southern Africa Labour and Development Research Unit

    Google Scholar 

  26. Southern Africa Labour and Development Research Unit: NIDS Fieldwork Manual - Wave 1. 2008, Cape Town: Southern Africa Labour and Development Research Unit - Development Research Africa

    Google Scholar 

  27. Perkins KA, Epstein LH, Marks BL, Stiller RL, Jacob RG: The effect of nicotine on energy expenditure during light physical activity. New Engl J Med. 1989, 320 (14): 898-903. 10.1056/NEJM198904063201404.

    CAS  Article  PubMed  Google Scholar 

  28. Zhang J, Kesteloot H: Anthropometric, lifestyle and metabolic determinants of resting heart rate. A population study. Eur Heart J. 1999, 20 (2): 103-110. 10.1053/euhj.1999.1230.

    CAS  Article  PubMed  Google Scholar 

  29. Mahanonda N, Bhuripanyo K, Leowattana W, Kangkagate C, Chotinaiwattarakul C, Panyarachun S, Sukrungreang C: Regular exercise and cardiovascular risk factors. J Med Assoc Thailand. 2000, 83 Suppl 2: S153-158.

    CAS  Google Scholar 

  30. Adler NE, Newman K: Socioeconomic disparities in health: pathways and policies. Health Aff. 2002, 21 (2): 60-76. 10.1377/hlthaff.21.2.60.

    Article  Google Scholar 

  31. Breslow RA, Smothers BA: Drinking patterns and body mass index in never smokers: national health interview survey, 1997–2001. Am J Epidemiol. 2005, 161 (4): 368-376. 10.1093/aje/kwi061.

    Article  PubMed  Google Scholar 

  32. Ullman JB: Structural equation modeling: reviewing the basics and moving forward. J Pers Assess. 2006, 87: 35-50. 10.1207/s15327752jpa8701_03.

    Article  PubMed  Google Scholar 

  33. StataCorp: Stata Statistical Software: Release 12. 2011, College Station: StataCorp LP

    Google Scholar 

  34. Muthén L, Muthén B: Mplus User’s Guide, 6th edition. 2011, Los Angeles: Muthén & Muthén

    Google Scholar 

  35. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL, Jones DW, Materson BJ, Oparil S, Wright JT, Roccella EJ: Seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure. Hypertension. 2003, 42 (6): 1206-1252. 10.1161/01.HYP.0000107251.49515.c2.

    CAS  Article  PubMed  Google Scholar 

  36. Steyn K, Gaziano TA, Bradshaw D, Laubscher R, Fourie J: Hypertension in South African adults: results from the demographic and health survey, 1998. J Hypertens. 2001, 19 (10): 1717-1725. 10.1097/00004872-200110000-00004.

    CAS  Article  PubMed  Google Scholar 

  37. Ardington C, Case A: Health: Analysis of the NIDS Wave 1 Dataset. Discussion Paper no. 2. 2009, Cape Town: Southern Africa Labour and Development Research Unit

    Google Scholar 

  38. Fleischer NL, Diez Roux AV, Alazraqui M, Spinelli H, De Maio F: Socioeconomic gradients in chronic disease risk factors in middle-income countries: evidence of effect modification by urbanicity in Argentina. Am J Public Health. 2011, 101 (2): 294-301. 10.2105/AJPH.2009.190165.

    Article  PubMed  PubMed Central  Google Scholar 

  39. Dressler WW, Grell GA, Gallagher PN, Viteri FE: Blood pressure and social class in a Jamaican community. Am J Public Health. 1988, 78 (6): 714-716. 10.2105/AJPH.78.6.714.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  40. Hutchinson J: Association between stress and blood pressure variation in a Caribbean population. Am J Phys Anthropol. 1986, 71: 69-79. 10.1002/ajpa.1330710109.

    CAS  Article  PubMed  Google Scholar 

  41. Charlton KE, Steyn K, Levitt NS, Peer N, Jonathan D, Gogela T, Rossouw K, Gwebushe N, Lombard CJ: A food-based dietary strategy lowers blood pressure in a low socio-economic setting: a randomised study in South Africa. Public Health Nutr. 2008, 11 (12): 1397-1406. 10.1017/S136898000800342X.

    Article  PubMed  Google Scholar 

  42. Department of Health Medical Research Council: South Africa Demographic and Health Survey 2003. 2003, Pretoria: Department of Health

  43. Kayima J, Wanyenze RK, Katamba A, Leontsini E, Nuwaha F: Hypertension awareness, treatment and control in Africa: a systematic review. BMC Cardiovasc Disord. 2013, 13: 54-10.1186/1471-2261-13-54.

    Article  PubMed  PubMed Central  Google Scholar 

  44. Matheson FI, White HL, Moineddin R, Dunn JR, Glazier RH: Neighbourhood chronic stress and gender inequalities in hypertension among Canadian adults: a multilevel analysis. J Epidemiol Commun H. 2010, 64 (8): 705-713. 10.1136/jech.2008.083303.

    Article  Google Scholar 

  45. Department of Health - Medical Research Council: South Africa Demographic and Health Survey 1998. Full report. 2002, Pretoria: Department of Health

    Google Scholar 

  46. Sodjinou R, Agueh V, Fayomi B, Delisle H: Obesity and cardio-metabolic risk factors in urban adults of Benin: relationship with socio-economic status, urbanisation, and lifestyle patterns. BMC Public Health. 2008, 8: 84-10.1186/1471-2458-8-84.

    Article  PubMed  PubMed Central  Google Scholar 

  47. Leibbrandt M, Woolard I, Finn A, Argent J: Trends in South African Income Distribution and Poverty since the Fall of Apartheid. OECD Social, Employment and Migration Working Papers, No. 101. 2010, Paris: OECD Publishing

    Book  Google Scholar 

  48. Wilson BL, Albright GL, Steiner SS, Andreassi JL: Cardiodynamic response to psychological and cold pressor stress: further evidence for stimulus response specificity and directional fractionation. Biofeedback Self-reg. 1991, 16: 45-53. 10.1007/BF01000445.

    CAS  Article  Google Scholar 

  49. Masala G, Bendinelli B, Versari D, Saieva C, Ceroti M, Santagiuliana F, Caini S, Salvini S, Sera F, Taddei S, Ghiadoni L, Palli D: Anthropometric and dietary determinants of blood pressure in over 7000 mediterranean women: the European prospective investigation into cancer and nutrition-Florence cohort. J Hypertens. 2008, 26 (11): 2112-2020. 10.1097/HJH.0b013e32830ef75c.

    CAS  Article  PubMed  Google Scholar 

  50. Sawaya AL, Sesso R, Florêncio de Menezes Toledo TM, Fernandes MTB, Martins PA: Association between chronic undernutrition and hypertension. Matern Child Nutr. 2005, 1 (3): 155-163. 10.1111/j.1740-8709.2005.00033.x.

    Article  PubMed  Google Scholar 

  51. Whitlock G, Clark T, Vander Hoorn S, Rodgers A, Jackson R, Norton R, MacMahon S: Random errors in the measurement of 10 cardiovascular risk factors. Eur J Epidemiol. 2001, 17 (10): 907-909. 10.1023/A:1016228410194.

    CAS  Article  PubMed  Google Scholar 

  52. Minor DS, Wofford MR, Jones DW: Racial and ethnic differences in hypertension. Curr Atheroscler Rep. 2008, 10 (2): 121-127. 10.1007/s11883-008-0018-y.

    Article  PubMed  Google Scholar 

  53. Brondolo E, Love EE, Pencille M, Schoenthaler A, Ogedegbe G: Racism and hypertension: a review of the empirical evidence and implications for clinical practice. Am J Hypertens. 2011, 24 (5): 518-529. 10.1038/ajh.2011.9.

    Article  PubMed  Google Scholar 

  54. Rosenthal T: The effect of migration on hypertension and other cardiovascular risk factors: a review. J Am Soc Hypertens. 2014,

    Google Scholar 

  55. Cuffee YL, Hargraves JL, Allison J: Exploring the association between reported discrimination and hypertension among African Americans: a systematic review. Ethnic Dis. 2012, 22 (4): 422-431.

    Google Scholar 

  56. Kato N: Ethnic differences in genetic predisposition to hypertension. Hypertens Res. 2012, 35 (6): 574-581. 10.1038/hr.2012.44.

    Article  PubMed  Google Scholar 

  57. Non AL, Gravlee CC, Mulligan CJ: Education, genetic ancestry, and blood pressure in African Americans and Whites. Am J Public Health. 2012, 102 (8): 1559-1565. 10.2105/AJPH.2011.300448.

    Article  PubMed  PubMed Central  Google Scholar 

  58. Thomas D: Education across generations in South Africa. Am Econ Rev. 1996, 86 (2): 330-334.

    Google Scholar 

  59. Kaufman JS, Cooper RS, McGee DL: Socioeconomic status and health in blacks and whites: the problem of residual confounding and the resiliency of race. Epidemiology. 1997, 8 (6): 621-628.

    CAS  PubMed  Google Scholar 

Pre-publication history

Download references


Dr Cally Ardington of the Southern Africa Labour & Development Research Unit provided helpful advice on the use of the NIDS dataset.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Annibale Cois.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

AC performed the analysis and drafted the manuscript. RE conceived the study and contributed to the drafting of the manuscript. Both authors read and approved the final manuscript.

Electronic supplementary material


Additional file 1: Statistical analyses and further results. Methodological details on statistical analyses. Further results. (PDF 262 KB)

Authors’ original submitted files for images

Below are the links to the authors’ original submitted files for images.

Authors’ original file for figure 1

Authors’ original file for figure 2

Authors’ original file for figure 3

Rights and permissions

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Cois, A., Ehrlich, R. Analysing the socioeconomic determinants of hypertension in South Africa: a structural equation modelling approach. BMC Public Health 14, 414 (2014).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


  • Systolic blood pressure
  • Diastolic blood pressure
  • Hypertension
  • Body mass index
  • Socioeconomic status
  • Sub-Saharan Africa
  • Structural equation modelling