Abstract
Objectives
We examined the extent to which the association between socioeconomic position (SEP) and later life prevalence of hypertension, diabetes and visual impairment in Nakuru, Kenya is mediated by health-related behaviour.
Methods
We used data from a community survey of 4,314 participants sampled from urban and rural areas in Nakuru, Kenya. Structural equation modelling was employed to estimate the direct and indirect—via health-related behaviour—effects of SEP on the three health outcomes.
Results
The accumulation of material resources was positively associated with hypertension and diabetes, whereas both education and material resources had a negative association with the prevalence of visual impairment. However, the observed health inequalities were not due to variation between SEP groups in health-related behaviour.
Conclusions
The pattern of associations between education, material resources and the three health outcomes varied, suggesting that in Kenya, unlike the observed pattern of inequalities in high income countries, different dimensions of SEP provide different aspects of protection as well as risk. Smoking and alcohol use did not appear to mediate the observed associations, in contrast with countries past the epidemiologic transition.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Introduction
Extensive research has shown that socioeconomic and socio-demographic factors are associated with health outcomes and the importance of these associations for public policy is widely recognised internationally (Mackenbach et al. 2008; Marmot et al. 1991, 2002). However, only recently has the issue of health inequalities for chronic diseases started receiving attention in Africa (Sodjinou et al. 2008; Addo et al. 2009; Bovet et al. 2002), with efforts to reduce inequality and improve health facing many challenges. Among others, these include the high incidence of infectious diseases, the growing burden of chronic diseases such as coronary heart disease and stroke, which are emerging as new leading causes of death in older people (Lopez et al. 2006), inadequate health systems, and limited human and material resources. Furthermore, little is known of the factors that determine health inequalities and the mechanisms through which they operate.
In high income countries past the epidemiologic transition (Omran 1971), theories of explanation of the association between socioeconomic position (SEP) and health essentially focus on three mechanisms (Smith et al. 1994; Ploubidis et al. 2011): The first is a neo-materialist one; suggesting that those with higher incomes are able to purchase better food, better housing, live in safer environments and have better access to health care (Kaplan et al. 1996; Lynch et al. 2000). The second emphasises behavioural or “lifestyle” factors, such as smoking, diet, alcohol consumption and appropriate use of health care, which may also vary with cognitive skill and access to information (Schrijvers et al. 1999). The third places more emphasis on psychosocial factors such as empowerment, relative social status and social integration, including exposure to stressful events that may result from low status and low autonomy in important arenas of life, such as work (Wilkinson 1997).
Despite the fact that previous studies have reported an association between SEP and various health outcomes (Addo et al. 2009; Bovet et al. 2002), the extent and pattern of health inequalities are relatively unknown and may be different to those observed in high-income countries. Furthermore, the explanatory framework that underlies health inequalities, which has been offered in countries past the epidemiologic transition (Omran 1971), has not yet been tested in Africa. The major aims of this study therefore are: (1) to investigate the associations between education and material resources with the prevalence of hypertension, diabetes and visual impairment in the older population of Nakuru, Kenya; (2) enhance the understanding of the underlying mechanism(s) of health inequalities in the older population of Kenya, by testing the importance of the health-related behaviour pathway in the association between education and material resources with three distinct health outcomes (see Fig. 1).
Methods
Population and sample recruitment
Nakuru District has a population of 1.6 million, one-third of which lives in urban areas, and it is broadly representative of Kenya in terms of ethnic diversity and economic position. During January 2007 to December 2008, a sample of 100 clusters of 50 people aged ≥50 years were selected across Nakuru District through probability proportionate to size sampling, using the electoral roll as the sampling frame. Clusters were classed as “rural” or “urban” using the classification of the district statistical office, which considered population size and access to social amenities in its definition. Households were selected within clusters through modified compact segment sampling (Turner et al. 1996). Further details of the survey are available elsewhere (Mathenge et al. 2010). A total of 5,010 eligible subjects were selected, of whom 4,396 (88 %) were examined. Our analytic sample included 4,314 participants (86 % response) and excluded participants with missing data since they constituted less than 2 % of our data. Ethical approval was granted by the London School of Hygiene and Tropical Medicine Ethics Committee, the Kenya Medical Research Institute Ethical Committee and Nakuru District Health Management Team. Informed consent was obtained from all participants.
Measures
Participants were interviewed by trained nurses. Information was collected on smoking, alcohol use demographic data, education and assets (building materials of the house—type of walls, roof, floor and toilet; ownership of household assets—radio, TV, fridge, phone, cupboard, sofa set, sewing machine, table, bicycle and vehicle; animal ownership—cows, sheep/goats). A nurse recorded the blood pressure of participants three times on the right arm of the participant, at least 5 min apart using the Omron digital automatic blood pressure monitor (model HEM907) after an initial period of 5 min of rest. The average of the last two readings was used as the measures of systolic and diastolic blood pressure (to nearest 1 mmHg). If there were large differences between the two readings (more than 10 mmHg systolic and 5 mm diastolic) or if both readings were extremely high, a fourth reading was taken and this was recorded as the study BP. A random finger-prick blood sample was taken to measure glucose (Accutrend GC system) and cholesterol levels (Accutrend GC system). These machines were both calibrated once per week using control solutions, and each time a new pack of strips was used. In addition, detailed ophthalmic measures were taken, including visual acuity which was measured using logMAR. According to WHO standards, participants with normal or near normal vision were classified as not being visually impaired. Visual impairment was defined as visual acuity <6/18 in the better eye with available correction; hypertension as systolic blood pressure (SBP) ≥140 mmHg or diastolic blood pressure (DBP) ≥90 mmHg, previous diagnosis by a physician or current use of antihypertensive medication; and diabetes as previous diagnosis by a physician, reported current medication or diet control for diabetes, or random blood glucose level ≥11.1 mmol/L (Tables 1 and 2).
Statistical analysis
A structural model for the variables depicted in Fig. 1 was formulated with all direct and indirect effects estimated simultaneously. We employed a formal—model-based—estimation of the indirect effects (mediation) of the effect of education and material resources on the three binary health outcomes that were jointly modelled in order to account for their intercorrelations. This allowed us to estimate the extent to which the effects of education and material resources on health are due to the latent summary of smoking and alcohol use. This would not be possible using standard multiple regression models, since the estimated effect of education and material resources on the three health outcomes would be simply adjusted for each other, therefore neglecting the indirect perspective. (De Stavola et al. 2006) According to our conceptual model in Fig. 1, education was entered as an exogenous variable, whereas latent summary variables for the accumulation of material resources and health-related behaviour were specified as intermediate variables. The accumulation of material resources as latent summary variable was derived from combining several assets. Previous work conducted in this setting showed that household assets were closely related to household per capita expenditure and household self-rated wealth (Kuper et al. 2008) and therefore the assets-based latent variable can be thought of as a wealth indicator.
The summary of health-related behaviour was derived combining indicators of smoking status and alcohol use, with high scorers being those who excessively drank alcohol and smoke. Both latent variable measurement models were estimated as unidimensional confirmatory factor analysis (CFA) models appropriate for the combination of binary ordinal and continuous indicators. The latent variables specified within the CFA models comprise common (shared) variance between the observed indicators and exclude their unique (not shared) variance as random error. The latent summaries can be thought of as accurate representations of material resources and health-related behaviour, since one of the properties of latent variable measurement models is their ability to capture unobserved heterogeneity (Rabe-Hesketh and Skrondal 2008). Mother tongue was classified as “Kikuyu” and “Kalenjin” and “Others” and used to represent “Ethnicity”. Age in years and gender were included as confounders. All models were estimated with the weighted least squares, mean and variance adjusted (WLSMV) estimator, with the “complex” command of the Mplus 6.12 software (Muthen and Muthen 1998–2010). Within this framework, regressions between predictors and continuous latent variables are modelled as linear regressions, whereas the regressions of the three binary health outcomes on all predictors and mediators are modelled as probit regressions. Model fit was assessed with the comparative fit index (CFI), the Tucker Lewis index (TLI) and the root mean square error of approximation (RMSEA). We note that for the CFI and TLI, values >0.90 are indicative of acceptable fit, and values >0.95 indicative of good fit, whereas for the RMSEA values <0.08 are indicative of acceptable and values <0.06 of good fit (Bollen 1989).
Results
The accumulation of material resources CFA model fitted the data, CFI = 0.97, TLI = 0.96, RMSEA = 0.04 and all standardised factor loadings representing the association between the observed assets and the latent summary were satisfactory (range 0.74–0.88). Similarly, the CFA model for alcohol use and smoking had good fit, CFI = 0.95, TLI = 0.94, RMSEA = 0.06, and both standardised loadings were satisfactory (0.68 and 0.73, respectively). We used these measurement models to develop the components of the conceptual model of Fig. 1. We estimated models stratified by area of residence due to the large differences in the prevalence and distribution of hypertension, diabetes and visual impairment that have been reported between urban and rural areas in Africa (Mathenge et al. 2010). In Tables 3 and 4, we present the estimated model parameters, their corresponding 95 % confidence intervals as well as standardised parameters derived from the path analytic model.
In rural areas, the summary variable representing health-related behaviour had a negative association with the prevalence of diabetes, b = −0.219 (−0.330 to −0.107), but was not associated with the prevalence of hypertension or visual impairment. The latent indicator of accumulated material resources was positively associated with the prevalence of hypertension, b = 0.226 (0.138–0.290) and with the prevalence of diabetes, b = 0.313 (0.185–0.431), but was negatively associated with the prevalence of visual impairment, b = −0.173 (−0.266 to −0.075) and with smoking and alcohol use, b = −0.257 (−0.347 to −0.160). We observed a positive indirect effect (via health-related behaviour) on the prevalence of diabetes b = 0.056 (0.016–0.096). However, we did not observe a significant indirect effect of material resources on either the prevalence of hypertension or visual impairment. Education was not associated with any of the three outcomes, or with smoking and alcohol use, but had, as expected, a strong graded association with the accumulation of material resources (higher educational level, more material resources), which was reflected in the significant indirect—via material resources—associations with the three outcomes.
In urban areas, the latent summary variable representing health-related behaviour was not associated with the prevalence of hypertension, diabetes and visual impairment. The accumulation of material resources was positively associated with the prevalence of hypertension, b = 0.299 (0.183–0.415) and with the prevalence of diabetes, b = 0.562 (0.404–0.719), as well as with health-related behaviour, b = −0.209 (−0.362 to −0.057), but not with visual impairment, b = 0.184 (−0.013 to 0.380). On the contrary, we did not observe a significant indirect effect of material resources on the three health outcomes. Education was not associated with the prevalence of hypertension and diabetes, but had a negative association with the prevalence of visual impairment. It was not associated with smoking and alcohol use, but, as in rural areas, had a strong graded association with the accumulation of material resources (higher educational level, more material resources), which was reflected in the significant indirect—via material resources—associations with the prevalence of hypertension and diabetes. It appears that better education leads to the accumulation of material resources, and in urban areas this relates to higher prevalence of hypertension and diabetes, while education has a direct protective effect against visual impairment.
Discussion
We found evidence of later life health inequalities in both rural and urban areas of Nakuru, Kenya, but the pattern of these inequalities varied between the SEP indicators and the three health outcomes considered here, suggesting that different dimensions of SEP provide different aspects of protection as well as risk. Furthermore, it appears that the measures of health-related behaviour we employed here do not mediate the association between SEP and the three health outcomes, suggesting that at least one important dimension of social causation theory does not adequately account for the pattern of inequalities observed in urban and rural areas of Nakuru, Kenya. Education was, as expected, positively associated with the accumulation of material resources, which in our study we believe largely reflects household consumption (Kuper et al. 2008). However, the pattern of direct effects of education and material resources with health was not consistent and varied by outcome. Education was not directly associated with the prevalence of hypertension and diabetes in either urban or rural areas and had a weak negative association with visual impairment only in rural areas. However, accumulated material resources were positively associated with hypertension and diabetes in both urban and rural areas, indicating that diseases of affluence are more common among people with higher SEP. This finding is in contrast with the pattern observed in high-income countries (Booth and Hux 2003), but consistent with findings in low-/mid-income countries in other world regions (Mendez et al. 2003), in other countries in Africa (Addo et al. 2009) as well as in historical European populations (Razzell and Spence 2006). We speculate that the observed difference in the pattern of health inequalities in Kenya compared to high-income countries which are past the epidemiologic transition is that participants with higher income and those living in urban areas are able to afford processed foods high in fats and salts and also experience a decrease in the level of physical activity since they are more likely to own a car and be employed in an office-based job (Razzell and Spence 2006; Kerry et al. 2005). These lifestyle differences not captured in our data may contribute to higher levels of obesity, which is a well known risk factor for both hypertension and diabetes (Kaplan 1989; Mokdad et al. 2003; Sorof and Daniels 2002; Steppan et al. 2001).
To test this hypothesis, we estimated a path analytic model where an indicator of obesity was added as a mediator of the association between material resources and the three health outcomes. Similarly with previous studies in Africa (Sodjinou et al. 2008), accumulated material resources were positively associated with obesity, which as expected had a positive association with the prevalence of hypertension and diabetes. Both indirect effects were statistically significant and stronger in urban areas, confirming that the positive association between material resources, hypertension and diabetes is partly due to obesity. However, the independent to obesity and other lifestyle factors, direct associations between the accumulation of material resources, hypertension and diabetes remained significant. We postulate that this finding could be due to the effects of high salt intake and physical activity on hypertension and diabetes (Law et al. 1991; Weinberger 1996; Batty et al. 2002; Healy et al. 2007).
Both education and the accumulation of material resources had a negative association with the prevalence of visual impairment, a finding consistent with what has been reported in high income countries (Wu et al. 2010) and also in studies on Africa (Dandona and Dandona 2001). Participants with high income have better access to health-care facilities, which is crucial for cataract blindness—the leading cause of blindness in Kenya (Mathenge et al. 2007)—that can be treated with a simple surgery. Educated participants are more likely to be aware of the availability of such procedures and are therefore more likely to be treated and are also more likely to be able to read and so seek treatment earlier to preserve their vision. However, this was evident only in urban areas, where perhaps access to health-care services is easier. Furthermore, SEP has been associated with the prevalence of other eye diseases that can potentially lead to blindness, such as trachoma which has been linked to poor sanitation (Pruss and Mariotti 2000).
The observed later life health inequalities were not due to variation between different SEP groups in the measures of health-related behaviour employed in our study. Education was not associated with smoking and alcohol use in both urban and rural areas, a finding in contrast with what has been widely reported in high income countries, where smoking and alcohol use are strongly socially patterned (Smith et al. 1994). The accumulation of material resources was weakly associated with health related lifestyle choices in both urban and rural areas, indicating that similarly to the pattern observed in countries after the epidemiologic transition, low income, represented here by the accumulation of material resources, is associated with smoking and alcohol use. However, the smoking and alcohol use summary variable was not consistently associated with the three health outcomes as was expected, since in urban areas it had a positive association with diabetes, with the pattern reversed in rural areas, indicating that at least for the time being the lifestyle explanation of health inequalities is not applicable in pre-epidemiologic transition countries.
The strengths of this study include the availability of a population-based sample and of observer measured health indicators for hypertension, diabetes and visual impairment. Furthermore, an appropriate modelling strategy was employed that allowed us to formally estimate direct and indirect effects of education and material resources on the three health outcomes. A major limitation is the lack of measures of psychosocial factors such as empowerment, relative social status, social integration and exposure to stressful events that are thought to mediate the association between SEP and health (Lynch et al. 2000; Wilkinson 1997) and the reliance on relatively crude measures of smoking and alcohol use in order to approximate health-related behaviour. Although we statistically adjusted for confounders such as age, gender ethnicity and stratified by area of residence, our results rely on correct specification of the model and may be influenced by unknown unmeasured confounders (Hernan et al. 2002). Another limitation is the cross-sectional nature of the study. It is plausible that previously diagnosed disease may influence behavioural patterns and to some extent attenuate our parameter estimates. Future longitudinal studies are needed to investigate the mechanism that underlies the association between SEP and health in Africa, particularly since major lifestyle changes occur as a result of rapidly urbanised populations. Studies that will include detailed dietary and physical activity information are particularly needed to investigate the role of obesity in the pattern of health inequalities and further disentangle the positive association between SEP indicators and the prevalence of hypertension and diabetes.
References
Addo J, Smeeth L, Leon DA (2009) Socioeconomic position and hypertension: a study of urban civil servants in Ghana. J Epidemiol Community Health 63(8):646–650. doi:10.1136/jech.2008.081828
Batty GD, Shipley MJ, Marmot M, Smith GD (2002) Physical activity and cause-specific mortality in men with Type 2 diabetes/impaired glucose tolerance: evidence from the Whitehall study. Diabet Med 19(7):580–588
Bollen KA (1989) Structural equations with latent variables. Wiley, New York
Booth GL, Hux JE (2003) Relationship between avoidable hospitalizations for diabetes mellitus and income level. Arch Intern Med 163(1):101–106
Bovet P et al (2002) Distribution of blood pressure, body mass index and smoking habits in the urban population of Dar es Salaam, Tanzania, and associations with socioeconomic status. Int J Epidemiol 31(1):240–247
Dandona R, Dandona L (2001) Socioeconomic status and blindness. Br J Ophthalmol 85(12):1484–1488. doi:10.1136/bjo.85.12.1484
De Stavola BL et al (2006) Statistical issues in life course epidemiology. Am J Epidemiol 163(1):84–96. doi:10.1093/aje/kwj003
Healy GN et al (2007) Objectively measured light-intensity physical activity is independently associated with 2-h plasma glucose. Diabetes Care 30(6):1384–1389
Hernan MA, Hernandez-Diaz S, Werler MM, Mitchell AA (2002) Causal knowledge as a prerequisite for confounding evaluation: an application to birth defects epidemiology. Am J Epidemiol 155(2):176–184
Kaplan NM (1989) The deadly quartet: upper body obesity, glucose intolerance, hypertriglyceridemia and hypertension. Arch Intern Med 149(7):1514–1520
Kaplan GA, Pamuk ER, Lynch JW, Cohen RD, Balfour JL (1996) Inequality in income and mortality in the United States: analysis of mortality and potential pathways. Br Med J 312(7037):999–1003
Kerry SM et al (2005) Rural and semi-urban differences in salt intake, and its dietary sources, in Ashanti, West Africa. Ethn Dis 15(1):33–39
Kuper H, Polack S, Eusebio C, Mathenge W, Wadud Z, Foster A (2008) A case–control study to assess the relationship between poverty and visual impairment from cataract in Kenya, the Philippines, and Bangladesh. PLoS Med 5(12):1716–1728. doi:e24410.1371/journal.pmed.0050244
Law MR, Frost CD, Wald NJ (1991) By how much does dietary salt reduction lower blood pressure. 3. Analysis of data from trials of salt reduction. Br Med J 302(6780):819–824
Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJ (2006) Global and regional burden of disease and risk factors, 2001: systematic analysis of population health data. Lancet 367(9524):1747–1757
Lynch JW, Smith GD, Kaplan GA, House JS (2000) Income inequality and mortality: importance to health of individual income, psychosocial environment, or material conditions. Br Med J 320(7243):1200–1204
Mackenbach JP et al (2008) Socioeconomic inequalities in health in 22 European countries. N Engl J Med 358(23):2468–2481
Marmot MG et al (1991) Health inequalities among British civil servants: the Whitehall-II study. Lancet 337(8754):1387–1393
Marmot M, Banks J, Blundell R, Lessof C, Nazroo J (eds) (2002) Health, wealth and lifestyles of the older population in England: the 2002 english longitudinal study of ageing. Institute for Fiscal Studies, London
Mathenge W et al (2007) Rapid assessment of avoidable blindness in Nakuru District, Kenya. Ophthalmology 114(3):599–605. doi:10.1016/j.ophtha.2006.06.057
Mathenge W, Foster A, Kuper H (2010) Urbanization, ethnicity and cardiovascular risk in a population in transition in Nakuru, Kenya: a population-based survey. BMC Public Health 10(1):569
Mendez MA, Cooper R, Wilks R, Luke A, Forrester T (2003) Income, education, and blood pressure in adults in Jamaica, a middle-income developing country. Int J Epidemiol 32(3):400–408. doi:10.1093/ije/dyg083
Mokdad AH et al (2003) Prevalence of obesity, diabetes, and obesity-related health risk factors 2001. Jama J Am Med Assoc 289(1):76–79
Muthen LK, Muthen BO (1998–2010) Mplus user’s guide, 6th edn. Los Angeles, CA
Omran AR (1971) Epidemiologic transition: theory of epidemiology of population change. Milbank Meml Fund Q 49(4):509–538
Ploubidis G, DeStavola B, Grundy E (2011) Health differentials in the older population of England: an empirical comparison of the materialist, lifestyle and psychosocial hypotheses. BMC Public Health 11(1):390
Pruss A, Mariotti SP (2000) Preventing trachoma through environmental sanitation: a review of the evidence base. Bull World Health Organ 78(2):258–266
Rabe-Hesketh S, Skrondal A (2008) Classical latent variable models for medical research. Stat Methods Med Res 17(1):5–32. doi:10.1177/0962280207081236
Razzell P, Spence C (2006) The hazards of wealth: adult mortality in pre-twentieth-century England. Soc Hist Med 19(3):381–405. doi:10.1093/shm/hkl048
Schrijvers CTM, Stronks K, van de Mheen HD, Mackenbach JP (1999) Explaining educational differences in mortality: the role of behavioral and material factors. Am J Public Health 89(4):535–540
Smith GD, Blane D, Bartley M (1994) Explanations for socio-economic differentials in mortality: evidence from Britain and elsewhere. Eur J Public Health 4(2):131–144. doi:10.1093/eurpub/4.2.131
Sodjinou R, Agueh V, Fayomi B, Delisle H (2008) Obesity and cardio-metabolic risk factors in urban adults of Benin: relationship with socio-economic status, urbanisation, and lifestyle patterns. BMC Public Health 8:84. doi:10.1186/1471-2458-8-84
Sorof J, Daniels S (2002) Obesity hypertension in children: a problem of epidemic proportions. Hypertension 40(4):441–447. doi:10.1161/01.hyp.0000032940.33466.12
Steppan CM et al (2001) The hormone resistin links obesity to diabetes. Nature 409(6818):307–312
Turner AG, Magnani RJ, Shuaib M (1996) A not quite as quick but much cleaner alternative to the Expanded Programme on Immunization (EPI) cluster survey design. Int J Epidemiol 25(1):198–203
Weinberger MH (1996) Salt sensitivity of blood pressure in humans. Hypertension 27(3):481–490
Wilkinson RG (1997) Socioeconomic determinants of health—health inequalities: relative or absolute material standards? Br Med J 314(7080):591–595
Wu RY et al (2010) Smoking, socioeconomic factors, and age-related cataract. Arch Ophthalmol 128(8):1029–1035
Acknowledgments
George B. Ploubidis is supported by a Medical Research Council (MRC) Population Health Science fellowship—G0802442.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Ploubidis, G.B., Mathenge, W., De Stavola, B. et al. Socioeconomic position and later life prevalence of hypertension, diabetes and visual impairment in Nakuru, Kenya. Int J Public Health 58, 133–141 (2013). https://doi.org/10.1007/s00038-012-0389-2
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00038-012-0389-2