Chronic non-communicable disease multimorbidity in South African adults: evidence from the who sage study
Date
2022
Authors
Chidumwa, Glory
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Background
Non-communicable diseases (NCDs) are the leading cause of global mortality, and in South Africa were estimated to account for 57.4% of the total burden of disease in 2016. Within an
individual, the co-existence of two or more chronic (at least three months) non-communicable, mental health or infectious disease is referred to as ‘multimorbidity’ (MM). While risk factors
for MM are present across the life course, the onset of most NCDs occurs in middle to older age. However, there is limited research that explores MM among middle aged and older adults
in South Africa.
Aim
The aim of this thesis was to investigate MM in middle-aged and older South African adults from the WHO SAGE cohort. This was addressed in three parts:
1. To determine the prevalence of multimorbidity and the co-occurrence of chronic diseases in a cohort of South African adults over the age of 50 years, and to identify the demographic, anthropometric and behavioural factors associated with different multimorbidity clustering.
2. To examine the spatial distribution of hypertension and diabetes jointly, and the distribution of shared unmeasured characteristics on hypertension and diabetes in South African
middle aged and older adults.
3. To determine the complex inter-relationships between socio-economic, sociodemographic, behavioural, and environmental factors associated with multimorbidity and depression in South African middle aged and older adults. This study will contribute to an expansion of the epidemiology of NCDs and biostatistical literature through the novel application of statistical techniques such as latent class analysis (LCA), bivariate joint shared component modelling, and the generalized structural equation model (gSEM) approach, used to address the research objectives above.
Methods
Cross-sectional secondary analysis of data collected as part of a panel study carried out by the WHO SAGE Wave 2 in South Africa in 2015 was completed. The current thesis included adults (≥18 years old for objective 1 and ≥40 years old for objectives 2 and 3) for whom data on 7 NCDs (angina, arthritis, asthma, chronic lung disease, depression, diabetes, and hypertension) and socioeconomic, demographic, behavioural, and anthropometric information were available. Further details of the South African SAGE sample are given in separate methods sections. Latent class analysis was used to identify groups and determine the co-occurrence of the NCDs. Bivariate joint shared component modelling was used to assess the clustering and association between diabetes and hypertension and to jointly model the shared and disease-specific geographical variation of hypertension and diabetes. Lastly, I utilized the logit models and gSEM to explore the association between socioeconomic, demographic, and behavioural factors, and multimorbidity and depression.
Results
The study used the WHO SAGE South Africa Wave 2 data collected in 2015 on 2761 participants aged 18 years and above. The majority of the sample were female (n=1846; 67%) The prevalence of multimorbidity was 21%. The LCA identified three latent classes which were named as follows: minimal MM risk (83%), concordant MM (i.e., expected, or typical clustering of hypertension and diabetes; 11%), and discordant MM (less typical clustering of combination of angina, asthma, chronic lung disease, arthritis and depression; 6%). Using the minimal MM risk group as the reference, female [Relative risk ratio (RRR) = 4.57; 95% Confidence Interval (CI) (1.64; 12.75); p-value=0.004] and older [RRR=1.08; 95% CI (1.04; 1.12); p-value<0.001] participants were more likely to belong to the concordant MM group. Tobacco users [RRR= 8.41; 95% CI (1.93; 36.69); p-value=0.005] and older [RRR=1.09; 95% CI (1.03; 1.15); pvalue=0.002] participants had a higher likelihood of belonging to the discordant MM group. As hypertension and diabetes commonly co-occur in South African adults the second study modelled the shared and disease-specific spatial distribution of these two NCDs using bivariate joint shared component modelling. The shared component of diabetes and hypertension had distinct spatial patterns with higher odds in the eastern districts of Kwa-Zulu Natal and central Gauteng province of South Africa. The shared component represents unmeasured influences such as health behaviour characteristics or social determinants of health in our population. My study further showed that the shared component for hypertension and diabetes, which may include ecological factors and environmental determinants such as population density, pollution, transport, power, and local food environment is more pronounced in certain South African provinces such as Gauteng and Kwa-Zulu Natal.
Using logistic regression and generalized structural equation modelling (gSEM) to explore the associations between socio-economic, socio-demographic, behavioural and environmental
factors, and risk of depression and multimorbidity, the results were as follows: In the unadjusted logistic regression analyses, feeling “unsafe” [aOR=2.04; 95% Confidence Interval: 1.25; 3.42], being female, [aOR=1.93; 95% Confidence Interval: 1.02; 3.62], and older age [aOR=1.05; 95% Confidence Interval: 1.02; 1.08] were associated with higher odds for multimorbidity. In addition, being female, belonging to the highest wealth tertile relative to those in the lowest tertile, and living in an urban area were significantly associated with higher odds of depression [OR=1.39; 95% Confidence Interval: 0.59; 3.29]. Similarly, in the gSEM model, where models are estimated concurrently, demographic factors [older age (aOR=1.03, 95% Confidence Interval: 1.01; 1.05) and being female (aOR= 3.02; 95% Confidence Interval: 1.88; 4.86)] and behavioural factors [individuals with history of tobacco avoidance (aOR=0.46; 95% Confidence Interval: 0.27;0.75), and good sleep quality (aOR=0.59; 95% Confidence Interval: 0.39;0.91)] were significantly associated with multimorbidity. Moreover, using the gSEM approach, multimorbidity had two-fold odds of depression and was statistically significant (aOR=2.41; 95% Confidence Interval: 1.36;4.28).
Discussion and conclusion
Results indicate that in my sample of middle aged and older South African adults 1 in 5 people above the age of 45 years have MM. Risk factors for multimorbidity included older age, female sex and tobacco use, but results show that these may differ depending on whether the diseases are concordant or discordant, which may suggest different avenues for intervention. Given the co-occurrence of NCDs, I underscore the need for the healthcare system to focus on managing multiple diseases rather than a vertical approach in managing
single diseases, specifically for hypertension and diabetes. In addition, policy-makers may potentially use our spatial results for purposes of more localised resource allocation and
prevention health programs in high burden hypertension and diabetes areas in South Africa. In addition, future efforts should focus on understanding the unmeasured shared component, which may include infectious diseases, or frequently co-occurring common conditions, and to evaluate clustering patterns.
Description
A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy (Public Health in Biostatistics) to the Faculty of Health Sciences, School of Public Health, University of the Witwatersrand, Johannesburg, 2022