Tuberculosis risk factors in South Africa, 2008 to 2017: A Generalised Estimating Equations approach

Abstract

Background: Although, death due to tuberculosis has been on the decline. In 2016, 124 000 people died of tuberculosis in South Africa and the disease was declared the leading cause of death by Statistics South Africa. Continued efforts to use research to create a nation free of tuberculosis are underway.  Methods: A repeated measures investigation was performed with the aim of identifying the persistent predictors and the long-term patterns of tuberculosis infection in South Africa for the period 2008 to 2017. The most suitable Generalised Estimating Equations that describe the population average probability of infection over time were applied to a sample of respondents taken from the National Income Dynamics Survey data, wave 1 to wave 5. The response variable was binary with the outcome of interest being the respondents that self-reported to have been diagnosed with tuberculosis. To improve estimation efficiency, the best working correlation matrix for this data was selected.  Results: We used a sample of 8510 individuals followed for five waves, of these, 3.7%, 2.54%, 4.15%, 5.72% and 5.99% for waves 1, 2, 3, 4 and 5 respectively, reported to have been diagnosed with tuberculosis. Findings revealed that the independent working correlation matrix with the model-based standard error estimates gave the most robust results for the repeated measures tuberculosis data in South Africa. Furthermore, over the years, the average probability of being diagnosed with tuberculosis was positively associated with being single, male, middle-aged (30- 59 years), black African, unemployed, smoking, lower education levels, lack of regular exercise, asthma, suffering from other diseases, lack of access to improved sanitation, lower household income and expenditure.  Conclusion: The probabilities of tuberculosis infection are independent within individuals over time. The inequalities in socioeconomic status in South Africa caused the poor to be more at risk of tuberculosis over time from 2008 to 2017.