The study followed the cross-sectional study design. Cross-sectional secondary data were used from the 2019 General Household Survey (GHS) and focussed on persons aged 60 years and older.

The setting of the study is South Africa, which comprising 9 provinces, each further subdivided into 52 districts, including 8 metropolitan areas and 226 local municipalities. As of 2022, the total population of South Africa stood at 62 million individuals.²⁶ Individuals classified as older persons, aged 60 and above, make up over 8.5% of the total population.²⁶ The Eastern Cape province had the highest number of older persons at 11.6%, followed by Western Cape at 10.7%, and the Northern Cape with the lowest at 10.1%.²⁶

The 2019 GHS, which is secondary cross-sectional data, was utilised, encompassing a weighted sample of 4 887 334 individuals aged 60 years and above. The decision to employ the 2019 GHS stemmed from concerns over the data integrity of more recent GHS iterations, which may have been impacted by the repercussions of the coronavirus disease 2019 (COVID-19) pandemic. The GHS is a household-based cross-sectional survey, representative at the national level, covering all nine provinces within the country. The target population for this survey comprised all private households across the nation.²⁷

The survey excluded the population living in institutionalised settings ‘such as students’ hostels, old-age homes, hospitals, prisons, military barracks’ and others. The survey used a two-stage stratified sampling design.²⁷ In the first stage, primary sampling units (PSUs) were selected, and in the second stage, dwelling units (DUs) were selected.²⁷ Primary sampling unit is a geographical or administrative unit used for sampling; it helps ensure geographical representation in the sample.^28,29 Whereas a DU is a physical unit where people reside and represent the actual places where data are collected.²⁷ The PSUs are selected first in the sampling process and thereafter DUs are selected within those PSUs.^28,29 The response rate of the 2019 GHS was 87.2%.²⁷ Further information about the GHS sampling, study setting and weighting can be found in the 2019 GHS metadata report.²⁷ Figure 1 explains the steps taken to reach our study sample.

The data were extracted from GHS 2019. The study’s dependent variable is based on individuals’ self-reported chronic disease diagnoses, derived from their self-reported health status. The participants were asked about their chronic disease diagnoses as reported by a physician: ‘Has a doctor/nurse/other healthcare worker at a clinic/hospital/private practice ever told (name) that he/she has/had any of the following?’²⁷ In the survey, conditions such as: (1) asthma, (2) diabetes, (3) cancer, (4) HIV (human immunodeficiency virus) and AIDS (acquired immunodeficiency syndrome), (5) hypertension or high blood pressure, (6) arthritis, (7) stroke, (8) heart attack or myocardial infarction, (9) tuberculosis, (10) mental illness, (11) epileptic seizure, (12) meningitis and sinusitis, (13) pneumonia, (14) bronchitis, (15) high cholesterol, (16) osteoporosis and (17) malaria were included.²⁷ The survey asked respondents to respond with a ‘yes’ or ‘no’ to each item in the list. This study focusses only on the following diagnosed diseases: diabetes, hypertension, cancer, arthritis and stroke.

We used Stata version 16 (StataCorp LLC, Texas, USA) to analyse the data in this study.³⁰ For this study, three types of analyses were used: univariate, bivariate and multivariate analysis. The univariate analysis includes descriptive statistics. For bivariate analysis, the chi-square test was utilised to show the correlation among diagnoses of chronic disease and selected independent variables. In the multivariate analysis, binary logistic regression was used to analyse the relationship between the selected background characteristics and being diagnosed with a chronic disease. We used the ‘svy’ command in Stata to adjust for the complex sampling structure of the data in the analyses. We further used the variance inflation factor (VIF) to test for multicollinearity in the explanatory factors. The multicollinearity test found no collinearity between the variables; the minimum VIF was 1.06, the maximum VIF was 2.05 and the mean VIF was 1.39.

We used secondary data from the 2019 GHS. The data collected by Statistics South Africa followed all the necessary ethical considerations. The collection of data by Statistics South Africa is guided by the fundamental principles of statistics (see more details at https://www.statssa.gov.za/?page_id=361).

The characteristics of the study sample are presented in Table 2. Based on the findings, older people in the age group 60–64 years were dominating, whereas those in the age group 75+ were less dominating. There were more females than males in the sample. The black population group constituted the largest sample in the study; there were fewer older persons from the Indian and Asian population group. In terms of marital status, the majority of the sample was married, while a few were cohabiting. Those with secondary education made up the largest percentage. The majority of the study sample had no disability difficulty. Over 57% of the study sample was from rich households. Over 29% of the study sample was from extended, female-headed households. Over 64% of the study sample was from urban areas, while 3% was from farm areas. The majority of the study sample was from Gauteng province.

The results presented in Table 2 indicate the frequency of chronic disease diagnoses as reported by individuals, categorised according to various demographic factors. The results revealed that age, population group, sex, marital status, level of education, disability status, household composition and province were associated with self-reported chronic disease diagnoses. Those aged 70–74 years had a higher prevalence (57.6%) of self-reported chronic disease diagnoses. Females had a higher prevalence (55.7%) of self-reported chronic disease diagnoses. The black population group had a higher prevalence (54.9%) of self-reported chronic disease diagnoses, while it was lower (40.5%) among the white population group. Those who were never married had a higher prevalence (49.0%) of self-reported chronic disease diagnoses, while it was low (5.9%) for those who were no longer married.

The prevalence of self-reported chronic disease diagnoses was lower for those with higher socioeconomic status. The findings showed that those with primary education had a higher prevalence (54.4%) of self-reported chronic disease diagnoses, while it was lower (41.1%) for those with higher education. In terms of household wealth status, those from average-wealth households had a higher prevalence (52.4%) of self-reported chronic disease diagnoses, while it was lower (49.0%) for those who were from rich households. Those who had a lot of difficulty, in terms of disability status, had a higher prevalence (57.0%) of self-reported chronic disease diagnoses. In terms of household composition, those who were from extended, female-headed households, had a higher prevalence (58.8%) of self-reported chronic disease diagnoses. There were geographical variations in chronic disease diagnoses. The findings showed that those who were from traditional areas had a higher prevalence (52.0%) of self-reported chronic disease diagnoses. In terms of the province, those who were from KwaZulu-Natal (57.3%) and Eastern Cape (55.8%) had a higher prevalence (57.0%) of self-reported chronic disease diagnoses, while it was lower (37.1%) among those from Limpopo (see the visual presentation by province in Figure 2).

Table 3 presents the results of the multivariate analysis exploring the factors influencing self-reported chronic disease diagnoses among South Africa’s older population. The analysis revealed several significant findings. Individuals aged 70–74 years were 1.65 times more likely to report chronic diseases (95% confidence interval [CI]: 1.39–1.97) compared to those aged 60–64 years. Similarly, individuals aged 75–79 years had a 1.37 times higher likelihood (95% CI: 1.11–1.70) of reporting chronic diseases compared to the reference group (60–64 years old). Those aged 80 and above were 1.40 times more likely (95% CI: 1.12–1.75) to report chronic diseases compared to the 60–64 age group. Females exhibited a significantly higher likelihood, being 1.78 times more likely (95% CI: 1.54–2.06) to report chronic diseases compared to males. When considering population groups, Indian and Asian older individuals were 0.37 times less likely (95% CI: 0.24–0.56) to report chronic diseases compared to the Black population group. Similarly, Coloured individuals were 0.51 times less likely (95% CI: 0.38–0.67), and White individuals were 0.39 times less likely (95% CI: 0.30–0.49) to report chronic diseases compared to the Black population group.

The study revealed that older individuals who were previously married were 1.75 times more likely (95% CI: 1.27–2.42) to report self-reported chronic diseases compared to those who were cohabiting. Similarly, those currently married exhibited a 1.39 times higher likelihood (95% CI: 1.02–1.90) of reporting chronic diseases compared to their cohabiting counterparts. Individuals with primary education attainment showed a 1.19 times higher likelihood (95% CI: 1.00–1.40) of reporting chronic diseases compared to those with no education. Those categorised as having ‘some difficulty’ in terms of disability status were 1.42 times more likely (95% CI: 1.23–1.63) to report chronic diseases compared to those reporting ‘no difficulty’ in disability. Similarly, older individuals classified as experiencing ‘a lot of difficulty’ in disability status were 1.32 times more likely (95% CI: 1.09–1.59) to report chronic diseases compared to those with ‘no difficulty’ in disability. Individuals from poor households had a 0.73 times lower likelihood (95% CI: 0.59–0.89) of reporting self-reported chronic diseases compared to those from rich households.

Likewise, individuals from households with average wealth exhibited a 0.83 times reduced likelihood [95% CI: 0.69–1.00] of being diagnosed with self-reported chronic diseases than those from rich households. In summary, the likelihood of being diagnosed with chronic conditions decreased as wealth status decreased. The results indicate that the likelihood of being diagnosed with self-reported chronic diseases was significantly predicted by household composition. The older individuals living in lone households that were female-headed exhibited a 0.52 times lower likelihood [95% CI: 0.38–0.73] of being diagnosed with self-reported chronic diseases compared to their counterparts in male-headed nuclear households.

Similarly, individuals of older age residing in male-headed extended households showed a 0.70 times decreased likelihood [95% CI: 0.58–0.85] of receiving diagnoses for self-reported chronic diseases than those residing in households that were nuclear male-headed. In households that were nuclear female headed, there was a 0.61 times reduced likelihood [95% CI: 0.45–0.83] of being diagnosed with self-reported chronic diseases compared to households that were nuclear male-headed. Additionally, individuals from households that were extended female-headed exhibited a 0.69 times lower likelihood [95% CI: 0.53–0.89] of being diagnosed with self-reported chronic diseases than those from households that were nuclear male-headed.

Older individuals who resided in the Western Cape had a 2.01 times higher likelihood [95% CI: 1.45–2.80] of being diagnosed with self-reported chronic diseases than older individuals from Limpopo. Similarly, those who resided in the Eastern Cape had a 2.24 times higher likelihood [95% CI: 1.80–2.80] of being diagnosed with self-reported chronic diseases than older individuals from Limpopo. Older persons residing in Gauteng had a 1.93 times higher likelihood [95% CI: 1.49–2.50] of being diagnosed with self-reported chronic diseases than older individuals from Limpopo. Individuals of older age residing in KwaZulu-Natal showed a 2.68 times increased likelihood [95% CI: 2.11–3.39] of being diagnosed with self-reported chronic diseases compared to those from the Limpopo province.

It is widely acknowledged in research that a certain degree of bias may be unavoidable in studies. The GHS (2019) excluded the population living in institutions (hospitals, old-age homes, etc.) from the sample. Excluding older individuals living in institutions means that the results may not fully represent this significant segment of the older population. Consequently, from this dataset, it is challenging to ascertain the living arrangements of this population, including those residing in old-age homes or hospitals. The GHS also did not include information on lifestyle-related factors (i.e., food the respondents eat, whether they exercise or not, whether they smoke or not, whether they drink alcohol or not, etc.). Access to health facilities could also be a limitation, in that some provinces have healthcare services that are closer to the people. This is not the case in predominantly rural provinces, where people have to travel long distances to access healthcare facilities. However, because of the methodology used to analyse this data, the findings of this study are generalisable to the population of persons aged 60 years and older in South Africa.