One in five South Africans are multimorbid: An analysis of the 2016 demographic and health survey.

Multimorbidity is a global research priority, yet relatively little is known about it in low and middle income countries. South Africa has the largest burden of HIV worldwide but also has a growing burden of non-communicable diseases; potentially leading to uncommon disease combinations. Information about the prevalence of multimorbidity and factors associated with it can assist in healthcare planning and targeting groups of people for interventions. This study aimed to determine the prevalence of multimorbidity by age and sex, as well as factors associated with multimorbidity in people 15 years and older. This study analyses the nationally representative 2016 South African Demographic Health Survey. The sample included 10 336 people who participated in the Adult Health questionnaire and approximately 7 961 people who provided biomarkers. Multivariate logistic regression was used to measure the association of multimorbidity with age, sex, living in an urban or rural area, education level, wealth level, employment status, body mass index, current alcohol or tobacco use. All analyses were conducted using STATA 15. Multimorbidity was present in 20.7% (95% CI: 19.5%- 21.9%) of participants; in 14.8% (95% CI: 13.4% - 16.3%) of males and 26.2% (95% CI: 24.7-27.7%) of females. Multimorbidity increased with age; with the highest odds in the 55-64 years old age group (OR: 24.910, 95% CI: 14.901-41.641, p < 0.001) compared to those aged 15-24 years. The odds of multimorbidity was also higher in young females compared to young males (OR: 2.734, 95% CI: 1.50-4.99, p = 0.001). Possessing tertiary education (OR: 0.722, 95% CI: 0.537-0.97, p = 0.031), being employed (OR: 0.813, 95% CI: 0.675-0.979, p = 0.029) or currently using alcohol (OR: 0.815, 95% CI: 0.686-0.968, p = 0.02) was protective against multimorbidity. Multimorbidity is prevalent within the South African population, with females and older adults being most affected. However, multimorbidity is also observed in younger adults and most likely driven by the high prevalence of HIV and hypertension.

Introduction

People living with more than one disease (also known as multimorbidity) have their lives impacted in many ways; including a reduced quality of life [1-6], an increased risk of dying [7-9] and an intensified need to utilise healthcare [10-14]. Despite these negative impacts, the area of multimorbidity remains under researched when compared to research afforded to single disease conditions [15]. This is particularly acute in low and middle income countries (LMICs) where 5% of multimorbidity research globally has taken place [15]. Little is known about multimorbidity in LMICs where disease burdens are thought to differ from countries with more established multimorbidity profiles.

South Africa is an upper middle-income country [16] with a quadruple burden of disease consisting of: HIV/AIDS and tuberculosis (TB); other communicable diseases, perinatal conditions, maternal causes, and nutritional deficiencies; non-communicable diseases (NCDs); and injuries [17]. South Africa has a very high HIV prevalence and it is not uncommon that people living with HIV also develop other chronic conditions. Given this HIV burden, information is needed on the prevalence of multimorbidity to plan for more responsive healthcare services. This information is valuable for planning purposes as health service delivery could be made more efficient around common disease clusters to the benefit of those living with multimorbidity. Knowing who is most affected by multimorbidity (i.e. determinants or factors that are common in those affected) can also be used to design interventions to target those individuals. While multimorbidity research has been emerging in the country for the past decade, few studies have reported the prevalence of multimorbidity and factors associated with it in a consistent and comparable manner [18]. The authors conducted a systematic review of multimorbidity prevalence studies in South Africa and found significant heterogeneity in the study designs as well as the estimates of prevalence [18]. Of the studies included [19-27], the prevalence of multimorbidity ranged from 3 to 87%. In addition, the factors associated with multimorbidity were disparate and at times contradictory. Among the factors that were occasionally associated with multimorbidity in South Africa were: age, being female, locality, education level, body mass index (BMI) and marital status.

Prevalence estimates form an important part of the information used for evidence-based health decision-making. Given the lack of studies conducted about multimorbidity in South Africa, we aimed to determine the prevalence of multimorbidity by age group and sex in the country using the 2016 Demographic and Health Surveys (DHS) 6 (SADHS 2016). In addition, this paper reports the process and results derived from a systematic analysis of the SADHS 2016 to establish factors associated with multimorbidity in the South African population. The SADHS 2016 is unique in South Africa in that it is a nationally representative survey which includes biomarkers for the measurement of HIV, HbA1c (diabetes), blood pressure and anaemia status.

Materials and methods

Sample and data source

National survey data is an important source of information about multimorbidity. National surveys represent a largely untapped resource that could shed light on multimorbidity in the general population. This is especially true for LMICs such as South Africa where limited information exists about multimorbidity. The DHS project, primarily funded by the United States Agency for International Development has conducted more than 230 nationally representative comparable household surveys in more than 80 countries since 1984 [28]. The DHS collects data on a range of topics such as fertility, contraception, maternal and child health, HIV, malaria and domestic violence [28]. For many countries, the DHS is an important source of information for policy making, monitoring and evaluation and as the country’s public health evidence base [28]. In terms of multimorbidity, the DHS collects information on self-reported health conditions and biological markers.

This paper presents a secondary analysis of national survey data from the SADHS 2016. The survey is nationally representative with the aim of providing up-to-date estimates of demographic and health indicators such as information on fertility levels, marriage, sexual activity, contraceptive use, nutrition, child mortality, aspects of child health, exposure to the risk of HIV infection, behaviour and health indicators [29]. The SADHS 2016 also collected information on anthropometry, anaemia, hypertension, HbA1c levels and HIV among adults 15 years and older.

The SADHS 2016 followed a stratified two-stage sample design and a total of 750 primary sampling units (PSUs) were selected and stratified by urban, traditional and farm areas. A fixed number of twenty dwelling units were randomly selected in each PSU. Of these dwelling units, sub-sampling occurred whereby half of the households were eligible for a South African-specific module on Adult Health that included the collection of biomarkers [29].

All participants signed consent forms to participate in the study SADHS 2016. For this secondary data analysis, the anonymised dataset with necessary permissions was obtained from the DHS programme. In addition, ethics clearance was granted by the Biomedical Research Ethics Committee of the University of the Western Cape (BM20/5/8) as part of the lead author’s PhD project.

Description of included variables

Multimorbidity is frequently measured by counting the number of co-existing conditions, using a predefined list of medical conditions [30, 31]. Various studies have used this technique when doing secondary data analysis [32]. Estimation of multimorbidity included: self-reported diseases (e.g. bronchitis/ COPD, heart disease, high blood cholesterol, stroke, TB in the last 12 months), biomarker disease (e.g. HIV, anaemia, high blood pressure) and a combination of the two (i.e. diabetes). Disease variables were coded as binary (disease absent ‘0’ or disease present ‘1’). An index variable was created where for each individual, the number of disease conditions present was counted. If there was information about a disease condition missing, this was counted as “no disease present”. The disease index variable was further categorised to create another variable, the Multimorbidity Index. This variable categorised individuals into either having “no multimorbidity” (no disease or only one disease present) or “multimorbidity present” (two diseases or more present).

Self-reported diseases

The study sample consisted of 10 336 youth and adults who completed the Adult Health module and were asked about the presence of several diseases (Table 1) e.g. “Has a doctor, nurse or health worker told you that you have or have had any of the following conditions”. The response to the questions were “No”, “Yes” or “Don’t know”, with the “Don’t know” response recorded as missing values. TB in the last 12 months was constructed from two other variables: whether a participant had ever had the disease and whether they had the disease in the last 12 months or more than 12 months ago.

Table 1

Survey questions on self-reported diseases.

Variable	Survey Question
Diabetes	Has a doctor, nurse or health worker told you that you have or have had any of the following conditions: diabetes or blood sugar?
Emphysema/ Bronchitis/COPD	Has a doctor, nurse or health worker told you that you have or have had any of the following conditions: chronic bronchitis, emphysema, or COPD?
Heart disease	Has a doctor, nurse or health worker told you that you have or have had any of the following conditions: Heart attack or angina/chest pains?
High blood cholesterol	Has a doctor, nurse or health worker told you that you have or have had any of the following conditions: high blood cholesterol or fats in the blood?
Stroke	Has a doctor, nurse or health worker told you that you have or have had any of the following conditions: stroke?
TB in the last 12 months	Has a doctor, nurse or health worker ever told you that you have TB?
	When was the last time you had TB?

The analysis included data where the variables were deemed to be “current” conditions. Disease conditions were excluded for the following reasons: (i) disease conditions that could not be assumed to be current at the time of the survey due to the way that the question was asked (ii) disease conditions that were considered to be acute or of short duration (iii) disabilities or injuries. Two clinicians assisted where the information was unclear. Further details are available in S1 Table in S1 File.

Physically measured diseases (biomarkers)

Of the people included, approximately 74.4% (n = 7 961) of people also had information on physically measured diseases. The following information was of interest to the analysis: diabetes (HbA1c), HIV status (dry blood spot), blood pressure measurements, anaemia (Hb), anthropometry (height and weight). Nurses collected blood specimens from finger pricks.

For diabetes, dry blood spots were analysed using a blood chemistry analyser which measures total haemoglobin concentrations [29]. A participant was assigned diabetic status if their HbA1c ≥ 6.5 mmol [33, 34]. Participants with normal HbA1c values but on medication to manage diabetes were also assigned diabetic status. For participants without HbA1c data, their disease status was based on their self-assessment of whether they had diabetes or not.

For HIV, dry blood spots were tested with an enzyme-linked immunosorbent assay (ELISA) and a second ELISA was done for confirmation [29]. The results of the first ELISA was included in this study.

For anaemia, nurses collected blood samples in a microcuvette and the analysis of haemoglobin was conducted on site. The SADHS 2016 anaemia results were adjusted for smoking status and altitude [29]. Anaemia levels below 7.0 g/dl were considered as severe anaemia. Moderate anaemia was considered levels between 7.0g/dl and 9.9g/dl. For pregnant women, mild anaemia were levels between 10.0 g/dl and 10.9 g/dl and between 10.0 g/dl and 11.9 g/dl for all other adult women [35]. Participants were then categorized either having no anaemia or having anaemia. The degree of anaemia was characterized as mild, moderate or severe.

Three blood pressure measurements were taken from participants using digital blood pressure monitors [29]. For this study, the first measurement was excluded and the average of the remaining repeated measurements were taken. The values were categorised as: hypertension absent (Systolic < 120 mmHg & diastolic < 80 mmHg), Pre-hypertension (Systolic: 120–139 mmHg or diastolic: 80–89 mmHg), Stage 1 Hypertension: (Systolic: 140–159 mmHg or diastolic: 90–99 mmHg), Stage 2 hypertension (Systolic ≥160 mmHg or diastolic ≥100 mmHg) [36]. Hypertension was coded as being absent (normal or pre-hypertension) or present (Stage 1 or Stage 2 hypertension). People on medication to manage hypertension were included in those that had hypertension. Data cleaning for diabetes [37] and hypertension [38] followed the procedures used in the Second South African Comparative Risk Assessment. Further details of data collection, cleaning and coding is listed in S2 Table in S1 File.

Other variables of interest

Systematic reviews identified the following characteristics (among others) as being related to multimorbidity: (i) Biomedical and individual: ageing, female, (ii) Socioeconomic: lower socioeconomic status, high-income group (in low and middle-income countries), lower education, (iii) Social and environmental: living in urban environments (iv) Behavioural: tobacco, overweight and obese [39]. For this study, the following variables were investigated as predictor variables: age category, sex, locality, highest education level, wealth index, employment status, BMI category, current smoker status and current alcohol drinker status.

The ages of participants were taken from DHS 2016 dataset. Participants under the age of 15 years were excluded. Where appropriate, age was analysed in 10-year age bands. The variable sex was included and participants were coded as male or female. Locality was included and coded as either urban or rural. Educational attainment was also included and described by the ‘highest grade or form you completed at that level’. This study divided the responses into three categories: primary school or less, secondary school, and tertiary education. Employment status was coded as employed (currently working) or unemployed.

This study made use of the SADHS 2016 wealth index. The wealth index uses principal component analysis to score households according to the types of goods that are owned and other characteristics [29]. The households were divided into five quintiles, from poorest (Quintile 1) to richest (Quintile 5).

This study also examined current alcohol and tobacco use. For current alcohol use, the responses to the following two questions were combined: “Have you ever consumed a drink that contains alcohol such as beer, wine, ciders, spirits, or sorghum beer?” and “Was this within the last 12 months?”. For tobacco use, the question “Do you currently smoke tobacco every day, some days, or not at all?”. Both variables were coded as binary (e.g. Yes/No).

The BMI of participants were also examined. Height and weight were measured using a digital scale and stadiometer [29]. BMI was calculated using the BMI STATA package. BMI was categorized as follows: underweight (15 .0 - <18.5 kg/m2), normal weight (18.5 - <25.0 kg/m2), overweight (25.0 - <30.0 kg/m2), obesity grade 1 (30.0 - <35.0 kg/m2), obesity grade 2 (35.0 - <40.0 kg/m2), obesity grade 3 (40.0 - <60.0 kg/m2) [40]. Data cleaning was done in accordance to another study [40]. Further details are listed in S2 Table in S1 File.

Analysis

The statistical analysis was done using STATA 15.0 (Stata Corporation, College Station, Texas, USA) software. The STATA survey set (‘svy’) of commands were used to account for the complex survey design. Sampling weights were calibrated against the Statistics South Africa mid-year population estimates [41].

For unweighted data (sample), frequencies were used to display categorical data. Age was analysed as a continuous variable while gender, locality, province, educational level and wealth index were analysed as categorical variables. Bivariate associations between locality, province, highest education level and wealth index by sex were assessed using Chi-square tests. The prevalence of having single disease conditions by sex was also assessed with Chi-square tests. For weighted data, multimorbidity status was described using histograms and box plots against age.

Regression methods were used to describe the relationship between a dependent variable and other predictor variables [42]. In this case, a multivariate logistic regression was employed because the dependent variable was binary (Multimorbidity absent = 0, Multimorbidity present = 1).

Crude odds ratios were estimated by only including the dependent variable and one predictor variable. Three models were constructed for logistic regression with multimorbidity as the dependent variable.Model 1 contained only demographic information (e.g. age and sex), while Model 2 contained sociodemographic information (e.g. age, sex, educational attainment, wealth index and employment status). The final model (Model 3) included all variables in the previous models but also included lifestyle or behavioural factors (e.g. alcohol use, tobacco use and BMI).

Model checking was performed using various statistical tests. The link test [43] was used to determine if there were specification errors. Interaction terms were added where necessary. Influential observations were checked using the Pearson residuals, deviance residuals and Pregibon leverage [44] on the unweighted model as these tests cannot be used on survey weighted data. Influential observations were dropped, and the model was refitted. The crude and adjusted odds ratios were reported with 95% CIs and p-values of less than 0.05 were considered as statistically significant.

Results

Sample description

There were 10 336 youth and adults included in the sample; with more females (59.2%) than males (Table 2). The median age of participants was 36 years (interquartile range: 24–52 years), with females being slightly older than males but this was not statistically significant. More than half of the sample resided in urban areas (55.0%) and most (64.5%) had completed secondary education. The majority of participants were Black African (84.7%), followed by coloured (9.6%), white (4.4%) and Indian/Asian (1.4%). Age, urban location and education did not differ between males and females. There were significant differences between the proportion of males and female participants in the sample, by province (p < 0.001) and wealth quintile (p = 0.018).

Table 2

Description of sample population (unweighted).

	Total (N = 10 336)% (n)	Male (N = 4210)% (n)	Female (N = 6126)% (n)	p-value’
Age* (Median years and IQR)	36 (24–52)	33 (22–49)	37 (25–54)	0.442
Urban location	55.0 (5 685)	55.2 (2 324)	54.86 (3 361)	0.735
Province:				<0.001
• Western Cape	7.29 (754)	6.65 (280)	7.74 (474)
• Eastern Cape	13.08 (1 352)	13.16 (554)	13.03 (798)
• Northern Cape	8.53 (882)	8.38 (353)	8.64 (529)
• Free State	9.97 (1 031)	9.12 (384)	10.56 (647)
• Kwa-Zulu Natal	15.2 (1 571)	14.32 (603)	15.8 (968)
• North West	10.5 (1 085)	11.97 (504)	9.48 (581)
• Gauteng	9.97 (1 031)	11.16 (470)	9.16 (561)
• Mpumalanga	11.8 (1 220)	12.23 (515)	11.51 (705)
• Limpopo	13.64 (1 410)	12.99 (547)	14.09 (863)
Education level				0.502
• Primary or less	26.26 (2 714)	25.65 (1 080)	26.67 (1 634)
• Secondary complete	64.51 (6 668)	65.11 (2 741)	64.1 (3 927)
• Tertiary	9.23 (954)	9.24 (389)	9.22 (565)
Wealth index				0.018
• Quintile 1 (Poorest)	20.3 (2 098)	20.45 (861)	20.19 (1 237)
• Quintile 2 (Poorer)	21.55 (2 227)	22.71 (956)	20.75 (1 271)
• Quintile 3 (Middle)	22.61 (2 337)	23.06 (971)	22.3 (1 366)
• Quintile 4 (Richer)	19.99 (2 066)	18.74 (789)	20.85 (1 277)
• Quintile 5 (Richest)	15.56 (1 608)	15.04 (633)	15.92 (975)
Employed	33.9 (3 506)	41.6 (1 751)	28.7 (1 755)	<0.001

*Age in years. ‘Categorical variables were tested using Chi-squared, continuous variables tested using Wilcoxon signed rank test.

All self-reported disease conditions were more common in females compared to males (Table 3). Females had a slightly higher prevalence of TB in the last 12 months compared to males, however, the difference was not statistically significant. Other than hypertension, all physically measured disease conditions were significantly more common in females than males. The prevalence of multimorbidity in the sample population was 22.9% (S3 Table in S1 File).

Table 3

Prevalence of single disease conditions by sex and method of measurement in South Africa for 2016 (unweighted data).

Disease condition	Total %(n/N)	Male %(n/N)	Female %(n/N)	p-value
SELF-REPORTED
Diabetes	4.5(459/10 292)	3.6(150/4 176)	5.1(309/6 116)	<0.005
Bronchitis/COPD	1.3(132/10 290)	1.0(40/4 177)	1.5(92/6 113)	<0.005
Heart disease	3.4(354/10 305)	2.4(101/4 183)	4.1(253/6 122)	0.015
Cholesterol	2.9(296/10 282)	2.3(94/4 167)	3.3(202/6115)	0.002
Stroke	1.4(146/10 309)	1.0(40/4 186)	1.7(106/6123)	0.001
TB in last 12 months	1.3(138/10 336)	1.3(53/4 210)	1.4(85/6126)	0.576
PHYSICALLY MEASURED (BIOMARKER)
HIV	19.9(1 307/6 584)	13.8(346/2 517)	23.6(961/4 067)	<0.005
Hypertension	46.2(3 678/7 961)	45.1(1 412/3 130)	46.9(2 266/4 831)	0.117
Anaemia	25.9(1 862/7 200)	17.7(489/2 769)	31.0(1 373/4 431)	<0.001
Diabetes (HbA1c)	12.4(839/6 763)	9.3(241/2 591)	14.3(598/4 172)	<0.001
PHYSICALLY MEASURED (BIOMARKER) AND SELF-REPORTED
Diabetes (self-report or HbA1c)	10.06(1 036/10 295)	7.35(307/4 178)	11.92(729/6 117)	<0.001

Prevalence of single diseases and multimorbidity

Table 4 shows the weighted prevalence of each included disease condition by sex. Of the self-reported diseases, diabetes had the highest prevalence (4.4%), followed by high cholesterol (3.5%), heart disease (3.1%), COPD or bronchitis (1.4%), stroke (1.4%) and TB in the last 12 months (1.2%). The prevalence of physically measured diseases was higher than that of self-reported diseases. Of the physically measured diseases, hypertension occurred most frequently (45.0%), followed by anaemia (24.7%), HIV (19.6%) and diabetes (11.7%).

Table 4

Prevalence of single disease conditions by sex and method of measurement in South Africa for 2016 (weighted data).

Disease condition	Total %(95% CI)	Male %(95% CI)	Female %(95% CI)
SELF-REPORTED
Diabetes	4.4(3.9–5.0)	3.7(3.0–4.5)	5.1(4.4–5.9)
Bronchitis/COPD	1.4(1.1–1.8)	1.1(0.8–1.6)	1.7(1.3–2.2)
Heart disease	3.1(2.7–3.5)	2.3(1.8–2.9)	3.8(3.3–4.5)
Cholesterol	3.5(2.9–4.2)	3.0(2.3–3.8)	4.1(3.4–4.9)
Stroke	1.4(1.1–1.7)	1.0(0.7–1.5)	1.7(1.4–2.1)
TB in last 12 months	1.2(0.9–1.5)	0.9(0.6–1.3)	1.5(1.0–2.0)
MEASURED (BIOMARKER)
HIV	19.6(18.2–21.1)	13.7(11.8–15.8)	24.5(22.7–26.4)
Hypertension	45.0(43.1–46.9)	44.1(41.5–46.7)	45.8(43.7–48.0)
Anaemia	24.7(23.2–26.3)	16.8(15.0–18.8)	31.3(29.2–33.5)
Diabetes (HbA1c)	11.7(10.7–12.8)	9.1(7.8–10.6)	13.9(12.6–15.4)
MEASURED (BIOMARKER) AND SELF-REPORTED
Diabetes (self-report or HbA1c)	9.1(8.4–9.9)	6.9(5.9–7.9)	11.1(10.2–12.3)

Note: Biomarker prevalence differed slightly from the DHS report due to the different data cleaning methods and cut-offs employed.

Diabetes was the only disease condition included in this study that was both physically measured and self-reported in the questionnaire. The prevalence of physically measured diabetes was more than double that of self-reported diabetes (11.7% versus 4.4%, respectively). This indicates that self-reported diabetes is most likely underreported. When combining the responses of the measured and self-reported diabetes, the composite prevalence was 9.1%. All diseases were more prevalent in females compared to males.

The number of diseases present in one individual ranged from zero to six. The difference in the prevalence in the number of diseases by sex was statistically significant (p<0.001). About 49% of the participants had none of the diseases included in the study, with more males compared to females being “disease-free” (55.8% versus 41.8%, respectively, p<0.001) (Table 5). Less than a third of participants (30.8%) had one disease and there was a difference between males and females (29.4% versus 32.1%, respectively, p = 0.0183). Multimorbidity was present in 21% of participants. The prevalence of multimorbidity in females was almost double that of males (26.2% vs 14.8%, respectively) and the difference between the sexes was statistically significant (p<0.001).

Table 5

Number of diseases in individuals by sex in South Africa for 2016 (weighted data).

Number of diseases	Total %(95% CI)	Male %(95% CI)	Female %(95% CI)
No disease	48.6(47.0–50.1)	55.8(53.5–58.1)	41.8(40.0–43.4)
1 disease	30.8(29.5–32.0)	29.4(27.5–31.2)	32.1(30.6–33.7)
2 diseases	14.1(13.2–15.1)	10.5(9.4–11.8)	17.4(16.2–18.7)
3 diseases	5.2(4.7–5.9)	3.5(2.8–4.3)	6.8(6.0–7.8)
4 diseases	1.1(0.8–1.3)	0.6(0.4–0.9)	1.4(1.1–1.9)
5 diseases	0.2(0.1–0.4)	0.1(0.1–0.3)	0.4(0.2–0.6)
6 diseases	0.07(0.02–0.19)	0.01(0.01–0.04)	0.01(0.01–0.02)
Multimorbidity (≥ 2 diseases)	20.7 (19.5–21.9)	14.8 (13.4–16.3)	26.2 (24.7–27.7)

Multimorbidity prevalence increased with increasing age in both males and females (Fig 1, S3 Table in S1 File). The prevalence of multimorbidity was consistently higher in females compared to males across the different age groups. Multimorbidity was present at lower levels: 3% for adolescents aged 15–19 years and 10% for 20–29 years old. In females, multimorbidity peaked at 47% in the 60–69 years old; whereas in males, it peaked at 40% in the 70–79 years old. Multimorbidity prevalence dropped slightly in the age group 80 years and over. However, the observed drop is most likely due to uncertainty introduced by a smaller number of people aged 80+ being included in the sample. People with multimorbidity tended to have an older median age compared to those with no multimorbidity (Fig 2), but this difference was not statistically significant.

Fig 1

Estimated multimorbidity prevalence by age group and sex in South Africa in 2016.

Fig 2

Multimorbidity status by age.

Factors associated with multimorbidity

The factors associated with multimorbidity were investigated through a logistic regression (Table 6). For the final model, outliers were dropped and the model was refitted due to its limited ability in predicting multimorbidity in young women with a low BMI (S1 Fig in S1 File). An interaction term between age and sex was added to the model to improve its fitness.

Table 6

Factors associated with multimorbidity.

Variable	UnadjustedOdds ratios(95% CI)	Final model (Model 3)Odds ratio(95% CI)
Age category (Reference: 15–24 year)
• 25–34 years	2.982 (2.407–3.695) *	3.923 (2.299–6.695) *
• 35–44 years	4.861 (3.769–6.269) *	8.417 (5.101–13.890) *
• 45–54 years	7.527 (5.844–9.694) *	14.165 (8.654–23.185) *
• 55–64 years	11.764 (8.837–15.662) *	24.910 (14.901–41.641) *
• 65+ years	14.181 (10.951–18.364) *	23.062 (13.719–38.766) *
Sex (Reference: Male)	2.038 (1.804–2.301) *	1.135 (0.831–1.551)
Age category and sex interaction
• 15–24#Female	-	2.734 (1.498–4.988) *
• 25–34#Female	-	1.896 (1.169–3.075) *
• 35–44#Female	-	1.340 (0.842–2.132)
• 45–54#Female	-	1.089 (0.676–1.755)
• 55–64#Female	-	0.866 (0.558–1.345)
• 65+#Female	-	1 (omitted)
Urban (Reference: Rural)	0.817 (0.721–0.925) *	1.107 (0.901–1.360)
Education (Reference: Primary)
• Secondary	0.423 (0.372–0.480) *	0.966 (0.819–1.140)
• Tertiary	0.323 (0.251–0.414) *	0.722 (0.537–0.970)*
Wealth index (Reference: Poorest)
• Poorer	0.995 (0.829–1.194)	1.067 (0.864–1.317)
• Middle	1.076 (0.874–1.324)	1.126 (0.867–1.464)
• Richer	1.036 (0.845–1.270)	1.034 (0.778–1.374)
• Richest	0.901 (0.713–1.138)	0.754 (0.545–1.044)
Employed (Reference: Not employed)	0.744 (0.643–0.861) *	0.813 (0.675–0.979)*
BMI (Reference: Underweight)
• Normal weight	0.961 (0.679–1.361)	0.892 (0.609–1.309)
• Overweight	1.779 (1.227–2.581) *	1.033 (0.680–1.571)
• Obesity group 1	2.536 (1.759–3.655) *	1.213 (0.793–1.854)
• Obesity group 2	2.94 (1.965–4.397) *	1.340 (0.840–2.137)
• Obesity group 3	3.518 (2.367–5.228) *	1.527 (0.956–2.438)
Current alcohol use (Reference: No current alcohol use)	0.571 (0.498–0.653) *	0.815 (0.686–0.968) *
Current tobacco use (Reference: No current tobacco use)	0.704 (0.592–0.838) *	0.893 (0.710–1.122)

The final model showed that the odds of being multimorbid increased with age, with the highest odds if the participant was in the 55–64 years of age group (OR: 24.910, 95% CI: 14.901–41.641, p<0.001), compared to 15–24 years old. Younger females (15–34 years) had larger odds of being multimorbid compared to males in the same age groups.

The odds of being multimorbid were reduced if an individual had tertiary education (OR: 0.722, 95% CI: 0.537–0.970, p = 0.031) compared to only having completed primary school education. Those that were employed had reduced odds of multimorbidity compared to those that were unemployed (OR: 0.813, 95% CI: 0.675–0.979, p = 0.029). Those that had used alcohol recently also reported lowered odds compared to those that were not using alcohol (OR: 0.815, 95% CI: 0.686–0.968, p = 0.02). BMI and current tobacco use were not significant when adjusted for age, sex and other variables. The wealth index was not a predictor of multimorbidity. Additional models can be found in S5 Table in S1 File.

Discussion

Using the DHS national survey, it was found that one in five South Africans aged 15 years or above was multimorbid. The prevalence of multimorbidity generally increased with age and reached 42% in the 60 years and older age groups. The prevalence of multimorbidity was higher in females compared to males, but the difference was larger in younger age groups. Our study corroborates other studies that have found high levels of chronic diseases in the sub-Saharan region. For example, an analysis of DHS surveys in 33 sub-Saharan African countries (excluding South Africa), found that there was a high prevalence of hypertension, anaemia, underweight, overweight and obesity in females 15 years or above [45].

Several other national surveys have been analysed to determine the prevalence of multimorbidity in South Africa [19–21, 46]. The prevalence estimates varied from 2.8% [21] to 63.4% [20], although these studies looked at differing age groups, used varying data collection methods and included different disease conditions. The 2003 World Health Survey which surveyed adults 18 years older found a standardised prevalence of 11.2% [19]. Two waves of the National Income Dynamic Surveys (2008 and 2012) found a low prevalence of 2.7% and 2.8%, respectively [21]. The same 2008 dataset was analysed using different methods but found a similar low prevalence of 4% [46]. Garin et al. [20] used the 2007/2008 World Health Organization Study on global AGEing (SAGE) and adult health and found a prevalence of 63.4% in adults over the age of 50 years. Most of these studies included self-reported diseases and physically measured hypertension. Self-reported diseases are likely to be underreported as people may be unaware that they have a disease. The DHS physically measured more diseases compared to the National Income Dynamic Surveys and World Health Survey (i.e. HIV, diabetes and anaemia) which may explain its ability to detect higher levels of multimorbidity. The number of disease conditions included in each study also varied. The Garin et al. [20] study was restricted to adults over the age of 50 years and included a larger number of disease conditions (e.g. depression, cognitive impairment, edentulism and obesity as a disease condition) and therefore reported higher prevalence of multimorbidity. A recent analysis of Wave 2 SAGE (2014/2015) [47] of adults aged 45 years and older, included fewer disease conditions than Garin et al. [20] (7 vs. 12); reported a multimorbidity prevalence of 21%. Another discrepancy to note is that the 2016 DHS was more recently conducted than the other national surveys.

In terms of factors associated with multimorbidity, an increasing age was associated with being multimorbid. This follows trends in reporting in international [39] and the South African literature on multimorbidity. Multimorbidity is often associated with older adults, especially in high income countries [48] due to shifting demographic trends whereby people are living longer, ageing and developing chronic diseases of lifestyle. However, our study had an interesting finding in observing that multimorbidity was present in 10% of young adults between the ages of 20–29 years. This is most likely attributed to the high prevalence of single disease such as HIV, anaemia and hypertension in South Africa. HIV is known to affect younger adults in South Africa. A South African national HIV prevalence survey indicated that 7.9% of people (4.8% of males, 10.9% of females) aged 15–24 years were HIV positive in 2017 [49]. Also, the prevalence of hypertension in South Africa is thought to be increasing. In young South Africans, hypertension is frequently associated with having a family history of the disease (suggesting a genetic component) and obesity or metabolic syndrome [50, 51]. In this study, approximately 32% of people with HIV under 30 years of age, also had hypertension (S3 Fig in S1 File). This has implications for young people in that they will have to be on lifelong treatment for both diseases.

The present study showed that having tertiary education decreased the odds of multimorbidity, this has been noted both locally and internationally [19, 20, 52]. However, a systematic review of education levels and multimorbidity in Southeast Asia found the association was inconsistent [53]. This study found that being employed decreased the odds of multimorbidity. Similar results were found in an analysis of social determinants and multimorbidity in South Africa [46]. Yet, this could also be interpreted to mean that healthier people are more likely to be employed. In a systematic review of multimorbidity and its impact on workers, multimorbidity was found to have a negative impact on work, worsening absenteeism and lowering employability [54]. The wealth index was not significantly associated with multimorbidity. The relationship between wealth and multimorbidity in this study may be unclear as the diseases included may have different patterns according to the individual disease. For example, HIV could be associated with being in a lower wealth quintile, while cardiovascular diseases such as diabetes could be associated with being in a higher wealth quintile. The same argument could be used to explain the findings on BMI. This study indicated that having a high BMI could be associated with multimorbidity but the findings were not significant. High BMI has been identified as associated with multimorbidity in other studies [55]. However, the inclusion of HIV and anaemia could mean that people with lower BMIs were also prone to being multimorbid. An interesting finding was that alcohol use was associated with decreased odds of multimorbidity. A study of binge drinking among adults in the United States found that binge drinkers tended to have lower levels of multimorbidity [56]. They related these findings to the ‘sick quitter’ hypothesis whereby adults stop drinking due to interactions with prescribed medications [57].

Limitations

The current analysis was limited to the data available and disease conditions asked about in the original survey. Additional disease conditions (e.g. cancer) could have been included in the analysis, but the survey in question only asked if the individual had “ever” had the disease. A strength of this study is that disease conditions were only included if the person could have been considered to have the disease at the time of the survey or at a time close to when the study took place. Many studies of multimorbidity include past and current disease conditions without distinction. Had there been included more disease conditions, the prevalence estimates would have most likely be higher. Also, we did not account for pregnancy status in our calculation of BMI.

The study is limited to a simple count of diseases to determine multimorbidity. Studies done with electronic health records or surveyed people specifically for multimorbidity may have taken the severity of diseases into account. Nonetheless, the DHS provides a robust source of data that could be analysed in other LMICs to generate information about multimorbidity where little is still known.

The analysis included self-reported and measured (biomarker) diseases. Self-reported diseases may have been underreported due to participants being unaware that they have a disease. In this study, the prevalence of measured (biomarker) diseases was higher than self-reported diseases. In addition, this study was cross-sectional by nature meaning that we cannot confer temporality.

Conclusion

This study showed that one in five South Africans, 15 years or above, are managing more than one disease condition. Multimorbidity started in adolescents and increased with age. Females were more frequently affected than males. It was found that having tertiary education and being employed lowered the odds of multimorbidity.

The high prevalence of multimorbidity needs to be addressed in South Africa. This could be done in a twofold manner: (a) by reducing the high prevalence of single diseases such as hypertension and (b) by simultaneously targeting people with existing diseases to reduce their chances of becoming multimorbid. More studies are needed to identify common disease clusters to assist in the endeavour of targeting high risk people. More studies are also needed to determine whether the trends in multimorbidity are changing in the country. For example, to understand whether policies aimed at diseases such as HIV and hypertension have helped to decrease multimorbidity. Also, information is needed on how emerging diseases such as COVID-19 may affect people with multimorbidity in South Africa.

(DOCX)

Click here for additional data file.

18 Feb 2022

Submitted filename: One in five_SADHS2016_manuscript_Edited_Kassa_Dec2021.docx

Click here for additional data file.

25 Feb 2022

Manuscript ID PONE-D-21-34287

Response to Reviewers

Dear Dr Pathak,

Thank you for providing us with the opportunity to submit a revised draft of the manuscript “One in five South Africans are multimorbid: An analysis of the 2016 Demographic and Health Survey” to PLOS ONE. We appreciate the time and effort that you and the reviewers have given us, and the chance to revise, improve and strengthen our manuscript.

Additional text was added to the Introduction and Discussion sections. The Methods section was restructured as Reviewer 2 suggested. Various tracked changes made by Dr Abdi were accepted. We have marked the changes within the manuscript using the tracked change function and insertions were marked in blue highlight. Please see our responses to reviewer comments below.

Comment Response Page & line number

Editor comments (Dr Mona Pathak)

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. Document was checked and meets the standards described in the attachments. N/A.

2. Please provide additional details regarding participant consent. In the ethics statement in the Methods and online submission information, please ensure that you have specified what type you obtained (for instance, written or verbal, and if verbal, how it was documented and witnessed).

If your study included minors, state whether you obtained consent from parents or guardians. If the need for consent was waived by the ethics committee, please include this information. Additional information was added to the Methods section and the section on Patient consent for publication:

“All participants signed consent forms to participate in the SADHS 2016. For this secondary data analysis, the anonymised dataset with necessary permissions was obtained from the DHS programme. In addition, ethics clearance was granted by the Biomedical Science Research Ethics Committee of the University of the Western Cape (BM20/5/8) as part of the lead author’s PhD project.”

Page 6, Line 118 – 122.

3. Please review your reference list to ensure that it is complete and correct.

If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. New references were added as indicated in the response to reviewers (below).

References were checked for completeness and errors. No errors were found.

No retracted papers are cited. N/A.

Reviewer 1 (Dr Kassa Demissie Abdi)

4. Reviewer #1: The manuscript is

- technically sound and contributing some information to the global scientific audience,

-used appropriate statistical analysis,

-presented relevant data, and

-written in standard English We thank Dr Kassa Demissie Abdi for the very helpful review, edits and comments made. N/A.

5. How data on the age, sex, residences, educational level and employment status of participants were synthesized from SADHS (2016)? Done. Details on age, sex, locality and employment status were added to the Methods section.

“The ages of participants were taken from DHS 2016 dataset. Participants under the age of 15 years were excluded. Where appropriate, age was analysed in 10-year age bands. The variable sex was included and participants were coded as male or female. Locality was included and coded as either urban or rural. Educational attainment was also included and described by the ‘highest grade or form you completed at that level’. This study divided the responses into three categories: primary school or less, secondary school, and tertiary education. Employment status was coded as employed (currently working) or unemployed.” Page 10, Line 189 – 195.

6. Better to use ≥ 6.5 mmol instead of >=6.5 mmol Done. Symbol inserted. Page 8, Line 151.

7. The results of the first ELISA was included in this study. Done. Wording changed to suggested wording. Page 9, Line 157-158.

8. Participants were then categorized either having no anaemia or having anaemia. The degree of anaemia was characterized as mild, moderate or severe. Done. Wording changed to suggested wording. Page 9, Line 164 – 166.

9. The first measurement was excluded and the average of the remaining repeated measurements were taken. Done. Wording changed to suggested wording. Page 9, Line 169 – 170.

10. Further details of data collection, cleaning and coding is listed in S2 Table Done. Wording changed to suggested wording. Page 9, Line 178-179.

11. Is it to mean variables? Done. Corrected to “Variables”. Page 15, Line 281.

12. Is it medically sound to combine them? Blood was only taken for a sub-sample of the participants - 6 763 participants provided HbA1c measurements whereas 10 292 participants provided information on self-reported diabetes status.

Instead of setting the diabetes information as missing for 3 259 of the participants that did not have HbA1c information, we combined the results with the self-reported diabetes data. We believe it was appropriate for a study of multimorbidity where it would be important to reduce the amount of missing information in the sample.

N/A.

13. Used. Alcohol use would have been expected to predict multimorbidity. Please, check for collinearity or confounding? We checked our model for collinearity and no problems were detected.

The model was adjusted for a series of important potential confounders, however, residual confounding is possible. Extra text was added to the Discussion to explain the concept of the ‘sick quitter’ hypothesis:

“An interesting finding was that alcohol use was associated with decreased odds of multimorbidity. A study of binge drinking among adults in the United States found that binge drinkers tended to have lower levels of multimorbidity [56]. They related these findings to the ‘sick quitter’ hypothesis whereby adults stop drinking due to interactions with prescribed medications [57].” Page 24, Line 419 – 423.

14. Make it clear i.e. higher or lower? Done. Sentence changed to:

“The prevalence of multimorbidity was higher in females compared to males, but the difference was larger in younger age groups” Page 22, Line 360.

Reviewer 2

15. The issue on multimorbidity is an important health challenge particularly in sub-Saharan Africa due to the associated consequences coupled with the sub-region’s weak health system. Therefore, I commend the authors for researching on the issue. The manuscript is well written and can be accepted for possible publication after addressing the following comments: We thank the reviewer for the positive comments. N/A.

16. The statement of the problem is not too solid.

The authors failed to point out clearly the actual limitations of existing studies.

What is the actual contribution of the present study to the literature? Is it the use of the national survey?

17. It is also important to let the reader know about few examples of existing studies on multimorbidity: highlighting on their focus and associated factors of multimorbidity. We thank the reviewer for the suggestions. We inserted a few sentences that both a) mentions existing studies and their problems b) more clearly states why we did the analysis and what the contribution of the study is.

“While multimorbidity research has been emerging in the country for the past decade, few studies have reported the prevalence of multimorbidity and factors associated with it in a consistent and comparable manner [18]. The authors conducted a systematic review of multimorbidity prevalence studies in South Africa and found significant heterogeneity in the study designs as well as the estimates of prevalence [18]. Of the studies included [19-27], the prevalence of multimorbidity ranged from 3 to 87%. In addition, the factors associated with multimorbidity were disparate and at times contradictory. Among the factors that were occasionally associated with multimorbidity in South Africa were: age, being female, locality, education level, body mass index (BMI) and marital status.”

“Prevalence estimates form an important part of the information used for evidence-based health decision-making. Given the lack of studies conducted about multimorbidity in South Africa, we aimed to determine the prevalence of multimorbidity by age group and sex in the country using the 2016 Demographic and Health Surveys (SADHS 2016). In addition, this paper reports the process and results derived from a systematic analysis of the SADHS 2016 to establish factors associated with multimorbidity in the South African population. The SADHS 2016 is unique in South Africa in that it is a nationally representative survey which includes biomarkers for the measurement of HIV, HbA1c (diabetes), blood pressure and anaemia status.” Page 2, Line 65 – 71.

Page 2, Line 73-81

18. The paragraphs on DHS in the introduction section (Lines 68 -80) should be moved to the methods section. Done. Paragraph was moved to Methods section. Page 4, Line 84 – 94.

19. The literature on the discussion of multimorbidity prevalence is quite lengthy. However, the authors did not relate the finding to other sub-Saharan African countries, particularly studies that have used the DHS.

While comparison with studies like SAGE and World Health Survey is good, the authors have to be mindful of the years in which these surveys were conducted. Done. In addition, we included the following:

“Our study corroborates other studies that have found high levels of chronic diseases in the sub-Saharan region. For example, an analysis of DHS surveys in 33 sub-Saharan African countries (excluding South Africa), found that there was a high prevalence of hypertension, anaemia, underweight, overweight and obesity [45].”

We noted that the years of the surveys included differed.

“Another discrepancy to note is that the 2016 DHS was more recently conducted than the other national surveys.” Page 22, Line 361 – 364.

Page 22, Line 386.

20. The authors failed to discuss some important findings of the study including the association between educational attainment, employment status and current alcohol use and multimorbidity. It is important to relate the findings to the literature and provide possible explanation as well. This section was expanded upon:

“The present study showed that having tertiary education decreased the odds of multimorbidity, this has been noted both locally and internationally [19,20,52]. However, a systematic review of education levels and multimorbidity in Southeast Asia found the association was inconsistent [53]. This study found that being employed decreased the odds of multimorbidity. Similar results were found in an analysis of social determinants and multimorbidity in South Africa [46]. Yet, this could also be interpreted to mean that healthier people are more likely to be employed. In a systematic review of multimorbidity and its impact on workers, multimorbidity was found to have a negative impact on work, worsening absenteeism and lowering employability [54]. The wealth index was not significantly associated with multimorbidity. The relationship between wealth and multimorbidity in this study may be unclear as the diseases included may have different patterns according to the individual disease. For example, HIV could be associated with being in a lower wealth quintile, while cardiovascular diseases such as diabetes could be associated with being in a higher wealth quintile. The same argument could be used to explain the findings on BMI. This study indicated that having a high BMI could be associated with multimorbidity but the findings were not significant. High BMI has been identified as associated with multimorbidity in other studies [55]. However, the inclusion of HIV and anaemia could mean that people with lower BMIs were also prone to being multimorbid. An interesting finding was that alcohol use was associated with decreased odds of multimorbidity. A study of binge drinking among adults in the United States found that binge drinkers tended to have lower levels of multimorbidity. They related these findings to the ‘sick quitter’ hypothesis whereby adults stop drinking due to interactions with prescribed medications [56].”

Page 24, Line 402 - 423

21. The methods section needs to be re-written and restructured for easy understanding. For example, under description of included variables, it is important to let the reader know what the outcome variable is, before explaining how it was measured. It is confusing to read paragraphs of sentences before knowing about the outcome variable.

Lines 168- 180 should come immediately after the title ‘description of included variables’.

Lines 182-187 should be moved to ‘Analysis’. Likewise, lines 190 -194 should be moved to ‘Analysis’ and Lines 202 – 213 should proceed line 194

Done. We thank the reviewer for the excellent suggestions. The text has been restructured as the reviewer suggested. Methods section.

22. Line 48 – delete through preceding having Done. Page 3, Line 46.

23. Lines 195-201 should also be summarized under ‘other variables of interest’ Done. Paragraph moved. Page 10, Line 181

24. What is the justification for measuring diabetes using both self reported and physical measurement?

Why didn’t you adopt the same procedure for other diseases such as hypertension? Blood was only taken for a sub-sample of the participants - 6 763 participants provided HbA1c measurements whereas 10 292 participants provided information on self-reported diabetes status.

Instead of setting the diabetes information as missing for 3 259 of the participants that did not have HbA1c information, we combined the results with the self-reported diabetes data. We believe it was appropriate for a study of multimorbidity where it would be important to reduce the amount of missing information in the sample.

The same procedure was not used for hypertension because we did not include self-reported hypertension in our study. Two clinicians independently rated which self-reported diseases were suitable to include in the analysis and they assessed that self-reported hypertension was not suitable due to the way in which the question was asked e.g. “Have you ever been diagnosed with hypertension?”. In their assessment, hypertensive status can be changed through diet and it would be inappropriate to conclude that if a person was once diagnosed with hypertension that they could be assumed to be hypertensive. In addition, the proportion of missing data for the hypertension biomarker was lower than that of HbA1c. N/A.

25. Line 121- rephrase the sentence “self-reported diabetes information was combined for participants without biomarker data” for clarity Done. Sentence changed:

“For participants without HbA1c data, their disease status was based on their self-assessment of whether they had diabetes or not.” Page 8, Line 152 -153.

26. Line 156, explain what coding the variables as binary implies. Ever or never? Yes or no? Done. Text added:

Both variables were coded as binary (e.g., Yes/No).

Page 11, Line 207.

27. Line 157- regarding the measurement of BMI, was the BMI of pregnant women taken? If yes why, if no, why not? The BMI of pregnant women was not adjusted for. Only 2.3% (n=117) of the women that had BMI measurements were pregnant, and of those 71.8% were normal weight and overweight.

We have added this as a limitation in the Discussion.

“Also, we did not account for pregnancy status in our calculation of BMI.” Page 25, Line 433.

28. Line 171-reference the sentence “Various studies have used this technique when doing secondary data analysis”. Done. A systematic review of systematic reviews was cited.

Johnston MC, Crilly M, Black C, Prescott GJ, Mercer SW. Defining and measuring multimorbidity: a systematic review of systematic reviews. European journal of public health. 2019 Feb 1;29(1):182-9. Page 6, Line 32.

29. Line 216- the use of “over” implies individuals 15 years were excluded from the study. Rephrase the sentence. Done. Sentence changed to:

“There were 10 336 youth and adults included in the sample…” Page 14, Line 271.

30. Line 220- insert had before completed Done. The word “had” was inserted. Page 14, Line 274

31. Line 222- delete “and” before did Done. Wording changed as suggested. Page 14

32. Line 223- clarify the sentence “there were significant difference in participation between males and females by province”. Participation in what? Done. The wording was changed to clarify that there was a difference in the proportion between males and females by province.

“There were significant differences between the proportion of males and female participants in the sample, by province (p < 0.001) and wealth quintile (p = 0.018)” Page 14, Line 278.

33. Line 277-delete the sentence Unclear. Line 277 was the title for Figure 2. N/A.

34. Line 300- the findings indicate that it peaked at aged 55-64 years. Hence kindly rephrase the sentence Fig.1 shows that multimorbidity prevalence was 42% in the 60 years and older age groups. The sentence was changed to:

“The prevalence of multimorbidity generally increased with age and reached 42% in the 60 years and older age groups…”

In addition, a sentence in the abstract was updated:

“Multimorbidity increased with age; with the highest odds in the 55 - 64 years old age group…” Page 22, Line 359.

Page 2, Line 34.

35. Line 334-335: “In our study, approximately 44% of people with HIV under 30 years of age, also had hypertension”. Which Table or figure indicates it. Thank you for picking this up. We added “data not shown.” Page 24, Line 400.

Submitted filename: Response to reviewer comments_final.docx

Click here for additional data file.

23 Mar 2022

24 Mar 2022

Manuscript ID PONE-D-21-34287R1

Response to Reviewers

Dear Dr Pathak,

Thank you for once again providing us with the opportunity to submit a revised draft of the manuscript “One in five South Africans are multimorbid: An analysis of the 2016 Demographic and Health Survey” to PLOS ONE.

We thank Reviewer 2 for the thorough review and for suggesting that data be added to the supplementary appendix. We have added an extra graph to the appendix and applied all the changes the reviewer suggested. Please see our responses to the reviewer comments below.

Comment

Editor comments (Dr Mona Pathak)

1. Please review your reference list to ensure that it is complete and correct.

If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Reference: References were checked for completeness and errors.

Reviewer 2

1. Line 246: delete the sub-title factors associated with multimorbidity

Response: Done. The sub-title was deleted. Line 221, Page 12.

2. Line 267: Cut the sentence "Three models were constructed for logistic regression with multimorbidity as the dependent variable and paste it at line 257 before Model 1

Response: Done. The sentence was moved to where the reviewer suggested. Line 227-228, Page 12.

3. Line 284: the sentence whiles females had a slightly higher prevalence of TB in the last 12 months, though...... is not complete. Kindly check and complete it.

Response: Done. The sentence was changed to:

Females had a slightly higher prevalence of TB in the last 12 months compared to males, however, the difference was not statistically significant. Line 256 – 257, Page 14.

4. Line 309: delete "disease" after six

Response: Done. Word was deleted.

Line 282, Page 16.

5.Line 357: change over the age 15 years to aged 15years or above

Response: Done. Wording changed as suggested. Line 330-331, Page 20.

6. Line 362-364: which group of people did the study focused on? adult men, women or both?

Response: Added text “in females 15 years or above”. Line 337, Page 21.

7. Line 333: if the data is available, kindly show it as an appendix to erase doubts among readers

Response: We thank the reviewer for insisting on adding the data. Upon relooking at the data, the quoted prevalence was for the unweighted whole population and not for those under 30 years. We have added the corrected weighted prevalence for those under 30 years (32%) and added a supporting figure in the supplementary appendix (Fig S3).

Line 373, Page 22. Figure S3 in appendix added.

8.Line 440: change over the age 15 years to 15 years or above

Response: Done. Wording changed as suggested. Line 420, Page 24.

Submitted filename: Response to reviewer comments_2nd revision.docx

Click here for additional data file.

16 May 2022

One in five South Africans are multimorbid: An analysis of the 2016 Demographic and Health Survey

PONE-D-21-34287R2

Dear Dr. Roomaney,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Carla Pegoraro

Division Editor

PLOS ONE

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

18 May 2022

PONE-D-21-34287R2

One in five South Africans are multimorbid: An analysis of the 2016 Demographic and Health Survey

Dear Dr. Roomaney:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr Carla Pegoraro

Staff Editor

PLOS ONE