Cohort Profile: Health and Ageing in Africa: A Longitudinal Study of an INDEPTH Community in South Africa (HAALSI)

Why was the Cohort set up?

Population ageing is a global phenomenon. The United Nations estimates that the world population aged over 60 will have increased 3-fold from 1950 to 2050, to reach 21% of the population.¹ This compositional shift is happening fastest in low- and middle-income countries (LMIC).² South Africa in particular is undergoing a dramatic demographic and epidemiological transition, and little is known about the socioeconomic determinants or consequences of transition. This study, following important findings in previous studies in Agincourt^3–6 and South Africa in general,^7–9 is set up to inform us about morbidity, mortality and aetiological factors shaping these trends. Various ageing studies, including the Studies on Global Ageing and Adult Health (SAGE) and the 2015 Global Burden of Disease, found that non-communicable diseases, driven mainly by population growth and ageing, have become leading causes of death and disability globally, including in LMIC such as South Africa.^10–14 At the same time, the share of the population 60 and above in South Africa is estimated to increase from 7.8% in 2012 to 14.8% in 2050,¹⁵ and the population aged 50 and over living with HIV will triple by 2040.¹⁶ We established the cohort ‘Health and Ageing in Africa: A Longitudinal Study of an INDEPTH Community’ in South Africa (HAALSI) in the INDEPTH Health and Demographic Surveillance System (HDSS) site of Agincourt, as a harmonized sister study to the Health and Retirement Study (HRS) in the USA¹⁷ and other similar studies worldwide, including ELSA in the UK,¹⁸ TILDA in Ireland,¹⁹ SHARE in Europe,²⁰ CHARLS in China²¹ and LASI in India.²² We aim to describe biological, social and economic determinants and consequences of health and ageing in rural South Africa, as well as to build capacity to explore cross-country differences in risk factors for health and well-being.

What does the study cover?

HAALSI is an interdisciplinary study aiming to longitudinally monitor social, economic and biological risks for chronic health conditions, whether infectious or non-infectious, in a random sample of adults in Agincourt, South Africa. HAALSI focuses on cardiovascular disease, HIV, cognitive functioning and dementia, as these are of special interest in South Africa as it undergoes profound epidemiological and demographic transitions.

Who is in the cohort?

HAALSI was created to establish a population-based longitudinal cohort of men and women aged 40 and over in a rural South African community.²³ The cohort consists of 5059 people (n = 2345 men and 2714 women). This relatively young starting age was chosen for two reasons. First, life expectancy at birth is low in South Africa, mainly due to HIV. Second, a central aim is to observe longitudinally the pre-disease pathways that evolve in middle age and affect later life health.

Study design

HAALSI is a population-based community observational study with longitudinal follow-up at 3-year intervals. Embedded in the study are several randomized experiments and evaluations of public policies. HAALSI includes virtually complete mortality ascertainment, using state-of-the-art algorithms for verbal autopsies.²⁴

Study setting

The study was conducted in the Agincourt sub-district in Mpumalanga Province, South Africa, where the MRC/Wits Rural Public Health and Health Transitions Research Unit has been running the Agincourt HDSS since 1992. The HDSS conducts an annual census of all households and collects vital events for all household members (births, deaths and migration), and residency status. Sociodemographic characteristics are collected in alternating years.²³ The study area consists of 31 villages and covers approximately 450 km²; the total population is approximately 116 000 people. The primary health care system consists of six clinics, two health centres and three district hospitals. Despite the Apartheid legacy of underdevelopment and inadequate education, the social situation of this community has improved in the past 22 years as South Africa experienced political change to a democratic governance system. However, there are still gaps in access to electricity, water and tarred roads.²³ Unemployment rates are high, leading to high rates of labour migration with reliance on remittances as an important source of income.²⁵ The demographic profile of the HAALSI cohort is typical of rural South Africa; life expectancy at older ages has improved in Agincourt as well as elsewhere in rural South Africa,^26,27 though continued high fertility has led to overall slower compositional ageing than in the national population.²⁸

Study population

Of the 116 000 people living in the study setting, 12 875 men and women met eligibility criteria for the study: aged 40 and older as of 1 July 2014 and permanently living in the study site for 12 months preceding the 2013 HDSS census. Using these inclusion criteria, a sampling frame of the 12 875 (8974 women and 3901 men) was identified and 6281 people were randomly selected to participate in HAALSI; gender-specific sampling fractions were developed to ensure a gender-balanced cohort.

Recruitment

Sampled individuals were interviewed at home between November 2014 and November 2015. Once identified, potential participants were asked to provide informed consent in xiTsonga, the local language, or in English. Participants unable to read had a witness and used an inked fingerprint as signature.

From the selected 6281 men and women, 5059 completed home interviews; a response rate of 85.9% (Figure 1). A total of 391 (6%) of the sample were ineligible due to death or out-migration from the surveillance area before interview. Of the remaining 5890 eligible, 353 (6%) were not found, 48 (1%) were unable to participate and 430 (7%) refused (Figure 1). Those who refused to participate were more likely to be women, were younger, had more education and were more likely to be native South Africans. A brief interview with a proxy was conducted for 116 (2.3%) participants who were too ill or unable to respond to the full interview.

Figure 1.

Flow-chart of HAALSI sample.

Follow-up

The HAALSI cohort has been contacted twice a year following baseline. At the beginning of each year, each participant is contacted by phone or home visit to verify phone number, address and vital status. Approximately 6 months later, the annual HDSS census is conducted, reaching all participants who still reside in the study area. HAALSI cohort members who have permanently moved outside the study site since baseline are contacted by phone and remain in the cohort.

Deaths of HAALSI participants are identified through these biannual contacts and reported to the verbal autopsy team (VA team). The VA team visits households of every deceased person within 12 months of the death, and interviews caregivers of the deceased using a World Health Organization (WHO) standardized VA questionnaire.²⁹ Probable cause of death is established using InterVA-4.²⁴

Field team training

The HAALSI baseline field team comprised experienced local fieldworkers and supervisors. The 1-month training included study objectives, household and individual computer-assisted personal interviews (CAPI), anthropometrics, performance measurements, dried blood spot and point of care blood-based measurements, and referrals to health facilities when indicated.

Quality control and quality assurance

Data were captured via CAPI during interview. To ensure data completeness and accuracy, internal checks were embedded in the system. Study team analysts produced weekly and monthly field check tables to support field-based teams for continuous progress and data quality monitoring.

Ethics

The study received ethical approvals from the University of the Witwatersrand Human Research Ethics Committee (ref. M141159), the Harvard T.H. Chan School of Public Health, Office of Human Research Administration (ref. C13–1608–02) and the Mpumalanga Provincial Research and Ethics Committee.

What has been measured?

The interview lasted 2.5–3 hours and consisted of household and individual questionnaires. A summary of all data collected is presented in Table 1. The household questionnaire included a household roster, consumption, income and assets. The individual questionnaire included sociodemographic items, self-reported health and health behaviours, and performance assessments of physical and cognitive function.

Table 1.

Data collected during household and individual interviews in the HAALSI study in Agincourt, South Africa

INDIVIDUAL
Demographics	Employment	Social conditions
Age Sex Education Marital status Religion Nationality Ethnicity	Employment status Occupation Wages Income for unemployment:  Insurance  Disability/pensions	Social networks, social support, interactions
		Caregiving/care receiving
		Psychological well-being
		Gallup (Well-being)
		Center for Epidemiological Studies Depression scale
		Post-traumatic stress disorder
		Expectations
		Survival
		HIV infection
Cognition	Self-reported health	Health behaviours
Orientation Word recall  Immediate  Delayed Numeracy, number series Self-rated memory	Vision and hearing Diagnosis and treatment:  Hypertension  Glycaemia  HIV  TB  Stroke  Angina  Myocardial infarction  High cholesterol  Kidney disease Male circumcision Brief Pain Inventory Pittsburgh Sleep Quality Index Activities of Daily Living	Tobacco use Alcohol consumption Diet (fruit, vegetables, bread, soft drinks) International Physical Activity Questionnaire Vigorous, moderate exercise, sedentary activity Sexual behaviour Morisky Medication Adherence Scale
Health care utilization	Physical examinations: function	Point of care: blood assays
Type of health facility used Number of visits	Blood pressure Pulse Grip strength Height and weight Body mass index Waist and hip circumference, Waist-hip ratio Walk time Balance	Total cholesterol, HDL, LDL, Triglycerides Glucose Haemoglobin HIV  Viral load  Antiretroviral therapy (3TC, FTC) High-sensitivity C-reactive protein
HOUSEHOLD
Members	Consumption	Income and assets
Age Sex Relationship to household head Education	Food Communication Transport Rent, mortgage Consumables Celebrations, funerals Education Health care Transfers (including charity)	Household member employment status Household member wages Business Subsidies, pensions Gifts Rentals Durable goods (TV, fridge …) Properties (dwelling, land …) Livestock

We collected anthropometric measurements and biomarkers via point of care and dried blood spots. Table 2 presents detailed descriptions of devices used to take these measurements, field procedures and thresholds used to categorize these objective measures.

Table 2.

Anthropometry, physical performance and point of care measures, procedures for data collection and threshold values

Anthropometry	Equipment	Field procedures	Thresholds: measurements
Blood pressure50	Omron©	Three measurements, 2 min apart, after 5-min rest. Final blood pressure: average of second and third measures	Hypertension: diastolic ≥ 90 OR systolic ≥ 140 OR on hypertension medication
Waist circumference51	SECA© flexible tape	Tape at the navel, waist measured at mid distance between the iliac crest and the lowest rib, on a horizontal plane with participant standing	Men: Women:	Increased: ≥ 94 cm AND < 102 cm; Substantially increased: ≥12 0 cm Increased: ≥80 cm AND < 88 cm; Substantially increased: ≥88 cm
Hip circumference	SECA© flexible tape	Tape at the hip joint, circle around the widest portion of the buttocks on a horizontal plane with participant standing
Waist/hip ratio51		Ratio of the waist and hip measurement	Men: Women:	Increased waist/hip ratio: ≥0.90 Increased waist/hip ratio: ≥0.85
Height	Genesis Growth Management Scale©, Patient Focus Africa	Measured in centimetres with one decimal point, using a height sensor placed on top of participant’s head connected via infrared to the weight scale
Weight	Genesis Growth Management Scale©, Patient Focus Africa	Measured in kilograms with one decimal point
Body mass index (BMI)52		Weight in kilograms / (height in metres)2	Underweight  < 18.5;	Overweight 25–29.9; obese > 30.00
			normal 18.5–24.9
Physical Performance
Walking course	Century digital timer: Jumbo©	Participant walks 2.5 m, and repeats, at their usual speed using any walking aid needed	Mean walk time (5 m/s) = 5/(time 1 + time 2)
Semi-tandem		Participant stands with the heel of one foot beside the other, touching the toe of the other foot and holds for 10 s
Tandem		If semi-tandem was completed: participant stands with the heel of one foot touching the toe of the other foot, with feet in one line and holds for 60 s i < 70 years of age, or 30 s if >70 years
Side by side		If semi-tandem was not completed: participant stands with both feet together, side by side, with the inside of both feet touching, and holds for 10 s
Grip strength	Smedley© Digital Hand Dynamometer (12–0286)	Participant sits upright with feet flat on the floor, legs uncrossed and elbow at a 90-degree angle with arm close to body and forearm parallel to the floor. Results of grip recorded in kg with one decimal point	Mean of the 2 measures per hand if difference between measures is <10 kg; strongest measure per hand if mean difference >10 kg
Point of care
Total cholesterol, LDL, HDL, triglycerides53,54	Cardio Chek© PA (Silver version)	Finger prick using PTS Panel #1710 lipid panel test strips	High total cholesterol >  6.21 mmol/L, high LDL  > 4.1 mmol/L	Low HDL < 1.19 mmol/L, high triglycerides > 1.7 mmol/L
Glycaemia55	CareSens© N Monitor	CareSens N blood glucose test strips	Diabetes:	Glucose ≥ 11.1 mmol/L no fasting Glucose ≥ 7 mmol/L fasting Glucose < 7 mmol/L on diabetes medication
Haemoglobin56	Hemocue© Hb 201 + Analyser	Finger prick using Hemocue Hb 201 + microcuvette	Men: Women:	Normal > 12.9 g/dl; mild anaemia ≤ 12.9 g/dl and ≥11 g/dl, moderate anaemia < 11 g/dl and ≥ 8 g/dl Normal > 11.9 g/dl, mild anaemia ≤ 11.9 g/dl and ≥ 11 g/dl, moderate anaemia < 11 g/dl and ≥8 g/dl, severe anaemia <8 g/dl

Although HAALSI follows HRS sister studies in balancing assessments of health and functioning with social, economic and behavioural conditions, it measures more deeply critical features of HIV/AIDS infection, cardiometabolic disorders, and family and social networks than do many comparable sister studies. The baseline assessment consists of seven sections described below.

Social conditions: early childhood, family, social networks, mobility, migration, household characteristics

HAALSI gathered information about participants’ demographics and family information: age, literacy, education, religion, marital status, timing of marriage and marital dissolution. Participants were asked about living children and their sex, age and current residence; and number, age and residence of grandchildren and siblings. The interview included questions about participants’ early life, place of birth, duration of residence in area, parents’ union status at participant’s birth, parents’ current vital state, age and residence. Participants were asked about paternal schooling and occupation.

The individual interview contained a rich set of questions on social networks and social support. Formal egocentric social network structure was modelled after the United States National Social Life, Health and Ageing Project (NSHAP),³⁰ in which participants nominate up to six individuals close to them and describe interactions.

Economic conditions and productivity

The household interview included: consumption and expenditures; labour income; business income; government transfers; remittances; housing characteristics; ownership of durable goods, land, livestock and financial assets; and food security. A wealth index was created from principal components analysis of household characteristics and ownership of household items, vehicles and livestock.³¹ Individual participants were asked about their own work status, working hours, income, unemployment, disability income and pensions.

Cognition and mental health

Specific measures included the eight-item Center for Epidemiological Studies Depression (CES-D) scale for depressive symptoms,³² a short screening scale for post-traumatic stress disorder³³ and Gallup World Values Survey questions on subjective well-being and life satisfaction.³⁴ In assessing cognitive functioning, we harmonized with HRS, including items on orientation, immediate and delayed recall, and numeracy.^35,36 Domain-specific cognitive assessments developed by Humphreys for low-literacy settings were administered with tablets to about half the HAALSI cohort.³⁷

Health

This was assessed primarily by self-report. Participants were asked about doctor, nurse or other health professional diagnosis and treatment of cardiovascular and metabolic conditions (high cholesterol, high blood pressure, stroke, heart failure, angina, myocardial infarction), diabetes, tuberculosis, HIV infection and kidney disease (Table 1). Health care utilization and expenditures questions were asked. Other indices include the Pittsburgh Sleep Quality Index³⁸ and Brief Pain Inventory.³⁹

Health behaviours

Participants were asked about tobacco use (present and past, quantity, frequency, duration, type of tobacco) and alcohol consumption (ever consumed, daily quantity when consumed, type of alcoholic drink and binge drinking behaviour). Show-cards and a table of equivalent alcohol units per drink were used to ensure accuracy. Show-cards were used to collect information on dietary consumption, frequency and quantity of fruit, vegetables, bread and soft drinks taken.

To capture physical activity, we administered the International Physical Activity Questionnaire (IPAQ)⁴⁰ which includes type of work, exercise (vigorous, moderate) and sedentary activity. For each activity, we enquired about amount of time spent during weekdays and weekends.

Information on sexual activity and partnerships included number and type of partners and condom use. Participants self-reported whether they had ever been tested for HIV, and disclosed their HIV status, knowledge of antiretroviral therapy (ART) and whether they were receiving ART.

Physical function and performance

Activities of daily living (ADLs) measures included difficulty in walking, eating, bathing, getting in/out of bed and using the toilet. Function measures included a 5-m timed walk and balance tests. We measured grip strength of both hands twice using the Smedley^© Digital Hand Dynamometer (12–0286).

Anthropometry and biomarkers

We collected a comprehensive set of measures including: weight, height, hip and waist circumferences; blood pressure; and point of care and dried blood spot (DBS) assays (Table 2).

Blood pressure (systolic, diastolic) and pulse were collected three times, 2 min apart, after the participant had been seated for 5 min, using the OMRON^© Automatic blood pressure monitor M6W. Final blood pressure and pulse were calculated using the average of the second and third readings. Hip and waist circumferences were measured in centimetres with participants in the standing position.

Eight blood drops were collected from a finger prick. Three blood drops were used to measure: point of care total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol and triglycerides (Cardio Chek^© PA Silver version); haemoglobin (Hemocue^© Hb 201 + Analyser); and glucose (CareSens^© N Monitor). Five dried blood spots (DBS) on Whatman 903 TM filter paper were kept at room temperature (approximately 23°C) for 1–3 weeks and then sent to Global Labs in Durban and stored at -20°C. DBS assays measured high-sensitivity C-reactive protein (hsCRP),⁴¹ HIV status and, when HIV-positive, viral load and traces of emtricitabine (FTC) and lamivudine (3TC) using levels higher than 0.02 µg/ml for positivity for both drugs.

The HIV results were determined by first conducting Vironostika Uniform 11 (Biomeriuex, France) screening assay. If positive, confirmation was done using Roche Elecsys, USA. If the confirmatory test was positive, the final result was considered positive and viral loads were calculated and reported. In those few cases that were weakly positive, final results were considered positive and viral load calculated.

What has it found?

Sociodemographic characteristics and health

A gender comparison of sample demographic characteristics and socioeconomic conditions including age, education, nationality, employment, marital union status, household composition, consumption per capita quintile rank and household asset index quintile rank is presented in Table 3. Results show a mean age for both men and women of 61.7 years. Women attained fewer years of education compared with men, fewer women are married and more widowed compared with men. More men are employed than women, fewer women are in single-member households compared with men and more women are in households ranked lowest on household consumption per capita.

Table 3.

Comparison of sociodemographic characteristics by gender among HAALSI participants

Variables	Male		Female
	N	%	N	%
Age group*
40–49	418	17.8	500	18.4
50–59	624	26.6	786	29.0
60–69	643	27.4	661	24.4
70–79	446	19.0	432	15.9
80+	214	9.1	335	12.3
Years of education*
No formal education	957	40.9	1349	49.9
Primary (1–7 years)	833	35.6	883	32.7
Some secondary (8–11 years)	314	13.4	260	9.6
Secondary or more (12+ years)	234	10.0	212	7.8
Nationality of origin
South African	1663	70.9	1865	68.8
Mozambican/other	682	29.1	844	31.2
Union status*
Never married	166	7.1	124	4.6
Separated/divorced	300	12.8	350	12.9
Widowed	276	11.8	1264	46.6
Currently married/cohabitating	1602	68.3	973	35.9
Household composition*
Living alone	330	14.0	204	7.5
Living with 1 other person	257	11.0	281	10.3
Living in 3–6 person household	1055	45.0	1383	51.0
Living in 7+ person household	703	30.0	846	31.2
Employment status*
Employed	443	18.9	362	13.4
Unemployed	1709	73.1	2010	74.2
Homemaker	186	8.0	335	12.4
Household consumption per capita*
Quintile 1 (lowest)	456	19.5	591	21.8
Quintile 2	444	18.9	580	21.4
Quintile 3	468	20.0	553	20.4
Quintile 4	464	19.8	511	18.8
Quintile 5 (highest)	513	21.9	479	17.7
Household asset index
Quintile 1 (lowest)	502	21.4	544	20.0
Quintile 2	455	19.4	546	20.1
Quintile 3	450	19.2	541	19.9
Quintile 4	457	19.5	550	20.3
Quintile 5 (highest)	481	20.5	533	19.6

Missing data for years of education: 17, nationality of origin: 5, union status: 4 and employment status: 14.

^*Chi-square P-value < 0.001.

Prevalence by gender of key health conditions, self-reported diseases, behavioural risk factors, function and cognitive measures is shown in Table 4. Below we discuss specific findings related to cognitive function, sexual behaviour and HIV, physical function and cardiometabolic risk factors.

Table 4.

Prevalence of cardiovascular risk factors, self-reported cardiovascular diseases, behavioural risk factors and physical function

Indicators	Male		Female
	N	(%)	N	(%)
HIV-positivea	483	(23.0)	565	(22.9)
Hypertensionb
Self-reported medication or systolic ≥ 140 or diastolic ≥ 90 mmHg**	1227	(54.3)	1616	(61.3)
Self-reported high blood pressure or systolic ≥ 140 or diastolic ≥ 90 mmHg**	1319	(58.4)	1768	(67.1)
Mean systolic BP	137.9		138.1
Mean diastolic BP	81.9		82.3
Diabetesc
Self-reported medication or glucose ≥ 11.1 mmol/L*	153	(7.2)	219	(8.8)
Self-reported medication or glucose > 7 mmol/L fasting or ≥ 11.1 mmol/L*	197	(9.3)	276	(11.1)
Self-reported diabetic or glucose > 7 mmol/L fasting or ≥11.1 mmol/L*	224	(10.5)	309	(12.4)
Anthropometric measurements
Overweight (BMI ≥ 25–29.9) or obese (BMI ≥ 30)d**	948	(44.0)	1751	(69.6)
High waist circumference (men ≥ 94 cm; women ≥ 80 cm)e**	725	(32.9)	2128	(83.7)
High waist/hip ratio (men ≥ 0.90; women ≥ 0.85)f**	1351	(61.6)	1859	(73.4)
Lipids
High cholesterol  ≥6.21 mmol/L)g**	79	(4.2)	196	(8.5)
High triglycerides (>2.25 mmol/L)h	388	(20.4)	504	(21.8)
High LDL (>4.1 mmol/L)i**	36	(2.1)	105	(5.0)
Low HDL (<1.19 mmol/L)j**	594	(31.3)	520	(22.5)
C-reactive proteink
Elevated CRP (>2 mg/dl)**	1009	(51.5)	1387	(59.2)
Anaemial
Moderate (<11 g/dl–≥8 g/dl)/severe anaemia (<8 g/dl)**	235	(11.4)	544	(22.5)
Self-reported measuresm
High cholesterol	10	(0.5)	20	(0.7)
High blood pressure**	797	(34.0)	1321	(48.7)
Stroke	64	(2.7)	85	(3.1)
Heart failure	12	(0.5)	21	(0.8)
Angina*	42	(1.8)	77	(2.8)
Myocardial infarction	10	(0.4)	11	(0.4)
Diabetes	145	(6.2)	192	(7.1)
Tuberculosis**	258	(11.0)	188	(6.9)
Kidney disease	97	(4.1)	117	(4.3)
Behavioural risk factorsm
Currently drinks alcohol**	912	(38.9)	259	(9.6)
Currently uses tobacco**	450	(19.2)	10	(0.4)
Physical functionn
Mean walk time (s/5 m)**	12.7		13.5
Activities of daily living (ADL)	210	(9.0)	246	(9.1)
Mean grip strength (kg)**	28.6		20.9

BP, blood pressure.

^a4560 consented to HIV testing and had valid dried blood spot results.

^b4895 had blood pressure readings.

^c4626 had glucose biomarker results.

^d4670 had valid height and weight measurements.

^e4744 had a valid waist measurement.

^f4728 had valid hip and waist measurements.

^g4195 had a valid cholesterol reading.

^h4214 had a valid triglyceride reading.

ⁱ3820 had a valid LDL cholesterol reading.

^j4212 had a valid HDL cholesterol reading.

^k4302 had CRP dried blood spot results.

^l4493 had valid haemoglobin results.

^mQuestionnaire responses were missing for: high cholesterol (6), high blood pressure (4), stroke (3), heart failure (3), angina (4), myocardial infarction (3), diabetes (6), tuberculosis (7), kidney disease (5), currently drinks alcohol (3), currently uses tobacco (5).

ⁿWalk time assessed for 4694 and grip strength assessed for 4699.

^*Chi-square/t-test P-value < 0.05; ^**chi-square/t-test P-value < 0.001.

Cognition

Our approach to assessing cognition rests on both: novel assessments using tablets for low literacy and numeracy tests;³⁷ and on standard assessments (attention, immediate and delayed recall) harmonized with sister studies in the US, Mexico, China and India. HAALSI provides an opportunity to test whether educational attainment is strongly associated with cognition—as has been reported in many other countries^42–44—in a setting where many people could not attend school. We report that a higher proportion of people with no formal education have low cognitive function in a number of domains compared with their counterparts with any formal education, regardless of age.³⁷ In an analysis of early life conditions, older adults with poor self-reported childhood health or whose father worked in unskilled manual labour had relatively poor cognitive outcomes.⁴⁵ These findings suggest that education can provide cognitive reserve, even in a setting where access to education was restricted.

Sexual behaviour and HIV

A recent HAALSI study⁴⁶ reports that many older adults are still sexually active. In contrast to stereotypes, more than half of HAALSI participants (57%) reported at least one sex partner in the past 2 years. The proportion was higher among men (77%) compared with women (40%), and generally decreased with age. Over one in 10 of these recent partners (12%) were classified as either casual or anonymous, and only a quarter of participants (25%) reported ever using condoms with their most recent partner. In an HIV-hyperendemic community like the Agincourt study area—with 23% HIV prevalence in this sample—these sexual behaviours are consistent with both HIV transmission risk and HIV acquisition risk.

Physical function

Measured physical performance in the HAALSI sample was associated with socioeconomic conditions—higher school attainment and increased household wealth were both strongly associated with higher hand grip strength and faster gait speed.⁴⁷ In order to place the HAALSI cohort in international context, we compared its functioning and self-reported physical health with HRS and sister studies in Mexico and China. HAALSI respondents had better self-reported health and lower rates of reported ADL limitation than most other countries.⁴⁷ However, the HAALSI sample had overall lower age-adjusted physical performance outcomes.⁴⁷

Cardiometabolic risk factors

Hypertension prevalence was high (58.4%), and significantly increased with age.⁴⁸ We observed high levels of overweight/obesity, affecting 70% of women and 44% of men. Total cholesterol levels were twice as high among women as compared with men (8.5% vs 4.2%) and women self-reported higher levels of most conditions including a higher prevalence of angina (2.8% vs 1.8%). The fact that self-reported levels are lower than measured levels is a reflection of the low level of awareness of some cardiovascular risk factors among HAALSI participants. In our recent paper on cardiometabolic risk,⁴⁸ we observed that HIV-negative people had higher levels of cardiometabolic risk factors than HIV-positive people, with the HIV-negative presenting higher prevalence of hypertension (men: 59.2% vs 38.7% and women: 67.2% vs 43.8%), diabetes (men: 10.9% vs 7.3% and women: 13.1% vs 7.9%), overweight (men: 29.3% vs 22.7% and women 28.7% vs 26.2); and obesity [body mass index (BMI) 30–34.9] (men: 12.9% vs 8.6% and women: 23.9% vs 18.6%). The absolute 10-year cardiovascular risk scores ranged from 7.7–9.7% for women and from 12.5–15.3% for men.⁴⁸ Comparing the cardiometabolic risks of the HAALSI cohort with the national South Africa National Health and Nutrition Examination Survey (SANHANES), we found that the HAALSI lipid profile is similar, although HDL levels and the waist hip ratio were higher in the HAALSI cohort and men reported a higher rate of smoking.⁴⁸

Measured blood pressure

The 2883 participants with high blood pressure (defined as systolic ≥ 140 mmHg, diastolic ≥ 90 mmHg, or self-reported antihypertensive medication use) were generally older (mean age of 64.1 ± 12.7 years). We found high rates of hypertension treatment among those who were aware, with nearly half of those treated having controlled blood pressure.⁴⁹ Multivariable regression results showed that hypertension awareness was associated with higher socioeconomic status, previous cardiovascular disease (CVD), non-immigrant status, literacy and physical limitation, and awareness was higher for women compared with men.^47–49 The HAALSI cohort showed higher levels of awareness, treatment rates and control levels than previously published data in the region, possibly due to increased awareness following previous studies in the area which focused on stroke and hypertension.^3–6

What are the main strengths and weaknesses?

The HAALSI cohort was established as a population-based study representative of the Agincourt sub-district. The strength of this cohort lies in the ability to identify a broad range of social, economic, behavioural and biological risk factors and their associations with a range of health outcomes through longitudinal follow-up. Though not a nationally representative sample, HAALSI findings are likely to be mirrored in similar rural communities. The fact that this study is embedded in the Agincourt HDSS, where the population has been followed since 1992, is a major strength. The field and research teams have extensive experience implementing fieldwork, and the HDSS allows for annual cohort follow-up, death registration and cause of death ascertainment. Future waves of HAALSI promise an exceptional combination of health, sociodemographic, cognitive and economic information that will enhance our understanding of the complex nature of rapid epidemiological and demographic transitions in this rural setting.

Can I get hold of the data? Where can I find out more?

The HAALSI baseline data are publicly available at the Harvard Center for Population and Development Studies (HCPDS) programme website [www.haalsi.org]. Data are also accessible through the Inter-university Consortium for Political and Social Research (ICPSR) at the University of Michigan [www.icpsr.umich.edu] and the INDEPTH Data Repository [http://www.indepth-ishare.org/index.php/catalog/113]. The release includes documentation of baseline data. Further details can be obtained by e-mailing the corresponding author of this paper.

Profile in a nutshell

• The HAALSI study addresses the knowledge gap regarding health and ageing in South Africa as it undergoes epidemiological and demographic transitions. It is designed as a harmonized sister study to the US Health and Retirement Study (HRS).

• In-person interviews were conducted from November 2014 through November 2015 in the Agincourt sub-district, Mpumalanga Province, South Africa, where the INDEPTH Agincourt Health and Demographic Surveillance System has been run since 1992. The cohort consists of 5059 men (n = 2345) and women (n = 2714) aged 40 and older, with planned longitudinal follow-up at 3-year intervals.

• HAALSI is closely harmonized with the global HRS sister studies in assessments of health, functioning and social, economic and behavioural conditions. HAALSI goes into more depth on HIV, cardiometabolic disorders, family dynamics and social networks. It is designed to identify social and economic determinants of health. Broad sections of the baseline assessment include: general health; physical function; cognition; mental health; anthropometrics and biomarkers; behavioural risks; social conditions; economic conditions; and productivity.

• HAALSI is a collaboration between the Harvard Center for Population and Development Studies and the MRC/Wits Rural Public Health and Health Transitions Research Unit, a member centre of the INDEPTH Network. Data are in the public domain.

Funding

The HAALSI study, funded by the National Institute on Aging (P01 AG041710), is nested within the Agincourt Health and Demographic Surveillance System site, supported by the University of the Witwatersrand and Medical Research Council, South Africa, and the Wellcome Trust, UK (grants 058893/Z/99/A; 069683/Z/02/Z; 085477/Z/08/Z; 085477/B/08/Z). The HAALSI study has been carried out through a collaboration between the Harvard Center for Population and Development Studies from Harvard T.H. Chan School of Public Health, MRC/Wits Rural Public Health and Health Transitions Research Unit from School of Public Health at the University of the Witwatersrand in South Africa, and the INDEPTH Network in Accra, Ghana.

Conflict of interest: None declared.