Work

Cohort Profile: South African Population Research Infrastructure Network (SAPRIN)

Year: 2021

Type: article

Abstract: The South African Population Research Infrastructure Network (SAPRIN) is a harmonized national network of health and demographic surveillance system (HDSS) nodes, constituted in 2016 with the aim of establishing a national resource and research platform to strengthen the evidence for informing health and social policy and programming at a time of stringent fiscal constraint. The three HDSS founding nodes are hosted by the Medical Research Council/Wits Rural Public Health and Health Transitions Research Unit (Agincourt) in Mpumalanga province, the University of Limpopo (DIMAMO) in Limpopo province and the Africa Health Research Institute (AHRI) in KwaZulu-Natal. In 2020, the SAPRIN population cohort covered 355 000 individuals for whom data on births and deaths, residency status and migration, household asset ownership status, and measures of well-being including employment, education and social protection are updated three times a year. Attrition is low due to keeping track of population dynamics, and refusal rates are usually below 1%. The SAPRIN longitudinal population dataset is publicly available at [http://saprindata.samrc.ac.za]. South Africa is striving to emerge from a legacy of gross social injustice and consequent health and socioeconomic inequality, to becoming a country where all residents have opportunities to build productive lives. However, recent declines in economic performance and unemployment, exacerbated by weaknesses in national and provincial level governance, coupled with colliding epidemics of HIV/AIDS and non-communicable diseases, have left the country’s leadership with serious, seemingly intractable challenges. Moreover as with most countries, the effects of stringent sociobehavioural responses to the coronavirus disease 2019 (COVID-19), with serious economic consequences, serve to amplify such challenges. At a national level, there has been a realization that reliable, relevant, and long-term research infrastructure is needed to generate data and evidence to underpin policies that can address these interconnected problems. The Department of Science and Innovation invested in a programme of national research infrastructure development, called the South African Research Infrastructure Roadmap.1 This includes the South African Population Research Infrastructure Network (SAPRIN) established in 2016. Building on the strengths of three existing Health and Demographic Surveillance Systems, SAPRIN is a harmonized network of population-based cohorts with data that are made accessible for the research community and national policy makers. The aim of the network is to provide: (i) ongoing, up-to-date, longitudinal health, demographic and socioeconomic data, representative of South Africa’s fast-changing poorer communities, for research and policy evaluation and interpretation and calibration of national data; (ii) an extensive, versatile and interdisciplinary research platform for researchers from universities, science councils and regional and international collaborations; (iii) a scientific evidence base for cost evaluation, policy making and targeting intervention programmes; and (iv) expanded human capacity for conducting research that is effectively linked with national, regional and international networks. Since SAPRIN’s inception in 2016, three existing health and demographic surveillance system (HDSS) nodes in South Africa have been brought together to form a harmonized national network, namely: MRC/Wits University Agincourt HDSS in Bushbuckridge District, Mpumalanga, established in 1992 with a 2019 population of 115 000 individuals2; University of Limpopo DIMAMO HDSS in the Capricorn District of Limpopo, established in 1995, with population of 8000 individuals, then expanded to 42 000 individuals in 2010, and expanded again in 2019 to 100 0003; and the Africa Health Research Institute (AHRI) HDSS in uMkhanyakude District, KwaZulu-Natal, linked to the University of KwaZulu-Natal,4 established in 2000 with a 2019 population of 140 000 individuals. The three founding nodes added to their capacities as needed to introduce a common scientific protocol. In 2019, SAPRIN released its first harmonized multicentre dataset, including the legacy data from the start of each of the founding three HDSS nodes. The population platform will be expanded to include urban nodes in the country’s two main metropolitan areas, Gauteng and the Western Cape. The new Gauteng node, GRT-INSPIRED, was established in 2020 and is run by a coalition of three universities, namely University of Witwatersrand, University of Johannesburg and University of Pretoria. With a common scientific protocol in place, the new HDSS nodes will be able to produce data harmonized with the existing platform. The Western Cape node, C-Sharp, will start in 2022. The SAPRIN aim, with long-term support from the Department of Science and Innovation, is to run seven nodes across South Africa which will incorporate dynamic, bi-directional migration flows linking poor rural communities with urban centres (see Figure 1). Representational map of SAPRIN, showing both founding and joining nodes in the South African setting Representational map of SAPRIN, showing both founding and joining nodes in the South African setting Each SAPRIN node is a fully functioning Health and Demographic Surveillance System (HDSS). Such ‘whole population cohorts’ provide longitudinal platforms that enable rigorous individual follow-up, systematic recording of all vital events (births, deaths, migrations) and the opportunity for a range of observational and interventional study designs.2–4 Individuals eligible for inclusion are all members of the households located within the stipulated geographical boundaries of a SAPRIN node—usually comprising several thousand households in an area conforming to the boundaries for decentralized health and social administration. Households and household membership are self-defined by a household informant in an interview at their place of residence (or a telephone interview). A resident household member is defined as an individual who intends to sleep most of the time at the dwelling place of the household, over a 4-month period. A non-resident member is defined as an individual who retains close economic and relationship ties to the household but does not physically reside with them for the majority of time. To ensure complete household membership, a rigorous system is in place for routinely updating all vital events including in- and out-migration, births and deaths of individuals in the population cohort. In about half of the cases, individuals tend to have a single period of residence from the time of first enumeration, at birth or in-migration, to their eventual death or out-migration. In the remaining cases, household members tend to transition from resident to non-resident through out-migration or movement within the HDSS area. Individuals returning from a period of non-residency continue with a new residency period and retain their original identification. Migration reconciliation takes place to ensure that new migrants are verified against known individuals, to prevent duplicate counting for the same individual. SAPRIN nodes collect data in three annual rounds, over a 45-week schedule: one by computer-assisted personal face-to-face interview and two telephone interviews. Frequency of follow-up is designed to minimize the risk of missing demographic events, especially pregnancies, which allows collection of pregnancy outcomes including stillbirths and neonatal deaths. Before 2019, the three HDSS nodes had different data collection frequencies, influenced by the scientific questions addressed as well as resources available, as shown in Table 1. Attrition is low due to keeping track of population dynamics. Refusal rates are low and usually below 1%. South African Population Research Infrastructure Network (SAPRIN) infrastructural development stages 70 000 individuals 15 000 households Household—proxy Individual 3 1 Household—proxy Individual 3 1 Household—proxy Individual 3 1 Agincourt HDSS node Dikgale HDSS node AHRI HDSS node Household proxy Individual 3 1 Agincourt HDSS node DIMAMO HDSS node AHRI HDSS node Household—proxy Individual 3 1 GRT-INSPIRED C-SHARP 70 000 individuals 15 000 households Household—proxy Individual 3 1 Household—proxy Individual 3 1 Household—proxy Individual 3 1 Agincourt HDSS node Dikgale HDSS node AHRI HDSS node Household proxy Individual 3 1 Agincourt HDSS node DIMAMO HDSS node AHRI HDSS node Household—proxy Individual 3 1 GRT-INSPIRED C-SHARP Dikgale HDSS, the forerunner of DIMAMO HDSS; Africa Centre DIS, a forerunner of Africa Health Research Institute, Population Intervention Programme; SAMRC, South African Medical Research Council; DIMAMO, an amalgam of the initial two letters of the three sub-districts covered by the field-site; GRT-INSPIRED, the Gauteng Research Triangle Initiative for the Study of Population, Infrastructure and Regional Economic Development; C-SHARP, Cape Town Surveillance through Healthcare Action Research Project. South African Population Research Infrastructure Network (SAPRIN) infrastructural development stages 70 000 individuals 15 000 households Household—proxy Individual 3 1 Household—proxy Individual 3 1 Household—proxy Individual 3 1 Agincourt HDSS node Dikgale HDSS node AHRI HDSS node Household proxy Individual 3 1 Agincourt HDSS node DIMAMO HDSS node AHRI HDSS node Household—proxy Individual 3 1 GRT-INSPIRED C-SHARP 70 000 individuals 15 000 households Household—proxy Individual 3 1 Household—proxy Individual 3 1 Household—proxy Individual 3 1 Agincourt HDSS node Dikgale HDSS node AHRI HDSS node Household proxy Individual 3 1 Agincourt HDSS node DIMAMO HDSS node AHRI HDSS node Household—proxy Individual 3 1 GRT-INSPIRED C-SHARP Dikgale HDSS, the forerunner of DIMAMO HDSS; Africa Centre DIS, a forerunner of Africa Health Research Institute, Population Intervention Programme; SAMRC, South African Medical Research Council; DIMAMO, an amalgam of the initial two letters of the three sub-districts covered by the field-site; GRT-INSPIRED, the Gauteng Research Triangle Initiative for the Study of Population, Infrastructure and Regional Economic Development; C-SHARP, Cape Town Surveillance through Healthcare Action Research Project. Through ongoing field-, telephone- and computer-based operations, the following individual and household indicators are regularly updated: births and deaths, residence and in- and out-migrations, which are updated three times per year; and measures of well-being, including household asset status, labour status, education and social protection, which are updated once per year. Each node follows the core protocol, summarized in Table 2 with more detail given in the Supplementary material, available as Supplementary data at IJE online. This forms the basis for the SAPRIN longitudinal population dataset. The public access databases are structured to enable longitudinal analysis of events occurring in time. Three types of dataset are available and updated each year: a basic residency file; a file with residencies divided by age, year and delivery date; and a file with time-varying attributes, such as education, employment, marital status and household asset status. Household asset measures include 35 ordinal variables on asset ownership in the categories of modern assets, livestock, power supply, house construction materials and water and sanitation. A household socioeconomic status indicator is constructed by normalizing the sum of scores in each category and summing the five category scores.5 Variablesa collected at South African Population Research Infrastructure Network (SAPRIN) Health and Demographic Surveillance System nodes Full list of variables given in the Supplementary material, available as Supplementary data at IJE online. Variablesa collected at South African Population Research Infrastructure Network (SAPRIN) Health and Demographic Surveillance System nodes Full list of variables given in the Supplementary material, available as Supplementary data at IJE online. In response to the emergence of the COVID-19 epidemic, in April 2020 SAPRIN implemented an active surveillance module in a random sample of more than 2000 households in the three nodes in Mpumalanga, Limpopo and KwaZulu-Natal. Respondents have been interviewed every 2–3 weeks to obtain a dynamic picture of COVID knowledge and related behaviours and household outcomes (see Supplementary material). The data have been used to evaluate the impact of non-pharmaceutical interventions implemented by government policy.6 Figures 2–5 were computed from the basic residency dataset at the SAPRIN Data Repository.7 The plots below were computed from the basic residency dataset [https://doi.org/10.23667/SAPRIN.SISEBD2020], downloaded from the SAPRIN Data Repository on 15 June 2020. Age-sex-residency population pyramids on 1 July 2017 for the three SAPRIN nodes separately and the combined population Age-sex-residency population pyramids on 1 July 2017 for the three SAPRIN nodes separately and the combined population Age-standardized mortality rate per 1000 individuals for males and females for the three SAPRIN nodes separately and combined Age-standardized mortality rate per 1000 individuals for males and females for the three SAPRIN nodes separately and combined Age-standardized in-migration rate per 1000 individuals for males and females for the three SAPRIN nodes separately and combined Age-standardized in-migration rate per 1000 individuals for males and females for the three SAPRIN nodes separately and combined Age-standardized out-migration rate per 1000 individuals for males and females for the three SAPRIN nodes separately and combined Age-standardized out-migration rate per 1000 individuals for males and females for the three SAPRIN nodes separately and combined Age-sex pyramids were constructed from harmonised HDSS data on the same day (1 July 2017) at the three SAPRIN nodes. Their similarity, despite coming from different provinces which exhibit heterogeneity of living environments, employment opportunities and culture, indicates that the nodal data can be combined. A comparison of the age/sex structure in the Agincourt population with the national census 2011 shows a good fit. Common features include the large sub-population of youth aged 20–25 years and a higher female sex ratio in older ages.8 The population pyramids include resident and non-resident household members, which reveals the extent of temporary, largely labour migration. With a median age of 20–24 years, there is a similar distribution of temporary migrants in the three nodes, balanced by sex. The temporary migration levels are especially high in the KwaZulu-Natal node. The least prominent temporary migrant age groups are in the teenage years, due to school attendance, and after the age of 60 when migrants retire, are eligible for a government pension and return to their rural homes. The location of all three nodes in former ‘Bantustans’ highlights the extreme lack of work opportunities in rural areas due to distance from labour markets. DIMAMO is closer to an urban centre, Polokwane, which explains why the proportion of temporary migrants of working ages is lower. Despite age standardization there are major shifts in mortality profiles over time. The series start at different points aligned with different starts of the HDSS operations: 1993 in Agincourt, 1997 in DIMAMO and 2000 in AHRI. The AIDS-TB epidemic affected all the populations, with an especially high impact and prevalence in KwaZulu-Natal.9 The HDSS findings highlight the decline in mortality associated with improved public health interventions introduced at different times by the respective provincial departments of health. These are supported by research contributed to by the SAPRIN nodes.10–13 Migration is defined here as moving into the dwelling of a household in a SAPRIN node from outside the node or moving out from a dwelling within the node to outside the node. As above, the migration trends begin in different years, as data became available. For all nodes there were high peaks of migration around the turn of the millennium, from 1999–2002, followed by a slow decline in migration rates for men and women over a 15-year period, with the lowest levels being current. Due to high levels of circularity, the periods with high migration rates have higher both in- and out-migration rates.14 The platform provides the opportunity for a range of observational and interventional study designs. Table 3 shows the current research data themes at each node, arising from their respective research portfolios. Nodal research data themes Research, trials and evaluation platform, including health, population and social transitions Respiratory virus transmission studies, including SARS-CoV-2, influenza, RSV Adolescent health and development, including HIV prevention, risk factors for chronic disease, anxiety and depression Adult and older adult health and well-being, including NCD and infectious chronic diseases, multimorbidity, cognition and dementias Social and environmental determinants of health, including migration, urbanization and health; and natural resources and food security Non-communicable diseases Mental health Strengthening the response to dementia Intervention to optimize adolescent BMI pre-conception, to address the of of COVID-19 on household food status and socioeconomic status risk and community Population HIV HIV and trials disease COVID-19 Social Population Research, trials and evaluation platform, including health, population and social transitions Respiratory virus transmission studies, including SARS-CoV-2, influenza, RSV Adolescent health and development, including HIV prevention, risk factors for chronic disease, anxiety and depression Adult and older adult health and well-being, including NCD and infectious chronic diseases, multimorbidity, cognition and dementias Social and environmental determinants of health, including migration, urbanization and health; and natural resources and food security Non-communicable diseases Mental health Strengthening the response to dementia Intervention to optimize adolescent BMI pre-conception, to address the of of COVID-19 on household food status and socioeconomic status risk and community Population HIV HIV and trials disease COVID-19 Social Population SAMRC, South African Medical Research Council; DIMAMO, an amalgam of the initial two letters of the three sub-districts covered by the non-communicable Nodal research data themes Research, trials and evaluation platform, including health, population and social transitions Respiratory virus transmission studies, including SARS-CoV-2, influenza, RSV Adolescent health and development, including HIV prevention, risk factors for chronic disease, anxiety and depression Adult and older adult health and well-being, including NCD and infectious chronic diseases, multimorbidity, cognition and dementias Social and environmental determinants of health, including migration, urbanization and health; and natural resources and food security Non-communicable diseases Mental health Strengthening the response to dementia Intervention to optimize adolescent BMI pre-conception, to address the of of COVID-19 on household food status and socioeconomic status risk and community Population HIV HIV and trials disease COVID-19 Social Population Research, trials and evaluation platform, including health, population and social transitions Respiratory virus transmission studies, including SARS-CoV-2, influenza, RSV Adolescent health and development, including HIV prevention, risk factors for chronic disease, anxiety and depression Adult and older adult health and well-being, including NCD and infectious chronic diseases, multimorbidity, cognition and dementias Social and environmental determinants of health, including migration, urbanization and health; and natural resources and food security Non-communicable diseases Mental health Strengthening the response to dementia Intervention to optimize adolescent BMI pre-conception, to address the of of COVID-19 on household food status and socioeconomic status risk and community Population HIV HIV and trials disease COVID-19 Social Population SAMRC, South African Medical Research Council; DIMAMO, an amalgam of the initial two letters of the three sub-districts covered by the non-communicable The work shows disease risk across the with in of coupled with adolescent and in older The and in of an cohort was designed to the social and economic that health in the and in women are at by in and The of dementia and on HIV status knowledge is the with and are more than in HIV status A comparison of HIV prevalence in older Agincourt and rates for men and for women through to age 70 in both South Africa has the HIV and programme in the with living with of whom are on In the country on health SAPRIN nodes, especially the AHRI node, contributed from improved the of to showing new HIV by in This the impact of government and and provides for The for within DIMAMO Population Health Research Centre and the are of the is at the and environmental risk factors for disease in the DIMAMO has with the University of Cape Town to a study on response to dementia in The study is at dementia and support migration linking rural and urban areas is in rural South is to the national and on and to be to migrants are and Health can be such as HIV or or such as migrants are from can or at the and for health and which can be a on the rural and health the COVID-19 South Africa in 2020, SAPRIN the of a population platform in place to to the scientific and social by the of April 2020 SAPRIN had an COVID-19 surveillance protocol in place, and by 2020 a long-term observational cohort was established that covered COVID-19 knowledge and and the economic and health including health, of the national The strengths of HDSS data from their longitudinal structure and complete and from being representative of and in which health, population and socioeconomic data are With of death and in- and out-migration following the initial the population are on a are of each residence in a study dwelling place, as well as membership of a social namely a household membership is which that the HDSS platform can and temporary migration. As an time analysis can be with of the population dynamic deaths and in- and out-migration both and temporary from the and high detail of data The of findings to a national population is not a that can be the HDSS does not have a study to ensure The of the sub-population to be in the national that population data can national mortality is that data collection for periods and in trials can produce a study communities that are can be by the of their and a not be from intervention effects in The for of of data collection is coupled with of and to individuals and communities linked to the are by in the work and are routinely in events where research are and The SAPRIN longitudinal population are publicly available and can be from [http://saprindata.samrc.ac.za]. for data or can be made to the The core protocol for SAPRIN is in with the of and the of as in the for The protocol was by the Research of the South African Medical Research In each node and from their for in the research platform. Supplementary data are available at IJE online. SAPRIN is by the Department of Science and Innovation as of the South African Research Infrastructure and is hosted by the South African Medical Research The to all the study and the at the three SAPRIN nodes, Data Data AHRI Data Data DIMAMO Data Data Centre for Infrastructure for Data designed the data for the by data data for the data and a first on the the on of the to be and to be for all of the

Source: International Journal of Epidemiology

DOI: 10.1093/ije/dyab261