International cohort studies
Cohort studies involve tracking the health of a given group of people over time. Look at some examples of international cohort studies to explore the reasons for this kind of research and the challenges it can present
We often think of surveillance as a price we pay to combat crime or terrorism, but it is also integral to public health. How can you plan medical systems unless you know how many people there are and what ails them?
Yet even this basic information can be elusive. Reliable census data, and registration of births and deaths, are taken for granted in many states. They are far from universal, however, and the more detailed data needed to track influences on health and disease can be even harder to come by.
In the Karonga district on the shores of Lake Malawi, for example – where about 220,000 people are served by a single doctor – it would be easy for an individual birth or death to escape notice. However, the area is now one of the most closely monitored sub-populations in Africa.
The Karonga Prevention Study began in 1979 as a cohort study of the kind familiar from epidemiological research in the northern hemisphere – following a particular group in the population and charting their illnesses. But the initial work, which focused on leprosy, was unusual among cohort studies in that it covered virtually everyone living in the area. Since then, the project – which received £3 million core funding from the Wellcome Trust in 2006 – has developed into a full-blown demographic surveillance system (DSS), which continues to keep detailed records of the entire population as a platform for studies of many other local health problems.
The Karonga operation is unusually large, involving more than 100 people in Malawi and at the London School of Hygiene and Tropical Medicine. However, such local population monitoring is now taking root more widely: a group of studies united under the umbrella of the INDEPTH Network currently covers 31 field sites in 17 different countries, which between them record data on 1.8 million people. Most are in sub-Saharan Africa, but other parts of Africa and Asia are also involved.
Where whole-country demographic data are not available, these smaller-scale efforts use field stations to deliver high-quality, long-term records of smaller groups. They aim to give a better picture of the disease burden, help to spot new problems such as emerging infectious diseases or drug resistance, and help other teams test out new approaches to healthcare – from new vaccines or drugs to health education and outreach in areas poorly catered for by existing medical systems.
Wherever they are, working on these projects is not like doing cohort studies in, say, the UK or the USA. Everything from training fieldworkers to data recording methods has to be tailored to local circumstances. Even establishing who lives where is a non-trivial task in many cases. Homesteads in rural areas used to be painstakingly plotted using grids overlaid on aerial photos. Nowadays, they can be pinpointed using the satellite-based global positioning system (GPS), and fieldworkers may have handheld GPS units.
Locating people, and tracking their migration, can be challenging. But trickier still is the vital issue of how they die. It is common to use specially trained workers to conduct a ‘verbal autopsy’. When they hear of a death, they track down surviving family and friends and try to persuade them to answer detailed medical, social and personal questions about the death. This is sensitive work in countries where infant mortality is often high. The routine may include careful checking that the person in need of an autopsy is actually dead, as declaring someone dead who is still living is a taboo in some communities.
Deaths aside, collecting other kinds of information also has to be done with attention to local conditions. Confidentiality and consent take more explaining when people are unfamiliar with the notion of research in the first place. Self-administered questionnaires are not much use if many people are illiterate. Follow-up has to be very active to update information in the database over time. Rural and urban populations may both be highly mobile.
People who stay in the same city may change address often, and even where phone networks operate, individual numbers frequently change. In addition, keeping the funding going for the long haul – which really generates payoffs from these projects – is, of course, another big problem.
But all these obstacles have not stopped many of the projects from delivering valuable results. The Agincourt Health and Socio-demographic Surveillance System in South Africa, for example, covers a relatively small area in the north of the country, home to 65,000 people in a score of villages. But the intensive study of these people, with data updated every year, supports a wealth of further research. Because the basic data are sound, it is easy to add new questions.
In recent years, there have been special studies of chronic cough as an indicator for tuberculosis and of disability. The whole study is linked to the Agincourt Health and Population Programme at the University of the Witwatersrand, Johannesburg, which includes work on problems as varied as violence among adults, the protein deficiency kwashiorkor among under-fives, poor health among refugees from Mozambique and how labour migration affects men’s sexual behaviour.
Projects such as the Karonga study also show how long-term follow-up and archived samples can generate new research. The team there mounted a retrospective cohort study of HIV-infected families for five years from 1996 to 2001, using 40,000 blood samples drawn in the 1980s.
Brazil, South Africa and Sweden
Countries with more developed bureaucracies and health systems than currently exist in parts of Africa still need their own cohort studies, too. The city of Pelotas in the far south of Brazil has seen the largest and longest birth cohort study in a low- to middle-income country. Cesar Victora and colleagues were worried that findings from studies in high-income countries were of doubtful use in a country where poor fetal growth, childhood infections and malnutrition were still common.
They began with the nearly 6,000 hospital births in Pelotas in 1982 and have followed up these children (and now their offspring) at intervals ever since. Again, the key feature is the richness of the data – with more than 2,000 variables recorded for each subject – and the quality of the follow-up. For men, this was made easier by the fact that Brazilian males have to report for compulsory military service at 18, so the team could concentrate on locating those who did not show up on time.
The Pelotas data have fed into research on the effects of poverty and the origins of obesity and asthma. The same team got the cash to launch two other studies, on a second birth cohort in 1993 and a third in 2004. They will soon know a lot more about intergenerational effects on health, as well as being able to monitor changes in the city’s health problems.
A similar study in South Africa began with 3,200 births in Soweto (outside Johannesburg) in a seven-week spell in 1990, just after Nelson Mandela’s release from prison. This intensive follow-up of ‘Mandela’s Children’ by Linda Richter and colleagues has attracted lots of attention nationally as an index of how the new nation is doing. More concretely, it has become an important source of data about when and how young people have sex as the country’s HIV epidemic has become more intense.
At the other extreme of life, new cohort studies are focusing on the very old in countries that established good record-keeping and health surveillance earlier on. The Uppsala Longitudinal Study of Adult Men in Sweden began compiling data on all men in the country born between 1920 and 1924. It began when the men were 50, and the survivors have been followed up at 60, 70, 77 and 82. The cohort has produced a wealth of studies on conditions such as heart disease, diabetes and bone deterioration. As with the other studies, the results will help doctors and health planners to cater for the future needs of the population, using predictions based on much better information than before.
A version of this article first appeared in ‘Wellcome Science’ (February 2007).Lead image:
Wellcome Library, London