We believe there is an urgent need for longitudinal cohorts based in sub-Saharan Africa (SSA). This conclusion is drawn from the fact that non-communicable diseases (NCDs) cause a large and growing disease burden (please see Box 1) –.
Box 1. Selected Statistics on Non-Communicable Disease Burden in Sub-Saharan Africa
Non-communicable disease (NCD) burden
Over 80% of deaths from non-communicable diseases worldwide are estimated to occur in low- and middle-income countries .
The World Health Organization projects that the number of deaths from ischaemic heart disease in the African region will double by 2030 .
In 2004 stroke was estimated to cause 3% of all deaths in Africa and 52% of vascular deaths .
The prevalence of diabetes mellitus in Africa is predicted to increase by 80% in 20 y .
One in five deaths from NCDs in adults over 45 y in Africa is estimated to be caused by cancer .
In the past, public health in SSA has focused on communicable diseases. The advent of HIV/AIDS reinforced this image of infections as SSA's major health burden. However, NCDs, including cardiovascular diseases, mental illnesses, trauma, cancer, and diabetes, are now major sources of morbidity and mortality and are projected to overtake infectious diseases by 2030 ,. We argue that SSA lacks adequate resources to respond to this problem.
Prospective cohort studies can be used to study multiple complex diseases and risk factors simultaneously over an individual's lifetime. Such studies have proved crucial in understanding the etiology, course, and outcome of NCDs in other populations and have informed the design of prevention programs. In addition, cohort studies provide an incomparable resource for the training of public health researchers. Because the payoff from cohort studies continues—and often grows—over time, they are a long-term investment in public health.
In order to highlight the potential impact of cohort studies in SSA, we compared published literature on NCDs from longitudinal studies in high-income countries to publications from Africa. Further, we estimated the costs of establishing cohort studies in SSA and describe the response needed to correct the disparities in research investment between SSA and the world's more wealthy regions.
What We Have Learned from Prospective Cohorts in the Wealthy Countries
In 2001, Pettiti prepared a list of 55 important findings from epidemiological studies for the Epidemiology Monitor . This list, reproduced and recently revised by Dr. Pettiti (personal communication, D. Pettiti) in Table 1, represents only a partial inventory of epidemiological discoveries and does not refer exclusively to cohorts but all epidemiological designs.
Colditz and Winn recently proposed criteria by which the success of large cohort studies can be judged . The criteria they suggest are: (1) discovery, (2) development, and (3) delivery. Discovery refers to the ability to explain disease etiology and is measured by numbers of publications and the impact factor of the journals in which they are published. Development refers to the provision of a basis for developing prevention and control measures for populations at risk. Development is measured by contribution to the determination of factors causing disease, providing scientific support for prevention, clinical guidelines and randomized trials, and quantification of the preventable burden of disease. Delivery refers to the implementation of findings from the discovery and development phase by clinicians, public health practitioners, policy makers, and the general public. Delivery is measured by health policies, industry applications, and public awareness. Although these criteria were applied to studies of cancer epidemiology, they are more broadly applicable. When applied to the Nurses' Health Study (NHS), a longitudinal cohort begun in 1976 with 121,700 US female registered nurses, this one study contributed importantly according to the Colditz and Winn's criteria. Because cohorts are able to focus on multiple outcomes, they can demonstrate the full public health impact of lifestyle factors.
International Disparity in Investment in NCD Research
International Disparity in Publications
We attempted to quantify numbers of publications in the following manner. We searched the PubMed on-line database  up to November 26, 2009. We included the Medical Subject Headings (MeSH) “cohort studies” in combination with “Africa South of the Sahara” and two NCDs: “stroke” and “diabetes mellitus.” Each condition was limited to being a major subject matter. We repeated each search, changing “Africa South of the Sahara” to “United States,” “China,” “India,” and “Europe.” We used the PubMed limit function to limit each search to those articles involving humans, having an abstract, and written in English. We counted citations retrieved by these strategies.
We estimated the number of people with diabetes and the number at risk for stroke (those with hypertension) in each area from the following sources. Total population was from The WHO's World Health Statistics 2008 . The prevalence of hypertension in the US was from the National Center for Health Statistics . The prevalence of hypertension in Africa –, Europe , China , and India  was estimated from several sources. The prevalence of diabetes for all countries and regions was from the International Diabetes Federation 2009 Diabetes Atlas .
We calculated the number of publications per million people for each condition in each geographic region (Figures 1 and 2). Continental Africa produced a tiny fraction of the publications from cohort studies per affected population compared to the US: 0.6% for stroke and 8% for diabetes.
International Disparity in Adults Enrolled in Long-Term Cohorts
No central database of numbers of adults enrolled in long-term cohorts of NCDs exists. We have relied on our knowledge of the epidemiology field and that of our colleagues to estimate the proportions of populations enrolled. This method is admittedly crude and we have assuredly missed many. For instance, large populations in North America and Europe are monitored through data linkage to the health care system and are regularly used as cohorts. However, we believe that one brief example can provide a conservative estimate of the magnitude of the disparity.
In the US there are 18 cohorts and 2.9 million participants that are part of The Pooling Project .
For Africa we included the following cohorts. The Birth to Twenty Cohort studies the health of young adults born in 1990 in Johannesburg-Soweto, South Africa . The Heart of Soweto is a health facility-based South African study that enrolled 4,162 cardiovascular disease cases in 2006. The Women's Health Study of Accra, Ghana enrolled 1,328 women to study the prevalence of communicable diseases and NCDs ,. The total cohort enrollment from these three sources is 8,763.
Using the populations of the US and Africa from the WHO's World Health Statistics 2008 , we found that compared to the US, Africa had 0.1% as many cohort enrollees. Given that there are certainly more US enrollees than we have counted, and unlikely to be large African cohorts that we have missed, the disparity is likely to be even larger. Recent calls to create even more US cohorts to study gene environment interactions , would increase this disparity.
INDEPTH is an organization that evaluates population demography in developing countries, including 12 in SSA . Because they assess primarily births, deaths, migration, and not risk factors or longitudinal data, we have not included them in our count of SSA cohorts. Although some have NCD projects, a massive input of resources would be required to convert them to true prospective cohorts.
International Disparity in Research Money Spent
In 2008 the Global Forum for Health Research published the document “Monitoring Financial Flows for Health Research 2007” , the preface of which reads:
Since  the landscape of health research for development has changed in important ways:
global expenditure on health research has more than quadrupled to over US$125 billion in 2003;
there are many more actors engaged in funding or conducting health research relevant to the needs of developing countries;
but the epidemiology of diseases has shifted substantially, so that many developing countries are now experiencing high burdens of non-communicable diseases such as cancer, diabetes, heart disease, and stroke, as well as continuing high burdens of infectious diseases and injuries.
As a result of these changes, the total global expenditure applied to research relevant to all the health problems of developing countries cannot be estimated with any meaningful degree of accuracy.
Yet Chapter 5 of this same document, called “Using Bibliometrics to Inform Cancer Research and Spending,” used publication numbers and estimated cost per paper to estimate total world cancer research expenditures. This provides assurance that the disparity in publications calculated previously can provide insight into the disparity in money spent.
International Disparity in Scientific Person-Power
Economic development is intrinsically linked to science and technology produced by academic, technical, and professional institutions . Currently, many SSA countries have small scientist pools because of limited research funding and few viable career paths. Lasting, indigenous capacity building in public health requires the development of scientific training and capabilities for research.
What Africa Can Teach Us
There is no reason to doubt that smoking, obesity, high salt intake, sedentary lifestyle and pollution will have adverse health effects in Africa as in other places.
Why then do we need to do cohort studies in Africa? We suggest the following reasons: to study population-specific disease burden; genetic heterogeneity; unparalleled geographic, social and cultural diversity; practices and secular trends that may be unique risk factors; and to stimulate political will for health promoting change.
Scientific findings from North America and Europe may not be applicable to other regions. Longitudinal studies track changes in risk factors over time and provide health planners with information on disease burden. One example would be hypertension prevalence. This has been considered a problem of more wealthy, sedentary and obese urban but not rural African populations . Yet blood pressure measurements from cross-sectional studies in several rural African locations have found a hypertension prevalence of 11%–25%, similar to that in the cities and in high income countries ,. A cross-sectional study of 1,500 rural villagers over age 15 in three nations (Tanzania, Malawi, and Rwanda) found 23% had hypertension that was independent of obesity (mean body mass index was 21 kg/m2) and associated with diet . Others have confirmed these findings ,. The assessment that hypertension is mainly a problem of cities is therefore wrong. Global dietary changes will affect the trajectory of hypertension prevalence in the future, and cohorts will give information on this trajectory that the cross-sectional studies cannot.
Diet is another location-specific cultural practice. Although we know from studies in high income countries that consumption of whole grain based on mostly wheat and rice is associated with lower risk of diabetes , this may not be applicable in East Africa where the staple is corn. We need to research local foods that confer protection to formulate interventions.
History is also location specific. With the exception of studies of starved Europeans during World War II , studies of the West's obesity epidemic have all been among populations with a lifetime of more than adequate energy intake. In Africa, under-nutrition remains common  yet coexists with the global obesity epidemic ,,. Evidence from high income countries shows that intrauterine growth retardation predisposes to increased risk of cardiovascular and metabolic diseases –. The effects of the collision of these two secular trends in Africa (recent under-nutrition and rapid introduction of obesity-enhancing lifestyle factors) cannot be fully anticipated.
The high prevalence of infectious diseases such as HIV in SSA along with NCDs is another location-specific opportunity. Inflammation common to infections is increasingly known to be important for cardiovascular, metabolic, and neoplastic disease –. SSA presents the opportunity to better understand this pathophysiology.
Adding to the complexity of locale-specific behavior and history is genetic diversity. Humans first evolved in East Africa, and thus African genetic diversity is greater than elsewhere . Studying the interaction between environmental factors and genes elucidates disease mechanisms. If one multiplies the extended range of cultural practices and historical circumstances by the genetic diversity found in Africa, the opportunities for scientific discovery become enormous (S. Dalal, M.D. Holmes, R. Ramesar. Advancing public health genomics in Africa through prospective cohort studies. Journal of Epidemiology and Community Health [In press]).
Even if smoking and intake of salt and refined grains has been shown detrimental in higher income populations, political will for change may be greater among policy makers, clinicians, and the public if shown in their own population.
Although cohort methodology developed in North America and Europe, high income countries may also gain from knowledge derived from African studies. For example, the obesity epidemic in the West is well developed; we have yet to apply our knowledge to interventions having substantial impact .
Estimated Cost of African Cohorts
We propose cohort studies of 50,000 to 100,000 people in three or four countries covering west, east, and southern Africa. The higher number would increase enrollment of Africans to 532 per million, still only 5% compared to the US.
These studies will include the collection and storage of biological specimens, repeated exposure measurements, and ongoing ascertainment of disease among participants. Disease ascertainment poses particular challenges in countries without strong health systems. It will be important to choose strategies that accommodate local circumstances.
We propose to include populations that are urban, rural, and include men, women, and children. We have four overarching goals: (1) to measure disease burden; (2) to study the etiology of NCDs, the distribution of risk factors, and attributable risks; (3) to provide short-term results by nesting randomized intervention trials within the cohorts; and (4) to stimulate academic development and collaboration.
Ideal study sites would be those where African colleagues have some history of cooperation with the population. Sites with established assessments or interventions may offer cost savings. However, such sites are not likely to easily accommodate the additional proposed layers of complexity, including multiple research designs, multinational collaborators, and managing new enrollees with ongoing study participants separately.
We estimated costs for establishing cohorts in two ways. The first was by comparison to the mail-based NHS costs for April 1, 2005 to March 31, 2006 in Table S1 (personal communication, S. Hankinson). Personnel costs are lower in Africa than in the US. We used the Source of International Wage Comparisons compiled by the financial firm UBS to calculate wage differences (Table S2) . Using this method, we estimated the cost of an African cohort of 100,000 participants to be $1.2 million per year, or $12/person/year (Table S3).
Costs could be higher without a reliable mail system; thus we estimated African cohort costs a second way. We applied the costs of a cohort of 775 patients followed for HIV medication compliance in rural Uganda (personal communication, D. Bangsberg). One year costs were $494,256, or $638/person/year.
Randomized Trial Costs
We used information from a randomized controlled trial (RCT) of single versus multiple doses of multivitamins in addition to ART for women with HIV in Tanzania to estimate the cost of a RCT (personal communication, W. Fawzi). The cost was $4 million for 5 y for 3,000 people. However, this included multiple laboratory measurements because participants were ill. Assuming half the cost for an intervention in well people, we estimate RCT costs to be as low as $135/person/year. We applied a range of $135–$270/person/year to our estimates.
The prevalence of hypertension in Africa is similar to the US ,. The Dietary Approaches to Stop Hypertension (DASH) trial significantly lowered blood pressure among 459 hypertensive Americans . An illustrative intervention would be to do an “African DASH” trial substituting culturally appropriate foods to lower blood pressure among 500 Africans lasting 2 y. We propose doing a total of five similar interventions addressing diseases of the epidemiologic transition over 10 y. The costs for 500 people×2 y×($135–$270)×5 similar type interventions = $675,000 to $1.35 million.
We assumed the costs of educating African scholars using tuition and living expenses at Harvard School of Public Health ,. We propose other concomitant models for training including lecturers in residence at African universities, short courses, and distance learning. We assumed a range of costs depending on the degree program (master's or doctorate) and whether the student was single or had a family. The cost per student given in Table S4 ranges from $110,000 to $269,000.
Total Cost Scenarios
Table S5 summarizes the assumptions of the range of costs explained in this section. We assumed cohorts would extend at least 10 y; in a recent review successful large cohorts experienced exponential publication growth after 10 y . We then applied high and low costs for the cohort, interventions, and for training, and combinations thereof in Table 2. The low estimate including interventions and training is $23.7 million for a cohort of 50,000 people in three countries each. The high estimate is for $2.56 billion for a cohort of 100,000 people in four countries each.
Comparisons are useful. PEPFAR cost $18.8 billion for 5 y only for HIV/AIDS . The Bill and Melinda Gates Foundation will spend $3.9 billion in 2009 on various causes including global health . Harvard University's fundraising for fiscal year 2006 was $600 million .
The UK Biobank, a cohort of 500,000 people with a baseline assessment only and 8 y of follow-up linked to a national health record, will cost $104 million . The similar Swedish LifeGene cohort of 500,000 people and with baseline assessment and 8 y of follow-up will cost $112 million (personal communication H.-O. Adami). Translated into costs per person per year, the UK Biobank is $26 and LifeGene is $28. The high and low estimates of the African initiative gives a range from $14 to $644/person/year for the added value beyond the European cohorts of repeated exposure measurements, embedded interventions, and academic capacity building. Other scenarios of different size and intensity could be entertained. For instance, a large company that already conducts periodic health examinations of employees might be economically converted to a cohort. Also, funding need not necessarily come from a single source. However, some costs for an African cohort will be higher than a similar cohort in Europe or North America. A generous budget for travel will be needed to allow newly trained personnel to return to the host institution for further contact and exchange of ideas, but also to enable experienced researchers to visit the sites frequently.
Yach defined the following obstacles impeding attention to NCDs in developing countries: policy makers lack evidence of disease burden, beliefs that NCDs affect only the wealthy and elderly, that they arise only from freely acquired risks, and that their control is expensive, ineffective, and a lower priority than infectious diseases .
These obstacles will need to be addressed if cohort studies of NCDs are to be sustainable. Health systems will have to be strengthened to absorb the disease burden uncovered by research. Training programs in public health disciplines are also necessary for long-term career development for the next generation of researchers.
We believe that starting large cohort studies in Africa is a 21st century design to serve public health in Africa. It combines surveillance, epidemiology, and clinical trials, and could yield information relevant to the NCD epidemic in wealthy countries. The funding for such an initiative will require political leadership, a long vision of the possibilities, and patience in awaiting results. We hope that the growing burden of NCDs and the unfulfilled need for research and action will spur funders of global health to include NCDs in Africa in their agenda.
2001 Noncommunicable diseases in sub-Saharan Africa: where do they feature in the health research agenda? Bull World Health Organ 79 947 953
2007 The burden and costs of chronic diseases in low-income and middle-income countries. Lancet 370 1929 1938
2008 Ischaemic heart disease in Africa. Heart 94 836 843
2007 Burden of stroke in black populations in sub-Saharan Africa. Lancet Neurol 6 269 278
5. 2009 International Diabetes Foundation 2009 Diabetes Atlas. Accessed Nov 25, 2009 http://www.diabetesatlas.org/map
2008 Part I: cancer in indigenous Africans–burden, distribution, and trends. Lancet Oncol 9 683 692
2004 The global burden of chronic diseases: overcoming impediments to prevention and control. JAMA 291 2616 2622
2006 Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med 3 e442 doi:10.1371/journal.pmed.0030442
2001 Epidemiology triumphs and stories of discovery. Accessed March 24, 2009 http://www.epimonitor.net/EpiMonday/Triumph62501.htm. In: Epimonitor.net, editor
2008 Criteria for the evaluation of large cohort studies: an application to the nurses' health study. J Natl Cancer Inst 100 918 925
11. PubMed. Accessed May 14, 2009 http://www.ncbi.nlm.nih.gov/sites/entrez. National Center for Biotechnology Information
12. World Health Statistics 2008 Accessed May 14, 2009 http://www.who.int/whosis/whostat/2008/en/index.html. World Health Organization
13. Health, United States, 2007, Table 70. Accessed May 18, 2009 http://www.cdc.gov/nchs/data/hus/hus07.pdf#070. National Center for Health Statistics, Centers for Disease Control
2004 Prevalence, detection, management, and control of hypertension in Ashanti, West Africa. Hypertension 43 1017 1022
2007 Hypertension in sub-Saharan Africa: a systematic review. Hypertension 50 1012 1018
2008 Hiding in the shadows of the HIV epidemic: obesity and hypertension in a rural population with very high HIV prevalence in South Africa. J Hum Hypertens 22 236 239
2008 Body size and blood pressure: an analysis of Africans and the African diaspora. Epidemiology 19 38 46
2005 Hypertension in sub-Saharan African populations. Circulation 112 3562 3568
2003 Hypertension prevalence and blood pressure levels in 6 European countries, Canada, and the United States. JAMA 289 2363 2369
2002 Prevalence, awareness, treatment, and control of hypertension in china. Hypertension 40 920 927
2004 Trends in hypertension epidemiology in India. J Hum Hypertens 18 73 78
22. The Pooling Project of Prospective Studies of Diet and Cancer. Accessed March 24, 2009 http://www.hsph.harvard.edu/poolingproject/. President and Fellows of Harvard College
23. Birth to Twenty. Accessed August 21, 2008 http://web.wits.ac.za/Academic/Health/Research/BirthTo20/. Witswatersrand, South Africa: University of the Witswatersrand
2007 Health of urban Ghanaian women as identified by the Women's Health Study of Accra. Int J Gynaecol Obstet 99 150 156
2007 Results of the Women's Health Study of Accra: assessment of blood pressure in urban women. Int J Cardiol 117 115 122
2004 The case for a US prospective cohort study of genes and environment. Nature 429 475 477
2004 Toward the last cohort. Cancer Epidemiol Biomarkers Prev 13 895 897
28. INDEPTH Network. Accessed May 15, 2009. http://www.indepth-network.org/
29. 2008 Monitoring financial flows for health research 2007 - behind the global numbers. Accessed Sept. 19, 2008 http://www.globalforumhealth.org/filesupld/monitoring_financial_flows_07/GlobalForum_xMonitoringFinancial%20Flows2007.pdf. In: Burke MA, de Francisco A, Matlin S, editors: Global Forum for Health Research
2005 Innovation: applying knowledge in development. Accessed May 14 2009 http://www.unmillenniumproject.org/documents/Science-complete.pdf. In: 2005 UNMP, editor: Task Force on Science, Technology and Development, United Nations Development Programme
31. Stewart de RamirezS
2008 Prevalence and correlates of hypertension in rural East Africa Boston MA Harvard Medical School
32. de MunterJS
2007 Whole grain, bran, and germ intake and risk of type 2 diabetes: a prospective cohort study and systematic review. PLoS Med 4 e261 doi:10.1371/journal.pmed.0040261
2008 Prenatal exposure to the Dutch famine is associated with a preference for fatty foods and a more atherogenic lipid profile. Am J Clin Nutr 88 1648 1652
2008 Maternal and child undernutrition: global and regional exposures and health consequences. Lancet 371 243 260
2007 Poverty, malnutrition, underdevelopment and cardiovascular disease: a South African perspective. Cardiovasc J Afr 18 321 324
2002 Where does the black population of South Africa stand on the nutrition transition? Public Health Nutr 5 157 162
2000 The role of size at birth and postnatal catch-up growth in determining systolic blood pressure: a systematic review of the literature. J Hypertens 18 815 831
2004 Birth weight and subsequent cholesterol levels: exploration of the “fetal origins” hypothesis. JAMA 292 2755 2764
2007 Is birth weight a risk factor for ischemic heart disease in later life? Am J Clin Nutr 85 1244 1250
2003 Is birth weight related to later glucose and insulin metabolism? A systematic review. Diabet Med 20 339 348
2001 Is obesity an inflammatory condition? Nutrition 17 953 966
2008 Evidence of systemic inflammation and estimation of coronary artery disease risk: a population perspective. Am J Med 121 S15 S20
2006 Chronic inflammation: a common and important factor in the pathogenesis of neoplasia. CA Cancer J Clin 56 69 83
1997 Alu insertion polymorphisms and human evolution: evidence for a larger population size in Africa. Genome Res 7 1061 1071
2006 Understanding and addressing the epidemic of obesity: an energy balance perspective. Endocr Rev 27 750 761
47. Prices and earnings: a comparison of purchasing power around the globe 2003 Accessed March 24, 2009 www.ubs.com/1/ShowMedia/media_overview/media_global/mediareleases?contentId=34588&. Zurich: UBS
1997 A clinical trial of the effects of dietary patterns on blood pressure. DASH Collaborative Research Group. N Engl J Med 336 1117 1124
49. Student Financial Services: Tuition and Fees from July 2008–June 2009. Accessed Jan. 5, 2009 http://www.hsph.harvard.edu/administrative-offices/student-financial-services/tuition-fees/tuition-and-fees-from-july-2008-june-2009/index.html. Harvard School of Public Health
50. Student Financial Services: Student budgeting. Accessed Jan. 5, 2009 http://www.hsph.harvard.edu/administrative-offices/student-financial-services/student-budget/. Harvard School of Public Health
51. Celebrating life: the U. S. President's emergency plan for AIDS relief: 2009 annual report to Congress highlights. Accessed May 14, 2009 http://www.pepfar.gov/documents/organization/113878.pdf. President's Emergency Plan for AIDS Relief
2009 Gates group plans to give more in 2009 despite losses. The New York Times New York The New York Times Company http://www.nytimes.com/2009/2001/2027/us/2027gates.html?_r=2001&ref=us
2006 Harvard fundraising reaches $595M in fiscal year '06. Harvard University Gazette: Harvard University. http://www.news.harvard.edu/gazette/2006/2009.2014/2099-fundraising.html
54. UK Biobank Limited report and financial statements for the year ended 30 September 2007. Accessed March 25, 2009 http://www.ukbiobank.ac.uk/docs/Annualreport.pdf. Cheshire