Efforts to increase the inclusion of African populations in genomic studies are beginning to take shape, mainly in recognition of the importance of African populations to our understanding of human genetic variation and its impact on disease, and partly in pursuance of equity and social justice in health. Notable initiatives include the African Partnership for Chronic Disease Research (including the Uganda GWAS study and the Durban Diabetes Study) the African American Diabetes Mellitus Study, and the Human Heredity and Health in Africa Consortium, all of which aim to increase the inclusion of Africans in genomic discovery. Together, these efforts will genotype under 100 000 African individuals, far less than the couple of millions of European-ancestry individuals that make up the existing body of genomic data.
Most current genomic studies in Africa are de novo studies. This is understandable in instances where highly specific phenotypes or previously unstudied populations are the target. However, if situated within existing epidemiological research studies and programmatic health data collection platforms, many of these studies could fulfil their primary objectives while creating new opportunities to refine and extend answers to previous research hypotheses. Examples of such platforms include established prospective cohort studies, Demographic and Health Surveys (DHS), Multiple Indicator Cluster Surveys (MICS) and the WHO STEPwise approach to Surveillance (STEPS) survey programmes – all of which are well developed in many African countries with high response rates and limited migration-related attrition. These platforms have generated rich biomarker and health outcome data. However, because of their observational nature, they are inadequate for assessing causality due to confounding, reverse causality and other sources of bias. Embedding genomic studies into these platforms provides opportunities to generate genotypic data linked to existing observational health and biomarker data. This allows for casual inference through mendelian randomization (MR) techniques while achieving efficiency by saving on resource needs required for de novo studies.
MR, as an approach to causal inference, is an instrumental variable estimation technique that uses genetic markers to extract causal effects of a biomarker on an outcome in observational studies. The underpinning idea in this approach is random allocation of genetic determinants of a given biomarker at gamete formation. The import of this is that a component of the biomarker’s effect is due to a randomly set genotype (unconfounded) which is not affected by the outcome (eliminating reverse causality). Thus, by using genetic determinants of a biomarker as instrumental variables, the causal effect of the biomarker on an outcome can be determined, provided the instruments satisfy set criteria. Well conducted MR studies have been useful in establishing causal associations, notable among which is that between alcohol consumption and cardiovascular disease.
Delineating causal biomarkers and exposures is important for informing the development of more focussed and impactful interventions to improve population health. Yet, the potential of existing cohort studies, DHS, MICS and STEPS surveys in Africa have not been well recognised in ongoing endeavours to expand genomic research on the continent to achieve the same. Careful planning and integration of current efforts to expand genomic studies in Africa within existing research, health surveillance and survey platforms, provide unique opportunities to efficiently leverage and exploit available data to elucidate causal links between exposures and outcomes to improve health on the continent.