Morten B. Andersen 2013

  Bjerregaard-Andersen, Morten. 2013. Establishing a twin cohort in Guinea-Bissau. Bandim Health Project, Statens Serum Institut. Faculty of Health Sciences, University of Southern Denmark. Odense University Hospital.


Twin studies are often used to investigate genetic vs. environmental risk factors for disease development, as well as controlling for shared environmental and genetic factors in intra-pair analyses. This is also the main reason for the rapidly growing number of twin registries.

Yet, there are major gaps in our twin knowledge. For Sub-Saharan Africa, only very limited twin data are available, with large-scale prospective twin registrations being particularly needed. This is a paradox since a considerable proportion of the world’s twins are born in Africa due to a high twinning rate. At the same time it is problematic, since the interactions between genes and environment may not be the same in Africa, partly due to high mortality, different nutrition and frequent infections. Finally, twins are at much increased risk in settings with inadequate health care.

We have therefore established a twin registry in Guinea-Bissau, West Africa. The overall purpose was to create a long-term twin registry to investigate risk factors for various diseases such as diabetes and metabolic syndrome, including the importance of the intrauterine environment and epigenetic influences. The registry has been established at the Bandim Health Project, a research station in the capital Bissau with a well-described background population.

For this PhD thesis, we aimed to investigate some basic aspects regarding twins in Guinea-Bissau during the first year of life, including twinning rate, mortality, causes of death and risk of hospitalization. Such data are required to improve twin health care, but they are equally important in characterizing our twin registry. The a priori hypothesis was that – unlike twins in high income settings – African twins may continue to have an elevated mortality into childhood due to increased risk of infections and more difficult socio-economic conditions.

A much debated matter concerning twins is the intrauterine environment, i.e. exposures to the fetus during pregnancy. Twins are very often born with low birth weight, which in numerous studies has been correlated to the development of chronic diseases later in life, e.g. diabetes and metabolic syndrome. This is typically referred to as the "fetal origins hypothesis". If twins – due to an unfavorable fetal environment – are prone to develop diabetes and other chronic conditions, twins may not be representative for the population in general. Consequently, the use of twin studies in e.g. metabolic research may be biased.

Though the debate about "fetal programming" and twins is of universal relevance, there are many aspects that are different in Africa. A substantial proportion of twins die very early and malnutrition may be common. Also, infections such as HIV and malaria may cause damage during fetal life. Findings from high income settings may therefore not be directly applicable to African twins, who experience a very different intrauterine environment. We investigated metabolic syndrome and diabetes among twins and singletons between 5-32 years in Guinea-Bissau.

Overall objective

- To establish a twin registry in Guinea-Bissau

Specific objectives in this PhD thesis

- To determine the twinning rate and perinatal twin mortality compared to singletons.

- To investigate post-perinatal infant mortality and hospitalization rate for twins compared to singletons.

- To describe the prevalence of metabolic syndrome and diabetes among young and adult twins compared to singletons.

Results and discussion

The frequency of twin births was 1.8% in the overall community, though 2.8% at the main National Hospital (likely due to specific referral of twin deliveries). The perinatal twin mortality was 22%, 2.71 (CI: 1.93-3.80) times higher than for singletons. Very low birth weight was the most important risk factor for perinatal death, but unrecognized twin pregnancy - comprising 65% of all twin pregnancies - was also of high relevance.

Excess twin mortality beyond the perinatal period continued during infancy, though predominantly in the first three months. Very low birth weight remained an important risk factor, but also loss of the co-twin perinatally, severe maternal illness during pregnancy and maternal HIV infection were associated with twin death, with the latter possibly being related to lack of breastfeeding. Sepsis, including neonatal sepsis, was a common cause of twin death after the perinatal period.

Despite higher twin mortality during the first months of life, we could not detect an elevated hospitalization rate for twins. This could be caused by difficult socio-economic conditions, but requires further investigation.

Among young and adult twins the prevalence of metabolic syndrome was 3.0%, while for singletons it was 3.6% (P=0.66). We found no diabetes cases among 574 twins, against one among 463 singletons. The prevalence of elevated HbA1c (6.0-6.4%) was 1.4% among twins vs. 2.4% among singletons (P=0.28). The average HbA1c was 5.3% for both groups, and mean HbA1c was similar for monozygotic and dizygotic twins (P=0.56). To some extent this supports large-scale register studies from Scandinavia, which found that adult twins did not have higher diabetes prevalence than singletons. Yet, comparisons should be made cautiously, both due to the young age group in our study (5-32 years), and because of the very high mortality among the weakest twins in Guinea-Bissau ("healthy survivor effect"). Further metabolic twin studies Africa are therefore warranted.


Overall, this thesis contributes unique twin data from Sub-Saharan Africa, including the consequences of being a twin in a low-resource setting with high mortality. In the future we expect that our registry can improve the understanding of the interplay between genetic, intrauterine and environmental risk factors for diabetes and other chronic conditions, including possible epigenetic influences.