Centre for Research Ethics & Bioethics

Ethics at the Beginning of Life, 2003

The V Annual Swedish Symposium on Biomedicine, Ethics and Society was held on June 2-3 in Sandhamn.

Predicting a risk and protection in a longitudinal study - meaning of prenatal and early life factors

Marjo-Riitta Järvelin, MD, MSc, PhD
Professor in Lifecourse Epidemiology
Department of Epidemiology and Public Health,
Imperial College London,
University of London, UK, and
Department of Public Health and General Practice,
University of Oulu, Finland

Abstract of keynote lecture

Many cross-sectional studies on the possible causes and markers of the most important chronic diseases have been carried out, but there are few longitudinal, general population based studies from early life exploring the risk for or protection from a disease or disability. The use of biological markers such as height and weight as indicators of the physical and economic health of a society has a long history. There are notes in the literature that birthweight was seen as a key indicator of population health already in the 19th century. In 1913 Pedagogical Anthropology, Maria Montessori referred to birthweight as an indicator of "the hygiene of generation" and the weight gain of the new born child as "a valuable prognostic of the child's life". Most research investigating prenatal and early life determinants and later disease risk, in terms of the foetal programming theory, has focused on cardiovascular disorders. There is increasing - though controversial - evidence that both markers of the foetal environment and childhood factors predict not only cardiovascular morbidity and mortality, but also levels of intermediate cardiovascular disease risk factors, and other chronic diseases such as mental diseases.

Previous longitudinal studies on adults have been based mainly on historical cohorts or on prospective "follow-back" designs, largely due to the fact that prospective studies are expensive and time consuming. These studies have, however, a number of limitations such as the retrospective ascertainment of data, possible misclassification, recall and selection bias, small sample size and inconsistency in the inclusion of covariates; particularly an incomplete or inadequate consideration of confounding operating throughout the lifecourse. Most prospective studies from birth or early childhood with a few exceptions have looked at relatively young ages, also these design are vulnerable to specific biases (losses to follow-up). The interpretation of the reported studies is further complicated by publication bias; this issue has recently been seriously addressed.

The "magnitude" of the relative risk or protection is affected by multiple factors, and especially in longitudinal studies, which have been continuing for tens of years, this is an even more complicated issue than in other studies. The high profile of genetic research will also increasingly influence longitudinal epidemiology. It is already noticeable that relative risks only a little above unity for a genetic marker get treated with considerably more excitement than if similar relative risks were associated with other exposures. The interaction of genomic and environmental factors is also much emphasized; this should not obscure the need to sometimes stand back and observe the larger view. The researcher is faced by clinical, research and social ethics when conducting and reporting these studies: How reliable and complete is the data? What is justified to report and when? Can we generalize what we have found? What is the risk at the individual level or how much of the variation in the response can be explained by the factor in question? In many occasions only a small fraction. What about population attributable risk? In population terms a rare exposure with a high associated relative risk may be less serious in the total number or proportion of disorders/deaths that it will cause than a very common exposure with a lower associated relative risk.

The varying evidence from our and other's studies that the foetal environment and early life circumstances predict the risk of chronic illness and (neuro)behavioural disorders in later life gives rise to many questions:

1. Whether birth measures and other early life factors are truly associated with outcomes or whether the observed correlations are due to incomplete or inappropriate control for confounding. If the association exists what kind of association is it?
2. To what extent does the correlation between fetal or early childhood factors and later morbidity contribute to, and how is it affected by, phenomena that take place later in life?
3. To what extent do fetal environment and perinatal outcomes act on adult health through independent or intermediary mechanisms? One of the most important current questions is the relative significance of genetic factors.
4. What are the proportional effects of biological and social factors and their mechanisms of action?

Organizers

The Vth annual Swedish symposium on Biomedicine, Ethics and Society was organized with financial support from the Swedish Foundation for Strategic Research (ELSA program), the Foundation for Health Care and Allergy Research, Ethics in Health Care Programme and the Swedish Research Council.

More information:

Centre for Research Ethics and Bioethics
Uppsala Science Park
SE-751 85 Uppsala
Fax +46 18 50 64 04

Josepine Fernow, Co-ordinator, josepine.fernow@crb.uu.se
Associate Professor Anna T. Höglundl, anna.hoglund@crb.uu.se