Many birth cohorts have been established worldwide. Some are general cohorts with multiple aims, while others are focused on e.g. environmental factors or specific disease risk factors. Some have been following children for many years, whereas others have just started. Some collect biological samples aimed at investigating biomarkers of exposure, susceptibility or response, as well as genetic determinants of disease; others do not collect any samples.
It is becoming increasingly clear that the full potential of the cohorts can only be realized through collaboration between them. A prerequisite for collaboration is well-documented cohorts. Furthermore information about design and data on the existing cohorts should be collected in a comparable form and be easily accessible. This website aims to serve this purpose.
Collaboration will enhance the scientific output from individual birth cohorts and their ability to influence health policy in the following ways:
Understanding inequalities in disease and health related behaviours
- Differences in prevalence of important health problems, such as those seen for childhood obesity, suggest that interventions may reduce these prevalences. Comparative studies taking advantage of existing birth cohort studies may provide important clues.
Greater and more efficient use of existing cohorts
- Individual birth cohorts commonly have many hundreds of different variables on individuals collected repeatedly. The potential contribution to science of these data go well beyond what any group of principal investigators could imagine doing. Collaboration provides opportunity for a wider range of expertise to address research questions of public health interest.
Improving causal inference through cross-cohort comparisons, replications, Mendelian randomization and family designs
- An important limitation to causal inference in any cohort study is potential confounding. As different populations have different distributions of socioeconomic factors and different cultures and lifestyles, confounding structures will differ markedly between the cohorts.
Associations found to be similar between cohorts provide more confidence that an observed association is not driven by confounding factors. Similarly, Mendelian randomization studies and family based studies are valuable for making causal inference, including in the area of developmental origins of later disease outcomes. However, these studies require very large sample sizes, which can only be achieved by collaboration.
Improving statistical power
- Large sample sizes are required for some of the new methods being used to strengthen detection of causal inference. In addition very large sample sizes are required to study rare but important disease outcomes in infancy and childhood such as congenital anomalies and childhood cancers. To determine child health and disease status one must fully understand how risk factors interact with each other (including gene-environment interactions). This requires very large sample sizes and necessitates large collaborations between cohorts with all relevant data.
Improving methodological approaches including protocols of biological and environmental sample collection and analysis
- By sharing ideas and know-how scientists involved with birth cohorts may improve the efficiency in data collection and analyses. We need not to replicate mistakes.