• Tong Gong, Bronwyn Brew, Arvid Sjölander, and Catarina Almqvist.
• 1 Department of Medical Epidemiology and Biostatistics, Karolinska Institute, Sweden.
• Scand J Public Health. 2017 Jul 1; 45 (17_suppl): 30-35.

AimsVarious epidemiological designs have been applied to investigate the causes and consequences of fetal growth restriction in register-based observational studies. This review seeks to provide an overview of several conventional designs, including cohort, case-control and more recently applied non-conventional designs such as family-based designs. We also discuss some practical points regarding the application and interpretation of family-based designs.MethodsDefinitions of each design, the study population, the exposure and the outcome measures are briefly summarised. Examples of study designs are taken from the field of low birth-weight research for illustrative purposes. Also examined are relative advantages and disadvantages of each design in terms of assumptions, potential selection and information bias, confounding and generalisability. Kinship data linkage, statistical models and result interpretation are discussed specific to family-based designs.ResultsWhen all information is retrieved from registers, there is no evident preference of the case-control design over the cohort design to estimate odds ratios. All conventional designs included in the review are prone to bias, particularly due to residual confounding. Family-based designs are able to reduce such bias and strengthen causal inference. In the field of low birth-weight research, family-based designs have been able to confirm a negative association not confounded by genetic or shared environmental factors between low birth weight and the risk of asthma.ConclusionsWe conclude that there is a broader need for family-based design in observational research as evidenced by the meaningful contributions to the understanding of the potential causal association between low birth weight and subsequent outcomes.

11 keywords...

hide…