How to Deal with Proxy-Reports

Children are not always able to fully respond to all questions posed in medical research due to, for example, limited understanding of the queried items or limitations in expressing their perception or views. This is clearly true for newborns and infants and likely true for toddlers but may also apply to children at older ages. In pediatric epidemiological research, it is thus customary to obtain proxy-reports rather than self-reports, most often from parents or explicitly from the mother. This chapter describes important aspects of proxy-reports that should be considered in pediatric epidemiology, including proxy perspective and proxy selection as well as quality of proxy-reports, their agreement with self-reports, and their influence on missing data, on information bias and on confounding. All interpretation of proxy-reports is context-dependent. The proxy perspective should be clearly defined and the best available proxy to inform about the items requested should be selected. Here, characteristics of the index subject, the proxy, and their relationship have to be taken into account. Ideally, pilot studies should be conducted to determine feasibility of proxy selection and quality of their reports.