Pitfalls in the use of register-based data for comparing adverse maternal and perinatal outcomes in different birth settings.
Jonge, A. de, Wouters, M.G.A.J., Klinkert, J., Brandenbarg, J., Zwart, J.J., Dillen, J. van, Horst, H.E. van der, Schellevis, F.G. Pitfalls in the use of register-based data for comparing adverse maternal and perinatal outcomes in different birth settings. BJOG: International Journal of Obstetrics and Gynaecology: 2017, 124, 1477-1480Download de PDF
Routinely collected registration data are an efficient source for conducting research. In particular, they are useful for studies with rare outcomes, as these require large sample sizes, which are difficult to obtain through targeted data collection. Registration data are often used to compare severe, adverse maternal and perinatal outcomes in different birth settings. However, registration data are primarily recorded for purposes other than research, such as supporting care processes. Subsequently, if these data are used for scientific studies, the analyses are, by definition, 'secondary' and therefore are likely to have more limitations compared to prospectively and purposely collected research data. We want to show how some of these limitations can lead to erroneous conclusions, using the Dutch perinatal register as an example. This may help researchers to avoid some of the pitfalls of using register based data and help service users, professionals and policy makers to recognise these limitations when interpreting study findings. This article is protected by copyright. All rights reserved.