Abstracts by keywords: cluster randomized trials
Health services research can be classified into a system with several levels and uses a range of study designs at each level (Hey/Maschewsky-Schneider 2006: Chapter 6; Pfaff et al. 2011). In general, studies are important instruments for gaining knowledge in health services research (e.g. Meyer et al. 2012; Karbach et al. 2012). Study designs with several hierarchy levels (so-called multilevel data structures) are increasingly being used. An important aspect in healthcare research is the need to adjust for possible significant confounders, “since healthcare-relevant parameters are the focus of the analysis” (Hey/Maschewsky-Schneider 2006: 251). This can result in study designs with more than one hierarchy level and additional consideration of covariates. These designs are referred to below as study designs with a complex hierarchical data structure. Examples are longitudinal studies to examine time courses (e.g. body weight) with additional consideration of influencing variables (e.g. gender, age at baseline) or cluster-randomized studies with additional consideration of other cluster characteristics (e.g. proportion of migrants, baseline value of the target variable). A specific example of such a study is URMEL-ICE, a cluster-randomized trial to examine the effectiveness of a school-based program to prevent obesity in primary school children (Brandstetter et al. 2012). In addition to the group membership (intervention group vs. control group), the baseline value and time interval between measurements were also taken into account in the analysis.