235 intervention. This non-blinding leads to a risk of performance bias. This occurs when participants report improved behaviour while not changing much, which may lead to an overestimation of the effect of the intervention. However, this is unlikely to explain all effects, as for instance the measurement of hypertension is quite objective. Causality and confounding Our research has several strengths with regard to causality and confounding. First, our quasi-experimental study design allowed a thorough analysis of the effectiveness of the intervention with minimal risks of bias resulting from confounding. Second, in our analyses we took measures to prevent confounding as well. For example, we performed a multi-level analysis to determine whether a two-level structure was present, with patients nested under HCPs. In addition, we performed sensitivity analyses without imputation, and without correction of eGFR. We also performed subgroup analyses based upon health literacy, health setting and use of the intervention. The sensitivity analyses led to similar results, while subgroup analyses showed the performance of the intervention in different subgroups. Several study limitations might have influenced the results of our quasiexperimental study. First, the timeframe of nine months might have been too short to address changes, for example in eGFR or self-management. Second, compared to the general populations, our sample showed good health behaviors at baseline, which might have imposed a ceiling effect. Third, participants in the control group may have made efforts to change their behaviors, because they knew they were in a study on kidney disease and self-management. If so, this might have inflated the estimated effect of the intervention. Fourth, the process evaluation revealed 25% of the patients used only one component of GoYK, potentially limiting the effectiveness of the intervention. Fourth, results in quasi-experimental studies can always be influenced by changes in policies or procedures within the participating health care organizations. After our study, we asked HCPs if there were new care protocols introduced, and we assessed if other studies on similar topics had taken place, yielding no major changes.
RkJQdWJsaXNoZXIy MTk4NDMw