Ecological momentary assessment of depressive symptoms and repetitive negative thinking in a randomized controlled treatment trial

Ecological Momentary Assessment (EMA) through mobile devices can provide new insights into the dynamic change of psychiatric disorders during the course of psychological treatments (Colombo et al., 2019; Shiffman, Stone, & Hufford, 2008; Vos et al., 2015). Especially affective or cognitive symptoms such as mood or Repetitive Negative Thoughts (RNT) (Ehring & Watkins, 2008) are characterized by high variability within days and across multiple days (McConville & Cooper, 1996; Peeters, Berkhof, Delespaul, Rottenberg, & Nicolson, 2006; Starr & Davila, 2012). It is therefore assumable that EMA allows for more reliable estimations of symptom change than standard questionnaires (SQs) that are collected less frequently and are currently the state of the art in psychological research. On the other side, EMA usually contains fewer items than SQs, because time and cognitive load on the respondent must be kept low, which in turn may lower the reliability and validity of a measuring instrument. So far, it is not clear whether outcome measures by EMA are more precise than SQs. To our knowledge, there is only one study (Moore, Depp, Wetherell, & Lenze, 2016) that has empirically investigated differences between EMA and SQs as measuring instruments for treatment response. The results indicate that treatment efficacy largely differs depending on which measurement method is used. Especially for depressive symptoms, EMA turned out to reveal higher effect sizes than SQs. When looking at the internal consistency of items, which is an indicator for the precision of a measure, EMA-items tended to show higher internal consistency than the same items measured with SQs, although this was not significant. Also, consistent with other studies, they found that correlations between EMA and SQ items were only of small to medium size. A possible explanation for these findings is that EMA is more sensitive to detect change in depressive symptoms than SQs, which are less frequently conducted and are known to be biased by recall. In this way, EMA can provide more precise estimations of treatment outcomes, which results in higher effect sizes. Besides the precision, the power to predict clinically relevant outcomes is an important evaluation criterion for measuring instruments. Therefore, building on Moores (2016) findings, we are interested in whether treatment outcomes measured by EMA can predict change in global functioning measured by clinical interviews better than SQs. Within an randomized clinical trial (RCT) investigating treatment efficacy of cognitive behavioral therapy (CBT), schema therapy (ST) and individual supportive therapy (IST) for depression, we assess self-reports about depressive symptoms and RNT three times daily with EMA in addition to weekly questionnaire assessments. RNT is a transdiagnostic factor with a central role in depression and is known to have a significant temporal fluctuation (Takano & Tanno, 2011). Further, we assess global functioning gobal functioning in a clinical interview pre- and posttreatment. Similar to Moore et al. (2016) we also calculate treatment effects separately for EMA and SQ data and compare the results.