Logo: to the web site of Uppsala University

Introduction: Idiographic network analysis, assessing associations between a system of variables of interest, also called nodes, is a proposed method for personalizing psychological treatment. The aim of this study was to test the centrality hypothesis that a treatment condition delivering interventions guided by the most central network node will yield better outcomes compared to a treatment condition delivering interventions guided by the least central node. We tested this using perceived causal networks (PECAN) and focusing on psychological inflexibility processes. Effects were examined in terms of pain interference, motivation, and pain intensity.

Method: We used a single case design with multiple baselines across six participants. Therapists were blind to treatment conditions. While participants were not blind to the responses that they provided for creating the PECAN, they were blind to the resulting network and the treatment conditions. Randomization was applied to baseline length and to whether the most central node or the least central node intervention came first.

Results: All participants had at least one outcome changing in beneficial directions in line with hypotheses. However, two participants also had one outcome each that changed in contradiction to the hypotheses.

Discussion: Adapting psychological treatment by matching interventions to the most central node in a perceived causal network looks promising. However, it is unlikely that this method will always be the best matching method. We need to keep exploring additional personalization methods and under which circumstances they are efficient.

Introduction:

Idiographic network analysis, where associations between multiple nodes are estimated, can potentially guide choice of interventions in psychological treatment. In this single case experiment using ecological momentary assessment data to estimate continuous time network models, we aimed to test the so-called centrality hypothesis. We did so by comparing effects of interventions guided by the most central node to those guided by the least central node. We used the composite level psychological inflexibility processes lack of openness, lack of awareness, and lack of engagement, alongside an interference outcome, as network nodes. Effects on pain interference, motivation, and pain intensity were examined.

Method:

We employed a multiple baseline design across six participants. Therapists and participants were blinded to participants' treatment conditions. Baseline length and order of treatment phases were randomized.

Results:

Four participants had an overall treatment effect on pain interference, but it was generally not possible to discern that one particular phase was more beneficial than another. For three participants, the picture was somewhat clearer, indicating one of the treatment phases as more beneficial, although the results for these participants were not consistently in line with hypotheses. Retrospectively examining other potential guidance methods for these three participants, we saw a potential in discrete time contemporaneous network models.

Discussion:

Current results are not in line with previous assumptions or research on idiographic network models for treatment personalization, although the previous research in this area is scarce. Future research should investigate alternative network models or estimation choices to determine the potential utility of data driven idiographic networks.