A protocol for assessing stability and invariance properties of two commonly used outcome measures in anxiolytic clinical trials

<ns3:p>Background Research has suggested that network analysis can be used to identify important pathology symptoms and inform targeted treatment plans that could lead to more efficacious outcomes in clinical trials. However, unless it can be demonstrated that network models are stable, including when accounting for moderating variables, NA-derived treatment plans may not be appropriate to implement. Aim We aim to assess the stability and invariance properties of two commonly used anxiety outcome measures to determine the suitability of NA methods to inform treatment plan design in clinical settings. Method Individual participant data (IPD) for large multi-trial samples will be accessed via Vivli.org. Exploratory graphical analysis will be used to model empirical networks pre- (baseline) and post-treatment (outcome) for the two most commonly used outcome measures in antidepressant clinical trials, namely the Hamilton Rating Scale for Anxiety (HAM-A) and the anxiety subscale of the Hospital Anxiety and Depression Scale (HADS_A). Bootstrapping and permutation techniques will be used to determine the stability and invariance properties of empirical networks in relation to a range of moderating variables, such as age, sex, treatment type and symptom severity. For networks that are unstable or partially invariant, we will examine item redundancy and remove non-performing items to pursue stable/invariant abbreviated models. Discussion This study will determine the suitability of applying NA methods in clinical trials. Findings could inform the way in which clinical trials, and other such research, are conducted. If outcome measures are stable and invariant, then NA methods will have demonstrable utility to inform more efficacious treatment plans. However, if NA is not found to be suitable, its validity as a robust analytical approach will be questionable.</ns3:p>