Multivariate determinants of self-management in Health Care: Assessing Health Empowerment Model by comparison between structural equation and graphical models approaches

Background: In public health one debated issue is related to consequences of improper selfmanagement in health care. Some theoretical models have been proposed in Health Communication theory which highlight how components such general literacy and specific knowledge of the disease might be very important for effective actions in healthcare system. Methods: This paper aims to investigate the consistency of the Health Empowerment Model by means of both a graphical models approach, which is a “data driven” method and a Structural Equation Modeling (SEM) approach, which is “theory driven”, showing the different information pattern that can be revealed in a health care research context. The analyzed dataset provides data on the relationship between the Health Empowerment Model constructs and the behavioral and health status in 263 chronic low back pain (cLBP) patients. We used the graphical models approach to evaluate the dependence structure in a “blind” way, thus learning the structure from the data. Results: From the estimation results dependence structure confirms links design assumed in SEM approach directly from researchers, thus validating the hypotheses which generated the Health Empowerment Model constructs. ConclusionS: This models comparison helps avoiding confirmation bias. In Structural Equation Modeling, we used SPSS AMOS 21 software. Graphical modeling algorithms were implemented in a R software environment.