A network approach to the analysis of psychosocial risk factors and their association with health

Network analysis to examine sex differences linked to emotional well-being in cardiovascular disease

Psychosocial dimensions are essential to guarantee an optimal approach to improve emotional well-being in patients with cardiovascular disease (CVD). There is evidence of sex differences regarding these dimensions. Thus, the connections between them are crucial to implement personalized therapies. Network model analyses were conducted on data from 593 CVD patients. The models included scores from the Hospital Anxiety and Depression Scale (HADS), positive (PA) and negative affect (NA), positivity (PS), satisfaction of life (SLS), social support (SS), self-efficacy on emotion regulation (RESE), cardiac self-efficacy (CSE) and the Short Form-12 Health Survey. The main sex differences were found in: PA-PS (p = 0.03), SS-RESE (p = 0.04), for which the positive associations are stronger for men than for women and PA-RESE (p < 0.01) for which the positive association is stronger for women than for men. These nuances should be considered to implement tailored and integrative therapies for each CVD patient.

Intersecting Relationships of Psychosocial and Structural Syndemic Problems Among People with HIV in South Africa: Using Network Analysis to Identify Influential Problems

In South Africa, little is known about interrelationships between syndemic problems among people with HIV (PWH). A better understanding of syndemic problems may yield important information regarding factors amenable to mitigation. We surveyed 194 PWH in Khayelitsha, outside of Cape Town, South Africa. We used network analysis to examine the frequency of 10 syndemic problems and their interrelationships. Syndemic problems among PWH in South Africa were common; 159 (82.8%) participants reported at least 2 co-occurring syndemic problems and 90 (46.9%) endorsed 4 or more. Network analysis revealed seven statistically significant associations. The most central problems were depression, substance use, and food insecurity. Three clusters of syndemics were identified: mood and violence; structural factors; and behavioral factors. Depression, substance use, and food insecurity commonly co-occur among PWH in sub-Saharan Africa and interfere with HIV outcomes. Network analysis can identify intervention targets to potentially improve HIV treatment outcomes.

Psychosocial work environment as a dynamic network: a multi-wave cohort study

While characteristics of psychosocial work environment have traditionally been studied separately, we propose an alternative approach that treats psychosocial factors as interacting elements in networks where they all potentially affect each other. In this network analysis, we used data from a prospective occupational cohort including 10,892 participants (85% women; mean age 47 years) and repeated measurements of seven psychosocial work characteristics (job demands, job control, job uncertainty, team climate, effort-reward imbalance, procedural justice and interactional justice) assessed in 2000, 2004, 2008 and 2012. Results from multilevel longitudinal vector autoregressive models indicated that job demands as well as interactional and procedural justice were most broadly associated with the subsequent perceptions of the work-related psychosocial factors (high out-Strength), suggesting these factors might be potentially efficient targets of workplace interventions. The results also suggest that modifying almost any of the studied psychosocial factors might be relevant to subsequent perceptions of effort-reward imbalance and interactional justice at the workplace.

Discovery of Depression-Associated Factors From a Nationwide Population-Based Survey: Epidemiological Study Using Machine Learning and Network Analysis

BACKGROUND

In epidemiological studies, finding the best subset of factors is challenging when the number of explanatory variables is large.

OBJECTIVE

Our study had two aims. First, we aimed to identify essential depression-associated factors using the extreme gradient boosting (XGBoost) machine learning algorithm from big survey data (the Korea National Health and Nutrition Examination Survey, 2012-2016). Second, we aimed to achieve a comprehensive understanding of multifactorial features in depression using network analysis.

METHODS

An XGBoost model was trained and tested to classify "current depression" and "no lifetime depression" for a data set of 120 variables for 12,596 cases. The optimal XGBoost hyperparameters were set by an automated machine learning tool (TPOT), and a high-performance sparse model was obtained by feature selection using the feature importance value of XGBoost. We performed statistical tests on the model and nonmodel factors using survey-weighted multiple logistic regression and drew a correlation network among factors. We also adopted statistical tests for the confounder or interaction effect of selected risk factors when it was suspected on the network.

RESULTS

The XGBoost-derived depression model consisted of 18 factors with an area under the weighted receiver operating characteristic curve of 0.86. Two nonmodel factors could be found using the model factors, and the factors were classified into direct (P<.05) and indirect (P≥.05), according to the statistical significance of the association with depression. Perceived stress and asthma were the most remarkable risk factors, and urine specific gravity was a novel protective factor. The depression-factor network showed clusters of socioeconomic status and quality of life factors and suggested that educational level and sex might be predisposing factors. Indirect factors (eg, diabetes, hypercholesterolemia, and smoking) were involved in confounding or interaction effects of direct factors. Triglyceride level was a confounder of hypercholesterolemia and diabetes, smoking had a significant risk in females, and weight gain was associated with depression involving diabetes.

CONCLUSIONS

XGBoost and network analysis were useful to discover depression-related factors and their relationships and can be applied to epidemiological studies using big survey data.

Psychosocial assessment of families caring for a child with acute lymphoblastic leukemia, epilepsy or asthma: Psychosocial risk as network of interacting symptoms

The purpose of this study is to assess psychosocial risk across several pediatric medical conditions and test the hypothesis that different severe or chronic pediatric illnesses are characterized by disease specific enhanced psychosocial risk and that risk is driven by disease specific connectivity and interdependencies among various domains of psychosocial function using the Psychosocial Assessment Tool (PAT). In a multicenter prospective cohort study of 195 patients, aged 5–12, 90 diagnosed with acute lymphoblastic leukemia (ALL), 42 with epilepsy and 63 with asthma, parents completed the PAT2.0 or the PAT2.0 generic version. Multivariate analysis was performed with disease as factor and age as covariate. Graph theory and network analysis was employed to study the connectivity and interdependencies among subscales of the PAT while data-driven cluster analysis was used to test whether common patterns of risk exist among the various diseases. Using a network modelling approach analysis, we observed unique patterns of interconnected domains of psychosocial factors. Each pathology was characterized by different interdependencies among the most central and most connected domains. Furthermore, data-driven cluster analysis resulted in two clusters: patients with ALL (89%) mostly belonged to cluster 1, while patients with epilepsy and asthma belonged primarily to cluster 2 (83% and 82% respectively). In sum, implementing a network approach improves our comprehension concerning the character of the problems central to the development of psychosocial difficulties. Therapy directed at problems related to the most central domain(s) constitutes the more rational one because such an approach will inevitably carry over to other domains that depend on the more central function.

Adverse childhood experiences (ACEs) and cardiovascular development from childhood to early adulthood: study protocol of the Niagara Longitudinal Heart Study

INTRODUCTION

Recent reviews have found substantial links between a toxic childhood environment including child abuse and severe household dysfunction and adult cardiovascular disease (CVD). Collectively referred to as adverse childhood experiences (ACEs), this toxic environment is prevalent among children, with recent Canadian estimates of child abuse at 27%-32%, and severe household dysfunction at 49%. Based on these prevalence rates, the potential effect of ACEs on CVD is more significant than previously thought. Yet, how ACEs amplify the risk for later CVD remains unclear. Lifestyle risk factors only partially account for this connection, instead directing attention to the interaction between psychosocial factors and physiological mechanisms such as inflammation. The Niagara Longitudinal Heart Study (NLHS) examines how ACEs influence cardiovascular health (CVH) from childhood to early adulthood. Integrating the stress process and biological embedding models, this study examines how psychosocial and physiological factors in addition to lifestyle factors explain the relationship between ACEs and CVH.

METHODS

This follow-up study combines three baseline studies from 2007 to 2012 that collected CVH measures including child blood pressure, heart rate, left ventricular structure and function, arterial stiffness indices and baroreflex sensitivity on 564 children. Baseline data also include anthropometric, biological, lifestyle, behavioural, and psychosocial measures that varied across primary studies. Now over 18 years of age, we will recruit and retest as many participants from the baseline studies as possible collecting data on ACEs, CVH, anthropometric, lifestyle and psychosocial measures as well as blood, saliva and hair for physiological biostress markers.

ETHICS AND DISSEMINATION

Ethics approval has been received for the NLHS follow-up. Written consent to participate in the follow-up study is obtained from each participant. Results testing all proposed hypotheses will be submitted for publication in peer-reviewed journals.