Path Analysis Models Integrating Psychological, Psycho-physical and Clinical Variables in Individuals With Tension-Type Headache

• Published in: The journal of pain Vol. 24; no. 3; pp. 426 - 436
• Main Authors: Liew, Bernard X.W., Palacios-Ceña, María, Scutari, Marco, Fuensalida-Novo, Stella, Guerrero-Peral, Angel, Ordás-Bandera, Carlos, Pareja, Juan A., Fernández-de-las-Peñas, César
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.03.2023
• Subjects: Bayes Theorem; Bayesian network; Headache; Humans; Pain; Pain Threshold; structural equation modelling; Tension type headache; Tension type headache; pain; structural equation modelling; Bayesian network
• ISSN: 1526-5900; 1528-8447
• DOI: 10.1016/j.jpain.2022.10.003

•A theoretical model proposes a relationship of psycho-physical and psychological factors in TTH.•A data-driven model associates poor sleep, psychological factors, and chronicity of headache.•This data-driven model could be leveraged in clinical trials investigating treatment of TTH. Tension type headache (TTH) is a prevalent but poorly understood pain disease. Current understanding supports the presence of multiple associations underlying its pathogenesis. Our aim was to compare competing multivariate pathway models that explains the complexity of TTH. Headache features (intensity, frequency, or duration - headache diary), headache-related disability (Headache Disability Inventory-HDI), anxiety/depression (Hospital Anxiety and Depression Scale), sleep quality (Pittsburgh Sleep Quality Index), widespread pressure pain thresholds (PPTs) and trigger points (TrPs) were collected in 208 individuals with TTH. Four latent variables were formed from the observed variables - Distress (anxiety, depression), Disability (HDI subscales), Severity (headache features), and Sensitivity (all PPTs). Structural equation modelling (SEM) and Bayesian network (BN) analyses were used to build and compare a theoretical (modeltheory) and a data-driven (modelBN) latent variable model. The modelBN (root mean square error of approximation [RMSEA] = 0.035) provided a better statistical fit than modeltheory (RMSEA = 0.094). The only path common between modelbn and modeltheory was the influence of years with pain on TrPs. The modelBN revealed that the largest coefficient magnitudes were between the latent variables of Distress and Disability (β=1.524, P = .006). Our theoretical model proposes a relationship whereby psycho-physical and psychological factors result in clinical features of headache and ultimately affect disability. Our data-driven model proposes a more complex relationship where poor sleep, psychological factors, and the number of years with pain takes more relevance at influencing disability. Our data-driven model could be leveraged in clinical trials investigating treatment approaches in TTH. A theoretical model proposes a relationship where psycho-physical and psychological factors result in clinical manifestations of headache and ultimately affect disability. A data-driven model proposes a more complex relationship where poor sleep, psychological factors, and number of years with pain takes more relevance at influencing disability.