Sex disparities in the emotional impact of chronic non-cancer pain using hierarchical clustering and machine learning tematics

Abstract

Introduction: Chronic Non-Cancer Pain (CNCP) disproportionately affects women and men, not only in prevalence but also in emotional, social, and functional burden. Biological sex differences interact with gender roles and societal expectations. Women report higher emotional distress, sleep disturbances, and caregiving-related limitations, while men face higher work-related disability and underreport emotional suffering due to cultural norms of stoicism. These disparities result in unequal clinical presentations and treatment outcomes, highlighting the need for gender-sensitive categorization and management of CNCP. Objectives: The primary objective was to identify sex-specific emotional impact profiles of CNCP patients through hierarchical clustering and machine learning. Materials and Methods: This mixed-methods study included 216 CNCP patients (69% women) from a Spanish tertiary hospital. Data were collected via structured interviews conducted by four pain experts using clinical scales and the internally validated Gender-Pain Questionnaire. A team of three psychosocial researchers organized and thematically categorized the emotional impact data. Quantitative clustering analyses were conducted using hierarchical clustering (Ward.D2 with Euclidean distance and Gower's method for mixed data) and supported by machine learning thematic classification. Results: 1/ Women were older, more likely to be homemakers or on work disability, and showed trends of higher anxiety. Men were more often prescribed morphine and antidepressants. 2/ Women reported higher reproductive role disruption, while men showed predominance in productive role impact. Mixed roles were more burdensome for women. 3/ For women, three clusters captured physical-emotional overload, psychosocial disconnection, and role-based distress. For men, three clusters highlighted emotional suppression, work-related loss, and social disintegration. 4/ Using weighted mixed data and thematic ML categorization, distinct emotional impact profiles by sex were found, with men clustering around emotional repression and productivity loss, and women around relational suffering and psychosocial vulnerability. Conclusions: This thesis reveals robust sex-based differences in the emotional and functional impacts of CNCP. Women suffer a broader spectrum of emotional strain linked to caregiving and social expectations, while men exhibit underrecognized emotional distress tied to productivity and social withdrawal. Hierarchical clustering combined with ML proved effective in defining distinct emotional profiles, offering valuable insights for implementing gender-sensitive clinical and public health strategies in CNCP care.