Quantifying welfare impairment from respiratory disease in fattening pigs using adaptive conjoint analysis

Abstract Respiratory disease remains highly prevalent in pig production and constitutes a major health and welfare challenge worldwide. However, most studies assessing pig welfare impairment from respiratory disease rely on isolated animal-based indicators, limiting integration across welfare domains. This study aimed to quantify welfare impairment from respiratory disease in fattening pigs (8–10 weeks of age) using indicators aligned with the Five Domains Model. A survey using adaptive conjoint analysis was conducted with animal welfare researchers ( n = 32) and practising swine veterinarians ( n = 22). Respondents completed pairwise comparisons of profiles defined by combinations of five welfare indicators—appetite, coughing, respiratory distress, play behaviour, and exploratory behaviour—each expressed at increasing severity levels. Based on respondents’ choices of which profile represented greater welfare impairment, indicator importance scores and level-specific contributions to pig welfare impairment from respiratory disease were estimated. Across respondent groups, severe respiratory distress was perceived as the largest contributor to pig welfare impairment from respiratory disease, with an estimated contribution of 29.8%, followed by repeated coughing (+ 22.7%) and complete loss of appetite (+ 19.2%). Reductions in play and exploratory behaviour were perceived to contribute comparatively less to welfare impairment from respiratory disease but remained relevant. Agreement between researchers and veterinarians was high, with statistically significant differences limited to occasional coughing and complete loss of play behaviour. In addition, profiles representing single-pathogen respiratory infections were constructed based on expert elicitation, providing an illustrative demonstration of how outcomes from adaptive conjoint analysis may be used to assess welfare impairment under different respiratory disease scenarios. The findings of this study may inform refinement of welfare assessment protocols, support the integration of welfare metrics into epidemiological models, and enable comparison of disease scenarios. However, results should be interpreted with caution, as they reflect perceived rather than directly measured welfare impairment and assume independence between biologically interrelated indicators. Future work should prioritise the application of welfare impairment estimates to biologically realistic welfare profiles in which interrelated indicators co-occur, while also exploring integration with observational approaches and extension to broader, multifactorial welfare contexts.