<p>Structural equation model analysis.</p>

Background Human Immunodeficiency Virus (HIV) and tuberculosis (TB) co-infection poses a significant challenge to public health systems due to its complex clinical management and high mortality. This study aimed to reveal the specific mechanisms through which knowledge influences practice in HIV/TB co-infection management among healthcare providers (HCPs) using structural equation modeling (SEM). Methods An exploratory cross-sectional study using convenience sampling was conducted from May to June 2025 involving healthcare providers (HCPs) across various medical institutions in Beijing, which included university-affiliated tertiary hospitals, specialized hospitals, and community health centers. Results A total of 565 valid questionnaires were collected, with 364 (64.42%) completed by medical doctors. The knowledge, attitude, and practice scores were 18.51 ± 7.75 (possible range: 0–30), 43.64 ± 5.51 (possible range: 10–50), and 30.75 ± 7.44 (possible range: 8–40), respectively. Spearman correlation analysis revealed significant positive correlations between knowledge and attitude (r = 0.500, P < 0.001), between attitude and practice (r = 0.584, P < 0.001) and between knowledge and practice (r = 0.592, P < 0.001). SEM analysis indicated that knowledge had a direct influence on both attitudes (β = 0.458, P = 0.002) and practices (β = 0.491, P = 0.011), while attitudes influenced practices (β = 0.272, P = 0.008). Furthermore, knowledge indirectly affected practices through attitudes (β = 0.124, P = 0.005). Conclusion In our sample, respondents demonstrated limited knowledge, generally positive attitudes, and moderately adequate practices regarding HIV and TB co-infection. These preliminary findings suggest that targeted educational interventions designed to enhance knowledge may effectively improve both attitudes and practical behaviors, though more rigorous research is needed to confirm these relationships.