Risk factors for Q fever incidence in South Korea: a comparative analysis using frequentist and Bayesian methods.

Objectives: This study investigated the principal determinants of human Q fever incidence and explored regional variation between metropolitan cities and provinces in South Korea.

Methods: Panel data on human Q fever incidence, livestock populations, and facility metrics were collected across 17 metropolitan cities and provinces from 2017 to 2024. Analytical approaches included frequentist models (ordinary least squares [OLS], random effects, fixed effects) and Bayesian models.

Results: Frequentist panel analysis indicated that slaughterhouse count was positively associated with Q fever incidence in both pooled OLS (β=1.20, p<0.001) and random effects models (β=1.03, p<0.001), but not in the fixed effects model (β=0.14, p=0.65). After correcting for serial correlation using Driscoll-Kraay standard errors, livestock population (β=0.55, p<0.01), livestock market count (β=-2.01, p<0.05), and livestock Q fever cases (β=-0.11, p<0.01) were significantly associated with human incidence. A Bayesian fixed effects model confirmed a significant relationship between slaughterhouses and human Q fever incidence (posterior mean: 0.87, 95% credible interval [CrI], 0.21-1.42), providing more stable inference with limited samples and allowing probabilistic uncertainty estimation. A Bayesian hierarchical model revealed a stronger association in metropolitan cities (posterior mean, 1.46; 95% CrI, 0.34-2.57) than in provinces (1.22), while livestock population remained significant in provinces (0.94, 95% CrI, 0.15-1.74).

Conclusion: In South Korea, slaughterhouse density was the main determinant of Q fever in metropolitan areas and livestock density was the primary risk factor in provinces. These findings underscore the need for region-specific preventive strategies and reinforce the value of a One Health approach.