HIV / STI
Development and Validation of Two Syphilis Treatment Completion Algorithms Using Routine Surveillance Data in British Columbia, Canada Justin Sorge* Justin Sorge Sorge Sorge Sorge Sorge Sorge Sorge Sorge Sorge Sorge Sorge Sorge BC Centre for Disease Control/Public Health Agency of Canada
Background British Columbia (BC), Canada, declared a syphilis outbreak in 2019 following sustained increases in infectious and congenital cases. Timely and complete treatment is critical to prevent onward transmission and severe outcomes, particularly during pregnancy. Monitoring treatment completion is therefore a key surveillance activity; however, treatment information is often inconsistently captured in routine public health data. To our knowledge, no studies have described the development of an automated algorithm to identify syphilis treatment completion, and no such algorithm is currently used in Canada.
Objective To develop and validate two independently constructed rule‑based algorithms to assign syphilis treatment completion status using routinely collected surveillance data in BC.
Intervention Using the DEVELOP RCD framework, interdisciplinary teams of epidemiologists and clinicians independently designed automated surveillance algorithms based on laboratory, treatment, and case management data. Validation analyses compared algorithm‑assigned treatment completion status with clinicians’ blinded categorizations, which served as the gold standard.
Outcomes The first algorithm had an overall sensitivity of 53% (95% CI: 45–61) and specificity of 94% (95% CI: 88–97). The second algorithm showed higher sensitivity (71%, 95% CI: 65–77) but lower specificity (78%, 95% CI: 67–86), reflecting different trade‑offs in misclassification.
Conclusion Interdisciplinary collaboration was essential to developing automated algorithms for identifying syphilis treatment completion. One algorithm prioritizes specificity and aligns with surveillance definitions, while the other prioritizes sensitivity and aligns with clinical management. These findings provide an evidence‑based foundation for scalable, automated monitoring of syphilis treatment completion in public health surveillance.