5F3 Evaluating the Impact of Preventive Behaviors and Individual Information Dissemination to Contain the Ongoing Syphilis Epidemic through Network-Based Simulation Models

Thursday, June 12, 2014: 8:20 AM
Caterina Scoglio, PhD1, Faryad Sahneh, PhD2, Fahmida Chowdhury, PhD3, Gary Brase, PhD2, Thomas Gift, PhD4, Robert P. Kohn, MPH5 and Kyle T. Bernstein, PhD, ScM6, 1Department of Electrical and Computer Engineering, Kansas State University, Manhattan, KS, 2Kansas State University, 3National Science Foundation, 4Division of STD Prevention, Centers for Disease Control and Prevention, Atlanta, GA, 5Applied Research, Community Health, Epidemiology, and Surveillance (ARCHES) Branch, Population Health Division, San Francisco Department of Public Health, San Francisco, CA, 6STD Prevention and Control Section, San Francisco Department of Public Health, San Francisco, CA

Background: San Francisco has experienced an ongoing syphilis epidemic among men who have sex with men (MSM) since the late 1990s. We developed a modified SIS (Susceptible-Infected-Susceptible) model to explore the potential impact of individual-based information dissemination of preventive behaviors.

Methods: Sexual network data from the San Francisco Department of Health partner services activities for early syphilis and human immunodeficiency virus (HIV) cases in 2013 were used to evaluate an expanded SIS model that allows susceptible individuals to become “Alert” (SAIS); aware of the health status of neighbors in their networks, individuals adopt cautious behaviors to reduce their infection rate.  This MSM sexual contact network has 435 nodes (individuals) and 454 links (sexual contacts). A self-infection rate models infections arriving from outside the model population. Additionally, the SAIS model incorporate the effect of information dissemination via an additional online network. Transition rates of the model are: β, infection rate; δ, recovery rate; βa , reduced-by-alertness infection rate; κ , alertness rate by contact network; μ , alertness rate by online information network; ε, self-infection rate.

Results: Simulations show that incorporating alertness and an information network reduces infection prevalence, versus an SIS model.  With no alertness (κ=μ=0), syphilis prevalence stabilizes at around 25%. With alertness via the contact network (κ=β, μ=0), syphilis prevalence decreases to 17%. Adding an online information dissemination network (κ=β, μ=β), reduces syphilis prevalence to 8%. Finally, we have performed a sensitivity analysis showing that when varying the parameter β in the range ±20%, the expected value of the standard deviation of the infected population fraction remains comparable, despite the parameter variations, over 1000 simulations for each value of β.

Conclusions: Network-based modeling of the syphilis epidemic suggests that “alertness” reduces prevalence. Utilization of online information dissemination can further reduce prevalence.  The practical usability of these results may be impacted by the reliability/accuracy of actual behavioral modeling.