Prediction Model to Maximize Impact of Syphilis Partner Notification-San Francisco, 2004-2008

Background: Syphilis cases increased 55% in San Francisco from 2007 (n = 354) to 2008 (n = 548). The San Francisco Department of Public Health interviews syphilis patients to identify sex partners needing treatment, but interviewing resources are limited. We developed and validated a model to prioritize interviews likely to result in treated partners. Methods: We included data from interviews conducted from July 2004 through June 2008. We used multivariate analysis to model the number of treated partners per interview in a random half of the data set. We applied the model to the other half, calculating predicted and observed proportions of partners successfully treated and interviews conducted if limiting interviews by syphilis patient characteristics. Results: In 1340 patient interviews, 1665 partners were named; of those, 827 (49.7%) were treated. Ratios of treated partners were significantly higher among patients aged <50 years, compared with >= 50 years (ratio 1.4; 95% confidence interval [CI], 1.0-1.9); patients with primary/secondary syphilis, compared with early latent (ratio 1.4; 95% CI: 1.1-1.8); and patients diagnosed at the municipal sexually transmitted disease clinic, compared with elsewhere (ratio 1.7; 95% CI: 1.4-2.1). Limiting interviews to patients aged <50 years would reduce interviews by 14% and identify 92% of partners needing treatment. Limiting interviews to primary/secondary syphilis patients would reduce interviews by 35% and identify 68% of partners needing treatment. Conclusions: Our model can provide modest efficiencies in allocating resources for syphilis partner notification. Health departments should consider developing tools to maximize impact of syphilis prevention and control activities.