On the Path to an Aberration Detection System for STD Surveillance

Wednesday, March 12, 2008: 11:30 AM
Northwest 4
Delicia Carey, PhD , Division of STD Prevention, Centers for Disease Control, Atlanta, GA
Ranell Myles, MPH, CHES , Division of STD Prevention, Centers for Disease Control, Atlanta, GA
Samuel Groseclose, DVM, MPH , Division of STD Prevention, Centers for Disease Control, Atlanta, GA

Background:
Routinely applying statistical aberration (signal) detection methods (ADMs) to syphilis surveillance data should translate into more timely response to emerging outbreaks. Prior to encouraging more widespread use of the methods by state and local STD programs, DSTDP investigated selected ADMs at the federal level to determine how they can be used to improve detection of and response to syphilis outbreaks.

Objective:
To apply selected ADMs (e.g., CDC-Early Aberration Reporting System [EARS]) to syphilis surveillance data at the federal level and describe their performance characteristics.

Method:
Because of different dynamics of male and female syphilis in the U.S. (i.e., substantial MSM transmission) and the role of primary and secondary syphilis (P&S) in transmission, three cumulative sums methods (differ by baseline, signal threshold and sensitivity) in EARS were applied to sex- and county-specific P&S data for the period 2001-2005. Initial analyses used data from two randomly selected states with counties that reported >10 cases per year in 2005. Parameters (e.g. date interval, stratification, and threshold) were defined based on data distribution. The analysis was performed using SAS.

Result:
Analysis of statewide P&S data revealed fewer signals compared to county-specific analysis. To address masking of local incidence trends, future analyses will be conducted on county-specific incidence. U.S. counties will be dichotomized by syphilis incidence: 1) low (for whom ≥ 1 case should stimulate investigation and response [manual threshold]) and 2) moderate-high (for whom ADMs might provide objective identification of increasing incidence).

Conclusion:
Counties and states with low syphilis incidence can use a manual threshold. Furthermore, county-level signals can be masked when data are accumulated and analyzed at the state level. However, additional evaluation is needed to better characterize false positive signals to improve effectiveness of investigation and response.

Implications:
ADMs can serve as STD program tools to focus the review and application of surveillance data for syphilis prevention.
See more of: Issues in STD Surveillance
See more of: Oral and Poster
<< Previous Abstract | Next Abstract