TP 11 Derivation and Validation of a Risk-Scoring Algorithm for Screening Asymptomatic Chlamydia and Gonorrhea

Tuesday, June 10, 2014
Pre-function Lobby (M2)
Titilola Falasinnu, MHS, School of Population and Public Health, University of British Columbia, Vancouver, BC, Canada, Mark Gilbert, MD, MHSc, FRCPC, Clinical Prevention Services, British Columbia Centre for Disease Control, Vancouver, BC, Canada, Jean Shoveller, PhD, School of Population and Public Health, University of British Columbia, BC, Canada and Paul Gustafson, PhD, Department of Statistics, University of British Columbia, BC, Canada

Background:  This study aims to derive and validate a risk-scoring algorithm to accurately identify asymptomatic patients at increased risk for chlamydia and gonorrhea (CT/GC) infection. 

Methods:  We examined risk assessment data (2000-2012) from clinic visits at two Vancouver, British Columbia sexual health clinics. We conducted multivariate logistic regression of 7 years of clinic visits data (2000-2006). We identified significant predictors of CT/GC infection from the final regression model and then weighted and summed their regression coefficients to create a risk score. The model’s performance was evaluated using the area under the receiver operating characteristic curve (AUC) and the Hosmer-Lemeshow (H-L) statistic. We examined the sensitivity and proportion of patients that would need to be screened at different cutoffs of the risk score. Temporal validation was assessed in clinic visits from 2007-2012. 

Results:  The prevalence of infection was 1.8% (n=13,791) and 2.1% (n=18,050) in the derivation and validation datasets, respectively. The final logistic regression model included male gender, younger age, non-white ethnicity, multiple sexual partners and previous CT/GC diagnosis. The model showed reasonable performance in the derivation (AUC, 0.74; H-L p=0.99) and validation (AUC, 0.64; H-L p=0.78) datasets. Possible risk scores ranged from -3 and 25. We identified a risk score cutoff point of ≥4 that detected cases with a sensitivity of 89% and 80% by screening 64% and 63% of the derivation and validation populations, respectively.

Conclusions:  This is the first study in sexual health contexts that derived and temporally validated a well-performing risk-scoring algorithm for screening asymptomatic CT/GC infection, an issue particularly salient in this field because of the shift to more sensitive diagnostic tests over this time period. These findings support the use of the algorithm for tailoring risk assessment to the specific circumstances of the patient and have important implications for reducing unnecessary screening and saving costs.