O037 Predicting STI risk among people attending sexual health services: Development of a triage tool targeting behavioural interventions among young people. (30th June 2016)

• Record Type: Journal Article
• Authors: King, Carina; Mercer, Cath; Furegato, Martina; Mohammed, Hamish; Copas, Andrew; Shahmanesh, Maryam; Gilson, Richard; Hughes, Gwenda
• Is Part Of: Sexually transmitted infections. Volume 92(2016)Supplement 1
• Page Start: A14
• Page End: A14
• Publication Date: 2016-06-30
• Subjects: Sexually transmitted diseases -- Periodicals; HIV infections -- Periodicals; 616.951005
• DOI: 10.1136/sextrans-2016-052718.36
• Languages: English
• ISSNs: 1368-4973

Abstract

Abstract : Background/introduction: There are very limited resources for delivering sexual health promotion within sexual health services (SHS). Aim(s)/objectives: Santé, a feasibility study for a trial of sexual risk reduction interventions, is developing a triage tool embedded within the electronic patient record to target interventions by risk score among young people (16–25 years) attending SHS. Methods: We used GUMCADv2, the national mandatory STI surveillance dataset (2013–2014 – Model 1), and the GUMCADv3 pilot (July-October 2015 – Model 2). Predictive logistic regressions for acute STI diagnosis were run. Model 1 only considered demographic and clinical variables; Model 2 also included enhanced behavioural data (number of partners, new partners, and condom use in the past 3 months). Results: 936, 251 and 619 patient-episodes were included in Models 1 and 2 respectively, of which 11% and 4% involved an STI diagnosis. In Model 1, predicted risk of STI diagnosis ranged between 1–47% (pseudo-R 2 : 1.9%). Referring the riskiest (highest decile) patients to more intensive interventions gives a sensitivity and specificity of 70% and 45%, respectively, and a positive predictive value (PPV) of 13% for STI diagnosis. In Model 2 the predicted risk of STI was 0–53% (pseudo-R 2 : 23%), and referring the riskiest patients demonstrated an improved sensitivity (76%), specificity (87%) and PPV (25%). Discussion/conclusion: Routinely collected surveillance data can be used to triage young people for targeted risk-reduction interventions, but this is more robust if behavioural data are taken into account. Addition of behavioural data to routine STI surveillance (GUMCADv3) is a powerful way to target sexual health promotion.