Modeling the cost-effectiveness of rapid STI tests for

Modeling the cost-effectiveness of rapid STI tests for the control of HIV and other STIs amongst female sex workers P Vickerman*, C Watts*, R W Peeling***, D Mabey*, M Alary**, *London School of Hygiene and Tropical Medicine **University of Laval ***WHO/TDR, Geneva HIVTools Research Group

Background In sub-Saharan Africa, gonococcal (Ng) and chlamydial (Ct) infections are usually managed using the syndromic approach However in women: – – Only a small proportion of infections are symptomatic and even the syndromic approach usually has poor sensitivity and specificity This results in many women not being treated and excessive over-treatment A possible alternative are rapid tests for Ng and Ct which are: – More specific (>90%) – Can be used on asymptomatic women – but currently have variable sensitivity (25 -85%)

Study aims Estimate the potential impact of the existing SIDA 2 HIV prevention intervention in Cotonou, Benin for 1995 to 1998 Determine the impact of the syndromic STI treatment component of the intervention for Ng/Ct Estimate the incremental cost-effectiveness of replacing the currently used syndromic algorithm with a combined rapid test for Ng/Ct with: – different assumptions about the cost and sensitivity of the rapid test

Steps in analysis Compile epidemiological, behavioural and intervention specific data from Cotonou, Benin and scientific literature Using data from Cotonou obtain different fits of a dynamic deterministic mathematical model to site specific HIV and STI prevalence data for 1995 -1998 Estimate the potential impact of the existing SIDA 2 HIV prevention intervention in Cotonou, Benin for 1995 -1998: – What is impact of STI treatment component of intervention? Estimate the incremental cost (US$ 2004) of replacing the existing syndromic algorithm with a rapid test, including: – – – Test costs Extra staff costs for training and undertaking tests Decrease in treatment costs due to less over-treatment Estimate the incremental impact and cost-effectiveness for 1995 -1998 of replacing the existing syndromic algorithm with rapid tests of: – High specificity (95%) but variable sensitivity (40 -80%)

Data Summary : Cotonou, Benin - West Africa (1998) • Sexually active population of 310, 000 • Nearly all men circumcised (~99%) • Average ~1. 6 sexual partners person per year • Commercial sex: § ~ 1% women of reproductive age sell sex § average 52 clients per month § 13 -29% sexually active men currently buy sex • Condom use: • Very low in regular partnerships; • Low in casual partnerships; • High in commercial sex (81% of commercial sex acts)

Ongoing intervention activity in Cotonou, Benin SIDA STD clinic est. in Cotonou in 1993 • Dedicated clinic for SWs, providing free clinical syndromic STD screening & treatment • reaches 250 SW (13%) every month • Modified syndromic approach based on clinical findings and microscope examination of vaginal discharge • Sensitivity of algorithm for Ng/Ct =48%, specificity=75% • Outreach to SW focussing on prevention activities and condom promotion: • From 1995 to 1998, condom use increased from 60% to 81% of acts in last week • SWs encouraged to come to STI clinic irrespective of symptoms

Prevalence of HIV and STI in Cotonou for 1995 -1998 Sex workers (SW) HIV 1995 49%‡ Clients (CL) HIV 1998 41%‡ 8. 4%** Female population HIV 1. 5%* 3%† Sex workers Ng/Ct 33%‡ 25%‡ Clients Ng/Ct 8%** Female population Ng/Ct <3%† † data from population survey of women, *data from ANC women in 1996, ‡ data from SWs attending the SIDA 2 STI clinic, ** CLs recruited at prostitution venues

Part 1: Impact of existing SIDA 2 HIV prevention intervention for 1995 to 1998

Predicted impact of SIDA 2 intervention over 4 years from 1995 -1998 Mean uncertainty FSWs general population 18% 22% 14 -21% 18 -25% Number of HIV infections averted 1523 1176 -1820 STI treatment aspect of intervention: Ng/Ct cases treated Ng/Ct cases averted in whole pop 1339 1181 -1709 18553 16871 -20447 % drop in HIV incidence in: % HIV infections averted due to STI treatment 23% 19 -26%

Part 2: Impact of replacing existing syndromic approach with rapid tests Four scenarios of rapid test for diagnosing Ng/Ct: – Sensitivity=50%, 60%, 70% and 80% – Specificity=95% Only concerned with impact of STI treatment component of intervention

SWs treated for Ng/Ct

Percentage of HIV and Ng/Ct infections prevented by STI treatment intervention

HIV infections averted

Part 3: Incremental cost-effectiveness of replacing existing syndromic approach with rapid tests -Modelled for range of test costs = $1 -4 -Evaluation of GC-Check rapid test in Cotonou in 2004 used to estimate additional personnel costs: – Extra 15 -20 minutes needed per test undertaken – Training of 3 hours needed for 2 staff (50 tests used in training)

Incremental cost-effectiveness per HIV infection averted (in US$ 2004)

Key messages Compared to syndromic management (48% sensitivity), a moderately sensitive (>60%) rapid test can substantially increase the impact of STI treatment interventions for SWs: – Most additional impact is amongst their clients Compared to syndromic management, rapid tests can dramatically decrease the number of women inappropriately treated for Ng/Ct – >75% if a 95% specific rapid test is used instead of current algorithm Even at $4 per test, moderately sensitive (>60%) rapid tests may be a highly cost-effective strategy for increasing the impact of STI treatment interventions. Even with highly sensitive rapid tests, SW targeted STI treatment interventions will only result in large decreases in HIV incidence if: – – Higher coverage levels amongst SWs are achieved and/or STI treatment is also provided to clients

Acknowledgements Funding from: Sexually Transmitted Diseases Diagnostics Initiative, WHO/TDR DFID funded AIDS Knowledge programme Article is in press in Sexually Transmitted Infections