CROI 2015 Program and Abstracts

Abstract Listing

Poster Abstracts

(P=0.34). Fifty (35.0%) HIV-infected and 9 (16.1%) HIV-uninfected women died during follow-up. Of these 59 deaths, 47 (79.7%) were attributed to cancer and 1 (1.7%) to toxicity of treatment; cause of death was unknown for 11 (18.6%). Median survival for HIV-infected women was shorter than HIV-uninfected women, median 16.6 versus 24.3 months, respectively (P=0.007), see Figure. Findings were similar after adjustment for stage, with HIV-infection associated with increased mortality (HR 2.66, 95% CI 1.3 - 5.5, P=0.008). Advanced cancer stage was strongly predictive of mortality (P=0.002). Among women with HIV, CD4 cell count or ART duration was not associated with survival.

Conclusions: Despite ART, HIV-infection is associated with increased mortality among women with cervical cancer. Overall survival is poor among women with HIV-associated cervical cancer in Africa and further research is needed to understand factors contributing to excess mortality. 712 Potential Cost-Effectiveness of Cervical Cancer Screening of HIV-Positive KenyanWomen Marita Mann ; Joseph Babigumira; Louis Garrison; Michael Chung University of Washington, Seattle, WA, US Background: As antiretroviral therapy is scaled up in Africa, HIV-positive women are living longer and increasingly likely to die from cervical cancer. Cervical cancer screening methods used in these settings include Papanicolaou smear (Pap), visual inspection with acetic acid (VIA), and human papillomavirus testing (HPV). Our objective was to assess the cost-effectiveness of these methods among HIV-positive women. Methods: The cost-effectiveness analysis was based on a trial of 500 HIV-positive women who underwent VIA, Pap smear, HPV, and gold-standard colonoscopy-directed biopsy at the Coptic Hope Center in Nairobi, Kenya. We used a decision tree to assess overall cost-effectiveness comparing Pap, VIA, and HPV. A Markov model consisting of 7 disease states projected life expectancy and costs following each strategy. The base case analysis included those with CD4 count 200-500 cells/mm 3 . We additionally modeled patients with low and high CD4 counts. We addressed the impact of parameter uncertainty using univariate and probabilistic multivariate sensitivity analysis. Costs included direct and indirect medical and non-medical costs from a semi-societal perspective. Results: VIA had lowest cost and highest life expectancy (due to reduced loss-to-follow-up) ($331, 17.2 LYs), followed by HPV ($563, 17.1 LYs), Pap ($622, 17.1 LYs) (Figure 1). CD4 level did not affect this rank order, though VIA at low CD4 showed the lowest cost ($111, 15.3 LY), while VIA at high CD4 produced most health gains ($285, 19.9 LY) [ICER: $37/LY]. Costs were sensitive to prevalence of cancer, sensitivity, age, and cost of cancer. Life expectancy was sensitive to age at screening. Results were robust to probabilistic sensitivity analysis.

Poster Abstracts

438

CROI 2015