Timely diagnosis of HIV infection reduces mortality in patients on HAART in Argentina. A Markov model approach

DARIO DILERNIA; CARINA CESAR; ALEJANDRO KROLEWIECKI; OMAR SUED; RUBEN MARONE; NATALIA LAUFER; MANUEL GÓMEZ CARRILLO; PEDRO CAHN; HORACIO SALOMÓN

México DF, México

Conferencia; AIDS 2008 - XVII International AIDS Conference; 2008

International AIDS Society

Background: In Argentina, diagnoses occur through voluntary testing or symptom-based case finding. Our objective was to predict the impact of expanding HIV testing in Argentina, a resource-limited country.Methods: We developed a time-dependent Markov model of infection, diagnosis, treatment and survival with parameters-dependent transition-probabilities developed from cohort studies. Local data was obtained from the CCASAnet/IEDEA cohort (N=2920), Ministry of Health, NexoNGO and the National Reference Center for AIDS. Analysis was performed by Monte-Carlo microsimulation trials. Outputs include time on clinical stages and therapy regimens, CD4-count, viral-load, infection-state and age at every life-time of each individual; mortality rates and proportion of unidentified infection at a population-level. Simulations were performed for current testing strategy and for a theoretical scenario with earlier diagnosis. Results: Diagnosis before onset of symptoms increase life expectancy by 10.7 years (52.9 vs. 63.6 for those diagnosed with AIDS, non-infected population= 76.9). Given a delay of 9 years from infection to diagnosis, proportion of unrecognized infection was 43.2% and the mortality rate in the first year of HAART 7.6%, characteristics of a resource limited setting (Lancet 2006; 367:817-24); simulations performed under a delay of 5 years showed rates of 23.8% and 2.1% respectively. Proportion of individuals with unaware infection needing treatment was 12% for the former condition, and 0.2% for the last. The model accurately predicts times to AIDS and death for untreated individuals, and survival curves for those initiating HAART compared to published data (Lancet 2002; 355:1131-1137, Ann Intern Med. 2007; 146:87-95). Conclusions: Our model correctly predicts current Argentina HIV-epidemic. Our results highlight the importance of implementing health policies aimed at detecting infections at early stages. In Argentina, detection within the first 5 years of infection would mean a reduction of AIDS-related mortality to less than 30% and initiation of therapy for 6000-8000 individuals unaware of their infection status.