Article ID: | iaor20043663 |
Country: | Netherlands |
Volume: | 20 |
Issue: | 1 |
Start Page Number: | 131 |
End Page Number: | 135 |
Publication Date: | Jan 2004 |
Journal: | International Journal of Forecasting |
Authors: | Oliveira M.M., Mexia J.T. |
Keywords: | time series & forecasting methods |
Both epidemic and endemic scenarios have been proposed to study the evolution of acquired immune deficiency syndrome (AIDS) in Portugal. This article presents a multistage model to study AIDS in which the stage of the infected patients precedes the stage of death by AIDS. The STATIS methodology (Structuration des Tableaux à Trois Indices de la Statisque) is ued to condense the information for these two stages. The results for the stage of infection enabled the choice of time dependent control variables for prediction of the number of deaths and the assessment of the effectiveness of anti-retrovirus treatments. These predictions point towards endemic scenarios and to anti-retrovirus treatments effectiveness lower than in other European countries. Moreover, when both series are jointly considered, the results obtained point to a lag of 2 to 3 years between infection by AIDS and death by connected causes. The multistage approach is more centered on actual data than other approaches (e.g. back-projection).