Secular trends in the survival of HIV-infected homosexual men in Amsterdam and Vancouver estimated from a death-included CD4-staged Markov model

Background The purpose of this study was to investigate secular trends in waiting times in CD4-based stages of human immunodeficiency virus (HIV) disease progression in two cohorts of homosexual men, one in Vancouver and one in Amsterdam. All HIV-positive men with two or more CD4 counts in their AIDS-free period between 1 January 1985 and 1 January 1997 were included in this study. Data regarding clinical AIDS diagnoses (using the 1987 Centers for Disease Control and Prevention [CDC] AIDS case definition) and death were collected through active follow-up, review of hospital records, and municipal/national registries. The Vancouver Lymphadenopathy-AIDS Study (VLAS), was started in November 1982 and had enrolment until December 1984. Both HIV-negative and HIV-positive men were followed at intervals of 3-6 months until 1986 and annually thereafter. The Amsterdam cohort study on HIV and AIDS (ACS) started in December 1984, has ongoing enrolment and follow-up of both HIV-negative and HIV-positive homosexual men. The HIV-positive men were followed at intervals of 3 months. Methods The CD4-based stage of an individual at each visit was determined using smoothed data. For each cohort and in each calendar time period, a CD4-based Markov model with death as the absorbing stage was fitted to the data. The parameters in this model were estimated using the method of maximum likelihood and confidence intervals were calculated using bootstrap methods. Results A total of 509 homosexual men participating in the VLAS were included in this study, providing 5356 visits. Some 292 men developed AIDS before 1 January 1997 and 219 died before this date. In all, 232 of the 239 deaths were AIDS related. Thirty-seven per cent of all visits were related to treatment. A total of 543 homosexual men participating in the ACS were included in this study, providing 10 043 visits; 277 men developed AIDS before 1 January 1997 and 250 died before this date. The date of AIDS diagnosis was known for 225 of the 250 deaths. Twenty per cent of all visits were related to treatment. We found that in both cohort studies the stage-specific waiting times were longer in the low CD4-based stages (stages 4, 5 and 6: i.e. CD4 count <500 cells per mm(3)) after March 1990 compared to waiting times before March 1990. The increase in mean waiting time in these stages with low CD4 count was 21%, 33% and 53%, respectively in the ACS and 20%, 2% and 29% in the VLAS. Because waiting times alone are not exclusive for progression in a reversible model we also calculated the stage-specific median incubation periods till death. Men spent considerably longer in these CD4-based stages after March 1990 compared to before March 1990. Conclusions Data from these population-based cohort studies showed that HIV disease progression in the calendar period where treatment was administered was slower for individuals in stages with low CD4 counts. We found no evidence for shortening of the incubation period that may have appeared from increasing virulence of the HIV in the population.