Return on investment in needle and syringe programs in Australia: report

3.2 Methodology for estimates of numbers of people living with HIV acquired through injecting drugs

Page last updated: 2002

3.2.1 Estimates and projections of HIV/AIDS incidence
3.2.2 Estimates of the number of injecting drug users living with HIV infection
3.2.3 Estimating the number of injecting drug users living with HIV without NSPs
3.2.4 Estimated effect of NSPs in reducing numbers of injecting drug users with HIV by disease stage

3.2.1 Estimates and projections of HIV/AIDS incidence

Estimates of past HIV incidence and future AIDS incidence as a result of injecting drug use were obtained using back-projection methods. The method uses observed AIDS incidence data (adjusted for reporting delay), and knowledge of the rate at which HIV infected people progress to AIDS, to reconstruct the likely pattern of past HIV incidence. It is then also possible to estimate future AIDS incidence. The form of back-projection used was that suggested by Becker et al (1991), as modified by Marschner and Watson (1992). Because of the relatively small numbers of AIDS cases reported due to injecting drug use, back-projection analyses were applied to annual AIDS counts.

The baseline rate of progression to AIDS was modelled using a Weibull-with-levelling distribution (Rosenberg et al. 1992), corresponding to a median time to AIDS of just under 10 years and a progression rate of 11.2% at four years (Alcabes et al. 1993). The extended definition of AIDS, adopted in Australia in January 1988, was assumed to result in a 10% increase in the rate of progression to AIDS (Rosenberg et al. 1992).

Because of the uncertainties surrounding both the effect of combination antiretroviral treatments in reducing the rate of progression to AIDS, and the numbers of people living with HIV infection taking up such treatments, back-projections were performed using the following methods. First, a back-projection based on AIDS cases diagnosed to the end of 1994 was performed to estimate the pattern of HIV incidence up to this time. Over this period only moderately effective antiretroviral treatments were available, assumed to correspond to an overall 10% reduction in the rate of progression to AIDS, so the pattern of past HIV incidence can be reliably reconstructed. Second, the effects of improved combination treatments since the beginning of 1995 were then estimated, based on the estimated pattern of HIV incidence, so as to closely approximate AIDS incidence observed between 1996 and 2000.
Top of page
The effects of improved combination treatments on reducing the overall rate of progression to AIDS were estimated based on cases of AIDS reported due to injecting drug use, and are summarised in the Table 3.2.1 below.

Projections of AIDS incidence from 2001 onwards were made by assuming that the effect of treatments on the rate of progression to AIDS continued at the year 2000 level.

In analyses HIV incidence was fixed at 20 cases per year from 1994 onwards. The level at which HIV incidence was fixed was decided on the basis of the number of HIV diagnoses and diagnoses of newly acquired HIV infection reported to the National HIV Database, and was also chosen to be consistent with the estimated HIV incidence obtained from the back-projection analyses.

Back-projection estimates of HIV incidence need to be adjusted for underreporting of AIDS diagnoses, and deaths prior to AIDS. Reporting of AIDS cases was thought to be relatively complete in Australia, with completeness estimated to be at least 95% (Grulich et al. 1999). Deaths among IDUs are estimated to be approximately 1% per annum (Thorley 1981; English et al. 1995). The median time to AIDS is thought to be just under 10 years, so, taken together, HIV incidence was inflated by 15% to allow for underreporting of AIDS and deaths prior to AIDS.

Table 3.2.1 Estimated percentage effect of combination antiretroviral treatments in reducing the overall rate of progression to AIDS between 1995 and 1999

Year19961997199819992000
Estimated reduction in progression rate (%)
77
69
60
52
44

3.2.2 Estimates of the number of injecting drug users living with HIV infection

Estimates of the number of IDUs living with HIV infection by disease stage (CD4+ cell count more than 500/Ml, a CD4+ cell count of less than 500/Ml and AIDS free, or living with AIDS) were based on the estimated pattern of past HIV incidence. The rate of progression to a CD4+ cell count fewer than 500/Ml was modelled using a similar Weibull-with-levelling distribution to that used to model the time from HIV infection to AIDS. The median time from HIV infection to a CD4+ cell count of 500/Ml was assumed to be 4 years, with 95% below 500/Ml by 10 years. Survival following AIDS among IDUs in Australia was reasonably consistent between 1988 and 1995. The effect of combination antiretroviral treatment in improving survival following AIDS from 1996 was assumed to be similar to the effect of treatment in reducing the rate of progression to AIDS in Table 3.2.1, and to continue at the year 2000 rate from 2001 onwards. Background death rates were based on ABS life tables, assuming that the mean age at HIV seroconversion among IDUs was 30 years, and that there were 3 male HIV-infected IDUs for each female HIV-infected IDU (ABS 1995).
Top of page

3.2.3 Estimating the number of injecting drug users living with HIV without NSPs

The effect of needle and syringe programs (NSPs) in reducing HIV transmission among IDUs has been estimated to correspond to an annual reduction in (logit) HIV prevalence of 0.28 (see Section 3.1).

HIV prevalence among IDUs in Australia between 1980 and 2000 was based on the estimated numbers of IDUs living with HIV described above, and estimates of the numbers of IDUs in Australia.

Numbers of IDUs in Australia were estimated as follows. The number of dependent heroin users in Australia in 1997 was assumed to be 75,000 (Hall et al. 2000). A reasonable fit to available estimates over the previous two decades was obtained by assuming a constant net 8% increase in dependent heroin users per year. To allow for injecting of other drugs, the total number of regular IDUs was assumed to be 33% greater than the number of dependent heroin users (i.e. 100,000 regular IDUs in 1997 (Law 1999). The number of occasional IDUs was assumed to be 175,000 in 1997 (Law 1999) with the same annual percentage increases.

NSPs were first introduced in Australia in late 1987. Hence, NSPs were assumed to have reduced HIV prevalence among IDUs from 1988 onwards. The pattern of HIV prevalence if NSPs had not been introduced was estimated by increasing (logit) HIV prevalence by 0.28 per year from 1988 onwards. From this, a pattern of HIV incidence if NSPs had not been introduced was derived.

Estimates of the numbers of IDUs living with HIV by disease stage if NSPs had not been introduced were obtained by applying the same models described above regarding rates of progression from HIV infection to CD4+ cell count <500 cells/Ml, to AIDS and survival before and following AIDS.

3.2.4 Estimated effect of NSPs in reducing numbers of injecting drug users with HIV by disease stage

To allow costing of the effect of NSPs in reducing the number of people living with HIV, estimates of the reduction in the number of people living with HIV by disease stage were obtained by subtracting the estimates obtained with NSPs from the corresponding estimates without NSPs. In these analyses, HIV incidence due to injecting drug use was assumed to cease from 2001 onwards, and estimates were projected forward until all people infected with HIV were estimated to have died.
Top of page