Appendix S1 Model formulation.

Figure S1 Death rates per month by age, separately for recent injectors and non-recent injectors. Recent injecting was defined as having reported injecting episodes for a period between interviews [in response to the question ‘have you injected since your last Amsterdam Cohort Study (ACS) visit?']. Lines give the best fit for these hazards by spline Poisson regression.

Figure S2 (a) Average duration of injecting. (b) Average duration of injecting at first Amsterdam Cohort Study (ACS) visit. (c) Combination antiretroviral therapy (cART) uptake. A stop-injecting rate of 0.016 combined with a relapse rate of 0.04 provided a good fit to the variables of injecting duration, as well as to the fraction of injecting drug users (IDU) currently injecting within the ACS (main text Fig. 1). With a lower stop-rate and no relapse-rate, a similar fraction of IDU currently injecting over time could be achieved. However, in this case the distribution of current injectors became more skewed to shorter times since first injecting and, combined with the ACS inclusion criterion of recent injecting, this led to a lower modelled average injecting duration, especially at the first ACS visit (results not shown). From about 2000 onwards, extra effort was put into recruiting younger drug users for the ACS. This bias is not included in the model, as only few IDU were recruited after this time, so that results would hardly be influenced. Within the ACS among 126 IDU with known HIV-seroconversion dates, cART was defined as at least three antiretroviral drug types. A 0.0125 probability per month from 1996 onwards to start cART provided a good fit to these data. Baseline model parameters (see Table 1). The average of 100 simulations is shown (± 1 standard deviation).

Table S1 Additional HIV-induced mortality, based on data from the CASCADE (Concerted Action on SeroConversion to AIDS and Death in Europe) Collaboration. From the mortality rates per group we have subtracted non-HIV-induced background mortality, the age-specific HIV-negative injecting drug users (IDU) mortality estimated from the Amsterdam Cohort Study (ACS).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.