Key parameters related to disease progression.

The hazard of virological failure (1a) is applied for first-line ART as such, and for second-line ART together with a resistance penalty factor (1b) which depends on the time spent on failing first-line ART. Immunological failure can happen through two independent hazard functions: the other is applied only to patients on virologically failing first- or second-line ART (2a), the other for all patients irrespective of the virological status or ART regimen (2b). For clinical failure, the hazard function (3a) is used as such for patients without virological and immunological failures, and the hazard ratios (3b, 3c) are applied for patients with the corresponding failures. HIV-related mortality is calculated from a competing risk analysis of observed mortality (4a) and loss to follow-up (4b) as well as the expected mortality among lost patients (4c). The parametric hazard function for mortality (4d) is used as such for patients without virological, immunological or clinical treatment failure, and the hazard ratios (4e, 4f, 4g) are applied to patients with the corresponding failures.CI, confidence interval; ART, antiretroviral therapy; HR, hazard ratio; LTFU, loss to follow-up; n/a, not applicable* Relative decrease in second-line efficacy per year spent on failing first-line ART** Observed mortality and LTFU rates on successful first-line ART were calculated from the data and used, together with background mortality and expected mortality among patients LTFU, to calculate the corrected HIV-related mortality for the cohort*** Weighted sum of two Weibull distributionsKey parameters related to disease progression.