Table 1. Main parameters used in the extended MARISA model.

Rate	Description	Source/Definition
Rates estimated in the previous MARISA model [18]
${\gamma }_{I\to D}^{k,elig/inel},{\gamma }_{D\to T_1}^{k,elig}$	Diagnosis rate, treatment initiation rate to NNRTI	Calibrated by fitting MARISA to Thembisa model (see [18]) from 2005 to 2016 (see S1 Text Section 2.2)
${\gamma }_{T_1\to S_1},{\gamma }_{T_1\to F_1},{\gamma }_{F_1\to S_1},{\gamma }_{S_1\to F_1}$	NNRTI suppression and failure rates	Estimated with individual epidemiological data from IeDEA-SA [19] (see S1 Text Table B)
${\gamma }_{T_2\to S_2},{\gamma }_{T_2\to F_2},{\gamma }_{F_2\to S_2},{\gamma }_{S_2\to F_2}$	PI suppression and failure rates	Estimated with data from IeDEA-SA [19] (see S1 Text Table B)
${\gamma }_{F_1\to T_2}^{k,elig/inel}$	Rate of switching from NNRTI to PI (before 2020)	Estimated with data from IeDEA-SA and then adjusted (see S1 Text Section 2.2)
Added rates
${\gamma }_{T_3\to S_3},{\gamma }_{T_3\to F_3},{\gamma }_{F_3\to S_3},{\gamma }_{S_3\to F_3}$	DTG suppression and failure rates	Calibrated with data from NAMSAL study [12] (see S1 Text Section 2.5)
γD→T3(t) (t≥2020)	DTG initiation rate (from 2020)	Same DTG initiation rate as for NNRTI (for DTG-inel. people) ${\gamma }_{D\to T3}≔{\gamma }_{D\to T1}^{inel}$
${\gamma }_{I\to D}^{k,elig}\left(t\right)$, ${\gamma }_{I\to D}^{k,inel}\left(t\right)$, (t≥2020)	Diagnosis rate from 2020 (distribution across DTG-eligibility classes)	${\gamma }_{I\to D}^{1,elig}\left(t\right)=p_1·{\gamma }_{I\to D}^1\left(2020\right)$ and ${\gamma }_{I\to D}^{1,inel}\left(t\right)=(1-p_1)·{\gamma }_{I\to D}^1\left(2020\right)$, for t≥2020 (see S1 Text Section 2.2)
${\gamma }_{F_1\to T_2}^{elig/inel}\left(t\right)$, (t≥2020)	Rate of switching from NNRTI to PI (DTG-inel.)	${\gamma }_{F_1\to T_2}^{inel}\left(t\right)={\gamma }_{F_1\to T_2},{\gamma }_{F_1\to T_2}^{elig}\left(t\right)=0$, for t≥2020 (see S1 Text Section 2.2)
${\gamma }_{F_3\to T_2}\left(t\right)$, (t≥2020)	Rate of switching from DTG to PI	${\gamma }_{F_3\to T_2}\left(t\right)≔{\gamma }_{F_1\to T_2}^{inel}\left(t\right)$, t≥2020
${\gamma }_{S_1\to S_3}\left(t\right)$, (t≥2020)	Switching rate from NNRTI to DTG (maintenance therapy)	${\gamma }_{S_1\to S_3}^k\left(t\right)≔1\mathrm{y}\mathrm{e}\mathrm{a}{\mathrm{r}}^{-1},t\ge 2020$
${\gamma }_{F_1\to T_3}\left(t\right)$, (t≥2020)	Switching rate from NNRTI to DTG (switch therapy)	${\gamma }_{F_1\to T_3}\left(t\right)=1\mathrm{y}\mathrm{e}\mathrm{a}{\mathrm{r}}^{-1},t\ge 2020$

DTG, dolutegravir; MARISA, Modeling Antiretroviral drug Resistance In South Africa; NAMSAL, New Antiretroviral and Monitoring Strategies in HIV-infected Adults in Low-income countries; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.