Prospects for HIV control in South Africa: a model-based analysis

Figures & data

Fig. 1 Mapping of effects of interventions and threats on epidemiological parameters in the Thembisa model. Dashed lines represent effects that are less direct or less well-established. *Corresponds to multiple parameters in Table 1. ANC = antenatal care, ART = antiretroviral treatment, EBF = exclusive breastfeeding, EID = early infant diagnosis, HCT = HIV counselling and testing, POC = point of care, PrEP = pre-exposure prophylaxis, sc ARVs = short-course ARVs, SWs = sex workers.

Table 1 Parameters included in uncertainty analysis

Parameter	Baseline value^a	Baseline period^a	Prior distribution	Prior mean, standard deviation	95% CI
1. HIV counselling and testing
1.1 Annual rate of first-time HIV testing in non-pregnant HIV-negative women at age 25	0.30	2011/12	Gamma (17.36, 69.4)	0.25, 0.06	0.15–0.38
1.2 Ratio of male HCT uptake to female HCT uptake at age 25 (HIV negative, non-pregnant)	0.68	2011/12	Gamma (94.4, 138.8)	0.68, 0.07	0.55–0.82
1.3 Time to the introduction of home-based HCT (in years after 2014/15)	–	–	Weibull (0.195, 0.55)	33.1, 65.1	0–21+^b
1.4 Annual rate of HIV testing through home-based HCT	–	–	Gamma (5.44, 15.56)	0.35, 0.15	0.12–0.70
1.5 Time to the introduction of intensified counselling for positives (in years after 2014/15)	–	–	Weibull (0.195, 0.55)	33.1, 65.1	0–21+^b
1.6 Coverage of intensified risk reduction counselling (of diagnosed HIV positives)	–	–	Uniform (0, 1)	0.50, 0.29	0.03–0.98
1.7 Reduction in unprotected sex following intensified counselling in HIV-positive adults	–	–	Beta (2.083, 2.083)	0.50, 0.22	0.10–0.90
2. Adult ART
2.1 Percentage of patients starting ART with CD4 <200 who are virally suppressed (VL <400 copies/ml)	0.77	2012/13	Beta (10.5, 3.136)	0.77, 0.11	0.52–0.94
2.2 Time to the introduction of universal ART eligibility (in years after 2014/15)	–	–	Weibull (0.195, 0.55)	33.1, 65.1	0–21+^b
2.3 Time to the introduction of POC CD4 testing (in years after 2014/15)	–	–	Weibull (0.195, 0.55)	33.1, 65.1	0–21+^b
2.4 Fraction starting ART immediately after diagnosis if POC CD4/universal ART available	0.40	2012/13	Beta (13.8, 9.2)	0.60, 0.10	0.40–0.79
2.5 Mean time to women starting ART if ART is not started at time of diagnosis (in months)	20	2013/14	Gamma (5.06, 0.281)	18.0, 8.00	5.9–36.7
3. Behaviour change
3.1 Average partner age difference in non-marital relationships	3.00	2011/12	Gamma (9, 3)	3.00, 1.00	1.37–5.25
3.2 Ratio of marriage rates in 2016/17 to those in 2011/12	1.00	2011/12	Gamma (25, 25)	1.00, 0.20	0.65–1.43
3.3 Ratio of sexual debut rates in 2016/17 to those in 2011/12	1.00	2011/12	Gamma (25, 25)	1.00, 0.20	0.65–1.43
3.4 Odds of condom use in marital relationships (relative to 1998)	1.78	2011/12	Gamma (14.97, 8.41)	1.78, 0.46	1.00–2.79
3.5 Odds of condom use in non-marital relationships (relative to 1998)	3.14	2011/12	Gamma (4.89, 1.56)	3.14, 1.42	1.00–6.47
3.6 Odds of condom use in SW–client relationships (relative to 1998)	3.80	2011/12	Gamma (3.84, 1.01)	3.80, 1.94	1.00–8.44
4. Prevention of mother-to-child transmission
4.1 Rate of retesting in late pregnancy	0.45	2011/12	Beta (9.02, 3.01)	0.75, 0.12	0.48–0.94
4.2 Fraction of newly diagnosed pregnant women who start ART prior to delivery	0.75	2011/12	Beta (49.73, 5.525)	0.90, 0.04	0.81–0.96
4.3 Proportionate increase in mean duration of ART prior to delivery (relative to pre-2010)	0.50	2011/12	Gamma (25, 35.71)	0.70, 0.14	0.45–1.00
4.4 Relative rate of short-course antiretroviral uptake if long-term ART not started prior to delivery	1.00	2012/13	Uniform (0, 1)	0.50, 0.29	0.03–0.98
4.5 Relative infectivity of HIV-positive women on long-term ART (breastfeeding)	0.20	2011/12	Beta (3.00, 12.00)	0.20, 0.10	0.05–0.43
4.6 Median duration of EBF in months, if EBF is initiated	2.00	2010/11	Gamma (16, 4)	4.00, 1.00	2.29–6.00
4.7 Fraction of mothers discontinuing EBF who stop breastfeeding completely	0.30	2010/11	Beta (5.16, 29.25)	0.15, 0.06	0.05–0.28
5. Medical male circumcision
5.1 Annual probability of MMC uptake in men in non-marital relationships	0.15	2013/14	Beta (1.880, 4.387)	0.30, 0.17	0.04–0.68
6. Pre-exposure prophylaxis
6.1 Effectiveness of PrEP	–	–	Beta (1.27, 1.9)	0.40, 0.24	0.03–0.88
6.2 Proportionate reduction in condom usage in PrEP users	–	–	Beta (0.8, 7.2)	0.10, 0.10	0.00–0.37
6.3 Time to the introduction of PrEP for sex workers (in years after 2014/15)	–	–	Weibull (0.195, 0.55)	33.1, 65.1	0–21+^b
6.4 Time to the introduction of PrEP for youth aged 15–24 (in years after 2014/15)	–	–	Weibull (0.195, 0.55)	33.1, 65.1	0–21+^b
6.5 Annual rate at which sex workers adopt PrEP if it is available	–	–	Gamma (2.25, 7.50)	0.30, 0.20	0.04–0.80
6.6 Annual rate at which sexually active youth adopt PrEP if it is available	–	–	Gamma (2.25, 7.50)	0.30, 0.20	0.04–0.80

^aNo baseline value or baseline period is specified for parameters that relate to interventions not yet introduced in South Africa. ^bA delay of more than 20 years means that the intervention is not introduced over the period considered in this analysis; hence the upper limit of the distribution is irrelevant. ART=antiretroviral treatment, EBF=exclusive breastfeeding, HCT=HIV counselling and testing, MMC=medical male circumcision, PrEP=pre-exposure prophylaxis, POC=point of care, SW=sex worker, VL=viral load.

Fig. 2 South African HIV incidence trends and progress towards the 90-90-90 targets. Solid lines represent averages from 1,000 simulations; dashed lines represent 95% confidence intervals (2.5 and 97.5 percentiles of distribution of model outputs). Shaded areas represent virtual elimination targets (panels a, b) and 90-90-90 targets (panels c through f). *Denominator is the number of births to HIV-positive mothers plus the number of mothers who seroconvert while breastfeeding, and numerator includes all cases of perinatal and postnatal transmission. ART = antiretroviral treatment. VL=viral load.

Fig. 3 Correlation coefficients between epidemiological parameters and HIV incidence measures over the 2015–2035 period. Zero represents no association, and the interval between the dashed lines represents correlation coefficients that are not significantly different from zero. A positive value represents a positive association between the parameter of interest and future HIV incidence. ART = antiretroviral treatment, BF = breastfeeding, EBF = exclusive breastfeeding, HBCT = home-based counselling and testing, HCT = HIV counselling and testing, MMC = medical male circumcision, POC = point of care, PrEP = pre-exposure prophylaxis, RR = relative rate.

Table 2 Multivariate analysis of epidemiological parameters affecting HIV incidence in 15–49-year-olds (2015–2035)

Unit change	RR (95% CI) of HIV incidence per unit change	p	Change required to achieve a 5% incidence reduction
10% increase in viral suppression (absolute)	0.857 (0.851–0.862)	<0.001	3.31%
10% increase in odds of condom use (ST partners), relative to 2011/12	0.972 (0.970–0.973)	<0.001	17.8%
10-year reduction in time to intensified counselling	0.916 (0.908–0.924)	<0.001	−5.82
10-year reduction in time to universal ART eligibility	0.924 (0.916–0.931)	<0.001	−6.45
10% increase in annual MMC uptake (absolute)	0.965 (0.961–0.968)	<0.001	14.2%
1-year reduction in ART delay following diagnosis	0.909 (0.899–0.918)	<0.001	−0.53
10-year reduction in time to youth PrEP	0.932 (0.925–0.940)	<0.001	−7.32
10% increase in coverage of intensified risk reduction counselling for HIV-diagnosed	0.983 (0.981–0.986)	<0.001	0.300
10-year reduction in time to HBCT	0.954 (0.947–0.962)	<0.001	−11.0
10% increase in effectiveness of intensified risk reduction counselling for HIV-diagnosed (absolute)	0.981 (0.978–0.984)	<0.001	7.0%
1-year reduction in partner age difference (ST partners)	0.968 (0.962–0.975)	<0.001	−1.59
10% increase in odds of condom use (SW–client relationships), relative to 2011/12	0.994 (0.993–0.995)	<0.001	87.4%
10% increase in efficacy of PrEP (absolute)	0.990 (0.987–0.993)	<0.001	50.5%
10% increase in annual HCT uptake in youth (absolute)	0.961 (0.950–0.972)	<0.001	12.8%
10% increase in rate of marriage, relative to 2011/12	0.989 (0.986–0.992)	<0.001	46.1%
10% increase in annual PrEP uptake by youth (absolute)	0.989 (0.986–0.993)	<0.001	47.3%
10% decrease in rate of sexual debut, relative to 2011/12	0.992 (0.988–0.995)	<0.001	−62.7%
10% increase in fraction starting ART immediately after diagnosis (absolute)	0.987 (0.981–0.994)	<0.001	40.5%
10% increase in annual HBCT uptake (absolute)	0.995 (0.990–1.000)	0.030	–^a
10% increase in male-to-female HBCT uptake ratio (absolute)	0.990 (0.980–1.000)	0.040	50.1%

Parameters are ordered according to statistical significance. Only parameters with p values <0.05 were included in the final model. ^aIt is not possible to achieve a 5% incidence reduction based only on changing this parameter.

ART=antiretroviral treatment, HBCT=home-based counselling and testing, HCT=HIV counselling and testing, MMC=medical male circumcision, PrEP=pre-exposure prophylaxis, ST=short-term, SW=sex worker.