Article

The Effect of Monitoring Viral Load and Tracing Patients Lost to

Follow-up on the Course of the HIV Epidemic in Malawi: A
Mathematical Model

ESTILL, Janne Anton Markus, et al.

Reference
ESTILL, Janne Anton Markus, et al. The Effect of Monitoring Viral Load and Tracing Patients
Lost to Follow-up on the Course of the HIV Epidemic in Malawi: A Mathematical Model. Open
Forum Infectious Diseases, 2018, vol. 5, no. 5

DOI : 10.1093/ofid/ofy092
PMID : 29977952

Available at:
http://archive-ouverte.unige.ch/unige:105774

Disclaimer: layout of this document may differ from the published version.
 Open Forum Infectious Diseases
MAJOR ARTICLE

The Effect of Monitoring Viral Load and Tracing Patients

Lost to Follow-up on the Course of the HIV Epidemic in
Malawi: A Mathematical Model
Janne Estill,1,2,3 Cliff C. Kerr,4,5 Nello Blaser,3,6 Luisa Salazar-Vizcaya,3,7 Lyson Tenthani,3,8 David P. Wilson,4 and Olivia Keiser1,3
1
Institute of Global Health, University of Geneva, Geneva, Switzerland; 2Institute of Mathematical Statistics and Actuarial Science (IMSV) and 3Institute of Social and Preventive Medicine (ISPM),
University of Bern, Bern, Switzerland; 4Burnet Institute, Melbourne, Australia; 5School of Physics, University of Sydney, Sydney, Australia; 6Department of Mathematics, University of Bergen,
Bergen, Norway; 7Department of Infectious Diseases, University Hospital Bern, Bern, Switzerland; 8Supporting Operational AIDS Research (Project SOAR), Population Council, Blantyre, Malawi

Background. Antiretroviral therapy (ART) reduces HIV transmission, but treated patients may again become infectious. We
used a mathematical model to determine whether ART as prevention is more effective if viral load (VL) is routinely monitored and
patients lost to follow-up (LTFU) traced.
Methods. We simulated ART cohorts to parameterize a deterministic transmission model calibrated to Malawi. We investigated
the following strategies for improving treatment and retention: monitoring VL every 12 or 24 months, tracing patients LTFU, or
a generic strategy leading to uninterrupted treatment. We tested 3 scenarios, where ART scale-up continues at current (Universal
ART), reduced (Failed scale-up), or accelerated speed (Test&Treat).
Results. In the Universal ART scenario, between 2017 and 2020 (2050), monitoring VL every 24 months prevented 0.5% (0.9%),
monitoring every 12 months prevented 0.8% (1.4%), tracing prevented 0.3% (0.5%), and uninterrupted treatment prevented 5.5%
(9.9%) of HIV infections. Failed scale-up resulted in 25% more infections than the Universal ART scenarios, whereas Test&Treat
resulted in 7%–8% less.
Conclusions. Test&Treat reduces transmission of HIV, despite individual cases of treatment failure and ART interruption.
Whereas viral load monitoring and tracing have only a minor impact on transmission, interventions that aim to minimize treatment
interruptions can further increase the preventive effect of ART.
Keywords. antiretroviral therapy; HIV; loss to follow-up; mathematical model; monitoring; transmission.

Antiretroviral therapy (ART) suppresses the HIV-RNA concen- and Test&Treat have been widely discussed, but the evidence
tration (viral load [VL]) in people living with HIV (PLHIV), is not conclusive. Some studies have suggested that successful
reducing transmission risk [1, 2]. Since 2016, the World Health Test&Treat programs could eradicate the epidemic, but others
Organization (WHO) has recommended that all PLHIV begin predict only minor benefit on the population level [4–6].
ART immediately [3]. Preventing transmission through treat- Treatment failures, poor adherence, and frequent dropout
ment—“treatment as prevention” (TasP)—was an argument from care impair the effect of TasP [7–9]. Viral load monitoring
for expanding eligibility for ART to wider groups of patients and tracing patients lost to follow-up (LTFU) can support viral
and ultimately to all PLHIV. An intensive TasP intervention suppression in treated patients. The WHO has recommended
called “Test&Treat” screens the population regularly for HIV routine VL monitoring as its preferred monitoring strategy
and immediately starts all patients who test positive on ART. In since 2013. For several years, VL monitoring in sub-Saharan
2014, UNAIDS launched its global “90-90-90” target, with the Africa was available only in South Africa and Botswana, and
goal of substantially reducing transmission. The benefits of TasP a few research sites. New testing technologies have made rou-
tine VL monitoring easier, but coverage remains limited [10].
Patients LTFU are frequently traced in sub-Saharan Africa
through phone calls or home visits to those who do not return
Received 15 November 2017; editorial decision 11 April 2018; accepted 24 April 2018. to pick up their antiretrovirals [11].
Correspondence: J. Estill, PhD, Institute of Global Health, University of Geneva, c/o IMSV,
University of Bern, Alpeneggstrasse 22, 3012 Bern, Switzerland ([email protected]). We have developed mathematical models to test the effect of
Open Forum Infectious Diseases® VL monitoring [12] and tracing patients LTFU [13] on reduc-
© The Author(s) 2018. Published by Oxford University Press on behalf of Infectious Diseases ing potential transmission of HIV. We found that these inter-
Society of America. This is an Open Access article distributed under the terms of the Creative
Commons Attribution-NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/ ventions could prevent patients on ART from transmitting the
by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any infection, but our analyses only evaluated a patient’s potential
medium, provided the original work is not altered or transformed in any way, and that the work
is properly cited. For commercial re-use, please contact [email protected]
for transmitting the virus. The future course of the HIV epi-
DOI: 10.1093/ofid/ofy092 demic depends on other important factors, such as transmission

Monitoring, Tracing, and the HIV Epidemic • OFID • 1

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018
 from untreated patients, behavioral preferences, and contact patient’s time on ART into 32 states (Figure 1A; Supplementary
patterns. In this study, we took the next step and developed a Table 1) that accounted for virological and immunological
transmission model to assess the potential effect of VL moni- treatment response (successful or failing), ART regimen (first-
toring and tracing on the HIV epidemic. or second-line), and retention (on or off ART). We did not
consider further treatment options beyond second-line. We
METHODS separated HIV-related and HIV-unrelated mortality. We lim-
Setting ited the number of off-ART episodes to 1 to simplify the struc-
We modeled the HIV epidemic of Malawi. In 2016, the esti- ture. Patients were simulated for 10 years after initiating ART.
mated adult HIV prevalence in Malawi was 9.2%, and about Virological and immunological treatment response was based
680 000 patients (65% of all PLHIV) were on ART [14]. Until on previous analyses of routine data from sites in South Africa
recently, Malawi relied on clinical monitoring and occasional with 6-monthly VL and CD4 monitoring (Supplementary
CD4 counts to monitor treatment response, but since 2011, the Table 2) [12, 19, 20]. The risk of virological failure corresponds
Ministry of Health has recommended monitoring VL at regular to a cumulative risk of 5.7% 1 year and 12.9% 5 years after
24-month intervals [15]. In 2016, 19% of all patients on ART ART initiation. One year after ART initiation, the sensitivity
in Malawi had had at least 1 VL test. Several sites trace patients of immunological criteria was 7% and specificity 12%; 5 years
who miss appointments [16]. after ART initiation, the sensitivity and specificity were 26%
and 45%, respectively. A resistance penalty factor was applied
Mathematical Model
to increase the risk of failure depending on how long the patient
The model consists of an individual-based simulation of disease
had spent on a failing regimen or off ART. Parameters related to
progression and a deterministic transmission model.
treatment interruptions and return to care with or without trac-
Disease Progression Simulation ing were derived from data from Lighthouse and Martin Preuss
We used the R package gems to develop an individual-based Centre clinics in Lilongwe, Malawi [13]. Patients switched to
simulation model for disease progression in patients who have second-line ART after either virological or immunological
started ART (“treatment model”) [17, 18]. We divided the treatment failure, depending on the monitoring strategy.

A B
Chronic Chronic
Primary
Virologically Successful Virologically Susceptible asymptomatic symptomatic Treated AIDS
Successful infection
failing 1st- 2nd-line failing 2nd- infection infection
1st-line ART
line ART ART line ART

Successful Virologically Successful Virologically

1st-line ART failing 1st- 2nd-line ART failing 2nd-
Discordant line ART line ART
Discordant
immunological Immunological immunological Immunological
failure failure failure failure
Male
15–49 Male
High risk 15–49 Male
Low risk 50–
Children
0–14
Female Female
In care (on ART) Female 15–49 50–
15–49 Low risk
High risk
Death

Interrupted treatment

Death

Back in care (on ART)

Death

Figure 1. Schematic representation of the mathematical model. A, Flow of patients in the treatment model. White boxes represent stages with suppressed viral load, and
gray boxes represent stages with continuously elevated viral load. “Discordant” immunological failure refers to a decline in CD4 cell count fulfilling the failure criteria under
suppressed viral load; this condition will not reverse upon switch to second-line therapy. The flow described on the upper half is applicable to patients on ART, including
those who returned after ART interruption. While progressing along the stages of treatment response (upper graph), the patients may also interrupt and restart treatment or
die (lower graph). B, Transmission model. The upper graph shows the course of the HIV infection, and the lower graph the flow through age, sex, and risk group. Black arrows
show flows between compartments, and gray lines show sexual contact patterns. Abbreviation: ART, antiretroviral therapy.

2 • OFID • Estill et al

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018
 The output of the simulations is a matrix of the entry times to also patients who interrupted ART. “AIDS” represents the last
all states for each patient. For each simulated patient, we sam- year of untreated HIV, where the patient has a CD4 cell count
pled the VL values for different stages (virologically successful, below 200 cells/μL or severe AIDS-defining opportunistic
failing, or off ART). We then calculated the mean log10 viral infections.
load over time in each strategy to inform the transmission The model is solved numerically. We calculated infec-
model [12]. We ran the model for the following nine strategies: tiousness from the estimated per-act transmission probabil-
ity and assumed frequency of unprotected sex acts (Table 1).
i. no VL monitoring, no tracing; Infectiousness was multiplied by 20 during the acute stage,
ii. no VL monitoring, active tracing; and by a factor that depended on the monitoring and retention
iii. no VL monitoring, uninterrupted treatment; strategy during ART. This factor was estimated for each sce-
iv. VL monitoring at 24-month intervals, no tracing; nario from the treatment model. We also allowed infectious-
v. VL monitoring at 24-month intervals, active tracing; ness to decrease over time. The progression of patients across
vi. VL monitoring at 24-month intervals, uninterrupted the stages of HIV was estimated from the literature.
treatment; We calculated the ART initiation rate at based on the diag-
vii. VL monitoring at 12-month intervals, no tracing; nosis rate, availability, and eligibility criteria for ART and the
viii. VL monitoring at 12-month intervals, active tracing; expected progression of the patient’s CD4 cell count (Table 1).
ix. VL monitoring at 12-month intervals, uninterrupted Antiretroviral therapy was generally unavailable until 2003,
treatment. provided to symptomatic patients only between 2003 and 2015,
“No VL monitoring” means that treatment failure is again and increasingly to asymptomatic patients after 2015. The
determined by clinical symptoms and CD4 counts only. The rate of ART initiation among symptomatic patients increased
WHO recommends VL monitoring at 12-month intervals, but in 2011 to match the change in CD4-based eligibility criteria
VL monitoring at 24-month intervals is current practice in (from <250 to <350 cells/μL). A separate ART initiation rate
Malawi. Treatment failure is confirmed 3 months after detec- was applied to women from 2011 on to take into account the
tion, and the patient is switched to second-line therapy after “Option B+” strategy of treating all pregnant and breastfeeding
a random delay. In strategies with no tracing, patients LTFU women [23].
may only return spontaneously. With active tracing, all patients We ran the simulation for 1975–2016 with the best available
LTFU are traced 3 weeks after missing their appointment parameter estimates to calibrate the model and compared the
and return to care at a higher rate than without tracing [16]. model’s outputs with observed data, including total population
Strategies with uninterrupted treatment represent the ideal but size and number of patients on ART each year from 2010. We
unrealistic scenario that treatment is never interrupted. This also compared our results to the estimates of the UNAIDS EPP/
scenario sets the theoretical limit for improving retention. In Spectrum model on prevalence and number of annual new
all scenarios, we assumed that patients retained in care adhered infections. We then calibrated the following parameters: per-act
to treatment, as in the data that were used for parameterization transmission probability, birth rate, and year and magnitude of
[12, 13, 19, 20]. decrease in infectiousness.
We considered 3 possible scenarios for treatment access from
Transmission Model 2017 (see Table 2 for input parameters):
We developed a deterministic transmission model (“trans-
1) Failed scale-up: Recommendations to treat all PLHIV
mission model”) to represent the HIV epidemic in Malawi
are not successfully implemented, and treatment
between 1975 and 2050. The model consists of 40 compart-
remains restricted to the sickest patients. Women can
ments (Figure 1B) representing HIV status (susceptible; pri-
start ART earlier because “Option B+” has already been
mary, asymptomatic chronic, or symptomatic chronic infection;
implemented.
ART; AIDS), age (children <15 years, adults 15–49 years, adults
2) Universal ART: The policy introduced in 2015 contin-
≥50 years), sex (not distinguished for children), and risk behav-
ues, and ART can be initiated during the asymptom-
ior (high or low, except for children, older adults, and AIDS
atic stage, although at a lower rate than in symptomatic
patients). “Acute infection” represents early HIV infection,
patients because of barriers to testing.
when the risk of onward transmission is highest [21]. During
3) Test&Treat: Intensive screening is added to the Universal
“asymptomatic chronic infection,” CD4 cell counts are expected
ART scenario, and ART initiation rate increases.
to be >350 cells/μL. In “symptomatic chronic infection,” CD4
cell count is below 350 cells/μL and the patient is in WHO clin- We ran each of the 3 access scenarios for all 9 monitoring
ical stage ≥2. Although there is no strict dependency between and retention strategies, (i)–(ix). We reported the number of
symptoms and clinical stage, these definitions roughly correlate expected new infections and AIDS-related deaths in 2020, 2030,
[22]. “ART” represents patients who ever started ART, including and 2050 and calculated the relative reduction in new infections

Monitoring, Tracing, and the HIV Epidemic • OFID • 3

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018
 Table 1. Prior Parameter Values of the Transmission Model: Fixed Parameters With Identical Values in All Scenarios and Strategies

Demographic Parameters Value Source

Birth rate, default value,a y-1 0.16 [24]

Non-HIV related mortality: children aged <15 y, y-1 0.0142 [25]
Non-HIV related mortality: males aged 15–<50 y, y-1 0.0059 [25]
Non-HIV related mortality: females aged 15–<50 y, y-1 0.0051 [25]
Non-HIV related mortality: males aged 50 y and above, y-1 0.0478 [25]
Non-HIV related mortality: females aged 50 y and above, y-1 0.0422 [25]
Mixing and sexual behavior
Proportion of young males engaging in high-risk behavior 0.10 Assumption
Proportion of young females engaging in high-risk behavior 0.05 Assumption
Mean duration of high-risk behavior among males, y 25 Assumption
Mean duration of high-risk behavior among females, y 10 [26, 27], assumption
Mean number of unprotected sex acts/y with regular partner 50 Assumption
Mean number of unprotected sex acts/y with casual partners: low-risk individuals 1 Assumption
Mean number of unprotected sex acts/y with casual partners: high-risk individuals 100 Assumption
Mixing (proportion of casual partners sampled exclusively from own risk group) 0.5 Assumption
Sexual transmission
Per-act transmission probability, male-to-female (chronic untreated infection), default valuea 0.00155 [2]
Per-act transmission probability, female-to-male (chronic untreated infection), default valuea 0.00079 [2]
Risk ratio for transmission probability during acute infection 20 [28]
Mother-to-child transmission
Probability of mother-to-child transmission if the mother is acutely infected 0.313 [29], assumption
Probability of mother-to-child transmission if the mother is chronically infected 0.250 [29, 30]
Probability of mother-to-child transmission if the mother is treated 0.050 [29], assumption
Natural progression of HIV
Mean duration of acute infection, y 0.25 [28]
Mean duration of asymptomatic stage, y 4.8 [31]
Mean duration of symptomatic stage before AIDS, y 5.2 [31]
HIV related mortality during symptomatic stage, y-1 0.1 Assumption
HIV related mortality during AIDS, y-1 1
Treatment
Introduction of ART, y 2003 [23]
Eligibility at CD4 <350 cells/μL 2011 [23]
Universal ART eligibility 2015
Introduction of “Option B+” 2011 [23]
Initial conditions in 1975b
Total population size 5 302 000 [32]
Male-to-female ratio among adults aged 15–<50 y 1:1 [32]
Male-to-female ratio among adults aged 50 y and above 47:53 [32]
Proportion of children aged <15 y 0.469 [32]
Proportion of people aged 50 y and above 0.025 [32]

Abbreviation: ART, antiretroviral therapy; y, year.

a
Default value was adjusted during the calibration using a constant coefficient (Supplementary Table 5 for values).
b
HIV prevalence and risk behavior in 1975 were determined in calibration (Supplementary Table 5).

compared with strategy (i) of the corresponding access scenario RESULTS

for 2 time windows: 2017–2020 and 2017–2050. The mean (log10 scale) viral load of the patients simulated in
Sensitivity Analyses the treatment model for 10 years ranged from 63 to 104 copies/
We conducted 2 sensitivity analyses to assess the impact of mL across the scenarios (Supplementary Table 4). The corre-
uncertainty around the parameters (Supplementary Table 3). sponding per-act transmission risk from patients who started
In the first analysis, we assumed that the risk of virological fail- ART was 14–17 times lower than the risk from chronically
ure would considerably increase over time. Second, we con- infected, untreated patients. The per-act transmission prob-
ducted an analysis where all-cause mortality was reduced from ability determined via calibration of the transmission model
2017 onwards, and treated patients could no longer proceed was 20% higher than the literature-based prior (Supplementary
to AIDS. Table 5).

4 • OFID • Estill et al

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018
 Table 2. Parameters of the Transmission Model: Parameters With Values Depending on Time Period and ART Initiation Scenario

2003–2010 2005–2014 2015–2016 Failed Scale-up Universal ART Test&Treat

Rate of starting ART, asymptomatic, adults 0 0 0.5 0.1 0.5 1

Rate of starting ART, symptomatic, adults 0.05 2 1 1 1 1
Rate of starting ART, asymptomatic, children 0 10 10 10 10 10
Rate of starting ART, symptomatic, children 0.3 10 10 10 10 10
Rate of starting ART due to PMTCT, asymptomatic, 0 0.2 0.2 0.2 0.2 0.2
women
Rate of starting ART due to PMTCT, symptomatic, 0.15 0.2 0.2 0.2 0.2 0.2
women
Rate of AIDS for patients on treatment 0.05 0.05 0.05 0.01 0.01 0.01

All rates are per person-year.

Abbreviations: ART, antiretroviral therapy; PMTCT, prevention of mother-to-child transmission.

The transmission model’s pre-2017 results were in line with relative differences remained the same across all scenarios of
observed data and the UNAIDS EPP/Spectrum predictions. treatment scale-up and retention/tracing.
Prevalence among adults aged 15–49 years followed the upper In the Universal ART scenario without viral load monitor-
limit of the UNAIDS estimates (Supplementary Figure 1) [14]. ing, actively tracing patients LTFU reduced the average number
The total number of PLHIV in 2016 was about 10% lower than of new infections in 2017–2020 by 0.3% from 53 400 to 53 300,
predicted by UNAIDS. The largest discrepancy was in annual or in 2017–2050 by 0.5% from 260 900 to 259 500 (Figure 3,
new infections, which our model predicted to be about 20% Table 3). When treatment interruptions were eliminated, the
higher until 2013. In the last few years, the new infections number of new infections decreased in 2017–2020 by 5.5% to
declined rapidly, going below the UNAIDS lower limit in 2016. 50 500, or in 2017–2050 by 9.9% to 235 100. The relative benefit
According to our model, 695 000 patients were on ART in 2016, of tracing and improved retention was similar across scenarios
in line with observed estimates (680 000). In 2016, we predicted of treatment scale-up and monitoring.
23 100 AIDS-related deaths, whereas UNAIDS predicted 24 000. The total number of PLHIV also decreased over time. In 2020,
In all scenarios, prevalence continued to decline. Prevalence we predicted 839 200–849 500 PLHIV, and in 2050, 465 300–
among adults aged 15–49 years was 7.8%–7.9% in 2020, 4.4%– 538 000 PLHIV. Because of the decreasing number of PLHIV,
4.8% in 2030, and 1.4%–1.7% in 2050, depending on the sce-
nario (Supplementary Figure 2). The number of annual new
New HIV infections in Malawi 2017–2050
infections also decreased rapidly, ranging 8800–13 400 in 2020, 18000
6400–9900 in 2030, and 3500–6100 in 2050 across the scenarios
(Figure 2, Table 3). The number of AIDS deaths was 12 000 in 16 000

2020, 7400–7900 in 2030, and 4600–5400 in 2050.

14 000
Of the factors that differed between scenarios, results were
most sensitive to the overall treatment and testing scenario. In 12 000
the Universal ART scenario, our model predicted 50 100–53 400
New infections

10 000
new infections between 2017 and 2020, or 231 600–260 900
between 2017 and 2050. In the Failed scale-up scenario, the 8000
ranges were 56 800–60 200 until 2020 (13% higher than with
6000
Universal ART) or 288 900–326 800 until 2050 (25% higher than
with Universal ART). With Test&Treat, the ranges dropped to 4000
46 200–49 400 until 2020 or 214 900–241 500 until 2050, about
2000
7%–8% lower than with Universal ART.
The differences between monitoring and retention strategies 0
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050

were smaller (Figure 3, Table 3). Assuming the Universal ART

scenario, current retention, no tracing, and no VL monitor- Year
ing, 53 400 patients were infected between 2017 and 2020, or
260 900 between 2017 and 2050. Monitoring VL at 24-month Figure 2. Number of annual predicted new HIV infections in Malawi between 2017
and 2050. Gray/black, Failed scale-up; pink/red, Universal ART; blue, Test&Treat.
intervals lowered these numbers by 0.5% to 53 200 until 2020,
Light color, no viral load monitoring; intermediate color, 24-monthly viral load moni-
or by 0.9% to 258 700 until 2050; and, at 12-month intervals, by toring; dark color, 12-monthly viral load monitoring. Dotted curves, no tracing; dashed
0.8% to 53 000 until 2020, or by 1.4% to 257 200 until 2050. The curves, tracing patients lost to follow-up; solid curves, no treatment interruptions.

Monitoring, Tracing, and the HIV Epidemic • OFID • 5

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018
 Total New Infections Total New Infections
Infections prevented by viral load monitoring, tracing or other
prevention interventions 2017–2050

216 705 (10.3%)

2017–2050

214 927 (11.0%)

240 265 (0.5%)
218 062 (9.7%)

239 495 (0.9%)

239 615 (0.8%)

236 160 (2.2%)

238 153 (1.4%)
1200

241 548 (ref)

1000

Prevented infections
800
Test&Treat

49 268 (0.3%)
46 553 (5.8%)

49 175 (0.5%)

49 190 (0.5%)
46 384 (6.1%)

49 013 (0.8%)

46 163 (6.6%)
2017–2020

48 772 (1.3%)
49 422 (ref)

600

400
New Infections

200
2020

9751
9705
8904

9678

9682
8854

9630

9559

8789
0

2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
Year
Total New Infections

233 591 (10.5%)

231 637 (11.2%)

259 498 (0.5%)
235 083 (9.9%)

258 650 (0.9%)

258 782 (0.8%)

254 983 (2.3%)

257 175 (1.4%)
2017–2050

260 908 (ref)

Figure 3. Number of annual new HIV infections prevented by routine viral load
Percentages in parentheses refer to the reduction compared with scenario (i) (first row) of the corresponding access scenario (Failed scale-up, Universal ART, or Test&Treat).
monitoring, tracing patients lost to follow-up, or other retention support interven-
tions in the Universal ART scenario. Light gray, no viral load monitoring; dark gray,
24-monthly viral load monitoring; black, 12-monthly viral load monitoring. Dotted
curves, no tracing; dashed curves, tracing patients lost to follow-up; solid curves,
Universal ART

no treatment interruptions.
New infections Total New Infections

53 262 (0.3%)
50 503 (5.5%)

53 168 (0.5%)

53 182 (0.4%)
50 331 (5.8%)

53 003 (0.8%)

50 106 (6.2%)
52 758 (1.2%)
2017–2020

53 419 (ref)

the total number of patients on treatment barely increased, even

with accelerated ART scale-up, and started to decrease after
2019. In 2020, 778 100–794 900 patients were on ART, dropping
infections

in 2050 to 457 700–519 400. Decreased incidence and faster

2020

2020
9955

10 762

10 766
9903

10 712

10 638

9835

ART scale-up lowered the number of patients on ART, which,

in 2050, was 8% higher with Failed scale-up, and 2% lower with
Number of Predicted HIV Infections in Malawi in Different Scenarios

Test&Treat, than with Universal ART.

Total New Infections

293 353 (10.2%)

291 420 (10.8%)

288 888 (11.6%)

324 965 (0.6%)

323 868 (0.9%)

324 039 (0.8%)

319 122 (2.3%)

321 959 (1.5%)

Sensitivity Analyses
2017–2050

326 790 (ref)

If the risk of virological failure increased over time, monitoring

strategy no longer had an influence on the number of new infec-
tions (Supplementary Table 6). If we assumed lower mortality,
the absolute number of new infections increased, but the rela-
Total New Infections
Failed Scale-up

tive benefit of the monitoring and retention support strategies

60 023 (0.3%)
57 218 (4.9%)

59 927 (0.4%)

59 942 (0.4%)
57 044 (5.2%)

59 760 (0.7%)

56 815 (5.6%)
59 511 (1.1%)
2017–2020

did not differ from the main analysis (Supplementary Table 7).
60 182 (ref)

DISCUSSION

In this modeling study, we found that the number of new HIV

Abbreviations: ART, antiretroviral therapy; VL, viral load.

infections will likely continue to decrease rapidly in all test-

New Infections

ing, treatment, retention, and monitoring scenarios. The most

13 363
13 314
12 438

13 284

13 289
12 384

13 232

13 154

12 312
2020

important factors associated with the speed of decrease were the

rates of ART initiation and treatment interruptions. VL moni-
toring and tracing lost patients reduced new infections, but only
v) 24-m VL monitoring, tracing
i) CD4 monitoring, no tracing

vii) 12-m VL monitoring, no

iv) 24-m VL monitoring, no

vi) 24-m VL monitoring, no

ix) 12-m VL monitoring, no

ii) CD4 monitoring, tracing

minimally. Interventions designed to keep patients in care with-

viii) 12-m VL monitoring,
iii) CD4 monitoring, no

out interruptions could be much more beneficial.

interruptions

interruptions

interruptions

The results of VL monitoring were expected. In an earlier

study, we predicted that routine VL monitoring could prevent
tracing

tracing

tracing
Table 3.

up to one-third of transmissions from treated patients [12]. But

the proportion of new infections attributable to patients on

6 • OFID • Estill et al

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018
 ART is low, so it is unsurprising that the overall effect of VL thus costing about $60 000 to preventing a single infection. This
monitoring was small. Our results are in line with other mode- is clearly more expensive than treating the infected patient for
ling studies. For example, in 2014, Braithwaite et al. noted that life. Annual VL testing may, however, also benefit the patient
VL monitoring was more cost-effective if its effect on transmis- and reduce transmitted drug resistance. A more complete per-
sion was considered, but extending ART eligibility, as the WHO spective is needed to assess cost-effectiveness.
recommended, had much more impact [33]. Some potential
advantages we did not include in our model could also increase Strengths and Limitations
the benefit of VL monitoring. Patients whose ART is failing des- We used a structurally simple, deterministic compartmental
pite good adherence may believe they are not infectious, and transmission model to produce results that closely match the
thus be likely to engage in unprotected sex [34]. Patients on fail- data and projections returned by other established mathe-
ing ART may also carry resistant strains of HIV. Their spread matical models. Our transmission model was informed by an
could limit the potency of available firstline regimens, increas- individual-based simulation of the progression of HIV under
ing long-term mortality [35]. ART, parameterized with routine cohort data. We expect that
Loss to follow-up remains a major problem in sub-Saharan our results are generalizable to many Southern and Eastern
Africa. Although much LTFU can be explained by undocu- African countries with similar epidemics and relatively high
mented deaths and transfers between facilities, about one- ART coverage [14, 41]. We also expect that the relative decrease
third of lost patients have probably stopped treatment or are in new infections with VL monitoring or tracing will be simi-
taking ART irregularly [16]. There is a broad range of reasons lar in other countries in this region. Our approach of linking a
contributing to interrupting treatment [36]. Especially during detailed simulation of disease progression to a relatively simple
the introduction of the Option B+, there were also concerns deterministic transmission model can also be adapted to other
about retention rates among women who start ART in the settings and questions.
asymptomatic stage. Experience from this program has, how- Our study also had several limitations. The data we used to
ever, shown that long-term retention is feasible even among parameterize the model covered only a few years of follow-up
patients who start ART in an early stage of the infection [37]. and may thus not be accurate for long-term projections. We
In a prior study, we found that tracing lost patients did not sub- also did not consider other future changes in care, such as
stantially reduce expected transmission from patients who had new antiretroviral regimens. New regimens may lead to sub-
started ART [13]. This study confirmed our finding. Tracing stantially lower failure rates. If third-line ART becomes widely
rarely locates patients who have moved or are traveling, or available in the future, it will likely further reduce the number
who provided an incorrect address or phone number. Many of new infections, influencing also the relative benefit of the
patients refuse to return to care, or interrupt treatment again monitoring and retention interventions. Apart from a resist-
later [16]. If treatment interruptions could be eliminated, the ance penalty factor, first- and second-line failures were sam-
overall number of new infections would drop by 5% in the next pled independently, ignoring the possible individual factors
few years, but perfect retention is unrealistic. Retention may that may result in failure through, for example, poor adherence.
be increased by further decentralizing treatment services, pro- Several parameters had to be estimated using assumptions and
viding larger supplies of ART per visit, or SMS reminders [16, calibration. We aimed to keep the transmission model as sim-
38–40]. ple as possible and did not divide patients according to VL
A recent survey found that, in Malawi, 89% of diagnosed strata during treatment. Instead, we calculated average infec-
PLHIV were on ART, but only 73% of all PLHIV had been tiousness for each stage of the disease, so our model may not be
diagnosed. Moreover, 91% of patients tested for VL in Malawi able to catch variation in infectiousness over time. We included
were virally suppressed [41]. Our results suggest that continu- only heterosexual transmission. Finally, we did not account for
ing screening to find more PLHIV in Malawi is the most effect- geographical variability, which can play a major role in the
ive strategy for meeting the ambitious 90-90-90 target. Our dynamics of the epidemic: district-level HIV prevalence in
model predicted that, by 2020, in all scenarios, at least 90% of Malawi ranges from 3% in some northern districts to over 18%
all PLHIV would be on ART. We may slightly overestimate the in the south.
number of PLHIV on ART, as our estimate includes patients There were also discrepancies between our and UNAIDS’
who have interrupted ART. But it is clear the 90-90-90 target modeling estimates. In particular, our model estimated the
can realistically be met in Malawi. annual new infections higher than UNAIDS. This resulted also
Cost-effectiveness calculations of VL monitoring and tracing in a slightly differing pattern in the total number of PLHIV. We
patients LTFU should account for benefits to both individuals predicted a moderate decrease since 2000, whereas UNAIDS
and the population. If VL test costs could be suppressed to $10 suggests that the number remained stable or even increased
[10], testing each patient annually would cost about $6 million. slightly. However, the UNAIDS estimates are also based
This would prevent only about 100 new infections each year, on models built on a limited amount of data. Our model’s

Monitoring, Tracing, and the HIV Epidemic • OFID • 7

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018
 projections on number of patients on ART matched closely with Asia: implications for ‘test and treat’ prevention strategy. AIDS Patient Care STDS
2013; 27:681–91.
the reported data. However, our estimate also includes patients 8. Wanyama JN, Nabaggala MS, Wandera B, et al. Significant rates of risky sexual
who have stopped ART. behaviours among HIV-infected patients failing first-line ART: a sub-study of the
Europe-Africa Research Network for the Evaluation of Second-line Therapy trial.
Conclusions Int J STD AIDS 2017; 29:287–97.
9. Cohen MS, Smith MK, Muessig KE, et al. Antiretroviral treatment of HIV-1
Malawi’s response to the HIV epidemic has been successful, prevents transmission of HIV-1: where do we go from here? Lancet 2013;
and we expect the number of new infections to continue to 382:1515–24.
10. Roberts T, Cohn J, Bonner K, Hargreaves S. Scale-up of routine viral load testing
fall, regardless of the strategies of ART initiation, monitoring,
in resource-poor settings: current and future implementation challenges. Clin
and retention support. To reach the 90-90-90 goal, scaling-up Infect Dis 2016; 62:1043–8.
ART and preventing treatment interruptions must be a prior- 11. McMahon JH, Elliott JH, Hong SY, et al. Effects of physical tracing on esti-
mates of loss to follow-up, mortality and retention in low and middle income
ity. Though only a small number of infections can be prevented country antiretroviral therapy programs: a systematic review. PLoS One 2013;
by viral load monitoring and tracing patients lost to follow-up, 8:e56047.
12. Estill J, Aubrière C, Egger M, et al; IeDEA Southern Africa. Viral load monitor-
the population-level effects of these interventions should be ing of antiretroviral therapy, cohort viral load and HIV transmission in Southern
included in future cost-effectiveness evaluations. Africa: a mathematical modelling analysis. AIDS 2012; 26:1403–13.
13. Estill J, Tweya H, Egger M, et al. Tracing of patients lost to follow-up and HIV
The substantial heterogeneity in the Malawian HIV epidemic transmission: mathematical modeling study based on 2 large ART programs in
across regions and population groups may require interventions Malawi. J Acquir Immune Defic Syndr 2014; 65:e179–86.
14. Joint United Nations Programme on HIV/AIDS (UNAIDS). AIDSinfo. Available
targeted to specific groups or regions to increase treatment effi-
at: aidsinfo.unaids.org. Accessed 31 October 2017.
ciency and acceptability. Test&Treat reduces transmission of 15. Malawi Ministry of Health. Clinical Management of HIV in Children and Adults.
HIV, despite individual cases of treatment failure and ART Available at: http://apps.who.int/medicinedocs/documents/s18802en/s18802en.
pdf. Accessed 31 October 2017.
interruption, but the full benefits of the treatment as a preven- 16. Tweya H, Feldacker C, Estill J, et al. Are they really lost? “True” status and rea-
tion strategy cannot be reaped unless ART is improved with sons for treatment discontinuation among HIV infected patients on antiretroviral
therapy considered lost to follow up in urban Malawi. PLoS One 2013; 8:e75761.
interventions that support retention and viral suppression. 17. Blaser N, Salazar Vizcaya L, Estill J, et al. Gems: an R package for simulating from
disease progression models. J Stat Softw 2015; 64:1–22.
Supplementary Data 18. Salazar Vizcaya L, Blaser N, Gsponer T. Gems: Generalized multistate simulation
model. Available at: http://cran.r-project.org/web/packages/gems/index.html.
Supplementary materials are available at Open Forum Infectious Diseases
Accessed 31 October 2017.
online. Consisting of data provided by the authors to benefit the reader,
19. Estill J, Egger M, Blaser N, et al; IeDEA Southern Africa. Cost-effectiveness of
the posted materials are not copyedited and are the sole responsibility of point-of-care viral load monitoring of antiretroviral therapy in resource-limited
the authors, so questions or comments should be addressed to the corre- settings: mathematical modelling study. AIDS 2013; 27:1483–92.
sponding author. 20. Estill J, Egger M, Johnson LF, et al; IeDEA Southern Africa Collaboration.
Monitoring of antiretroviral therapy and mortality in HIV programmes in
Acknowledgments Malawi, South Africa and Zambia: mathematical modelling study. PLoS One
2013; 8:e57611.
We thank Kali Tal for commenting on and editing the manuscript. 21. Powers KA, Ghani AC, Miller WC, et al. The role of acute and early HIV infection
Financial support. This work was supported by the Swiss National in the spread of HIV and implications for transmission prevention strategies in
Science Foundation (grant numbers PP00P3_163878, PDFMP3_137106). Lilongwe, Malawi: a modelling study. Lancet 2011; 378:256–68.
Potential conflicts of interest. All authors: no reported conflicts of 22. Edathodu J, Ali B, Alrajhi AA. CD4 validation for the World Health Organization
interest. All authors have submitted the ICMJE Form for Disclosure of classification and clinical staging of HIV/AIDS in a developing country. Int J
Potential Conflicts of Interest. Conflicts that the editors consider relevant to Infect Dis 2009; 13:243–6.
the content of the manuscript have been disclosed. 23. Harries AD, Ford N, Jahn A, et al. Act local, think global: how the Malawi experi-
ence of scaling up antiretroviral treatment has informed global policy. BMC
Public Health 2016; 16:938.
References 24. The World Bank. World Bank Open Data. Available at: data.worldbank.org.
1. Cohen MS, Chen YQ, McCauley M, et al; HPTN 052 Study Team. Prevention Accessed 31 October 2017.
of HIV-1 infection with early antiretroviral therapy. N Engl J Med 2011; 25. Lopez AD, Mathers CD, Ezzati M, et al. Global and regional burden of disease
365:493–505. and risk factors, 2001: systematic analysis of population health data. Lancet 2006;
2. Quinn TC, Wawer MJ, Sewankambo N, et al. Viral load and heterosexual trans- 367:1747–57.
mission of human immunodeficiency virus type 1. Rakai Project Study Group. N 26. Elmes J, Nhongo K, Ward H, et al. The price of sex: condom use and the determi-
Engl J Med 2000; 342:921–9. nants of the price of sex among female sex workers in eastern Zimbabwe. J Infect
3. World Health Organization. Consolidated guidelines on the use of antiretro- Dis 2014; 210(Suppl 2):S569–78.
viral drugs for treating and preventing HIV infection: Recommendations 27. Musyoki H, Kellogg TA, Geibel S, et al. Prevalence of HIV, sexually transmit-
for a public health approach. Available at: http://apps.who.int/iris/bitstr ted infections, and risk behaviours among female sex workers in Nairobi, Kenya:
eam/10665/85321/1/9789241505727_eng.pdf?ua=1. Accessed 31 October 2017. results of a respondent driven sampling study. AIDS Behav 2015; 19(Suppl
4. Granich RM, Gilks CF, Dye C, et al. Universal voluntary HIV testing with imme- 1):S46–58.
diate antiretroviral therapy as a strategy for elimination of HIV transmission: a 28. Blaser N, Wettstein C, Estill J, et al. Impact of viral load and the duration of pri-
mathematical model. Lancet 2009; 373:48–57. mary infection on HIV transmission: systematic review and meta-analysis. AIDS
5. Eaton JW, Johnson LF, Salomon JA, et al. HIV treatment as prevention: system- 2014; 28:1021–9.
atic comparison of mathematical models of the potential impact of antiretroviral 29. Drake AL, Wagner A, Richardson B, John-Stewart G. Incident HIV during preg-
therapy on HIV incidence in South Africa. PLoS Med 2012; 9:e1001245. nancy and postpartum and risk of mother-to-child HIV transmission: a system-
6. Iwuji C, Orne-Gliemann J, Balestre E, et al. The impact of universal test and atic review and meta-analysis. PLoS Med 2014; 11:e1001608.
treat on HIV incidence in a rural South African population: ANRS 12249 TasP 30. Semba RD, Miotti PG, Chiphangwi JD, et al. Maternal vitamin A deficiency and
trial, 2012–2016. Paper presented at: Programme and Abstracts of the 21st mother-to-child transmission of HIV-1. Lancet 1994; 343:1593–7.
International AIDS Conference; 2016; Durban, South Africa. FRAC0105LB. 31. Wandel S, Egger M, Rangsin R, et al. Duration from seroconversion to eligibility
7. Guy R, Wand H, McManus H, et al; Australia HIV Observational Database for antiretroviral therapy and from ART eligibility to death in adult HIV-infected
(AHOD) and Treat Asia HIV Observation Database (TAHOD). Antiretroviral patients from low and middle-income countries: collaborative analysis of pro-
treatment interruption and loss to follow-up in two HIV cohorts in Australia and spective studies. Sex Transm Infect 2008; 84(Suppl 1):i31–6.

8 • OFID • Estill et al

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018
 32. United Nations, Department of Economic and Social Affairs, Population Division. 37. Haas AD, Tenthani L, Msukwa MT, et al. Retention in care during the first 3 years
World Population Prospects: The 2010 revision. Available at: http://www.un.org/ of antiretroviral therapy for women in Malawi’s option B+ programme: an obser-
en/development/desa/publications/world-population-prospects-the-2010-revi- vational cohort study. Lancet HIV 2016; 3:e175–82.
sion.html. Accessed 31 October 2017. 38. Nhavoto JA, Grönlund Å, Klein GO. Mobile health treatment support inter-
33. Braithwaite RS, Nucifora KA, Toohey C, et al. How do different eligibility guide- vention for HIV and tuberculosis in Mozambique: perspectives of patients and
lines for antiretroviral therapy affect the cost-effectiveness of routine viral load healthcare workers. PLoS One 2017; 12:e0176051.
testing in sub-Saharan Africa? AIDS 2014; 28(Suppl 1):S73–83. 39. Vogt F, Kalenga L, Lukela J, et al. Brief report: decentralizing ART supply for sta-
34. Cassell MM, Halperin DT, Shelton JD, Stanton D. Risk compensation: the ble HIV patients to community-based distribution centers: program outcomes
Achilles’ heel of innovations in HIV prevention? BMJ 2006; 332:605–7. from an urban context in Kinshasa, DRC. J Acquir Immune Defic Syndr 2017;
35. Bertagnolio S, Perno CF, Vella S, Pillay D. The impact of HIV drug resistance on 74:326–31.
the selection of first- and second-line ART in resource-limited settings. J Infect 40. McCormick NM, Li N, Sando D, et al. Implementation and operational research:
Dis 2013; 207(Suppl 2):S45–8. risk factors of loss to follow-up among HIV-positive pediatric patients in Dar es
36. Tweya H, Gugsa S, Hosseinipour M, et al. Understanding factors, outcomes and Salaam, Tanzania. J Acquir Immune Defic Syndr 2015; 70:e73–83.
reasons for loss to follow-up among women in Option B+ PMTCT programme in 41. ICAP at Columbia University. PHIA Project. Available at: http://phia.icap.colum-
Lilongwe, Malawi. Trop Med Int Health 2014; 19:1360–6. bia.edu/. Accessed 31 October 2017.

Monitoring, Tracing, and the HIV Epidemic • OFID • 9

Downloaded from https://academic.oup.com/ofid/article-abstract/5/5/ofy092/4987343

by guest
on 08 June 2018