Open access Original research

Cost-­effectiveness of implementing HIV

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
and HIV/syphilis dual testing among
key populations in Viet Nam: a
modelling analysis
David Coomes  ‍ ‍,1,2 Dylan Green,1,2 Ruanne Barnabas,2,3 Monisha Sharma,2
Magdalena Barr-­DiChiara,4 Muhammad S Jamil,4 R Baggaley,4
Morkor Newman Owiredu,4 Virginia Macdonald,4 Van Thi Thuy Nguyen  ‍ ‍,5
Son Hai Vo,6 Melanie Taylor,4,7 Teodora Wi,4 Cheryl Johnson,4,8 Alison L Drake  ‍ ‍1,2

To cite: Coomes D, ABSTRACT

Green D, Barnabas R, Objectives  Key populations, including sex workers, men
STRENGTHS AND LIMITATIONS OF THIS STUDY
et al. Cost-­effectiveness of who have sex with men, and people who inject drugs, ⇒ Our model parameters are informed by empiric data
implementing HIV and HIV/ including demographic, behavioural and biological
have a high risk of HIV and sexually transmitted infections.
syphilis dual testing among data from government sources, surveys, surveil-
We assessed the health and economic impacts of different
key populations in Viet Nam: a
HIV and syphilis testing strategies among three key lance, publicly available reports, databases and
modelling analysis. BMJ Open
2022;12:e056887. doi:10.1136/ populations in Viet Nam using a dual HIV/syphilis rapid peer-­reviewed literature.
bmjopen-2021-056887 diagnostic test (RDT). ⇒ We assess the impact of five testing scale up sce-
Setting  We used the spectrum AIDS impact model to narios using both HIV rapid diagnostic test (RDT) and
► Prepublication history and dual HIV/syphilis RDT and conduct sensitivity analy-
simulate the HIV epidemic in Viet Nam and evaluated five
additional supplemental material ses to evaluate uncertainty in model results.
for this paper are available
testing scenarios among key populations. We used a 15-­
year time horizon and a provider perspective for costs. ⇒ Due to limited data, we make assumptions regard-
online. To view these files,
please visit the journal online Participants  We simulate the entire population of Viet ing the timing and uptake of HIV and syphilis testing
(http://dx.doi.org/10.1136/​ Nam in the model. among key populations that may be inaccurate.
bmjopen-2021-056887). Interventions  We modelled five testing scenarios among ⇒ Our model conservatively assumes that increased
key populations: (1) annual testing with an HIV RDT, (2) syphilis testing and treatment will not impact syphi-
Received 04 October 2021 annual testing with a dual RDT, (3) biannual testing using lis prevalence, which is currently unknown.
Accepted 16 July 2022 dual RDT and HIV RDT, (4) biannual testing using HIV RDT
and (5) biannual testing using dual RDT.
Primary and secondary outcome measures  The populations, are at higher risk of acquiring
primary outcome is incremental cost-­effectiveness ratios. both HIV and syphilis. HIV incidence is
Secondary outcomes include HIV and syphilis cases. significantly higher among key populations
Results  Annual testing using a dual HIV/syphilis RDT compared with the general population in
was cost-­effective (US$10 per disability-­adjusted life year all geographical regions; however, differ-
(DALY)) and averted 3206 HIV cases and treated 27 727 ences vary substantially by region and by key
syphilis cases compared with baseline over 15 years.
population.1 While key populations and their
Biannual testing using one dual test and one HIV RDT
sexual partners represent approximately
(US$1166 per DALY), or two dual tests (US$5672 per DALY)
both averted an additional 875 HIV cases, although only 25% of new HIV cases in sub-­Saharan Africa,
the former scenario was cost-­effective. Annual or biannual they represent 80% of new HIV cases in the
HIV testing using HIV RDTs and separate syphilis tests rest of the world.2 Recent data suggests that
were more costly and less effective than using one or two syphilis incidence, while generally remaining
dual RDTs. stable in low-­ income and middle-­ income
© Author(s) (or their Conclusions  Annual HIV and syphilis testing using dual countries (LMICs), is increasing among key
employer(s)) 2022. Re-­use RDT among key populations is cost-­effective in Vietnam populations, particularly MSM.1 3 4 WHO HIV
permitted under CC BY-­NC. No and similar settings to reach global reduction goals for HIV
commercial re-­use. See rights testing guidelines recommend HIV retesting
and permissions. Published by
and syphilis. at least annually for key populations and
BMJ. more frequent testing (3–6 months) for those
For numbered affiliations see with high ongoing risk.5 WHO guidelines for
end of article. INTRODUCTION syphilis screening depend on population
Correspondence to
Key populations, including people who inject and setting. Laboratory-­based syphilis testing
Dr Alison L Drake ; drugs (PWID), men who have sex with men remains common, however rapid diagnostic
​adrake2@​uw.e​ du (MSM), sex workers (SW) and transgender tests (RDTs) for syphilis are increasingly

Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887 1

Open access

available and may be used to improve access to testing among key populations using the Goals model within

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
and treatment, including among key populations who Spectrum, as previously described.13 Briefly, Spectrum
are disproportionately affected by both HIV and syph- is a deterministic, compartmental mathematical model
ilis.6 With the introduction of prequalified dual HIV/ of HIV transmission stratified by sex and age. Transmis-
syphilis RDTs, and the recent WHO recommendation to sion is simulated through male-­female and male-­male sex
offer dual HIV/syphilis testing in antenatal care (ANC) acts, needle sharing for injection, and maternal-­to-­child
settings,7 it is important to evaluate how further integra- transmission with specific transmission probabilities for
tion and expansion of dual HIV/syphilis testing could each route. One can further specify parameters for low-
benefit key populations. and medium-­risk groups, as well as high-­risk categories
Since 2015, WHO has recommended immediate initia- including FSW, MSM, and PWID. Low-­risk heterosexuals
tion of antiretroviral therapy (ART) for all people living are those in stable couples while medium-­risk heterosex-
with HIV (PLWH)8 and the United Nations 95-­ 95-­
95 uals are those that engage in casual sex but are not in
targets aim to diagnose 95% of PLWH, provide 95% of a high-­risk group (high risk groups: MSM, FSW, PWID).
PLWH who know their status with ART, and ensure 95% Each of these high-­risk categories is nested within their
PLWH on ART are virally suppressed.9 Despite prog- parent categories and interact with one another. For
ress towards these goals—in 2019 81% of PLWH knew example, among MSM, a proportion is assumed to also
their HIV status and 67% were on ART—this progress is have female sexual partners. These high-­risk categories
uneven; only two-­thirds of key populations are aware of can be parameterised to have differential rates of partner-
their HIV status.2 While key populations lag behind the ship and uptake of interventions. The model was param-
general population in all phases of testing, linkage to eterised with demographic, behavioural and biological
treatment, and viral suppression, the largest gap exists in data from government sources, surveys, surveillance,
testing.10 WHO has also developed a global strategy on publicly available reports, databases and peer-­reviewed
sexually transmitted infections (STIs) which aims for a literature.
90% reduction in syphilis incidence by 2030, and 70% of To estimate syphilis burden, we used key population size
key populations to have access to STI and HIV services, estimates from the Goals model and population-­specific
including prevention, testing, and treatment.11 Increased estimates of prevalence12; we estimate the number of
syphilis testing and treatment may reduce syphilis burden persons in key populations testing positive and treated
among key and general populations, as well as HIV inci- for syphilis infection under each scenario. This model
dence since early symptomatic syphilis increases risks assumes that syphilis testing and treatment does not
of HIV acquisition and transmission.3 Currently, WHO impact syphilis prevalence, although increased screening
recommends syphilis testing for pregnant women and could potentially result in reduced, unchanged or
key populations, however, the optimal frequency of syph- increased syphilis prevalence depending on coverage.14 15
ilis testing is unknown and recommendations on syphilis For both HIV and syphilis, disability-­adjusted life years
testing for other populations are not available. (DALYs) are calculated for each scenario. Model key
In Viet Nam, the national HIV prevalence is <1% in parameters are shown in table 1.
the general population, and significantly higher in key
populations, with prevalence ranging between 3% and Settings and populations
13% among PWID, MSM and female SW (FSW). Simi- We modelled three key populations: MSM, PWID and
larly, syphilis prevalence among MSM (6.7%) and FSW FSW (and their clients) within the HIV epidemic in Viet
(2.1%) are also higher than that of the general popu- Nam, using national level HIV prevalence and syphilis
lation (0.3%).12 With budgetary constraints in HIV/ prevalence estimates for each key population (table 1).
STI programmes and the health sector, identifying cost-­
effective strategies for targeted HIV and syphilis testing Scenarios
among key groups in Viet Nam is crucial to inform poli- Our baseline scenario estimates annual HIV testing based
cymakers seeking to optimise resource allocation to maxi- on current WHO recommendations and estimated HIV
mise population health. We modelled the health impacts testing rates among key populations,5 and syphilis testing
and costs associated with varying frequencies of HIV and based on observed uptake. In the baseline scenario, we
syphilis testing for key populations, using test scenarios assume 50% of individuals in key populations test annu-
that include a dual HIV/syphilis RDT. ally for HIV, and syphilis screening with a non-­treponemal
test (rapid plasma reagin, RPR) occurs at rates specific to
each key population (table 2).16–18 Individuals who test
METHODS positive using RPR are given a confirmatory treponemal
Model (treponema pallidum haemagglutination, TPHA) test.
We used the AIDS Impact Model within the Spectrum We modelled HIV RDTs as WHO recommend, limiting
software package (V.5.76) to simulate the HIV epidemic the use of lab-­based testing such as western blot, espe-
in Viet Nam from 2020 to 2035. The model estimates cially in hard-­to-­
reach populations, to increase access
annual HIV incidence, AIDS mortality and disability. We and limit the loss to follow-­up.19 We considered alter-
simulated the impact of increasing HIV testing frequency native scenarios with varying testing frequency and test

2 Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887

 Open access

annual testing with dual RDT, (3) biannual testing (two

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
Table 1  Model parameters for spectrum input and cost-­
effectiveness analysis of HIV and syphilis testing scale up times per year), first with dual RDT and then with HIV
among key populations in Viet Nam RDT, (4) biannual HIV testing with RDT and baseline
Model parameter Value
syphilis RPR testing and (5) biannual testing with dual
RDT (table 2). We assumed 75% test acceptance for the
HIV Prevalence
first test in all scenarios except baseline, and 90% of those
 MSM (incl. TGW)* 10.8% who accepted the first test would accept the second test
 PWID** 12.7% in all scenarios that include biannual testing. All individ-
 FSW* 3.6% uals who test positive for syphilis using the dual test are
Syphilis prevalence12 then tested using RPR and TPHA per current Viet Nam
 MSM (incl. TGW) 6.7% country guidelines.
 PWID 0.3% We modelled increases in testing coverage by adjusting
 FSW 2.1%
the percent of people living with HIV (PLWH) on ART;
in scenarios with increased testing there is a higher prob-
Baseline syphilis test acceptance
ability that an individual in the model will initiate ART
 MSM (incl. TGW)16 49 50 27%
throughout the year. The baseline scenario assumes 95%
 PWID17 16% ART coverage among PLWH by 2028 (4.8% increase per
18 51
 FSW 35% year) based on recent ART scale-­up in Viet Nam; test
Syphilis DALYs averted52 coverage increases by 6.0% per year with annual HIV
 DALYs averted per syphilis case treated 0.04 testing (HIV RDT or dual RDT), and by 7.2% per year
ART with biannual testing. Maximum test coverage is 95%
 2019 coverage†† 70% for each model. All models assume ART coverage of
 Annual scale-­up† 4.8%
66% of men and 72% of women living with HIV in 2020
based on estimates from the Viet Nam HIV-­AIDS Tech-
 Transmission reduction efficacy‡ 70%
nical Working Group. Modelled HIV incidence per year
 Mortality reduction efficacy‡ 80%
is shown in online supplemental figure S1. We assume
Other prevention universal treatment among those who test positive for
 Condom use† 50% syphilis, individuals treated cannot become renfected
 Condom efficacy‡ 80% within the same year,20 and no changes to syphilis preva-
 PrEP coverage (MSM incl. TGW)† 5% lence under test case scenarios.
 PrEP efficacy‡ 90%
 PrEP adherence‡ 80% Costs
Costs†*
Testing cost inputs include cost per HIV RDT test, RPR,
TPHA, and dual HIV/syphilis RDT. We used local data
 HIV lay test§ $4.50
on the personnel, commodities and transport costs asso-
 Syphilis RPR§ $6.28
ciated with lay testing and estimate costs (table 1). ART
 Syphilis TPHA§ $10.26 costs include personnel, commodities, clinical follow-­up
 HIV/syphilis dual test§ $6.50 and laboratory monitoring. This analysis includes the
 ART¶ $285 costs of intervention delivery and treatment (Benzathine
 Syphilis treatment $6.50 penicillin G and ART) but does not consider additional
 Time horizon 2020–2035 averted sequelae costs such as the treatment of opportu-
 Discount rate 3% nistic infections due to uncontrolled HIV. All costs are
from the provider’s perspective and reported in 2019 US
*2018 Viet Nam HIV Sentinel Surveillance.
†Assumed.
dollars.
‡Spectrum model prior.
§Testing costs include labour, incentives, travel costs and test kits. Primary Cost-effectiveness
cost driver between tests is the cost of the test kit.
¶ART cost includes labour, laboratory monitoring costs, antiretroviral drugs Health impact was measured in DALYs averted, HIV
(ARVs) and other recurring costs. infections averted, syphilis infections treated and AIDS-­
**2019 Viet Nam HIV Sentinel Surveillance.
††Based on information from in-­country source. related deaths averted over the 15-­ year time horizon.
ART, antiretroviral therapy; DALY, disability-­adjusted life-­year; FSW, This time horizon was chosen because it reflects current
female sex workers; MSM, men who have sex with men; PrEP, pre-­exposure
prophylaxis; PWID, people who inject drugs; RPR, rapid plasma reagin; TGW, HIV programme planning in Viet Nam. HIV outcomes
transgender women; TPHA, treponema pallidum haemagglutination assay. are modelled for the entire population of Viet Nam while
syphilis outcomes are specific to key populations. Costs
and health benefits were discounted at 3% annually per
type (separate HIV and syphilis RPR, or a combined dual standard health economic evaluations.21 Incremental
syphilis/HIV RDT) among key populations from 2020 costs were calculated as costs incurred and averted by
to 2035. Scenarios modelled include: (1) annual HIV the testing strategy. We uused WHO guidelines for cost-­
testing with RDT and baseline syphilis RPR testing, (2) effectiveness threshold: less than gross domestic product

Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887 3

Open access

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
Table 2  HIV/syphilis testing scenarios among key populations in Viet Nam
Proportion of key population receiving HIV or syphilis testing per year
2 syphilis
Scenario 1 HIV test 2 HIV tests 1 syphilis test tests
Baseline 50% – 35% (FSW), 27% (MSM), 16% (PWID) –
1. One HIV RDT 75% – 35% (FSW), 27% (MSM), 16% (PWID) –
2. One dual HIV/syphilis RDT 75% 75% –
3. One HIV RDT and one dual HIV/syphilis 75% 68% 75% –
RDT
4. Two HIV RDTs 75% 68% 35% (FSW), 27% (MSM), 16% (PWID) –
5. Two dual HIV/syphilis RDT 75% 68% 75% 68%

The table cells show the proportion of key populations in Viet Nam that receive each test per year. If not specified, the proportion refers to all key
populations.
FSW, female sex workers; MSM, men who have sex with men; PWID, people who inject drugs; RDT, rapid diagnostic test.

per capita is considered cost-­

effective in Viet Nam the baseline scenario. The discounted cost of imple-
(US$2715 in 2019).22 menting this scenario over 30 years is US$62.9 million
compared with US$31.0 million for the baseline scenario.
Model calibration and sensitivity analyses The testing cost of implementing biannual testing using
Models were calibrated to national HIV prevalence data the dual RDT is approximately four times the cost of
for each key population. Monte Carlo sensitivity anal- baseline testing with an HIV RDT (US$67.6  million
yses were conducted to evaluate robustness of results to vs US$16.5  million, respectively), but an estimated
changes in: HIV and syphilis testing coverage, scenario US$8.0 million in HIV treatment costs would be averted
programme uptake rate, HIV and syphilis testing cost, by biannual HIV testing and US$11.8 million in syphilis
HIV and syphilis treatment cost, average years on ART testing costs would be averted by using the dual RDT. The
and time horizon. Online supplemental table S1 shows cost of biannual testing with an HIV RDT and continuing
the model parameters, ranges and distributions used in to test for syphilis with RPR is higher than biannual
the sensitivity analysis. testing with one dual RDT and one HIV RDT (US$53.4
Patient and public involvement vs US$51.9 million USD, respectively), but using one dual
It was not appropriate or possible to involve patients or test in biannual testing treats an estimated 28 000 more
the public in the design, conduct, reporting or dissemi- cases of syphilis over 15 years while averting the same
nation plans of our research. number of HIV cases.
Annual testing with the dual RDT is cost-­ effective
compared with the baseline scenario (US$10 per DALY
RESULTS averted) (figure 1). Annual HIV testing with HIV RDT
Increasing annual HIV test coverage from 50% (base- is more expensive and averts fewer DALYs than with the
line) to 75% using an HIV RDT (scenario 1) or a dual dual RDT (strongly dominated). The next most efficient
RDT (scenario 2) is projected to avert 3,206 HIV infec- scenario is biannual testing using one dual RDT and one
tions and 660 AIDS-­related deaths by 2035 in Viet Nam HIV RDT, which is cost-­effective (US$1166 USD per DALY
(table 3). Annual testing using dual RDT led to treat- averted). Biannual testing with HIV RDT is less effective
ment of an additional 27 727 syphilis cases over 15 years and more costly than biannual testing using one dual
compared with using HIV RDT, but the number of HIV RDT and one HIV RDT, while biannual testing using the
infections averted was the same. HIV testing with either dual RDT provides additional health benefits but is not
HIV or dual RDT biannually (scenarios 3, 4 and 5) was cost-­effective (US$5672 USD per DALY averted). Despite
projected to avert an additional 875 HIV infections and slightly higher initial costs, the discounted cost of annual
183 AIDS-­related deaths by 2035 compared with annual testing with a dual RDT becomes less than that of current
testing. Testing using a dual HIV/syphilis RDT biannually testing within 2 years, due to decreased ART costs associ-
among key populations is projected to lead to an addi- ated with HIV averted (online supplemental figure S2).
tional 60 674 syphilis cases treated by 2035, compared Sensitivity analyses including all scenarios found that
with annual testing using a dual RDT. an annual dual RDT (scenario 2) is cost-­saving in 52%
The most effective strategy was biannual testing among of the simulations and either cost-­saving or cost-­effective
MSM, PWID, and FSW with the dual RDT, which was (at US$2715 per DALY averted) in all simulations (online
projected to avert 4,081 HIV infections (7% of total infec- supplemental table S2). Biannual testing using one dual
tions), 76,632 HIV DALYs (0.04% of total HIV DALYs) RDT and one HIV RDT was cost-­ effective in 86% of
and treat 88 401 cases of syphilis by 2035 compared with simulations, but cost saving in only 1% of simulations

4 Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887

 Open access

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
Table 3  Estimated HIV and syphilis infections, and cost-­effectiveness of increased HIV and dual HIV/syphilis testing among
key populations in Viet Nam from 2020 to 2035
Scenario
1 HIV and 1
Baseline 1 dual test 1 HIV test dual 2 HIV tests 2 dual tests
HIV New HIV infections 57 902 54 696 54 696 53 821 53 821 53 821
AIDS deaths 13 877 13 217 13 217 13 034 13 034 13 034
Total HIV DALYs 174 567 240 174 508 007 174 508 007 174 490 608 174 490 608 174 490 608
Syphilis Total cases treated 88 953 116 680 88 953 116 680 88 953 177 354
Total DALYs treated 2831 3713 2831 3713 2831 5644
Incremental HIV infections – 3206 0 875 0 0
cases averted
averted
HIV DALYs averted – 59 233 0 17 399 0 0
Syphilis cases – 27 727 −27,727 27 727 −27,727 88 401
treated
Syphilis DALYs – 882 −882 882 −882 2813
averted
Total DALYs averted (HIV and – 60 115 −882 18 281 −882 2813
syphilis)
Costs Net costs $31 036 672 $31 659 182 $33 094 783 $51 942 954 $53 378 555 $62 896 039
(USD)
HIV testing $16 491 955 – $24 683 204 $22 142 027 $46 825 230 –
HIV treatment – -$6 133 138 -$6 133 138 -$7 991 393 -$7 991 393 -$7 991 393
averted
Syphilis testing $14 084 698 $1 535 409 $14 084 698 $1 535 409 $14 084 698 $2 333 826
Syphilis treatment $460 019 $603 395 $460 019 $603 395 $460 019 $917 162
Dual testing – $35 653 516 – $35 653 516 – $67 636 444
Total incremental costs – $622 510 $1 435 601 $18 848 171 $1 435 601 $9 517 484
ICERs (cost per DALY averted) – $10 Dom $1166 Dom $5672

Each scenario refers to the number of tests per year. The baseline scenario assumes that 50% of key populations are tested for HIV each year
and syphilis testing rates are specific to each subpopulation (FSW, MSM and PWID). Scenarios including one test per year assume a 75% test
acceptance rate, and those that include two tests per year assume a 75% test acceptance rate for the first test, and a 68.5% test acceptance rate
for the second test. Incremental cases averted, total DALYs averted and ICERs compare each scenario to the previous one.
DALY, disability-­adjusted life-­years; FSW, female sex workers; ICERs, incremental cost-­effectiveness ratios; MSM, men who have sex with men;
PWID, people who inject drugs.

(figure 2). Biannual testing using two dual RDTs was cost-­ tests would avert additional DALYs attributed to syphilis,
effective in 45% of simulations and cost saving in 31% although this latter scenario was not found to be cost-­
of simulations as compared with biannual testing with effective. Increasing the frequency of HIV testing to one
one dual RDT and one HIV RDT. In univariate sensi- or two tests per year using only HIV RDTs (scenarios 1
tivity analysis adjusting costs, our scenarios that involve and 4), while continuing to screen for syphilis using RPR,
one dual RDT (scenarios 2 and 3) remain cost-­effective was not efficient compared with other strategies.
even after all costs (testing and treatment) are increased Implementing biannual testing substantially increases
by 50% (US$16 and US$1705 USD per DALY averted, testing costs, but also prevents more HIV infections, there-
respectively). fore averting more HIV healthcare costs, including ART
and hospitalisations. Increasing test frequency may be
DISCUSSION cost saving or cost-­effective, although it incurs consider-
In this modelling analysis, we found that implementing able costs in the near term while costs averted may not be
annual testing among key populations with the dual observed for many years. Annual testing using a dual RDT
RDT at 75% coverage was cost-­effective, averted more can help offset some near-­term costs as it is less expen-
HIV infections and treated more syphilis cases compared sive than using HIV RDT and syphilis RPR. Policy-­makers
with annual testing using HIV RDT at 50% coverage must weigh the health impact and cost-­ effectiveness
and current syphilis testing in Viet Nam. While bian- of different testing scenarios over time against current
nual testing with one dual RDT and one HIV RDT was affordability; however, using the dual RDT will help
projected to be more costly, it would avert more HIV integrate syphilis testing within existing HIV testing
and syphilis related DALYs, and using dual RDT for both programmes, improving programme efficiencies.23

Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887 5

Open access

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
Figure 1  Efficiency frontier presenting the total disability-­adjusted life-­years (DALYs) and costs for five testing scenarios
among key populations. The solid line indicates the scenarios that are not dominated by other scenarios. Dominated indicates
that a scenario is either more costly and less effective or has a higher ICER than a scenario that is more effective. The ICERs for
the non-­dominated scenarios are shown. ICER, incremental cost-­effectiveness ratio; RDT, rapid diagnostic test; .

Implementation of dual RDT is occurring in some both fund dual RDT in ANC,25 and there are multiple
settings; preliminary reports indicate that 49 countries dual RDTs prequalified by the WHO.26 The use of dual
have adopted policies to use dual HIV/syphilis RDT RDT during ANC could be a model for improving HIV/
in ANC, and 15% of reporting counties have policies STI integration, particularly among those at high risk for
to support their use in key populations, although the both HIV and syphilis, such as key populations, however,
extent of implementation is unknown.24 The President’s there are operational challenges associated with inte-
Emergency Plan for AIDS Relief and the Global Fund grating HIV and STI programmes and delivering

Figure 2  Sensitivity analysis of non-­dominated scenarios using a Monte-­Carlo simulation of the cost-­effectiveness of HIV/
syphilis dual testing among key populations in Viet Nam. Plot shows 10 000 iterations in which 17 key parameters were
randomly adjusted. All points below the green line are cost-­effective at US$2715 per DALY averted and those below the solid
black line (y-­intercept) are cost saving. Only non-­dominated scenarios are shown in this figure; cost-­effectiveness of 1 dual test
is compared with baseline, 1 HIV test and 1 dual test is compared with 1 dual test, and 2 dual tests is compared with 1 HIV test
and 1 dual test. DALYs, disability-­adjusted life-­years; ICER, incremental cost-­effectiveness ratio; RDT, rapid diagnostic test.

6 Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887

 Open access

person-­ centred diagnosis, treatment and prevention not only through additional commodity procurement

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
services.27 but also for health systems, programme coordination,
Benefits of the dual test are its potential to cost-­ and outreach. Policy-­ makers may likely benefit from
effectively reach more at-­risk individuals at the point of targeting limited testing resources towards high-­ risk
care. Annual or biannual testing can enable earlier iden- groups such as key populations.
tification of HIV-­positive individuals for faster ART initia- Dual RDTs may also increase syphilis testing frequency
tion and prevention of onward transmission. Annual HIV and coverage among key populations who are more likely
testing for key populations is recommended by WHO, and to access HIV testing than testing for syphilis. Previous
more frequent testing (every 3–6 months) may be advised research has shown that coupling rapid syphilis testing
for those with individual risk factors, including those in ANC may also increase HIV test coverage in LMICs,
using pre-­exposure prophylaxis (PrEP) and key popula- particularly in settings where HIV test coverage is low.38
tions presenting with STIs.19 Individuals presenting with This strategy may be similarly effective at increasing test
syphilis symptoms should also test for HIV, and using the coverage for both diseases among key populations, as well
dual RDT is less costly as compared with a syphilis RPR as augment current ANC testing by reaching women in
and HIV RDT. As policy-­makers scale up PrEP among key key populations who present late or are missed by ANC
populations in Viet Nam, including at least one dual RDT services. While there is a lack of data on dual RDTs among
in the testing algorithm may be more cost-­effective than key populations, models of dual RDT during ANC have
using HIV RDTs alone. In addition, using dual RDT tests been shown to be cost saving or cost-­effective among both
can facilitate lay providers to offer both HIV and syphilis key populations and the general population of pregnant
testing for their community.28 women.7 39 While dual RDTs are likely more effective in
Our results were robust to sensitivity analyses, suggesting the context of ANC since testing can avert more adverse
that testing annually or biannually using dual RDTs outcomes associated with congenital syphilis and mother-­
remains cost-­effective if testing costs increase and HIV to-­child HIV transmission, we find dual RDTs may also be
prevalence decreases. In scenarios involving dual RDT, cost-­effective among non-­pregnant key populations.
the majority (>98%) of benefits, as measured in DALYs, Our results are consistent with previously published
come from averting HIV infections rather than treating models that show expanded testing and early access to
syphilis due to the relatively large burden of disease from ART for key populations in Viet Nam will cost-­effectively
HIV compared with syphilis. However, since the cost of a reduce the country’s HIV burden.40 41 Additionally,
dual RDT is only slightly higher than the cost of an HIV models from both low-­ resource and high-­ resource
RDT, it is cheaper to use a dual RDT than separate HIV countries suggest HIV testing every 3–6 months among
RDT and syphilis RPR tests in situations where both tests key populations can be cost-­ effective in concentrated
are recommended. epidemics.42 43 However, HIV risk within key popula-
Increased HIV testing can reduce HIV-­ associated tions is not homogenous; further targeting of higher-­risk
morbidity and mortality and transmission from PLWH groups within key populations may be needed to achieve
through early detection and initiation of ART. While efficient testing regimens. While we examine the impact
models suggest high ART coverage would result in of increased testing frequency among key populations
substantial declines in HIV incidence,29 30 empiric data as a whole, previous research has described the benefits
from countries with population-­ level viral suppres- of targeting high-­risk groups within key populations.44
sion exceeding 73% (eg, Australia, eSwatini and Thai- Individuals who engage in risky behaviours, such as those
land) have observed less significant reductions in with more sexual partners, practicing unprotected sex
HIV incidence relative to predictions from mathemat- or needle/syringe sharing may benefit from additional
ical models.31 Similarly, when high ART coverage was testing or linkage to HIV prevention such as PrEP and
achieved in a series of cluster-­randomised trials in sub-­ harm reduction interventions. Further research is needed
Saharan Africa, it resulted in decreased population-­level on the optimal testing intervals for higher-­risk groups of
HIV incidence; however, this decrease was insufficient key populations.
to end HIV as a public health threat.32–35 These discrep- Globally, approximately one-­third of key populations
ancies may in part be attributed to delayed diagnosis are not aware of their HIV status. Programmes focusing
and ART initiation following infection,36 37 and gaps in on HIV testing and treatment among FSW and PWID
the 95-­95-­95 targets for some population groups, for in Viet Nam have shown success in reducing HIV preva-
example young men and key populations. Additional lence in these groups; however, less than a third of MSM
barriers may include poor coverage of evidence-­based reported testing for HIV in 2015, likely contributing to
prevention interventions and persistent structural increases in HIV prevalence among this group in the past
barriers, particularly for key populations. More frequent decade.45 Annual syphilis testing among key populations
HIV testing strategies could increase earlier diagnosis in Viet Nam is similarly low, ranging from 16% among
and initiation on ART and focusing testing and linkage PWID to 36% among FSW.16–18 Due to high dual preva-
efforts on key populations could reduce the access and lence of HIV and syphilis among key populations, dual
coverage disparities in these groups. However, more testing is a promising strategy to increase testing coverage
frequent testing will also increase programme costs, and linkage to care.

Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887 7

Open access

Our analysis has several limitations. We did not include Nam have recently been achieved through social media

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
the cost of scaling-­up and training providers in admin- campaigns, perhaps providing a guide for cost-­effectively
istering dual RDTs. However, RDT are easy to use and increasing testing uptake among key populations.45 48
can be administered by a lay provider, and rapid results We also did not consider the burden that increased test
can minimise loss to follow-­up. Overall, dual RDTs have coverage and frequency may have on the health system;
been shown to have adequate performance in field however, as testing may be conducted effectively using lay
settings in Viet Nam among key populations.46 Dual providers, increased testing may not substantially impact
RDTs may also increase HIV test coverage as they can be the provision of other services.47 Although targeting key
easily conducted by community health workers outside populations in lower prevalence regions may be more
of healthcare settings, and they may be more acceptable difficult and costly, these results are robust to increased
to some members of key populations who are concerned costs and it will likely remain an effective use of resources.
about stigma associated with testing.47 Dual RDTs may Research that focuses on province-­specific estimates of
also expand syphilis testing uptake, as most syphilis cases cost and impact would likely find that focusing on high-­
in Viet Nam are currently diagnosed at provincial hospi- burden areas is more cost-­ effective; however, health
tals. Despite this, some additional training, supervision policy, financing, guideline development and implemen-
and support will be needed to scale-­up dual RDT use tation continue to be led nationally in Viet Nam. There-
among key populations. fore, national-­level evidence is needed to direct decision
We did not explicitly model HIV testing or diagnosis making.
in this analysis as HIV testing uptake is not an adjustable We assume that syphilis screening will not impact syph-
model parameter in Spectrum. We instead modelled ART ilis prevalence rates. Increased screening may reduce
coverage, which required assumptions about the link prevalence by increasing early treatment, but syphilis
between testing frequency and ART coverage. Since data screening also has the potential to increase prevalence
on the impact of retesting on population HIV incidence as individuals with latent syphilis are unlikely to transmit
is limited, we made conservative assumptions about the the infection to others unless they are treated and then
frequency of linkage to care and ART use following infected again. Thus, we believe our estimates of infec-
retesting. We assumed that HIV testing frequency would tions averted and cost-­ effectiveness are conservative.
increase in Viet Nam among key populations in the base- Finally, there is limited data on population size, HIV
line scenario but testing frequency would increase more and syphilis prevalence, and health seeking behaviours
quickly under the other scenarios. Because of this, we among key populations. We based our model input on
believe our estimates of the impact of increased testing estimates included in published literature as well as Viet
frequency are conservative. Due to the lack of evidence Nam country sources.
on the impact of retesting on population HIV incidence,
a model that explicitly includes testing rates as a param-
eter would also need to rely on assumptions concerning
the relationship between testing behaviour and ART CONCLUSIONS
enrollment. Our study suggests that annual or biannual HIV and syph-
Some model assumptions regarding the timing of HIV ilis testing among key populations in Viet Nam using a
and syphilis testing may be inaccurate. Timing of testing is dual RDT will increase HIV and syphilis detection and
an important component from both a technical analytical treatment, while remaining cost saving or cost-­effective.
perspective and guideline development process. In truth, Integrating HIV and other STI testing can streamline
there are a nearly limitless number of permutations of services as well as expand testing and help countries with
frequency and spacing of retests. We chose even spacing epidemics concentrated in key populations reach 95-­95-­95
as it is easily interpretable at all levels of research, policy, targets. Future collection of empirical data, including
and service delivery. This maximal spacing between tests conducting budget impact studies, would be useful to
is expected to have the largest impact at the population determine the impact of HIV and syphilis screening
level, assuming risk is evenly spread across the calendar among key populations on ART uptake as well as HIV
year. We assume regular testing intervals for the entire and syphilis incidence, particularly in concentrated HIV
population in each scenario, but it is possible—and epidemics.
entirely sensible—for people who had a risky sexual
Author affiliations
encounter or who are experiencing symptoms to seek 1
Department of Epidemiology, University of Washington, Seattle, Washington, USA
more frequent retesting than biannually. We assumed in 2
Department of Global Health, University of Washington, Seattle, Washington, USA
scenarios that included a dual RDT, additional syphilis 3
Department of Medicine, University of Washington, Seattle, Washington, USA
screening tests would not be conducted. However, PLWH 4
Department of Global HIV, Hepatitis and STI Programmes, World Health
who know their status and present for syphilis screening Organization, Geneva, Switzerland
5
Viet Nam Country Office, World Health Organization, Hanoi, Viet Nam
do not need an HIV test. 6
Viet Nam Authority for HIV/AIDS Prevention and Control, Government of Viet Nam
We did not include the costs of outreach to achieve Ministry of Health, Hanoi, Viet Nam
increased test coverage of key populations. Consider- 7
Division of STD Prevention, Centers for Disease Control and Prevention, Atlanta,
able expansions of first time testing among MSM in Viet Georgia, USA

8 Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887

 Open access
8
Clinical Research Department, London School of Hygiene & Tropical Medicine, 9 UNAIDS. 90-­90-­90 an ambitious treatment target to help end

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
London, UK the AIDS epidemic, 2016. Available: https://www.unaids.org/en/​
resources/909090
10 Hakim AJ, MacDonald V, Hladik W, et al. Gaps and opportunities:
Contributors  CJ and ALD devised the project and the main conceptual ideas. measuring the key population cascade through surveys and
DC, DG and RB parameterised the model. DC and DG carried out the model services to guide the HIV response. J Int AIDS Soc 2018;21 Suppl
implementation. VM and SHV provided model feedback. All authors, including DC, 5:e25119.
DG, RB, MS, MB-­D, MSJ, RB, MNO, VM, VTTN, SHV, MT, TW, CJ and ALD, provided 11 World Health Organization. Global health sector strategy on sexually
critical feedback and helped shape the research, analysis and manuscript. transmitted infections 2016-­2021, 2016. Available: https://www.who.​
int/reproductivehealth/publications/rtis/ghss-stis/en/
Funding  This study was funded by WHO-­USAID: GHA-­G-­00-­09-­00003; NIH/NIAID: 12 World Health Organization. Global health Observatory data
K01 AI116298. Repository: data on syphilis. Available: https://apps.who.int/gho/​
data/node.main.A1357STI?lang=en [Accessed 4 Jun 2020].
Disclaimer  The views expressed in this manuscript are those of the authors and
13 Stover J, Brown T, Puckett R, et al. Updates to the Spectrum/
do not necessarily represent the official position of the WHO or the US Centers for Estimations and projections package model for estimating trends
Disease Control and Prevention. and current values for key HIV indicators. AIDS 2017;31 Suppl
Competing interests  None declared. 1:S5–11.
14 Tuite AR, Testa C, Rönn M, et al. Exploring how epidemic context
Patient and public involvement  Patients and/or the public were not involved in influences syphilis screening impact: a mathematical modeling study.
the design, or conduct, or reporting, or dissemination plans of this research. Sex Transm Dis 2020;47:798-­810.
15 Tuite A, Fisman D. Go big or go home: impact of screening coverage
Patient consent for publication  Not applicable. on syphilis infection dynamics. Sex Transm Infect 2016;92:49-­54.
Ethics approval  Not applicable. 16 Justumus P, Colby D, Mai Doan Anh T, et al. Willingness to use the
Internet to seek information on HIV prevention and care among men
Provenance and peer review  Not commissioned; externally peer reviewed. who have sex with men in Ho Chi Minh City, Vietnam. PLoS One
2013;8:e71471.
Data availability statement  Data are available on reasonable request. Extra data
17 Nguyen TA, Hoang LT, Pham VQ, et al. Risk factors for HIV-­1
are available by emailing, DC, ​dcoomes@​uw.​edu. seropositivity in drug users under 30 years old in Haiphong, Vietnam.
Supplemental material  This content has been supplied by the author(s). It has Addiction 2001;96:405–13.
not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been 18 Ngo AD, Ratliff EA, McCurdy SA, et al. Health-­seeking behaviour for
peer-­reviewed. Any opinions or recommendations discussed are solely those sexually transmitted infections and HIV testing among female sex
workers in Vietnam. AIDS Care 2007;19:878–87.
of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and 19 World Health Organization. Consolidated guidelines on HIV testing
responsibility arising from any reliance placed on the content. Where the content services. Geneva, 2019. Available: https://www.who.int/publications/​
includes any translated material, BMJ does not warrant the accuracy and reliability i/item/978-92-4-155058-1
of the translations (including but not limited to local regulations, clinical guidelines, 20 Feldman J, Mishra S. What could re-­infection tell us about R0? A
terminology, drug names and drug dosages), and is not responsible for any error modeling case-­study of syphilis transmission. Infect Dis Model
and/or omissions arising from translation and adaptation or otherwise. 2019;4:257–64.
21 Edejer TT-­T, Baltussen R, Adam T. Making choices in health: who
Open access  This is an open access article distributed in accordance with the guide to cost effectiveness analysis. World Health Organization,
Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which 2003.
permits others to distribute, remix, adapt, build upon this work non-­commercially, 22 World Bank. World Bank Data: GDP per capita (current US$) -
and license their derivative works on different terms, provided the original work is Vietnam. Available: https://data.worldbank.org/indicator/NY.GDP.​
properly cited, appropriate credit is given, any changes made indicated, and the use PCAP.CD?locations=VN [Accessed 16 Dec 2020].
is non-­commercial. See: http://creativecommons.org/licenses/by-nc/4.0/. 23 Ong JJ, Fu H, Smith MK, et al. Expanding syphilis testing: a scoping
review of syphilis testing interventions among key populations.
ORCID iDs Expert Rev Anti Infect Ther 2018;16:423–32.
24 UNAIDS Joint United Nations Programme on HIV/AIDS, WHO. Laws
David Coomes http://orcid.org/0000-0003-1815-1969
and policies analytics.
Van Thi Thuy Nguyen http://orcid.org/0000-0001-8766-3082 25 U.S. President’s Emergency Plan for AIDS Relief (PEPFAR). PEPFAR
Alison L Drake http://orcid.org/0000-0002-4178-7706 2021 country and regional operational plan (COP/ROP) guidance for
all PEPFAR countries, 2021. Available: https://www.state.gov/wp-​
content/uploads/2020/12/PEPFAR-COP21-Guidance-Final.pdf
26 The Global Fund. List of HIV diagnostic test kits and equipments
REFERENCES classified according to the global fund quality assurance policy,
1 UNAIDS Joint United Nations Programme on HIV/AIDS. Global AIDS 2021. Available: https://www.theglobalfund.org/media/5878/psm_​
update 2019 - Communities at the centre. Geneva, 2019. Available: productshiv-who_list_en.pdf
https://www.unaids.org/en/resources/documents/2019/2019-global-​ 27 Broyles LN, Boeras D, Peeling RW. Implementation of dual maternal
AIDS-update HIV-­Syphilis testing: the devil is in the details. Lancet Glob Health
2 UNAIDS Joint United Nations Programme on HIV/AIDS. Global AIDS 2021;9:e595.
update 2020: seizing the moment. Geneva, 2020. Available: https:// 28 Nguyen V, Anh L, Thong N. High prevalence of HIV, syphilis and HCV
www.unaids.org/en/resources/documents/2020/global-aids-report among key populations and partners: results from an integrated
3 Cameron CE. Syphilis vaccine development: requirements, multiple diseases testing led by community in Viet Nam. In: Poster
challenges, and opportunities. Sex Transm Dis 2018;45:S17. exhibition IAS 2019. Mexico City, Mexico, 2019.
4 Tsuboi M, Evans J, Davies EP, et al. Prevalence of syphilis among 29 Granich RM, Gilks CF, Dye C, et al. Universal voluntary HIV
men who have sex with men: a global systematic review and meta-­ testing with immediate antiretroviral therapy as a strategy for
analysis from 2000-­20. Lancet Glob Health 2021;9:e1110–8. elimination of HIV transmission: a mathematical model. Lancet
5 World Health Organization. Consolidated guidelines on HIV testing 2009;373:48–57.
services. Geneva, 2015. Available: https://www.ncbi.nlm.nih.gov/​ 30 Eaton JW, Johnson LF, Salomon JA, et al. Hiv treatment as
books/NBK316021/ prevention: systematic comparison of mathematical models of the
6 World Health Organization. WHO guideline on syphilis screening and potential impact of antiretroviral therapy on HIV incidence in South
treatment for pregnant women, 2017. Available: https://www.who.int/​ Africa. PLoS Med;9:e1001245.
reproductivehealth/publications/rtis/syphilis-ANC-screenandtreat-​ 31 UNAIDS. AIDSinfo | UNAIDS. Available: https://aidsinfo.unaids.org/
guidelines/en/ [Accessed 27 May 2020].
7 Rodriguez PJ, Roberts DA, Meisner J, et al. Cost-­Effectiveness of 32 Iwuji CC, Orne-­Gliemann J, Larmarange J, et al. Universal test
dual maternal HIV and syphilis testing strategies in high and low and treat and the HIV epidemic in rural South Africa: a phase
HIV prevalence countries: a modelling study. Lancet Glob Health 4, open-­label, community cluster randomised trial. Lancet HIV
2021;9:e61–71. 2018;5:e116–25.
8 World Health Organization. Consolidated guidelines on the use 33 Makhema J, Wirth KE, Pretorius Holme M, et al. Universal testing,
of antiretroviral drugs for treating and preventing HIV infection expanded treatment, and incidence of HIV infection in Botswana. N
2016;2016. Engl J Med 2019;381:230–42.

Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887 9

 Open access

34 Havlir DV, Balzer LB, Charlebois ED, et al. Hiv testing and treatment 44 Reitsema M, Steffers L, Visser M, et al. Cost-­Effectiveness of increased

BMJ Open: first published as 10.1136/bmjopen-2021-056887 on 11 August 2022. Downloaded from http://bmjopen.bmj.com/ on June 19, 2023 by guest. Protected by copyright.
with the use of a community health approach in rural Africa. N Engl J HIV testing among MSM in the Netherlands. AIDS 2019;33:1807–17.
Med 2019;381:219-­229. 45 Green KE, Vu BN, Phan HT, et al. From conventional to disruptive:
35 Hayes RJ, Donnell D, Floyd S, et al. Effect of Universal Testing and upturning the HIV testing status quo among men who have sex with
Treatment on HIV Incidence - HPTN 071 (PopART). N Engl J Med men in Vietnam. J Int AIDS Soc 2018;21 Suppl 5:e25127.
2019;381:207–18. 46 Withers K, Bristow C, Nguyen M, et al. A field evaluation of a rapid
36 Akullian A, Bershteyn A, Jewell B, et al. The missing 27. AIDS dual immunoassay for human immunodeficiency virus and syphilis
2017;31:2427–9. antibodies, Hanoi, Vietnam. Int J STD AIDS 2019;30:173–80.
37 Abdool Karim SS. HIV-­1 Epidemic Control - Insights from Test-­and-­ 47 Vu BN, Green KE, Thi Thu Phan H, et al. Lay provider HIV testing:
Treat Trials. N Engl J Med 2019;381:286–8. a promising strategy to reach the undiagnosed key populations in
38 Swartzendruber A, Steiner RJ, Adler MR, et al. Introduction of rapid Vietnam. PLoS One 2018;13:e0210063.
syphilis testing in antenatal care: a systematic review of the impact 48 Nguyen VTT, Phan HT, Kato M, et al. Community-­led HIV testing
on HIV and syphilis testing uptake and coverage. Int J Gynaecol services including HIV self-­testing and assisted partner notification
Obstet 2015;130 Suppl 1:S15–21. services in Vietnam: lessons from a pilot study in a concentrated
39 Gliddon HD, Peeling RW, Kamb ML, et al. A systematic review and epidemic setting. J Int AIDS Soc 2019;22 Suppl 3:e25301.
meta-­analysis of studies evaluating the performance and operational 49 Clatts MC, Goldsamt LA, Giang LM, et al. Sexually transmissible
characteristics of dual point-­of-­care tests for HIV and syphilis. Sex infection and HIV prevention and treatment for young male sex
Transm Infect 2017;93:S3–15. workers in Vietnam: findings from the sheath intervention. Sex Health
40 Kato M, Long NH, Duong BD, et al. Enhancing the benefits of 2016;13:575-­581.
antiretroviral therapy in Vietnam: towards ending AIDS. Curr HIV/ 50 Bao A, Colby DJ, Trang T, et al. Correlates of HIV testing among
AIDS Rep 2014;11:487–95. transgender women in Ho Chi Minh, Vietnam. AIDS Behav
41 Kato M, Granich R, Bui DD, et al. The potential impact of expanding 2016;20:371-­378.
antiretroviral therapy and combination prevention in Vietnam: 51 Magnani R, Riono P, Saputro E, et al. Sexual risk behaviours, HIV and
towards elimination of HIV transmission. J Acquir Immune Defic other sexually transmitted infections among female sex workers in
Syndr 2013;63:e142-­9. Indonesia. Sex Transm Infect 2010;86:393–9.
42 Kazemian P, Costantini S, Kumarasamy N, et al. The cost-­ 52 Cassini A, Colzani E, Pini A, et al. Impact of infectious diseases on
effectiveness of human immunodeficiency virus (HIV) preexposure population health using incidence-­based disability-­adjusted life
prophylaxis and HIV testing strategies in high-­risk groups in India. years (DALYs): results from the burden of communicable diseases
Clin Infect Dis 2020;70:633–42. in Europe study, European Union and European economic area
43 Cipriano LE, Zaric GS, Holodniy M, et al. Cost effectiveness of countries, 2009 to 2013. Euro Surveill 2018;23.
screening strategies for early identification of HIV and HCV infection
in injection drug users. PLoS One 2012;7:e45176.

10 Coomes D, et al. BMJ Open 2022;12:e056887. doi:10.1136/bmjopen-2021-056887

 BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance
Supplemental material placed on this supplemental material which has been supplied by the author(s) BMJ Open

Cost-effectiveness of implementing HIV and HIV/syphilis dual testing among key

populations in Viet Nam: a modeling analysis

Supplemental material

Table S1: Parameters and probability distributions for Monte Carlo simulation
Table S2: Sensitivity analysis of all scenarios using a Monte Carlo simulation
Figure S1: Estimated yearly HIV incidence under baseline, annual, and biannual HIV testing
Figure S2: Cost pressure analysis of testing scenarios

Coomes D, et al. BMJ Open 2022; 12:e056887. doi: 10.1136/bmjopen-2021-056887

 BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance
Supplemental material placed on this supplemental material which has been supplied by the author(s) BMJ Open

Table S1. Parameters and probability distributions for Monte Carlo simulation. Table shows the
baseline model parameter values and the probability distributions used for random draws of 17
variables for 10,000 Monte Carlo simulations. Beta distributions were used for all proportion
parameters. For the beta distribution, the alpha and beta parameters were calculated as the baseline
value multiplied by 100, except for the impact parameter where an alpha and beta of 25 was used.
Gamma distributions were used for all other parameters. For the gamma distribution, the alpha
parameter was calculated as the square of the baseline parameter divided by the square of the standard
deviation. The beta parameter was calculated as the square of the standard deviation divided by the
baseline parameter.

Baseline
Model Parameter value Distribution St. Dev alpha/beta

Baseline HIV test acceptance 50% Beta N/A 50

Recommended test acceptance 75% Beta N/A 75

Additional test drop off rate 10% Beta N/A 10

HIV Lay Test Cost $4.50 Gamma $3.00 N/A

Syphilis RPR1 Cost $6.28 Gamma $3.50 N/A

Syphilis TPHA Cost $10.26 Gamma $5.00 N/A

HIV/Syphilis Dual Test Cost $6.50 Gamma $3.50 N/A

ART Treatment Cost $285 Gamma $50 N/A

Syphilis Treatment Cost $6.50 Gamma $3.50 N/A

Avg Years on ART 25 Gamma 3 N/A

Time Horizon 2035 Gamma 3 N/A

Impact 100% Beta N/A 25

Baseline Syphilis test acceptance FSW 35% Beta N/A 35

Baseline syphilis test acceptance MSM 27% Beta N/A 27

Baseline syphilis test acceptance PWID 16% Beta N/A 16

Baseline syphilis prevalence FSW 2.1% Beta N/A 2.1

Baseline syphilis prevalence MSM 6.7% Beta N/A 6.7

Baseline syphilis prevalence PWID 0.3% Beta N/A 0.3

Coomes D, et al. BMJ Open 2022; 12:e056887. doi: 10.1136/bmjopen-2021-056887

 BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance
Supplemental material placed on this supplemental material which has been supplied by the author(s) BMJ Open

Table S2. Sensitivity analysis of all scenarios using a Monte Carlo simulation. Table shows the percentage
of simulations (10,000 iterations) in which each scenario is cost-effective (at $500 or $2,715 per DALY
averted), cost-saving, or less-effective. Less effective scenarios are both less effective and more costly as
compared to the scenario above. Scenarios are arranged in order of increasing cost and each scenario is
compared to the one immediately above; 1 Dual HIV/syphilis RDT is compared to the baseline scenario.
Cost-effective Cost-effective
Less
Scenario ($500) ($2,715) Cost-saving effective

1 Dual HIV/syphilis RDT 85% 100% 52% 0%

1 HIV RDT 1% 1% 1% 98%

1 Dual HIV/syphilis RDT & 1 HIV

RDT 32% 84% 18% 0%

2 HIV RDTs 1% 1% 1% 98%

2 Dual HIV/syphilis RDTs 45% 52% 44% 0%

DALY=disability adjusted life-year, RDT=rapid diagnostic test

Coomes D, et al. BMJ Open 2022; 12:e056887. doi: 10.1136/bmjopen-2021-056887

 BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance
Supplemental material placed on this supplemental material which has been supplied by the author(s) BMJ Open

Figure S1. Estimated yearly HIV incidence under baseline, annual, and biannual HIV testing.
Figure shows modeled incidence under each scenario for the entire adult population of Viet
Nam. The baseline scenario assumes 95% ART coverage among PLWH by 2028 (4.8% increase
per year). Annual HIV testing models a 6.0% ART coverage increase per year, and biannual
testing models a 7.2% ART coverage increase per year. Maximum test coverage is 95% for each
model. All models assume ART coverage of 66% of men and 72% of women living with HIV in
2020. Scenarios are implemented in 2020 and modeled through 2035.

Coomes D, et al. BMJ Open 2022; 12:e056887. doi: 10.1136/bmjopen-2021-056887

 BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance
Supplemental material placed on this supplemental material which has been supplied by the author(s) BMJ Open

Figure S2. Cost pressure analysis of testing scenarios. Figure shows the discounted cost over
time of each scenario. Costs are discounted 3% with a time horizon from 2020 – 2035. Baseline
costs include testing costs assuming that 50% of key populations are tested for HIV each year and
syphilis testing rates are specific to each sub-population (FSW, MSM, and PWID), and syphilis
treatment costs. All other scenarios include the cost of HIV treatment averted compared to the
baseline scenario, testing costs, and syphilis treatment costs. Scenarios including one test per
year assume a 75% test acceptance rate, and those that include two tests per year assume a 75%
test acceptance rate for the first test, and a 68.5% test acceptance rate for the second test. Each
scenario refers to the number of tests per year. RDT=rapid diagnostic test, USD=United States
dollars

Coomes D, et al. BMJ Open 2022; 12:e056887. doi: 10.1136/bmjopen-2021-056887