Articles
The cascade of HIV care among key populations in Indonesia:
Lancet HIV 2018; 5: e560–68
Published Online
August 21, 2018
http://dx.doi.org/10.1016/
S2352-3018(18)30148-6
See Comment page e539
Center for Public Health
Innovation, Faculty of
Medicine, Udayana University,
Bali, Indonesia
(P P Januraga DrPH); Kirby
Institute, UNSW, Sydney, NSW,
Australia (J Reekie PhD,
Prof M Law PhD,
Prof D A Cooper DSc,
Prof J M Kaldor PhD); Faculty of
Medicine, Padjajaran
University, West Java,
Indonesia (T Mulyani SKom,
B W Lestari MSc, S Iskandar PhD,
R Wisaksana PhD); Center for
Tropical Medicine, Faculty of
Medicine, Universitas Gadjah
Mada, Yogyakarta, Indonesia
(N A Kusmayanti MPH,
Y W Subronto PhD); Faculty of
Medicine, Udayana University,
Bali, Indonesia
(D N Widyanthini MPH);Yayasan
Kerti Praja, Denpasar, Bali,
Indonesia
(Prof D N Wirawan MPH); AIDS
Research Center, Atma Jaya
Catholic University of
Indonesia, Jakarta, Indonesia
(L V Wongso SPsi,
A G Sudewo MPsi,
E Sukmaningrum PhD); and
WHO Country Office of
Indonesia, Jakarta, Indonesia
(T Nisa MS, B R Prabowo MD)
*Prof D A Cooper died on
March 18, 2018
Correspondence to:
Dr Pande Putu Januraga, Center
for Public Health Innovation,
Faculty of Medicine, Udayana
University, Bali 80232, Indonesia
januraga@unud.ac.id
e560 www.thelancet.com/hiv Vol 5 October 2018
a prospective cohort study
Pande Putu Januraga, Joanne Reekie, Tri Mulyani, Bony Wiem Lestari, Shelly Iskandar, Rudi Wisaksana, Nur Aini Kusmayanti,
Yanri Wijayanti Subronto, Desak Nyoman Widyanthini, Dewa Nyoman Wirawan, Lydia Verina Wongso, Anindita Gabriella Sudewo,
Evi Sukmaningrum, Tiara Nisa, Bagus Rahmat Prabowo, Matthew Law, David A Cooper*, John M Kaldor
Summary
Background Indonesia has had low uptake of HIV testing and treatment. We did a study to estimate the cascade of
HIV care in key populations and identify predictors of outcomes at key cascade steps.
Methods We used an observational cohort study design to recruit and follow up men who have sex with men (MSM),
female sex workers, transgender women (known as waria in Indonesia), and people who inject drugs (PWID)
diagnosed with HIV in four locations in Indonesia: Bali, Bandung, Jakarta, and Yogyakarta. Recruitment, baseline,
and follow-up visits were done at collaborating clinical services, including both primary care sites and hospitals.
Inclusion criteria for participants included identifying as a member of a key population, age 16 years or older, not
previously tested positive for HIV, and HIV positivity at baseline. All participants were offered treatment as per
national guidelines, with the addition of viral load testing and completion of study-specific forms. Estimates were
calculated of proportions of participants linked to care, commencing treatment, adherent to treatment, and who
achieved virological suppression. We used logistic regression to investigate characteristics associated with
antiretroviral therapy (ART) initiation and viral suppression and Cox regression to identify factors associated with loss
to follow-up. This study is registered with ClinicalTrials.gov, NCT03429842.
Findings Between Sept 15, 2015, and Sept 30, 2016, 831 individuals were enrolled in the study, comprising 637 (77%)
MSM, 116 (14%) female sex workers, 27 (3%) waria, and 51 (6%) PWID. Of those enrolled, 703 (84·6%, 95% CI
82·1–87·1) were linked to HIV care and 606 (86·2%, 83·7–88·8) who were linked with care started ART. Among
participants who started treatment, 457 (75·4%, 71·8–78·9) were retained in care, of whom 325 (71·1%, 66·7–75·2)
had a viral load test about 6 months after enrolment, with 294 (90·5%, 86·7–93·4) of those tested (294 [35%,
32·1–38·7] of the original cohort) virally suppressed. 146 (24%) of 606 who started treatment were lost to follow-up.
People who enrolled at sites that offered both testing and treatment had a higher likelihood of treatment initiation
than those who enrolled at sites offering testing only (p<0·0001 by multivariate analysis), and participants who had
been linked to care and had a high school or university education were significantly more likely to achieve viral
suppression than those with a primary school or lower level of education (p≤0·029 by mulivariate analysis).
Interpretation HIV cascade data among key populations in Indonesia show very poor rates of retention in treatment
and viral suppression. Site and individual characteristics associated with initiating and continuing treatment suggest
an urgent need to develop and implement effective interventions to support patients in achieving viral suppression
among all people with HIV.
Funding Australian Government Department of Foreign Affairs and Trade, WHO, and Indonesian Government.
Copyright © 2018 Elsevier Ltd. All rights reserved.
Introduction Many studies have found gaps in the HIV cascade of
Since the wide adoption of effective antiretroviral therapy care, which links the sequential steps of testing, referral
(ART), HIV-related morbidity and mortality has sub­ to care, uptake of ART, and maintenance of treatment.4
stantially reduced and HIV is now regarded as a Health service interventions have therefore focused on
manageable chronic illness.1 Guidelines recommend minimising leaks in the cascade to ensure the best
immediate initiation of ART after HIV diagnosis, possible health outcomes.5–7 Design and implementation
regardless of the individual’s clinical or immunological of such interventions require a good understanding of
status.2 Furthermore, ART has been shown to substantially the factors that affect each step in the cascade.4,7
reduce the risk of onward HIV transmission to sexual With an estimated 622 000 people living with HIV at the
partners,3 so has been endorsed as a key strategy for end of 2016, Indonesia is one of few countries with an
prevention and treatment. However, despite major increasing number of new HIV infections.8,9 Although
efforts, global coverage of ART was still only 53% in 2016, HIV prevalence in the general Indonesian population is
with 47% coverage in Asia-Pacific countries.1 estimated to be low (0·5%),10 HIV has particularly affected

Research in context
Evidence before this study
Indonesia is one of the few countries in the world that has an
increasing number of new HIV infections. In the Asia-Pacific
region, the HIV epidemic is concentrated among the key
populations of men who have sex with men (MSM), female sex
workers, people who inject drugs (PWID), and transgender women
(waria). UNAIDS has set ambitious 90-90-90 targets for the
proportions of people with HIV who are aware of their status,
referred to care, and have effective viral suppression. Estimates
suggest that less than half of people in key populations with HIV in
the Asia-Pacific region are aware of their HIV status and that only
around a third are referred for treatment. We searched PubMed
from inception until Dec 17, 2017, for publications in English using
the terms “HIV”, “cascade”, or “continuum of care”, and “Asia” or
“Indonesia”. Our search returned two systematic reviews, one in
MSM from 2017 and one in female sex workers from 2014.
Both reviews reported a paucity of evidence on the HIV cascade for
these populations, particularly in low-income and middle-income
countries. Our search also found a study from Cambodia, with a
specific focus on female sex workers, which found that 83% had
initiated ART but only 23% achieved viral suppression.
A questionnaire administered by the European Centre for Disease
Prevention and Control in 2014 across 55 countries in Europe and
central Asia found a paucity of consensus between countries on
definitions of the HIV care cascade, with only 16 reporting data on
all four cascade parameters (estimated numbers living with HIV,
and proportions diagnosed, referred for ART, and with
undetectable viral loads). One cross-sectional study done in
2012–13 reported estimates of cascade parameters in seven
countries in Asia, including Indonesia. However, the study relied
on self-reported outcomes and did not stratify estimates by key
populations. The Thai National AIDS Universal Coverage
key populations at increased risk of HIV infection,
including people who inject drugs (PWID; 39·5%), direct
sex workers (7·2%) and indirect sex workers (1·6%), men
who have sex with men (MSM; 12·8%), and transgender
women (known as waria in Indonesia, 7·4%).8,9,11 The
provinces of West Java, Jakarta, and Bali have the highest
incidences of HIV in the country.10
The Indonesian Ministry of Health has responded to the
growing HIV epidemic with a continuum of HIV care
known as Layanan Komprehensif Berkesinambungan (ie,
continuum of care), an approach first adopted in 2012 that
aims to integrate prevention, treatment, and care for all
populations. This programme authorises doctors to
immediately offer patients ART after HIV diagnosis;
however, coverage of prevention services among key
populations is believed to have remained at less than
55%,10 and only an estimated 10–20% of people with HIV
were receiving ART at the end of 2016.8
The HIV Awal (early) Testing and Treatment in
Indonesia (HATI) study was developed to prospectively
evaluate strategies to improve the cascade of HIV care in
www.thelancet.com/hiv Vol 5 October 2018 e561
Articles
Programme reported on its cascade for patients involved in the
programme from 2000 to 2014, although, again, they did not
consider the cascade by key population.
Added value of this study
To our knowledge, this study is the first prospective, longitudinal
investigation from any country in Asia of the cascade of HIV care
among MSM, female sex workers, PWID, and waria. We found
that the proportion of participants who were retained in care and
achieved viral suppression was substantially lower among female
sex workers than among the other key populations. Individuals
who were diagnosed with HIV at a health-care facility that also
offered ART were more likely to initiate treatment than those
who were diagnosed at a facility that only offered testing. We also
highlight methodological issues involved in the direct
measurement of cascade parameters, including accounting for
people who are undiagnosed in key populations and improving
patient tracking during follow-up.
Implications of all the available evidence
The paucity of evidence on the HIV care cascade among key
populations in low-income and middle-income countries is
concerning. Our study finds strikingly poor rates of retention in
treatment and viral suppression in Indonesia, a country with
the world’s fourth largest population and a fast-growing HIV
epidemic in several key populations. To achieve the global
UNAIDS 90-90-90 targets, attrition at each step of the care
cascade needs to be well defined and understood, so that future
interventions can be suitably tailored and targeted to meet the
needs of the most affected key populations in each country.
Such interventions could involve simplifying ART initiation
procedures and developing support systems for treatment
adherence that will be piloted in this study’s second phase.
key populations. In HATI’s baseline cohort phase, the
first direct estimates of patient retention in cascade steps
have been generated for Indonesia, which, like many
other countries, has previously relied on modelling and
programmatic data to provide this information.4,12 Here
we report the first year’s cascade results. The subsequent
intervention phase (ongoing at the time of writing this
Article) is not the subject of this report.
Methods
Study design and participants
We did a multicentre, observational cohort study in Bali,
Bandung, Jakarta, and Yogyakarta in Indonesia. These
sites were chosen because HIV prevalence among key
populations in these locations is high,8,13 services for HIV
prevention and treatment were already in place, and each
location had a university-affiliated coordinating unit with
experience in HIV/AIDS research among key populations.
These four coordinating units implemented the study and
oversaw its operation through collaboration with local
clinics. In Bali, the coordinating unit was Yayasan Kerti
 Articles
Praja, which has operated a large sexually transmitted
infection and HIV clinical service in Denpasar, Bali, for
key populations since 1992. In Yogyakarta, the coordinating
unit was Dr Sardjito Hospital and in Bandung, the unit
was Hasan Sadikin Hospital, both of which are tertiary
facilities that provide HIV clinical services. In Jakarta, the
coordinating unit was the AIDS Research Center at Atma
Jaya Catholic University.
Each coordinating unit established its own network of
satellite clinics for recruitment and follow-up of
See Online for appendix participants (appendix p 1). Satellite clinics were selected
by site staff because of their records in recruiting key
populations for HIV testing. 21 clinics were recruited as
satellite sites. The five clinics in Bali and three in
Yogyakarta were government-run community health
centres (known as puskesmas) or non-government
clinics that referred patients to their associated
coordinating unit for ART. Of the six clinics in Bandung,
two were hospitals that offered ART, and four were
puskesmas and non-government clinics that referred
patients to the coordinating unit for ART. Of the seven
clinics in Jakarta, six were puskesmas and one was a
hospital, all of which offered ART. All puskesmas and
private clinics in the study were also able to do mobile
HIV testing for key population groups.
Participants were eligible for inclusion if they were from
a key population (ie, PWID, female sex workers, MSM,
and waria), aged 16 years or older with no upper age limit,
had not previously tested positive for HIV, and were able
to provide written informed consent. Eligible individuals
were screened at a participating clinic for HIV infection by
use of three rapid HIV antibody tests (SD Bioline HIV-1/2
3.0, Standard Diagnostics, Abbott, Chicago, IL, USA;
Advanced Quality One Step Anti-HIV [1 & 2] Test, InTec
Products, Xiamen, China; and VIKIA HIV 1/2,
bioMérieux, Marcy-l’Étoile, France) and those who were
HIV postive were invited to enter the study. Because of the
small number of newly positive PWID recorded, and a
high prevalence of PWID known to be HIV positive, the
inclusion criteria for this group were expanded to also
include those who had been previously diagnosed with
HIV but were ART naive.8 Exclusion criteria were a clinical
condition that could be adversely affected by study
procedures, immediate ART contraindication, or
participation in another study with a similar focus.
Each site recruited specific key populations on the
basis of the populations frequently seen by clinics. The
Bandung site aimed to recruit female sex workers, MSM,
waria, and PWID; the Bali and Yogyakarta sites aimed to
recruit female sex workers, MSM, and waria; and the
Jakarta site only recruited PWID. Female sex workers
were defined on the basis of having engaged in sex work
in the past 12 months, MSM were defined as men who
self-reported anal sex with a man in the past 12 months,
waria were defined on the basis of self-identification, and
PWID were defined on the basis of self-reported history
of ever having injected drugs.
e562 www.thelancet.com/hiv Vol 5 October 2018
HATI received ethical approval from the ethics
committee of the National Institute for Health Research
and Development (NIHRD), Ministry of Health Indonesia,
and the UNSW Human Research Ethics Committee. All
participants provided written informed consent.
Procedures
Recruitment and follow-up procedures are presented in
the appendix (pp 1–2). People with HIV attending a study
site who met inclusion criteria were offered study
enrolment by clinic staff. Apart from completion of
study-specific enrolment forms and viral load testing,
treatment at all study sites followed national guidelines
and were applied uniformly at participating clinical sites
regardless of study enrolment status. These guidelines
include the offer of immediate ART initiation, including
referral to treatment sites, after CD4 cell count and pre-
ART diagnostic tests. The Ministry of Health covered all
costs of ART for study participants. CD4 cell count and
viral load testing were provided free of charge to study
participants at ART initiation and every 6 months
thereafter.
Standardised data collection forms were completed for
each participant at eligibility screening, registration to
the study, first visit to an ART treatment site, and follow-
up visits. Information was collected on demographic
characteristics (date of birth, self-identified gender,
marital status, education, and employment). Information
collected at follow-up visits included CD4 cell count,
stage of HIV disease, HIV-related conditions (eg,
tuberculosis and other opportunistic infections), viral
load, ART regimens, and body-mass index (BMI). A
short behavioural questionnaire was completed at entry.
Outcomes
The events of interests were the number of participants
who enrolled in the baseline cohort, were linked with
HIV care (ie, received at least one HIV-related service at a
study treatment site), initiated ART at a study treatment
site, were retained in care (ie, had at least two outpatient
visits at least 90 days apart after ART initiation), had a
viral load test at around 6 months after ART initiation (ie,
at 5–9 months after ART initiation), achieved viral
suppression (ie, less than 200 copies per mL), and were
lost to follow-up (ie, 180 days without any visit to clinic
after starting ART). These definitions have been reported
elsewhere.4,12,14–16
Statistical analysis
Statistical analyses included all participants recruited
into the baseline cohort for whom at least 6 months had
elapsed since recruitment. We summarised the
demographic characteristics of the cohort using
descriptive statistics.
The target sample size across all locations was
600 MSM and 600 female sex workers who were positive
for HIV, with the aim to recruit on average 200 individuals
 Articles

from each of these key populations in each year (here we

Number of participants (n=831)
report phase one, year 1, of our study). For PWID we
aimed to recruit 200 people across the study period. Our Locations

prospective sample sizes were determined such that any Bali 281 (34%)
changes in study outcomes would be detected. Because Bandung 377 (45%)
the size of the waria population is considerably smaller Jakarta 30 (4%)
than the other key populations in this study, we had no Yogyakarta 143 (17%)
specified target for the HIV-positive waria sample. Key populations
We calculated the proportion of participants who MSM 637 (77%)
proceeded through specific steps in the cascade on the FSWs 116 (14%)
basis of the number of participants for whom relevant Waria 27 (3%)
outcomes took place, divided by the number who had PWID 51 (6%)
attained the preceding step. Specifically, we defined the Place of enrolment
proportion of participants transferred to care as the total Testing site only 223 (27%)
number who reached event two, divided by the number Testing and treatment site 608 (73%)
who were recruited (at event one). The proportion of Age (years)
participants who commenced ART was the number who Mean (SD) 29·1 (7·9)
reached event three divided by the number who reached Median (IQR) 27·3 (23·1–32·9)
event two. We calculated the outcomes for the cohort Age group (years)
overall and for each of the key populations. We calculated 16–19 38 (5%)
person-years of follow-up for patients who were referred 20–29 461 (55%)
to care and who started treatment. 30–39 234 (28%)
We used univariate and multivariate logistic regression 40–59 98 (12%)
analysis to identify factors associated with the likelihood of Gender
ART initiation and viral suppression at about 6 months Male 681 (82%)
after enrolment. Factors included in all models were Female 123 (15%)
location, key population, place of enrolment, age, marital Transgender 27 (3%)
status, education, and employment. Additionally, we Marital status
investigated the clinical characteristics of those who Currently married 88 (11%)
initiated ART as predictors of viral suppression at about
Never married 625 (75%)
6 months. These characteristics included ART regimen,
Divorced 116 (14%)
BMI, CD4 cell count, viral load, stage of HIV disease, and
Data missing 2 (<1%)
HIV-related conditions at the time of ART initiation. We
Highest level of education
used Cox regression to identify factors associated with loss
Primary school or less 71 (9%)
to follow-up, and calculated the hazard ratio (HR) associated
Junior high school 138 (17%)
with each factor using univariate and multivariate analyses.
Senior high school 433 (52%)
We did all analyses using Stata Statistical Software: release
University or academy 188 (23%)
12 (Stata/SE 12).
Data missing 1 (<1%)
The study is registered with ClinicalTrials.gov,
Employment
NCT03429842.
Employed 646 (78%)

Role of the funding source Unemployed or student 183 (22%)

The study was funded by Australian Government Data missing 2 (<1%)

Department of Foreign Affairs and Trade (DFAT), WHO, Data are n (%) unless otherwise stated. MSM=men who have sex with men.
and the Indonesian Government (Ministry of Health). FSWs=female sex workers. PWID=people who inject drugs.

DFAT and the Indonesian Ministry of Health had a seat
on the advisory committee, but had no role in the study.
WHO was a collaborator in all aspects of the study,
including study design, data collection, data analysis,
data interpretation, and writing of the report. The
corresponding author had full access to all the data in the
study and had final responsibility for the decision to
submit for publication.
Results
Between Sept 15, 2015, and Sept 30, 2016, 831 people
infected with HIV were recruited and 606 (73%)
Table 1: Baseline demographic characteristics
participants started ART, resulting in 358·3 person-years
of follow-up. Most participants were men and the
predominant key population was MSM (table 1). Bandung
and Bali contributed most participants, and 78% of
female sex workers came from Bali (appendix p 3). Most
participants were in the age group 20–39 years, and the
median age was 27·3 years (IQR 23·1–32·9). The median
time from first HIV diagnosis to recruitment for
participants other than PWID was 0 days (range 0–271),

www.thelancet.com/hiv Vol 5 October 2018 e563

 Articles

did. Most participants were employed and had never been

Started Lost to Univariate analysis Multivariate analysis
ART follow-up (HR, 95% CI; p value) (HR, 95% CI; p value) married (table 1).
(n=606) (n, %) Of 831 individuals recruited, 703 (84·6%, 95% CI
Locations 82·1–87·1) attended an ART site after HIV diagnosis.
Bali 225 72 (32%) 1 1 CD4 counts were lower than expected, with 481 (73%) of
Bandung 233 52 (22%) 0·59 (0·38–0·90; 0·015) 0·58 (0·33–1·01; 0·052) 662 participants tested having a count of less than
Jakarta 21 4 (19%) 0·56 (0·17–1·79; 0·329) 0·15 (0·02–1·04; 0·054)
350 cells per μL and 106 (15%) of 701 with available HIV
Yogyakarta 127 19 (15%) 0·27 (0·13–0·56; 0·001) 0·29 (0·13–0·63; 0·002)
stage data having HIV stage 3 or 4. Of 703 participants
Age
who accessed an ART treatment site after enrolment,
606 (86·2%, 95% CI 83·7–88·8) started treatment
Per year older 606 147 (24%) 0·97 (0·95–0·10; 0·052) 0·96 (0·93–0·99; 0·019)
(table 2), with 503 (83%) given fixed dose combination of
Key populations
tenofovir, lamivudine, and efavirenz. The median time
MSM 466 95 (20%) 1 1
from study registration to treatment initiation was 7 days
FSWs 87 37 (43%) 2·94 (1·85–4·66; <0·0001) 1·49 (0·67–3·28; 0·328)
(IQR 3–17).
Waria 19 6 (32%) 2·40 (1·04–5·56; 0·041) 2·17 (0·84–5·61; 0·110)
457 (75·4%, 95% CI 71·8–78·9) of 606 participants
PWID 34 8 (24%) 1·38 (0·59–3·18; 0·457) 2·27 (0·65–8·02; 0·200)
who initiated ART were retained in care. Of whom,
Education
325 (71·1%, 66·7–75·2) had a viral load test at about
Primary school or less 48 20 (42%) 1 1
6 months after initiating ART, with 294 (90·5%,
Junior high school 97 34 (35%) 0·86 (0·44–1·66; 0·649) 0·99 (0·47–2·08; 0·978)
86·7–93·4) of 325 participants tested found to be virally
Senior high school 308 71 (23%) 0·41 (0·22–0·74; 0·003) 0·62 (0·30–1·27; 0·193)
suppressed (ie, 294 [35%, 32·1–38·7] of the original
University or academy 153 22 (14%) 0·23 (0·11–0·49; <0·0001) 0·42 (0·17–1·03; 0·059)
cohort).
Employment Of 703 participants who attended an ART site, most
Employed 474 116 (24%) 1 1 had been diagnosed with HIV at a primary care centre,
Unemployed or student 131 31 (24%) 1·05 (0·66–1·70; 0·826) 1·16 (0·67–2·02; 0·589) 302 (43%) at a puskesma, 239 (34%) at a non-government
Marital status community centre, and 111 (16%) at a hospital. 474 (67%)
Currently married 66 14 (21%) 1 1 had their baseline visit at the facilities where they were
Never married 455 102 (22%) 1·06 (0·53–2·12; 0·874) 0·90 (0·42–1·95; 0·705) recruited. Comparing key populations (figure), the lowest
Divorced 84 31 (37%) 2·29 (1·06–4·95; 0·035) 0·99 (0·41–2·37; 0·638) proportion of participants who were retained in care
Place of enrolment were female sex workers.
Testing site only 78 24 (31%) 1 1 We found participants were more likely to start
Testing and treatment site 528 123 (23%) 0·85 (0·48–1·53; 0·596) 0·88 (0·46–1·70; 0·705) treatment if they enrolled at sites that offered both testing
Stage of HIV and treatment than if they enrolled at a site offering
Stage 1–2 519 126 (24%) 1 1 testing only (table 3). Participants with an education level
Stage 3–4 87 21 (24%) 0·99 (0·55–1·77; 0·971) 1·18 (0·59–2·38; 0·638) beyond high school were more likely to start treatment
CD4 count at baseline (cells per μL) than those with a lower level of education (table 3).
<200 208 51 (25%) 1 1 Participants from Bandung were significantly less likely
200–349 220 42 (19%) 0·74 (0·45–1·21; 0·230) 0·61 (0·36–1·06; 0·077) to begin care than participants from Bali, whereas
350–499 105 24 (23%) 0·97 (0·54–1·73; 0·921) 0·75 (0·40–1·38; 0·349) participants from Yogyakarta were significantly more
≥500 53 20 (38%) 1·73 (0·93–3·21; 0·084) 0·96 (0·47–1·94; 0·907) likely to begin care than those from Bali (table 3). We
ART regimens found no significant association between initiation of
Contains efavirenz 550 138 (25%) 1 1 treatment and other demographic characteristics after
Other combinations 56 9 (16%) 0·75 (0·35–1·62; 0·469) 0·55 (0·23–1·28; 0·163) adjusting for other variables.
Level of education was significantly associated with
Likelihood of loss to follow-up was determined by use of univariate and multivariate Cox regression analysis.
ART=antiretroviral therapy. FSWs=female sex workers. HR=hazard ratio. MSM=men who have sex with men. viral suppression, with participants who had a junior and
PWID=people who inject drugs. senior high school and university qualifications having a
Table 2: Associations between individual characteristics and loss to follow-up among people who started
significantly higher likelihood of suppression than those
treatment with an educational level of primary school or less
(table 4). Patients with a more advanced disease stage
when starting treatment were less likely to achieve
and for PWID was 24 days (0–4360). Waria and PWID had suppression than those with a less advanced disease
a higher median age (34·0 years [IQR 31·4–36·1]) than stage (table 4). Older participants were more likely to
did MSM (26·4 years [22·7–31·5]) and female sex workers achieve suppression than younger participants (table 4).
(29·2 years [25·0–34·2]). Over half of participants had Multiple logistic regression of only MSM and female sex
high school, university, or academy education (table 1) . worker data gave the same results (appendix p 4).
More MSM (545 [86%] of 637) and PWID (34 [68%] of 50) Among the 606 participants who started ART,
had a high school or higher level of education than female nine (<1%) were reported by their friends, family, or
sex workers (30 [26%] of 116) and waria (12 [44%] of 27) outreach workers to have died, 38 (<1%) were recorded

e564 www.thelancet.com/hiv Vol 5 October 2018

 Articles

as transferred to another treatment site, and 147 (2%) Enrolled Linked to care Started ART Retained in care
were lost to follow-up. 43% of female sex workers (37 of Viral load test at 6 months Supressed viral load
87) were lost to follow-up, compared with 20–32% 100 100 100 100 100 100
91 90
in the other populations. Older participants were 90 85 85
83
significantly less likely to be lost to follow-up than 80 75
73 73
younger participants across all key populations (table 2). 70
70 67
Participants from Yogyakarta were significantly less

Proportion (%)
60 55 57 56 55
likely to be lost to follow-up after starting ART than 50 45
43
those from Bali (table 2). 40
39
35
39
30
30 27 26
Discussion 20
23 22 20

In our prospective study of the HIV care cascade among 10

key populations in Indonesia, we found that referral to
0
care took place for a high proportion of those diagnosed Overall (n=831) MSM (n=637) FSW (n=116) Waria (n=27) PWID (n=51)
with HIV, but saw a substantial decrease in the proportion
Figure: The cascade of HIV care by key population groups
of referred participants who then initiated treatment and Proportion of each key population remaining in care at each stage of the cascade. Viral load suppression defined as
were subsequently retained in care with viral suppression. 200 copies per mL or less. ART=antiretroviral therapy. MSM=men who have sex with men. FSWs=female sex
Few studies have explored the HIV care cascade in workers. PWID=people who inject drugs.
high-risk populations. A study done in India investigated
the HIV care cascade among MSM and PWID by use of a Enrolled Started Univariate analysis Multivariate analysis (OR,
cross-sectional study design,17 and a small cohort study (n=831) ART (n, %) (OR, 95% CI; p value) 95% CI; p value)
from Cambodia reported the care cascade in female sex Locations
workers.18 To our knowledge, this study is the first to Bali 281 225 (80%) 1 1
prospectively measure the cascade of HIV care in key Bandung 377 233 (62%) 0·40 (0·28–0·58; <0·0001) 0·44 (0·27–0·72; 0·001)
populations at high risk of HIV infection in a country in Jakarta 30 21 (70%) 0·58 (0·25–1·34; 0·202) 0·31 (0·70–1·38; 0·123)
the Asia-Pacific region. A systematic review19 of treatment Yogyakarta 143 127 (89%) 1·97 (1·09–2·59; 0·025) 3·83 (1·84–7·99; <0·0001)
as prevention for HIV among MSM found one report Age
from the Asia-Pacific region regarding the HIV care Per year older 831 606 (73%) 1·00 (0·98–1·02; 0·938) 0·98 (0·95–1·01; 0·147)
cascade that was not based on a prospective study. A
Key populations
systematic review20 of articles reporting treatment uptake,
MSM 637 466 (73%) 1 1
attrition, adherence, and outcomes among female sex
FSWs 116 87 (75%) 1·10 (0·70–1·74; 0·697) 1·50 (0·63–3·54; 0·359)
workers with HIV found very few reports from low-
Waria 27 19 (70%) 0·87 (0·37–2·03; 0·750) 2·01 (0·66–6·08; 0·218)
income and middle-income countries. In the Pacific
PWID 51 34 (67%) 0·73 (0·40–1·35; 0·319) 0·90 (0·26–3·07; 0·870)
region, only Australia has reported estimates of the HIV
Education
care cascade for all people with HIV.21
Primary school or 71 48 (68%) 1 1
A key strength of our study was our use of less
comprehensive and standardised data collection methods Junior high school 138 97 (70%) 1·13 (0·61–2·10; 0·690) 1·58 (0·70–3·57; 0·268)
from a wide variety of clinical sites, which included Senior high school 433 308 (71%) 1·18 (0·69–2·02: 0·546) 1·20 (0·56–2·61; 0·636)
primary care clinics and hospitals. This diversity allowed
University or 188 153 (81%) 2·09 (1·12–3·89; 0·019) 2·96 (1·24–7·08; 0·015)
us to recruit individuals diagnosed with HIV from academy
different groups in the participating locations.16 Employment
By multivariate analysis, individuals who enrolled at Employed 646 474 (73%) 1 1
sites that offered both testing and treatment had an Unemployed or 183 131 (72%) 0·91 (0·63–1·32; 0·630) 0·78 (0·48–1·26; 0·308)
increased likelihood of starting treatment compared with student
those attending sites that offered testing only. This result Marital status
indicates the need to expand treatment access to primary Currently married 88 66 (75%) 1 1
care.22 Since 2015, the Indonesian Ministry of Health has Never married 625 455 (73%) 0·89 (0·53–1·49; 0·663) 0·80 (0·39–1·61; 0·531)
endeavoured to move HIV treatment into primary care by Divorced 116 84 (72%) 0·87 (0·47–1·65; 0·678) 0·81 (0·35–1·89; 0·627)
providing training to staff at puskesmas and encouraging Place of enrolment
back-referral of patients for treatment continuation at Testing site only 223 78 (35%) 1 1
these centres. A qualitative study23 in Bali found that Testing and 608 528 (87%) 12·27 (8·54–17·62; <0·0001) 18·80 (12·18–29·00; <0·0001)
perceived competency and workload, requirement of treatment site
laboratory testing as per guide­lines, and fear of stigma
Likelihood of ART initiation was determined by use of univariate and multivariate logistic regression analysis.
among patients when accessing voluntary counselling ART=antiretroviral therapy. MSM=men who have sex with men. FSWs=female sex workers. PWID=people who inject
and testing and ART services were barriers to integrating drugs.
ART in puskesmas. Strategies to simplify procedures for
Table 3: Associations between individual characteristics and initiation of ART
starting and monitoring treatment in primary care,

www.thelancet.com/hiv Vol 5 October 2018 e565

 Articles

either an association between acquisition of HIV infection

Linked Suppressed Univariate analysis (OR, Multivariate analysis (OR,
to HIV viral load 95% CI; p value) 95% CI; p value) and education, or that the study clinics are preferentially
care (n, %) attracting members of the key populations who have a
(n=703) high level of education. Although a high level of education
Locations is often associated with income, we believe that costs of
Bali 262 103 (39%) 1 1 treatment do not play a role in treatment uptake and
Bandung 284 127 (45%) 1·32 (0·92–1·91; 0·134) 1·20 (0·77–1·89; 0·419) retention, because ART and voluntary counsel­ ling and
Jakarta 30 6 (20%) 0·47 (0·18–1·27; 0·136) 2·74 (0·45–16·85; 0·276) treatment services are available for free in all study sites.
Yogyakarta 127 58 (46%) 0·99 (0·64–1·54; 0·984) 0·88 (0·53–1·46; 0·620) We also found that high educational attainment was a
Age predictor of starting treatment. Collectively, these findings
Per year older 703 294 (42%) 1·01 (0·99–1·03; 0·237) 1·03 (1·01–1·06; 0·019) indicate the need for health providers to make accurate
Key populations information available to service users, particularly those
MSM 528 250 (47%) 1 1 with a low level of education, so that they can understand
FSWs 106 27 (25%) 0·40 (0·24–0·65; <0·0001) 0·62 (0·28–1·39; 0·245) the benefits of treatment and maximise treatment uptake
Waria 23 7 (30%) 0·52 (0·20–1·33; 0·174) 0·67 (0·23–1·93; 0·454) and adherence.25 Programmes to increase the quality of
PWID 46 10 (22%) 0·32 (0·15–0·70; 0·004) 0·29 (0·07–1·26; 0·099)
consultations before and during treatment can improve
Education
outcomes.26 Consistent with previous studies,27,28 our
Primary school or less 63 12 (19%) 1 1
findings suggest that older patients were less likely to be
lost to follow-up than younger patients (ie, for every year
Junior high School 111 36 (32%) 1·99 (0·90–4·37; 0·089) 2·64 (1·10–6·30; 0·029)
increase in age, likelihood decreased). Developing targeted
Senior high School 360 156 (43%) 3·36 (1·66–6·84; 0·001) 3·82 (1·65–8·83; 0·002)
interventions for younger people with HIV by use of
University or 168 90 (54%) 4·68 (2·22–9·86; <0·0001) 5·18 (2·11–12·73; <0·0001)
academy innovative approaches, including mobile technology, could
Employment be considered to improve retention in care.29
Employed 551 233 (42%) 1 1 A key limitation of our study is the representativeness
Unemployed or 150 61 (41%) 0·90 (0·61–1·32; 0·581) 0·89 (0·57–1·28; 0·611)
of the participants and generalisability of the care
student continuum in these key populations across Indonesia.
Marital status Although our study included a variety of sites with high
Currently married 80 30 (38%) 1 1 reported numbers of patients with HIV, we were not able
Never married 521 233 (45%) 1·23 (0·73–2·06; 0·440) 0·91 (0·48–1·73; 0·781) to compare their characteristics with those of people with
Divorced 100 31 (31%) 0·67 (0·35–1·29; 0·228) 1·10 (0·48–2·54; 0·817) HIV from key population groups diagnosed at other non-
Place of enrolment study sites or who remain undiagnosed. The structure of
Testing site only 110 29 (26%) 1 1 Indonesia’s HIV information system did not allow us to
Testing and 593 265 (45%) 1·65 (1·01–2·70; 0·045) 1·45 (0·83–2·55; 0·191) distinguish patients recruited into the study from others
treatment site who were diagnosed with HIV across all participating
Stage of HIV sites who did not participate in the study. Because of
Stage 1–2 595 265 (45%) 1 1 differences in patient-related factors and quality of
Stage 3–4 106 29 (27%) 0·43 (0·30–0·71; 0·001) 0·45 (0·25–0·81; 0·008) clinical care and outreach support for key populations,
CD4 count at baseline (cells per μL) proportions of patients who are linked with care, retained
<200 237 99 (42%) 1 1 in care, and lost to follow-up could differ compared with
200–349 244 118 (48%) 1·35 (0·92–1·97; 0·122) 1·19 (0·78–1·82; 0·429) other sites in the study cities, and other cities in the
350–499 114 52 (46%) 1·10 (0·69–1·76; 0·686) 0·98 (0·58–1·65; 0·944) region.27
≥500 67 21 (31%) 0·77 (0·41–1·42; 0·394) 0·77 (0·39–1·51; 0·442) However, using available data at participating clinics
ART regimens and the Bali Health Office, we found that the
Contains efavirenz 550 268 (49%) 1 1 281 individuals we recruited in Bali made up 88% of
Other combinations 56 26 (46%) 0·91 (0·52–1·58; 0·743) 0·90 (0·49–1·64; 0·733)
320 individuals who had been newly diagnosed with HIV
at study clinics during the study period, and 70% of
Viral load suppression defined as 200 copies per mL or less. Likelihood of viral suppression was determined by use of 401 individuals who had been newly diagnosed among
univariate and multivariate logistic regression analysis. ART=antiretroviral therapy. FSWs=female sex workers.
MSM=men who have sex with men. OR=odds ratio. PWID=people who inject drugs.
key population groups across all clinical sites in Bali in
the same period. Information from lead investigators at
Table 4: Association between individual characteristics and attaining viral load suppression among other coordinating units suggest that similarly high
participants linked to HIV care
recruitment was achieved across participating clinics in
the other study cities, and that the participants in the
including simpler laboratory algorithms and access to cohort were likely to be representative of people seen at
point-of-care equipment for monitoring treatment those clinics.
outcomes, could allow the expansion of this option.24 The generalisability of our results is probably most valid
Our data indicate that most of the study population, for MSM, because they were the largest key population
particularly MSM, had a high education level, suggesting recruited in three cities (Bangung, Bali, and Yogyakarta).

e566 www.thelancet.com/hiv Vol 5 October 2018


This predominance could reflect an increasing epidemic
of HIV among MSM in Indonesia,8,10 or a greater
willingness of this population to seek testing and
treatment. Our study was unable to recruit substantial
numbers of female sex workers in Bandung and
Yogyakarta, despite evidence that HIV prevalence in these
cities differs little from the prevalence in Bali, where most
female sex workers in this study were recruited.8 Two
reasons for this discrepancy potentially exist. First, the
main Bali study site is a long established local non-
governmental organisation with years of experience in
outreach and clinical services for female sex workers.
Second, Bali has allowed brothel complexes (although
they are not legal), by contrast with the other cities where
local authorities have closed them. Consequently, findings
from the female sex worker group might be less
generalisable to other parts of Indonesia than they are for
the MSM group. Nevertheless, attention should focus on
improving the number of female sex workers who enter
the HIV care cascade, particularly because the sex-work
industry in Indonesia involves a lot of stigma and mobility,
and low education.30 The small number of waria and
PWID recruited preclude us from making separate
conclusions for these populations.
Another key limitation of our study was our inability
to measure the first step in the cascade, defined as the
proportion of people with HIV who have been tested
and found to be HIV positive. Using the national
estimate that 44% of people with HIV have been tested,31
our figures in this study for the next two steps of the
HIV care cascade indicate that substantial effort will be
needed for Indonesia to fulfil the 90-90-90 commitment.
Furthermore, in the absence of a mechanism to
comprehensively track patients’ move­ ments, deaths,
and other events, some participants could have been
erroneously classified as lost to follow-up when they
might have died or moved to another care site in the
same area that was not involved in the study.27 Hence,
an expanded system to track patients across sites is
needed.
In conclusion, we have provided, to our knowledge, the
first direct, prospective evidence of the status of the
continuum of care in Indonesia for people with HIV from
key populations in Bali, Bandung, Jakarta, and Yogyakarta.
With cascade of care parameters far worse than those
needed to reach the global 90-90-90 commit­ment, much
remains to be done. The participating sites in our study
were overseen by highly experienced HIV clinical teams;
as such, ART programme sites in other parts of the
country where staff might be less experienced could be
even further from reaching the target commitments. The
site and individual characteristics associated with
initiating and continuing treatment suggest an urgent
need to develop and implement effective interventions to
support patients in achieving viral suppression among all
people with HIV. These interventions could involve
simplifying ART initiation procedures so that testing sites
www.thelancet.com/hiv Vol 5 October 2018 e567
Articles
could offer treatment and develop support systems for
treatment adherence (eg, a reminder system that uses
mobile technology, or individual adherence consultations).
The study’s second phase, involving new interventions
with the aim of improving the HIV care cascade, will be
important to pilot and evaluate strategies for testing and
treatment that could then be adopted more widely in
Indonesia, and elsewhere.
Contributors
PPJ did the data analysis and wrote the manuscript draft. JR helped PPJ
to analyse the data and provide key input into the manuscript
development. TM coordinated data management of the study. DNWir,
RW, YWS, ES, TN, BRP, ML DAC, and JMK contributed to development
of the study design, data collection methods, and data interpretation.
BWL, SI, NAK, and AGS led data collection at each site. DNWid and
LVW coordinated data management at each site. All authors approved
the final version of the manuscript.
Declaration of interests
SI and BRP report grants from WHO during the conduct of the study.
ML reports grants from Boehringer Ingelhiem, Gilead Sciences,
Merck Sharp & Dohme, Bristol-Myers Squibb, and Janssen-Cilag,
ViiV HealthCare, and personal fees from Gilead Sciences, and Sirtex Pty,
outside the submitted work. All other authors declare no competing
interests.
Acknowledgments
We dedicate this report to the memory of our dear friend and colleague
Prof David A Cooper who initiated and guided the project, and who died
in March, 2018. We thank Bradley Mathers and Amit C Achhra for their
contribution to the study design and implementation of the first phase
of the HATI study. We thank the Ministry of Health of Indonesia for
supporting the study team from the outset. We also thank the clinical
and outreach staff at all participating health services, and representatives
of community organisations who provided crucial advice on study
design and conduct. We thank the study participants who gave their time
and knowledge. This work was supported by the Australian Government
Department of Foreign Affairs and Trade, WHO, and the Indonesian
Government.
References
1 UNAIDS. UNAIDS data 2017. Geneva: UNAIDS, 2017.
2 INSIGHT START Study Group. Initiation of antiretroviral therapy
in early asymptomatic HIV Infection. N Engl J Med 2015;
373: 795–807.
3 Cohen MS, Chen YQ, McCauley M, et al. Antiretroviral therapy for the
prevention of HIV-1 transmission. N Engl J Med 2016; 375: 830–39.
4 MacCarthy S, Hoffmann M, Ferguson L, et al. The HIV care
cascade: models, measures and moving forward. J Int AIDS Soc
2015; 18: 19395.
5 Muhamadi L, Tumwesigye NM, Kadobera D, et al. A single-blind
randomized controlled trial to evaluate the effect of extended
counseling on uptake of pre-antiretroviral care in eastern Uganda.
Trials 2011; 12: 184.
6 Losina E, Touré H, Uhler LM, et al. Cost-effectiveness of preventing
loss to follow-up in HIV treatment programs: a Côte d’Ivoire
appraisal. PLoS Med 2009; 6: e1000173.
7 Alvarez-Uria G, Pakam R, Midde M, Naik PK. Entry, retention, and
virological suppression in an hiv cohort study in india: description
of the cascade of care and implications for reducing HIV-related
mortality in low- and middle-income countries.
Interdiscip Perspect Infect Dis 2013; 2013: 384805.
8 Ministry of Health of Indonesia. Estimation and projection of
HIV/AIDS in Indonesia 2015–2020. Jakarta: Ministry of Health of
Indonesia, 2017. http://siha.depkes.go.id/portal/files_upload/
ESTIMATES_AND_PROJECTION_OF_HIVAIDS_IN_
INDONESIA_2015___2020.pdf (accessed Aug 13, 2018).
9 WHO Regional Office for South-East Asia. Progress report on HIV
in the WHO South-East Asia region 2016. India: World Health
Organization Regional Office for South-East Asia, 2016.
http://apps.searo.who.int/PDS_DOCS/B5282.pdf (accessed
Aug 13, 2018).
 Articles
10 AIDS Data Hub. Evidence to action: Indonesia country profile at a
glance: National AIDS Commission of Indonesia, 2011.
http://www.aidsdatahub.org/Country-Profiles/Indonesia (accessed
July 17, 2018).
11 Januraga PP, Wulandari LP, Muliawan P, et al. Sharply rising
prevalence of HIV infection in Bali: a critical assessment of the
surveillance data. Int J STD AIDS 2013; 24: 633–37.
12 Kilmarx PH, Mutasa-Apollo T. Patching a leaky pipe: the cascade of
HIV care. Curr Opin HIV AIDS 2013; 8: 59–64.
13 Ministry of Health Republic of Indonesia. IBBS 2011. Integrated
biological and behavioral surveillance. Ministry of Health Republic
Indonesia, 2011. http://www.aidsdatahub.org/sites/default/files/
documents/IBBS_2011_Report_Indonesia.pdf (accessed July 7, 2018).
14 Hull MW, Wu Z, Montaner JS. Optimizing the engagement of care
cascade: a critical step to maximize the impact of HIV treatment as
prevention. Curr Opin HIV AIDS 2012; 7: 579–86.
15 Kohler P, Schmidt AJ, Cavassini M, et al. The HIV care cascade in
Switzerland: reaching the UNAIDS/WHO targets for patients
diagnosed with HIV. AIDS 2015; 29: 2509–15.
16 Mugavero MJ, Amico KR, Horn T, Thompson MA. The state of
engagement in HIV care in the United States: from cascade to
continuum to control. Clin Infect Dis 2013; 57: 1164–71.
17 Mehta SH, Lucas GM, Solomon S, et al. HIV care cascade among
hard to reach populations in India: need to expand hiv counseling
and testing. Conference on Retroviruses and Opportunistic
Infections; Boston, MA, USA; March 3–6, 2014. 1063.
18 Muth S, Len A, Evans JL, et al. HIV treatment cascade among
female entertainment and sex workers in Cambodia: impact of
amphetamine use and an HIV prevention program.
Addict Sci Clin Pract 2017; 12: 20.
19 van Griensven F, Guadamuz TE, de Lind van Wijngaarden JW,
Phanuphak N, Solomon SS, Lo YR. Challenges and emerging
opportunities for the HIV prevention, treatment and care cascade in
men who have sex with men in Asia Pacific. Sex Transm Infect 2017;
93: 356–62.
20 Mountain E, Mishra S, Vickerman P, Pickles M, Gilks C, Boily M-C.
Antiretroviral therapy uptake, attrition, adherence and outcomes
among HIV-infected female sex workers: a systematic review and
meta-analysis. PLoS One 2014; 9: e105645.
21 Keen P, Gray RT, Telfer B, et al. The 2016 HIV diagnosis and care
cascade in New South Wales, Australia: meeting the UNAIDS
90-90-90 targets. J Int AIDS Soc 2018; 21: e25109.
e568 www.thelancet.com/hiv Vol 5 October 2018
22 Clouse K, Pettifor AE, Maskew M, et al. Patient retention from HIV
diagnosis through one year on antiretroviral therapy at a primary
healthcare clinic in Johannesburg, South Africa.
J Acquir Immune Defic Syndr 2013; 62: e39–46.
23 Sugiana M, Sutarsa N, Pradnyaparamita Duarsa D. Barriers to
integrating antiretroviral therapy services into community health
centre: a qualitative study in Badung regency.
Public Health Prev Med Arch 2015; 3: 11 (in Indonesian and English).
24 Stevens W, Gous N, Ford N, Scott LE. Feasibility of HIV
point-of-care tests for resource-limited settings: challenges and
solutions. BMC Med 2014; 12: 173.
25 Glass TR, Battegay M, Cavassini M, et al. Longitudinal analysis of
patterns and predictors of changes in self-reported adherence to
antiretroviral therapy: Swiss HIV Cohort Study.
J Acquir Immune Defic Syndr 2010; 54: 197–203.
26 Thompson MA, Mugavero MJ, Amico KR, et al. Guidelines for
improving entry into and retention in care and antiretroviral
adherence for persons with HIV: evidence-based recommendations
from an International Association of Physicians in AIDS Care
panel. Ann Intern Med 2012; 156: 817–33.
27 Fleishman JA, Yehia BR, Moore RD, Korthuis PT, Gebo KA.
Establishment, retention, and loss to follow-up in outpatient HIV
care. J Acquir Immune Defic Syndr 2012; 60: 249–59.
28 De La Mata NL, Ly PS, Nguyen KV, et al. Loss to follow-up trends in
HIV-positive patients receiving antiretroviral treatment in Asia from
2003 to 2013. J Acquir Immune Defic Syndr 2017; 74: 555–62.
29 Haberer JE, Kiwanuka J, Nansera D, Wilson IB, Bangsberg DR.
Challenges in using mobile phones for collection of antiretroviral
therapy adherence data in a resource-limited setting. AIDS Behav
2010; 14: 1294–301.
30 Januraga PP, Mooney-Somers J, Ward PR. Newcomers in a
hazardous environment: a qualitative inquiry into sex worker
vulnerability to HIV in Bali, Indonesia. BMC Public Health 2014;
14: 832.
31 Ministry of Health of Indonesia. Quarterly report on HIV-AIDS
situation in Indonesia, 4th. Jakarta: Ministry of Health of Indonesia,
2017. http://siha.depkes.go.id/portal/files_upload/Laporan_HIV_
AIDS_TW_4_Tahun_2017__1_.pdf (accessed Aug 14, 2018; in
Indonesian).