Global Health Action

ISSN: 1654-9716 (Print) 1654-9880 (Online) Journal homepage: http://www.tandfonline.com/loi/zgha20

Time to anti-retroviral therapy among people

living with HIV enrolled into care in Myanmar: how
prepared are we for ‘test and treat’?

Kyaw Zin Linn, Hemant Deepak Shewade, Kyaw Ko Ko Htet, Thae Maung
Maung, San Hone & Htun Nyunt Oo

To cite this article: Kyaw Zin Linn, Hemant Deepak Shewade, Kyaw Ko Ko Htet, Thae Maung
Maung, San Hone & Htun Nyunt Oo (2018) Time to anti-retroviral therapy among people living with
HIV enrolled into care in Myanmar: how prepared are we for ‘test and treat’?, Global Health Action,
11:1, 1520473, DOI: 10.1080/16549716.2018.1520473

To link to this article: https://doi.org/10.1080/16549716.2018.1520473

© 2018 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group.

View supplementary material

Published online: 30 Nov 2018.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=zgha20
 GLOBAL HEALTH ACTION
2018, VOL. 11, 1520473
https://doi.org/10.1080/16549716.2018.1520473

ORIGINAL ARTICLE

Time to anti-retroviral therapy among people living with HIV enrolled into
care in Myanmar: how prepared are we for ‘test and treat’?
Kyaw Zin Linna, Hemant Deepak Shewade b,c
, Kyaw Ko Ko Htetd, Thae Maung Maung e
, San Hone*a
and Htun Nyunt Oo*a
a
National AIDS Programme, Department of Public Health, Ministry of Health and Sports, Nay Pyi Taw, Myanmar; bInternational Union
Against Tuberculosis and Lung Disease (The Union), South-East Asia Office, New Delhi, India; cInternational Union Against Tuberculosis
and Lung Disease (The Union), Paris, France; dDepartment of Medical Research, Pyin Oo Lwin branch, Pyin Oo Lwin, Myanmar;
e
Department of Medical Research, Yangon Head Quarter, Myanmar

ABSTRACT ARTICLE HISTORY

Background: Among people living with HIV (PLHIV) enrolled into care, time to anti-retroviral Received 25 June 2018
therapy (ART) has not been studied in Myanmar. To inform progress, we conducted this Accepted 3 September 2018
operational research among treatment-naive PLHIV (≥18 years) enrolled during a period of
RESPONSIBLE EDITOR
three years (2014–2016) at Pyin Oo Lwin, Myanmar.
Stig Wall, Umeå University,
Objectives: To determine (i) the time from HIV diagnosis to ART initiation (time to ART) and Sweden
associated factors and (ii) the association between time to ART and attrition (loss to follow-up
and death) from ART care. KEYWORDS
Methods: This was a retrospective cohort study involving a record review of secondary ART initiation; time to ART;
programme data. The PLHIV were followed up to 5 December 2017 for ART initiation and attrition; HIV care cascade;
up to 31 March 2018 (date of censoring) for attrition during ART. SORT IT; operational
research
Results: Of 543 enrolled, 373 (69%) were found to be eligible and initiated on ART. Of 373,
245 (67%) were initiated within 6 weeks of enrolment. The median enrolment delay (from
diagnosis) was 4 (IQR: 1, 14) days and median ART initiation delay (from ART eligibility) was 20
(IQR: 13, 36) days. The median time to ART (excluding the time interval in pre-ART care) was
29 (IQR: 18, 55) days and was significantly long in those with prevalent TB and CD4 count ≥
500/mm3 at enrolment. Among 373, the annual incidence density of attrition was 12.8% (0.95
CI: 10.2, 15.7). Attrition was common in first 100 days. Time to ART (after excluding time
interval in pre-ART care) was not significantly associated with attrition.
Conclusion: The programme appears to be on track to initiate ART as soon as possible in a
‘test and treat’ scenario (implemented since September 2017) subject to interventions to
reduce ART initiation delay.

Background care); (iii) eligibility to ART initiation; (iv) retention

while on ART [4]. The corresponding possible delays
Globally, the new human immunodeficiency virus
are: HIV diagnosis delay, enrolment delay, delays
(HIV) infections and acquired immunodeficiency
during eligibility assessment and pre-ART care, and
syndrome (AIDS)-related deaths are declining.
ART initiation delay (Figure 1). Time to ART (from
However, in 2016 there were 1.8 million people
HIV diagnosis) has been recommended as an indica-
newly infected with HIV and 1 million AIDS-related
tor to supplement existing HIV cascade indica-
deaths. Of 36.7 million people living with HIV
tors [4,5].
(PLHIV), 19.5 million are on anti-retroviral therapy
Studies globally have reported a long time to enter
(ART) [1]. With the aim of ending the HIV epidemic
into care and/or initiate ART after HIV diagnosis [5–
by 2030, the Joint United Nations Programme on
11]. The enrolment and ART initiation delay
HIV/AIDS in 2013 set an ambitious 90-90-90 target:
(Figure 1) can be due to underlying factors like client
(i) 90% of PLHIV know their status, (ii) 90% of
readiness/refusal, physical disability and physician
PLHIV who know their status receive ART, (iii)
concerns to defer ART initiation during the treat-
90% of PLHIV on ART have suppressed viral loads
ment of co-morbidities especially anti-TB treatment
by 2020 [2,3].
[6,8,10].
The HIV care cascade before ‘test and treat’
Myanmar is one of the 35 countries that account
included four stages: (i) from HIV diagnosis to enrol-
for 90% of new HIV infections globally and one of
ment; (ii) enrolment to ART eligibility (pre-ART
the 6 countries in Asia identified as a priority for the

CONTACT Kyaw Zin Linn [email protected] National AIDS Programme, Department of Public Health, Ministry of Health and Sports, Nay
Pyi Taw, Myanmar
*
These authors contributed equally to this work
Supplemental data for this article can be accessed here.
© 2018 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits
unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
2 K. Z. LINN ET AL.

Figure 1. Conceptual framework of delays in HIV care cascade, Myanmar, 2014–2016*.

HIV = human immunodeficiency virus, ART = anti-retroviral therapy
*Eligibility assessment delay and time spent in pre-ART care will not be applicable in a ‘test and treat’ setting (implemented since September 2017).

global fast-track strategy to end the epidemic by 2030. Methods

The HIV epidemic is concentrated among key popu-
Study design and participants
lations and the prevalence in the adult population is
0.6% (2016) [12]. Between 2014 and 2016, Myanmar This was a retrospective cohort study involving a record
has done a commendable job in increasing the access review of secondary programme data. We included all
to ART (from 85,600 to 130,000), which is one of the PLHIV enrolled into care in a Pyin Oo Lwin ART centre
factors that has resulted in reductions in annual new (under NAP) between 1 January 2014 and 31 December
infections (12,000 to 11,000) and AIDS-related deaths 2016. PLHIV < 18 years of age and pregnant/post-partum
(9700 to 7800) [12,13]. women (at enrolment) were excluded. PLHIV re-
Studies from Myanmar have documented low reten- enrolled into care after a previous episode of loss to
tion in pre-ART care and high retention once on ART follow-up (either during pre-ART care or during ART
[14–16]. PLHIV initiated on ART had low CD4 counts care) and PLHIV transferred in (either during pre-ART
at baseline indicating delays before ART initiation [14– or ART care) from other ART centres were also excluded.
16]. Hence, there was a need for systematic assessment
of time to ART and associated factors which has not
Setting
been studied in the National AIDS Programme (NAP)
setting. Globally, the association of the time to ART General setting
with attrition has not been studied. Myanmar is situated in south-east Asia and has an
With the World Health Organization (WHO)’s estimated population of 51 million, of whom 70% live
‘test and treat’ strategy, which Myanmar currently in rural areas. It has 1 union territory and 14 states and
implements (since September 2017), the delay during regions. They are administratively divided into districts
eligibility assessment and pre-ART care (enrolment and further into 412 townships/sub-townships [21].
to ART eligibility) becomes irrelevant as all enrolled The NAP is one of the vertical programmes under the
PLHIV are eligible for ART [17,18]. A new HIV care Disease Control Division, Department of Public Health,
cascade has therefore been recommended where the Ministry of Health and Sports. Myanmar has 124 ART
stage from enrolment to eligibility has been removed centres and 173 decentralized sites for ART services.
and an additional stage of early retention in ART care Pyin Oo Lwin district is one of the seven districts
introduced [19]. Therefore, time to ART will now in the Mandalay region. It has a combination of plain
include enrolment delay and ART initiation delay. and hilly terrain. It has 5 townships and a population
As Myanmar gears up to implement the ‘test and of 255,000. Under NAP, it has two ART centres (one
treat’ strategy, reduction in enrolment and ART at Pyin Oo Lwin and another at Mogoke) and one
initiation delays is key to ensure a short time to decentralized site. The ART centre at Pyin Oo Lwin is
ART after HIV diagnosis [20]. located at a district-level referral hospital.
Hence, to inform progress we conducted this
operational research during a period of three Care of PLHIV under NAP (2014–2016)
years (2014–2016) before implementation of ‘test A flow chart describing the care of PLHIV is shown in
and treat’. We aimed to determine (i) the time Figure 2. A screening test for HIV, a confirmation test
from HIV diagnosis to ART initiation (time to (‘three test’ policy) and enrolment was done at a decen-
ART) and associated factors, and (ii) the associa- tralized site. Eligibility assessment and treatment initia-
tion between time to ART and attrition (loss to tion were done at an ART centre. Decentralized sites
follow-up and death) from ART care in Pyin Oo initiated ART only in key populations and TB-HIV co-
Lwin, Myanmar. infected patients. After enrolment, PLHIV were assessed
 GLOBAL HEALTH ACTION 3

Figure 2. Flow chart on process of ART care for people living with HIV in ART clinics of the National AIDS Programme, Myanmar
(2014–2016)*†.
HIV = human immunodeficiency virus, ART = anti-retroviral therapy; AIDS = acquired immunodeficiency syndrome; HBV = hepatitis B virus;
HCV = hepatitis C virus; TB = tuberculosis; WHO = World Health Organization
*Box in pink colour is relevant during ‘test and treat’ strategy (implemented since September 2017). Box in white colour will not be relevant
after introduction of ‘test and treat’ strategy.†Enrolled during 2014–2016 and followed up for ART initiation up to 5 December 2017; followed
up until 31 March 2018 for attrition.

for eligibility for ART, which was the presence of any one TB/HIV co-infection, if the patient was not on anti-
of the following: (i) WHO clinical staging 3 or 4; (ii) CD4 TB treatment, anti-TB treatment was initiated first,
count below the threshold (≤ 350 cell/mm3 before July followed by ART within two to eight weeks of start-
2014, ≤ 500 cell/mm3 from July 2014); (iii) special ing anti-TB treatment. For those initiated on ART,
populations (TB or Hepatitis B co-infection, sero-discor- information regarding CD4 count and WHO staging
dant couples, key populations, HIV-positive pregnant at the time of medical eligibility and date of medical
women). eligibility were recorded in a pre-ART register (black
PLHIV not assessed for eligibility, not eligible for book), ART register (red book) and ART card (white
ART and not started on ART despite eligibility within card). Patient visited the ART centre or the decen-
six weeks of enrolment were considered as ‘PLHIV tralized site on a monthly basis. Information regard-
under pre-ART care’. The PLHIV who were assessed ing ART care was recorded in an ART register (red
and not yet eligible for ART were usually clinically book) and ART card (white card). Though patients
healthy and followed up at three-monthly intervals. were screened for TB at enrolment and follow-up
They received multivitamins, drugs for prophylaxis of visits (sputum examination and chest radiograph if
opportunistic infections, and counselling for psychoso- positive on symptom screen), isoniazid preventive
cial support. Pre-ART outcomes (ART initiation, death, therapy was not implemented during the study per-
loss to follow-up, transfer out, still in pre-ART care) iod. All the services were provided free of cost [22].
were recorded in the pre-ART register (black book).
If a patient was eligible for ART (at enrolment or
Data variables, sources of data and data
during pre-ART care), baseline investigations along
collection
with counselling sessions (three times over two
weeks) were prescribed and a follow-up date was The PLHIV were followed up to 5 December 2017 for
given after two weeks. During this two-week period, ART initiation and up to 31 March 2018 (date of
cotrimoxazole preventive therapy (CPT) was started censoring) for attrition during ART.
(if eligible: CD4 count < 350 per mm3 and/or WHO The following variables were collected at enrol-
stage 3 or 4). Since the most common initial side ment from the pre-ART register: date of enrolment,
effect of cotrimoxazole and anti-retroviral therapy is date of HIV confirmation test, age, sex, risk factors
a rash, it was recommended to start CPT first and to for HIV infection, entry point to ART centre, literacy
initiate ART two weeks later if the individual does status, employment status, CPT/anti-TB treatment
not develop a rash with cotrimoxazole. In cases of date. The time period until six weeks after the date
4 K. Z. LINN ET AL.

of enrolment was taken as the enrolment period (‘at estimates (forward stepwise method). Unadjusted
enrolment’). If initiated on ART, the following vari- and adjusted prevalence ratios (aPR) with 0.95 con-
ables were additionally collected from the pre-ART fidence interval (CI) were used to summarize (infer)
register, ART register and white card: date of ART, the association.
date of medical eligibility, CD4 count at eligibility Cumulative rates (%) and incidence density (per
and WHO staging at eligibility. 100 person-years/per year [%]) of attrition (among
Treatment outcomes (alive and still on treatment, ART cohort) were calculated. Survival analysis was
transfer out, death and loss to follow-up) along with undertaken to analyse the hazards of attrition after
the dates of outcome were collected. Attrition was ART initiation. The risk time of PLHIV in ART care
defined as death and loss to follow-up combined. started from their date of ART initiation and ended
The date of censoring was the date of outcome for with the date of attrition or censoring, whichever
PLHIV still on treatment. For death and loss to occurred earlier. Cox regression model was used to
follow-up, more than 90 days after the last visit was determine the association between various delays
taken as the date of outcome. For transfer out, the last (one model for each delay) and attrition in ART
visit date was considered. care after adjusting for confounders. Proportional
hazards assumption was confirmed by using ‘estat
phtest’ function in STATA. Unadjusted and adjusted
Data management and statistics
hazard ratios (0.95 CI) were used to summarize
Data were single-entered into Excel (Microsoft, (infer) this association.
Redmond, WA, USA) and analysed using STATA
(version 12.1 STATA Corp., College Station,
TX, USA). Results

Derived variables Background and clinical characteristics

Year of enrolment was derived from the date of A total of 543 PLHIV were enrolled. Table 1 shows
enrolment. PLHIV initiated on ART were classified their background and clinical characteristics. Of 543,
as initiated within six weeks of enrolment or after six 319 (59%) were male, 194 (36%) belonged to the age
weeks of enrolment. Based on the date of enrolment group 25–34 years and 318 (59%) were heterosexuals.
and date of initiation of TB treatment, we classified At enrolment, TB was prevalent in 52 (10%) people
TB as prevalent (on TB treatment at enrolment or and 289 (53%) were on CPT. Among 52 with pre-
initiated within six weeks of enrolment) or incident valent TB, 27 were already on anti-TB treatment at
(initiated TB treatment after six weeks of enrolment) the date of enrolment and 25 were started on TB
among the enrolled cohort. Similarly, CPT status was treatment within 6 weeks of the date of enrolment.
determined at enrolment. Of 491 without prevalent TB, 6 (1.2%) developed
Enrolment delay was calculated from the date of incident TB post-enrolment.
HIV confirmation to date of enrolment. Delay in Out of 543, 373 (69%) were found to be eligible
eligibility and/or time period in pre-ART care was and initiated on ART and 245 (45%) were initiated at
calculated among the ART cohort (date of medical enrolment. Baseline characteristics of the 373 PLHIV
eligibility was recorded only for those initiated on initiated on ART are given in Table 2. Of 373, 78
ART) from the date of enrolment to date of eligibility. (21%) had WHO clinical stage of more than 2 and the
In addition, ART initiation delay (from date of elig- median CD4 count at was 206 (IQR: 109, 332)
ibility to date of ART initiation) and time to ART per mm3.
(from date of diagnosis to date of ART initiation)
were also calculated among the ART cohort
(Figure 1).
Time to ART: magnitude and risk factors
Data analysis Within 7 days of diagnosis, 26 (8%) PLHIV were
Baseline characteristics of the enrolled and ART initiated on ART. The median enrolment delay and
cohort were summarized using mean (standard ART initiation delay were 4 days (IQR: 1 to 14 days)
deviation – SD), median (inter-quartile range – and 20 days (IQR: 13 to 36 days), respectively. The
IQR), frequency and proportions. Among the median time to ART was 41 days (IQR: 25 to
enrolled, ART initiation was summarized using fre- 92 days). Assuming ‘test and treat’ during the study
quency and proportions. Delay in days was summar- period, we calculated the median time to ART after
ized using median (IQR). excluding the delay in eligibility and/or time interval
Risk factors associated with various types of delays in pre-ART care which was 29 days (IQR: 18 to
(median was used as cut-off) were assessed using 55 days) (Table 3). This was nothing but enrolment
modified Poisson regression with robust variance delay and ART initiation delay combined.
 GLOBAL HEALTH ACTION 5

Table 1. Socio-demographic and clinical characteristics of

people living with HIV enrolled at a public sector ART centre Table 2. Socio-demographic and clinical characteristics of
in Pyin Oo Lwin, Myanmar (2014–2016) (N = 543). people living with HIV initiated on ART at a public sector
ART centre in Pyin Oo Lwin, Myanmar (2014–2016)
Variables N (%)
(N = 373)†.
Total 543 (100.0)
Sex Variables n (%)
Male 319 (58.7) Total 373 (100.0)
Female 224 (41.3) Sex
Age in years Male 212 (56.8)
18–24 48 (8.8) Female 161 (43.2)
25–34 194 (35.7) Age in years
35–44 191 (35.2) 18–24 28 (7.5)
≥ 45 110 (20.3) 25–34 122 (32.7)
Mean (SD) 36.7 (9.7) 35–44 137 (36.7)
Risk factors ≥ 45 86 (23.1)
Heterosexual 318 (58.6) Mean (SD) 37.6 (9.7)
MSM 3 (0.6) Risk factors
FSW 7 (1.3) Heterosexual 269 (72.1)
IDU 61 (11.2) MSM 2 (0.5)
Blood transfusion 12 (2.2) FSW 4 (1.1)
Mother to child 1 (0.2) IDU 48 (12.9)
Missing 141 (26.0) Blood transfusion 10 (2.7)
Literate Mother to child 1 (0.3)
No 76 (14.0) Missing 39 (10.5)
Yes 377 (69.4) Literate
Missing 90 (16.6) No 62 (16.6)
Employed Yes 309 (82.8)
No 125 (23.0) Missing 2 (0.6)
Yes 331 (61.0) Employed
Missing 87 (16.0) No 103 (27.6)
Entry point Yes 268 (71.8)
VCT 180 (33.1) Missing 2 (0.6)
STI 1 (0.2) Entry point
TB clinic 2 (0.4) VCT 125 (33.5)
Outpatient 158 (29.1) STI 1 (0.3)
Inpatient 67 (12.3) TB clinic 2 (0.5)
Private 24 (4.4) Outpatient 145 (38.9)
Self-referred 18 (3.3) Inpatient 58 (15.6)
Drug Treatment Unit 3 (0.6) Private 22 (5.9)
Missing 90 (16.6) Self-referred 18 (4.8)
CPT at enrolment* Drug Treatment Unit 1 (0.3)
No 254 (46.8) Missing 1 (0.3)
Yes 289 (53.2) WHO stage
TB at enrolment* Stage 1 155 (41.6)
No 491 (90.4) Stage 2 75 (20.1)
Yes 52 (9.6) Stage 3 61 (16.4)
Year of enrolment Stage 4 17 (4.6)
2014 170 (31.3) Missing 65 (17.4)
2015 171 (31.5) CD4 count
2016 202 (37.2) <50 24 (6.4)
HIV = human immunodeficiency virus, ART = anti-retroviral therapy, 50–99 53 (14.2)
MSM = male who has sex with men, FSW = female sex worker, 100–199 99 (26.5)
IDU = intravenous drug users, STI = sexually transmitted infection, 200–349 104 (27.9)
TB = tuberculosis, VCT = voluntary counselling and testing, 350–499 48 (12.9)
CPT = cotrimoxazole preventive therapy ≥ 500 35 (9.4)
*Until six weeks from date of enrolment. Missing 10 (2.7)
Median (IQR) 206 (109, 332)
Year of enrolment
2014 93 (24.9)
2015 127 (34.1)
No baseline factors were significantly associated 2016 153 (41.0)
with enrolment delay and ART initiation delay (data HIV = human immunodeficiency virus, ART = anti-retroviral therapy,
not shown). Prevalent TB at enrolment and CD4 MSM = male who has sex with men, FSW = female sex worker,
IDU = intravenous drug users, STI = sexually transmitted infection,
count at ART eligibility were significantly associated TB = tuberculosis, VCT = voluntary counselling and testing,
with a long time to ART (≥ 30 days, after excluding WHO = World Health Organization; CPT = cotrimoxazole preventive
the delay in eligibility and/or time interval in pre- therapy
†
Enrolled during 2014–2016 and followed up for ART initiation up to 5
ART care). When compared to PLHIV without pre- December 2017.
valent TB, those with prevalent TB had a 42% higher *Until six weeks from date of enrolment.
chance of delay (aPR [0.95 CI]: 1.42 [1.13, 1.79]).
When compared to PLHIV with a CD4 count less
than 50, those with a count ≥ 500 had a 59% higher Other factors like year and literacy were not signifi-
chance of delay (aPR [0.95 CI]: 1.59 [1.00, 2.52]). cantly associated with delay (Table 4).
6 K. Z. LINN ET AL.

Table 3. Time taken (days) for ART initiation from HIV diag- Table 5. Outcomes among people living with HIV initiated on
nosis among people living with HIV enrolled at a public ART at a public sector ART centre in Pyin Oo Lwin, Myanmar
sector ART centre in Pyin Oo Lwin, Myanmar (2014–2016). (2014–2016) (N = 373)†.
Time taken (days) Initiated at Initiated after
in care cascade Eligible enrolment* enrolment Overall
from HIV and Outcomes n (%) n (%) n (%)
diagnosis assessed* Median (IQR) Range
Total 245 (100) 128 (100) 373 (100)
Enrolment delay 472 4 (1, 14) 0, 1105 Still on treatment 151 (61.6) 88 (66.7) 239 (64.1)
(a) Transferred out 36 (14.7) 13 (10.2) 49 (13.1)
Eligibility delay 373 4.5 (0, 13.3) 0, 559 Death** 27 (11.0) 14 (10.9) 41 (11.0)
and/or time Loss to follow-up** 31 (12.7) 13 (10.2) 44 (11.8)
interval in pre-
ART care (b) HIV = human immunodeficiency virus, ART = anti-retroviral therapy
†
ART initiation 373 20 (13, 36) 0, 673 Enrolled during 2014–2016 and followed up for ART initiation up to 5
delay (c) December 2017; followed up until 31 March 2018 for attrition.
Time to ART 330 39 (24, 74) 0, 1105 *Until six weeks from date of enrolment.
(a + b + c) **Included under attrition.
Time to ART after 330 29 (18, 55) 0, 1105
excluding the
delay in Table 6. Cumulative incidence of attrition rate among people
eligibility and/
or time interval
living with HIV initiated on ART at a public sector ART centre
in pre-ART care in Pyin Oo Lwin, Myanmar (2014–2016) (N = 373)†.
(a + c) Person- Incidence
HIV = human immunodeficiency virus, ART = anti-retroviral therapy; days of density
IQR = inter-quartile range follow- per year
*543 were enrolled and 373 were initiated on ART as on 5 December Cohort up Attrition^ (%) (0.95 CI)
2017. Those who were eligible and dates were available were included. Total 237,628 83 12.8 (10.2, 15.7)
0–100 days 33,545 43 46.7 (34.7, 63.1)
101–200 days 30,011 8 9.9 (4.7, 19.3)
> 200 days 174,072 32 6.6 (4.8, 9.5)
Table 4. Factors associated with long time to ART* (≥
30 days) after excluding the time interval in pre-ART care HIV = human immunodeficiency virus, CI = confidence interval
†
Enrolled during 2014–2016 and followed up for ART initiation up to 5
among people living with HIV initiated on ART at a public December 2017; followed up until 31 March 2018 for attrition.
sector ART centre in Pyin Oo Lwin, Myanmar (2014–2016) ^
Includes loss to follow-up and death.
(N = 373)†.
Crude PR Adjusted PR#
Variables (0.95 CI) (0.95 CI) Attrition while on ART: magnitude and risk
Entry point at enrolment factors
VCT Ref Ref
STI/TB/Drug Treatment Unit -^ -^ The outcomes of PLHIV stratified by ART initia-
Outpatient 0.87 (0.71, 1.08) 0.86 (0.70, 1.06)
Inpatient 0.94 (0.72, 1.22) 0.89 (0.69, 1.16)
tion within six weeks of enrolment and after six
Private/self-referred 0.70 (0.47, 1.03) 0.70 (0.48, 1.03) weeks of enrolment are provided in Table 5.
Missing -^ -^ Among 373, cumulative attrition was seen in 85
Prevalent TB at enrolment
No Ref Ref (23%) which was contributed by 41 (11%) deaths
Yes 1.39 (1.12, 1.71)** 1.42 (1.13, 1.79)** and 44 (11.8%) loss to follow-up. The annual inci-
CD4 count (microlitre) at ART
start dence density of attrition was 12.8% (0.95 CI: 10.2,
<50 Ref Ref 15.7). Attrition was common in first 100 days of
50–99 1.06 (0.66, 1.70) 1.22 (0.77, 1.93)
100–199 1.11 (0.72, 1.72) 1.25 (0.83, 1.91) ART. (Table 6 and Figure 3)
200–349 1.15 (0.75, 1.78) 1.34 (0.87–2.05) After adjusting for potential confounders (CD4
350–499 1.13 (0.70, 1.80) 1.36 (0.85, 2.19)
≥ 500 1.31 (0.82, 2.10) 1.59 (1.00, 2.52)**
count, WHO staging, literacy and employment status,
Missing -^ -^ entry point, HIV risk factors, year, prevalent TB, CPT
Constant - 0.46 (0.30, 0.70) use at enrolment, age and sex), the association of var-
HIV = human immunodeficiency virus, ART = anti-retroviral therapy, ious delays before ART and attrition have been sum-
STI = sexually transmitted infection, TB = tuberculosis,
VCT = voluntary counselling and testing, PR = prevalence ratio, marized in Table 7. The time to ART (after excluding
aPR = adjusted prevalence ratio, CI = confidence interval the delay in eligibility and/or time interval in pre-ART
†
Enrolled during 2014–2016 and followed up for ART initiation up to 5
December 2017. care) was not significantly associated with attrition.
*Combination of enrolment delay and ART initiation delay. We categor- However, long enrolment delay (≥ 5 days) was signifi-
ized the delay based on the median value.
#
Modified Poisson regression with robust variance estimates (forward cantly associated with less attrition.
stepwise method); other variables at enrolment (age, sex, HIV risk
factor, literacy, employment status, cotrimoxazole preventive therapy
status, year of enrolment) and at ART eligibility (WHO clinical staging)
were considered in the model. However, they were excluded by the Discussion
model as they did not significantly improve the model prediction.
^
Limited sample (n ≤ 10) in this sub-category. This is the first study from Myanmar looking at time
**p < 0.05 to ART from HIV diagnosis. The major strength of
 GLOBAL HEALTH ACTION 7

Figure 3. Kaplan Meier survival curve^ among people living with HIV initiated on ART* at a public sector ART centre in Pyin Oo
Lwin, Myanmar (2014–2016): overall and stratified by time to ART after excluding the delay in eligibility and/or time interval in
pre-ART care*†.
HIV = human immunodeficiency virus, ART = anti-retroviral therapyLog rank test P value = 0.325, ^Event of interest is attrition which includes
loss to follow-up and death.*Combination of enrolment delay and ART initiation delay, categorized based on the median value.†Enrolled during
2014–2016 and followed up for ART initiation up to 5 December 2017; followed up until 31 March 2018 for attrition.

this study is that we documented the entire HIV care ART-related services (both before and after ART
cascade from enrolment to treatment outcome and uptake) even in contexts where these services are
attempted to study time to ART in a hypothetical ‘test freely available [6,23]. The ‘Test and treat’ strategy
and treat’ scenario. We also made an attempt to that has been implemented in Myanmar eliminates
assess the association between pre-ART delays and the need for pre-ART care [18]; however, the
ART treatment outcomes. There were some key barriers to accessing free ART was beyond the
findings. scope of this study.
First, two-thirds of enrolled PLHIV were eligi- Second, the time to ART (after excluding the delay
ble and initiated on ART during the follow-up in eligibility and/or time interval in pre-ART care)
period and half were initiated on ART within six was around a month and was largely contributed by
weeks of enrolment. It was related to a policy of ART initiation delay. Less than 1 in 10 were initiated
assessment for eligibility for ART initiation during on ART within a week of diagnosis. The time to ART
the study period. Though we do not know the was lower than reported elsewhere globally and simi-
status of those not on ART, a previous study in lar to a study from Malawi (for the year 2011) [5–11].
Myanmar has detected alarmingly high attrition However, this has to be interpreted with caution as
during pre-ART care [15]. There is evidence that the eligibility criteria might vary based on the study
PLHIV often encountered barriers in accessing site and period.
8 K. Z. LINN ET AL.

Table 7. Confounder adjusted association between various many PLHIV with a high CD4 count will be eligible
delays before ART initiation and attrition (death and loss to for ART. This requires attention by the programme.
follow-up) among people living with HIV initiated on ART at Fourth, the annual attrition rate was higher than
public sector ART centre in Pyin Oo Lwin, Myanmar (2014–
2016) (N = 373)†. that reported previously in a large cohort of PLHIV
Delay type* Attrition** HR (0.95 CI) aHR (0.95 CI)# managed by NAP in collaboration with a non-gov-
Enrolment ernment organization (12.8% versus 8.8%). High
delay (a) attrition in the initial 100 days was similar to this
Yes 7.4 0.64 (0.41, 0.99) 0.60 (0.37, 0.99)^
No 12.1 Ref Ref previous study where attrition was significantly
Eligibility delay and time on higher in the initial 180 days [14]. Time to ART
pre-ART care (b)
Yes 12.4 0.93 (0.60, 1.43) 0.72 (0.42, 1.24)
(after excluding the delay in eligibility and/or time
No 13.1 Ref Ref interval in pre-ART care) was not associated with
ART initiation attrition. Enrolment delay (≥ 5 days) was a protec-
delay (c)
Yes 12.0 0.88 (0.57, 1.35) 0.71 (0.42, 1.20) tive factor for attrition. This should be interpreted
No 13.5 Ref Ref with caution. This analysis was limited by the
Time to ART
(a + b + c) exclusion of factors like socio-economic status,
Yes 10.6 0.72 (0.47, 1.12) 0.72 (0.44, 1.17) marital status, sero-discordant status, alcohol/sub-
No 15.3 Ref Ref
Time to ART after excluding stance abuse, ART regimen, distance of residence
the delay in eligibility and/or from ART centre, baseline weight and patient’s
time interval in pre-ART care
(a + c)
knowledge and attitude from the adjusted analysis.
Yes 11.7 0.81 (0.52, 1.24) 0.67 (0.40, 1.13) These were not collected.
No 14.2 Ref Ref Finally, in addition to not collecting an exhaustive
HIV = human immunodeficiency virus, HR = hazard ratio; aHR = adjusted list of confounders, there were other limitations as
hazard ratio; CI = confidence interval
†
Enrolled during 2014–2016 and followed up for ART initiation up to 5 well. We were not able to describe the pre-ART care
December 2017; followed up until 31 March 2018 for attrition. cascade (how many were assessed at enrolment; of
*Median delay was used to categorize each delay type.
**Incidence density per year (%)/incidence density per 100 person-years, those assessed, how many were eligible; and of those
#
attrition includes death and loss to follow-up. eligible, how many were initiated on ART) including
Cox regression (enter method); adjusted for CD4 count, WHO staging,
literacy and employment status, entry point, HIV risk factors, year, the pre-ART outcomes. Details regarding eligibility
^
prevalent TB, CPT use at enrolment, age and sex. (staging and CD4 count) were recorded only if a
p < 0.05
PLHIV was initiated on ART. Therefore, we did not
have information on CD4 counts and WHO staging
at enrolment. The programme needs to improve
ART initiation delay was probably related to the recording (especially CD4 counts and WHO staging
waiting time for completing two weeks of CPT (CPT at enrolment) as it has moved on to the ‘test and
at enrolment was 53%) and two to eight weeks of treat’ strategy.
anti-TB treatment (10% had prevalent TB at enrol-
ment). Prevalent TB at enrolment was also an inde-
pendent predictor of long time to ART (after
Future research
excluding the delay in eligibility and/or time interval
in pre-ART care). With ‘test and treat’, PLHIV with a Fox MP et al. in their systematic review (studies from
higher CD4 count at enrolment will become eligible 2008 to February 2015) found that the evidence on
for ART, reducing the number of people on CPT and interventions in sub-Saharan Africa to improve rate
with prevalent TB at enrolment significantly. This and time of ART uptake was limited and of mixed
will help reduce the ART initiation delay. However, quality. They recommended more implementation
all PLHIV after eligibility also received counselling research to improve uptake in a manner that is effi-
sessions for two weeks before ART initiation [22]. cient for providers and effective for PLHIV without
Under ‘test and treat’, WHO recommends ART as affecting treatment outcomes [24]. The findings of
soon as possible if the patient provides consent, and this study provide the evidence base for more imple-
this issue of two weeks’ counselling needs to be mentation research to reduce time to ART in a ‘test
addressed by the programme [17]. and treat’ setting.
Third, a high CD4 count has been previously asso- This study was limited to one ART centre in the
ciated with delay in enrolment and long time to ART Mandalay region; one of the better-performing cen-
[7,9]. The long time to ART may be explained by tres. Similar studies are urgently recommended in
delay in becoming eligible for ART. In our study, a other part of Myanmar before a policy decision can
high CD4 count was associated with long time to be made. Qualitative studies are recommended in
ART after excluding the delay in eligibility and/or Myanmar to explore the provider and patient per-
time interval in pre-ART care. With ‘test and treat’, spectives into barriers to early ART uptake.
 GLOBAL HEALTH ACTION 9

Conclusion Ethics and consent

We studied the time to ART among PLHIV in a The study was approved by the Ethics Review Committee
three-year period before the ‘test and treat’ of the Department of Medical Research, Ministry of Health
and Sports, Myanmar, and the Ethics Advisory Group of
approach was implemented in Myanmar. This pro- International Union Against Tuberculosis and Lung
vides necessary baseline information to support the Disease (The Union), Paris, France. Since the study
national programme to reduce time to ART during involved review of existing records, the waiver of individual
the implementation of ‘test and treat’. The NAP informed consent was sought and approved by the ethics
appears to be on track to implement ‘test and committees..
treat’ subject to interventions to reduce the ART
initiation delay by addressing the issue of ART
counselling which itself induces a two-to-three- Funding information
week delay and addressing the high-risk group of The author(s) received no specific funding for the conduct
‘people with high CD4 counts’. We recommend of this study. This study was conducted in operational
including routine indicators within the programme settings using resources available within programme.
to measure linkage to care (enrolment) among all However, the training programme, within which this
paper was developed, and the open access publication
diagnosed, ART uptake among those enrolled (both
costs were funded by the Department for International
proportion and time) and early ART outcomes (in Development (DFID), UK and La Fondation Veuve Emile
first 100 days) as per the recently recommended Metz-Tesch (Luxembourg). The funders had no role in
updates in the HIV care cascade [19]. study design, data collection and analysis, decision to pub-
lish, or preparation of the manuscript.

Acknowledgments
Paper context
This research was conducted through the Structured In Myanmar, time to anti-retroviral therapy from HIV
Operational Research and Training Initiative (SORT IT), diagnosis has not been studied. In the anti-retroviral ther-
a global partnership led by the Special Programme for apy centre of Pyin Oo Lwin, the median time to anti-
Research and Training in Tropical Diseases at the World retroviral therapy after excluding the period in pre-anti-
Health Organization (WHO/TDR). The model is based on retroviral therapy care was 29 days and largely contributed
a course developed jointly by the International Union by delay in ART initiation after eligibility. As Myanmar has
Against Tuberculosis and Lung Disease (The Union) and begun implementing ‘test and treat’, this calls for interven-
Medécins sans Frontières (MSF/Doctors Without Borders). tions to reduce delay in starting anti-retroviral therapy
The specific SORT IT programme which resulted in this after enrolment.
publication was jointly organized and implemented by the
Centre for Operational Research, The Union, Paris, France;
the Department of Medical Research, Ministry of Health
and Sports, Myanmar; the Department of Public Health, Data availability statement
Ministry of Health and Sports, Myanmar; The Union The dataset used in this study (STATA data file) along with
Country Office, Mandalay, Myanmar; The Union South- the programme file used for analysis (STATA do file) and
East Asia Office, New Delhi, India; and the Burnet codebook have been provided as a supplementary annex (S
Institute, Australia. We are grateful to the programme Annex).
staff under the National AIDS Programme, Myanmar, in
Pyin Oo Lwin who are responsible for maintaining the
records. ORCID
Hemant Deepak Shewade http://orcid.org/0000-0002-
Author contributions 8242-1911
Thae Maung Maung http://orcid.org/0000-0002-1265-
KZL was the principal investigator; HDS, KKKH, TMM 3813
were the SORT IT mentors; SH and HNO are the senior
authors. KZL, KKKH, HDS and TMM were involved in
data collection and entry; KZL, HDS, KKKH analysed the
data and prepared the first draft of the paper. All authors References
were involved in the conception, design, inference of
[1] UNAIDS Data 2017. UNAIDS. Geneva (Switzerland):
results, providing critical review to the manuscript and
UNAIDS; 2017.
approval of the manuscript from this protocol.
[2] UNAIDS. 90-90-90An ambitious treatment target to
help end the AIDS epidemic. Geneva (Switzerland):
UNAIDS; 2014.
Disclosure statement [3] Word Health Organization (WHO). Health in 2015
from Millennium Development Goals (MDG) to
No potential conflict of interest was reported by the Sustainable Development Goals (SDG). Geneva
authors. (Switzerland): WHO; 2015.
10 K. Z. LINN ET AL.

[4] World Health Organization. Regional Office for the [15] Minn Oo M, Gupta V, Aung TK, et al. Alarming
Eastern Mediterranean. Nasr city, Cairo, Egypt: HIV attrition rates among HIV-infected individuals in
test–treat–retain cascade analysis: Guide and tools; 2014. pre-antiretroviral therapy care in Myanmar,
[5] Medland NA, Chow EPF, McMahon JH, et al. Time 2011–2014. Glob Health Action. 2016;9:31280.
from HIV diagnosis to commencement of antiretro- [16] Sabapathy K, Ford N, Chan KN, et al. Treatment
viral therapy as an indicator to supplement the HIV outcomes from the largest antiretroviral treatment
cascade: dramatic fall from 2011 to 2015. PLoS One. program in Myanmar (Burma): a cohort analysis of
2017;12:e0177634. retention after scale-up. J Acquir Immune Defic
[6] Hoehn N, Gill MJ, Krentz HB. Understanding the Syndr. 2012;60:e53–62.
delay in starting antiretroviral therapy despite recent [17] World Health Organization (WHO). Treat all people
guidelines for HIV patients retained in care. AIDS living with HIV, offer antiretrovirals as additional
Care. 2017;29:564–569. prevention choice for people at “substantial” risk
[7] Sloan DJ, van Oosterhout JJ, Malisita K, et al. [Internet]. 2015 [cited 2018 May 22]. Available from:
Evidence of improving antiretroviral therapy treat- http://www.who.int/mediacentre/news/releases/2015/
ment delays: an analysis of eight years of programma- hiv-treat-all-recommendation/en/.
tic outcomes in Blantyre, Malawi. BMC Public Health. [18] National AIDS Program; Department of Public
2013;13:490. Health; Ministry of Health and Sports. Guidelines for
[8] Robertson M, Wei SC, Beer L, et al. Delayed entry into the clinical management of HIV infection in
HIV medical care in a nationally representative sam- Myanmar. Nay Pyi Taw, Myanmar: National AIDS
ple of HIV-infected adults receiving medical care in Program; 2017.
the USA. AIDS Care. 2016;28:325–333. [19] Fox MP, Rosen S. A new cascade of HIV care for the
[9] Takah NF, Awungafac G, Aminde LN, et al. Delayed era of “treat all.”. PLOS Med. 2017;14:e1002268.
entry into HIV care after diagnosis in two specialized [20] National AIDS Programme; Department of Public
care and treatment centres in Cameroon: the influence Health; Ministry of Health and Sports. Myanmar
of CD4 count and WHO staging. BMC Public Health. national stragegic plan on HIV and AIDS (2016-
2016;16:529. 2020). Nay Pyi Taw, Myanmar: National AIDS
[10] Khoury Z, Silva RS, Villela W. Factors associated with Program; 2016.
a delay in seeking HIV/AIDS treatment in São Paulo, [21] Myanmar Information Management Unit; Ministry
Brazil. AIDS Behav. 2015;19:679–683. of Immigration and Population; Myanmar. The
[11] Sobrino-Vegas P, Rodríguez-Urrego J, Berenguer J, 2014 Myanmar population and housing census
et al. Educational gradient in HIV diagnosis delay, [Internet]. Nay Pyi Taw, Myanmar; 2014 [cited
mortality, antiretroviral treatment initiation and 2018 May 22]. Available from: http://themimu.info/
response in a country with universal health care. census-data.
Antivir Ther. 2012;17:1–8. [22] National AIDS Program; Department of Public
[12] HIV and AIDS data hub for Asia-pacific region. Health; Ministry of Health and Sports. Guidelines for
Evidence to action.HIV and AIDS data hub for Asia- the Clinical Management of HIV infection in
pacific region. Evidence to action. Country profiles: Myanmar. Nay Pyi Taw, Myanmar: National AIDS
myanmar. Key facts on HIV [Internet]. 2017 [cited Program; 2014.
2018 May 22]. Available from: http://www.aidsdata [23] Ankomah A, Ganle JK, Lartey MY, et al. ART access-
hub.org/Country-Profiles/Myanmar. related barriers faced by HIV-positive persons linked
[13] UNAIDS. County profile: myanmar [Internet]. 2017 to care in southern Ghana: a mixed method study.
[cited 2018 May 22]. Available from: http://www. BMC Infect Dis. 2016;16:738.
unaids.org/en/regionscountries/countries/myanmar. [24] Fox MP, Rosen S, Geldsetzer P, et al. Interventions to
[14] Thida A, Tun STT, Zaw SKK, et al. Retention and risk improve the rate or timing of initiation of antiretro-
factors for attrition in a large public health ART viral therapy for HIV in sub-Saharan Africa: meta-
program in Myanmar: a retrospective cohort analysis. analyses of effectiveness. J Int AIDS Soc.
PLoS One. 2014;9:e108615. 2016;19:20888.