Original research article

International Journal of STD & AIDS

2019, Vol. 30(1) 20–28
Relationship between self-efficacy and ! The Author(s) 2018

HIV testing uptake among young men Article reuse guidelines:

sagepub.com/journals-permissions
who have sex with men in Myanmar: a DOI: 10.1177/0956462418791945
journals.sagepub.com/home/std
cross-sectional analysis

Minh D Pham1,2 , Poe Poe Aung3, Paul A Agius1,2,4,

Naanki Pasricha1, Swai Mon Oo3, Waimar Tun5,
Ashish Bajracharya6 and Stanley Luchters1,2,7

Abstract
Men who have sex with men (MSM) are disproportionally affected by the HIV epidemic. Self-efficacy is an important
individual psychosocial factor associated with access to and use of health and HIV-related services. We estimated HIV
testing prevalence and examined the relationship between HIV testing self-efficacy and self-reported HIV testing behav-
ior among young MSM (YMSM) in Myanmar. We enrolled 585 MSM aged 18–24 years from six urban areas using
respondent-driven sampling (RDS) technique. RDS analyses were performed to provide estimates for the key outcome
of interest. More than a third (34.5%) had never been tested for HIV, whereas 27.5% and 38.0% had their most recent
HIV test more than three months and within the past three months from the time of interview, respectively. Young MSM
who reported high self-efficacy (adjusted relative risk ratio [ARR]¼7.35, 95%CI ¼ 2.29–23.5) and moderate self-efficacy
(ARR ¼ 8.61, 95%CI ¼ 3.09–24.0) were more likely to report having tested for HIV in the past three months compared
to their counterparts who reported low self-efficacy. Findings highlight a positive association between self-efficacy and
HIV testing uptake, indicating a potential causal relationship. Further research is needed to examine the direction of this
association and inform future public health interventions targeting YMSM in Myanmar.

Keywords
Self-efficacy, HIV testing, men who have sex with men, multinomial logistic regression, Myanmar
Date received: 10 May 2018; accepted: 2 July 2018

Introduction
Compared to the general population, men who have
sex with men (MSM) are at greater risk of HIV infec-
tion in nearly all studied contexts.1–4 Compared to their 1
Burnet Institute, Melbourne, Victoria, Australia
age peers in the general population, young MSM 2
Department of Epidemiology and Preventive Medicine, Faculty of
(YMSM) are more likely to engage in high-risk behav- Medicine Nursing and Health Science, Monash University,
iors such as having condomless anal intercourse and Melbourne, Australia
are more vulnerable to HIV infection.5,6 Published 3
Burnet Institute, Yangon, Myanmar
4
data showed a global ongoing high HIV incidence— Judith Lumley Centre, La Trobe University, Melbourne,Victoria, Australia
5
Population Council, Washington DC, USA
particularly among adolescents and YMSM, with an 6
Population Council, Phnom Penh, Cambodia
overall trend suggesting a greater burden of HIV infec- 7
International Centre for Reproductive Health, Department of
tion in low- and middle-income countries (LMICs).7–11 Obstetrics and Gynecology, Faculty of Medicine and Health Sciences,
It is recommended that MSM, including YMSM, who Ghent University, Belgium
engage in high-risk behaviors should have an HIV test
Corresponding author:
every three months.12 Minh D Pham, Burnet Institute, 85 Commercial Road, Melbourne,
Self-efficacy is defined as individual’s belief in his or Victoria 3004, Australia.
her capacity to execute behavior necessary to produce Email: [email protected]
Pham et al. 21

specific performance attainment.13 Perceived self- between April and June 2015. Details of project inter-
efficacy for specific health behavior is a social cognitive ventions and the evaluation study design, including
factor that has been highlighted in the Health Action study population, data collection, and study measures
Process Approach. This model of health behavior have been reported elsewhere.24 Briefly, potential par-
change stipulates that the adoption and maintenance ticipants who self-identified as YMSM aged 18–24
of health behaviors require two distinct processes: years, living in six urban areas (Pathein,
motivation to change and self-regulation. Within Mawlamyine, Bago, Kalay, Magwe, and Thanlyin) of
these processes, different social cognitive predictors Myanmar and reported having sexual intercourse with
may emerge and self-efficacy appears as the only pre- other men in the past six months were recruited into the
dictor that is equally important in both processes.14 study using respondent-driven-sampling (RDS).14
Previous research has shown that health-related self- A total of 585 MSM were recruited by 33 seed respond-
efficacy contributes to the likelihood of having access ents to participate in face-to-face interviews with a
to health care and is associated with health outcomes: trained MSM data collector using study tools adapted
higher self-efficacy is associated with having multiple from a validated questionnaire for a sexual health
points of connection to the health care system,15 while study among MSM in Kenya.25 The questionnaire
lower self-efficacy is associated with higher burden of was translated to local language by a researcher who
chronic diseases.16 is fluent in both English and Burmese. The final ques-
Recent studies have examined the importance of tionnaire was reviewed by the research team to ensure
psychosocial factors in HIV testing uptake;17 however, clarity and consistency between English and
data on the relationship of self-efficacy and HIV testing Burmese versions.
behavior, particularly from low-resource settings, are In order to examine factors associated with uptake
limited and not consistent. This is partly because dif- and the temporal nature of HIV testing among
ferent measures of self-efficacy had been used among YMSM, we constructed an outcome variable for the
different study populations in different settings.18–20 main outcome of interest “HIV testing status” with
For example, a study among HIV-negative incarcerat- three groups including never tester (never been tested
ed adult men reported that men with high HIV coping for HIV), nonrecent tester (having most recent HIV test
self-efficacy are more likely to be tested for HIV,21 more than three months ago), and recent tester (having
while another study among females using shelter serv- an HIV test in the past three months). Other study
ices for intimate partner violence found that general measures include: (i) main exposure “Self-efficacy for
self-efficacy for handling difficult situations, measured HIV testing” measured using a single-item response on
by 10 items general self-efficacy scale (GSE), has no a five-point Likert scale to the statement “I feel confi-
effects on the acceptance of HIV testing.22 dent that I could get tested for HIV” (1 ¼ strongly dis-
Our previous study showed that self-efficacy was agree; 5 ¼ strongly agree), respondents with responses
correlated with HIV testing uptake among YMSM in 1–3, 4, and 5 were coded as having low, moderate, and
Myanmar, a LMIC with a concentrated HIV epidemic high self-efficacy, respectively; and (ii) potential con-
among key populations.23 In this present analysis, we founders including socio-demographics (sexual identi-
aim to estimate HIV testing prevalence and examine ties: gay- and nongay-identified MSM, age, ethnicity,
the relationship between different levels of HIV testing education, occupation, relationship status), sexual
self-efficacy and self-reported HIV testing behavior of behavior (multiple sexual partners, engaged in transac-
YMSM aged 18–24 years from six urban areas tional sex, reported sexually transmitted infection [STI]
of Myanmar. symptoms), and other psychosocial variables related to
HIV testing (HIV-related knowledge, experienced
Methods social stigma due to sexual orientation, disclosure of
sexual orientation, knowing sexual partners’ HIV
Study design status, having a close relative/friend infected
with HIV).
This is a cross-sectional study using data from the eval-
uation of the Link-Up project, a global HIV and sexual
reproductive health project led by the International
Statistical analysis
HIV/AIDS Alliance, with YMSM in Myanmar. The Descriptive statistics were used to summarize the char-
project implemented a combined community-based acteristics of study participants by outcome of interest
peer education and clinic-based service intervention (HIV testing status). Respondent-driven sampling
to improve the health of YMSM. The interventions (RDS) analyses for outcome of interest using RDS-II
were implemented for six months from October 2014 (Volz-Heckathorn) estimators26 with final reported
and data for the evaluation study were collected sample including seeds27 were performed to provide
22 International Journal of STD & AIDS 30(1)

population estimates28 for coverage of HIV testing never been tested for HIV (never testers) were nongay
among YMSM in Myanmar. There were 12 partici- MSM, while only 43% of participants who had an HIV
pants including one seed who reported being HIV pos- test in the past three months (recent testers) belonged
itive. To keep the sampling structure and recruitment to this sexual identity group. Among the recent testers,
matrix of the RDS sample intact, given a small number 58% reported moderate and 39% reported high self-
of HIV-positive participants, all self-reported HIV-pos- efficacy. For the nonrecent and never testers, the per-
itive participants were included in the analyses. A sen- centages of YMSM who reported moderate and high
sitivity analysis was performed to assess the potential self-efficacy were 54%, 41%, 45%, and 37%,
impact of inclusion of seeds and self-reported HIV-pos- respectively.
itive participants on the final RDS estimates. RDS-population estimates including seed data
Nested multinomial logistic regression models were showed that 38.0% (95%CI: 33.1–43.0) of YMSM in
built to examine the relationship between self-efficacy Myanmar have had an HIV test in the past three
and HIV testing status, with YMSM who had never months, 27.5% (95%CI: 22.9–32.0) had their HIV
tested for HIV treated as the outcome reference test more than three months before the time of inter-
group. Cluster robust standard errors29 were specified view, and 34.5% (95%CI: 29.7–39.5) had never been
to account for clustering of the sample around the tested (Table 2). Sensitivity RDS analysis without seed
seeds. Independent effects for three specific groups of data provided similar point estimates with wider 95%
potential confounding variables—socio-demographic, CIs. Similar results were found with 12 HIV-positive
psychosocial, and sexual behavior factors—were also participants dropped from the analysis.
estimated in multinomial regression modeling. We Bivariate analyses suggested that self-efficacy and all
used univariable-filtering approach for variable selec- other variables except “education” and “reported STI
tion. Only variables that were significantly associated symptoms in the past 12 months” were significantly
with the outcome of interest at p < 0.1 in bivariate associated with HIV testing status (Table 1). Results
analyses were included in multivariate analyses. Post- of multivariate multinomial regression analyses
estimation joint Wald tests were used to assess the sta- (Table 3) showed an association between self-efficacy
tistically significant contribution of covariates on HIV testing status among YMSM. The introduction
across outcomes. of potential confounders into the regression models did
Interactions between self-efficacy and other covari- not attenuate the associations. In the final model, Joint
ates that might influence HIV testing status were Wald tests showed that self-efficacy (p ¼ 0.001), socio-
assessed. More specifically, interactions between self- demographic (p < 0.001), and psychosocial factors
efficacy and sexual identity were assessed to examine (p < 0.001) were independently associated with HIV
whether the effects of self-efficacy on HIV testing status testing but sexual behavior factors were not associated
were moderated by sexual identity. A complete case with HIV testing conditional on the other fac-
approach to missing data was applied. All statistical tors (p ¼ 0.067).
analyses were undertaken using Stata version 13.1 YMSM who reported moderate self-efficacy (adjust-
(STATA Corp, College Station, TX, USA) and Stata ed relative risk ratio [ARR] ¼ 8.61, 95%CI: 3.09–24.0)
user written command package SPost13.30 and high self-efficacy (ARR ¼ 7.35, 95%CI: 2.29–23.5)
were more likely to have tested for HIV in the past
Ethics three months (recent testers) compared to their peers
The study was approved by the Department of Medical who reported low self-efficacy. Similarly, YMSM who
Research in Myanmar, the Population Council reported moderate self-efficacy (ARR ¼ 6.09, 95%CI:
Institutional Review Board in USA, and the Alfred 2.08–17.80) and high self-efficacy (ARR ¼ 6.50, 95%
Hospital Human Research Ethics Committee CI: 2.08–20.4) were more likely to report being a non-
in Australia. recent tester compared to their peers who reported low
self-efficacy.
In further analyses, we introduced an interaction
Results term to explore whether the association between HIV
A total of 585 MSM were enrolled in the study. Table 1 testing and self-efficacy was moderated by sexual iden-
presents the characteristics of the study participants. tity. Although we found the effect of self-efficacy on
The mean age was 20.8 years (SD ¼ 1.9). Ninety-one HIV testing uptake was more pronounced for those
percent of participants self-identified as Burmese ethi- identifying as gay (nonrecent testing: moderate self-
nicity and had completed middle school or higher, 58% efficacy ARRs ¼ 8.3 vs. 6.3, high self-efficacy
identified themselves as nongay MSM, and 42% as gay ARRs ¼ 8.6 vs. 6.1; recent testing: moderate self-
MSM. Eighty-four percent of participants who had efficacy ARRs ¼ 23.1 vs. 4.1, high self-efficacy
Pham et al. 23

Table 1. Characteristics of young MSM aged 18–24 years in six urban areas of Myanmara disaggregated by HIV testing status,
October 2014 to June 2015.
HIV testing statusb
(n/N, %col)
p-Values
Recent tester Nonrecent tester (v2 test of
(tested for HIV (tested for HIV independence
in past three prior to three except
All (n/N, % col) months) months ago) Never tester otherwise noted)

Social-demographics
Age (years) mean, SD 20.8 (1.9) 20.6 (1.9) 21.4 (2.0) 20.4 (1.9) <0.001c
Sexual identity <0.001
Nongay identified 337/585 (57.6) 104/240 (43.3) 86/170 (50.6) 147/175 (84.0)
Gay identified 248/585 (42.4) 136/240 (56.7) 84/170 (49.4) 28/175 (16.0)
Ethnicity 0.004
Non-Burma 50/576 (8.7) 29/238 (12.2) 16/167 (9.6) 5/171 (2.9)
Burma 526/576 (91.3) 209/238 (87.8) 151/167 (90.4) 166/171 (97.1)
Education 0.842
Primary school 54/583 (9.2) 24/239 (10.0) 15/169 (8.9) 15/175 (8.6)
Middle/high school 412/583 (70.7) 170/239 (71.1) 122/169 (72.2) 120/175 (68.6)
College or University 117/583 (20.1) 45/239 (18.9) 32/169 (18.9) 40/175 (22.8)
Employment 0.027
Unemployed 135/584 (23.1) 48/239 (20.1) 34/170 (20.0) 53/175 (30.3)
Employed 449/584 (76.9) 191/239 (79.9) 136/170 (80.0) 122/175 (69.7)
Relationship 0.032
Never in a relationship 486/585 (83.1) 191/240 (79.6) 139/170 (81.8) 156/175 (89.1)
Current/previously in 99/585 (16.9) 49/240 (20.4) 31/170 (18.2) 19/175 (10.9)
a relationship
Psychosocial factors
Self-efficacy for HIV testing <0.001
Low 47/585 (8.0) 7/240 (2.9) 8/170 (4.7) 32/175 (18.3)
Moderate 310/585 (53.0) 140/240 (58.3) 92/170 (54.1) 78/175 (44.6)
High 228/585 (39.0) 93/240 (38.8) 70/170 (41.2) 65/175 (37.1)
Have good HIV-related knowledged 0.007
Yes 217/572 (37.9) 104/238 (43.7) 65/165 (39.4) 48/169 (28.4)
No 355/572 (62.1) 134/238 (56.3) 100/165 (60.6) 121/169 (71.6)
Have close relative/friend infected with HIV <0.001
Yes 311/585 (53.2) 143/240 (59.6) 108/170 (63.5) 60/175 (34.3)
No 274/585 (46.8) 97/240 (40.4) 62/170 (36.5) 115/175 (65.7)
Know sexual partner HIV status 0.003
Yes 138/585 (23.6) 70/240 (29.2) 42/170 (24.7) 26/175 (14.9)
No 447/585 (76.4) 170/240 (70.8) 128/170 (75.3) 149/175 (85.1)
Experienced social stigma due to sexual orientation in the past 12 months 0.016
Yes 139/585 (23.7) 65/240 (27.1) 46/170 (27.1) 28/175 (16.0)
No 446/585 (76.3) 175/240 (72.9) 124/170 (72.9) 147/175 (84.0)
Disclose sexual orientation to others 0.047
Yes 405/585 (69.2) 178/240 (74.2) 117/170 (68.8) 110/175 (62.9)
No 180/585 (30.8) 62/240 (25.8) 53/170 (31.2) 65/175 (37.1)
Sexual behaviors and STI
Had multiple (3) male sexual partners in the past 12 months 0.003
Yes 385/573 (67.2) 172/235 (73.2) 115/167 (68.9) 98/171 (57.3)
No 188/573 (32.8) 63/235 (26.8) 52/167 (31.1) 73/171 (42.7)
Engaged in transactional sex (selling sex to other) in the past 30 days 0.034
Yes 96/585 (16.4) 49/240 (20.4) 28/170 (16.5) 19/175 (10.9)
No 489/585 (83.6) 191/240 (79.6) 142/170 (83.5) 156/175 (89.1)
Reported STI symptoms in the past 12 months 0.188
Yes 106/585 (18.1) 51/240 (21.3) 30/170 (17.6) 25/175 (14.3)
No 479/585 (81.9) 189/249 (78.7) 140/170 (82.4) 150/175 (85.7)
a
Data from all study participants are included in descriptive analysis except otherwise as noted due to missing data.
b
Values are expressed as mean (standard deviation) for age and or n/N (percent) for all other variables.
c
One-way ANOVA.
d
HIV-related knowledge was measured by six common questions on HIV transmission/prevention (e.g. Can a person get HIV from a mosquito bite?),
participants who answer all questions correctly were categorized as having good HIV-related knowledge.
24 International Journal of STD & AIDS 30(1)

Table 2. RDS population estimatesa of HIV testing prevalence among young MSM aged 18–24 years in six urban areas of Myanmar,
October 2014 to June 2015.

Population estimates Population estimates Population estimates

calculated using data calculated excluding calculated excluding HIV
Sample from all participants seed participants positive participants
estimates
HIV testing statusb (N%) % 95%CI % 95%CI % 95%

Recent tester 240 (41.0) 38.0 33.1–43.0 36.6 31.5–42.0 38.6 33.9–43.7
Nonrecent tester 170 (29.1) 27.5 22.9–32.0 25.6 21.0–30.4 26.1 21.7–30.6
Never tester 175 (29.9) 34.5 29.7–39.5 37.8 32.5–43.0 35.3 30.3–40.4
a
Population estimates calculated using RDS II (Volz–Heckathorn) estimators, 95% CIs obtained using bootstrapping method with 1000 replications.
b
Recent tester: Tested for HIV in past 3 months; Nonrecent tester: Tested for HIV prior to 3 months ago; Never tester: Never tested for HIV.

Table 3. Multinomial logistic regression modeling exploring the association between HIV testing statusa and self-efficacy, adjusted for
socio-demographic, psychosocial, and sexual behavior factorsb among young MSM aged 18–24 years in six urban areas of Myanmar,
October 2014 to June 2015 (n ¼ 548).

Adjusted relative risk ratio – ARR (95%CI),d p-values

Recent tester Nonrecent tester Recent tester Joint Wald v2

vs. Never tester vs. Never tester vs. Nonrecent tester (degree of
freedom),
Variablesc ARR (95%CI) p-values ARR (95%CI) p-values ARR (95%CI) p-values p-values

Self-efficacy v2(4) ¼ 19.0,

Moderate self-efficacy 8.61 (3.09–24.0) <0.001 6.09 (2.08–17.8) 0.001 1.41 (0.54–3.69) 0.480 p¼0.001
High self-efficacy 7.35 (2.29–23.5) 0.001 6.50 (2.08–20.4) 0.001 1.13 (0.45–2.81) 0.793
Socio-demographic factors v2(10) ¼ 91.8,
Sexual identity (gay-identified) 4.39 (2.59–7.45) <0.001 3.07 (1.96–4.81) <0.001 1.43 (0.91–2.26) 0.125 p<0.001
Ethnicity (non-Burma) 4.18 (1.36–12.9) 0.013 3.34 (0.91–12.2) 0.069 1.26 (0.61–2.58) 0.536
Employment 1.89 (1.01–3.58) 0.048 1.30 (0.63–2.69) 0.484 1.46 (0.81–2.63) 0.204
status (employed)
Currently or previously 1.53 (0.89–2.63) 0.121 1.11 (0.58–2.10) 0.761 1.39 (0.84–2.30) 0.205
in a relationship
Age 0.91 (0.81–1.02) 0.109 1.19 (1.06–1.33) 0.004 0.76 (0.68–0.86) <0.001
Psychosocial factors v2(8) ¼ 49.0,
Have good 1.81 (1.15–2.87) 0.011 1.59 (0.99–2.57) 0.056 1.14 (0.80–1.63) 0.472 p<0.001
HIV-related knowledge
Have close relative/friend 2.20 (1.51–3.21) <0.001 2.75 (1.90–3.97) <0.001 0.80 (0.53–1.21) 0.294
infected with HIV or died
of AIDS
Know partner HIV status 1.56 (0.87–2.79) 0.140 1.13 (0.72–1.79) 0.596 1.37 (0.86–2.19) 0.182
Experienced social stigma 0.95 (0.49–1.84) 0.877 1.07 (0.46–2.47) 0.881 0.89 (0.57–1.39) 0.609
due to sexual orientation
in the past 12 months
Sexual behavior factors v2(4) ¼ 8.8,
Had multiple (3) male 1.83 (1.05–3.19) 0.034 1.45 (0.73–2.88) 0.289 1.26 (0.70–2.27) 0.442 p¼0.067
sexual partners in the
past 12 months
Engaged in transactional sex 1.33 (0.68–2.60) 0.402 1.27 (0.59–2.73) 0.536 1.05 (0.64–1.69) 0.856
(selling sex to others) in
the past 30 days
a
Recent tester: Tested for HIV in past three months; Nonrecent tester: Tested for HIV prior to three months ago; Never tester: Never tested for HIV.
b
Potential confounders to be included in the multivariate multinomial regression model if found significantly associated with outcome at p < 0.1 in
bivariate analyses.
c
Reference group for self-efficacy is low-self-efficacy, that for sexual identity is nongay identified, that for ethnicity is Burma ethnic, and that for
employment status and currently or previously in a relationship is unemployed and never in a relationship, respectively. Psychosocials and sexual
behaviors variables are binary (Yes/No) and “No” are reference groups.
d
Robust standard errors to account for clustering by seeding.
Pham et al. 25

ARRs ¼ 14.4 vs. 4.3); jointly these moderated effects that self-efficacy to protect oneself from exposure to
were not statistically significant (p ¼ 0.312). risks of HIV infection was not associated with HIV
testing in the past 12 months.36 Studies among MSM
in Australia,33 Italy,13 and Hong Kong31 have shown
Discussion
that self-efficacy was positively associated with self-
Findings from this study shows that HIV testing self- reported HIV testing. A study among young people
efficacy positively correlates with HIV testing behavior aged 18–24 years in South Africa found that HIV
of YMSM aged 18–24 years in Myanmar. This finding risk reduction self-efficacy was not a determinant of
is in line with findings from other studies among MSM HIV testing,37 while another study among sexually
conducted in high resource settings showing that self- active young people aged 15–24 years in Thailand
efficacy was positively associated with acceptance and reported that self-efficacy for HIV testing was associ-
frequency of HIV testing among adult MSM.18,31–33 It ated with ever being tested for HIV.38 These findings,
is noted that in these studies, HIV testing self-efficacy although useful and indicative, cannot be generalized
was discussed; however, the measurements of self- to YMSM populations in low-resource settings.
efficacy were varied across these studies with authors In this study, we found no clear evidence on the
using either a single-item statement to measure partic- difference between the impact of high and moderate
ipant’s confidence in obtaining an HIV test if self-efficacy on HIV testing uptake. However, as differ-
desired31,32 or an opportunistically constructed scale ent specific self-efficacy beliefs and measurements were
with five or eight items to measure the extent to employed and there is a paucity of available data on the
which participant’s thought that having an HIV test relationship between self-efficacy and HIV testing
was under their control if they choose to.18,33 among key populations in low-resource settings, con-
Traditionally, the use of single-item measures of solidation of evidence for guiding public health inter-
cognitive variables has been criticized due to concern ventions is a challenge. Therefore, future research may
regarding their psychometric properties. However, cur- benefit from using standardized, validated self-efficacy
rent research has shown that single-item measures scales39 to measure self-efficacy as a global construct
of latent constructs such as self-efficacy, compared to underlying a basic belief in one’s capabilities to exercise
well-established multi-item measures, has equal or even control over their own functioning in various challeng-
superior predictive utility in predicting health-related ing situations rather than self-beliefs related to specific
outcomes. A single-item measure of self-efficacy was behaviors or learnt behaviors. There is also a need to
found consistently correlated positively with a well- develop and validate a standardized self-efficacy scale
established 20-item measure and consistently predicted specific to HIV testing for research examining the rela-
relapse to substance use in a sample of substance use tionship between this psychosocial factor and HIV test-
disorder treatment-seeking young adults in the United ing behavior. This may help to provide needed
States.34 In another study among female university stu- information and facilitate the synthesis of evidence to
dents in the UK, researchers found that a single-item draw conclusions on effects of self-efficacy on HIV test-
measure of self-efficacy was a significant predictor of ing and inform the design of public health interventions
well-being as was the validated 10-item measure of gen- to improve HIV testing uptake among target
eral self-efficacy.35 These findings suggest that using populations.
single psychometric measures may offer a valid Results of our study support the argument that the
approach to investigating overall well-being as well inclusion of seed data in RDS analyses can provide
as other health-seeking behavior and health- unbiased estimates. We show that when using RDS
related outcomes. estimator II, the inclusion of seed data in the final anal-
Previous research in low-resource settings has exam- ysis can produce accurate population estimates. We
ined the correlations between self-efficacy related to estimate that only 38% of YMSM in Myanmar have
specific behaviors with HIV testing practice among been tested for HIV in the past three months while
the general population and the results were mixed. more than a third (34.5%) has never been tested.
A community survey in Namibia20 reported that great- This finding raises an alarm for the national health
er self-efficacy for HIV prevention behaviors (ability to policy makers in Myanmar where the majority of
enact behaviors preventing HIV infection) was associ- newly-diagnosed HIV-infected cases were found
ated with lifetime HIV testing. Similarly, one study among young key populations, particularly YMSM
found that HIV prevention self-efficacy was positively aged less than 25 years who experience an HIV infec-
associated with previous HIV testing among the gener- tion rate five times higher than that of their peers in the
al adolescent and adult population in eight urban dis- general population.40 This is also an indication that the
tricts of Malawi.19 However, another community- reported HIV prevalence of 6.6% among MSM in
based study in 11 southern districts of Malawi reported Myanmar41 may be underestimated and innovative
26 International Journal of STD & AIDS 30(1)

approaches to improve HIV testing coverage among and inform public health interventions targeting
this marginalized population are urgently needed if YMSM in Myanmar.
the first 90 of the 90-90-90 target (90% of HIV infected
people are diagnosed) is to be achieved. Recent studies Acknowledgements
have shown that new approaches to HIV testing such The authors would like to acknowledge the International
as home-based or HIV self-testing could be effective HIV/AIDS Alliance and Marie Stopes International for
and efficient in reaching more first time testers and their kind collaboration and support, and the Dutch
promoting frequent testing among MSM who were Ministry of Foreign Affairs (BUZA) for their financial sup-
unreached by conventional testing methods.42–45 Such port of the Link-Up project, which aims to improve the
approaches may well be worth considering for sexual and reproductive health and rights (SRHR) of one
Myanmar in the years to come. million young people affected by HIV across five countries
Our study findings add to the evidence that self- in Africa and Asia. The project was implemented by a con-
efficacy for HIV testing is positively associated with sortium of partners led by the International HIV/AIDS
HIV testing behavior of YMSM who are at higher Alliance. The Population Council, in partnership with
risk of HIV infection but have limited access to HIV Burnet Institute implemented this research in Myanmar.
Our special thanks go to participants involved in the study
services they need.46,47 The positive association found
and the screeners from the study sites. We also acknowledge
in this study may indicate a potential causal relation-
the data collection team from Myanmar Business Coalitions
ship between self-efficacy and HIV testing and public
on AIDS (MBCA) and the Burnet Institute Myanmar
health interventions to improve self-efficacy may help research team. The authors gratefully acknowledge the con-
to promote HIV testing behaviors among YMSM in tribution to this work of the Victorian Operational
Myanmar. Our findings warrant further research to Infrastructure Support Program received by the Burnet
determine the effects of individual psychosocial factors, Institute. Stanley Luchters is a recipient of National Health
particularly self-efficacy on HIV testing practice among and Medical Research Council of Australia (NHMRC)
high-risk populations in LMICs. This is an important Career Development Fellowship. Minh D Pham received sup-
area of research because improving HIV testing cover- port via an International Postgraduate Research Scholarship
age and increasing the proportion of people at elevated (IPRS) from the Commonwealth of Australia and the
risk of HIV infection who know their HIV status is a Victorian International Research Scholarship (VIRS) from
key strategy to reach the 90-90-90 targets in countries State Government of Victoria, Australia.
with concentrated HIV epidemics among key
populations.48 Declaration of conflicting interests
The cross-sectional design of this study does not The authors declared no potential conflicts of interest with
allow us to determine the impact of self-efficacy on respect to the research, authorship, and/or publication of
HIV testing behavior. It limits our ability to generalize this article.
the findings and draw conclusive causal relationship
between self-efficacy and HIV testing among YMSM Funding
in Myanmar. The direction of this relationship and its
The authors disclosed receipt of the following financial sup-
implications for public health interventions are sub-
jected to further examinations in future studies. Our port for the research, authorship, and/or publication of this
data are self-reported and therefore can suffer from article: This work was supported by the Government of the
recall and social desirability bias. The measure of Netherlands’ Ministry of Foreign Affairs through its Sexual
self-efficacy using a single-item statement specifically and Reproductive Health and Rights (SRHR) Fund through
related to HIV testing behavior does not allow us to a subcontract from the International HIV/AIDS Alliance
make direct quantitative comparisons with findings under the Link-Up project.
from other studies. Nevertheless, this study provides
much-needed data related to HIV testing behavior of ORCID iD
a young population at high risk of HIV infection in Minh D Pham http://orcid.org/0000-0002-5932-3491
Myanmar. It contributes to the pool of evidence on
self-efficacy and HIV testing behavior of YMSM and References
offers directions for future HIV research in Myanmar 1. Baral S, Sifakis F, Cleghorn F, et al. Elevated risk for
as well as other countries in similar contexts. HIV infection among men who have sex with men in low-
In conclusion, we found that more than a third of and middle-income countries 2000-2006: A systematic
young MSM aged 18–24 years in Myanmar have never review. PLoS Med 2007; 4: e339.
tested for HIV. HIV testing self-efficacy positively cor- 2. Beyrer C, Baral SD, van Griensven F, et al. Global epi-
relates with HIV testing behavior and future research is demiology of HIV infection in men who have sex with
needed to determine the direction of this association men. Lancet 2012; 380: 367–377.
Pham et al. 27

3. Beyrer C, Baral SD, Collins C, et al. The global response systematic review of global studies. AIDS Behav 2016;
to HIV in men who have sex with men. Lancet 2016; 20: 880–918.
388: 198–206. 18. Adam PC, de Wit JB, Bourne CP, et al. Promoting reg-
4. Beyrer C, Sullivan P, Sanchez J, et al. The increase ular testing: an examination of HIV and STI testing rou-
in global HIV epidemics in MSM. AIDS 2013; tines and associated socio-demographic, behavioral and
27: 2665–2678. social-cognitive factors among men who have sex with
5. Salomon EA, Mimiaga MJ, Husnik MJ, et al. Depressive men in New South Wales, Australia. AIDS Behav 2014;
symptoms, utilization of mental health care, substance 18: 921–932.
use and sexual risk among young men who have sex 19. Berendes S and Rimal RN. Addressing the slow uptake
with men in EXPLORE: Implications for age-specific of HIV testing in Malawi: The role of stigma, self-
interventions. AIDS Behav 2009; 13: 811–821. efficacy, and knowledge in the Malawi BRIDGE
6. WHO. HIV and young men who have sex with men. Geneva, Project. J Assoc Nurses Aids Care 2011; 22: 215–228.
Switzerland: WHO, 2015. http://www.who.int/hiv/pub/tool 20. Creel AH and Rimal RN. Factors related to HIV-testing
kits/hiv-young-msm/en/ (accessed 3 May 2017). behavior and interest in testing in Namibia. AIDS Care
7. McCormack S, Dunn DT, Desai M, et al. Pre-exposure 2011; 23: 901–907.
prophylaxis to prevent the acquisition of HIV-1 infection 21. Andrinopoulos K, Kerrigan D, Figueroa JP, et al. HIV
(PROUD): Effectiveness results from the pilot phase of a coping self-efficacy: A key to understanding stigma and
pragmatic open-label randomised trial. Lancet 2016; HIV test acceptance among incarcerated men in Jamaica.
387: 53–60. AIDS Care 2010; 22: 339–347.
8. Liu G, Lu H, Wang J, et al. Incidence of HIV and syph- 22. Ratcliff TM, Zlotnick C, Cu-Uvin S, et al. Acceptance of
ilis among men who have sex with men (MSM) in Beijing: HIV antibody testing among women in domestic violence
An Open Cohort Study. PLoS One 2015; 10: e0138232. shelters. J HIV/AIDS Soc Serv 2012; 11: 291–304.
9. Wang Y, Huang Y, Chen H, et al. Incidence and corre- 23. Geibel S, King’ola N, Temmerman M, et al. The impact
lates of HIV and syphilis in a prospective cohort of men of peer outreach on HIV knowledge and prevention
who have sex with men in Mianyang, China, over a behaviours of male sex workers in Mombasa, Kenya.
36-month period. Sex Health 2015; 12: 546–555. Sex Transm Infect 2012; 88: 357–362.
10. Price MA, Rida W, Mwangome M, et al. Identifying at- 24. Aung PP, Ryan C, Bajracharya A, et al. Effectiveness of
risk populations in Kenya and South Africa: HIV inci- an integrated community- and clinic-based intervention
dence in cohorts of men who report sex with men, sex on HIV testing, HIV knowledge, and sexual risk behavior
workers, and youth. J Acquir Immune Defic Syndr 2012; of young men who have sex with men in Myanmar.
59: 185–193. J Adolesc Health 2017; 60: S45–S53.
11. van Griensven F, Holtz TH, Thienkrua W, et al. 25. van Griensven F and de Lind van Wijngaarden JW.
Temporal trends in HIV-1 incidence and risk behaviours A review of the epidemiology of HIV infection and pre-
in men who have sex with men in Bangkok, Thailand, vention responses among MSM in Asia. AIDS 2010;
2006-13: An observational study. Lancet HIV 2015; 24: S30–S40.
2: e64–e70. 26. Volz E and Heckathorn D. Probability based estimation
12. Wilcock B and Frommer M. HIV testing among gay men theory for respondent driven sampling. J Off Stat 2008;
and other men who have sex with men. Discussion paper 24: 79–97.
Australian Federation of AIDS Organizations. https:// 27. Johnston LG, Hakim AJ, Dittrich S, et al. A systematic
wwwafaoorgau/publication/afao-briefing-paper-hiv-test review of published respondent-driven sampling surveys
ing-among-gay-men-men-sex-men/ (2014, accessed 19 collecting behavioral and biologic data. AIDS Behav
January 2017). 2016; 20: 1754–1776.
13. Prati G, Breveglieri M, Lelleri R, et al. Psychosocial cor- 28. Schonlau M and Liebau E. Respondent driven sampling.
relates of HIV testing among men who have sex with men Stata J 2012; 12: 72–93.
in Italy: A cross-sectional study. Int J STD AIDS 2014; 29. Williams RL. A note on robust variance estimation
25: 496–503. for cluster-correlated data. Biometrics 2000; 56: 645–646.
14. Salganik MJ and Heckathorn DD. Sampling estimation 30. Long Scott J and Freesee J. Regression models for cate-
in hidden populations using respondent-driven sampling. gorical dependent variables using stata. 3rd ed. Texas,
Sociol Methodol 2004; 34: 193–239. USA: Stata Press, 2014.
15. Thompson T, Mitchell JA, Johnson-Lawrence V, et al. 31. Gu J, Lau JT and Tsui H. Psychological factors in asso-
Self-rated health and health care access associated with ciation with uptake of voluntary counselling and testing
African American men’s health self-efficacy. Am J Men’s for HIV among men who have sex with men in Hong
Health 2017; 5: 1385–1387. Kong. Public Health 2011; 125: 275–282.
16. Finney Rutten LJ, Hesse BW, St Sauver JL, et al. Health 32. Song Y, Li X, Zhang L, et al. HIV-testing behavior
self-efficacy among populations with multiple chronic among young migrant men who have sex with
conditions: The value of patient-centered communica- men (MSM) in Beijing, China. AIDS Care 2011;
tion. Adv Ther 2016; 33: 1440–1451. 23: 179–186.
17. Evangeli M, Pady K and Wroe AL. Which psychological 33. Jamil MS, Guy RJ, Bavinton BR, et al. HIV testing self-
factors are related to HIV testing? A quantitative efficacy is associated with higher HIV testing frequency
28 International Journal of STD & AIDS 30(1)

and perceived likelihood to self-test among gay and bisex- 41. NAP. Global AIDS response progress report Myanmar,
ual men. Sexual Health 2017; 2: 170–178. Jan 2014-Dec 2014. Naypyitaw: National AIDS
34. Hoeppner BB, Kelly JF, Urbanoski KA, et al. Program, Ministry of Health, Myanmar, 2015. http://
Comparative utility of a single-item vs. multiple-item www.unaids.org/en/dataanalysis/knowyourresponse/
measure of self-efficacy in predicting relapse among countryprogressreports/2015countries (accessed 31
young adults. J Subst Abuse Treat 2011; 41: 305–312. March 2017).
35. Williams GM and Smith AP. Using single-item measures 42. Johnson C, Baggaley R, Forsythe S, et al. Realizing the
to examine the relationships between work, personality, potential for HIV self-testing. AIDS Behav 2014;
and well-being in the workplace. Psychology 2016; 18: S391–S395.
7: 753–767. 43. Yan H, Yang H, Raymond HF, et al. Experiences and
36. Kaufman MR, Rimal RN, Carrasco M, et al. Using correlates of HIV self-testing among men who have sex
social and behavior change communication to increase with men in Jiangsu province, China. AIDS Behav 2015;
HIV testing and condom use: The Malawi BRIDGE 19: 485–491.
Project. AIDS Care 2014; 26: S46–S49. 44. Ferrand R, Ford N and Kranzer K. Maximising the ben-
37. Peltzer K and Matseke G. Determinants of HIV testing efits of home-based HIV testing. Lancet HIV 2015;
among young people aged 18–24 years in South Africa. 2: e4–e5.
Afr Health Sci 2013; 13: 1012–1020. 45. Qin Y, Liu F, Tang W, et al. HIV self-testing among
38. Musumari PM, Tangmunkongvorakul A, high-risk men who have sex with men in China: A
Srithanaviboonchai K, et al. Prevalence and correlates cross-sectional study. Lancet 2016; 388: S76.
of HIV testing among young people enrolled in non- 46. Adam PC, de Wit JB, Toskin I, et al. Estimating levels of
formal education centers in Urban Chiang Mai, HIV testing, HIV prevention coverage, HIV knowledge,
Thailand: A cross-sectional study. PLoS One 2016; and condom use among men who have sex with men
11: e0153452. (MSM) in low-income and middle-income countries.
39. Schwarzer R and Jerusalem M. Generalized self-efficacy J Acquir Immune Defic Syndr 2009; 52: S143–S151.
scale. In: J Weinman, S Wright and M Johnston (eds) 47. UNAIDS. National HIV Legal Review Report: Review
Measures in health psychology: A user’s portfolio causal of Myanmar’s legal framework and its affect on access to
and control beliefs. Windsor, UK: NFER-NELSON, health and HIV service for people living with HIV and
1995, pp. 35–37. key affected populations. Yangon, Myanmar, 2014.
40. UNFPA, UNESCO and WHO. Sexual and reproductive 48. UNAIDS. 90-90-90 An ambitious treatment target to
health of young people in Asia and the Pacific: A review of help end the AIDS epidemic. Geneva, Switzerland.
issues, policies and programmes. Bangkok: UNFPA, 2015. Joint United Nations Programme on HIV/AIDS, 2014.
http://unesdoc.unesco.org/images/0024/002435/243566E. http://www.unaids.org/en/resources/documents/2014/90-
pdf (accessed 31 March 2017). 90-90 (accessed 31 March 2017).