Driving factors of retention in care among HIV-positive MSM and transwomen in Indonesia: A cross-sectional study

Little is known about the prevalence of and factors that influence retention in HIV-related care among Indonesian men who have sex with men (MSM) and transgender women (transwomen, or waria in Indonesian term). Therefore, we explored the driving factors of retention in care among HIV-positive MSM and waria in Indonesia. This cross-sectional study involved 298 self-reported HIV-positive MSM (n = 165) and waria (n = 133). Participants were recruited using targeted sampling and interviewed using a structured questionnaire. We applied a four-step model building process using multivariable logistic regression to examine how sociodemographic, predisposing, enabling, and reinforcing factors were associated with retention in care. Overall, 78.5% of participants were linked to HIV care within 3 months after diagnosis or earlier, and 66.4% were adequately retained in care (at least one health care visit every three months once a person is diagnosed with HIV). Being on antiretroviral therapy (adjusted odds ratio [AOR] = 6.00; 95% confidence interval [CI]: 2.93–12.3), using the Internet to find HIV-related information (AOR = 2.15; 95% CI: 1.00–4.59), and having medical insurance (AOR = 2.84; 95% CI: 1.27–6.34) were associated with adequate retention in care. Involvement with an HIV-related organization was associated negatively with retention in care (AOR = 0.47; 95% CI: 0.24–0.95). Future interventions should increase health insurance coverage and utilize the Internet to help MSM and waria to remain in HIV-related care, thereby assisting them in achieving viral suppression.


Introduction
Men who have sex with men (MSM) and transgender women (transwomen) are disproportionately burdened by HIV infection worldwide. Compared to the general adult population, PLOS

Study design and population
We analyzed cross-sectional baseline data from a study conducted by the Asia Pacific AIDS Positive Network (APN+), a regional network of people living with HIV from 11 countries in Asia and the Pacific. Data were collected by trained interviewers from 1,655 Indonesian people living with HIV from December 2012 to February 2013. In this paper we focus on the MSM and waria subsample (n = 298; 165 MSM and 133 waria). Participants were administered a validated questionnaire developed by the APN+. The original questionnaire was in English, and translated to Bahasa Indonesia. The questionnaire was then back-translated into English to ensure consistency and quality. Participants were recruited using a targeted sampling method [30], a technique that has been widely used for sampling hard-to-reach groups. First, at least one person from each highrisk group with HIV (including MSM and waria) was selected to be a seed at each study site. These people were identified either through the contact of a local support group for people living with HIV or through a local community-based organization. Each seed was surveyed and then asked to recruit another participant to the study. Then this participant was surveyed and asked to recruit another participant to the study. This process was repeated until the desired sample size at each study site was reached.
Inclusion of the participants was based on the following criteria: 1) an HIV-positive individual residing in the study sites; 2) aged between 18-50 years; and 3) self-reported diagnosed with HIV infection at least three months prior to the date of interview. Participants were informed about the study objectives and the procedures prior to being surveyed. Those who voluntarily signed the informed consent form were then given the interviewer-administered survey. Study procedures were approved by the Research Ethics Board of Atma Jaya University Indonesia, Jakarta.

Measures
The primary outcome measure in our analysis was retention in HIV care, which was measured using the following question: "After being HIV-positive, how frequently did you visit your doctor/nurse/health worker?" Response options included: once a week; once a month; once every 2-3 months; once every 4-6 months; only every 7-12 months; or only when I am sick). We dichotomized participants into having adequate and inadequate retention in care. We defined adequate retention as at least one health care visit every three months once a person is diagnosed with HIV. This definition is in accordance with guidelines and prior studies [8,10,31]. As additional information, we also measured linkage to care using the following question: "After diagnosis, how long did it take you to meet the doctor/nurse/health worker?" Response options consisted of: right after diagnosis, same day; number of years/months/days after diagnosis; or not visited yet.
To explore the determinants of retention in HIV care, we analyzed the impacts of three groups of influencing factors: predisposing, enabling, and reinforcing factors based on the PRECEDE-PROCEED model [29]. Predisposing factors included HIV treatment literacy, which was defined as the level of understanding on all aspects of ART, including types of ART drugs, ART side effects, treatment adherence, HIV drug resistance and other related topics. To measure this, an overall literacy score was applied using 25 "True" or "False" questions (Cronbach's alpha = 0.90; higher scores indicate greater knowledge). The remaining predisposing factors were: disease history (e.g., in the past 6 months, did you suffer from any disease/health problem? yes versus no); ever diagnosed with TB (have you ever been diagnosed with TB after you were HIV+? yes versus no); alcohol drinking (do you currently drink alcohol? yes versus no); illicit drug use (have you ever used any illicit drugs? Yes versus no); smoking (do you smoke? yes versus no); and unsafe sex in the past 6 months (in the past 6 months, have you had sex with your spouse/someone other than your spouse? yes versus no); how frequently did you use a condom when you had sex with him/her?).
Enabling factors included having medical insurance (are you enrolled in any kind of health insurance program? yes versus no); and being a member of/affiliated to an HIV-related organization (are you a member of/affiliated to any HIV-related organization? yes versus no); internet use for HIV information (have you ever used internet to find HIV-related information? yes versus no), and amount of internet use (in the past 7 days, how many hours did you spend on the internet?).
Reinforcing factors included ART status (are you taking ART/HIV medicines now? yes versus no); disclosure to steady partner (have you ever disclosed your HIV+ status to your spouse? yes versus no); disclosure to individuals beyond family & steady partner (have you ever disclosed your HIV+ status to anyone except your spouse, a close family member, and your doctor? yes versus no); stigma & discrimination experience (in the last 12 months, how often have you been excluded from social events or activities; or been verbally insulted, harassed and/or threatened; or been physically assaulted; or been denied health services because of your HIV status?); and social support, which was measured using a validated 12 items scale [32]. These items addressed instrumental and emotional social support from family, friends, and significant other (e.g. "my family really tries to help me"; "I can count on my friends when things go wrong"; and "there is a special person with whom I can share my joys and sorrows"). Responses ranged from "strongly disagree" to "strongly agree" on a five-point scale (Cronbach's alpha = 0.90; higher scores indicate greater social support). Sociodemographic characteristics were also measured. This included risk group (MSM or waria), age, education, income, and place of residence.

Analyses
We analyzed data in Stata version 14.0 (StataCorp, College Station, Texas). We first examined differences between MSM and waria in sociodemographic characteristics, linkage to care, and retention in care, as well as predisposing, enabling, reinforcing factors using chi-square tests for categorical variables and independent t-test or Mann-Whitney U tests (when data were not normally distributed) for continuous variables. We then examined the associations of sociodemographic characteristics as well as predisposing, enabling, reinforcing factors with our primary outcome, retention in HIV care using univariate logistic regression models. When univariate p-values were less than 0.20, we included these factors in multivariable models. We fit multivariable logistic regression models to identify factors associated with retention in care. We used a four-step model building process, based on the PRECEDE-PROCEED Model [29]. We added factors cumulatively: first, we added sociodemographic characteristics (Model 1), followed by predisposing factors (Model 2), enabling factors (Model 3), and reinforcing factors (Model 4). Assumptions for logistic regression were met (i.e., non-collinearity, linearity in the logit, no outliers, independence of observations). Table 1 describes the sociodemographic characteristics of the sample. Compared to waria participants, MSM participants were significantly younger, were better educated, had a higher monthly income, and were less likely to have ever engaged in sex work. Table 2 indicates that the majority of respondents (78.5%) were linked to HIV care within 3 months after receiving their HIV diagnosis, and this did not differ between MSM and waria. Almost two-thirds (66.4%) of respondents had adequate retention in health care (i.e., they visited a healthcare facility at least once every three months since being diagnosed with HIV). Retention in HIV care was higher among MSM than waria participants (72.7% vs 58.7%, respectively, p < 0.05). Table 2 also shows the prevalence of predisposing, enabling, and reinforcing factors, as well as differences in these factors between MSM and waria. Regarding predisposing factors, 67.8% of participants were diagnosed with HIV more than 12 months prior to the survey, 6.4% reported having health problems in the past six months, 23.2% had unprotected sex in the past six months. Compared to waria, MSM were less likely to use alcohol, use illicit drugs, smoke, and ever have TB. MSM had a higher HIV-treatment literacy than waria. Regarding enabling factors, only 24.8% had medical insurance. Compared to waria, MSM were less likely to have medical insurance, to be a member of or affiliated with an HIV-related organization, more likely to use mobile phones, more likely to search for HIV information on the internet, and had higher internet usage.

Predisposing, enabling, and reinforcing factors
Regarding reinforcing factors, more than half of all respondents were already on ART (56.4%), with MSM less likely than waria to have been on ART. Compared to waria, MSM were less likely to disclose their HIV status beyond family and partners, were less likely to experience stigma and discrimination, and had lower social support.

Univariate differences for retention in care
Retention in care was associated with several of the sociodemographic, predisposing, enabling, and reinforcing factors that were considered ( Table 3). As noted in Table 2, MSM were more likely to have adequate retention (73%) than waria (59%). Compared to those who had inadequate retention in care, participants with adequate retention had a significantly higher monthly income, and were less likely to have engaged in sex work throughout their lifetime.
Compared to those who had inadequate retention in care, participants with adequate retention were less likely to drink alcohol, smoke, and have unprotected sex. They also had higher HIV-treatment literacy. Such participants were also more likely to have medical insurance, more likely to search for HIV information on the internet and more likely to have been on ART. Further they were less likely to experience stigma and discrimination but they also reported having less social support.

Multivariable associations of retention in care
Model 1 (Table 4) indicates that none of the sociodemographic variables were significantly associated with retention in care. Model 2 shows that after adjusting for sociodemographics, having any health problems (adjusted odds ratio [AOR] = 4.45; 95% confidence interval [CI]: 1.05-18.5) and higher HIV-treatment literacy (AOR = 1.12; 95% CI: 1.10-17.9) were significantly associated with higher odds of adequate retention in care. These two factors remained statistically significant when enabling factors were included in Model 3. Controlling for sociodemographic, predisposing, and enabling factors, being a member of or affiliated with an HIV-related organization was associated with lower odds of retention in care (AOR = 0.50; 95% CI: 0.27-0.91).
In Model 4, which added reinforcing factors, having been on ART (AOR = 6.00; 95% CI: 2.93-12.3) was associated with higher odds of adequate retention in care. Respondents who searched for HIV-information on the Internet (AOR = 2.15; 95% CI: 1.00-4.59) and had medical insurance (AOR = 2.84; 95% CI: 1.27-6.34) were also associated with higher odds of retention in care. Being a member of or affiliated with an HIV-related organization remained associated with lower odds of adequate retention in care (AOR = 0.47; 95% CI: 0.24-0.95), whereas having any health problems or higher HIV treatment literacy were no longer associated with retention in care in Model 4.

Discussion
This study demonstrated high rates of early linkage to care and a moderate rate of adequate retention in HIV care among HIV-positive MSM and waria in Indonesia. Our analysis showed that two predisposing factors, i.e. having other health problems and higher literacy of HIV treatment seem to facilitate to the higher odds of adequate retention in care among our study participants when accounting for sociodemographic characteristics and enabling factors. For the enabling factors, using the internet for HIV information and having medical insurance The relatively high rates of linkage to care among our study participants is comparable to that among general HIV-positive patients in Indonesia [33], suggesting that there might not be disparities in linkage to care between MSM and waria populations compared to other groups in Indonesia. Also, the prevalence of MSM and waria in our study that have adequate retention in HIV care is similar to the prevalence of retention in care of all HIV-positive individuals in Indonesia taken together [17]. Our results show a higher proportion of adequate retention in care in our sample than HIV-positive MSM and transgender women in other global locations [5,31,34,35]. This is to the best of our knowledge the first study in Indonesia documenting retention in care of HIV-positive MSM and waria and its social aspects. A sizeable number of people from these difficult to reach groups were interviewed, allowing to capture an overall view of retention in care among these groups. Nevertheless, this study also has several limitations. Firstly, the survey used a non-random sampling approach in urban settings, so findings may not be representative of the entire population of MSM and waria groups in Indonesia. The use of targeted sampling which is similar to snowball sampling may also have limited our study participants to certain social network. Secondly, MSM and waria were pooled in the analysis, while descriptive analysis showed that these groups differed on several aspects. The sample sizes, however, were too small to perform subgroup analysis with adequate power. For the same reason we could not stratify participants by time of diagnosis, which might be important because standards of clinical care for people newly diagnosed with HIV may be different from those given to people who have been living with HIV longer [5]. Lastly, this study is subject to social desirability due the nature of self-reported data collection, although all precautions were taken during data collected to reduce this bias to a minimum.
Participants who had higher HIV-treatment literacy and who had other health problems were more likely to be retained in care, but these associations lost significance when taking ART status into account. It is common that HIV patients seek care only when they experience symptoms of disease [36], and our results appear to confirm this tendency that participants who are in good health do not seek health care regularly. Our finding also supports the notion that sufficient literacy or knowledge about the benefits of HIV treatment could increase Retention in care of HIV-positive MSM and waria in Indonesia motivation to seek and remain in treatment. To our knowledge, prior studies have not examined the association between knowledge and retention in care, although some studies have shown that a lack of knowledge about HIV treatment is a barrier to HIV testing [37] and ART use [38].
Knowledge about HIV care and treatment may be obtained by searching HIV-related materials through any source of information, including the internet. Our results show a positive Retention in care of HIV-positive MSM and waria in Indonesia association between using the internet to look for HIV information and adequate retention in care. This finding supports prior studies that identified the strong potential of internet use for HIV-positive individuals in the HIV care continuum [39][40][41][42]. Our study further revealed that medical insurance predicted greater odds of retention in HIV care. This implies that financial reasons may prevent people with HIV in Indonesia from routinely accessing HIV care. In lower-middle income countries, a fee-for-service program was associated with a lower probability that people with HIV will continue treatment after ART initiation [43]. Likewise, a systematic review showed that having private insurance was associated with higher utilization rates of health services among people with HIV, even in high income countries [44].
A surprising finding in our study is that participants who are attached to an HIV-related organization are less likely to be retained in care. We assumed such attachment could bring support for people with HIV, in turn enabling them to be retained in care. HIV-related organizations usually have solid networking with local HIV care and treatment services [45,46], and thus would encourage people with HIV to be adequately retained in care [5]. Nevertheless, our findings indicate that having such support may possibly keep them away from health care. Maintaining a healthy status to keep the doctor away is a basic health concept among Indonesian people for financial reasons. It is possible that participants with support from HIV-related organizations perceive themselves as being in good health and find it unnecessary to visit HIV care too often [47]. More insight into this phenomenon is needed in order to improve retention to care among these individuals.
As expected, we found that ART status had a positive association with retention in care. This aligns with other studies indicating a positive association between ART initiation with retention in care [48], and with a higher utilization of health services [44,49]. Our analysis further showed that once HIV-positive MSM and waria are on ART, predisposing factors are no longer influential towards retention in care. In Indonesia, people on ART are required to visit an HIV care facility every month for a one-month supply of ART pills. A buffer stock will only be given if they have an acceptable reason e.g., going out of town. This strategy may encourage people with HIV to visit HIV care regularly.
Our findings suggest that it is important to ensure that adequate information on HIV care is available online [39,42]. We identified a large proportion of internet users among our participants, particularly among MSM. Internet should thus be considered as one alternative medium to improve adherence. In view of the high rate of mobile phone use, future research on the efficacy of mobile phone based interventions [50], either through text messaging [51][52][53][54] or smart-phone applications [55] is worth considering. It is also important to guarantee that HIV care is affordable for people with HIV. It is crucial to lower the cost of HIV care or to give people with HIV access to a health insurance program. Being on ART gives a greater likelihood that people with HIV remain in HIV care, therefore a strategy seems appropriate that puts newly diagnosed MSM and waria on ART immediately [56,57]. Lastly, more knowledge is needed to comprehend how organizational support influences retention in care of people with HIV in Indonesia.

Conclusions
Our results fill gaps on data about two crucial stages in the HIV treatment cascade, i.e., linkage to and retention in care, specifically for HIV-positive MSM and transgender women in Indonesia. This study describes how sociodemographic characteristics and social determinants influence retention in care. As highlighted in our study, future interventions should carefully consider socioeconomic and cultural barriers and use internet-based technology to improve retention in care, and ultimately viral load suppression in these two vulnerable populations.