One size does not fit all: HIV prevalence and correlates of risk for men who have sex with men, transgender women in multiple cities in Papua New Guinea

Background Biobehavioral data about men who have sex with men (MSM) and transgender women (TGW) in Papua New Guinea (PNG) are limited to those who sell sex. Information about those MSM and TGW who do not sell sex is necessary to guide HIV prevention and treatment efforts. Methods We conducted respondent-driven sampling (RDS) surveys among MSM and TGW in Port Moresby, Lae, and Mt. Hagen, PNG from in 2016 and 2017. Eligibility criteria was: aged > 12 years, born male, could speak English or Tok Pisin and had oral or anal sex with another person born male in the past 6 months. Participants were interviewed face-to-face and offered rapid HIV testing. Weighted data analysis was conducted using RDS-Analyst (v. 0.62). Results We enrolled 400 participants in Port Moresby, 352 in Lae, and 111 in Mt. Hagen. In the last six months, 73.2% of MSM/TGW in Port Moresby, 77.9% in Lae, and 75.9% in Mt. Hagen, had a concurrent sexual partnership. Upwards of 70% of MSM/TGW in all three cities had sex with a woman in the same period. Less than half of MSM/TGW had ever tested for HIV. HIV prevalence among MSM/TGW was 8.5% in Port Moresby and 6.9% in Lae. Among participants in Mt. Hagen it was 1.3%. HIV was associated with not having sex with a woman in the last six months and sexually transmitted disease symptoms in the last 12 months in Port Moresby and Lae. In Port Moresby, it was also associated with an uncut foreskin, and in Lae with earning income in the formal sector and being unable to rely on other MSM or TGW to accompany them to healthcare services. Conclusions The large proportion of MSM and TGW with concurrent sexual partnerships, combined with the low testing coverage, indicates strong potential for the spread of HIV. The different correlates of HIV in Port Moresby and Lae highlight the importance of conducting surveys in multiple locations and using data to develop locally appropriate interventions even within a country. Electronic supplementary material The online version of this article (10.1186/s12889-019-6942-7) contains supplementary material, which is available to authorized users.


Background
Nearly four decades since the emergence of HIV, there are still settings and populations whose HIV epidemics we know little about [1,2]. Papua New Guinea (PNG) is one such example. It was previously described as having a generalized epidemic. The expansion of antenatal clinic surveillance data, as well as biobehavioural surveys, have resulted in a better understanding of PNG's HIV epidemic and a consensus has emerged that with an HIV prevalence estimated at 0.9%, the country is experiencing an epidemic concentrated the key populations of sex workers, men who have sex with men (MSM), and transgender women (TGW) [3][4][5][6][7][8]. The data that exist on MSM and TGW in PNG focuses on those who sell sex, rather than all MSM and TGW. This new understanding of the epidemic in PNG necessitates the collection of data on key populations to inform estimates and guide the country's HIV response.
Male-to-male sex remains illegal and highly stigmatized in PNG, hampering HIV service provision and uptake for these men. One of the first respondent-driven sampling (RDS) surveys in the world was conducted among MSM in PNG in 2005 but it did not assess HIV prevalence [7]. It did, however, highlight low condom use and experiences of discrimination among MSM. A 2010 survey of MSM had similar findings but similarly did not include HIV testing [8]. In the same year, a survey of females, males, and transgender women engaged in sex work in the capital of Port Moresby found HIV prevalence in these populations of 19.0, 8.8, 23.7%, respectively [9]. These findings helped reorient PNG's HIV response to focus on key populations, but until now there has been little information to guide the epidemic response [10,11].
Data about MSM and TGW in PNG are needed to guide the country's HIV strategy and service provision. In conjunction with the PNG National Department of Health and National AIDS Council Secretariat, we conducted a respondent-driven sampling (RDS) biobehavioural surveys (BBS) of MSM and TGW in Port Moresby to fill this information gap. Here we report on HIV prevalence and correlates of HIV infection.

Community consultation
Community consultation was undertaken with MSM and TGW in Port Moresby, Lae, and Mt. Hagen to guide survey preparation and build trust with local stakeholders. Community consultation indicated that recruiting MSM and TGW would be challenging and both populations found it acceptable to combine the two populations into one survey sample.

Study population, setting, and design
We conducted RDS BBS of MSM and TGW in Port Moresby from June to October 2016, in Lae from January to June 2017, and Mt. Hagen from August to December 2017. These cities were selected for the survey because they are the most populous and include the national capital (Port Moresby), the main economic city and port (Lae), and a city that is at the intersection of roads and natural resource extraction activities in the country (Mt. Hagen). RDS is a variant of snowball sampling that can be used to produce sampling weights and approximate a random sample [12][13][14]. Eligibility criteria were: age > 12 years, spoke English or Tok Pisin, born biologically male, had oral or anal sex with a male in the past 6 months, and be in possession of a valid study coupon.

Recruitment
Recruitment started with four seeds in Port Moresby, four in Lae, and five in Mt. Hagen. Twelve additional seeds were added in Port Moresby, 15 in Lae, and 4 in Mt. Hagen to facilitate recruitment. Seeds were purposely selected to create diversity with respect to: age, sexual/gender identity, place of residence, region of origin, marital status, receipt of a unique object for size estimation, and affiliation with a non-governmental or community-based organization.

Data collection
Candidate participants were screened for eligibility and those eligible were asked to provide verbal informed consent. Non-blood specimens were collected for sexually transmitted infection and tuberculosis testing after which participants engaged in a computer-assisted personal interview (Open Data Kit, Washington, US).
After the interview, participants received pre-test HIV counseling before providing written informed consent for HIV testing. Fifteen milliliters of blood was taken through venipuncture for HIV testing and, if positive by confirmatory testing, CD4 T-cell count and molecular HIV viral load testing were also conducted at the point-of-care. The PNG national algorithm for HIV testing was used: Determine HIV-1/2 (Alere, MA, USA) followed by confirmatory testing with HIV 1/2 Stat-Pak (Chembio, Medford, NY). Discordant results between the two tests were deemed inconclusive, and participants were advised to retest in 3 weeks. HIV external quality assurance panels were provided by The Royal College of Pathologists of Australasia. We used the Chembio DPP Syphilis Assay to test for syphilis (Chembio, Medford, NY). All test results were returned to participants at the end of the first study visit. HIV-positive participants were offered an escort by a peer navigator to HIV treatment services of their choosing that had been sensitized to work with key populations. Those with an active syphilis infection initiated treatment at the survey site and were given a referral to complete treatment. Study staff were trained to identify and refer all sexually exploited persons under the age of 18 years to partner organizations experienced in providing counseling, health, social, and other protective services to these populations.
While waiting for test results, participants in Port Moresby received 3 coupons with which to recruit peers. Midway through data collection the number of coupons was increased to 4 to increase recruitment. Participants in Lae received three coupons to recruit peers and in Mt. Hagen they received four. In all locations, participants received 45 PNG kina for their first visit (about $14 US) and 10 PNG kina (about $3 US) per successful recruit plus 5 PNG kina (about $1.50 US) for transportation at their second visit. All participants were also provided with information on HIV and other sexually transmitted infections, condoms, and lubricants.

Data measures
The questionnaire used as its foundation the WHO, CDC, UNAIDS, and FHI360 Biobehavioral Survey Guidelines for Populations at Risk for HIV [15]. Interview domains included demographics, sexual history and identity, condom use, stigma, social cohesion, violence, HIV knowledge, history of sexually transmitted diseases, penile modification, internalized homophobia, and uptake of health services. The two-item Patient Health Questionnaire (PHQ-2) was used to screen for depression [16]. Internalized homophobia was defined based on responses to five questions regarding their feelings about sexual attraction to men. Comprehensive awareness of HIV was based on the United Nations Joint Programme for HIV/AIDS definition of correctly answering three questions and rejecting two myths regarding HIV [17]. Partner concurrency was defined as overlapping sexual partnerships with partners of any sex where sexual intercourse with one partner occurred between two acts of intercourse with another partner, limited to the last three sexual partners in the last six months. The full questionnaires can be found in the Additional files 1 and 2 (FSW Questionnaire and MSM TGW Questionnaire).

Data analysis
Our analysis characterizes MSM and TGW and correlates of HIV infection among these populations in each of the three survey cities. Odds ratios (OR) and 95% confidence intervals (CI) were calculated for bivariate comparisons and a p-value < 0.1 was the threshold for inclusion in multivariate analysis to identify a final model. To remain in the final model, the least significant variable was eliminated until all remaining variables were significant at the 0.05 level. Additional variables were removed to lower the relative standard error.
In Port Moresby, the full model included education, gender identity, disclosed sexual identity to non-MSM, ever cut foreskin, sex with a woman in the last 6 months, self-reported sexually transmitted disease (STD) symptoms, used the internet or mobile application to meet people, and last contact with peer outreach.
In Lae, the full model included main source of income, disclosure of sexual behaviors to non-MSM, can rely on other MSM or TGW accompany them to the doctor, experienced physical violence in the last 12 months, sex with a woman in the last 6 months, had self-reported STD symptoms, and last contact with peer outreach. No model was produced for Mt. Hagen because the actual sample size was too small and convergence was not reached for HIV [18].
Data were weighted and analyzed using Respondent-Driven Sampling Analyst (RDS-A) version 0.62 (Los Angeles, CA) and SAS version 9.3 (Carey, NC). We utilized Giles' Successive Sampling Estimator in RDS-A. All data presented are RDS-adjusted population estimates unless otherwise indicated. Weights were imported from RDS-A, and survey logistic procedures were used in SAS to identify correlates of HIV infection.

Ethical approval
This survey was approved by the PNG National Department of Health's Medical Research Advisory Committee (MRAC), the Research Advisory Committee of the National AIDS Council Secretariat (RAC), the PNG Institute of Medical Research's Institutional Review Board (IRB), and the Human Research Ethics Committee at UNSW Sydney, Australia. The protocol was reviewed according to the Centers for Disease Control and Prevention's (CDC) human research protection procedures and was determined to be research but CDC was not engaged. A letter of support was provided by Kapul Champion, the peer led civil society for sexually diverse men and transgender people. In all three cities male identity was more common than transgender identity and the majority of MSM and TGW were between the ages of 20-29 years ( Table 1). The median age was highest in Port Moresby and lowest Less than half of MSM and TGW had ever tested for HIV (41.8% in Port Moresby, 32.1% in Lae, and 28.7% in Mt. Hagen) and of those who had, over half in each city tested more than six months ago. Survey-related testing found HIV prevalence among MSM and TGW was 8.5% (95% CI: 5.0-11.9) in Port Moresby and 6.9% (95% CI: 3.6-10.3) in Lae, and 1.8% among survey participants in Mt. Hagen. 1 Prevalence of active syphilis was 4.0, 8.5, and 2.5%, respectively, in the three cities. Prevalence of lifetime syphilis infection was 10.1, 21.1, and 8.3%, respectively (data not shown).

Discussion
Our survey is the first biobehavioral HIV survey of MSM and TGW in PNG that includes both sex workers and non-sex workers. The large proportions of MSM and TGW with concurrent sexual partnerships or who sell sex, combined with the low testing coverage and condom use at last anal sex with a male or transgender woman, suggests substantial potential for the spread of HIV among these populations and their female sexual partners.
The different correlates of HIV in Port Moresby and Lae highlight the importance of conducting surveys in multiple locations and most importantly using such data to develop locally appropriate interventions even for the same population within a country. Having an uncut foreskin is associated with greater risk of HIV in Port Moresby but there is no association in Lae. In contrast, not being able to rely on peers for accompaniment to a clinic or hospital is not associated with HIV in Port Moresby but is associated with increased risk of the disease in Lae. This may be because there are fewer HIV and sexual health services for MSM and TGW in Lae than in Port Moresby, and the prospect of accessing services alone at facilities with staff not trained to work with MSM and TGW may prove daunting. Employment in the formal sector was also associated with HIV in Lae but not in Port Moresby, suggesting the need for greater targeting of HIV prevention and testing service in Lae compared to Port Moresby. In both cities, not having sex with a woman in the last six months was associated with HIV infection. While it is important to access all MSM and TGW for HIV services, given the low engagement of MSM and TGW in PNG with outreach workers and HIV testing, prioritization of limited HIV resources to MSM and TGW who do not have sex with women could reach more people as these individuals are often most accessible to KP services.
A high prevalence of sexually transmitted infections has been well documented in PNG, and we similarly documented a high prevalence of self-reported STD symptoms among MSM and TGW [19,20]. Though active syphilis infection was not associated with HIV, in both Port Moresby and Lae, HIV was associated with having at least one STD symptom in the last 12 months. As STI can contribute to both the transmission and acquisition of HIV, those seeking treatment for STD symptoms should routinely be offered HIV testing [21,22].
The lower proportion of MSM and TGW traveling outside of Port Moresby compared to other cities is likely due to the fact that Port Moresby is not connected to the rest of the country by road, whereas Lae and Mt.
Hagen are on the main highway. This may also lend itself to preventing the spread of HIV from the capital to Lae and Mt. Hagen. As the use of the internet and mobile applications to meet sexual partners has rapidly expanded in PNG and will likely continue to do so, these media platforms may be an effective way of reaching MSM and TGW with information about HIV and services. The relative recency (last 12 months) of interaction of MSM and TGW in Port Moresby and Lae with peer outreach workers suggests the potential for community-based HIV testing or provider-assisted self-testing.
Our findings are limited by the cross-sectional nature of our study and the low sample size in Mt. Hagen which we believe is due to the particularly heteronormative culture of this highlands city. Given that the lack of legal protections for MSM and TGW (which do not exist in PNG) are associated with higher HIV prevalence in a global meta-analysis and that HIV prevalence in the general population is highest in Mt. Hagen, we suspect that our finding of 1.8% prevalence among survey participants is an underestimate [23]. As interview data were self-reported in face-to-face interviews, there may be some response bias. The use of audio-computer assisted self-interviews rather than face-to-face interviews may have helped decrease this bias [24].

Conclusions
Our survey was able to reach MSM and TGW that were not engaged in HIV services, revealing that MSM and TGW will access HIV services provided in a safe, affirming environment, providing further evidence to guide service providers and policy makers. As the first HIV biobehavioral survey in PNG among MSM and TGW, this study reveals the substantial vulnerabilities and HIV risks of these populations in PNG and their limited access to and uptake of routine HIV services. Without action, HIV prevalence stands to increase among MSM and TGW in PNG. Addressing the HIV epidemic in PNG will require continued and expanded engagement with MSM and TGW across the country, respecting and responding to their different needs and vulnerabilities,