Using risk-tracing snowball approach to increase HIV case detection among high-risk populations in Cambodia: an intervention study

Over the past two decades, Cambodia has been successful in its fight to slow the spread of HIV [1]. HIV prevalence among the general population decreased from an estimated 1.6% in 1998 to 0.6% in 2015 [2, 3]. The estimated number of new HIV cases per year also decreased from roughly 4400 in 2005 to 700 in 2015 [2, 3]. Despite this success, as of 2015, an estimated 74,000 Cambodians were living with the virus [4]; with about 27% of these individuals (~19,000) being unaware of their HIV-positive status [2].

Early HIV diagnosis and initiation onto antiretroviral therapy (ART) may prevent ongoing spread of HIV. Knowledge of one’s HIV-positive status has been shown to be associated with increased condom use [5]. ART initiation, when resulting in viral load suppression, can reduce the risk of HIV infection between 89% to 96% to uninfected sexual partners of an HIV positive individual [6‐10].

The Government of Cambodia has committed to ending the AIDS epidemic by 2020 [3]. It means, by the year 2020, the annual number of new HIV infections must fall below 300 [3]. To successfully reduce the number of new HIV infections, innovative approaches are needed to reach people living with HIV who are unaware of their HIV status and link them to early diagnosis and treatment. One such approach, the Risk Tracing Snowball Approach (RTSA) has been shown to be effective in detecting new HIV cases that would otherwise not have been found [11‐13]. RTSA is an adaptation of the Respondent Driven Sampling method, a form of sampling often used in bio-behavioural surveillance studies, in which study recruitment relies on peer-referrals [14]. RTSA, which is also sometimes referred to as a “Peer-Driven Intervention,” has been successfully used to increase the rate of newly undiagnosed HIV infected individuals among high-risk heterosexuals in the US [13]. It has also been found to be an effective approach to rapidly identify newly undiagnosed HIV individuals among people who use and inject drugs in Greece [12] and among heterosexual couples in China [11]. However, it has never been implemented in Cambodia context.

The main objective of this study was to evaluate the effectiveness of RTSA in increasing the rate of newly identified HIV cases among high-risk populations, including female entertainment workers (FEW), men who have sex with men (MSM), transgender (TG) women, people who inject drugs (PWID), people who use drugs (PWUD), and the general population who have social connections to key populations. The second objective was to evaluate the effectiveness of RTSA, as compared to the walk-in group, in increasing the number of HIV tests and early HIV case detection by comparing CD4 count between clients from the two groups.

This study found that the rate of newly identified HIV positive cases generated by RTSA was significantly lower compared to the rate in the traditional walk-in approach. However, we found that the absolute number of HIV cases detected during RTSA implementation (RTSA and walk-in approach combined) was much higher than that in the six-month period prior to the RTSA implementation and the six-month period after the RTSA implementation ended. This finding indicates that a combination of RTSA and a traditional walk-in approach may increase new case detection among high-risk populations, especially key populations.

The lower rate of newly identified HIV positive cases among the RTSA clients than that in the walk-in clients is not surprising. Individuals who walked in to get tested are more likely to do so because of their self-perception of their risky behaviors, or because they felt sick. It is also plausible that when more individuals are tested, the HIV rate will naturally go down.

Even though the rate of newly identified HIV positive cases for the RTSA group was lower than the rate obtained from the clinics’ routine data (walk-in group), this rate should still be considered acceptable for four reasons. First, the combination of the routine walk-in testing and RTSA during RTSA implementation boosted the absolute number of HIV new case detection from 32 in the six-month period prior to the RTSA implementation to 52 cases in the six-month period of RTSA implementation. Second, CD4 count in RTSA group was higher than that in the walk-in group. This may infer that RTSA leads to earlier detection of people living with HIV. Third, the rate of newly identified HIV positive cases in RTSA group was much higher than the rate of newly identified HIV positive cases reported nationwide in programs using an outreach HIV testing approach among key populations—FEW, MSM, TG women, and PWID-- which has never reached more than 1.0% since the program implementation was started in 2013 [15, 16]. Fourth, the rate of newly identified HIV positive cases in RTSA group is comparable to the national HIV prevalence in each key population: FEW (3.8% vs. 4.6%), MSM (1.7% vs. 2.3%), TG women (2.2% vs. 5.9%), and PWUD (1.7% vs. 4.1%), except PWID (0% vs. 25%) [17‐20].

Other than being in a high-risk group (TG women, MSM) and experiencing tested for HIV, our study did not find any factors associated with the rate of newly identified HIV positive cases in either groups—walk-in and RTSA. The non-significant differences may be explained by the following reasons. First, as with any other self-reporting measures, risky behaviors may have been underreported due to social desirability. This was also observed among the walk-in group by counsellors. Thus, levels of risks may have been misclassified. In contrast, the RTSA group may have over-reported their risks to make their recruiters eligible for financial incentives. It should be reminded that all recruited individuals can get HIV testing and receive 2.5 USD transportation compensation regardless their risk status and HIV testing result. However, seeds or recruiters can only get the incentive when their recruits were screened at risk for HIV infection. Second, our questionnaire has never been validated. We may not have asked the right questions to identify risks. And third, the small sample size may have contributed to the non-statistically significant associations. Regardless what might be the reasons, using a risk-screening tool to identify risky clients may have low validity, especially when it confounded by the use of monetary incentives. Excluding those who reported low risk to HIV infection may also have influenced our ability to further elucidate risk factors.

Theoretically, people are more likely to refer those who are similar to them [14]. This theory is not necessarily true considering HIV testing results in this study. HIV negative clients may be happier after getting the test result and might be willing to refer their peers. In contrast, positive clients were more likely to stop the recruitment after they learned about their HIV-positive status. As shown in the network graph, most of the positive clients did not recruit anyone else to join the study. This nature has blocked the power of network referral. If HIV-positive clients could recruit more people, they may recruit their sexual partners and their peers who had similar risks. However, newly detected HIV-positive individuals need some time to stabilize their emotional condition first, or they will never recruit other people because they do not want others to suspect or learn about their HIV status.

The strength of this study is the use of well-managed patients tracking system in Chhouk Sar clinics where the study was conducted. The online fingerprint system can track HIV test interval of individual clients. Clients could not be tested multiple times at one or another Chhouk Sar clinic within 3 months. However, our study also has several limitations. First, we could only trace clients in our system. We could not manage if they were tested elsewhere. Second, the self-reported risky behaviors may not be reliable due to the use of monetary incentives. However, identifying sexual risky behaviors was not the main objective of the study. Rather, it was the rate of newly identified HIV positive cases that detected by three-test algorithm —Determine Test, Stat-Pak, and Uni-Gold [21]. Third, the identification of newly detected cases was based on counsellors’ skills. In this sense, we cannot deny the possibility of inclusion of formerly detected HIV clients in the newly identified HIV positive group. However, because all newly identified HIV-positive clients were closely followed-up until they got enrolled in the treatment, only minor discrepancies were likely to happen. Lastly, our study ended up with inconclusive result whether the RTSA is cost-effective because the study design did not allow further investigation on this important aspect. To address this shortcoming, cost-effectiveness modelling should be conducted. This kind of study will be able to assess the number of people who will be prevented from HIV infection resulting from early detection by RTSA, and unit cost to invest to keep a person uninfected with HIV per year – known as Disability-Adjusted Life Year (DALY) [22].

The rate of newly identified HIV positive cases among the RTSA group was lower than that found within the walk-in group, but the inclusion of RTSA method in addition to the traditional walk-in method boosted new HIV case detection. The higher CD4 counts found in the RTSA group compared to that in the walk-in group may reveal that RTSA could detect HIV cases earlier than walk-in approach could do. Further research is needed to understand whether RTSA is a cost-effective intervention to prevent ongoing spread of the HIV.