Announcing the availability of oral HIV self-test kits via text message to increase HIV testing among hard-to-reach truckers in Kenya: a randomized controlled trial

Truckers in ub-Saharan Africa are at high risk for HIV due to high levels of risk behavior. A study in 1991 among a sample of truckers in Kenya found 61% reported having visited sex workers, only 32% had ever used condoms [1], and 18% tested HIV-positive [2]. A 2015 study among truckers in Kenya found that 56% had paid for sex in the past 6 months, 47% had a regular partner along their trucking route in addition to a wife or girlfriend at home, and only 14% had always used condoms during sex in the past 6 months [3]. Studies among truckers in other countries have found high HIV prevalence, including 23% in Nigeria [4], 15.4% in Mozambique [5], and 26% in South Africa [6]. Because truckers cross national and international borders for work, high HIV prevalence among truckers can lead to HIV spread across these borders as well as from casual and commercial partners along trucking routes to wives and main partners at home [7, 8].

Few studies have looked at HIV testing among truckers. One study among 1881 truckers in South Africa in 2003–2004 found that only 38.2% had ever been tested for HIV [6]. HIV self-testing might be a more acceptable and convenient HIV testing option for truckers. In a 2015 study [3, 9], 305 truckers were recruited from the waiting room of two roadside wellness clinics in Kenya, randomized to be offered either a provider-administered blood (finger-prick) HIV test (Standard of Care [SOC]), or an oral HIV self-test as an option in addition to the standard HIV test. Those in the intervention group were also allowed to pick-up self-test kits 3–6 months later. In this study, HIV testing rates were higher in the intervention group the first time HIV testing was offered immediately following randomization (Odds Ratio [OR] = 2.8, p = 0.002), but not during the 6-month follow-up (OR = 1.0, p = 0.972). The reason for this discrepancy could be that the self-test increases testing when it is offered to someone who is already in the clinic, but it is not sufficient motivation to bring people to the clinic to get a test kit.

A number of studies among other population groups have found that direct distribution of HIV self-test kits can increase testing rates. A study in Kenya that gave pregnant women self-testing kits to take home to their male partners found a 1.6 times higher testing rate among those partners compared to when women gave their partners an invitation card to come to the clinic for HIV testing (p < 0.001) [10]. A similar study in Uganda had a similar impact (risk ratio [RR] = 2.1) [11], and two studies in Zambia and Uganda that offered self-test kits to female sex workers through peer educators found higher testing rates compared to the SOC (referral to a clinic for testing), although the difference was only significant for the Uganda study (RR = 1.3, P < 0.001 in Uganda and RR = 1.1, P = 0.11 in Zambia) [12, 13]. A home-based (door-to-door) HIV testing study in Zambia found that offering a self-test in addition to the standard provider-administered HIV test increased testing rates from 55.1 to 60.4% [11].

However, direct distribution may be unfeasible in some situations, especially for truckers who are mobile and have unpredictable schedules. Therefore, we conducted a randomized controlled trial to assess whether alerting truckers via text message about the availability of HIV self-testing at clinics located at major transit hubs in Kenya would increase HIV testing rates. We conducted a similar study among female sex workers in Kenya and found that those who were notified about the availability of HIV self-testing via text message were significantly more likely to come to a clinic for HIV testing (OR 1.9, p = 0.001) compared to those who received text messages about HIV testing in general (not about self-testing) [14]. For this study, we used similar methods and specifically targeted truckers who were not already accessing HIV testing regularly to see if alerting them about the availability of HIV self-testing would increase HIV testing rates among these hard-to-reach truckers.

For this study, we used methods similar to those we used in the study among female sex workers in Kenya [14]. Below we briefly describe the main design characteristics of the study.

HIV testing in this sample of hard-to-reach truckers was very low, at 1.3% in both the enhanced SOC and the SOC groups during the study period. Texting about the availability of HIV self-test kits increased HIV testing rates to 3.5% (OR = 2.7, or 4.1%, OR = 3.2 when including the administrative record data). This increase in testing rate is not very different from the study that recruited truckers from the waiting room of two North Star Alliance clinics in Kenya (OR = 2.8) [3, 9]. However, participants in the previous trucker study were testing at much higher rates than those in the current study, in part by design as we purposely selected irregular testers in this study. Thus, the relative increase in testing rate was fairly consistent across the two trucker studies that differed in terms of how the self-test was initially introduced (by the counselor when the client was already in the clinic versus via text message) and in terms of the HIV testing history of the participants. However, the absolute difference in testing rates varied widely between the two groups (2.2 per 100 more truckers testing in this study versus 14.9 per 100 truckers in the previous study) because the baseline (or SOC) testing rate in this study was so much lower. It is important to note that, in this study, five HIV-infections were identified in the intervention group. This may suggest that the intervention encouraged some to test who were not going to test in the SOC program, and that these people were more likely to be HIV-positive, either because they had not tested in a long time, if ever, and thus the pool of unidentified infections among those testing in the intervention group was greater than that among those testing in the SOC groups, or because those at higher risk were more likely to test under the intervention than the SOC. As participants were randomized to the three study arms, there is no reason to suspect that this higher rate of HIV-positive results in the intervention group was due to systematic differences in the participants in that study arm compared to the other two arms, although with chance, it is possible. It is also unlikely that the rate of false positives was higher among those in the intervention group, as the specificity of the self-test kit is actually slightly higher (99.9%) [18] than that of the SOC test (98.5%) [19].

Importantly, our study found that announcing the availability of HIV self-test kits via text did result in more clients coming to the clinics to access these kits and direct distribution is not the only effective distribution model. Our study among female sex workers who were irregular HIV testing also found that the text alert about the availability of HIV self-test kits increased testing rates (OR 1.9, p = 0.001) [14]. Thus if this basic text-message alert about HIV self-test kit availability worked among our hard-to-reach, high risk groups such as truckers and sex workers who are not accessing HIV testing regularly, it may work even better in the general population. Use of text messaging to announce the availability of HIV self-testing could be an easy and cost-effective strategy to increase uptake as countries roll out self-testing. HIV self-testing was approved in Kenya in May 2017, a few months after the completion of our study, with test kits being made available for a cost of around $8 US [16]. Text messages might be a useful mechanism to inform the public about where self-test kits can be obtained as they become available in new locations. Due to the character limits, the text messages we sent were fairly simple; however, videos depicting the self-testing process and package inserts are now available online in multiple African languages, including Kiswahili [20]. With the proliferation of smart phones in Kenya [21], links to demonstrations of the self-testing process could be sent to people so they can see what the self-test is and how it works. Providing more information about the self-test might further increase in test uptake over what we achieved.

This study specifically targeted hard-to-reach truckers and, while it resulted in an increased testing rate, the vast majority of truckers in the sample, even in the intervention group, did not test. Additional research is needed to explore the reasons why these hard-to-reach truckers do not test and what additional types of programs might encourage more to test. It could be that in order to reach the hardest-to-reach, a number of different new program options will be needed, with clinic-distribution of self-test kits being only one such program. It is possible that we would have had a greater increase in HIV testing if we had distributed the HIV self-test kits directly to the truckers. Clinic-based distribution may not reach those who are unable to access or do not feel comfortable going to clinics. Other studies that used direct or secondary distribution mechanisms to bring HIV self-test kits directly to potential users found increased testing rates with risk ratios from 1.1 to 2.1 [10‐13], which are actually somewhat lower than what we found. However, given the difficulty in reaching some truckers, such as those in this study, direct distribution of self-test kits, coupled with clinic pick-up as an option, might be worth exploring. The North Star Alliance already conducts outreach at truck stops and direct distribution of self-test kits might be added to that program to try to further improve testing rates among the hard-to-reach.

The majority of the study participants in the intervention group who tested chose to self-test (64.5%), but the SOC test was chosen by over a third of participants. Among those who chose to self-test, about two-thirds opted to do so in the clinic with supervision and one-third took a test kit for home use. Thus it seems that people vary in their preferences around HIV testing, which suggests that offering choices may increase the chance that a client can find an acceptable HIV testing option. Similar diversity in preferences was found in our previous study [3, 9], and in a Zambian study [11]. The popularity of self-testing with supervision evident in all these studies warrants further exploration. It could be that some people feel the need for guidance their first time self-testing, and going forward they would instead self-test unsupervised at home, or it could also be that some people’s ideal HIV test would be an oral provider-administered test, and supervised oral self-testing was the closest they could get to that in these studies. A discrete choice experiment we conducted in our previous study among truckers suggested that people have strong preferences regarding type of test (blood versus oral) and form of counseling (in-person versus over the phone) and these preferences differed by HIV testing history; however, preferences regarding who administers the HIV test as well as the testing location were not strong [22]. Thus, future studies might explore different combinations of test choices, such as provider-administered oral tests and self-administered blood tests, to try to determine which tests are the most popular and what the array of testing choices should be in order to maximize testing uptake.

This study had some limitations. Our text message was short and some may not have understood it or were unclear about what a self-test was. This may have weakened the potential impact of the intervention. In addition, the EHRS data may have included errors and misclassifications, causing us to include some who were ineligible, exclude some who were eligible, and misclassify the outcome for some participants. The large number of duplicate phone numbers in the system as well as the missing HIV-testing data on five participants who self-tested suggests a fair amount of error in the system. It is also possible that some may have tested during follow-up at clinics not part of this system and those would be misclassified in our data. However, the proportion choosing the standard HIV test in the intervention group (11/750) is similar to the proportion in the two SOC groups (10/748 and 10/762), suggesting that the standard test may appeal to a consistent proportion of truckers in all groups and therefore it seems likely that the portion testing in other clinics would also be similar due to randomization. Therefore these errors should be non-differential and bias our results towards the null. Finally, our results cannot be generalized to all truckers as we selected a sample meeting specific criteria, including a failure to test for HIV regularly, and from a clinic-based health record system.

Announcing the availability of HIV self-testing via text message significantly increased the HIV testing rate among a sample of hard-to-reach truckers, but many still did not access testing. Additional research is needed to identify the combination of HIV testing programs that will further increase testing rates in this group.