The effects of HIV self-testing kits in increasing uptake of male partner testing among pregnant women attending antenatal clinics in Kenya: a randomized controlled trial

Introduction HIV self-testing could add a new approach to scaling up HIV testing with potential of being high impact, low cost, confidential, and empowering for users. Methods Pregnant women attending antenatal clinics (ANC) and their male partners were recruited in 14 clinics in the eastern and central regions of Kenya and randomly allocated to intervention or control arms at a ratio of 1:1:1. Arm 1 received the standard of care, which involved invitation of the male partner to the clinic through word of mouth, arm 2 received an improved invitation letter, and arm 3 received the same improved letter and, two self-testing kits. Analysis was done using adjusted odds ratios (aOR) at 95% confidence intervals (CI) to calculate and determine effects of HIV self-testing in increasing uptake of male partner testing. Results A total of 1410 women and 1033 men were recruited; 86% (1217) women and 79% (1107) couples were followed up. In arm 3, over 80% (327) of male partners took HIV test, compared to only 37% (133) in arm 2 and 28% (106) in arm one. There was a statistical significance between arm one and two (p-value=0.01) while arm three was statistically significant compared to arm two (p-value<0.001). Men in arm three were twelve times more likely to test compared to arm one (aOR 12.45 (95% CI 7.35, 21.08)). Conclusion Giving ANC mothers test kits and improved male invitation letter increased the likelihood of male partner testing by twelve times. These results demonstrate that HIV self-test kits could complement routine HIV testing methods in the general population.


Introduction
Sub-Saharan Africa accounts for over 70% of all HIV related morbidity and mortality globally while in 2016, 27 million of the 36.7 million People Living with HIV (PLHIV) were in sub-Saharan Africa. Kenya has the fourth largest HIV epidemic globally. In 2012 alone, an estimated 1.6 million people were living with HIV and close to 58,000 people died of HIV related illness [1]. HIV and AIDS prevalence peaked at 11% in the mid-nineties and has since dropped and stabilized to below interventions [2]. Kenya has a generalized epidemic, meaning it affects all sections of society including children, young people, adults, women and men. HTC is a very effective strategy in linking those infected with HIV into care and treatment and making appropriate referrals to other HIV support services [3]. Kenya has adopted a number of strategies including provider initiated testing and counselling (PITC), outreach testing and counselling, home based testing and counselling. Other strategies that have proved effective are integration of HTC in antenatal care and reproductive health [4].
Kenya has in the recent years adapted and now included in the national HIV Testing Services (HTS) guidelines, HIV self-testing approach which is defined as any form of testing in which an individual collects his or her own sample then performs a simple HIV test; and is, therefore, the first to know the results. Self-testing could add a new approach to support scaling up testing with a potential to be high impact, low cost, confidential, and empowering for users. Unregulated sales of test kits and evidence of informal self-testing by health workers and community health workers indicate a demand for selftesting. Out-of-facility approaches to offering testing in the community and the workplace are means of bringing access to testing closer to clients [3,5]. Distance has always been cited as a barrier to HIV testing and, these methods could be complementary means to scale up HIV testing. There is ample evidence documenting the impact of male involvement on the various components of PMTCT and HTS programs [6]. Men play an important role in reducing women's risk of acquiring HIV [7], women's utilization of services, including testing for HIV [8] and in promoting a woman's health [9]. Male partners also influence women's treatment decisions, including whether she receives medication [10] and whether she adheres to infant feeding advice [11].
Many women in Kenya fear disclosing their HIV positive status to their male partners because of stigma or discrimination, many at times these same women will end up experiencing some form of physical or psychological violence, and/or even death [12]. Thus, the potential for violence as a consequence of disclosure remains an important question to address empirically. Uptake of facility-based HIV testing among pregnant women attending ANC is as high as 88% [1], but approaches to promote male involvement during ANC services, such as having male-only clinics on special days or evening clinics have been tried without much success [13]. There is limited information on male involvement in HTC at the antenatal setting in the developing world [14]. Correlates of male HIV testing behavior and the reasons for obtaining HIV tests have been examined in sub-Saharan Africa using population-based studies and within workplace and clinic-based HTC trials [15]. Reporting HIV risk behavior patterns generally appeared to be associated with increased testing behavior, but with varying socio-demographic characteristics across settings. Eastern and central regions of Kenya have the lowest testing rates among male partners in Kenya [2]. Stigma and discrimination is also widespread in these regions, which is a barrier to seeking HIV services among those who test positive. [1]. Women feel burdened by clinics' request for their partner to visit health facilities for testing [16]. Fear of disclosing HIV results to male partners results into many ANC women feeling less empowered to ask their partners to undergo an HIV test, rather, women prefer the request come from the clinic staff themselves [6]. The current nationwide practice in Kenya to encourage male involvement is by sending a letter or word of mouth home with the pregnant woman, regardless of her HIV status, which invites her partner to come to the facility for involvement in ANC. The letter does not mention the need for HIV testing of the partner, nor does it contain important and potentially motivating information such as the rate of married individuals who are HIV-positive but have an HIV-negative partner in Kenya i.e. discordance rate (45%). The KAIS 2012 report showed that 71% of Kenyans indicated a willingness to use an HIV self-test kit (74.1% of males and 67.3% of females) [1].
The aim of this study was to determine the effects of HIV self-testing kits on uptake of male partner testing among pregnant women attending antenatal clinics in Kenya. and control for the additional information that must be provided to test-kit recipients. Arm 3 (intervention arm) were given the same improved letter as arm 2 describing the benefits of HIV testing.

Methods
Participants were also given demonstrations on how to use the oral test kits then they were provided with two oral self-test kits,

counseling, communications materials and clear instructions on how
to: self-test, introduce the self-test kit to their partners and report use of tests through mobile phones or during their next ANC visits.
Participants in all three-study arms received more counseling on PMTCT, linkages to care and support groups if tested positive, and counseling on partner testing in general. All clients were informed how they would be followed up with either SMS messages or one on one by the study assistants where mobile use was not feasible. They were also given information on who to call if, for example, they had questions on the HIV self-testing, what the results meant or resources for GBV.
Outcomes: the primary outcome was the proportion of male partners who tested for HIV within three months as reported by the female partners. The secondary outcomes were acceptability of self-testing by the male partner, incidence of GBV, linkage to care and man's selfreport of HIV testing to supplement the woman's self-report, and measure concordance in reporting within the couple.
Sample size: sample size was calculated by comparing study arm one and study arm two, and study arm two and study arm three.
Sample size calculation comparing study arm 1 and study arm 3 was not performed because it was expected that the difference between the two groups would always be larger than any other comparison. In all sample size calculations, level of significance was assumed at 5% and power was at a minimum of 80%. Groups 1 and 2 sample sizes were calculated based on equivalence test. It was assumed that 5% was the limit of equivalence, i.e. any difference bigger than 5% made the two groups not equivalent. If there were no difference between study arm 1 and study arm 2, 5% limit of equivalence, n=475 per group for 80% power to detect difference of more than 5%. 475 per group ANC clients were required to be 80% sure that the limits of a two-sided 95% confidence interval were excluded in difference between the two groups of more than 5%. Study arm two and study arm three sample sizes were calculated by assuming that arm 3 would have an uptake of partner HIV testing of 20%, while study arm 2 would reach at least 11% (the upper limit of equivalence). Based on these, 250 ANC clients were required to have an 80% chance of detecting, as significant at the 5% level, an increase in the partner HIV testing measure from 11% in study arm 2 to 20% in study arm 3.
To avoid imbalance between the study groups, study recruited a similar number of participants for study arm 3, as in Groups 1 and 2. box and open to determine the assigned group. Figure 1 illustrates the study process.  Table 1 presents the participants (women, men and couples) recruited and followed up in each of the three study arms.

Results
Baseline demographic data: Table 2 and Table 3 presents participants characteristics across the three study arms (women and men). These characteristics were well balanced across the three study arms except for the testing history and the educational level by the male partner (P=0.009 and P=0.006 respectively).
HIV testing acceptability rates among male partners: in this study, 83% (327) of the males in arm three where test kits were provided to the ANC mother accepted testing for HIV as compared to those in arms two and one 37% (133) and 28% (106) respectively.

Prevalence of gender based violence (GBV), intimate partner violence (IPV), gender inequity and other harm associated with introduction of oral HIV self-testing on male partner
testing: on further investigation to establish prevalence of GBV, IPV, gender inequity and other social harm associated with the introduction of self-testing on males among women in the study, there was no incidence of GBV or IPV reported during the study period. Further there were no significant differences across the arms in terms of men's reactions towards GBV and IPV. Generally, over 80% of men interviewed in the three study arms did not support any form of gender-based violence.

HIV testing among ANC clients
Person conducting HIV test at home: all women in arm one who reported testing, tested in the clinics or VCT centers. Among those women in arm two and three who tested for HIV at home, 100% [5] and 99.7% (310) respectively self-tested and they did not involve a counsellor to conduct the test.

Effectiveness of location of HIV testing in improving couple
testing: a total of 560 couples tested together for HIV with 321, 133, and 106 couples testing together in arm 3, 2, and 1 respectively. In arm one and two 99.1% (105) and 97.7% (130) of the couples tested at clinic/VCT respectively. In arm 3, 86.6% (278) of the couples tested at home. Home testing was statistically significant in improving couple HIV testing (p-value<0.001).

Usability of test kits by the ANC mothers and their
partners: usability of test kits by ANC mother in arm three together with their partners was assessed using three parameters; understanding of user instructions for HIV self-testing kits, reading test results and taking the cheek swab. Participants with primary and secondary school education and those aged 18-34 years and 35 years and older reported it was very easy to understand user instructions for HIV self-testing kit, read test results and take the cheek swab as illustrated in Figure 2and

Discussion
The study finding showed that there were significant differences in testing rates (P=0.01) between arm one and two while arm three was superior to arm two with significantly higher testing rates (P<0.001).
There were high testing and acceptability rate among the men who got the test kit and the improved partner invitation letter in arm three compared to other two study arms (83%, 37% and 28% respectively).
KAIS 2012 showed overall, 71% of respondents would be willing to use such a kit if it were to be available to them, more men (74%) than women (67%) indicated willingness to use an HIV self-test kit [1].
This study findings aligns with existing literature on HIV self-testing, which suggest users (including the general population) may prefer oral fluid-based HIV Rapid Diagnostic Test (RDT) to finger stick/whole blood-based HIV RDT because they are reportedly easier to perform and are perceived as less painful as long as they are provided with sufficient HIV prevention information [17]. This will provide an opportunity for HIV testing among clients who dislike needle pricks.
Male partners of clients in arm three (those provided with improved letter and self-testing kits) were twelve times more likely to test for HIV compared to men in arm one (control). During enrolment women in arm three and two got brief counselling sessions on how to counsel the partner, discuss HIV and demonstrate the kits use to the partner.
Several studies have demonstrated that when given the opportunity to participate in sexual and reproductive health programmes, such as family planning and the PMTCT programmes, men preferred to be positively involved in promoting the health of their families and communities [9]. There were no incidence of GBV or IPV reported during the study period. In arm three where a test kit was provided,

Competing interests
The authors declare no competing interests.

Authors' contributions
Tom Marwa is the lead author in this manuscript, he was involved in the study design, data collection and writing of the manuscript; Sarah Karanja contributed in analysis and interpretation of data. Justus Osero and Alloys Orago reviewed the manuscript. All authors approved the final version of the manuscript submitted to the journal.

Acknowledgments
We acknowledge ANC mothers and other organizations for participating in this study; APHIAPLUS KAMILI project, Kenyatta University School of Public Health and 14 health facilities where the study data was collected. Table 1: number of women, men and couples recruited and followed up in each of the three study arms