Of 475 men screened for study eligibility, 215 were from Kalangala while 260 were from Buvuma. After eliminating those who were not eligible (Figure 2), 189 (88%) of the social network members in Kalangala and 211 (81%) of those in Buvuma were enrolled into the study for a total of 400 respondents. A majority of those who were not eligible for study enrolment in both districts were recent HIV-negative testers (i.e. last tested for HIV within three months).
Baseline characteristics
Table 1 shows the characteristics of the 400 social network members that were enrolled into the study at baseline. A greater majority of the men in both districts (272, 67.9%) were aged between 18 and 34 years; nearly two-thirds (257, 64.2%) had primary education, while more than half (233, 58.2%) were currently married. Nearly 163 (41%) of social network members could not read text written in Luganda, the main local language spoken in their community; with a higher proportion of men in Buvuma being unable to read text written in the local language than men in Kalangala district (Buvuma: 110, 52.1% vs. Kalangala: 53, 28.0%; P<0.0001). Slightly over half of the male social network members (227, 56.7%) were engaged in fishing or fishing-related activities. A majority of the respondents (328, 82.0%) had ever tested for HIV.
Feasibility of the intervention
Figure 3 shows the number of HIV self-test kits given to the peer-leaders by the study team and the number of kits that they gave out to their social network members. Overall, peer-leaders received 800 kits to distribute to their social network members. Within two months, peer-leaders had distributed 782 kits (97.7%) to their social network members.
At the subsequent follow-up visit in September 2022, we interviewed 361 men (90.2%) of the 400 social network members enrolled at baseline. When asked if they received any kits from their peer-leaders, nearly all men (355, 98.3%) responded in the affirmative. Of these, 176 (49.6%) received the kits from the distributor’s home, 95 (26.8%) had the kits delivered to their homes, 59 (16.6%) had the kits delivered to their own work places, and 15 (4.2%) picked the kits from the distributor’s work place. Male social network members from Buvuma were almost three times as likely to report that they received kits from the distributors’ homes than those in Kalangala (odds ratio [OR] = 2.79; 95% confidence interval [95%CI]: 1.77, 4.39) but those in Kalangala had twice the odds of having the kits delivered to their homes than those in Buvuma (OR = 2.07; 95%CI: 1.24, 3.45). After receiving the kits, 265 (74.6%) reported that they were contacted by their peer-leaders to check if they had used them to self-test for HIV while 183 (51.5%) reported that they were provided with information on the existing referral networks that they could use in case they self-tested HIV-positive and needed to link to HIV care. When asked about how comfortable they were to receive HIV self-test kits from their peer-leaders, 312 (87.9%) reported that they were comfortable or very comfortable to do so. When asked if they would recommend that trained peer-leaders continue to distribute HIV self-test kits in the fishing communities, 348 (98.0%) responded in the affirmative with a similar proportion of men in Buvuma (177, 97.8%) and Kalangala (171, 98.3%) responding in this way (Table 2).
Acceptability of the intervention
Table 3 shows HIV self-testing uptake among men who received HIV self-test kits from their peer-leaders. Of the 355 men who reported that they received HIV self-test kits from their peer-leaders, 352 (99.1%) reported that they used them to self-test for HIV. Of these, 159 (45.2%) used the kits on the same day of receiving them; 75 (21.3%) used the kits on the next day, while 52 (14.8%) used the kits within 2-4 days after receiving them. Of those that self-tested for HIV, 104 (29.5%) reported that they self-tested under the direct supervision of someone else; however, 340 (96.6%) of all self-testers performed the test on their own.
Table 4 shows the HIV self-test results read by both the respondent and members of the study team. In both districts, 51 men (14.5%) were HIV-positive; 297 men (84.4%) were HIV-negative, three men (0.8%) had indeterminate results while one (1) man (0.3%) reported that they did not know or remember their results. HIV prevalence was higher in Kalangala (34, 19.5%) than in Buvuma (17, 9.5%).
In general, HIV prevalence increased with increasing age but decreased with increasing level of education. For instance, HIV prevalence increased from 4/109 (3.7%) among those aged 18-24 years to 25/87 (28.7%) among those aged 35-44 years and 9/30 (30.0%) among those aged 45+ years. On the other hand, HIV prevalence was 6/30 (20.0%) among those with no education; reduced to 34/219 (15.5%) among those with primary education and was 11/99 (11.1%) among those with post-primary education. By marital status, HIV prevalence was highest among those who had ever married (19/52, 36.5%) followed by those who were currently married (26/204, 12.7%) and was lowest among the single/never-married (6/92, 6.5%). By occupation, HIV prevalence was highest among those engaged in fishing (27/130, 20.8%) and fishing-related activities (12/66, 18.2%) than in those engaged in business/other commercial activities (7/62, 11.3%) or other occupations (5/90, 5.6%).
Preliminary effects of the intervention
Of the 51 men that tested HIV-positive, 16 men (31.4%) were newly diagnosed while 35 (68.6%) were already aware of their HIV-positive status. Of the 16 first-time HIV-positive testers, 14 men (87.5%) went for confirmatory HIV testing, seven (50%) were confirmed as HIV-positive, while five (71.4%) were initiated on ART (Table 5). The highest proportion of newly diagnosed men was observed among those aged 18-24 years (3/4, 75.0%), those who were singe/never-married (4/6, 66.7%), those reporting 2+ sexual partners (12/31, 38.7%) and those who had never tested (3/9, 33.3%). The proportion of newly diagnosed HIV-positive men was higher in Buvuma than Kalangala (7/17, 41.2% vs. 9/34, 26.5%; respectively) but this difference was not statistically significant (OR=1.94, 95%CI: 0.47, 7.81).