In this study, we found that web-based HIVST have reached many first-time testers similar with previous studies among cis-MSM and TGW [25-29]. Reporting and linkage to care and prevention rates were high but ART and PrEP initiation were sub-optimal. While a minority (12.5%) asked for assistance, a higher proportion of testing reactive for HIV among those who opted for DAH. Choosing online approach was associated with lower odds of DAH. Almost half (48.16%) were willing to share kits, more likely among those ever tested and residing in urban area. Stricter community quarantine measures did not affect reactivity rate, but it was associated with increased likelihood of opting for DAH and willingness to distribute.
It is interesting that there seems to be no difference in reactivity rate between first-time and ever testers, which disagrees with a previous study among individuals at high risk for HIV, only 34.2% of whom were from KP, in Senegal . In the Philippines, all of those who come for HIV testing are required to undergo risk reduction counselling [30, 31]. Our finding suggests that current intervention may have had marginal impact on reducing risk as those who had history of HIV test are as likely to test reactive as first-time testers. Testing is a good avenue to educate and previous studies have shown that higher HIV knowledge is associated with being ever tested among cis-MSM and TGW [32-35]. Amidst the tendency for counselling and education to be more self-driven with HIVST, risk reduction messaging could be strengthened.
Some implementers have argued to maximize the unassisted nature of HIVST , especially as studies have shown that unassisted HIVST appears to be acceptable and feasible [36, 37]. However, a minority in our cohort (12.5%) opted for DAH. Our study shows two interesting findings. Firstly, reactivity proportion among those who opted DAH was significantly higher than those who did not, similar to a study among TGW in Thailand , albeit not consistent with cis-MSM in the same study  and with a study done among people at high risk for HIV, 60.4% were females, in Congo . Despite the inconsistencies, the anxiety regarding linkage to care and the desire for assistance have been reported among cis-MSM and TGW in the country , and other KP elsewhere . Additionally, some transgender people have been documented to be needing some form of support during HIVST, mostly from social networks . Secondly, while it may be intuitive that testing for the first-time is associated with higher odds DAH, similar with previous studies [6, 23, 40], it was the opposite in our bivariate model. It may be possible that participants were enticed with the privacy and independence HIVST offers. Upon controlling for other variables, however, the association appears to be insignificant, although trending towards lower odds of DAH, similar with previous studies among TGW , partners of female sex workers, and adolescents [41-43]. Taken both altogether with DAH’s association with better retention rate  and higher ART initiation rates , there is benefit in offering DAH albeit an unpopular choice. As foresight, as the Philippines might roll-out oral fluid-based test, assistance may be warranted as Filipino KPs are accustomed to blood-based tests and may raise unfamiliarity.
Participants in the online approach being less likely to ask for DAH may be explained by the cis-MSM and TGW’s preference for privacy, confidentiality, and convenience in HIVST . Although web-based HIVST approaches provide the aforementioned , offering DAH may deal with issues of perceived lack of assistance or support in virtual platforms [46, 47]. Conversely, our findings suggest that choosing off-line approach was associated with higher odds of DAH. This may be explained by the implementers being KP-friendly CBOs, which may have alleviated the unfriendly nature of some HIV testing sites known to lead MSM to opt for otherwise, especially among those unaware of their status . KP-friendly community-based interventions have been previously evaluated to facilitate uptake of services across the HIV care cascade in the Philippines [8, 49] and elsewhere [50-52].
Secondary distribution have been shown to increase the reach, positivity yield, and cost-efficiency of HIV testing among cis-MSM [53, 54], including those unsure of their gender identity . Peer involvement could synergize with other facilitators known to HIVST . Similar to previous studies which showed decreased distribution [24, 56], we found that being never tested were associated with lower likelihood of wiliness to distribute. This may be explained by the relatively young cohort, with median age of 26 years old being not too far away from the median age for first HIV test among cis-MSM and TGW based on the national biobehavioral serologic survey in 2018 . Moreover, being younger (<24 years old) was found to be associated with lower likelihood of willingness to distribute in the bivariate analysis, although non-significant upon adjusting for other variables. This may be due to the prevailing sociocultural factors leading to concepts like sex, especially non-conforming sexual acts, as taboo, and is only for adults [57-62], making sexually transmitted infection and HIV prevention interventions elusive to young people, especially among sexual and gender minorities .
To our knowledge, this is among the first study on HIVST to control for both location and time. In terms of location, residing in urban areas was associated with increased odds of willingness to distribute and this may be explained by the dense clustering of KP , higher access to queer culture  and HIV education , and higher acceptability to HIV interventions  compared to rural areas. In terms of time, characterized by the extent of community quarantine protocols, it is expected that willingness to distribute was higher during maximal restrictions. This may be explained by the Filipino value of “bayanihan”, communal effort, which have been shown to be important for resiliency in disasters [66-68]. However, the actual incidence of secondary distribution decreased in other countries amid the COVID-19 pandemic . Furthermore, surprisingly, stricter restrictions were associated with higher likelihood of DAH in our study. Studies on the impact of COVID-19 on the HIV service delivery in the Philippines have been limited and we could only speculate that the perceived limited access to healthcare services amid a time of public health crisis may have reinforced dependence with health providers. It was also suggested that DAH during the COVID-19 reinforces ensuring support, linkage, and retention . Both factors considered, however, emphasize the fact that to fully provide differentiated service delivery, context must be considered .
Our study contributes to the growing body of evidence on the feasibility, acceptability, safety, and impact of web-based interventions to improve HIVST uptake and linkage to appropriate services. Likewise with one systematic review , the intervention increased uptake of HIV testing, even among populations difficult to reach and first-time testers. However, unlike the said review , there were issues with linkage to appropriate services. In terms of ART initiation among those who tested reactive, as with a previous trial of web-based HIVST among cis-MSM and TGW in Thailand, ART initiation has been limited, which was attributed inability of some CBOs in Thailand to initiate ART . This is opposite with a study with Vietnam where CBO-led HIVST and treatment services, some with aided with online intervention, led to high confirmatory and ART initiation rates at 90% . In our study, only one CBO was capable of starting ART among their clients. Although linkage to confirmatory testing is high (87.9%), its considerable difference with ART initiation (60.3%) may be explained by logistical issues brought by the COVID-19 restrictions. This reinforces the idea of decentralization of HIV services to KP-friendly CBOs. In terms of PrEP initiation among those who tested non-reactive, while previous studies found improved PrEP uptake in web-based HIVST interventions [71, 73], PrEP initiation in our cohort was minimal at 0.3%. PrEP was only nationally rolled-out in 2020 and, unlike ART, it is neither free nor covered by any health insurance. Cost has been consistently determined as a significant barrier to PrEP uptake . Moreover, likewise with ART, this is a biomedical intervention that may have had logistical challenges to deliver during amid the COVID-19 pandemic. Nevertheless, upscaling uptake of prevention is crucial especially that it was found that cis-MSM and TGW sustained some high risk sexual behavior even during the COVID-19 pandemic .
The primary strength of this study is the high reporting rate partly due to its web-based delivery (97.5%), similar with other web-based HIVST interventions , which allowed many data points to be used to explore associations and ensured precision on how the data were collected. Furthermore, to our knowledge, this is the first association study to consider the potential influence of quarantine restrictions on HIV service delivery in the Philippines.
Meanwhile, it is important to acknowledge some study limitations. Firstly, although we have reassured confidentiality, the highly sensitive nature of the data collected may have been influenced by social desirability. Moreover, likewise with a previous study , the willingness to distribute HIVST kits were collected at baseline and, hence, may be influenced by the uncertainty of their HIV status. Secondly, the risk of web-based convenience sampling may have led to self-selection. Although some have been invited offline and the models controlled for this variable to balance this risk, both techniques still have inherent disadvantages. Generalizing the findings of our study must be done with caution. Lastly, there are limitations of in the use of stepwise backward elimination. Although it prevents overfitting and allow different combinations of variables [75-78], there are considerable variance when different samples are used  and there is potential for inappropriate variables to be included in the model [75, 76, 78]. We dealt with these by ensuring there is considerable number of events per variable [76, 79] and exploring a priori predictors, respectively. Thus, we are confident that are models predict our outcomes within the context of our study.
Although we have established in our study the efficacy and safety of web-based HIVST in terms of reaching and testing, there remains to be gaps in terms of improving its linkage to appropriate services. Amid the inherent challenges of online-to-offline transition, further studies may explore online strategies for linkage and retention.