The search used 5 electronic databases that specialize in health research: MEDLINE, EMBASE, CINAHL, Web of Science, and PsycINFO. The search terms were developed through experimentation with the support of a specialist librarian, using a population, intervention, context, and outcome model adapted for qualitative research [26,27]. This resulted in the following 4 search term categories: (1) Population: individuals with or at risk of STIs (eg, chlamydia); (2) Intervention: testing or screening for STIs (eg, test); (3) Context: online (eg, internet); and (4) Outcome: qualitative perception or experience (eg, interview).

An example list of search terms is included in Multimedia Appendix 1.

The search period spanned from January 1, 2005, to December 31, 2018. We chose 2005 because this was when internet-based testing emerged, and fewer than half of UK households had access to the internet prior to this period [28,29]. The search was limited to studies published in English, as there were insufficient resources available to arrange translation.

Studies were eligible for inclusion if they (1) reported user (or prospective user) perceptions or experiences of any aspect of internet-based testing for STIs, either hypothetically or in practice, and how this affected whether users might access it; (2) collected the relevant data using qualitative methods, including the qualitative component of mixed methods research and free text responses to questionnaires; (3) were published in English in peer-reviewed academic journals between 2005 and 2018; and (4) collected data in countries defined as high-income countries by the World Bank.

Inclusion was limited to high-income countries due to their similar STI profiles, health care infrastructure, and rates of internet access [30,31]. Self-sampling for HPV was not included within the scope of this review, as it is normally conducted in the context of cancer screening and is not usually a component of STI screening [32].

The search results from each database were combined and duplicates were removed. Initially, all studies were screened via their titles and abstracts to determine eligibility, with potentially eligible studies subsequently being read in full. These stages of the screening process were both undertaken in full by the lead reviewer (TS), with a second reviewer (IK) screening a random 20% (267/1332 for title and abstract screening; 16/79 for full-text screening) of studies at each stage to determine interrater reliability. Any disagreements were resolved through discussion, and a third reviewer was brought in if consensus could not be reached. Reviewers always erred on the side of inclusion to ensure all relevant data were identified.

The quality of the studies was critically appraised using the Critical Appraisal Skills Programme checklist for qualitative research [33]. The checklist provides a holistic overview of the rigor of research, and any studies found to be methodologically weak were planned to be included in a sensitivity analysis once the synthesis was complete.

The included studies underwent thematic synthesis following an adapted version of a framework developed by Thomas and Harden [34] for qualitative systematic reviews. The results section of each study was uploaded to NVivo (QSR International), with each line of text that addressed the research question being coded according to its meaning and content. All relevant data were coded at least once, and once this process was complete, codes were grouped into descriptive themes that captured the meaning of multiple initial codes. These were in turn developed into broader analytical themes, which go beyond the findings of the included studies. The lead reviewer (TS) undertook all analysis, with support and interrater reliability being provided on 20% of included studies by a second reviewer (JW) and feedback being provided from the lead reviewer’s supervisors (FG and JR).

A total of 11 studies met the inclusion criteria for the review. The initial search identified 1735 studies for screening, which reduced to 1332 studies once duplicates were removed. The search and screening process is outlined in Figure 1, along with the number of studies removed at each stage in the process.

The initial rate of agreement between reviewers was 96.2% for title and abstract screening and 93.3% for full-text screening.

An overview of the included studies is provided in Table 1. All were published between 2006 and 2018. One was conducted in Australia, with the remainder split equally between the United Kingdom and United States.

Of the 11 studies included, 2 studies reported the experience of users who had accessed internet-based testing, with the remaining 9 exploring perceptions among potential users of hypothetical services. Of these, 2 studies explored the views of women, 2 explored the views of men who have sex with men (MSM), and 2 explored the views of men whose sexual orientation was unspecified. The remaining 3 studies explored the views of both men and women.

Young people aged younger than 30 years were exclusively recruited to 4 of the studies, including one of the studies that explored the views of women, one that explored the views of men, and one that explored the views of MSM. This latter study also had exclusively Black participants; another 4 studies reported over 45% of participants as people of color.

The results of the critical appraisal are outlined in Figure 2. All but 3 [35,42,43] of the studies met at least 9 of the 10 appraisal criteria, with the remainder [17,36-41,44] meeting at least 6. All of the studies were deemed valuable, had clear aims and statements of findings, and had appropriate methodologies, study designs, and data collection methods. However, most did not provide enough information to determine whether they met all of the criteria, with only 3 [36-38] reporting sufficient consideration of the researcher-participant relationship. Separately, 3 studies [35,38,42] described their analysis with limited detail, which meant it was not clear how themes were derived from the data and the methods were not replicable.

A total of 12 themes were identified, which were organized into 4 categories. These are outlined in Table 2.

This review and synthesis identified a wide range of perceptions and experiences of internet-based testing held by users and potential users. There was a clear finding that internet-based testing is attractive due to its convenience and the fact that it alleviates concerns around stigma associated with being tested in a clinic, and many participants were drawn to it because it allowed them to avoid elements of face-to-face testing that they disliked. However, there was also a concern among some participants relating to the privacy of internet-based testing, the self-sampling process, and the fact that internet-based testing would be missing positive aspects of face-to-face testing. There was no universally accepted medium for results delivery, but preference was largely motivated by perceptions of convenience and concerns over privacy. Overall, internet-based testing appears to be acceptable despite some reservations expressed about it.

This review is the first attempt we are aware of to bring together qualitative data that relate to this growing medium of STI testing. The inclusion of data from 1115 participants—a substantial number for qualitative research—is a notable strength, as is the fact that the data were synthesized using a well-established and transparent method. All of the studies were assessed to be of satisfactory quality, imbuing confidence in the results, and the fact that all studies were undertaken in 3 countries with similar socioeconomic profiles enhances the generalizability of the results.

The synthesis was limited by the small number of studies eligible for inclusion, and the quantity of data varied widely between these studies. This meant that although there was a relatively strong consensus among included studies on most themes, it was difficult to draw definitive conclusions on subpopulations. Only 2 of the studies collected data from people who had actually used internet-based testing, for example, and one of these collected data only on their experience receiving results, which limited the distinctions that could be drawn between their findings and the findings of the 9 studies that collected data from potential users of hypothetical services. This similarly affected the findings on perceptions held by different sociodemographic populations, and the review may also have benefited from studies undertaken in a wider number of geographical settings and published in languages other than English. The included studies are also unlikely to have recruited people from vulnerable populations, such as homeless people or those with serious mental health issues, meaning these populations’ views may not have been captured in the review. The screening of studies for inclusion was limited, as only a proportion of results were screened by 2 authors, and although including studies published between 2005 and 2018 ensured that all relevant data were included, this relatively long time frame means that findings from some of the earlier studies may now be less relevant due to considerable changes in internet usage over this period.

Internet-based STI testing differs from many other digital health interventions (such as telehealth, patient portals, and remote monitoring), as it does not involve two-way communication with clinical staff and is designed as a one-off engagement with the health care system rather than part of the management of a long-term condition. This review offers the opportunity to explore whether users of STI-testing services interact with the service in the same way as those using other digital health interventions, and one notable difference is the fact that internet-based testing is highly associated with convenience. This contrasts with the findings of a review by O’Connor et al [45] about the factors affecting more general engagement with digital health interventions, which found that many people could be deterred from accessing interventions if they felt they lacked the time or energy to do so, which may be because such interventions are frequently targeted towards individuals who are expected to have more sustained engagement [46].

However, other findings were similar to those found for alternative digital interventions, including the finding that some participants had privacy concerns over electronically providing the personal data required for access and that other participants appreciated the anonymity offered by digital interactions over face-to-face ones [45,47]. This concern over the provision of personal data parallels evidence on engagement with digital media unrelated to health care, such as social media, which regularly forces users to weigh privacy concerns against perceived benefits of use [48,49]. The normalization of social media can frequently lead users to overlook their privacy concerns, however, and this is unlikely to apply in the same way when people access internet-based testing, given that it is less likely to feature as often in their day-to-day lives [48-50].

The stigma associated with sexual health also undoubtedly heightens privacy concerns held by users of internet-based testing [23]. The concerns over privacy and anonymity identified in this review simultaneously highlight the potential of digital interventions to overcome stigma as a barrier to accessing health care—in the context of both sexual health and other conditions—and the underresearched phenomenon of the role that stigma plays when using internet-based interventions [23,45,51,52]. Our finding that many people are concerned about aspects of the internet-based testing process that could allow others to know they had used sexual health services emphasizes the role providers have in mitigating that risk and suggests a need for further exploration of the role that stigma plays when individuals access, or consider accessing, internet-based health care. It is noteworthy that this finding was prominent in studies with ethnically diverse populations and that women were found to be particularly concerned about the stigma associated with clinics and interacting with clinic staff, as users of internet-based testing are disproportionately women and White [53-55].

The concern identified that internet-based testing may deprive users of important aspects of face-to-face testing is significant, particularly as internet-based testing is one of the few health interventions that allows users to have no direct contact with clinical staff. Although it did not appear to deter the majority of participants from the prospect of using internet-based testing, it aligns with evidence from other contexts that digital health care is seen as supplementary and that service users are willing to use it provided it does not replace face-to-face care [51]. This may also be the case for the delivery of test results remotely, a topic on which limited evidence exists and to which this review therefore makes a notable contribution [56]. Our finding that people who had used internet-based testing were satisfied to receive their results by SMS is in agreement with other research, which has found high levels of satisfaction among patients who have received results electronically [57]. That we found no consensus on preferred media for results delivery among potential users corresponds with conflicting data from other sexual health studies, suggesting that service users’ preferences when conceptualizing hypothetical services may not be an accurate predictor for what they find acceptable when they start using them [58-60].

Providers of internet-based testing may wish to emphasize the approach’s comparative convenience and privacy compared with face-to-face testing in order to improve uptake, as these appear to be the most appealing aspects of the service. Uptake may also be improved through attempts to alleviate concerns around self-sampling, for example, by providing reassurance about discomfort and emphasizing that the sensitivity and specificity of self-sampling is comparable to samples obtained in a clinic. It may also be worthwhile for providers to consider patient confidentiality in results delivery, for example, by ensuring that text messages are worded so they could not inadvertently reveal a diagnosis in a phone notification, and to ensure that patients receive adequate signposting to support if results are delivered electronically.

This study has identified a wide range of perceptions and experiences of internet-based testing by actual and potential users, including both positive and negative comparisons with clinic-based testing. There is a clear need for further qualitative research exploring in depth the experiences of people who have accessed internet-based testing, given the paucity of data on this, and for qualitative research on internet-based testing that explicitly gathers the views of populations that are at high risk of STIs or that have testing behaviors that require more in-depth understanding. There would also be value in further research on attitudes towards communicating results, given that no consensus could be found on a preferred medium.