Koekenbier, Rik H. MSc*; Davidovich, Udi PhD*; van Leent, Edwin J. M. MD, PhD†; Thiesbrummel, Harold F. J. MD‡; Fennema, Han S. A. MD, PhD‡
SYPHILIS IS A REEMERGING SEXUALLY transmitted infection (STI) among men who have sex with men (MSM). Between 1998 and 2004 the annual number of infectious syphilis cases in Amsterdam increased from 35 to 240. In 2004, 84% (202 of 240) of the new infections occurred among MSM.1 To respond to the current epidemic of syphilis among MSM, the Amsterdam Health Service sought a way to lower the threshold for syphilis testing. To achieve this and to increase the test capacity, it introduced a Web site, accessed at syfilistest.nl, which allows MSM to make an appointment for a syphilis test with an accredited testing laboratory, anonymously, free-of-charge and without a referral person (such as a general practitioner). After a week, test results and advices can be obtained online. Participants who test positive are invited to the STI clinic for further examination and treatment.
Online-mediated testing for MSM is feasible because the Internet is a very popular medium in this target group. Data of the Amsterdam Cohort Studies2 and the Amsterdam monitoring study3 suggest that 80% to 90% of gay men increasingly use the Internet. Other studies show that MSM use the Internet to search for and meet sex partners,4 and are also willing to be informed about health issues on the Web sites they visit.5,6 Similar to a syphilis Web site in San Francisco,7 the Amsterdam site is the first of its kind in Europe.
We assessed the feasibility of the online procedures and the efficacy of online-mediated testing in detecting MSM with syphilis infections, especially those needing treatment (early and late syphilis). Regarding efficacy, we compared data from the Web site with data of the Amsterdam STI clinic. We also investigated whether the online test site would attract MSM who never visited the STI clinic in Amsterdam before. Finally, we examined whether a limited bannering period would create enough awareness for the Web site to remain functional after the bannering period was ended.
Our study targeted sexually active MSM, who live in the Amsterdam metropolitan area. During the study period of 15 months, from October 2004 till January 2006, participants were recruited actively during the first 4 months and passively during the remaining 11 months. Active recruitment consisted of online banners (Internet advertisements) on chat sites and lifestyle sites that are popular among gay men (Fig. 1). To recruit gay men in Amsterdam the banners on chat sites were placed in Amsterdam chat boxes, but the banners placed on lifestyle sites were visible to everyone. The banners displayed pictures of syphilis such as the treponema bacteria or syphilis ulcers with text warning about the return of syphilis. Similar banners were successful in recruitment of MSM for online-mediated syphilis testing in San Francisco.7–9 During the passive recruitment, no bannering or advertising took place, and knowledge of the site was passed predominantly through word-of-mouth communication. Demographic data on each participant consisted only of self-reported date of birth and residential postal code. To establish Web site efficacy in getting MSM tested, we compared the online visitors with visitors of the STI clinic. All MSM who visit the STI clinic receive a standard set of STI tests, including a syphilis test. Therefore, all MSM attending the STI clinic in the same period as the online study were included in the comparison.
An overview of the procedures is schematically presented in Figure 2. Once participants clicked on the banners, they were directed to the entry page of the Web site, which was divided into 2 fields: an information field on the left side of the screen and an action field on the right. The information field was designed to motivate men to get tested. Men who did not live in or near Amsterdam were advised to visit their general practitioner or a public health centre nearby their place of residence. The information section described syphilis, its health consequences, transmission routes, symptoms, and treatment options, as well as the ongoing epidemic among gay men. Information was also provided regarding the online testing procedures.
On the right side of the screen men could take direct action by following 3 steps to obtain an online referral for a syphilis test. Steps 1 and 2 were to fill in a username of their choice as well as their date of birth and postal code. In step 3, using a drop-down menu, the participants could choose the laboratory, the day, and time they preferred to get tested.
They were then provided with a referral letter to a testing laboratory, along with a list of addresses of the participating laboratories. Each referral letter carried a unique identifying code generated by the computer. Participants were asked to print their referral letter and the list of addresses before visiting the laboratory. After presenting the referral letter and its identifying code at the laboratory, the participants had their blood drawn for serological testing by treponema pallidum particle agglutination (TPPA) assay. If the TPPA was positive, venereal disease research laboratory (VDRL) test and fluorescent treponemal antibody (FTA) absorption assay were performed.
The laboratory sent the test results to the STI clinic, where they were uploaded online after diagnosis by a medical practitioner of the STI clinic. Participants could obtain the test results online, 7 days after testing, using their unique identifying code. Those who tested positive were requested to visit the STI clinic and were prioritized to confirm the test results by an interview and retesting. Exclusive of the syphilis retest these men also received a total STI check, including tests on human immunodeficiency virus (HIV), gonorrhea, chlamydia, and hepatitis B. If necessary, they received treatment and support with tracing and notification of partners. Participants who tested negative were online advised on the practice of safe sex and regular STI testing (every 6 months). If the test results were not obtained online within 4 weeks, a telephone number was shown that participants could call to obtain their test results.
For comparison of proportions between the bannering period and the nonbannering period and between the online data and the STI clinic data we used a normal approximation with Yates continuity correction. We used Yates continuity correction because of the low numbers in the comparing groups.10
To establish whether the online-initiated testing procedures are feasible, we measured how many of the men tested at the laboratory would subsequently obtain their results online, and how many with positive test results would appear at the STI clinic for treatment. The number of men tested was registered at the STI clinic by the received number of laboratory results. Of those tested, we registered the number that obtained their results online and the number of days elapsed between downloading a referral letter and obtaining the results online. The men who tested positive were registered at the STI clinic. Those who tested positive but did not show up at the STI clinic were defined as “no-shows.”
To assess the efficacy of the Web site in detecting MSM with a syphilis infection, we compared the percentage of tests that were found positive through the Web site with the percentage test that were found positive at the STI clinic during the same period. Men who tested positive were categorized as early syphilis (in this study early syphilis exists of primary, secondary, and early latent), latent (late) syphilis, and old or already treated syphilis, using the diagnostic criteria from the STI clinic.1 Because we were interested in finding the men who needed treatment, the percentages of early and (late) latent syphilis were the most meaningful.
Usage as Affected by Bannering.
The usage of the Web site during the 4 months with bannering versus the following 11 months without bannering served as the indicator of the long-term effect of bannering. Usage was registered as the number of sessions conducted on the Web site. We defined a session as a series of clicks through the site by an individual during one visit. A session is initiated when the visitor arrives at the site, and it ends when the browser is closed or when it automatically deactivates after 10 minutes of inactivity. We also identified how people reached the site by registering the external references, the referring URLs (external web pages), that brought traffic to our site. Sessions without an external reference (when a visitor logged on directly to our site through a bookmark or by typing in the URL directly) were designated as “no-referral.” Furthermore, we registered the number of downloaded referral letters when a participant printed a referral letter. To indicate the rate of testing among the more serious users who completed their online test request, we calculated the percentage of MSM tested from the total of downloaded referral letters.
During the study period of 15 months 25,671 Web site visits were registered. These visitors downloaded a total of 898 laboratory referral letters and of these, 93 men got tested. Ninety percent (90 of 93) obtained their test results from the Web site. Men collected their test results on average 10 days (SD 8.7) after downloading the referral letter. Of the 93 men tested, 14 (15.1%) had a positive serology (Table 1). Of these 14 positive men, 4 did not show up at the STI clinic to confirm their test results. Of the 4 no-shows, 1 did not collect his test results online. All 4 had a serology suggestive of an already-treated syphilis or a late latent syphilis (TPPA+/VDRL−), but without an interview and a retest, this could not be confirmed. Of the 10 positive men who came to the STI clinic, 33% (3 of 10) had never visited the STI clinic before.
We used the total online trial data for comparison with the STI clinic data. During the 15 months of trial, 5852 MSM were tested for syphilis within the standard screening program of the STI clinic. The demographic characteristics (e.g., age and zip code) did not significantly differ between the online and clinic samples. The clinic group had a higher percentage of men with a positive serology (TPPA+) compared with the online group; 21.9% (1284 of 5852) versus 15.1% (14 of 93). However, the online group included more men with an early syphilis or late syphilis (Table 1). Combining early and late syphilis, it was found that the overall percentage of men that needed treatment was higher in the online group (50%, 7 of 14 online vs. 25%, 319 of 1284 STI clinic, P <0.03). The percentage of men with an already-treated syphilis was higher among men tested at the STI clinic than among men tested through the Web site (75% 965 of 1284, vs. 21%, 3 of 14, P <0.01).
Usage as Affected by Bannering
In the first month of the project, 6622 Web site visitors were counted. This number declined to 2264 at the end of the 4-month bannering period. In the nonbannering period of 11 months, the number of sessions was relatively constant, with a monthly average of 679 (Fig. 1).
Of the visitors in the bannering period, 65% (n = 11,465) were recruited to the Web site by banners, whereas 13% (n = 2333) were no-referral; they directly typed in the URL of the Web site. In the nonbannering period, 55% (n = 4114) of the visitors were no-referral, whereas the rest were linked through from Web sites referring to syfilistest.nl with a link, or used a search engine with words like syphilis and STI testing.
The percentage of people who visited the Web site and downloaded a referral letter was higher in the bannering period than in the nonbannering period (4%, 761 of 18194 vs. 1.8%, 137 of 7477 P <0.01). In the overall period 10.4% (93 of 898) of the men who downloaded a referral letter got tested. Interestingly, the percentage of men getting tested after they downloaded a referral letter was higher in the nonbannering period than in the bannering period (22.6% 31 of 137 vs. 8.1% 62 of 761, P <0.01). The percentage of people who visited the Web site and were tested was almost the same in both periods (Table 2). Nevertheless, the monthly average of men getting tested was higher in the bannering period compared with the nonbannering period: 16 versus 3.
The evaluation of the feasibility and efficacy of syfilistest.nl showed that the Web site detected a significantly higher percentage of men who need treatment for syphilis (early or late syphilis) than did the routine STI screening program of the STI clinic. This online approach therefore may show promise in detecting patients with a possible infectious syphilis. However, in a similar study conducted by the San Francisco Department of Public Health,7 the percentage of men found with an early syphilis was lower in the online sample compared with the STI clinic (2.3% vs. 3.0%).
The process of arranging a test online and obtaining the test results online seems feasible, because 96% of those tested obtained their results within a mean period of 10 days (SD 8.7) from downloading a referral letter. A small minority of those who tested positive did not visit the STI clinic for further examination and treatment, but none of those had a serological profile indicating an infectious syphilis. It is possible that some of these no-shows actually did visit the clinic but failed to register as users of the online site or, alternatively, they sought treatment elsewhere, e.g., by a general practitioner. These scenarios are likely, because these men went through all the effort of blood-testing and checking their result online.
The average of 4548 sessions a month in the pilot period, of which 65% were attracted by banners placed on Amsterdam chat sites and Web sites for gay men, shows that interest for the Web site was created within the target group using minimal effort. Even during the nonbannering period of 11 months, the mean number of sessions a month was 679. This constant number of visitors for almost a year, without any advertising and with 55% of those visitors typing in the URL directly suggests that the existence of the Web site was sustained by word-of-mouth. However, the short period of bannering was not sufficient to evoke more extensive use of the site as we had hoped. To create more Web site awareness and to increase the number of tests performed, one should probably engage in longer and more intense promotion activities than a few months of bannering we conducted. Results of a study in San Francisco7 showed that, in addition to bannering, a print campaign (bus posters, palm cards, and so on) could increase the number of tests performed. However, print campaigns and banners are expensive and with the data presented in this study we cannot confirm that they will elevate the usage over a longer period of time.
Although more men downloaded a referral letter in the bannering period, the percentage of men getting tested after they downloaded a referral letter was significantly higher in the nonbannering period. The reason for this could be that visitors to the site in the nonbannering period were actively searching for information or testing options for syphilis and therefore had a strong intrinsic motivation to test. In neither period did we collect reasons for testing or downloading referral letters, so such hypotheses could not be confirmed.
Although our first aim was to get as many men tested as possible, we were concerned that the low threshold of our online approach would attract the worried-well. However, the high percentage of users who needed treatment implies that we managed to reach a good proportion of the right population. In addition, one third of men who got positive results online and reported to the STI clinic had never visited such a clinic before. This suggests that the online program could reach a population that normally would not visit the STI clinic.
Based on the success of syfilistest.nl we have now expanded the online test services, which are now available for MSM. Through a new national testing and prevention site, MantotMan.nl men can now arrange in a similar manner to that of the syphilis site a complete testing package for syphilis, chlamydia, gonorrhea, and HIV with an option to opt-out for HIV testing. In this way we can offer nearly the same standard of care and high-quality STI screening for MSM online as it is offered to them offline through STI clinics. To minimize the chance of missing early infections men online are first screened for eligibility to use the service: only men without symptoms or who are not notified by a partner with an STI can make use of the online testing package. All others are directly referred online to the STI clinic where they receive a physical examination and/or treatment. This online service could help to further alleviate the heavy workload at the STI clinic, because no precious time must be spent on men with higher chance of negative serology.
There are a number of limitations to our study. First, the number of sessions on the Web site might include multiple visits, and we therefore might have overestimated the number of individuals who downloaded referral letters. This could actually have caused us to underestimate the success rates for syphilis testing, because men could have downloaded several letters before using one for testing. It could be, therefore, that a higher proportion of men who downloaded a referral letters actually tested for syphilis than we have estimated. Second, for interpretations of Web site efficacy, it should be kept in mind that our data set has the limitations of a small sample size. Yet despite the small numbers, the site found a higher percentage of men with syphilis infections who required treatment, compared with the STI clinic.
This is the first study to examine an Internet approach to syphilis testing in Europe. It shows “Online testing” to be a potentially useful public health tool in the fight against syphilis and potentially other STIs. The Internet clearly emerged as a medium where people can find information about STIs and take direct action to address their concerns. The online testing approach should remain complementary to the existing prevention and screenings options, and although it is not meant to replace standard of care, we have shown it to possess unique added value.