With roughly 238 million internet users in the United States and over one billion users worldwide,1 the internet is a significant part of daily life and has provided a new environment for interaction among internet users. Included in these interactions are opportunities for seeking and meeting sexual partners. The use of the internet as a venue for meeting sexual partners first came to the attention of public health officials after an outbreak of syphilis in San Francisco was traced back to an AOL chat room used predominately by men who have sex with men (MSM).2 Since then, several studies have found that those who seek sex partners online are at greater risk for sexually transmitted infections (STI) and HIV infection.3^–7 However, most studies have used markers for STI/HIV rather than infection status itself, including number of sex partners,4 unprotected (anal) intercourse5 7^–10 or history of STI.11 12 Furthermore, in many of these studies, risk behaviours have been compared between people seeking sex online and people who were not. As the latter group more than likely includes individuals at generally lower risk, online sex seeking may have been a marker, rather than the cause, of risky behaviour. Finally, most of the online sex partner research has been conducted among MSM, as this behaviour is more prevalent in this population.

It is unclear from these studies whether seeking sexual partners on the internet is simply a marker of high-risk sexual behaviours, as we have previously suggested,4 or whether it actually increases one’s risk of acquiring an STI. To our knowledge, no study has evaluated the association between seeking sexual partners online and current STI status. To help elucidate these issues, we analysed data from patients attending an urban STI clinic with the aim of examining the association between self-reports of online sex partnering and current STI, specifically infection with Neisseria gonorrhoeae (GC) and Chlamydia trachomatis (Ct). We postulated that these effects would vary among different sexual orientations and therefore examined each of these groups separately. Finally, to avoid only comparing those who had internet sex-seeking experience with those who never sought sex partners online, we compared those with recent versus those with more distant or non-existent online sex-seeking experiences.

MATERIALS AND METHODS

Study design

We conducted a retrospective case-control study of a convenience sample of patients attending the Denver Metro Health Clinic. We limited our analyses to patients who were tested for either Ct or GC during their clinic visit. Cases were defined as any patient testing positive for Ct/GC between August 2006 and March 2008. Controls were patients who tested negative for Ct/GC during that same period. Throughout the study period patients were asked about sexual encounters with partners met on the internet, our exposure of interest, as part of their routine clinical examination.

We extracted all relevant data from existing medical records of patients attending the STI clinic during the study period. For patients with multiple visits, we included only their first visit in the analysis. All data utilised in this study existed before the start of research and no further contact with patients was necessary. Study staff did not obtain identifiable information about any patient or their sex partner(s). This study was exempt from review by the Colorado Multiple Institutional Review Board.

Main outcome and exposure measures

The main outcome measure was current Ct/GC infection. Ct and GC were detected with a nucleic acid amplification test using strand displacement amplification (Becton Dickinson, Spark, MD. USA). Samples were collected from urine specimens and cervical, rectal or pharyngeal swabs. Patients testing positive for either Ct or GC, at any anatomical site, were considered to have a current infection.

The exposure of interest, sexual encounters with internet partners, was assessed with a single item in the electronic medical record: “Sex with internet partner occurred”: with four response options—never, within the past 4 months, 4–12 months ago and more than 12 months ago. The exposure was evaluated by comparing cases and controls reporting an internet partner within the past 4 months to those reporting an internet partner longer than 4 months ago or never.

Statistical analysis

All analyses were conducted using Stata version 9.0 (StataCorp, College Station, TX, USA). An a priori decision was made to examine the data separately by sexual orientation to account for differences in baseline risk behaviours and potential differences in the use of the internet to find partners. We stratified our data into the following categories: men who have sex with men (MSM), men who have sex with women (MSW) and women. Cases and controls were effectively matched by gender and self-reported sexual orientation. Men who reported having sex with both men and women were included in the MSM group and women who identified as gay or bisexual were included with the women who identified as straight, given their small numbers. Using generalised linear models with log link and Poisson error distribution,13 unadjusted and adjusted risk ratios (RR) and 95% confidence intervals (CI) were estimated to describe the association between recent sexual encounters with internet partners and current Ct/GC infection for each group. Risk ratios are presented given that the outcome is not a rare event.14

Patients’ race/ethnicity, age, number of partners in the past 3 months, history of GC and Ct, and current HIV status, all collected from the electronic medical record, were examined for their potential confounding influence on the main association. Variables which appeared to confound the association between recent internet partners and current infection were included in the starting multivariable model. We used a backwards elimination strategy based on change-in-estimate criteria (less than 10%) to remove variables that did not confound the main association. We developed three separate models by sexual orientation to evaluate the association between recent sex with an internet partner and current Ct/GC infection.

RESULTS

Study population

A total of 23 836 medical records were reviewed, of which 20 180 represented a single or first visit. Of those, 15 103 had a valid Ct/GC result, 15 911 had internet partner information and 15 989 included sexual orientation information, resulting in a final dataset of 14 955 observations. A total of 2802 patients were infected with Ct/GC for an overall prevalence of 19% (table 1). Stratified by sexual orientation, the prevalence of Ct/GC infection was 17% for MSM, 21% for MSW and 16% for women. The prevalence of the main exposure, sexual encounters with internet partners, also varied by sexual orientation. Approximately 40% of MSM patients reported ever having a sexual encounter with a partner they met on the internet with 23% reporting the behaviour in the past four months. The proportion of patients reporting an internet sex partner was much lower among MSW and women, with 5% and 4%, respectively, reporting ever having an internet partner and 3% and 2%, respectively, reporting an internet partner in the past four months.

Patient demographics differed by sexual orientation. MSM patients were predominately white, while MSW and female patients were predominately non-white. A little more than half of MSM patients were under the age of 30 while almost two-thirds of MSW and three-quarters of women were under the age of 30. Roughly 63% of MSM reported having more than one partner in the past three months compared to 45% of MSW and 33% of women. One-fifth of MSM reported a past GC infection and 31% of women reported a past Ct infection.

Bivariable analysis

A total of 339 MSM, 192 MSW and 98 women reported having a sexual encounter with a person they met on the internet within the past four months. Among MSM 25% of currently infected cases reported having a recent internet partner, compared to 1% of infected MSW cases and 1% of infected female cases (table 2).

Restricted to MSM patients, 63% of cases were white and 61% were under the age of 30 (table 2). Approximately 74% of MSM cases reported having more than one sex partner during the past three months compared to 61% of MSM controls (RR: 1.21, 95% CI 1.11 to 1.32). Of the currently infected MSM patients, 25% reported a previous GC infection and 14% reported a previous Ct infection. Roughly 12% of MSM cases were also currently infected with HIV compared to 8% of controls (RR: 1.55, 95% CI 1.05 to 2.28).

Examining MSW patients, 84% of cases were non-white compared to 61% of controls (RR: 1.38, 95% CI 1.34 to 1.42), while 80% of cases were under the age of 30 compared to 61% of controls (RR: 1.30, 95% CI 1.26 to 1.34). Roughly 52% of cases compared to 43% of controls reported having more than one sexual partner in the past three months (RR: 1.22, 95% CI 1.16 to 1.29). Almost 14% of currently infected MSW cases reported having a previous GC infection and 19% reported a previous Ct infection. Only one MSW case currently infected with Ct/GC was also infected with HIV.

For female patients, 73% of cases were white compared to 64% of controls (RR: 1.16, 95% CI 1.11 to 1.21) while 88% of cases compared to 71% of controls were under the age of 30 (RR: 1.23, 1.20 to 1.27). Approximately 40% of cases reported having more than one sexual partner in the past three months compared to 32% of controls (RR: 1.19, 95% CI 1.08 to 1.30). Among currently infected cases, 13% reported a prior GC infection and 36% reported a prior Ct infection. There were three cases of current HIV infection among Ct/GC infected female patients.

Multivariable modelling

Evaluating the exposure by comparing those reporting having sex with an internet partner within the past four months to those reporting an internet partner longer than four months ago or never, the unadjusted estimate of the association with current Ct/GC infection for the MSM model was not significant (RR: 1.12, 95% CI 0.84 to 1.49; table 3). None of the covariates we evaluated appeared to confound the main association. The unadjusted risk of current infection among MSW was significant with the risk of infection for those reporting having a recent internet partner being 0.58 (95% CI 0.39 to 0.87) less than those reporting having a more distant internet partner or never having an internet partner. Adjusting for race and the number of sex partners in the last three months, which appeared to confound the association, did not appear to attenuate the protective effect (adjusted RR: 0.66, 95% CI 0.44 to 0.98). The unadjusted estimate for the female model was also protective although not significant (RR: 0.71, 95% CI 0.39 to 1.29). Adjusting for race did attenuate this association (RR: 0.81, 95% CI 0.45, to1.48).

To evaluate the association between recent internet sex partners and current Ct/GC infection further, we restricted the data to patients reporting ever having a sexual encounter with an internet partner and compared those reporting the behaviour in the past four months to those reporting the behaviour longer than four months ago. The restricted analysis yielded similar results (data not shown). However, the unadjusted estimate for the MSW model was not significant although still protective (RR: 0.70, 95% CI: 0.40 to 1.23).

DISCUSSION

During the past decade, finding sex partners on the internet has become an important topic in the STI literature and is generally viewed as a high-risk sexual behaviour, implying that men and women who have sex with partners found online are at higher risk for the acquisition or transmission of STI/HIV.4 15^–17 However, to our knowledge, there are no reports in the literature that have directly evaluated the association between internet sex seeking and the direct presence of STI. In our study, conducted in an STI clinic where patients are routinely asked about internet sex partners and are routinely evaluated for gonorrhoea and chlamydial infections, we did not find any evidence that this behaviour enhanced the risk for these infections.

There are a number of limitations to our study. First, this was a cross-sectional study of a clinic-based convenience sample. No inferences can thus be made of any cause-effect relation and the study findings may not be generalisable outside the STI clinic setting. It is important to bear in mind that STI clinic patients may differ from the general population with regard to factors that are associated with infection as well as STI clinic attendance, mainly socioeconomic factors.18 Second, all behavioural assessments were conducted in the context of a routine clinical interview that uses a check box format in a computerised medical record rather than a formal behavioural questionnaire. Thus, clinicians may have asked questions in many different ways, potentially limiting the validity of the results. Third, there was only a single question asking whether a person had had sex with a person first met on the internet. As a result, we don’t know what proportion of sex partners was first met online and what specific behaviours were associated with internet partners and non-internet partners. We also don’t know any specifics about the type of internet site. Whether partners were sought for the primary purpose of having sex or if partners were sought for a romantic relationship that may have resulted in having sex subsequently are two very different behaviours.

None the less, our study should raise some doubts about whether internet sex partnering per se conveys STI risk, a doubt that has been raised in a handful of other studies examining HIV risk among MSM.9 10 19 20 We suggest that this behaviour may be rather heterogeneous and that the risk for STI depends on the purpose of the person going online as well as the internet venue used for this purpose. In certain circumstances, such as the highly sexually charged environment of internet chat rooms for gay men or heterosexual “swingers”, where online interactions may be quickly followed by offline sexual encounters, the risk for STI may increase. Still, online interactions in these venues may allow for some safer sex negotiations, including serosorting—that is, making decisions about intended sexual behaviours based on discussions of mutual HIV serostatus.21^–23 Such negotiations may even be more outspoken when online partnering is initiated on dating sites where the primary purpose is to find a more long-term romantic partner. As suggested by a recent study by Padgett, interactions on such sites are often followed by extensive email contact that may include negotiations about sexual interactions as well as preventive intentions (like condom use) before the in-person encounter actually occurs, thus lowering the risk for STI.24

Despite the growing number of wide-ranging internet dating and networking sites in the past several years, a marked differential between MSM, MSW and women with regard to the extent to which they seek and meet sexual partners on the internet remains.25 26 MSM continue to utilise the internet as a primary means for seeking sexual partners compared to MSW and women. It is possible that over the years and with increased utilisation of various internet sites as a means to deliver prevention messages,27^–30 an increased awareness of the importance of safer sex practices is reaching those at highest risk. This increased awareness and the potentially less risky environment of the internet may be a factor in our observation of decreased risk among MSM.

In conclusion, online partnering did not appear to increase STI risk in our clinic and may even have had a protective effect for some. Future research should explore in greater detail the different ways people use the internet to find partners, if and how this process may increase or reduce STI risks, and, finally, how this context can be used for prevention interventions.