This online study examined the effects of a 90-min Internet-based safer sex intervention on sexual behavior of MSM. The results partially supported our hypotheses. Consistent with previous studies (e.g., Baker et al. 2003
), participants in both the control and experimental groups reduced levels of risky sex from baseline to follow-up. We hypothesized that, compared to participants in the control group, those in the experimental group would report a greater reduction in unprotected sex. This was not evidenced without regard to partner serostatus. However, focusing specifically on unprotected sex with partners of positive or unknown serostatus, we found that, compared to the control group, the experimental group did in fact report significantly fewer unprotected acts. We also hypothesized that reductions would be observed for the most risky sexual practices, especially
URAI. This was also not observed. Numbers of unprotected acts with risky partners decreased more for the intervention group than for the control group for all sexual practices, except
URAI. One possible explanation is that perhaps both groups already knew that URAI with HIV-positive or serostatus-unknown partners was a particularly risky activity and mere completion of the baseline assessment was enough to reduce it. However, the intervention group showed greater reductions in other types of unprotected activities with riskier partners.
The additional benefit observed among participants who engaged in sexual behaviors with riskier partners was expected based on the intervention's strong emphasis on negotiated safety (e.g., discussing serostatus and a focus on reducing risk with HIV-positive and status-unknown partners). Discussing serostatus with a partner is an important risk-reduction strategy. An important effect of the intervention could be to impart to participants the skills or the motivation to reduce or eliminate sex with partners of positive or unknown status. Through discussion, a sexual partner whose serostatus was unknown at baseline may have been recategorized as known-negative at follow-up. Alternatively, participants may have increased their refusal of unprotected sex with partners whose acknowledged serostatus was unknown or positive. Unfortunately, our measures of sexual activities were not specific enough to discriminate between these possibilities. Nonetheless, the assessment of different types of sexual activities by partner serostatus was strength of the present study. Future studies should examine to what extent reductions in unprotected sexual behavior are due to increases in skills versus motivation and by what mechanism reductions with riskier partners in particular are achieved.
Given that the control intervention, a relaxation skills and stress management website for MSM, was devoid of targeted sexual risk-reduction information, it is noteworthy that, across all partner types, the control group reported a reduction in sexual risk behaviors comparable to that of the intervention condition. Mere completion of the baseline measures may have resulted in this effect. Participants in both groups completed a detailed assessment of sexual behavior, and research has shown that assessment itself can serve as an intervention. For example, Kypri et al. (2007)
found that students who completed a 10-min online assessment of drinking patterns and associated problems reported significantly lower alcohol consumption and alcohol-related problems 12 months later as compared to those who did not complete an assessment. Presumably, completing a detailed assessment draws individuals’ attention to behaviors that they recognize as unsafe, motivating them to change their behavior. The effect of attention on participant behavior (i.e., the “Hawthorne Effect,” Roethlisberger and Dickson 1939
) could also partially explain the change seen in the control group. It is also possible, though we would suspect less likely, that relaxation skills-acquisition by the control group conferred some benefit with regard to sexual risk reduction.
The results of the present study should be considered in light of its limitations. First, consistent with Internet research in general, the present study suffered from low rates of participation by minorities and those of lower socioeconomic status. Recent research has shown that minority groups, those with lower levels of education, and those over 65 continue to be less likely than other groups to use the Internet; however, those gaps are closing (Fox and Pew 2005
). While recruiting for this study, we attempted to draw members of ethnic minority groups by targeting advertising to cities with higher percentages of African Americans and Latinos and by advertising on minority-specific Internet sites. Despite these efforts, the number of minority participants in our study remained low, as is common with Internet-based research. Reaching minorities is especially important in the area of HIV prevention research because minority MSM represent some of those at highest risk and in most need of preventative interventions. In order for the field of Internet-based intervention research to progress further, a more diverse audience must be accessed.
Another limitation of the present study was that it was designed in such a way so as to preclude intent-to-treat analysis (Lachin 2000
). Those who completed the baseline measures were not provided a link to complete the follow-up measures unless they completed the tutorial to which they were assigned. Thus, no data are available that speak to whether or how the outcome variables would have changed over time among participants who completed only part of the intervention or no intervention at all. Moreover, because the intervention was delivered remotely, there is no reliable information about “dosage” and no assurance that the intervention was completed as intended. For example, although we do know that each enrolled participant clicked on all the pages of the tutorial to which he was assigned, we cannot know to what extent he was truly engaged with the material.
The brevity of the intervention may have limited its effectiveness. It is possible that a longer, more in depth intervention would be more effective; however, we felt that individuals would be more likely to participate in a shorter intervention. Future studies should explore the optimal length for online HIV prevention interventions, but this may be difficult to measure as individuals typically start, stop, and return later to online programs. While this is a limitation for research, the ability to leave and return when one is more motivated is a strength of web-based programs that is generally not easily accomplished with in-person interventions.
In addition to enrolling more minority participants, future studies should assess use of safer sex skills and strategies other than condom use. For example, frequency of conversations about serostatus could be assessed if the intervention specifically targets that behavior, as this one did. More information regarding the effectiveness of specific components of the intervention (e.g., motivational enhancement vs. skills training) would also be useful in addition to data regarding the interaction between individual characteristics and particular intervention components. This intervention attempted to educate young MSM about HIV risk factors including drug and alcohol misuse, mental health issues such as posttraumatic stress disorder and depression, and intimate partner violence. It may be that HIV prevention interventions need to be developed specifically for individuals with these issues. Despite its limitations, the present study provides important preliminary evidence that brief interventions delivered via the Internet can indeed affect HIV risk behavior.