We utilized data from the Barriers to Online Prevention Research survey of US MSM collected between March 19, 2009 and April 16, 2009. The methods have been previously reported [31]. Briefly, participants were recruited from MySpace, a large social networking site, using banner advertisements. Advertisements were directed at male users of MySpace18 years or older who resided within the United States. Once the banner advertisement was clicked, individuals were screened for eligibility and provided informed consent. Eligible participants included men 18 years or older who had at least one male sex partner in the past year and were residents of the United States. Eligible men were then asked to complete a Health Insurance Portability and Accountability Act-compliant online survey on SurveyGizmo (Boulder, CO). The survey took approximately 30 to 45 minutes to complete, and no compensation was provided to participants. The study protocol was approved by the Emory University Institutional Review Board.

Participants were asked a series of questions regarding demographics, sexual history, most recent sex act, and most recent sexual partner. Specifically, participants were asked to report the number of male sexual partners in the last 12 months and if they were high or buzzed during last act sex act. Participants were also asked questions pertaining to their most recent sex partner, such as if he was a main or casual partner. A main partner was defined as someone that the participant felt committed to above all others, and a casual partner was defined as one whom the participant did not feel committed to above all others. With regard to their last sexual partner, participants were asked whether they engaged in anal and/or oral sex with him, if a condom was used, and if the condom broke during last anal intercourse. A copy of the survey items relevant to the present study can be seen in Multimedia Appendix 1.

Our outcome measure, condom breakage, was assessed via a categorical response to two separate questions designed to capture both receptive and anal insertive sex acts. Men were first asked if they had receptive anal intercourse, insertive anal intercourse, or both during their last sex episode. Based on their responses, men were asked separately about condom use when they were the receptive and/or insertive partner. Questions were: “Did [your last sexual partner] use a condom the last time you had receptive anal sex?” and “Did you use a condom the last time you had insertive anal sex?” Responses for both questions included “He (I) did not use a condom,” “He (I) used a condom part of the time,” “He (I) used a condom the whole time,” “He (I) used a condom but it broke,” “Don’t know,” or “Prefer not to answer.” Condom breakage was defined as answering “He (I) used a condom but it broke” for either insertive or receptive anal sex acts with the last male sex partner of the participant.

Data were analyzed using SAS version 9.4 (SAS Institute). Participants included in the final analysis reported engaging in receptive or insertive anal intercourse with their last male partner in the past 12 months and using a condom during last anal intercourse. Bivariate analyses were conducted to examine unadjusted correlates of condom breakage. For continuous variables, a Wilcoxon rank-sum z test (two-sided) was used due to the nonnormally distributed nature of the variables. Statistically significant covariates (P value ≤.05) were included in the final model, as were variables found to be associated with condom breakage in previous studies [20-22,25,32,33].

The dependent variable, condom breakage, was calculated as a proportion and modeled the event level for both receptive and insertive anal intercourse at last sex. Thus, one participant could contribute two observations (one for receptive and one for insertive sex) from his last sexual episode. Multivariable analysis was conducted using multivariable logistic regression controlling for repeated observations. Odds ratios and 95% confidence intervals were calculated for categorical variables. For continuous variables, odds ratios and confidence intervals per unit of 5 were calculated.

Results from our multivariable analysis of condom breakage among US MSM revealed an overall condom breakage rate of 4%. Condom breakage rates did not differ between participants who reported receptive and insertive anal intercourse. Younger age, a greater number of sexual partners reported in the last 12 months, and being buzzed or high at last sex with a casual partner were associated with condom breakage.

The overall condom breakage rate of 4% is in line with previous studies [10,14-18], but there is considerable variability among breakage rates from prior studies [33-40]. Golombok et al found a condom failure rate of 2% in a group of 283 homosexual couples in the United Kingdom, but the study focused on sexual activity among long-term couples, which does not represent the MSM population [19]. Our sample was younger, and more than half of sex partners were reported to be casual partners. A 6-month condom breakage risk of 31% was found in a cohort study of MSM in Atlanta, GA [35]. The high frequency of condom breakage found in the aforementioned Atlanta cohort study compared to other studies is likely due to the longer recall period (6 months), allowing for a higher number of sex acts to occur. Further, the study revealed 40% of black MSM reported breakage or incomplete use; the population was twice as likely to report condom breakage as white MSM. Similar point estimates were seen from our analysis; however, our sample size was not sufficient to assess the relationship. The wide variability of condom failure rates among these studies is most likely a reflection of the diverse population of MSM and sample sizes under study [36-40].

For every 5 male sexual partners, we found that the odds of condom breakage increased by 3%. There are likely user characteristics of participants with large numbers of sexual partners not captured by our survey that explain the statistical association with condom breakage. Participants with a larger number of male sexual partners might have engaged in more aggressive coital behaviors than those with fewer sexual partners, leading to greater stress on the condom [41,42]. Further, participants with a large number of male sexual partners may have a predisposition to inappropriately use lubricant, resulting in condom failure [43]. MSM who report high numbers of sexual partners represent a risk group commonly targeted for behavioral interventions, as having multiple sexual partners is an established risk factor for HIV acquisition [38]. Results from our study suggest that behavioral prevention interventions targeted to this high-risk group should also include more thorough condom education.

Several studies demonstrated that drug and alcohol use is associated with an increased risk of HIV acquisition among heterosexual men engaging in vaginal intercourse and MSM engaging in anal intercourse [44,45]. Alcohol use during sex is higher among casual partners compared to main partners; however, few studies have assessed the association between condom failure and partner type, modified by alcohol and drug use before or during sex [46,47]. Results from our study suggest that being high or buzzed during sex with a casual partner was associated with increased odds of condom breakage compared to being high or buzzed with a main partner. Alcohol and drug use may prolong ejaculation and thereby prolong sex, increasing the risk of condom failure [27]. Moreover, alcohol and drug use may impede the proper application and usage of condoms, increasing condom failure rates [27]. These results suggest that public health prevention interventions need to increase awareness of the effects of drug and alcohol use during sex, specifically targeting MSM who engage in sex with casual partners.

There are several limitations to this study. Most notably, our respondents are not representative of all MSM in the United States. Due to the nature of the survey, participants had to be proficient with computers, making them more likely to have a higher education and income than the general population of MSM. Our study used a cross-sectional study design to evaluate condom breakage at last intercourse and therefore did not capture condom use or breakage over time. Safe sex behaviors fluctuate over time [31], making a prospective study more appropriate to capture the time-dependent nature of condom use and failure. We did not ascertain a history of lubrication use or lubrication use during last sex act. Therefore, our sample may represent men who were less likely to use lubrication, and thus, factors associated with condom breakage may actually be factors associated with lubrication use. The majority of studies that have evaluated condom failure define it as condom breakage or slippage during sex [36-39]. Our study only used condom breakage as a measurement of condom failure. Omission of condom slippage data resulted in our inability to report condom failure more generally. Our study uses older data; however, our finding of a condom breakage rate of 4% is in line with previous and more recent studies [10,14-18]. Finally, responses may have been affected by social desirability bias and recall bias, resulting in misclassification of outcomes [48].

Results from our study highlight condom breakage rates among a diverse sample of MSM from the United States. While condoms remain a strong component of prevention policy, our findings indicate condoms may not suffice as a sole means of reducing HIV transmission risk within the context of a high per-act transmission risk of anal sex [6]. Combining condoms with other prevention modalities such as preexposure prophylaxis can help mitigate risk that occurs after condom breakage [49]. Even in the context of preexposure prophylaxis, more efforts should be made to improve condom use practices that might lead to condom failure [49]. MSM who report a high number of sexual partners as well as those who use drugs and/or alcohol before or during sex with a casual partner are at an increased risk for condom failure and would benefit from targeted condom education programs to help mitigate their risks. As innovation continues to occur in the design of condoms, it will be important to consider if there are new types or designs of condoms that might decrease condom breakage during anal intercourse [49]. To best understand when, why, and how condom failures occur, further studies are needed to capture the time-dependent nature of condom use; these should include high-risk sexual behaviors; precoital factors, such as condom storage; and coital factors, such as duration of sex act and use of lubricants. As HIV continues to disproportionately affect United States MSM and anal intercourse remains a significant mode of HIV transmission, condom failure among this population needs to be addressed through multiple approaches.