Male circumcision (MC) has recently become a significant component of human immunodeficiency virus (HIV) prevention efforts after the results of three randomized controlled trials (RCT) demonstrated that circumcised men have an approximately 60% less chance of becoming infected with HIV through unprotected vaginal sex. [1]–[3] Following the publication of results from the three RCTs the Joint United Nations Programme on HIV/AIDS (UNAIDS) and the World Health Organization (WHO) recommended that MC be recognized as an intervention for HIV prevention in countries with low MC rates, high HIV prevalence, and where heterosexual sex is the main route of transmission. [4]

Despite WHO and UNAIDS recommendation, and the promotion of MC as an HIV prevention method in sub-Saharan Africa, questions remain about whether promoting MC as an HIV prevention intervention will result in decreases in HIV incidence. A key concern is that promoting MC may lead circumcised men to develop a false sense of complete protection against HIV and engage in risk compensation by halting or decreasing previous protective behaviors such as condom use or partner reduction, causing the protective effects of MC to be reduced or negated. [5], [6]

Risk compensation is the notion that individuals modify their behavior in response to real or perceived changes in risks. Pinkerton has defined risk compensation as “any behavioral change that acts to offset a reduction in risk resulting from other changes”. [7] Risk homeostasis theory has been used to describe this phenomenon. [8] Accordingly, sexual risk compensation in the context of MC is characterized by someone knowing that they are less likely to contract HIV and responding to this reduced risk by increasing behaviors associated with greater risk of HIV transmission, such as not using condoms, increasing the number sexual partners, increasing the frequency of higher-risk sex (e.g. anal sex versus vaginal sex), or increasing the frequency of sex with high-risk partners (e.g. sex workers).

The terms risk compensation and sexual disinhibition have been used interchangeably in the public health literature. The concept of sexual risk compensation implies that a change in potential harm has occurred and is in turn being countered by a change in one's level of risk behavior. Sexual disinhibition on the other hand denotes that some sort of restraint has been removed, and subsequent behavior is not related to a change in the level of risk, but rather a lowering or removal of such a moderator. The effect of alcohol illustrates disinhibition since it may lower the restraints of certain behaviors and can result in people engaging in sexual behavior that they may be less likely to perform when not inebriated, but the risks of contracting HIV related to sex remain constant. MC lowers the probability of men contracting HIV and any increase in risk behavior attributable to this known protection should be considered risk compensation. A preference for the term risk compensation, rather than disinhibition, has been expressed by Hogben and Liddon [9] and Cassell et al. [5]

Data from the three African RCTs show varying evidence of changes in sexual risk behavior after circumcision. The Ugandan RCT reported no risk compensation among participants while the South African trial reported an increase in the number of sexual encounters between 4 and 21 months after circumcision but no decline in condom use or increase in the number of sex partners. [1], [3] The Kenyan trial reported a decrease in risk behavior among both the circumcised and uncircumcised groups, with the uncircumcised men reporting less unprotected sex and greater condom use at 24 months follow-up. [2] Additionally, a subgroup of 1,319 Kenya RCT participants were recruited for a study examining risk compensation over a 12-month period using an 18-item behavioral risk propensity scale and found a significant reduction in sexual risk behavior in both uncircumcised and circumcised men, and no statistically significant difference in the incidence of sexually transmitted infections (STIs) or sexual behavior in the two groups. [10]

A prospective cohort study conducted from 2002 to 2004, before the dissemination of RCT findings, looked at sexual risk behavior following MC in western Kenya and found no differences between circumcised and uncircumcised men in the average number of risky sex partners per month or the frequency of condom use over a 12-month follow-up period. [11]

Given the variability of results from studies examining sexual risk compensation in HIV vaccine trials, [12] HIV prevention interventions, [7] STI prophylaxis, [13] antiretroviral therapy, [14], [15] and condom promotion programs, [16] and the dearth of research on sexual behavior change related to MC, the objectives of this study were to: 1) explore how MC influences sexual risk perceptions and behaviors and 2) identify and describe individual level factors that could facilitate or reduce sexual risk behaviors related to male circumcision. In this paper we characterize the sexual perceptions and behaviors of circumcised men and provide context for sexual behavior change related to MC among men in Kisumu, Kenya.

Between March and November 2008 we conducted individual in-depth qualitative interviews and focus group discussions among women and circumcised and uncircumcised men to explore sexual risk perceptions and behaviors related to male circumcision. For the purpose of this paper only interviews conducted with circumcised men were used. Men who were recently circumcised provided first hand accounts of their pre- and post-circumcision sexual behaviors. Men circumcised in their youth provided information on whether learning that circumcision is protective against HIV influenced their sexual risk behaviors.

Table 1 presents respondent demographic characteristics. Among the 30 circumcised male respondents interviewed (N = 30) the median age was 25. Most respondents were not married and not living with a sexual partner (63%), had completed primary school (90%), worked in the service industry (e.g. transportation or security guard) (50%), reported earning 2,000 to 10,000 Kenyan Shillings (approximately US$30–$150) per month (70%), and were from the Luo ethnic group (83%). Among respondents' sexual behaviors, 60% reported having one sexual partner in the past 30 days; 73% said they had 2 or more sexual partners in the past year, 67% said that they used a condom during their last sexual encounter, 1 (3%) reported being HIV positive, and none reported having sex with men in the past 12 months.

Respondents were circumcised between 1986 and 2008. A majority of respondents were circumcised from 16 to 30 years of age (80%). The youngest age at the time of circumcision was seven years old and the oldest was thirty-three. The period of time that respondents had already been circumcised ranged from 2 weeks to 22 years, the median being one year. Of the 30 circumcised men, 11 were circumcised at the Kisumu RCT's clinic (average number of years having been circumcised = 2.4, range = <1–6 years), 10 were circumcised at the sexual and reproductive health NGO's clinic in Kisumu (average number of years having been circumcised = 0.8, range = <1–2 years), 7 were circumcised in hospitals throughout Kenya (average number of years having been circumcised = 10.4, range = <1–22 years), and 2 were circumcised in ethnic group ceremonies (average number of years having been circumcised = 16.4, range = 14–19 years). Twenty-three (77%) of the respondents were circumcised in Kisumu.

The results of this study illustrate that MC does not necessarily lead to risk compensation, and that there are different behavior change outcomes. Previous studies exploring risk compensation associated with MC have been quantitative in nature, conducted during an RCT, and have provided limited detail about the context of risk compensation and protective sexual behaviors. Our study conducted among RCT participants and men circumcised in other venues provides details of motivations for changing or maintaining sexual behaviors after circumcision. This study helps in understanding individual sexual risk perceptions and behaviors that may change during the scale-up of MC programs. Our results underscore the need for HIV prevention counseling to be integrated with MC services.

A majority of men reported adopting protective sexual behaviors or not changing their sexual behavior after getting circumcised, or after learning of MC's protective effects against HIV. Several factors were associated with adopting protective behaviors: becoming aware of one's sexual risk behavior, HIV and MC counseling and education, HIV testing, and a desire to remain HIV negative. Some respondents reported that MC accompanied by pre-circumcision HIV counseling and testing influenced them to adopt protective sexual behaviors, including reducing high-risk behavior. In other instances counseling seems to have made men aware that some of their past behaviors increased their risk of contracting HIV, which influenced them to adopt risk reduction strategies. All men who mentioned HIV counseling and testing as influencing their behavior had gone through a MC program established for HIV prevention purposes. However, respondents who engaged in risk compensation also received counseling.

Respondents who maintained their sexual behavior may have established limits that they thought were effective in avoiding HIV infection, such as using condoms consistently or being faithful to one partner, and may have resolved to continue with the same sexual practices. We also observed an inverse form of risk compensation from a man who acknowledged high-risk sexual behavior and got circumcised in order to reduce his chance of acquiring HIV. It is likely that other men may not feel the need to change their high-risk behavior because, by being circumcised, they decrease their chances of contracting HIV and therefore will maintain previous sexual behavior. Given the varied sexual behaviors reported by study participants, our results support the guideline that counseling should be tailored to the individual. [22]

All respondents who did not attend pre- or post-circumcision counseling maintained their sexual behaviors. These men were more likely to be circumcised at a younger age, before their sexual debut, in hospitals, or as part of ethnic group customs. They did not report receiving risk reduction counseling and did not discuss the process of getting circumcised as instrumental in changing their sexual behavior, which may be because these men established HIV risk reduction routines earlier in their sexual lives prior to knowing that MC was protective against HIV. Importantly, they reported being aware of MC's partial protection against HIV, suggesting that information on MC was reaching the general public and may be effective in mitigating risk compensation.

The fact that little risk compensation was reported may be due to several factors, including the effects of counseling, HIV testing, knowledge that MC only offers partial protection, and condom availability. All respondents who demonstrated risk compensation were circumcised at clinics that offered MC as an HIV prevention measure roughly one year before their interviews, and received some counseling before getting circumcised. The respondent who reported sex without a condom after getting circumcised also indicated that this sexual behavior was similar to his pre-circumcision behavior. The men who reported increasing the number of sexual partners after circumcision reported using condoms consistently with their new partners, potentially offsetting some of the risk of having concurrent partners. In light of these findings, MC and HIV counseling and testing programs should work to identify individuals who are disposed to engage in risk compensation during counseling sessions by asking men about their plans to change their sexual behavior following circumcision and emphasizing that MC only offers partial protection.

Behavior change facilitated by the physical and mechanical aspects of removing the foreskin may also have implications for HIV and STI prevention. Respondents reported the ability to increase the number of rounds of sex after being circumcised, which could potentially lead to greater exposure to HIV and STIs through an increased frequency of sexual acts. In the Orange Farm, South Africa trial, the investigators reported an increase in the number of sexual encounters, but no change in the number of sex partners. [1] This may reflect an increase in the number of rounds per partner as suggested by the circumcised participants in our study. Despite the greater number of encounters in the circumcised men, there was still a 60% reduction in HIV with circumcision in the South African trial, as well as a reduction in some other STIs. [1]

Other reported benefits of circumcision among respondents were that it made condoms easier to use and eliminated penile cuts. This is consistent with findings from the Kisumu RCT cohort in which circumcised men had a 39% reduction in the odds of reporting bleeding, cuts, scratches, or abrasions on their penis acquired during intercourse. [23] Reducing or eliminating such penile trauma could be a secondary benefit of male circumcision in reducing the risk of HIV transmission. Also of importance is that some men found it easier to use condoms after being circumcised, which may encourage them to use them more frequently.

Responses illustrate that behavior change is a dynamic process and examining sexual risk behaviors is not a linear exercise. Circumcised men may not just adopt positive behaviors or only increase risk behaviors. Men who reduced the number of sexual partners to one may in turn stop using condoms with that partner. We also saw that respondents who increased the number of concurrent sexual partners subsequently increased condom use with their new partners. That MC may play a role in promoting an increase in concurrent sexual partners is important to recognize since concurrent partnerships have been shown to be instrumental to the spread of HIV in sub-Saharan Africa. [24], [25] Also, if these new partners become longer term partners, condom use is likely to wane, making concurrent partnerships a significant HIV risk. [26]–[29] Since MC is not a stand-alone HIV prevention intervention it is vital that messages emphasizing continued condom use and partner reduction accompany MC programs and services.

While counseling and testing was reported by respondents as being instrumental in influencing their behavior, the content and frequency of the counseling provided to circumcised men was not a primary area of inquiry of our study. Therefore we have limited data describing what components of counseling were effective in bringing about behavior change. This is an area that merits further research. Nevertheless, since boys and men in various settings in sub-Saharan Africa are likely to be circumcised for many reasons, in addition to HIV and STI prevention, and counseling can have an impact on sexual risk behavior, as seen in our study, (see also Kamb et al. [30] and Denison et al. [31]) counseling should be made available with MC services. In settings where MC is not provided as an HIV prevention measure it is unlikely that hospital staff and ethnic group circumcisers are trained to provide HIV counseling. Counseling could be realized by providing training to clinicians and traditional circumcisers, placing counselors in clinical and traditional circumcision ceremony settings, or establishing referral systems between MC settings and HIV counseling and testing clinics.

The findings from this study should be viewed within the context of the following limitations. The results are based on self-reported experiences. People are sometimes reluctant to share personal information regarding their sexual behaviors, so it is possible that respondents withheld information or fabricated answers during the interviews. Recall bias also may have affected responses. It is possible that some of the men we interviewed were not circumcised since we relied on respondents' reports of their circumcision status. However, since uncircumcised men were eligible for the study, there was no reason for respondents to lie about their circumcision status. Eleven of the men were participants of the Kisumu RCT examining male circumcision's protective affect against HIV and were tested for HIV and received HIV prevention counseling prior to getting circumcised and after circumcision every six months for two years. Such intense and repeated counseling is unlikely to be representative of the level of counseling other adult men being circumcised receive. This research was conducted in the early stages of scaling up MC, so respondents in our study may have been those most motivated to seek protection offered by MC and more motivated to avoid risk compensation. Additionally, because this research was carried out in western Kenya mostly among the Luo, an ethnic group that traditionally does not circumcise yet are surrounded geographically by circumcising groups, these results may not be applicable in other countries or regions where circumcision is less widely practiced.

Despite these limitations, several important findings emerged from our data. Respondents have illustrated that protective behavior change can result from men being circumcised. Counseling placed within or related to MC programs appears to be influential in promoting protective behavior change among participants. We conducted interviews with respondents who had been circumcised for less than a year up to 22 years, and it is promising to see that men circumcised for longer periods of time exhibited similar sexual behavior and circumcision knowledge to those more recently circumcised, because it shows that information about MC is reaching the general population. In addition to the direct benefit circumcision provides in reducing HIV transmission there may also be secondary benefits for individuals such as reduction in penile cuts during sex and easier condom use, resulting in lower risk of HIV infection.