Three randomized control trials (RCTs) in Orange Farm, South Africa, Kisumu, Kenya, and Rakai, Uganda, demonstrated the efficacy of adult male circumcision on reducing female-to-male HIV transmission [1–3]. A meta-analysis found that the pooled relative risk for circumcised compared with uncircumcised men was 0.44 [confidence interval (CI) 0.33–0.60] , suggesting that the efficacy of male circumcision for preventing HIV infection is likely as high as what could be achieved by an eventual vaccine .
Given the strength of these findings, modeling studies have been implemented to assess population-level impact and cost-effectiveness of large-scale male circumcision programs on HIV [5–8]. Such studies estimate that with universal coverage in sub-Saharan Africa, approximately 2 million new infections would be averted in the first 10 years and an additional 3.7 million infections in the following 10 years . The largest impact of scaled-up male circumcision programs is projected for southern Africa, where HIV prevalence is high and male circumcision rates are low . Londish and Murray estimate that the prevalence of HIV in the average African population could be reduced by approximately six percentage points within 10 years for countries scaling to complete male circumcision coverage of the male population.
Although these modeling studies have shown that implementation of large-scale male circumcision programs can significantly reduce the population-level transmission of HIV and are cost-effective, the realization of the predicted beneficial effects is uncertain and dependent upon a number of factors, including rapid program scale up, a high percentage of the male population circumcised, no behavioral risk compensation after male circumcision, and little sexual behavior during the wound-healing period. Many of the parameters in models of male circumcision impact, however, currently lack quantification because male circumcision programs are just starting to scale up in many sub-Saharan African countries.
The prevalence of male circumcision in the adult population in Zambia in 2007 was 13% and the Government of Zambia has embarked on a national program to scale up male circumcision services. To maximize the public health benefit of the male circumcision program, the Ministry of Health (MOH) has targeted HIV-negative men aged 13–39 years for circumcision, while not refusing male circumcision to HIV-positive uncircumcised men . There are approximately 2.6 million HIV-negative, uncircumcised men aged 13–39 years in Zambia according to estimates from the 2010 census (E. McCarthy, Clinton Health Access Initiative, personal communication, 15 August 2011). The Zambian male circumcision program is one of the largest and most developed in the eastern and southern Africa region . Over 100 000 male circumcisions have been conducted in Zambia since January 2009, with approximately 61 000 male circumcisions conducted in 2010. This article uses observational data from Zambia's male circumcision program to measure the prevalence of sexual behavior in the 6-week healing period and to identify risk factors for early resumption of sex. Estimates of the population-level impact of sex during the wound-healing period on HIV transmission and acquisition for men and women are also obtained.
The Population Council is implementing a prospective study of behavioral risk compensation post male circumcision in Zambia. The household-based survey is representative of men and women 15–29 years of age in the catchment areas where the national male circumcision program is being scaled-up. Baseline data were collected between November 2010 and March 2011; approximately 3000 men and 1300 women were interviewed. A separate substudy designed to examine sexual behavior among male circumcision clients in the period following circumcision was conducted from August to October 2010. Prior to their circumcision, 248 men aged 15–29 years were recruited to participate. The baseline sample size was estimated to statistically detect a 10% prevalence of unprotected sexual behavior within 6 weeks after surgery, with 0.80 power and an alpha of 0.05. Men were sampled proportional to client flow aggregated to the provincial level from male circumcision service sites operational at the time of the study. At the time, male circumcision services were largely offered in urban and peri-urban areas, with limited outreach services to health clinics in rural areas. Men were interviewed by a trained enumerator and asked a series of questions about their socio-demographic characteristics, attitudes toward male circumcision, sexual behavior, risk perceptions, and recent history of sexually transmitted infections. Nonsensitive questions were administered via computer-assisted personal interviews; sensitive questions regarding sexual activity were administered via audio computer-assisted self-interviews.
Participants were interviewed 6 weeks later at the male circumcision site or at their homes. A 6-week healing period is recommended to allow for complete wound healing  and clients are counseled prior to their circumcision and at their review visits to abstain from sex during the period. To reduce the likelihood that these instructions might engender socially desirable responses, participants in this study were not asked directly: ‘Have you had sex since your circumcision?’ Rather, the questionnaire was structured so as to establish first whether the respondent was currently sexually active, and only subsequently were questions asked about the timing of the resumption of sexual activity after male circumcision. To determine current sexual activity, the participant was asked whether he had had sex in the previous 4 weeks.
Study protocols were reviewed and approved by the University of Zambia Biomedical Research Ethics Committee, the Zambian MOH and the Institutional Review Board of the Population Council. Written informed consent was obtained from all adult participants. For participants under the age of 18 years, written assent was obtained after obtaining informed consent from the parent/guardian. Consent forms did not ask for permission to link study data with client data collected by service providers, hence the HIV status of study participants was not known.
Descriptive statistics are provided for the sample of male circumcision clients, as well as for a sample of uncircumcised urban men from the prospective study. Multivariate analyses were restricted to men from the male circumcision client sample who participated in the baseline and follow-up interviews. To assess whether there were selection biases in those interviewed at 6 weeks, differences in the baseline and follow-up sample characteristics were assessed using t-tests for continuous variables and χ 2-tests for proportions. Logistic regression was used to evaluate risk factors for sexual activity during the recommended period of abstinence, with estimates reported as odds ratios. Poisson regression was used for the number of sexual acts in the previous 4 weeks, with estimates reported as incidence rate ratios. Huber–White standard errors were estimated to adjust for clustering at the facility level from which participants were sampled at baseline. The regression models also controlled for male circumcision site type (static or outreach), although for the sake of brevity, the estimated effects of site type are not presented here. All analyses were implemented in STATA 11.2 (StataCorp LP, College Station, Texas, USA).
A static-network deterministic transmission model was constructed to represent a cohort of men being circumcised over a 1-year period to estimate how sex during the wound-healing period would influence the rate of new HIV infections among the men and their sexual partners. The model classifies men according to their HIV status, the number of sexual partners they currently have (one or two), and the HIV status of those partners. If the man being circumcised is uninfected, and has infected sexual partners, the influence of the circumcision operation on his chance of becoming infected is calculated. If the man being circumcised is already HIV infected and has uninfected sexual partners, the influence of the circumcision on the chance that he infects a partner is calculated. Men can be involved in zero, one, or two partnerships at the same time over the year period and the risk of HIV transmission in each partnership in 1 year is computed as a function of the number of unprotected sex acts and the probability of transmission per sex act. The probability of transmission/acquisition of HIV per sex act includes an elevated risk during wound healing. It is further assumed that a fixed proportion of men do not have sex during the wound-healing period, but resume normal levels of sexual activity after wound healing. End-points captured in the model are limited to the first new HIV infections among the men and up to two of the sexual partners in the first year following circumcision (further chains of infection generated by a new infection are not captured). The model is available for download from the author's website (https://www.sugarsync.com/pf/D666097_702796_966523). The values for key parameters are specified in Table 1.
Of the 248 men who participated at baseline, 225 or 91% were interviewed 6 weeks later. Table 2 provides an overview of the demographic characteristics and risk behaviors of the male circumcision client sample at baseline and at follow-up. On average, male circumcision clients at baseline were 20.9 years old, had completed 10.5 years of schooling, were predominantly single, had a regular sex partner, lived in urban areas, and were not in the formal employment sector. No significant differences in terms of observable characteristics were found between the characteristics of men at baseline and 6 weeks in bivariate comparisons. Multivariate analysis indicated that men who were older and not formally employed were significantly more likely not to be interviewed (P < 0.05, data not shown).
To assess the characteristics of early circumcision adopters, characteristics of the male circumcision client sample and uncircumcised urban men from the household cohort were compared. Men in the male circumcision client sample were significantly more likely to be younger, more educated, single, to have a regular sex partner, and to own a mobile phone. They were less likely to be employed in the formal sector. Significant differences in sexual behavior are also evident between the male circumcision client and household sample of men. Men in the household sample had a higher mean number of lifetime partners and higher prevalence of casual partners in the previous 4 weeks; the latter finding is likely due to the fact that men in the household sample were significantly less likely to have a regular sexual partner. Men in the male circumcision client sample, however, had a significantly greater prevalence of sex with more than two partners in the previous 4 weeks.
Table 3 shows the risk behaviors during the recommended period of abstinence. Approximately 24% of the men who were circumcised reported resuming sex within 6 weeks; 46% of the men who had sex did so in the first 3 weeks after surgery (22% in the first week – data not shown). In men who reported having had sex in the 6-week period, the prevalence of risky sexual behavior was high, with 82% reporting at least one unprotected sex act and 37% having had sex with two or more partners. The mean number of sex acts in the previous 4 weeks prior to the follow-up interview among all men (14% of whom had never had sex) is less than one. Among men having sex in the healing period, the average number of sex acts in the previous 4 weeks was nearly 3, implying less than one sex act per week on average.
Table 4 shows the results of the multivariate analyses assessing risk behavior in the healing period adjusting for baseline demographic characteristics and risk behaviors. As can be seen in the top panel, few demographic variables account for the variance in sexual activity after the male circumcision healing period. Age increases the odds of having unprotected sex within 6 weeks with a 20% increment in the odds for each advancing year. The age variable is significant for four of the five outcome measures at the P less than 0.10 level and significant at the P less than 0.05 level for two indicators. Education reduces the odds of unprotected sex in the period by approximately 10–20% per year of schooling, but the effect is statistically significant only for predicting unprotected sex in the first 3 weeks. Participant exposure to sexual partners through marriage or living with a partner, increases the odds of having sex, unprotected sex, and frequency of sex, but is statistically significant only for frequency of sexual activity.
The multivariate results in the lower panel of Table 4 indicate that sexual risk factors measured at baseline are good predictors of sexual activity in the healing period. Both lifetime partners and reporting unprotected sex in the previous 4 weeks at baseline are significant risk factors for the resumption of sex, although only the latter of the two is significant at the P less than 0.05 level. The latter result suggests that men who report unprotected sex at baseline have more than twice the odds of having sex in the healing period. The number of lifetime partners reported at baseline indicates a 30% increase in the odds of having two or more partners and unprotected sex, as well as a 20% increase in the rate ratio of having an additional sex act for each additional lifetime partner (P < 0.01). Reporting a casual sex partner indicates more than a five-fold increase in the odds of having unprotected sex and having two or more partners in the past 4 weeks. Overall, these results demonstrate that baseline risk behaviors are useful for predicting risky behavior during the wound-healing period.
If 24% of the 61 000 men circumcised in 1 year were to resume sex during the wound-healing period, the model estimates there could be 69 more new HIV infections that year than would occur when compared to a benchmark of sexual abstinence during wound healing. According to Fig. 1a, 32 of those extra infections would be among men and 37 would be among the female partners of the men who were circumcised. Yet, even allowing for these extra HIV infections caused by early resumption of sex, the net impact of the circumcision intervention in the year is a substantial reduction in the number of HIV infections overall. The model estimates that there could be approximately 230 fewer HIV infections that year than would occur without the circumcision intervention, taking into consideration the impact of early resumption of sex (Fig. 1b). The vast majority (∼95%) of these averted infections are among men.
For women, the net impact of the male circumcision intervention in 1 year is sensitive to the prevalence of sex during wound healing. Abstinence during wound healing provides intrinsic protection to women, whereas early resumption of sex carries added risk. For instance, if more than 30% of men have sex early, within 1 year's time the intervention would generate more new infections among women than it averted (Fig. 1b). The observed proportion of men having sex during wound healing (24%) is just below this threshold. If the prevalence of sex during the healing period was to increase, to, say 50%, there would be 30 extra infections among women in the first year of the intervention. For men and the population overall, even at this high prevalence, the intervention still brings a net reduction in new infections, but at the expense of an increase in infections for women.
The number of excess HIV infections could be reduced if fewer men had sex during the wound-healing period, which would particularly benefit women (Fig. 1a). For instance, if prevalence of early sex were halved to 12%, there would be 26 infections averted among women in the first year instead of 10 (Fig. 1b). The impact of the intervention could also be enhanced by reducing the number of HIV-infected men who undergo circumcision. For example, if 1% of men being circumcised were HIV infected (instead of 6%) then the deleterious effects of sex during wound healing would be greatly mitigated – extra infections caused by the intervention would be reduced among women by 85%.
At the time of this study, male circumcision services in Zambia were situated in a limited number of locations within urban and peri-urban areas in four of the nine provinces of Zambia. Bivariate comparisons of sexual behavior between the male circumcision client sample and urban men from the household cohort indicate less-risk behaviors for some indicators at baseline among men seeking circumcision; however, the results are not consistent. Caution is warranted in interpreting these results as the sample of urban uncircumcised men from the household cohort may not be representative of the catchment areas of male circumcision sites operating at the time. By utilizing the longitudinal structure of the cohort study to compare early, late, and nonadopters of male circumcision, more information will be available to draw conclusions regarding circumcision uptake. Multiple rounds of data are required for such a comparison and are not yet available.
This study focused on the early resumption of sexual behavior after surgery among men aged 15–29 years undergoing adult male circumcision in Zambia. The prevalence of sexual activity and, in particular, risky sex during the wound-healing period in the Zambian context is not trivial and is consistent with levels found elsewhere [1,19]. Nearly one-quarter of men report having had sex during wound healing, with the prevalence reaching 30% among men who were already sexually active at baseline. Additionally, among those men resuming sexual activity, very early resumption of sexual activity (within the first 3 weeks), unprotected sex and multiple sexual partners within the healing period are common. Multivariate analyses identified risk factors associated with sexual activity in the healing period and found that baseline reports of risky sexual behaviors are predictive of early resumption of sex.
A modeling approach using 2010 program data assessed the impact of post male circumcision sex before complete wound healing on HIV infections averted over 1 year in Zambia. Although early resumption of sex is of concern, the model indicates that the negative impact of such behavior will not offset the net benefits of the circumcision program, even in the first year. With more benefits expected to accrue in later years, the impact of the male circumcision program on reducing HIV acquisition for men is expected to be high. Whether the impact of the program on women in the year after male circumcision is positive or negative, however, depends upon the prevalence of sex during the healing period. Even relatively small increases in the levels of early sex can have a deleterious impact on women to a point where new infections exceed averted infections in that year. A substantial number of infections in women could be prevented if more men could be dissuaded from such behavior or if only HIV-negative men were circumcised.
Large-scale male circumcision programs emphasize in information materials and counseling that men undergoing male circumcision should refrain from sex and masturbation for the duration of the healing period [20,21]. Although data on the percentage of clients resuming sexual activity early are available from the three male circumcision RCTs (4% in Kisumu, 5% in Rakai, and 23% in Orange Farm) and a related safety trial in Rakai (11.6%) [14,22], there are few behavioral data on post male circumcision sexual activity currently available for large-scale male circumcision programs. A recent study in Kenya is an exception and found that 31% of circumcised men aged 18 years and older resumed sexual activity prior to 6 weeks after surgery . Modeling studies that have assessed the impact of early resumption of sexual activity have relied either upon assumptions about behavior or data from the three RCTs [6–8]. This study overcomes those limitations by integrating data from a large-scale male circumcision program in Zambia with modeling of its impact on HIV incidence in assessing how interventions to reduce early resumption of sex might affect both men and women.
A few caveats are necessary. First, the model assesses the impact of male circumcision on HIV in 1 year and does not capture the longer term benefits to men and women. Also, if men resume sex, they were assumed to have sex throughout the healing period and with a consistent coital frequency. These assumptions could lead to an underestimation of the beneficial impact of male circumcision and an overestimation of the deleterious impact on women. Second, if sexual behavior is underreported by men, the model would lead to an underestimation of the negative effect of early resumption of sex. Third, to minimize social desirability bias in the reporting of sex, the follow-up questionnaire established whether sex had occurred by framing the question in terms of sex in the previous 4 weeks. Only those men who answered ‘yes’ were subsequently asked how soon after their circumcision they resumed sex. For any men who had sex in the first 2 weeks after male circumcision, but not in the 4 weeks prior to their follow-up interview, this approach would underestimate the prevalence of sex. The potential for this bias is thought to be minimal, as those who are sexually active very early on are likely to remain so.
The study findings suggest that the prevalence of risky sexual behavior during the wound-healing period is relatively high. For this reason, programs need to continue to emphasize to clients the risks associated with early resumption of sex. Identifying men who already engage in risky sexual behavior when they present for circumcision and targeting their counseling accordingly might be effective. Programs should also investigate the impact and costs of interventions to reduce sexual behavior in the wound-healing period, either increasing messages to men directly or, perhaps more importantly, reaching out to women with the message that engaging in sex prior to complete wound healing may heighten their risk of acquiring HIV.
We gratefully acknowledge the Zambian MOH and the University of Zambia Research Ethics Committee for their review and approval of the study. The research team would like to acknowledge collaborating institutions and their staff for their active support, in particular, Dr Mannasseh Phiri, Steve Gesuale, Hayden Hawry, and Nicholas Shiliya (Population Services International – Society for Family Health, Zambia); Marie Stopes International (MSI-Zambia); and Jhpiego.
We gratefully acknowledge the contributions of the study coordinator Kelvin Munjile, the data collection team members and the staff at participating health facilities. The research team is especially grateful to the participants, parents, and guardians who generously gave their time to participate in the study. Significant support was provided by Population Council staff, including Barbara Miller, Samir Souidi, Stan Mierzwa, and Kavitha Valasa.
Principal contributions: P.C.H., B.S.M., K.D., and P.E.T. designed and executed the study. T.H. and G.G. developed and implemented the mathematical model. P.C.H. conducted the empirical analysis of study data. P.C.H. and T.B.H. wrote the first draft of the manuscript, with revisions and comments from all co-authors resulting in the final draft of the article.
Conflicts of interest
This study was conducted under a subgrant from Population Services International (PSI) through the Male Circumcision Partnership, which is sponsored by PSI with support from the Bill & Melinda Gates Foundation. T.B.H. receives funding support from the Wellcome Trust and the Qatar National Research Foundation (NPRP 08-068-3-024).
The authors report no conflicts of interest regarding the content of this article.
1. Auvert B, Taljaard D, Lagarde E, Sobngwi-Tambekou J, Sitta R, Puren A. Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: the ANRS 1265 Trial
. PLoS Med
2. Bailey RC, Moses S, Parker CB, Agot K, Maclean I, Krieger JN, et al. Male circumcision for HIV prevention in young men in Kisumu, Kenya: a randomised controlled trial
3. Gray RH, Kigozi G, Serwadda D, Makumbi F, Watya S, Nalugoda F, et al. Male circumcision for HIV prevention in men in Rakai, Uganda: a randomised trial
4. Mills E, Cooper C, Anema A, Guyatt G. Male circumcision for the prevention of heterosexually acquired HIV infection: a meta-analysis of randomized trials involving 11 050 men
. HIV Med
5. Londish GJ, Murray JM. Significant reduction in HIV prevalence according to male circumcision intervention in sub-Saharan Africa
. Int J Epidemiol
6. White RG, Glynn JR, Orroth KK, Freeman EE, Bakker R, Weiss HA, et al. Male circumcision for HIV prevention in sub-Saharan Africa: who, what and when?
7. Hallett TB, Singh K, Smith JA, White RG, Abu-Raddad LJ, Garnett GP. Understanding the impact of male circumcision interventions on the spread of HIV in southern Africa
. PLoS One
8. Hallett TB, Alsallaq RA, Baeten JM, Weiss H, Celum C, Gray R, et al. Will circumcision provide even more protection from HIV to women and men? New estimates of the population impact of circumcision interventions
. Sex Transm Infect
9. Williams BG, Lloyd-Smith JO, Gouws E, Hankins C, Getz WM, Hargrove J, et al. The potential impact of male circumcision on HIV in Sub-Saharan Africa
. PLoS Med
10. Ministry of Health (MOH). National male circumcision strategy and implementation plan 2010–2020
. Lusaka: World Health Organization; 2010.
11. World Health Organization. Progress in male circumcision scale-up: Country implementation and research update
. Geneva: WHO and UNAIDS.2010.
12. World Health Organization, Joint United Nations Programme on HIV/AIDS. Operational guidance for scaling up male circumcision services for HIV prevention
. Geneva: WHO and UNAIDS; 2008.
13. Wawer MJ, Gray RH, Sewankambo NK, Serwadda D, Li X, Laeyendecker O, et al. Rates of HIV-1 transmission per coital act, by stage of HIV-1 infection, in Rakai, Uganda
. J Infect Dis
14. Mehta SD, Gray RH, Auvert B, Moses S, Kigozi G, Taljaard D, et al. Does sex in the early period after circumcision increase HIV-seroconversion risk? Pooled analysis of adult male circumcision clinical trials
15. Weiss HA, Halperin D, Bailey RC, Hayes RJ, Schmid G, Hankins CA. Male circumcision for HIV prevention: from evidence to action?
16. Wawer MJ, Makumbi F, Kigozi G, Serwadda D, Watya S, Nalugoda F, et al. Circumcision in HIV-infected men and its effect on HIV transmission to female partners in Rakai, Uganda: a randomised controlled trial
17. Weiss HA, Hankins CA, Dickson K. Male circumcision and risk of HIV infection in women: a systematic review and meta-analysis
. Lancet Infect Dis
18. Central Statistical Office (CSO), Ministry of Health (MOH), Tropical Diseases Research Centre (TDRC), University of Zambia, Macro International Inc. Zambian demographic and health survey 2007
. Calverton, MD: CSO and Macro International Inc.; 2009.
19. Herman-Roloff A, Bailey RC, Agot K, Ndinya-Achola JO. Medical male circumcision for HIV prevention in Kenya: a study of service provision and adverse events
. In: Poster presented at the XVIII International AIDS Society Conference
; 18–23 July; Vienna, 2010.
20. Society for Family Health. Male circumcision (MC) take-home booklet
. Lusaka, Zambia: Society for Family Health; 2009.
21. Society for Family Health. Male circumcision (MC) information booklet
. Lusaka, Zambia: Society for Family Health; 2009.
22. Kigozi G, Gray RH, Wawer MJ, Serwadda D, Makumbi F, Watya S, et al. The safety of adult male circumcision in HIV-infected and uninfected men in Rakai, Uganda
. PLoS Med
:e116. doi: 10.1371/journal.pmed.0050116.