The when and how of male circumcision and the risk of HIV: a retrospective cross-sectional analysis of two HIV surveys from Guinea-Bissau

Introduction Male circumcision (MC) reduces the risk of HIV, and this risk reduction may be modified by socio-cultural factors such as the timing and method (medical and traditional) of circumcision. Understanding regional variations in circumcision practices and their relationship to HIV is crucial and can increase insight into the HIV epidemic in Africa. Methods We used data from two retrospective HIV surveys conducted in Guinea-Bissau from 1993 to 1996 (1996 cohort) and from 2004 to 2007 (2006 cohort). Multivariate logistical models were used to investigate the relationships between HIV risk and circumcision status, timing, method of circumcision, and socio-demographic factors. Results MC was protective against HIV infection in both cohorts, with adjusted odds ratios (AORs) of 0.28 (95% CI 0.12-0.66) and 0.30 (95% CI 0.09-0.93), respectively. We observed that post-pubertal (≥13 years) circumcision provided the highest level of HIV risk reduction in both cohorts compared to non-circumcised. However, the difference between pre-pubertal (≤12 years) and post-pubertal (≥13 years) circumcision was not significant in the multivariate analysis. Seventy-six percent (678/888) of circumcised males in the 2006 cohort were circumcised traditionally, and 7.7% of those males were HIV-infected compared to 1.9% of males circumcised medically, with AOR of 2.7 (95% CI 0.91-8.12). Conclusion MC is highly prevalent in Guinea-Bissau, but ethnic variations in method and timing may affect its protection against HIV. Our findings suggest that sexual risk behaviour and traditional circumcision may increases HIV risk. The relationship between circumcision age, sexual behaviour and HIV status remains unclear and warrants further research.


Introduction
The HIV epidemic in Guinea-Bissau has changed drastically over the past 20 years. Guinea-Bissau has had the highest prevalence of HIV-2 in West Africa for many years, while HIV-1 was absent only three decades ago. In urban Guinea-Bissau, the HIV-1 prevalence has risen from 2.3% in 1996 to 4.6% in 2006, while HIV-2 has decreased from 7.4% to 4.4% in the same period [1]. Guinea-Bissau is affected by a generalised epidemic of HIV-1 and HIV-2 [2].
It is currently estimated that between 3.7-5.8% of the country's adult population is infected with HIV [3,4]. This prevalence is disturbingly high compared to neighbouring countries such as Senegal and Guinea Conakry, which both have HIV prevalence below 2% [4]. The difference in circumcision prevalence has repeatedly been suggested as one of the main causes for the evident contrast between high HIV prevalent countries in Southern and East Africa versus lower prevalent countries in West Africa [5,6]. Furthermore, studies have shown that HIV prevalence is generally lower in regions with high prevalence of traditionally practiced MC [5,7]. Findings from previous observational studies and three randomised control trials have reported up to a 60% risk reduction of HIV infection during heterosexual intercourse after voluntary medical MC (VMMC) [8][9][10][11]. Furthermore, studies have found that the protective effect of VMMC is sustained after a period of 6 years [12,13]. The World Health Organization and Joint United Nations Programme on AIDS have recommended adult male circumcision as a principal method of prevention of heterosexually acquired HIV infection in men from endemic HIV settings with low circumcision prevalence [14,15]. While great financial and logistical efforts have been made in certain regions of Sub-Saharan African, little attention has been paid to male circumcision and the HIV epidemic in West Africa [16]. Furthermore, few studies have looked at the role of circumcision in communities with diverse ethnic and cultural backgrounds. In Guinea-Bissau, an estimated 80 percent of the male population is circumcised [14]. MC is predominantly conceived as a traditional rite of passage and is practiced among all of the various ethno-linguistic groups living in the country. Despite its frequency, there is considerable variation in the age of circumcision. While the Balanta ethnic group perform circumcision ceremonies in village groups every 4-6 years and usually at a later stage in life (approximately 40 years), other ethnic groups such as the Fula and Mandinga generally perform MC between 6 and 13 years of age [17]. Studies have shown that early circumcision (during infancy and pre-puberty) may provide partial protection against the Sexually transmitted infections (STIs) that are known to be more prevalent in uncircumcised men by the time the men become sexually active [18]. Meta  Univariate and multivariate logistical regression models were used to examine the associations between circumcision status, circumcision age, and HIV. Circumcision age was dichotomised into men who were circumcised before or at age 12 years (pre-pubertal) and men circumcised at age 13 or older (post-pubertal). In a subanalysis using data from the 2006 cohort, we determined the association between traditional circumcision vs. medical circumcision and the risk of HIV infection. The results are reported as crude and adjusted odds ratio (COR/AOR) with the corresponding 95% confidence intervals (CI).

Ethics and consent:
The two cohort studies from which this secondary data analysis was performed were approved by the Guinea-Bissau Government Ethics Committee and the Danish Central Scientific Ethics Committee. All participants were counselled and gave informed verbal consent prior to HIV testing.

Prevalence of circumcision
A total of 1,014 males from the 1996 cohort and 954 males from the 2006 cohort were included in this study. All of the major ethnic groups in Guinea Bissau are represented in our analysis and reflect the distribution in Bissau. In the 1996 cohort, 89% (n=904) were circumcised, and the prevalence of circumcision had increased to 93% (n=888) ten years later. A significantly higher proportion of the 2006 cohort were circumcised compared to the 1996 cohort among men aged 14-24 years (p-value = 0.006) and men aged 35 and above (p-value = 0.031). The overall prevalence of circumcision in both cohorts was above 80% for all ethnic groups except the Balanta (only 65% were circumcised in the 1996 cohort and 69% in the 2006 cohort). In both cohorts, the Balanta and Papel comprised over 70% of those uncircumcised. Muslims in both cohorts had a circumcision rate of approximately 99%. As expected, the prevalence of circumcision increased with age. A complete overview of the ethnic groups and other socio-demographic factors, circumcision status and HIV prevalence is shown in Table 1.

Age of circumcision
The overall median age of circumcision (MAC) increased from 13 years in the 1996 cohort (IQR 10-17, range 1-53) to 14 years in the 2006 cohort (IQR 10-18, range 0-49). In both cohorts, the Papel and the Balanta had a notably higher MAC compared to the other ethnic groups. Table 2 shows the median age of circumcision by ethnic group. For the Balanta, only 34% in the 1996 cohort and 37% in the 2006 cohort were circumcised before the median age of sexual debut. We found that over 78% of the Manjaco, Mancanha, Fula and Mandinga were circumcised before age of sexual debut in both cohorts.

HIV, ethnicity and religion
The overall HIV prevalence in the 1996 cohort was 6.5% (66/1014), and 67% (n=44) of those men were infected with HIV-2. In the 2006 cohort, we found an overall HIV prevalence of 6.4% (61/954).  Table 3 shows HIV prevalence associated with circumcision status and other covariates for both cohorts.

Association between circumcision and HIV
In both cohorts, the prevalence of HIV was higher for uncircumcised compared to circumcised men, i.e., 9.1% vs. 6.2% in the 1996 cohort and 7.6% vs. 6.3% in the 2006 cohort. When adjusted for age, ethnicity, civil status, education and history of STIs, we found that circumcision was protective against HIV in the 1996 cohort and the 2006 cohort with OR's of 0.28 (95% CI 0.12-0.66; p=0.004) and 0.30 (95% CI 0.09-0.93; p=0.037), respectively.

HIV and circumcision age
In the 1996 cohort, HIV prevalence was 6.3% for men circumcised at age ≤ 12 (pre-pubertal), 6.1% for men circumcised at age ≥13 (post-pubertal) and 9.1% among those uncircumcised. Ten years later, HIV prevalence was 6.8% for men circumcised at age ≤ 12 (pre-pubertal), 6.0% for men circumcised at age ≥13 (postpubertal) and 7.6% among those uncircumcised. The highest risk reduction of HIV infection was associated with post-pubertal circumcision in both the 1996 cohort and the 2006 cohort models with AORs of 0.25 (95% CI, 0.10-0.62) and 0.29 (95% CI, 0.09-0.93), respectively. Table 3 shows the AORs for age of circumcision and risk of HIV. In a second model, we examined behaviour-related variables such as age groups, previous military duty, history of travel and history of STIs in both cohorts. In the 2006 cohort model we also had data to include condom use (ever) and alcohol use (Table 4) in the 35+ group. The HIV prevalence for men circumcised traditionally was 7.7% compared to 1.9% circumcised medically.
Traditional method of circumcision tended to be correlated with increased HIV risk when adjusted for ethnicity and age (AOR 2.7; 95% CI: 0.91-8.12).

Discussion
In this study, we examined the association between MC, age and method of circumcision and the risk of HIV in Bissau, Guinea Bissau.
Our findings are in line with the literature that has shown that MC is protective against HIV [8][9][10][11]. However, our data did not show increased protection based on early circumcision, contrary to previous research [23]. Other studies have found an inconsistent and non-significant relationship between circumcision age and HIV status [24, 25]. A higher HIV prevalence among ethnic groups who practice early circumcision may imply ethnic differences in risk behaviour. Among ethnic groups in Guinea-Bissau, the foreskin is linked to a lack of cleanliness, and according to some tribes, was the main reason some women might feel sexual repulsion [17].This may imply that those who were circumcised early may be more sexually desirable and therefore more sexually active. Among the Balanta, who practice late circumcision, sexual relations between an uncircumcised man and a virgin women was regarded as hazardous and could result in a disease with symptoms similar to HIV/AIDS, which may deter some men from engaging in sexual relations [17].

Competing interests
The authors declare that they have no competing interests.

Authors' contributions
MSO, CW and DNR designed the study. OL, ZJdS and PAA supervised data collection and handling. DNR, CW and MSO were responsible for data analysis. OL, ZJdS and PAA assisted with questions and comments throughout the entire process. All authors approved the final version of the manuscript prior to submission.

Acknowledgments
The authors would like to thank all of the researchers and research assistants at the Bandim Health Project for their contributions to this study. Furthermore, we would like to thank the Department of Infectious Diseases, Odense University Hospital for funding for this project.