Study design and eligibility and recruitment
We used baseline cross-sectional data on men collected from a previous study, therefore, this study is a secondary data analysis. The full details of the parent study are described elsewhere (14), and brief details will be described in this study. The parent study investigated HIV incidence and predictors of inconsistent condom use among adult men enrolled into an HIV vaccine preparedness study in Rustenburg South Africa and was conducted between May 2012 and Jun 2015. In the parent study, men aged 18-49 years of age were recruited from urban and rural area of Rustenburg by the research staff. The research staff used awareness campaigns and community meetings to recruit men into the HIV vaccine preparedness study. Further, the research staff approached men in various settings such as the central business district, townships, clinics, shopping centres, taxi ranks, taverns and car wash stations. Snowball sampling was then used to recruit to recruit friends and acquaintances of the enrolled men into the study. Men who consented to participate were tested for HIV and completed a screening questionnaire for high risk sexual behaviour. Participants reporting previous STI diagnosis, multiple sexual partners, having sexual intercourse with a men (MSM) having a new sexual partner or having sexual intercourse with a partner who was known to have HIV positive were classified as “high-risk”. Only participants who were HIV negative were enrolled into the parent study. The same eligibility criteria were used for this sub-study except that men who were MSM were not included in this study.
Setting
Rustenburg city is a mining town in the Bojanala Platinum District in the North West province, 173 kilometers from Johannesburg. The Tshwane ethnicity and diverse internal and external migrants are the predominant populations in this community (14). The migration population is due to the mining industry which attracts many people seeking employment. Further, most of the migrant mine workers are male and are housed in single sex male hostels, and this is associated with high numbers of female sex workers (15). This may increase the prevalence of STIs in this population (27).
Sample Size and power
The baseline data on 339 heterosexual men collected over 6 months were available for analysis. Post –hoc power calculation based on the prevalence of STIs being 30%, prevalence of all types of circumcision being 34%, and among those circumcised, 15% having STIs while among the uncircumcised group a 30% prevalence of STIs, yielded 84.8% power to detect a difference.
Data collection
Data were collected using a validated interviewer-administered questionnaire. Data collected included socio-demographic data, risky sexual behaviour and circumcision status and assessment of STIs. Inconsistent condom use as was categorized as never, frequently, or sometimes using condoms with partners in the last three months, while consistent condom use was always using a condom with all partners.
Assessment of circumcision
Using the interviewer-administered questionnaire, participants were asked whether they were circumcised and if so, whether they were circumcised traditionally or medically. There was no distinction between cultural and traditional circumcision.
Assessment of STIs
Assessment of sexually transmitted infections was done in the parent study using syndromic assessment. This included a brief physical examination and assessment of vital signs, weight, height, examination external genital, and rectal examinations for relevant symptoms (14). Clinicians employed by the study conducted the syndromic assessment.
Data analysis
All analyses were performed using Stata version 15 (Stata Corp (2017) Stata Statistical Software: Release 15. College Station, TX: Stata Corp LLC). Numerical data were presented using mean and standard deviation (SD) when variables were normally distributed, otherwise median and inter-quarter ranges (IQR) were reported. Frequencies and percentages were used to summarize categorical variables.
Demographic data were compared between the circumcision categories using Kruskal Wallis tests, chi-squared or Fisher’s exact tests. The prevalence of STIs overall was presented with 95% confidence intervals (95%CI), and compared between men who were not circumcised, those who were traditionally circumcised and those who were medically circumcised. The association between risk factors including circumcision and STI prevalence was assessed initially using univariate log binomial regression analysis and crude relative risks and 95% confidence intervals reported. Circumcision as well as the other risk factors which were found to show an association with STIs at <= 0.2 level of significance were included in the predictive multivariable log binomial regression model. Backward selection based on likelihood ratios was used to arrive at a final model. The magnitude of the association between identified factors and outcomes was summarized using adjusted relative risks and 95% confidence intervals.
Ethics considerations
The study was approved by the Health Research Ethics Committee at Stellenbosch University, (Reference No: S19/07/134), with a waiver of informed consent, as no new data were collected from participants. Permission to use the data was obtained from the Arum Institute.