The South African 2012 national population-based household survey was conducted using second-generation surveillance survey approach. The survey methodology is described in detail elsewhere [21]. Briefly, a multistage stratified systematic probability sample of 15 visiting points (VPs) or households was drawn from each of the randomly selected 1000 enumeration areas (EAs) sampled from 86,000 EAs based on the 2001 census EAs. The selection of EAs was stratified by province and locality type defined as urban formal, urban informal, rural formal (including commercial farms), and rural informal localities. Out of 15,000 VPs that were sampled 13,083 were found to be valid. Persons of all ages living in selected South African households and hostels were all eligible to participate in the study. A total of 42,950 individuals in the valid households were eligible to be interviewed, and 38,431 agreed to be interviewed and 28,997 out of these agreed to provide blood specimens for HIV testing [21].
A detailed questionnaire soliciting information related to demographic characteristics, knowledge, attitudes, practice, behaviors and circumcision type was administered. This analysis is based only on adult data, and the analysis focused on males aged 15 years and older who participated in the survey. All youth and adults who participated provided either written or verbal consent, including parent/guardian informed consent for youth under 18 years of age and youth verbal assent to have a blood specimen taken
The survey protocol was approved by the Human Sciences Research Council’s Research Ethics Committee (REC: 5/17/11/10) as well as by the Associate Director of Science of the National Centre for HIV and AIDS, Viral Hepatitis, STD and TB Prevention at the USA’s Centers for Disease Control and Prevention (CDC) in Atlanta.
Measures
The data was used to extract information on males who were circumcised (either medically or traditionally), and those who were not circumcised. Therefore the primary outcome variable has three levels, VMMC, TMC and no circumcision, which makes it a multinomial outcome.
Demographic measures controlled for in the analysis included age (15–19, 20–24, 25–49, 50 years and older, race (Black Africans or other races), locality type (urban formal, urban Informal, rural informal, rural formal areas) and province (Western Cape, Eastern Cape, Northern Cape, Free State, KwaZulu-Natal, North West, Gauteng, Mpumalanga, Limpopo). Explanatory variables included questions on self-perceived risk of HIV (no or yes), and sexual behaviors which included sexual debut (<15 or ≥15 years of age), number of sexual partners in the last 12 months (one partner, two partners or more than two partners), condom use at last sex with the most recent sexual partner (no or yes), and alcohol use based on Alcohol Use Disorder Identification Test (AUDIT) scores [22].
Statistical analysis
Descriptive statistics were used to summarize demographic characteristics, HIV risk perception, and risky behavior by circumcision type, and differences between categorical variables were assessed using Chi-square test. Bivariate and multivariate multinomial logistic regression models were used to assess factors associated with each type of circumcision using no circumcision group as a reference category. The relative risk ratio (RRR) computed as the exponentiated coefficient from -mlogit- was used as a measure of association and reported with 95 % confidence intervals (CI) and a p ≤ 0.05. The “svy” command was used to take into account complex design of the survey. All statistical analysis was conducted using Stata software version 11 (Stata Corp., College Station, TX, USA).