Study setting
The study took place in Hillbrow, a densely populated area in a residential suburb in the inner city of Johannesburg. Hillbrow is a suburb for many people in transit, with a significantly high population density of immigrants—a melting pot of internal migrants from different provinces in South Africa as well as immigrants from different countries. Based on the South African Census 2011 [6], Hillbrow has a total population of 74,131, 24,857 households, and a population density of 68,418.11 per km².
Study population
The participants for the study were selected from a population of consenting male and female immigrant youth, 18–34 years of age, residing in selected households in Hillbrow in 2019.
Study design
A cross-sectional study which involved a household survey conducted in Hillbrow in December 2019 using structured interviewer-administered questionnaires. Data on background information such as individual sociodemographic characteristics; migratory history, knowledge of SRH and access to SRH services were collected from respondents.
Sample size
The sample size for frequency in a population with a correction factor was used to calculate the sample size of immigrant youth needed for this study to account for the effect of intra-cluster correlation. The design effect of 1.25 was used to correct for estimated sample variance and account for within group differences among the respondents. The determination of sample size involved multi-stage sampling technique. Using simple random technique, 9 of the enumeration areas (EAs) were selected out of the 30 EAs in Hillbrow. A full list of all the streets in the selected EAs was mapped, and this formed the sampling frame for the study. Using a simple random sampling technique, a list was drawn from the sampling frame and the streets drawn were the location of the study. The dwelling units in the selected streets were not randomly selected due to security challenge in getting access to some of the properties. In dwelling units where access was given, a flat was randomly selected as a starting point and every 10th flat was visited. When a flat contained two or more households, then only one respondent per household was included. The questionnaire was administered to male and female immigrants from 18–34 years of age. A total of 467 immigrants responded to the survey.
Study instrument
The questionnaire was structured into themes to capture the different characteristics such as sociodemographic, sexual behaviour, knowledge, and information about SRH and migration characteristics of the respondents. Most of the questions were adapted from the survey questionnaire by the African Population and Health Research Centre (APHRC) and the 2016 South African Demographic and Health Survey [25–26], after an extensive literature review. The questionnaire was pilot tested and revised before the field work was undertaken.
Survey measures
Outcome variables
The outcome variables measured were KFP and access to government health facility for SRH services. KFP was assessed from two variables. The first variable was having heard of a family planning method. The question asked was “Have you ever heard of any of these family planning methods?” The contraceptive methods listed were daily pill, intrauterine device (IUD), male condom, female condom, injectable/Depo-Provera, implants, emergency contraception, vasectomy, and tubal ligation. A “Yes” or “No” response was provided for each contraceptive method. The second variable was knowledge of place to obtain a contraceptive method. The question asked was “Do you know a place in this area where you can obtain any of the following family planning method?” The contraceptive methods listed were daily pill, IUD, male condom, female condom, injectable/Depo-Provera, implants, emergency contraception, vasectomy, and tubal ligation. A “Yes” or “No” response was also provided for each contraceptive method. Factor analyses were done on the variables “Having heard of a family planning method” and “Knowledge of where to obtain a family planning method”. The new variables generated were “Heard of contraceptive method” and “Know place to obtain a contraceptive method”. These new variables had Yes/No responses. These two variables were re-classified and used to generate the variable “Knowledge of family planning” with a “Inadequate knowledge” or “Adequate knowledge” categories.
The second outcome variable was access to SRH services provided by a government health facility. The question asked was: “Have you visited any government clinic/health facility in this area for SRH services? This had a Yes/No response.
Explanatory variables
The explanatory variables include variables on sociodemographic, migration and SRH characteristics. The sociodemographic variables examined in this study were gender identified as biological male or female. Age in completed years was further categorised in groups from 18–24, 25–29, and 30–34. Highest level of education attained was categorised as incomplete primary, complete primary, incomplete secondary, complete secondary/matric, incomplete tertiary education (technikon/university), completed tertiary and post-graduate education. Incomplete primary and complete primary were combined and recoded as “Primary”; incomplete secondary and complete secondary/matric were combined and recoded as “Secondary”, while incomplete tertiary education (technikon/university), complete tertiary and post-graduate education were combined and recoded as “Tertiary”. Marital status was classified into single, which include those who are not in union (never married/separate/divorced) and married, which include those who are married or cohabiting. With regards wealth index, this was derived using principal component analysis of several household ownerships. The variable generated was categorised into poor, middle and rich quintiles. In order to assess the sources of information on SRH issues, respondents were asked about their sources of information. The options listed were posters, television/radio, health centre/community clinic, private clinic, government hospital, parents, peer educators, teachers, and friends. Each had a Yes/No option.
Migration status was assessed by asking the respondent about his/her current migration status in South Africa. The options listed were naturalised South African citizen, permanent resident, work permit, study permit, refugee, asylum seeker, no paper (undocumented). These were then categorised as documented (South African citizen, permanent resident, work permit, study permit, refugee), asylum seekers and undocumented. The second migration variable was to assess experiences of discrimination in the neighbourhood. The question asked was “Have you ever experienced discrimination since you came to Hillbrow?” The options were Never, Sometimes, Often, Always and No response. This variable was classified into a Yes or No. The last migration variable was social support in South Africa. The question asked was “Do you have anyone here in South Africa who you trust and with whom you can discuss very important and private issues?” This question has a Yes or No response. Respondents were further asked if they received information about family planning. The question asked was “Have you received any information in the past six months about family planning specifically? This had a Yes or No response. The last explanatory variable was about current use of a family planning method. The question asked was “Are you or your partner currently doing anything or using any method to delay or avoid getting pregnant?” This had a Yes or No response.
Data analysis
Data analysis was conducted using STATA MP 14.1. Bivariate logistic regression analysis was done to get the crude (unadjusted) odd ratios with p-value set at p≤0.05, and 95% confidence interval. The bivariate analyses showed the association between the outcome variables and each of the explanatory variables. Multivariate logistic regression (adjusted odds ratio) was carried out to determine the factors predicting the outcome variables in the adjusted models. Separate models were used for male, female, and total respondents in order to show gender differences in the outcome variables. The adjusted model combined all the significant variables at bivariate level. The results are tabulated for presentation.