Study design, population and procedures
Data are from a cross-sectional integrated bio-behavioral survey (IBBS) assessing knowledge, attitudes, and behavioral and biological risk factors associated with HIV infection among FSW in five cities across Cameroon. The objective of the IBBS was to translate findings into programmatic interventions and to propose strategies for reducing burden of disease. FSW were recruited for this study using respondent-driven sampling, which is a strategy employed when individuals in the target population are hard-to-reach and when no known sampling frame exists [11]. Methods for respondent-driven sampling have been described previously [11]. A total of 6 seeds [Yaoundé: 2, Douala: 1, Bertoua:1, Bamenda:1, Kribi:1] resulted in the recruitment of 2255 participants. Participants received 2000 Central African Franc (XAF) for their participation and 500 XAF, approximately 0.85 USD, for every eligible participant they recruited for up to three recruits. FSW were recruited from Yaoundé, Douala, Bertoua, Bamenda, and Kribi between December 2015 and October 2016.
Eligible participants were women who were 18 years or older, assigned the female sex at birth, and reported selling sex as a primary source of income in the last 12 months. Women were included in the current analysis if they were HIV seropositive and had at least one living child at the time of the study. All study participants completed written informed consent in either English or French prior to enrolling in the study. Ethical approval for the study was obtained from the National Research Ethics Committee in Cameroon, along with the Johns Hopkins School of Public Health Institutional Review Board.
Following provision of written informed consent, women completed a 45 to 60-min interviewer-administered questionnaire, answering questions related to demographics, sexual and reproductive health history, human rights abuses, and utilization of HIV prevention and treatment services. Blood draws and HIV testing were conducted according to national procedures. For HIV testing, all women were given a first-line rapid test: Alere Determine™ HIV-1/2 Ag/Ab Combo Rapid Test Kit. If nonreactive, the test result for that participant was recorded as HIV-negative. If reactive, a second rapid test, OraSure OraQuick® HIV- 1/2 was administered. If reactive for the OraQuick HIV- 1/2 test, the test result for that participant was recorded as HIV- positive. If nonreactive test result for the participant was recorded as HIV-negative. Pretest and posttest counseling was provided per national guidelines. Those testing positive for HIV were referred to treatment and care facilities for further management.
Outcome
The primary outcome examined in these analyses was the proportion of a woman’s children tested for HIV before the age of five. The proportion of children tested was calculated by taking the number of children who were tested for HIV before age five [“How many of your children were tested for HIV before they were 5 years old?”] and dividing it by the total number of living children a woman reported having [“How many of your biological children are currently living?”]. Both questions were answered via self-report.
Covariates of interest
The main independent variable of interest was ANC attendance during a woman’s last pregnancy, defined as “yes, attended” or “no, did not attend.” Women were asked “The last time that you were pregnant, did you go to the clinic for antenatal care during your pregnancy? This is care that you receive from health care providers during your pregnancy to ensure that you and your baby are well, and to promote a healthy pregnancy.” Other covariates of interest included site of recruitment, age, highest level of education completed, income, years selling sex, future pregnancy intentions, parity, network size, and awareness of HIV status, all of which, other than site of recruitment, were collected via self-report. Future pregnancy intentions were dichotomized and women were asked “Do you plan to or hope to have more children in the future?” To get at network size, women were asked “how many female sex workers have you met in the past week?”
Statistical analyses
Among FSW living with HIV and who had at least one living child, characteristics were compared by recruitment site using Fisher’s exact tests. Logistic regression was used to examine the associations between ANC attendance and other predictors and the odds of a child having been tested for HIV before the age of five. Both crude and adjusted associations were examined. Covariate predictors were selected for inclusion in the final multivariable model based on a consideration of both statistically significant associations with the outcome in univariate models (p < 0.10) and a priori hypothesized relationships. As all variables had less than 1% missing data, complete case analysis was utilized.
Sensitivity analysis
The association between ANC attendance and the odds of a child having been tested for HIV before the age of five was also examined among a restricted sample of women who had children 5 years old or younger at the time of the study. By restricting to women with children under five, a more proximal measure of risk was assessed given that testing guidelines are rapidly evolving and children born within the past 5 years may reflect current child testing trends.