Background: Unintended pregnancies are associated with negative consequences to both mother and baby. Female Sex Workers (FSWs) are at high risk of unintended/unplanned pregnancies. However, prevalence of pregnancy planning and its predictors among FSWs has not been comprehensively investigated. This study was designed to determine contraceptive use, the prevalence, and predictors of pregnancy planning among FSWs in Uganda. Methods: In this cross-sectional study, 819 FSWs attending most at risk populations initiative (MARPI) clinics were recruited using systematic sampling and interviewed with a pretested questionnaire that included collection of data on pregnancy intention using the London Measure of Unplanned Pregnancy (LMUP). Data were analysed using STATA version 14.0. Multinomial logistic regression model was used to identify predictors of pregnancy planning, Results: Of the 819 study participants, only 90 (11.0%) had planned pregnancies. Overall, 462 (56.4%) were hazardous alcohol users and 335 (40.9%) abused drugs; 172 (21.0%) had been raped in the last 2 years and 70 (40.7%) of these accessed emergency contraception post-rape. Dual contraception use (condom and other modern method) was 58.0%. Having a non-emotional partner as a man who impregnated the FSW compared to emotional partner was significantly associated with less planned relative to unplanned pregnancy, (aRR = 0.15 95%Cl =0.08, 0.30), so was lack of reported social support compared to support from friends, (aRR = 0.44; 95% CI = 0.22-0.87), keeping all factors constant in the model. Being raped (aRR = 0.51; 95% CI = 0.31-0.84) or abuse of substances (aRR = 0.65; 95% CI = 0.45-0.93) were significantly associated with lower ambivalence relative to unplanned pregnancy but not with planned relative to unplanned pregnancy. Conclusion: Compared to women in the general population, pregnancy planning was low among FSWs amidst modest use of dual contraceptive. There is an urgent need to promote dual contraception among FSWs to prevent unplanned pregnancies especially with non-emotional partners, drug users, and post-rape.
A cross-sectional study design was used. Participants were recruited from four hospitals where clinics for the ‘most at risk populations initiative’ (MARPI), including FSWs, have been established, in the four regions of Uganda including Central, Northern, Western and Eastern. The MARPI clinic in central region was established in 2008 while the other three clinics were established in 2015. The MARPI clinics offer free reproductive health services including Human Immunodeficiency Virus (HIV) testing, prevention, treatment, care, support and management of other sexually transmitted infections; cancer screening, and family planning services [21]. The clinics do not provide maternal health care services. A sample size of 379 women was calculated using the formula for single population cross-sectional studies [22], and assuming a prevalence of unplanned pregnancy of 44%, a 5% level of significance, and an error of 0.05. The estimated prevalence of unplanned pregnancy estimate of 44% was obtained from a study conducted among FSWs in 2012 in Gulu district in Uganda [15]. The calculated sample size was adjusted with a design effect of 2.0 to compensate for inter-cluster variation to obtain a sample size of 758. This was further adjusted for an anticipated non-response rate of 5% on any study variables [23], to obtain a final sample size of 800 participants. However data was collected among 819 FSWs due to concurrent enrolment across sites. The calculated sample size was allocated to the four MARPI clinics in proportion to their registered FSWs clientele. The FSWs recruited from central, Eastern, Western and Northern MARPI clinics were 517, 90, 112, and 100 respectively. FSWs were eligible for inclusion into the study if they were aged between 15 and 49 years and had been pregnant within the 2 years preceding the date of interview for this study. In addition, FSWs who could not consent because of illness or intoxication with alcohol and/or drugs at the time of screening were excluded. MARPI Clinic in the central region receives about 20–30 FSWs per day whereas the other three sites receive 8–15 FSWs per day. We planned to enrol 6 FSWs from MARPI Clinic in the central region and 2 from the other three MARPI clinics per day over a period of 4 months. To enrol the required number of FSWs over the 4 months’ period, systematic sampling was used where every third FSW registering at the reception at MARPI Clinic in the central region and every second FSW in the other MARPI clinics were approached by research assistants. They introduced the study and obtained written informed consent to administer the screening tool. After screening, written informed consent was sought from eligible FSWs. The study was conducted between May and August 2017. We used the LMUP tool [20] which has been psychometrically validated in general populations in both low and high income countries [13, 19, 24–26]. The LMUP tool was reviewed and other variables found in literature to be associated with pregnancy planning were added to develop the questionnaire. The LMUP was translated into the local languages of Luganda, Acholi, Lugisu, and Runyankole and was then validated among FSWs. The revised questionnaire was further pre-tested among FSWs who had not sought services from any MARPI clinics in the district of Mukono, east of Kampala District. During pre-testing, we checked for the understanding of the various questions by FSWs and made the necessary revisions in the wording. The final pre-tested questionnaire was used to collect data by experienced research assistants who had been trained for 3 days before data collection. The LMUP comprises six questions capturing information on a woman’s circumstances during the most recent pregnancy with respect to use of contraceptives, timing of pregnancy, pregnancy intention, wanting to have a baby, discussion with the man who fathered the last pregnancy, and preconception preparation. Each question was scored on 0–2 scale, with a total score of 0–12 [20]. Each point increase represents an increase in pregnancy planning effort. Data were collected on We included age, marital status, and education. We also collected data on economic indicators including household properties such as ownership of a radio, television, bicycle, motorcycle, home ownership, cell phone, regular phone, computer, an income generating activity, an indoor bathroom, water source, electricity, car, generator, and solar power source. These included the number of living biological children, duration of working as a FSW and main place of recruiting clients. The main work place was defined as the venue for recruiting clients including streets, entertainment places, residence as well as use of phone. FSWs were also asked if they had ever tested for HIV and the most recent result. Where available, clinic records were used to classify the HIV status of each FSW. Clinic records were used in preference to the self-reported status by the FSW. In case records were missing, HIV status was categorised as unverified. Data were collected on partner type including emotional and non-emotional client (either regular or occasional clients). An emotional partner was defined as a man with whom a FSW felt an emotional attachment after a sexual encounter even if he did not give money or gifts all the time after sexual intercourse [27]. A non-emotional partner was defined as a paying client towards whom the FSW felt no emotional attachment, or a rapist. Rape was defined as forceful non-consensual penile-vaginal sexual activity during the last 2 years. Data were collected on contraceptive use and consistent condom use with paying clients. Consistent condom use with paying clients was defined as using condoms all the time with men who paid for vaginal sex. The FSWs were asked if they had someone to provide social support in case they wanted to get pregnant. Social support was defined as any support (emotional, informational, affectionate, tangible and positive social interaction) provided by trusted and reliable person [28]. The supporters were categorised as friend, relative, health provider and no supporter. Data were also collected on hazardous alcohol use based on “The Alcohol Use Disorders Identification Test” (AUDIT) Score ≥ 7 [29]; and ever abuse of drugs. Completed questionnaires were stored in lockable cabinets with access to only authorised study staff. Double data entry was done using EpiData software. Data were then exported to STATA version 14.0 for analysis. The socio-demographic characteristics of participants are described using frequencies with corresponding percentages or as medians with corresponding inter quartile ranges (IQR), or as means with corresponding standard deviations, as appropriate. By using principal component analysis, five wealth quintiles were built from household properties as a measure of socio-economic status. Because the LMUP scores exhibited a bi-modal distribution (see Fig. 1), scores were grouped into three categories, with scores from 0 to 3 categorized as “unplanned” pregnancy, scores from 4 to 9 categorized as “ambivalent” pregnancy planning, and scores from 10 to 12 categorized as “planned” pregnancy. This categorization is consistent with published advice on use of the LMUP tool [20, 30]. The prevalence of pregnancy planning was calculated as the percentage in each of the three categories. Statistical analyses included the chi-square test or Fisher’s exact test, to assess the statistical significance of the association between the different categories of pregnancy planning and each of the independent variables. Additionally, for purposes of comparison with previous studies that considered pregnancy planning as a binary outcome [31], FSWs with LMUP scores less than 10 (including both unplanned and ambivalent) were considered as “unplanned pregnancy category”. Distribution of LMUP scores among female sex workers attending MARPI Clinics Potential predictor variables investigated included those identified from the literature that have been found to be associated with pregnancy planning [13, 32]. The multinomial logistic regression model was used to estimate the variations in the probability of planning for a pregnancy across the categories. The model assumes independence among the dependent variable choices [33]. The dependent variable was “pregnancy planning status” with three categories including unplanned (0–3), ambivalent (4–9), and planned (10–12). Then, we estimated the relative risk ratios with corresponding 95% confidence intervals (95% CI) for all independent variables per category of the dependent variable with exception of unplanned pregnancy which was considers as reference category. As a first step, all variables were included in the model, and then manual stepwise backward elimination was used to remove variables not significantly associated with pregnancy planning. Variables were removed one at a time, starting with those with the largest p-value, until only variables significantly associated with the outcome were left in the model. A 5% level of statistical significance (α = 0.05) was used to retain variables significantly associated with pregnancy planning. Further, we investigated if any of the variables removed from the model confounded the relationship between any of the variables significantly associated with the outcome by checking if putting back such variables in the model changed the odds ratio of any of the variables retained in the model by at least 10%. If putting back any variable in the model changed the relative risk ratio by at least 10%, the variable was retained in the model regardless of its strength of association with the outcome variable. Each variable was investigated for confounding one by one.
N/A