Menu
Menu
Menu
Menu
Background: In developing countries, facility-based delivery is recommended for maternal and neonatal health, and for prevention of mother-to-child HIV transmission (PMTCT). However, little is known about whether or not learning one’s HIV status affects one’s decision to deliver in a health facility. We examined this association in Zimbabwe. Methods: We analyzed data from a 2012 cross-sectional community-based serosurvey conducted to evaluate Zimbabwe’s accelerated national PMTCT program. Eligible women (≥16 years old and mothers of infants born 9-18 months before the survey) were randomly sampled from the catchment areas of 157 health facilities in five of ten provinces. Participants were interviewed about where they delivered and provided blood samples for HIV testing. Results: Overall 8796 (77 %) mothers reported facility-based delivery; uptake varied by community (30-100 %). The likelihood of facility-based delivery was not associated with maternal HIV status. Women who self-reported being HIV-positive before delivery were as likely to deliver in a health facility as women who were HIV-negative, irrespective of when they learned their status – before (adjusted prevalence ratio (PRa) = 1.04, 95 % confidence interval (CI) = 1.00-1.09) or during pregnancy (PRa = 1.05, 95 % CI = 1.01-1.09). Mothers who had not accessed antenatal care or tested for HIV were most likely to deliver outside a health facility (69 %). Overall, however 77 % of home deliveries occurred among women who had accessed antenatal care and were HIV-tested. Conclusions: Uptake of facility-based delivery was similar among HIV-infected and HIV-uninfected mothers, which was somewhat unexpected given the substantial technical and financial investment aimed at retaining HIV-positive women in care in Zimbabwe.
We analyzed data from a 2012 baseline cross-sectional survey conducted to evaluate Zimbabwe’s accelerated national PMTCT program implemented in 2011. The objective of that evaluation was to assess the population-level impact of the PMTCT program on MTCT and HIV-free child survival at 9–18 months postpartum. [22] The methods have been published in detail elsewhere. [17, 23, 24] In brief, infants (alive or deceased) born 9–18 months prior (henceforth ‘index babies’) and their mothers or caregivers (≥16 years old) were eligible for the community-based baseline survey. The infants’ age range was chosen to meet the objectives of the impact evaluation. [17] For this analysis, we excluded data regarding caregivers (n = 349, 3.9 % of 9018 participants) and only analyzed data on living biological mothers who were present at the time of the survey (n = 8662), as data collected about deceased (n = 55) and unavailable mothers (n = 294) did not include information about their place of delivery. The survey was conducted in April-September 2012 in five of Zimbabwe’s ten provinces (Harare, Mashonaland West, Mashonaland Central, Manicaland, Matabeleland South). These regions include both major ethnic groups (i.e., Shona, Ndebele), some of the largest cities in Zimbabwe, and rural areas with higher and lower HIV prevalence. Study participants were identified using a two-stage stratified cluster design. Firstly, of the 699 health facilities offering PMTCT services in these five provinces, we randomly selected 157 facilities, proportionate to the number of facilities in each district. Secondly, we identified all eligible infants from the catchment areas of these 157 facilities and sampled a known fraction of them proportionate to the size of the target population in each catchment area. Eligible mother-infant pairs were identified based on information pooled from community health workers and immunization registers from selected and neighboring facilities (to identify women residing in sampled facilities who accessed services at facilities nearby). Further, those mothers identified using community health workers and immunization registers were asked to identify other infants in their neighborhood who were born in the previous two years. Trained interviewers visited the houses of potentially eligible mother-infant pairs (identified as explained above), verified their eligibility, administered the questionnaire in the participant’s preferred language (English, Shona or Ndebele) and collected dried blood spot samples for HIV antibody testing for infants and mothers. Specifically, participating mothers answered anonymous interviewer-administered questionnaires, capturing the mothers’ demographic characteristics, healthcare utilization and place of delivery for the index baby. Maternal samples were stored at room temperature and tested for HIV-1 antibody in batches, using AniLabsytems EIA kit (AniLabsystems Ltd, OyToilette 3, FIN-01720, Vantaa, Finland). Positive specimens were confirmed using Enzygnost Anti-HIV 1/2 Plus ELISA (Dade Behring, Marburg, Germany) and discrepant results were resolved by Western Blot. [25] To assess the place of delivery of the index baby, participating mothers were asked “where did you give birth to your baby?”. We categorized women into two groups: i) mothers who delivered in healthcare facilities e.g., clinic, health center, hospital (henceforth ‘facility-based delivery’) and ii) mothers who delivered at home or elsewhere e.g., someone else’s home (henceforth ‘home-based delivery’). For this paper, we measured maternal HIV status in two ways. Firstly, our analyses used the mothers’ self-reported HIV status before delivery, based on the assumption that women’s healthcare behavior could only have been influenced by their known HIV status at the time of delivery (rather than unknown and laboratory-assessed status). Self-reported HIV status distinguished between mothers who did not know their status before delivery, those who reported they were HIV-negative before delivery, those who reported they were HIV-infected before the pregnancy, and those who learned they were HIV-infected while pregnant with the index baby. Secondly, we examined the association between the uptake of facility-based delivery and the mother’s laboratory-assessed HIV status at the time of the survey. We examined utilization of health services during the pregnancy (i.e., ANC, HIV testing) as these are key services preceding labor and delivery and thus represent possible opportunities to inform women of the benefits of facility-based delivery. We examined several covariates for inclusion in the multivariate models as potential confounders: province of residence, urban/rural status, age, highest educational level, religion, marital status, parity, the decision-maker regarding the place of delivery (i.e., mother, father, other), the sex of the person who makes important household decisions (i.e., female, male, both), the number of sellable assets present in the household (i.e., livestock, bicycle, motorcycle, car/truck, scotch cart, wheel barrow, phone, radio, television) and household-level food security as another indicator for household economic status. Household food security was assessed based on questions from the Household Food Insecurity Access Scale on anxiety and uncertainty about household food supply, insufficient food quality and insufficient food intake; we distinguished between three categories: food security, moderate food insecurity and severe food insecurity. These variables have been selected as covariates for inclusion in the multivariate models because previous studies have shown these factors to be associated with uptake of facility-based delivery in sub-Saharan Africa. [26] First, we described the uptake of facility-based delivery in Zimbabwe at the individual and community levels. At the individual level, we estimated the uptake of facility-based delivery in our sample and by province. We also assessed the aggregate-level uptake of facility-based delivery within each catchment area, the proxy for community. Community-level analyses include data from 156 catchment areas; we excluded one catchment area where only 4 mother-infant pairs were recruited in the study. Second, we examined the association between maternal HIV status and the uptake of facility-based delivery through univariate, bivariate and multivariate analyses. We constructed unadjusted and adjusted Poisson regression models with uptake of facility-based delivery (i.e., yes, no) as the outcome and self-reported HIV status before delivery as the exposure. In Poisson models with cross-sectional data, the exponentiated parameter estimates represent prevalence ratios, a conservative and more interpretable measure of association (compared to odds ratio) if the outcome is common, [27–30] as is the case for facility-based delivery (77 %). In building the adjusted model we checked for statistical interactions between each covariate and the uptake of facility-based delivery, as well as multicollinearity between the variables included in the models. Third, we explored healthcare utilization in our sample of recent mothers. Specifically, we distinguished between six categories of women, corresponding to the six possible combinations of the following two variables: receipt of ANC during the pregnancy (i.e., yes, no) and self-reported HIV status before delivery (i.e., not tested, HIV-negative, HIV-positive). For each of these six groups we computed the absolute and relative frequency of home-based deliveries. All analyses were conducted in STATA 12; we used the STATA svy commands, which allowed us to weight the data to account for the two-stage stratified cluster design and the survey non-response, and to adjust for catchment area-level clustering. The Medical Research Council of Zimbabwe and the ethics committees of the University of California, Berkeley and University College London approved the study protocol. Written informed consent was obtained from all participants prior to their participation. All participants received a gift (i.e., laundry soap and petroleum jelly) worth approximately $5USD. Women were able to receive their HIV test results at the local health facility up to 3 months after the survey, using a card with their unique identifier barcoded.
N/A
