Background: Maternal mortality is still a challenge in Uganda, at 336 deaths per 100,000 live births, especially in rural hard to reach communities. Distance to a health facility influences maternal deaths. We explored women’s mobility for maternal health, distances travelled for antenatal care (ANC) and childbirth among hard-to-reach Lake Victoria islands fishing communities (FCs) of Kalangala district, Uganda. Methods: A cross sectional survey among 450 consenting women aged 15–49 years, with a prior childbirth was conducted in 6 islands FCs, during January-May 2018. Data was collected on socio-demographics, ANC, birth attendance, and distances travelled from residence to ANC or childbirth during the most recent childbirth. Regression modeling was used to determine factors associated with over 5 km travel distance and mobility for childbirth. Results: The majority of women were residing in communities with a government (public) health facility [84.2 %, (379/450)]. Most ANC was at facilities within 5 km distance [72 %, (157/218)], while most women had travelled outside their communities for childbirth [58.9 %, (265/450)]. The longest distance travelled was 257.5 km for ANC and 426 km for childbirth attendance. Travel of over 5 km for childbirth was associated with adolescent girls and young women (AGYW) [AOR = 1.9, 95 % CI (1.1–3.6)], up to five years residency duration [AOR = 1.8, 95 % CI (1.0-3.3)], and absence of a public health facility in the community [AOR = 6.1, 95 % CI (1.4–27.1)]. Women who had stayed in the communities for up to 5 years [AOR = 3.0, 95 % CI (1.3–6.7)], those whose partners had completed at least eight years of formal education [AOR = 2.2, 95 % CI (1.0-4.7)], and those with up to one lifetime birth [AOR = 6.0, 95 % CI (2.0-18.1)] were likely to have moved to away from their communities for childbirth. Conclusions: Despite most women who attended ANC doing so within their communities, we observed that majority chose to give birth outside their communities. Longer travel distances were more likely among AGYW, among shorter term community residents and where public health facilities were absent. Trial registration: PACTR201903906459874 (Retrospectively registered). https://pactr.samrc.ac.za/TrialDisplay.aspx?TrialID=5977.
During January to May 2018, we enrolled women into a cross-sectional survey, selected based on age (15 to 49 years at survey time), being pregnant or history of a pregnancy outcome (live birth, still birth or abortion) in the past 6 months. The survey was part of an intervention aimed at improving maternal health through capacity strengthening of community health workers. Women were enrolled from six purposively selected hard to reach Ugandan islands in Kalangala district. The islands were selected from 12 islands where the authors had previous research experience based on being most hard-to-reach [31]. Exhaustive methods are published elsewhere [31, 32]. In this study we focused on those women who reported a previous birth, to understand their mobility for ANC, childbirth, distances travelled and associated factors. We collected global positioning system (GPS) coordinates (latitude and longitude) for women’s baseline household locations using open data kit (ODK) collect [33]. Women were asked the names of health facilities where they accessed ANC and childbirth services for the most recent birth. If a woman attended more than one health facility, the highest-level facility was considered. Health facilities GPS coordinates were documented using Google maps. Mapping of households and health facilities was done using Quantum Graphic Information System (QGIS) software version 3.16.3 with a coordinate reference system (CRS) of world geodetic system (WGS) 84, geodetic parameter dataset (EPSG) code number 4326 [34]. Straight-line distances in kilometers (km) between each woman’s household location and the facility they attended for ANC and or childbirth were calculated by QGIS distance matrix, using the Universal Transverse Mercator (UTM) CRS of standardized WGS 84, EPSG code number 32,636 [34, 35]. Straight-line distance despite being a less accurate (does not account for environmental conditions, time and effort that might impact on the real distance) measure of distance travelled, this method provides a suitable alternative and has been previously used to assess ease of access to health services in remote settings [36–39]. Household to ANC facility distances were complete for 218 women, while 250 women had household to childbirth location distances completed, and these were used in the distance to ANC and birth attendance analysis. This analysis aimed at answering the following questions: The primary dependent variable was mobility for birth attendance, dichotomized into whether or not a woman had the most recent birth within or outside her community of residence. Distance from household to birth facility was also a dependent variable, dichotomized into whether a woman moved up to 5 km or over 5 km from her household to a childbirth facility during the most recent childbirth. Women’s socio-demographic characteristics were summarized using frequency tables and compared with the dependent variables (mobility for birth attendance and distance from household to ANC or childbirth facility within 5 km and over 5 km), using chi-square and Fisher Exact tests for categorical variables and median, range for continuous variables. We defined adequate distance to maternal health facility as having travelled within 5 km from the women’s households [40]. The selection of 5 km is based on previous work in low- and middle-income countries, indicating that being within 5 km of obstetric care facilities was related to heath facility births [41–43]. The Uganda Health Sector Development plan 2015/16 to 2019/20 also aimed at improving access to health through ensuring that at least 85 % of the population are within 5 km access to a health facility [40]. Uganda’s current strategy for improving health service delivery also involves upgrading and construction of health facilities at subcounty level to attain a within 5 km walking distance to a health facility [44]. Adjusted odds ratios (AOR) of mobility for birth attendance and distance to childbirth facility were estimated using multivariable logistic regression modeling, testing for associations with the independent variables. A priori selection of independent variables to include in the multivariable models was based on previous literature and biological plausibility. Independent variables included in the bivariable analysis were residence community with or without a public health facility, age groups, duration of community stay, religious affiliation, marital status, highest education, partner’s highest education, main occupation, participant’s health decisions maker, lifetime births, pregnancy planned, history of pregnancy loss, number of ANC visits, at least four ANC visits attendance, receipt of ANC components, skilled birth attendance, and type of childbirth facility. Additionally, those variables found to have a bivariable statistical significance at an alpha (α) of ≤ 0.2 were included. The final best suited independent variables in the model were those with the lowest P-value, lowest model Akaike’s information criterion and Bayesian information criterion values. All analyses were done using STATA® version 15 [45]. Tables were created using asdoc, a STATA program written by Shah [46]. Strengthening the reporting of observational studies in epidemiology (STROBE) guidelines for cross sectional studies were followed in this article [47].