Predictors for achieving adequate antenatal care visits during pregnancy: a cross-sectional study in rural Northwest Rwanda

Background: Inadequate antenatal care (ANC) in low-income countries has been identified as a risk factor for poor pregnancy outcome. While many countries, including Rwanda, have near universal ANC coverage, a significant proportion of pregnant women do not achieve the recommended regimen of four ANC visits. The present study aimed to explore the factors associated with achieving the recommendation, with an emphasis on the distance from household to health facilities. Methods: A geo-referenced cross-sectional study was conducted in Rutsiro district, Western province of Rwanda with 360 randomly selected women. Multiple logistic regression analysis including adjusted odd ratio (aOR) were performed to identify factors associated with achieving the recommended four ANC visits. Results: The majority (65.3%) of women had less than four ANC visits during pregnancy. We found a significant and negative association between distance from household to health facility and achieving the recommended four ANC visits. As the distance increased by 1 km, the odds of achieving the four ANC visits decreased by 19% (aOR = 0.81, P = 0.024). The odds of achieving the recommended four ANC visits were nearly two times higher among mothers with secondary education compared with mothers with primary education or less (aOR = 1.90, P = 0.038). In addition, mothers who responded that their household members always seek health care when necessary had 1.7 times higher odds of achieving four ANC visits compared with those who responded as unable to seek health care (aOR = 1.7, P = 0.041). Furthermore, mothers from poor households had 2.1 times lower odds of achieving four ANC visits than mothers from slightly better-off households (aOR = 2.1, P = 0.028). Conclusions: Findings from the present study suggest that, in Rutsiro district, travel distance to health facility, coupled with socio-economic constraints, including low education and poverty can make it difficult for pregnant women to achieve the recommended ANC regimen. Innovative strategies are needed to decrease distance by bringing ANC services closer to pregnant women and to enhance ANC seeking behaviour. Interventions should also focus on supporting women to attain at least secondary education level as well as to improve the household socioeconomic status of pregnant women, with a particular focus on women from poor households. The present study was conducted in Rutsiro district (Supplementary file 1), one of the districts in Northwest Rwanda. Rutsiro has one district hospital, 18 health centers, and 36 health posts that served the population of 324, 654 in 2017 [25]. All maternal and child health services are provided by these health facilities, which are all, except two upgraded health posts (known as second-generation health posts), government owned (District Health Office, personal communication). Based on the Rwanda Demographic and Health Survey (RDHS), Rutsiro district performs poorly on child health, with 29% infant mortality rate in the ten-year period preceding RDHS 2019/20, and 44.4% stunting (short for child’s age) among children under 5 years of age [14]. Access to health services is also challenging in Rutsiro district due to long travel distances as a result of the nature of district’s landscape which hinders investments in transport infrastructure [26]. According to national survey data, the average perceived travel time to the nearest health facility in Rutsiro district is estimated to be more than 90 min, compared to the national average perceived travel time of about 60 min [27]. The data used in this study were collected as part of a cross-sectional survey conducted between September 2018 and January 2019 to investigate factors associated with child stunting in Rutsiro district. Details on sample size estimation and participant selection for the main cross-sectional study are described in detail elsewhere [28]. Briefly, the district was first divided into 3 zones based on main road network connecting the district to its neighbouring districts. Then, 18 villages were randomly selected (6 villages from each zone). In each of these villages, monthly growth monitoring lists were obtained from community health workers and used to compile a sampling frame from which participants were randomly selected. Eleven mothers who refused to participate or were not found in their homes were replaced by selecting the next name on the list. Mothers were eligible if (1) they had a child aged 6–23 months; (2) the child had no overt signs of illness; and (3) the mother was in the two lowest socioeconomic categories (out of four categories based on the Rwandan Government classification [29]). Of the 400 selected participants, 40 (10%) were excluded from the analysis due to missing data. Analysis was thus performed on 360 participants with completed data. The household data was collected using a structured pre-tested questionnaire pre-programmed on the CommCare platform [30] and data were digitally captured using a tablet. The survey was conducted in the participants’ homes through face-to-face interview. The outcome variable of our analysis was achieving the recommended four ANC visits, which was a binary choice variable defined based on the number of ANC visits achieved during pregnancy with the reference child. Four or more ANC visits was coded as 1, less than four visits was coded as 0. Our independent variable of interest was the distance from household to the nearest health center. GPS coordinates for the location of both the households and health centres were recorded using a handheld Geographic Position System (GPS) device (Tremble Juno SB Handheld), allowing measurement of the distance from households to the nearest health centers. The distance (crow fly) was calculated using ArcGIS software (version 10.4), with an assumption that a household accesses the nearest health center. The calculations resulted into 360 routes connecting the 360 households to the closest health facilities. Other independent variables included respondent and household characteristics. Respondent characteristics included age (coded as < 24 years, 25–29 years, ≥ 30 years), education level (primary education or less, some secondary education), marital status (coded as single/divorced/widowed, married/living with partner), and child’s birth order of the reference child (coded as first child, second child, third child and above). Household characteristics included socioeconomic group (coded as poor (lowest socioeconomic category) and slightly better-off (second lowest socioeconomic category), based on the Rwandan government classification [29]), household hunger level (coded as no /little hunger, moderate/severe hunger based on the Household Hunger Scale (HHS) following Ballard et al.’s methodology [31]), household size (continuous variable), number of children under 5 years of age (continuous variable). The HHS is an indicator of household hunger that is calculated by asking questions as to whether or not a specific condition associated with the experience of food insecurity ever occurred during the previous 4 weeks (30 days). We also included in our analysis three variables related to how respondents interacted with health services during pregnancy. The first was whether the respondent was a beneficiary of a supplementary food program during pregnancy (yes = 1, no = 0). The Government of Rwanda provides free fortified blended supplementary foods to pregnant and lactating women from households in the lowest socioeconomic quartiles [32]. The second was whether all household members were covered by a community health insurance (yes = 1, no = 0). The third was response to the question “do you and other household members always seek health care when necessary” (yes = 1, no = 0). Given that even households with community health insurance may not always seek health care due to additional costs associated to travel [33], information on whether respondent household’s members always seek health care when necessary was considered for analysis. Binary logistic regression was performed to assess independent associations between the outcome variable and the exposure variable of interest (i.e., distance – crow fly – from household to the nearest health facility), as well as individual predictor variables. Multiple logistic regression analysis was applied to determine the influence of distance and other independent variables on achieving four ANC visits represented as dichotomous outcome. All variables were considered as potential predictors and, thus, were included in the model. The magnitude of the effect of the independent variables associated with achieving the recommended four ANC visits was assessed through adjusted odds ratios (aOR). The significance of association was assessed through the corresponding 95% confidence intervals (CI). All analyses were performed using Statistical Package of Social Sciences (SPSS) version 25 (IBM).