Our study showed that 20.16% of 47,244 women aged 15–49 years in the Philippines delivered at home. This proportion was higher than what was reported among urban mothers (12.49%) but lower than what was reported in the rural communities (26.16%). This has been the lowest proportion of home deliveries since the 1993 NDHS [26] and mirrors the improvements made in reducing the infant mortality rate, under-5 mortality rate, and maternal mortality rate in the country [1, 9]. Despite this progress, surveys and the present analysis continue to show that women still deliver at home where there is low possibility of being assisted by skilled birth attendants. More importantly, delivering at home could delay life-saving treatment should complications due to childbirth occur [8].
There are a number of factors that influence the choice of women to deliver at home, including those identified by our study: age, marital status, mother’s education, husband’s or partner’s education, wealth index, parity, birth order, women’s decision-making power, ANC visits, and emergency preparedness. Our findings suggest these factors yield similar results among the overall, urban, and rural populations, but with a relatively greater effect observed for the rural population in most of the study factors considered similar to the findings of a previous study [38]. The present study further highlights existing disparities between women in the rural communities with more home deliveries despite comparable proportions of rural to urban women on key variables such as having adequate ANC visits and being prepared financially during pregnancy for emergencies. These likely reflect the inequalities associated with health service delivery, specifically in the distribution of health professionals and low investments in the health sector infrastructure [22]. A local study on the trends in neonatal mortality and child health inequality used facility-based delivery as a proxy of health service delivery, and determined that it is an effective but complex intervention that requires a fully functional system [22]. The government has since made efforts to make healthcare services more accessible through the MNCHN Strategy and the newly passed Universal Health Care Act [9, 11]. Our study also identified wealth through assets and wealth proxied by having emergency funds during pregnancy as important factors influencing the choice of home delivery. The odds of home delivery is most pronounced in the poorest wealth category with a slightly larger effect among urban women. The observed association between home delivery and wealth index is consistent with the results of other studies [28, 31, 34, 35]. Women of low economic status are likely to choose birth at home because of associated costs with health facility delivery [35]. The National Health Insurance Program (NHIP) covers health expenses during antenatal period and delivery of women. However, there are other costs related to pregnancy and delivery that need to be considered and acknowledged when interpreting our findings and when advocating for institutional delivery. In our analyses, we found that lack of emergency funds during pregnancy resulted in an increased likelihood for home delivery among rural women, but found the reverse for urban women. Especially in rural areas and geographically isolated and disadvantaged areas (GIDA), transportation costs to the facility may be expensive as health facilities usually require long-distance travels that could discourage women from seeking care at these facilities. The present analysis was unable to account for time and distance for travel, but this theory on distance being associated with place of delivery has been refuted by a local study, with most home deliveries being close to health facilities [36]. A geospatial analysis in Indonesia supports this finding that area of residence is not associated with place of delivery [39]. Increasing health insurance coverage is likely effective in increasing facility-based deliveries [23], but we also need to account for cultural context and maternal satisfaction when we interpret findings on this topic and make recommendations for policy [24, 36, 40].
Women’s decision-making power in healthcare also influences the choice of place of delivery. We found that women in both urban and rural settings whose healthcare decisions were based on their husband, partner, or someone else were more likely to give birth at home. This is consistent with previous reports that there is a negative effect on the use of institutional services when women themselves are not the ultimate decision maker [28, 41–43]. This observed relationship however is more complex as others have documented an increase in home deliveries among women who make decisions for themselves [21, 44]. Additionally, in our study population, there were more married women than those living with their partners, and we found cohabiting decreases the odds of giving birth at home among rural women but not among urban women although evidence is weak. This may be related to empowerment of women in making decisions as more women living with their partners made healthcare decisions on their own. However, we were not able to account for the beliefs of the women and their partners that could influence the association and could explain the slightly higher odds for home delivery observed among urban women. Also related is education, which is sometimes used as proxy measures for women’s autonomy [42]. Most women and their partners in our study received a secondary education or lower with an increased odds of home delivery among those with primary or lower education relative to those with higher education. The association is attenuated among rural women. Most studies on pregnancy and delivery study the role of education in influencing women’s healthcare decisions [21, 27, 28, 30–32, 34–36, 41]. Possible explanations that support our results include educated individuals having better access to health service information and having the ability to evaluate and apply such information, making them more health literate and thus more likely to seek care during pregnancy. These individuals are also likely to belong in richer wealth categories, allowing them to access quality health services better and with greater ease. Women who belong in poorer wealth categories in our study also had fewer ANC visits, and having inadequate ANC visits increases the odds of home delivery among urban and rural women. Our findings on decision-making power, the role of partners and networks, and the role of education highlight the importance of co-creation and patient participation in health service delivery involving not only the mother and her partner, but her parents and other family members, and close networks as well [30, 31]. This includes developing innovative strategies that leverage social capital and networks of women such as providing incentives to traditional birth attendants whom women go to for healthcare to refer them to health facilities [45]. Beyond empowering women and making health information and services more accessible, providing women with better formal education is also as important as it influences health outcomes of both mother and child [42].
Our study showed that each increase in age of women decreased the odds of home delivery in our study. This is in conflict with other studies that reported increase in maternal age increased the odds of home delivery [30, 31]. Previous experiences in health services may have been positive among our study population. It is also possible that these women had pregnancy complications that required an institutional delivery or that older women have greater autonomy and decision-making power. However, only decision-making power was measured in our study with complications and past experiences not measured in the NDHS. Because there is still a considerable proportion of women who deliver at a relatively young age, the reproductive health program may need to integrate and emphasize the importance of institutional deliveries. In our study, we also found that a significant proportion of women reported parity of at least two and a relatively high birth order number. The odds of home delivery increased with parity of two or more for both urban and rural women. For birth order, each increase in birth order also increased the odds of home delivery for both populations. Previous literature similarly reported that multiparity is associated with home delivery with the following possible explanations: larger family sizes demand more time and resources from the mother that could limit her ability to avail of health services, and uneventful deliveries with traditional birth attendants may lower risk perceptions of home delivery [27, 35]. Because the risk of complications increases with each additional pregnancy [46], women should be supported throughout pregnancy and delivery to encourage them to continue having institutional deliveries for all their pregnancies.
The study has methodological limitations that should be considered. Marital status in the model does not take into consideration data from women who are single, separated, widowed, or divorced. We restricted our analyses to women with partners so we could study the association between partner’s education and women’s decision-making power, which have been identified in the literature as important determinants of place of delivery [19, 21, 27–31]. Exclusion of information from women without partners at the time of the survey resulted to minimal loss of information with only less than 4% of single and separated, widowed, or divorced women with the outcome among the entire population of 47,244 women. Our study may also have unmeasured confounding that could affect the observed associations as our analyses relied on data collected and reported in the 2017 Philippine NDHS Births Recode. Some possible factors based on previous studies include receipt of health information during pregnancy, knowledge of danger signs, transport availability, time and distance to a health facility, past experiences, and history of complications. Despite this, we controlled for the effect of other known important factors such as wealth index, parity, birth order, ANC visits, and emergency preparedness. Fourth, if there are systemic differences in the women in the urban and rural communities relative to the women included in the study, there could be potential for selection bias. We also excluded observations due to missing data, which could also be a source of selection bias. Additionally, because MMR remains high in the country remains high [1, 5, 9], our study could potentially have Neyman bias[47] where those with poor maternal outcomes or those who have died due to complications during pregnancy or delivery could not be included in the survey. However, because this is a secondary analysis, we are not able to quantify the biases present arising from various sources such as non-participation. There also could be recall bias present because the information collected in the survey was based on self-reporting. However, the present analysis was restricted to women with one birth in the last year preceding the survey and no longer included information from other births from previous years minimizing potential for recall issues. Our study also focused mainly on variables that are ‘static’ and unchanged over time that could further minimize this problem. As typical in cross-sectional study designs, our study cannot establish a clear temporal association between the study factors and place of delivery. Despite this, our study provides updated contextual evidence on key determinants of home delivery in the Philippines, which is one of the countries that accounts for 80% of child deaths worldwide [2].