Introduction

Maternal mortality is one of the most daunting health challenges facing the world today [1]. Sustainable Development Goal 3 is the reduction of the global maternal mortality ratio to less than 70 deaths per 100,000 live births by 2030 [2]. In Tanzania, which has an estimated population of 56 million, the maternal mortality ratio remains unacceptably high, at 556 deaths per 100,000 live births in 2015–2016 [3]. There has been no notable progress over the past decade despite the existence of proven solutions and increased support at both the national and international levels [3–5].

One important global strategy to reduce maternal and perinatal deaths in low- and middle-income countries, such as Tanzania, is to increase the number of women giving birth in a health facility [6–9]. Currently, less than two-thirds of births in Tanzania take place in health facilities (63%), but the Government of Tanzania aims to reach 80% skilled birth attendance by 2020 [3]. A mid-term review of the national strategic plan to accelerate the reduction in maternal, newborn, and child deaths in Tanzania (2008–2015) showed a disparity in skilled birth attendance coverage across the country’s six zones, with the lowest rates in Western and Lake Zones [6]. The two regions that are the subject of this study, Mara and Kagera, are located in the Lake Zone, in northwestern Tanzania along the western and eastern shores of Lake Victoria. The organization of health services and infrastructure is similar in the two regions, and both have around six health workers per 10,000 population [6]. The regions of the Lake Zone have some of the worst maternal health indicators in the country and have been prioritized by the government for strategic investments and interventions given their poor coverage of key reproductive, maternal, newborn and child health (RMNCH) interventions, such as skilled birth attendance, postnatal care, and family planning [6]. According to the 2015–16 Tanzania Demographic and Health Survey (TDHS), 97% of women in the Lake Zone attended at least one antenatal care (ANC) visit during pregnancy, but the institutional delivery rate was only 49.8% [3].

Researchers have identified several factors that may influence institutional delivery rates and skilled birth attendance, although there is limited information from northwestern Tanzania, particularly in Kagera and Mara. Studies have found an association between institutional delivery rates and socio-demographic factors, including woman’s education, wealth [7–9], residence [8,10,11], and parity [8,10–12]; care during pregnancy, including early antenatal care (ANC) visits [9], number of ANC visits [11,13], ANC messages regarding the importance of institutional delivery [9], and counseling on and/or knowledge of danger signs of pregnancy and labor [12,14]; and practices such as birth preparedness/complication readiness [11,14,15] and decision making on health care [8,9,12,16].

Since 2014, the government has prioritized selected interventions in Kagera and Mara regions to improve maternal health indicators, including ANC attendance, deliveries in health facilities with assistance from skilled providers, access to emergency obstetric and newborn care services, and facility responsiveness to stock-outs of RMNCH commodities [6]. To support targeted interventions, our study investigated factors that are associated with institutional delivery in Kagera and Mara. We examined various socio-demographic variables, utilization and delivery of ANC services, and male involvement. The findings can be useful in guiding the design and implementation of programs and policies to promote facility-based childbirth in northwestern Tanzania.

Materials and methods

Results

Socio-demographic factors, ANC attendance, and place of delivery

The 1,214 mothers in the study sample were about evenly divided between Kagera and Mara regions. Half (53.1%) were aged 20–29, more than three-quarters (79.7%) had at least a primary education, most (88.1%) were in union, and most (78.3%) had at least two children (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Bivariate analysis of mother’s socio-demographic characteristics and ANC attendance, by place of delivery (N = 1,214).

https://doi.org/10.1371/journal.pone.0209672.t001

Two-thirds (67.3%) of the women gave birth in a health facility and the rest at home. A greater proportion of women from Kagera (75.3%) than Mara (59.1%) gave birth in a health facility (p<0.001). The likelihood of delivering in a health facility increased significantly with educational level (p<0.001) and household wealth (p<0.001). As parity increased, the proportion of women giving birth in a facility decreased (p<0.001); 79.9% of women delivering their first child gave birth at a facility, compared with 59.7% of women delivering their fifth or higher child. Neither age nor marital status influenced place of delivery. A small minority of women did not make any ANC visits (6%) and were significantly less likely to deliver in a facility than women who made at least one ANC visit.

Content and coverage of ANC visits

Most women received counseling on at least five of 11 possible topics (71.6%) and received at least four of nine essential services (72.3%) during ANC visits. Just over half of the women (52.3%) made four or more ANC visits. More than half of the women who attended ANC (61.1%) said their male partner came to at least one ANC visit. Women were significantly more likely to deliver at a facility as the number of counseling topics (p<0.001), services received (p<0.001), and ANC visits (p<0.001) increased, or were accompanied by their male partner to at least one ANC visit (p<0.05) (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Among women who attended ANC, counseling and service coverage and number of visits, by place of delivery (N = 1,141).

https://doi.org/10.1371/journal.pone.0209672.t002

Knowledge of pregnancy danger signs, decision making, and birth preparations

Among women who attended ANC services at least once, only 28.5% of women knew at least three pregnancy danger signs. Substantial numbers of women played no role in making decisions regarding their own health care (44.8%) or where to give birth (33.9%). Birth preparations were common, with 55.3% of women engaging in all four practices (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Knowledge of pregnancy danger signs, health care decision making, husband’s participation, and birth preparations, by place of delivery (N = 1,214).

https://doi.org/10.1371/journal.pone.0209672.t003

Facility deliveries were higher for women who had greater knowledge of pregnancy danger signs (p<0.001), participated in health care decision making (p<0.05), or practiced more components of birth preparedness (p<0.001).

Multivariable logistic regression

After adjusting for all variables, including those related to ANC, five variables remained statistically significant in the extended multivariable logistic regression. Institutional delivery was independently associated with living in Kagera region (p<0.001), giving birth to a first child (p<0.05), being in the highest wealth quintile (p<0.01), making four or more ANC visits (p<0.01), and making birth preparations (p<0.001) (Table 4). However, this model was limited to the 1,141 women who attended ANC.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Extended multivariable logistic regression of factors associated with institutional delivery among women that attended ANC (N = 1,141).

https://doi.org/10.1371/journal.pone.0209672.t004

The final model included all women, those who attended ANC services as well as those who did not. After controlling for the variables that were significant in the bivariate analysis, six variables remained significant predictors of delivery at a health facility in the multivariable logistic regression. Facility deliveries were more likely to occur among women who: lived in Kagera (p<0.001), were delivering their first child (p<0.05), belonged to the top two wealth quintiles (p<0.01), knew three or more pregnancy danger signs (p<0.001), made at least four ANC visits (p<0.01), and made any birth preparations (p<0.001) (Table 5). Notably, the final model did not include all variables that were significant in the bivariate analysis. Some had to be excluded because only women who attended ANC could provide responses; these were number of topics covered, number of services received, and presence of a male partner during ANC visits.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Multivariable logistic regression of factors associated with institutional delivery among all respondents (N = 1,214).

https://doi.org/10.1371/journal.pone.0209672.t005

Discussion

Our study shows a significant association between institutional delivery and three socio-demographic variables (region, number of children, and household wealth) and three variables reflecting women’ interactions with the health care system (number of ANC visits, knowledge of pregnancy danger signs, and birth preparations). Our results suggest that the content of ANC visits and male involvement in those visits also have an impact on whether women give birth in a health facility or at home.

Women living in Kagera were more likely to give birth in a health facility than women in Mara. Regional variations in the utilization of maternal health services within countries have been reported in other parts of Africa [8,10,11,19], and experts have urged that these differences be taken into consideration when creating policies and designing programs to improve institutional delivery rates and maternal and neonatal outcomes [19–21]. In Tanzania, Armstrong and colleagues reported subnational variations in care at birth and demonstrated an association between institutional deliveries and workforce and facility density, quality of care, and availability of essential commodities at facilities [22]. While most national strategic documents for maternal and newborn health in Tanzania consider Mara and Kagera to be homogenous parts of the Lake Zone [23], our results suggest that Kagera and Mara may differ on important parameters. Our study shows that even though both are located along Lake Victoria, women from these regions differ when it comes to accessing and using maternal health services. The results suggest a need for more locally targeted interventions to increase facility deliveries. A follow-up study to examine factors contributing to differences between the two regions could provide additional insights.

Women in the top two household wealth quintiles were more likely to give birth in a health facility than those in lower quintiles. Wealth has been associated with institutional deliveries in many low-income countries [10,24,25], and previous studies in Tanzania confirm the link, including the 2015–16 TDHS [3] and a study in one district of Kagera [9]. The impact of wealth is especially important in the Lake Zone, where fewer than 3 people in 10 belong to the highest two wealth quantiles in Tanzania, compared with more than 8 in 10 people in the Eastern Zone [3]. These economic disparities call for national policies and programs promoting facility-based births to take more assertive actions in communities with the greatest needs.

All women, regardless of the number of times they have given birth, are encouraged to deliver in health facilities where they can receive emergency obstetric care, if needed. However, we found that women of higher parity were less likely to give birth in health facilities. This finding is in agreement with other studies in Tanzania and East Africa [3,8]. Although multipara women tend to normalize childbirth and, hence, may be less likely to seek care during labor, we should also consider whether women’s previous interactions with health facilities have an impact. While our study did not measure this, the literature suggests that poor experiences during past deliveries may discourage women from returning for the next birth [9,26,27]. A study in rural, western Tanzania showed that the quality of care during childbirth, including having a respectful provider, influenced women’s decision about whether to deliver at a health facility or at home [28]. Accordingly, improving the experience of care for women during childbirth could encourage them to use health facility services in subsequent pregnancies.

Our study’s findings also suggest that quantity, as well as quality, of women’s interactions with a health facility during pregnancy have an impact on institutional delivery rates. Facility deliveries increased with the number of ANC visits in the multivariable analysis. Accessing ANC, particularly making at least four ANC visits, has been a strong predictor for institutional delivery in various studies [8,10], including in Tanzania [3]. But it is not just the number of visits that matters. To promote “a positive pregnancy experience” with the health system, the World Health Organization recommends that women have at least eight ANC “contacts” that offer meaningful interactions with a provider and serve as a platform for health promotion and education as well as screening and diagnosis [29]. In our study, women were more likely to deliver in a health facility if they knew at least three pregnancy danger signs and if they made birth preparations, both of which are the products of effective health education and promotion during ANC visits. Previous studies in Ethiopia, Uganda, and Nepal have found a similar association [8,11,14,30]. In addition, the bivariate analysis pointed to the importance of other variables that reflect the quality of ANC interactions: the number of services received and the number of counseling topics covered.

A multi-country study of four African countries underscored the important role of ANC service providers in promoting skilled birth attendance [13]. However, limited time with health providers during ANC visits may limit the exchange of information [31,32]. For example, 4 in 10 women in a study in Kenya did not receive counseling on birth preparedness during ANC visits [33], and less than half of women in Tanzania, Ghana, and Burkina Faso were counseled on most danger signs [34]. Our study also revealed substantial gaps in the services and counseling provided during ANC. These findings call for deliberate efforts to ensure that pregnant women receive adequate counseling and services according to standard protocols during ANC visits.

ANC providers can educate women and their families on danger signs and guide them through a process to develop individual birth plans, which can prepare them for an institutional delivery and to make timely decisions to seek health care in case of an emergency. Despite the proven effectiveness of birth preparedness, however, the majority of women in most low-income countries, including Tanzania, do not engage in this practice [8, 14, 35]. In part, this may be due to the poor quality of education and counseling for pregnant women in Tanzania [36, 37] and other low-income countries [31, 34]. Programs should be working on health education and promotion strategies to ensure that all pregnant women are well informed of danger signs so that they can access timely care. They should also encourage women to prepare for an institutional delivery, while recognizing that women may have good reasons to plan for a home delivery instead, such as experiences with poor care or a lack of supplies at the facility [9, 22, 26, 38].

Husbands’ or male partner’s involvement in maternal health services has been demonstrated to have a positive impact on women’s utilization of services and health outcomes, including ANC, labor and delivery, and postnatal care [12, 39–42]. In our study, women were more likely to give birth at a facility if their male partner accompanied them to at least one ANC visit. Similarly, a study in South Sudan, showed that lack of husbands’ support led to poor utilization of ANC services [43]. In Demographic and Health Surveys of 28 developing countries, the most common reason given by women for not delivering in a facility was that a decision maker at the household level—often a man—considered it “not necessary” [7]. Accordingly, maternal health policies and programs should consider men to be an integral part of pregnancy care. Male partner involvement in ANC can inform men about ways to support pregnant woman, including preparations for delivery at a health facility.

Although greater education has been associated with institutional delivery in studies from India, Ethiopia, and other regions of Tanzania [8,10,11], it was not a predictor in this study. One reason may be that the vast majority of women in this study had little or no education; only 10% had secondary education. Mother’s age and marital status also were not associated with the place of delivery in our study, contrary to the findings of other research [10,11]. The homogeneity of the participants in this study (almost 9 in 10 were in union and over 70% were aged 20–34) may have made it more difficult to observe this kind of association.

Conclusion

In an effort to reduce maternal mortality in Tanzania, this study examined factors that influence the place of birth (facility versus home) in two regions of Tanzania with lagging RMNCH indicators. Policymakers and program designers can use the results from these under-studied regions to develop effective interventions for promoting institutional delivery. Our results provide further evidence that giving birth in a health facility is significantly associated with socio-demographic variables. The results suggest that national policies and programs for promoting institutional delivery in Tanzania should consider regional variations, even within supposedly homogenous zones. To reduce maternal mortality, programs must also reach out to lower income and higher parity women to prompt more women in these categories to give birth in health facilities.

Our study also provides additional evidence that programs should also focus on the number and quality of women’ interactions with the health care system and male involvement in ANC. This should lead to improved understanding of the danger signs in pregnancy and the importance of birth preparedness, which can increase institutional deliveries. And by increasing institutional deliveries, Tanzania can drive down its long stagnate maternal mortality rates.

Supporting information

Acknowledgments

The authors gratefully acknowledge the Principal Investigator of the study (Mark Kabue) and other co-investigators at Jhpiego (John Charles, Rita Mutayoba, Julius R. Majula, and Novatus B. Munaku), Johns Hopkins University (Saifuddin Ahmed), and regional health management team members in Mara and Kagera (Stanley E. Kajuna and Herman E. Kabirigi) for their role in the design and implementation of the study. We also thank Ráz Stevenson at the United States Agency for International Development (USAID) in Tanzania for her inputs into the design of the study and conceptualization of this article. The authors also acknowledge the involvement of Mustafa Njozi (formerly at Jhpiego) and Diwakar Mohan of the Johns Hopkins University in supporting data analysis. We would like to acknowledge and thank Adrienne Kols, Judith Fullerton, and Elizabeth Thompson for their review of the manuscript. We would like to acknowledge the dedication of temporary data collectors and their supervisors who walked from house to house to interview women involved in the study. We highly appreciate the support of local government authorities at various levels in Kagera and Mara for being supportive during implementation of the study. Most importantly, we would like to thank all the women who participated in this study.