Determinants of facility-based childbirth among adolescents and young women in Guinea: A secondary analysis of the 2018 Demographic and Health Survey

Introduction Maternal mortality remains very high in Sub-Saharan African countries and the risk is higher among adolescent girls. Maternal mortality occurs in these settings mainly around the time of childbirth and the first 24 hours after birth. Therefore, skilled attendance in an enabling environment is essential to reduce the occurrence of adverse outcomes for both women and their children. This study aims to analyze the determinants of facility childbirth among adolescents and young women in Guinea. Methods We used the Guinea Demographic and Health Survey (DHS) conducted in 2018. All females who were adolescents (15–19) or young women (20–24 years) at the time of their most recent live birth in the five years before the survey were included. We examined the use of health facilities for childbirth and its determinants selected through the Andersen health-seeking model using descriptive analysis and multilevel multivariable logistic regression. All descriptive and analytical estimated were produced by adjusting for the survey sampling using the svy option, including adjustment for clustering, stratification and unequal probability of selection and non-response (individual sample weights). The subpopulation option was also used to account for the variance of estimations. Results Overall, 58% of adolescents and 57% of young women gave birth in a health facility. Young women were more likely to have used private sector facilities compared to adolescents (p<0.001). Factors significantly associated with a facility birth in multivariable regression included: secondary or higher educational level (aOR = 1.86; 95%CI:1.24–2.78) compared to no formal education; receipt of 1–3 antenatal visits (aOR = 9.33; 95%CI: 5.07–17.16) and 4+ visits (aOR = 16.67; 95%CI: 8.82–31.48) compared to none; living in urban (aOR = 2.50; 95%CI: 1.57–3,98) compared to rural areas. Women from poorest households had lower odds of facility-based childbirth. There was substantial variation in the likelihood of birth in a health facility by region, with highest odds in N’Zérékoré and lowest in Labé. Conclusion The percentage of births in health facilities among adolescents and young women in Guinea was 58%. This remains suboptimal regarding the challenges associated maternal mortality and morbidity issues in Guinea. Socio-economic characteristics, region of residence and antenatal care use were the main determinants of its use. Efforts to improve maternal health among this group should target care discontinuation between antenatal care and childbirth (primarily by removing financial barriers) and increasing the demand for facility-based childbirth services in communities, while paying attention to the quality and respectful nature of healthcare services provided there.


Introduction
Maternal mortality remains very high in Sub-Saharan African countries and the risk is higher among adolescent girls. Maternal mortality occurs in these settings mainly around the time of childbirth and the first 24 hours after birth. Therefore, skilled attendance in an enabling environment is essential to reduce the occurrence of adverse outcomes for both women and their children. This study aims to analyze the determinants of facility childbirth among adolescents and young women in Guinea.

Methods
We used the Guinea Demographic and Health Survey (DHS) conducted in 2018. All females who were adolescents (15)(16)(17)(18)(19) or young women (20-24 years) at the time of their most recent live birth in the five years before the survey were included. We examined the use of health facilities for childbirth and its determinants selected through the Andersen healthseeking model using descriptive analysis and multilevel multivariable logistic regression. All descriptive and analytical estimated were produced by adjusting for the survey sampling using the svy option, including adjustment for clustering, stratification and unequal probability of selection and non-response (individual sample weights). The subpopulation option was also used to account for the variance of estimations.

Results
Overall, 58% of adolescents and 57% of young women gave birth in a health facility. Young women were more likely to have used private sector facilities compared to adolescents (p<0.001). Factors significantly associated with a facility birth in multivariable regression included: secondary or higher educational level (

Introduction
Reducing maternal mortality to achieve Sustainable Development Goal 3 (SDG 3) is a major public health challenge. Despite efforts over the past 25 years, maternal mortality remains disproportionately high in low-and middle-income countries where the need to address the unmet need for family planning is most significant, particularly among adolescents [1]. The World Health Organization (WHO) estimated that 12 million adolescents aged 15-19 years give birth each year globally. The problem is most prevalent in sub-Saharan Africa (SSA), with variations between countries [2]. The recent Demographic and Health Survey (DHS 2018) in Guinea reported a fertility rate of 120 live births per 1,000 adolescents aged 15-19 [3]. This rate is higher than the West African average of 115 per 1,000 [4]. Giving birth during adolescence carries a higher risk of adverse outcomes for both the girl and the baby due in part to the mother's biological and physiological immaturity [5]. An important strategy to reduce these risks is to improve skilled birth attendance in an enabling environment [6]. However, despite this particular need, a study using data from 34 countries has shown that sub-Saharan adolescents who are having their first birth are less likely to give birth in a health facility than older women [7]. Between 2000 and 2018 the global proportion of women who gave birth in a health facility is estimated to have risen from 52 to 76% and this increase is thought to have made a significant contribution to the estimated global declines in maternal and neonatal mortality. The most recent estimates for countries in sub-Saharan Africa collated by UNICEF indicate national range from <20% in Somalia or South Sudan to >90% in countries such as Botswana, Malawi, Rwanda, DRC and South Africa [8]. In Guinea, 56% of adolescents used health facilities for the most recent live birth during the DHS 2018 recall period [3]. Factors associated with adolescents' use of facilities for childbirth are not clear across countries. In a study analyzing DHS datasets from 29 SSA countries, Doctor et al. (2018) found that the likelihood of facility-based delivery increases with maternal age [9]. However, Adde et al. (2020), in a most recent analysis including 28 SSA countries found no difference between adolescents and older women in facility-based childbirth [10]. Among women in reproductive age in SSA an analysis of DHS from nice countries showed that higher education, older age, receiving antenatal care, distance to health facility and media exposure were associated with a higher likelihood of facility-based childbirth [11]. A systematic literature review of determinants of facility-based childbirth in SSA additionally indicated that parity, urban/rural residence, household wealth were also consistently associated with higher use of facilities for childbirth [12]. Among adolescents, in a systematic review including 27 studies, Mekonnen et al. (2015) found that distance to the health facility and community factors such as the proportion of educated women in the community and the rate of ANC use in the community were associated with the use of maternal health services [13]. In Niger, Rai et al. (2013) found that the education of both spouses, belonging to a social group such as the Deermal/Songhai, and having some decision-making autonomy increased women's chances of having safe deliveries [14].
Guinea did not achieve the Millennium Development Goal 4 and is unlikely to achieve the Sustainable Development Goal 3.1 by 2030 (SDG 3.1). The government of Guinea is pursuing an agenda to achieve a maternal mortality ratio (MMR) reduction from 576 per 100,000 live births in 2017 [1] to 343 per 100,000 live births by 2024 [15]. The country has been implementing a policy of free maternal and neonatal emergency health care in public health facilities since 2011 [16]. Adolescents are a priority target for the reduction of maternal mortality and morbidity in the country. However, there are no specific efforts or strategies targeting adolescents [15]. Factors that influence the choice of location for childbirth among adolescents in the context of free maternal health services in Guinea are not well understood. The objective of this study is to estimate the levels and determinants of facility-based childbirth among adolescents and young women aged 20-24 years in Guinea. We used the 20-24 years old women as the reference or comparison category in this study for two main reasons. First, it has a high age-specific fertility rate and therefore contributes to the analysis with a relatively large sample size and second, while this is the age group adjacent in age to adolescents in terms of cultural and generational proximity and experiences, including the first birth, it faces among the lowest risks of maternal mortality and morbidity. The results of this study can support the development of targeted strategies to increase the use of maternal health services among adolescents in Guinea.

Data and population
The data used for this study come from the 2018 Demographic and Health Survey (DHS), the fifth to be conducted in Guinea since 1999. The DHS is a cross-sectional nationally representative household survey which used multilevel cluster sampling. Its purpose was to provide comparable data across time and countries on various indicators, including fertility, adult and child mortality, and use of maternal healthcare. The DHS generally uses standard model questionnaires with the possibility of adaptation and addition of optional modules according to country needs. For this study, we included all female respondents residing in sampled household, with at least live birth in the survey's five-year recall period, if they were 15-19 years (adolescents) or 20-24 years old (young women) at the time of their most recent live birth. Analyses of childbirth location were restricted to the circumstances of the most recent live birth.

Outcome variable
We used the place of birth as reported by women as the outcome variable. It was categorized into 'home birth' (respondent's home, other home, and other location) and facility-based birth (government hospital; government health center; government health post; private hospital or clinic, and other private medical facility). For a more detailed analysis of health facility type we considered births in the public versus the private sector. Within the public sector, we focused separately on government hospitals which provide the highest level of emergency obstetric and newborn care and serve as referral centers.

Independent variables
We used Anderson's Behavioral Model of Health Services utilization [17] (Fig 1) to identify relevant and available variables. We included woman's age at index birth (15-19; 20-24), woman's highest educational level at survey (no formal education, any primary, and any secondary or higher). The usual age of primary school completion in Guinea is 12 years [18]. We also included woman's marital status at survey (married/cohabiting or not), women's religion (Muslim, Christian, other), and number of ANC visits during the pregnancy preceding the index birth (none, 1-3, 4 or more). Women with 'don't know' responses on number of ANC visits were included in the 1-3 category because we assumed that these women had some ANC but could not recall the exact number of visits. The "4 or more" category aligns with the policy at the time of data collection.
Place of residence as specified on the DHS sampling strategy (rural, urban), region of residence (Conakry, Boké, Kindia, Mamou, Labé, Faranah, Kankan, and NZérékoré) and women's perception of distance to the health facility (big problem or not) were used as proxies for health service availability. Exposure to media (listens to radio or not) was used as a proxy access to health information. We used household wealth quintile (poorest, poorer, middle, richer, and richer) as defined by the DHS to capture financial access to care. Perceived need factors were mother's parity (no previous birth, 1 to 2 previous births, 3 or more births) and wantedness of the index pregnancy (wanted, mistimed/unwanted).

Statistical analysis
The data were processed and analyzed using Stata 16.1 software (StataCorp, College Station, TX USA). Two levels of statistical analysis were applied. The first level consisted of describing the characteristics of the included sample. We obtained absolute numbers and percentages of the characteristics under study for adolescents, young women, and the whole sample. We also compared the prevalence of facility-based childbirth by type of health facility between adolescents and young women using Pearson's Chi-squared test.
We then analyzed the determinants of facility-based childbirth among the combined sample of adolescents and young women. First, we conducted a bivariate logistic regression analysis to assess the relationship of each independent variable with adolescents' or young women's use of health facilities for their most recent birth. We initially ran one multilevel multivariable logistic regression model for each age group, but we found no difference between the characteristics associated with the outcome of interest between the two age groups. Therefore, we constructed a single multilevel multivariable logistic model with both age groups. The multilevel model was run to account for the unobserved characteristics at cluster or community level (second level) and region level (third level) during the data collection in addition to the women individual characteristics (first level). No cut-off was used for inclusion of variables in the multivariable models. All the variables selected through the theoretical framework adapted from Andersen's Behavioral Model of Health Care Utilization were included after assessing for collinearity between independent variables. The assessment for collinearity intended to determine the relevance of the variables selected for inclusion in the analyses. As a general rule, a mean-variance inflation factor (VIF) score below five is tolerated. In contrast, a mean score greater than or equal to five suggests that the regression coefficients could be misestimated. The VIF reported by the test in our analysis remained below five, so in line with the afore mentioned general rule. The Likelihood ratio test (LRtest) was also performed for each variable with more than two categories to assess how its inclusion impacted the over model. Regressions were performed at the 5% threshold, and the results were reported with a 95% confidence interval (95% CI). Differences were considered statistically significant at p<0.05. All descriptive and analytical estimated were produced by adjusting for the survey sampling using the svy option, including adjustment for clustering, stratification and unequal probability of selection and non-response (individual sample weights). The subpopulation option was also used to account for the variance of estimations.

Ethics considerations
A formal request for analysis of all data was made to MEASURE DHS through the online platform, and permission was granted. The original data was collected with ethical approval from the National Ethics Committee for Health Research of Guinea and ICF's International Review Board.

Characteristics of the samples
The sociodemographic characteristics of adolescents and young women included in this study are presented in Table 1. We included 934 females who were 15-19 at the time of their most recent live birth in the five years before the survey. Their mean age at the most recent birth was 17.9 years (SD = 1.3). Nearly 84% of these adolescents were married/cohabiting, 63% had no formal education, and for 65% this was their first birth. Overall, almost 89% of included adolescents reported at least one ANC visit during their most recent pregnancy, but fewer than two-fifths (36%) received 4 or more ANC visits. The majority of adolescents were Muslim (86%) and lived in rural areas (70%). However, more than half did not perceive the distance to health facility as a big problem (52%). Overall, 1220 young women aged 20-24 at the time of their most recent live birth in the five years before the 2018 DHS were included in the analysis. Their mean age at the most recent birth was 22.7 years (SD = 1.5), 92% were married/cohabiting, 63% had two or three previous births, and 70% reported no formal education. Nearly 88% of young women received at least one ANC for their most recent pregnancy, but fewer than two-fifths (38%) completed four or more ANC visits.

Proportions of facility-based delivery among adolescents and young women
Overall, about 58% adolescents and young women in Guinea gave birth to their most recent baby in a health facility. This proportion was similar in adolescents (58%) and young women (57%). No difference was noted between the two age groups in the use of government hospitals for childbirth (p = 0.332). However, young women used private sector facilities more (7.8%) compared to adolescents (3.5%, p<0.001) as shown in Table 2. Table 3 shows the percentage of births occurring in health facilities stratified by the independent variables.

Determinants of facility-based childbirth among adolescents and young women
In bivariate analysis, education level, marital status, parity at survey, exposure to media, number of ANC visits, perceived distance to health facility, religion, region, residence, and household wealth quintile were associated with facility-based childbirth in the combined sample of adolescents and young women.
In multivariable multilevel analysis ( Table 3), those with a secondary or higher educational level had 86% higher adjusted odds of facility-based childbirth (95%CI:1.24-2.78) compared to women with no formal education. Compared to those who had no ANC, women who had 1-3 ANC visits (aOR = 9.33; 95%CI: 5.07-17.16) and 4 or more visits (aOR = 16.67; 95%CI: 8.82-31.48) had higher odds of facility-based childbirth. Christian women had over two-fold higher odds (aOR = 2.40; 95%CI: 1.31-4.42) of facility birth compared to Muslim women. Adolescents and young women living in the region of NZérékoré had nearly five-fold higher odds (OR = 4.98; 95%CI: 2.52-9.85) of facility-based delivery as compared to women in Labé. Living in an urban area doubled the odds of giving birth in a health facility (aOR = 2.50; 95%CI: 1.57-3.98). The odds of facility birth increased with each wealthier household quintile. Compared to adolescents and young women living in the poorest household, those in the richest households had six-fold higher odds facility-based childbirth (aOR = 6.06; 95%CI: 3.21-11.42). Women who reported having their first birth had 47% higher adjusted odds of facility childbirth compared to those with one or two previous births (p = 0.005).

Discussion
This study examined recent levels and determinants of facility-based childbirth among adolescents and young women in the Republic of Guinea. The overall facility-based childbirth level found was 58%, comparable among adolescents and young women. The study also found that about 6% of women included in the analysis gave birth in the private sector. This was more frequent in young women than adolescents. In addition, most adolescents and young women had initiated prenatal visits, although few reached the recommended four or more ANC visits (37% of the combined sample). In multivariable analyses, education level, number of ANC visits, religion, residence, administrative region, and household wealth index were associated with facility-based delivery. The strongest predictors of higher facility-based childbirth were the use of 4 or more ANC visits, household wealth, and residence in the NZérékoré region. In addition, compared to respondents who were having their second or third child, women giving birth to their first child had higher adjusted odds of giving birth in a health facility. The overall levels of facility-based childbirth among adolescents and young women found in this study (58%) were statistically higher than what was reported from the same DHS among all women of reproductive age in Guinea (53%). In Guinea, one of the most important policies improving access to maternal health services during this period is the 2011 user fee removal policy [16]. As a result, the percentage of all births occurring in health facilities has increased from 40% in 2012 to 53% in 2018. This increase is even more pronounced in the under 20 years age group (from 41% to 58%, p<0.001) [19,20]. A similar increase has previously been noted in other sub-Saharan African countries, including Uganda [21].
The higher percentage of facility-based childbirths among respondents aged 15-24 compared to the general population may be explained by parity. A higher percentage of births to adolescents and young women are first births compared to older age groups and first births are perceived to be at greater risk of complications than subsequent births and are therefore more likely to take place in facilities.
Despite the increase in facility-based childbirth observed in our study as compared to 2012, the overall prevalence of facility-based childbirth among adolescents and young women in Guinea is far behind the average for sub-Saharan African countries (65%) [10], including neighboring countries such as Senegal (84%) and Mali (71%) [22,23]. Furthermore, we believe that this result is sub-optimal in light of the challenges associated with high maternal mortality and morbidity in the country and is hampering the achievement of Sustainable Development Goal 3 by 2030.
Our study identified large inequalities in the use of health facilities for childbirth across household wealth quintiles and geographic regions. Compared to adolescents and young women living in the poorest households, those from richer and richest wealth quintiles were more than five times as likely to deliver in a health facility, adjusted for other determinants. This suggests that despite the user fee removal policy, women and their families continue to face substantial financial barriers to accessing childbirth care, potentially including payment of informal fees and the indirect costs of seeking care, such as transportation, food, and sometimes housing. These findings are consistent with reports from other sub-Saharan African countries [7,24,25]. In regard to geographical inequalities, we found that while adolescents and young women living in Conakry had the highest bivariate odds of facility-based delivery, it was the region of NZérékoré in which the highest adjusted odds were found (compared to Labé). While an in-depth analysis for understanding of the reasons for this finding is necessary, we suspect that local health system actors are implementing national guidelines differently across the country, depending on local challenges such as the availability of financial and human resources. Similar disparities have been reported in Burkina Faso, Niger, Nigeria, Ghana, and Senegal [26]. Next, we draw attention to the fact that the increasing number of births occurring in health facilities, among all women in general, and adolescents and young women in particular, might have consequences for the quality of care provided at childbirth without a concomitant increase in resources. Current challenges of the health sector include health infrastructure, staffing, and funding. For instance, in 2019, only 51% of the public sector health facilities had an adequate physical state of functionality [15]. According to a Guinean Ministry of Health report from 2014, there were approximately 108 obstetrician-gynecologists, 409 midwives, and 1189 nurses for the population of almost 12 million [27]. While our study did not assess the quality of care women received during childbirth, we report patterns consistent with concerns about the quality of care. For example, while only a small percentage of women 15-24 used private sector facilities for childbirth (6%), this level was twice as high among young women compared to adolescents, which might be partly related to previous experience with maternal health services (young women have on average higher parity). Previous research in Guinea showed that lack of staffing, equipment, and space within health facilities generates a negative feedback loop resulting in high levels of disrespect and abuse during childbirth in health facilities. This included women being slapped by attendants, being verbally abused for not complying with health workers' requests, and giving birth on the floor [28]. A study of three urban hospitals reported that nine out of ten episiotomies were performed without consent, and fewer than 10% of women had access to a companion of choice during labour and childbirth [29]. Importantly, this study of four countries, including Guinea, found that young and less educated women had the highest rates of disrespectful care.
Our study findings also have important implications about the continuity of maternal care during pregnancy and childbirth and beyond during the women's life-course. Most adolescents and young women in the sample reported receiving some ANC, but only a third had the recommended four or more ANC visits (37%). Receiving four or more ANC visits was associated with about 17 times higher adjusted odds of childbirth in a health facility compared to no ANC. This value ranges from 9 to 31 due probability to the sample size. While this association might be partly confounded by higher use of both services by women with increased risks (e.g., twin pregnancies), the findings highlight the fact that timely initiation and contact with ANC during pregnancy is a critical factor for the continuity of maternal care during childbirth and the postnatal period. This is the case for high-quality ANC services as shown in a continuum of care analysis from Guinea [30]. ANC is an opportunity for a woman to discuss what her pregnancy implies for herself and her child with a skilled provider, ask questions, and develop a relationship of trust with the health system. It is also an opportunity for the providers to detect potential risk factors and educate the woman on the benefits of continued health service utilization [31].
We found that primiparity was associated with a higher facility-based delivery. This is consistent with studies conducted in sub-Saharan African contexts [12]. One reason for this could be the perceived higher health risk attributed to the first pregnancy in such contexts [32]. We hypothesize that after the first delivery in a facility where the woman may experience poor quality services and mistreatment, she would choose to deliver in a private facility, if she affords, or within her community for the subsequent deliveries [33]. However, a study on facility-based delivery including DHS from 34 sub-Saharan African countries reported that among women having their first birth, the youngest had significantly lower likelihood of delivering in a facility [7]. In any case, experiencing poor quality disrespectful care at the beginning of reproductive life carries negative consequences for the woman's future use of health services, her wellbeing and survival, and that of her child. This is especially the case in Guinea where the total fertility rate of 4.8 [20] and high maternal mortality rate (576/100,000 live births in 2017) combine into a very high lifetime risk of maternal death of 1 in 35 [1].

Limitations
This study benefited from a recent nationally representative sample of women of childbearing age. However, we note some limitations. First, the study's cross-sectional nature does not allow for a temporal relationship between the independent variables and the outcome of interest. Second, there is also the possibility of recall bias that characterizes the DHS, resulting from the retrospective nature of the information collected. Third, because some variables such as ANC were only available for the most recent live birth, we analyzed this subset of births. This might have resulted in underrepresenting the experience and determinants of women with higher number of births in the recall period. Fourth, we used a subsample of women of reproductive age by reducing the sample size to women aged 15-24 years old. This might have also resulted in misestimating the experience and determinants of women aged 15-24 years old at the regional level. Fifth, the analysis did not include some relevant variables that might have considerably impact women's willingness and ability to give birth in a health facility, such the quality of care offered or the distance or travel time to facility. Last, while the DHS data are most recent, they capture the situation from 4-9 years prior to this analysis, and might no longer accurately represent the current situation.

Conclusions
This study showed an overall facility-based births rate of 58% among the study population in Guinea, comparable between adolescents and young women. This rate was higher than in the general population but remains suboptimal as compared the 90% goal of facility-based childbirth. A multilevel multivariate analysis was used to account for unobserved community and regional characteristics in the analysis of determinants of facility-based childbirths among adolescents and young women. The strongest determinants of higher facility-based childbirth identified in the study were the use of ANC visits, household wealth, and residence in the N'Zérékoré region. In the pathway to achieve the SDG targets by 2030, there is a need to address inequalities related to financial and geographic accessibility given that the poorest women are still unable to afford the costs of childbirth in health facilities. Also, positive lessons for the regions achieving higher levels should be documented and scaled up. Another gap identified in this study relates to the quality of care across the continuum of care, which requires more investment and accountability in the health sector. Extending beyond the care received at childbirth, attention to accessible, acceptable, affordable, and high-quality maternal care must be paid to antenatal care as the first step in the continuum of care and an important determinant of the use of facilities for childbirth. This is particularly important among adolescents and young women whose well-being will be affected by the care experience and health outcomes experiences at the beginning of their reproductive lives.