Utilization of Kenya’s free maternal health services among women living in Kibera slums: a cross-sectional study

Introduction This study was aimed at determining factors affecting utilization of public health facilities by pregnant Kenyan women living in Kibera slums, Nairobi since the implementation of the Free Maternal Service (FMS) Program in 2013. Methods This was a cross-sectional study done on 396 women who delivered between 2014 and 2015. Interview questions addressed socio-demographic characteristics, perception of quality of care in public health facilities, awareness of the FMS Program, antenatal care (ANC) and delivery service utilization. Results 43.9% delivered in a public health facility, 30.3% in a private non-profit health facility (NGO), 22.7% in a private health facility and 3.0% at home. Of the 97% of the women who delivered in a health facility, only 43.9% delivered in a public health facility despite these facilities having free maternal services. Factors that favoured the Free Maternal Service uptake included a positive perception of the public health facility, living within close proximity, learning about the Program from a support group and a short waiting time before being examined by the doctor. On the other hand, safe delivery, quality of service, accessing a health facility on foot, ANC attendance at a private and a non-profit health facility were associated with low uptake of the free maternal services. Conclusion The uptake of the Free Maternal Service program can improve if the Kenyan government directs its efforts towards changing women’s perception on quality of care in public health facilities and to improve access to health facilities in slum areas of Nairobi.

.The presence of a skilled birth attendant (SBA) during child birth, readily accessible care in emergency cases and effective systems of communication and referrals, are crucial interventions in improving maternal health [3,4]. Women living in the slums of Nairobi, the capital city of Kenya, were at a disproportionately higher risk of maternal mortality. For instance, the MMR in two of Nairobi's slums was estimated to be 706 per 100,000 live births [5]. Lack of access to quality maternal care partly contributed to the high maternal risks in these slums [6]. A recent survey showed that 83% of women in these slums delivered at a health facility [7]. However, previous studies documented that slums were mainly served by privately owned and unlicensed informal health facilities with limited skilled personnel and equipment. Most formal health facilities were located outside the slums [8,9]. Other barriers that hindered access to SBAs included poor decision makingat the family level regarding health, limited physical access to formal health facilities, high cost of health services and fear of experiencing negative attitudes from health care workers at formal health facilities [10][11][12]. The Kenyan government with the goal of increasing access to SBAs, implemented the Free Maternal Service (FMS) Program on June 1, 2013. This new policy exempted women from paying for delivery services at public health facilities [13]. After the introduction of this policy, other researchers studied the barriers to its effective implementation from a healthcare worker's perspective [14,15].
However, as far as is known, no study has investigated the uptake and access of skilled delivery services in public health facilities by women residing in the slums of Nairobi since the implementation of the FMS Program. Thus the aim of this study was to assess the levels of utilization and factors influencing the uptake of SBAs in public health facilities by expectant women residing in Kibera slums, Nairobi, for delivery since the implementation of the FMS Program.

Study design and setting
This cross-sectional study was conducted in May 2016 in three of the seven administrative geographical sub-locations of Kibera slums in Nairobi, Kenya namely: Gatwekera, Makina and Laini-Saba.
Kibera, the largest slum in Kenya has an estimate of approximately 170,078 inhabitants.

Sample size estimation
Using the estimated proportion of deliveries attended to by SBAs (62%) in Kenya, as reported by the most recent demographic data available at the time of the study [2], the sample size was calculated using the Cochrane formula [16]. The sample size was thus estimated at 363 participants.

Sampling procedure
A multiple stage sampling technique was used to recruit the 396 study participants. First, the three sub-locations: Gatwekera, Makina and Laini-Saba used to recruit participants, were randomly selected from the seven sub-locations in Kibera. A probability proportional to the population was used to determine the number of women required in each sub-location. Finally, the women to be included in this study were consecutively sampled from their households.
Women were eligible if they had lived in Kibera slums for a minimum of one year prior to delivery and had given birth in 2014 or 2015. Where a woman had delivered more than one child between 2014 and 2015, data collection was based on their last delivery. If a household had two women who qualified for the study, the participant was chosenthrough balloting.

Data collection instrument
The data was collected using a structured questionnaire administered by the researcher and trained research assistants. It was adapted from the household survey carried out by the World Bank and the African Population and Health Research Centre (APHRC) of Nairobi as part of a multi-faceted Maternal Health Study [17,18]. In order to meet the language needs of all the study participants, the questionnaire was prepared in English and Kiswahili (Kenya's national language) using a forward and backward translation technique. The questionnaire addressed information on socio-demographic characteristics, patterns regarding antenatal care and delivery service utilization, women's perception of quality of care in public health facilities, and the women's awareness of the FMS Program. Prior to data collection, pretesting of the questionnaire was done in other unselected sub-locations. The principal investigator and a field supervisor closely followed the data collection process and checked for completeness and consistency of the questionnaires.

Variables
The dependent variable of interest was the dichotomous variable; Although not a key independent variable, the potential influence of socio-demographic characteristics on a woman's place of delivery, was included in the analysis to give a more comprehensive picture.
Socio-demographic characteristics were assessed on: age, religion, marital status, number of children, education level, employment and income level.

Statistical analysis
Statistical Package for Social Sciences (SPSS) versions 21 was used to analyze data. To present the women's characteristics and their use of delivery care, descriptive statistics was used. The association between the dependent and independent variables was assessed using bivariate and multivariate analysis. The former was performed using chi-square and logistic regression analysis tests. The confidence interval (CI) was set at 95% and results considered significant at a p value of < 0.05. For the logistic regression analysis, the odds ratio was also taken into account. To control for potential confounders, independent variables found to be significant at the bivariate analysis were included in some multiple logistic regressions.

Ethical clearance
Ethical approval for this study was obtained from the University of Nairobi and Kenyatta National Hospital's Ethics Review Committee.
Informed consent was obtained from all participants after full explanation of the study design and purposes. In cases where a participant was below 18 years, consent was obtained from the parents/guardians.

Socio-demographic characteristics of women who delivered in 2014 and 2015 in Kibera slums
The enrolled women in the study were mainly in the 20-34 years age bracket, of Christian faith, married and literate. About 30% of these women had one child, 35% had two children and 20.3% had three children. Majority of the respondents were unemployed. Of those employed, 18.7% earned between 0 and 5,000 Kenyan Shillings and 14.9% earned between 5,001 and 10,000 Kenyan Shillings (Table 1).

Place of delivery
A total of 396 women who had given birth in 2014 and 2015 were enrolled in the study. The results showed that 384 (97%) delivered at a health facility. Of the total respondents, 174 (43.9%) delivered at a public health facility, 120(30.3%) at an NGO health facility, 90 (22.7%) at a private health facility and 12(3.0%) delivered at home ( Figure 1). Of the 12 women who delivered at home, 11 had planned a health facility delivery, of which 3 in a public one, 2 in a private one and 6 at an NGO health facility.

Main reasons for not giving birth in a public health facility
Long distance from the public health facility (42.6%), perceived poor quality of care (23.8%) and negative attitude of the health care workers' (15.8%) were among the reasons advanced by women for not using public health facilities for delivery. Another 13.9% cited other reasons including fear of: being charged for delivery, child being stolen from facility, and death of baby as experienced during previous delivery.

Bivariate analysis for the utilization of skilled birth attendants in public health facilities
The results of the bivariate analysis (Table 1), indicated that marital status was significantly linked to the utilization of public health facilities. As shown in Table 2, women who had a complication during their most recent pregnancy (p < 0.0001), also women who had complications during the previous pregnancy (p < 0.0001), also women who were referred during labour (p < 0.0001) had a higher utilization of public health facilities compared to private health facilities (NGO or private for profit health facilities). Women who chose a health facility because they expected good quality of services (p < 0.0001) and because it was near (p < 0.0001), had a higher utilization of private health facilities (NGO or private for profit health facilities). Table 3 depicted that awareness of the FMS Program was not significantly associated with utilization of public health facility delivery. Women who attended ANC at a public health facility had a higher utilization of public health facilities for delivery compared to women who attended ANC at a private or an NGO health facility (p < 0.0001). The results in Table 4 showed that delivery in a public health facility was more likely if women had a positive perception (OR = 4.9, 95% CI = 2.5-9.7) of public health facilities within their closest proximity.
The results in Table 5 showed that the time taken to get to the health facility, the waiting time between arrival at the health facility and before examination by the doctor/nurse and mode of transport used to get to the health facility, was significantly associated with the utilization of public health facilities (p < 0.0001). The study results showed that at 47%, walking to a health facility was the most common means of transport used by the women (Data not shown). Only 15.3% of these women who travelled by foot delivered at a public health facility compared to 84.7% of women who delivered in either a private or an NGO health facility (p < 0.0001).

Discussion
The study assessed utilization of public health facilities since the implementation of the FMS Program. Considering the most recent delivery a mother had, the study showed that 97% of women delivered in a health facility (public, NGO or private health facility).
This estimate was not only higher than the 88.5% observed in a 2012 Nairobi Cross-sectional Slums Survey [7], but also the national average (62%) [2]. This could imply that the fee exemption policy offered by the Kenyan government encourages women to deliver in health facilities, whilst reducing cost barriers as observed in other developing countries, Ghana and Laos [19,20]. However, only 43.9% of the women chose to deliver in a public health facility. The results of our study showed that utilization of public health facilities was significantly higher among women who were referred or had had complications during their most recent or previous pregnancies.
The authors further noted that NGO health facilities also offered free delivery services in Kibera slums. This could have deterred some women from seeking delivery care in public health facilities. In contrast to a previous study on user fees which found that women were more likely to seek professional delivery care when they are aware of the free delivery services offered [21], this study suggests otherwise. This is because there was no statistically significant Perceived low quality of care is a major barrier to the utilization of maternal health services and can lead to a first delay in deciding to seek care [22][23][24]. This analysis revealed that women with a positive perception of public health facilities had a significantly increased likelihood of delivering at the same in comparison to women with a negative perception. Additionally, women who chose a health facility because they expected good quality of services and a safe delivery, were more likely to deliver at a private or NGO health facility. Previous research and reports have addressed the issue of poor quality of care and shown that public health workers are often disrespectful, unfriendly and neglectful [25][26][27]. This issue of poor quality of care may be linked to a generalized problem of the healthcare delivery system [25].
Our results on factors affecting access to public health facilities, showed that majority of the women accessed the health facility on foot which was found to significantly decrease the likelihood of delivering at a public health facility. Previous qualitative studies in similar contexts have found that poor road networks within Nairobi slums made transportation facilities inaccessible [8,10]. The inability to access appropriate means of transportation during labour may have deterred a significant proportion of women from accessing public health facilities for delivery. Additionally, the limited availability of public health facilities in the slums [28] likely contributes to the way in which the private sector is meeting the women's need for maternal health care. Moreover, women who mentioned choosing a health facility because it was in close proximity, showed an association with delivering at NGO or private health facilities. This study however, did not measure the actual physical distance to assess how it affects women's access to delivery services in public health facilities. Moreover, longer waiting times was significantly associated with public health facility delivery utilization. Perhaps women who are aware of the longer waiting times in public health facilities are less inclined to seek professional delivery care in these facilities.
Further research conducted on the policy is recommended to evaluate the women's' satisfaction level with the FMS program.
Future research should identify whether women are satisfied with the maternal service provisions at public health facilities and whether they would return for future delivery or recommend someone to deliver there.

Strengths and limitations of the study
As far as is known, this was the first analysis that determined the levels of utilization of SBAs in public health facilities and assessed the factors that influenced its utilization since the implementation of the FMS Program. One of the limitations in this cross-sectional study, was the inability to provide a cause-effect relationship. Recall bias was a potential limitation since the data collected was selfreported by the women. However, it is hoped that this may not have been an issue because recall bias is less likely with issues regarding pregnancy, which are viewed as less sensitive [29]. In addition, there was no validation of the information provided by the women. Data was collected by community health volunteers who are known within the community and, this could have led the women to provide sociable desirable answers leading to bias. Some eligible candidates such as women who work, may have been missed during the sampling process but, attempt was made to regulate this by collecting data during different times of the day.
Moreover, women who delivered at home or lost their child during delivery, could have refused to participate in the study.

Conclusion
This study showed that the proportion of women in the study area utilizing health facilities for deliveries was very high. However, the

Competing interests
The authors declare no competing interests.

Authors' contributions
Angela Owiti is the principal author. She designed the study, analyzed the data, interpreted the results and participated in the writing of the paper. Julius Oyugi and Dirk Essink participated in the formulation of the study design, supervised the statistical analysis and edited the manuscript. All authors read and approved the final manuscript.

Acknowledgments
The authors wish to thank all the research assistants who took part in the data collection and the women who agreed to participate in this study. The study was funded by Angela Owiti and did not receive any financial assistance from an institution/organization for the same.