Choice between home and health institutional births in the era of Kenya’s maternity subsidy

Abstract
 
 Background Adverse maternal and neonatal outcomes disproportionately afflict low and middle-income countries, which experience high-unmet need for safe and skilled attendance at birth. This study sought to investigate how choices for place of birth were made in Kenya during the era of a national maternal subsidy. Methods The study employed data from the Kenya Demographic Health Survey 2014 and involved data from women who experienced a birth around the time when the maternity subsidy was in place. After testing for multi-collinearity of variables and addressing endogenous endogeneity using two-stage residual inclusion, probit regression models were estimated. The choice for place of birth was employed as a binary outcome variable. Results Overall, data from 4,772 women were included in the analysis. The women’s mean age was 27.7 years and majority (83.8%) were married or staying with their sexual partners. Among these women, 2748 (57.5%) had elected institutional births. Regression analysis illustrated that woman’s age, the woman and partner’s education, economic empowerment, low parity, low county poverty headcount rate and access to medical insurance. Further, access to complementary reproductive services such as antenatal care and family planning and the existence of the maternity subsidy were associated with increased likelihood to choose deliveries in health facilities. Conclusions The existence of the maternal subsidy confers women increased potential to elect health institutions as a preferred place for birth, although this was influenced by other factors. These findings imply that investments, which prevent teen pregnancies, reduce domestic and national poverty, increase education attainment, expand autonomy of women in decision making and promote access to the continuum of reproductive health services can optimize choice making during the existence of the maternity subsidy favoring access to safe institutional births.



Introduction and Background
Kenya, a lower middle-income country in SSA, has experienced both a high maternal mortality ratio (MMR) ranging between 360-750 deaths/100,000 births and a high neonatal mortality rate (NMR) ranging between 19.9-39.1/1,000 births. Kenya launched a maternity subsidy in terms of free maternity services (FMS) in 2013 through a budgetary allocation of Kshs. 3.7 billion. These funds were to be channeled to county government-run health facilities, faith-based and national referral hospitals and sought to provide direct re-imbursement of service providers in public and selected private facilities for any skilled deliveries conducted. A 22% and 17% point rise in skilled birth attendance for normal and caesarian deliveries respectively was the anticipated immediate outcome. The subsidy sought to remove service-related costs, through fee exemption to encourage poor women to access skilled attendance at birth.
Anecdotal evidence reports increased interest among women to access institutional births. The subsidy has received accolades from policy makers (1)(2). However, it is plagued by multiple implementation challenges which prevent optimal attainment of its goals (3)(4)(5). Many women continued to deliver at home despite the fee-for-service exemption. It seems that how women and their families make decisions about where to deliver is determined by factors other than the anticipated cost of the services. This study investigates these factors in order to unearth determinants of choice of place of birth between home and institutional delivery during the era when the subsidy is actively being implemented. The study uses nation-wide data from the Kenya Demographic and Health Survey (KDHS) 2014, which was collected using a robust sampling criteria and sought to examine decision making for place of birth and compare these during and prior to the subsidy.

Analytic framework
According random utility theory (RUT), when a woman is presented with two alternatives regarding where to deliver their baby (at home or in a health institution), a rational individual will prefer the choice which maximizes utility. Choice theory assumes a decision-maker is adequately informed of the benefits and risks for each choice and is rational. This study uses random utility principles to determine the choice of place of delivery during the era of an existent subsidy. The differences in the utility between a home and institutional delivery is determined by the probability of selecting one of the two choices and can be presented as a probability function which can be derived using a probit model.
A probit model applies in decision-making circumstances when the dependent variable has binary categories and demands that there should always be a reference category against which the probability of the other choice is computed. Additionally, a probit model assumes that the errors terms are normally distributed. The probability that a random woman will choose to deliver in a health facility can be computed using equation 1.

…………………………………………………………………………………….(1)
Where y* is the place of birth; are the estimated coefficients, are vectors representing the unique characteristics of the woman as well as the socio-economic and health system factors; and the error term . The signs and magnitude of the coefficients for each of these characteristics can be estimated and the marginal effects for the two choices of place of birth can be computed. Further, the probability of a woman choosing either a home or an institutional delivery can be expressed using a multivariate regression model using her unique characteristics.
These include individual maternal factors, socio-economic factors, perceived benefits, accessibility of health institutions and access to an economic subsidy. months after maternity subsidy was approved). The study analyzed secondary data containing no identifiable information on the respondents, which is freely available to the public upon fulfilling a data request and no ethical considerations were required.

Study findings
This study analyzed data from 4,772 women who experienced a delivery during the period of interest among 9,892 women who completed the full women questionnaire. Table 1 summarizes the frequencies, mean, standard deviation, minimum and maximum values for all variables.
Women included in the analysis were of medium age with a mean of 27.8 years (standard deviation=6.477) and age ranging between 15 -49 years.  Results of the probit model in which 2SRI was employed are summarized on Table 3.  On the other hand, variables such as women from large-sized households, a high number of experienced births, owning land, alone or jointly with spouse, being married, being affiliated to Christian religion and originating from a county with a high poverty headcount rate were negatively associated with choosing an institutional birth.

Average marginal effects
Average marginal effects for choice between a health facility and home delivery were estimated using the explanatory variables and are presented on Table 4.

Discussion of the findings
The choice women make about where to deliver is positively or negatively influenced by multiple factors including individual factors, household factors, economic factors and access to reproductive health services including ANC as posited by Gabrysch and Campbell (6) and are discussed as follows.

Women individual factors
A significant and increasing probability for women to choose a health facility delivery with an increase in their own age was observed. Studies identify that increasing maternal age increases the risk for adverse maternal outcomes such as stillbirths and peri-natal deaths and the need for interventions including a need for caesarian delivery. This might prompt the women to plan for a health facility delivery (7). It is probable that as the women grow older, they anticipate these risks and as a precautionary measure plan for a health facility delivery. In addition, increasing age grants women autonomy, expanded space to articulate their preferences with confidence, and experience reduced reliance of relatives to make critical decisions. Further, older women are more likely to have been persuaded (through their own experiences) that health institutions are acceptable and safe for delivery.
Education appears to exert a strong positive influence on decision-making for place of birth.
Attainment of post-primary education by the women and their partners shows higher probabilities of electing a delivery in a health facility in concurrence with other studies (8).
Education's capacity to empower women and their partners to negotiate and make rational decisions has been proven by investigators in similar socio-economic settings (9,10). It is plausible that the ability to process the benefits of a health facility delivery versus the complications, which might occur from a home delivery increase with higher education attainment further driving women and their spouses to make risk averse decisions. These findings validate the value of education as an investment to achieving better health outcomes.
The role of the shared decision-making and influence of the women's social environment including family and household members portrays mixed results. Surprisingly, women who were married appeared to be less likely to deliver in a health facility, a concurrence with other studies which suggest the dominance of men and their extended family in negatively influencing decision-making that guarantees safety for pregnant women (9)(10)(11)(12). Further, an increase in the household size, a crude measure of domestic burden and dependency, seems to have a negative influence on the choice for a safe place for delivery. Conversely, increasing age of the household

Social environment
The findings from this analysis identify the household as a critical determinant of the choice of place of birth. Women from poor households were found to have a higher probability of not delivering in a health facility, which validates studies in the SSA region. A population-based study conducted in Tanzania established comparable results (16). Even though delivery services have been subsidized, this does not mean that access to services is absolutely cost free. Similarly, a study conducted in Kenya associated access to delivery services with resource needs to cater for transport, meals, medical items absent at the facilities and opportunity costs related to income loss when relatives accompany the women to the health facility (17). Although these costs might appear miniscule, compared to what is covered by the subsidy, poor households experience disproportionate negative consequences when they incur minimal costs. These findings imply that women from poor households begrudgingly persevere home deliveries even when their lives are at risk.
Similar to our study, a Tanzanian study (16) identified that women from households that possessed health insurance were more likely to delivery in a health facility. These findings are consistent with analysis conducted using data from KDHS 2008-9, which highlighted that 7% of women who possessed insurance at that time were 23 percentage points more likely to deliver in a health facility (18). Coverage by insurance remains a critical influencer of accessing health services (19). It is commendable that a 5% increase in insurance coverage was observed between the 2008-9 and 2014 demographic surveys. Apart from this, it appears that the higher concentration of health institutions in urban centres in Kenya is favorable for women making choices in favor of health facility deliveries. The distance women would have to cover to reach a health facility determines if they will choose a facility for delivery and as Mwaliko et al (2014) argue, investing in building and equipping comprehensive reproductive facilities is a meaningful approach to entice women to consider seeking professional assistance during birth.

Interaction with the health system
This study establishes that women who made previous favorable decisions and accessed adequate ANC, attended ANC in a health facility, and used FP methods tended to make consistent decisions when selecting the place of delivery. Women who used FP are likely to have spaced their pregnancies allowing them to have sufficient resources to access ANC, afford a health facility delivery and sustain utilization of beneficial reproductive services. Access to reproductive services provides clinicians with opportunities to offer health education, identify danger signs in pregnancy and recommend specific interventions to the women including facility delivery, thus optimizing future utilization. Furthermore, increased experience with health services assists to dissuade existing cultural and negative beliefs about the usefulness and appeal of health facilities further expanding agency for utilization.
This study has established that with increasing number of births, women shy away from making decisions that favor deliver in a health facility. These findings are comparable to studies, which document that women with three or more children are two times less likely to deliver in a health facility (20). We contend that women who experienced several uncomplicated deliveries possess limited motivation to deliver in a health facility. Further, women are more likely to experience easier deliveries with increasing parity, which increases complacency and reduces the initiative to seek assistance during subsequent births. In circumstances where negative experiences during facility delivery are rampant, women with multiple births are likely to lose their patience and avoid health facilities. Multiple successful births translate to increased dependency, higher domestic expenditures, and reduced fiscal space that reduce willingness to incur cost to access services when cheaper alternatives exist.
More women appear to choose health facility deliveries during the time of the subsidy than before. This is very encouraging news for policy makers because the subsidy appears to yield the intended outcomes replicating results observed in other counties. Subsidies in Rwanda, Burkina Faso, Ghana, Nigeria and Ethiopia resulted in impressive outcomes in skilled birth attendance during delivery (21)(22)(23). These Kenyan results show good promise given that this study reports data for the first 15 months of the subsidy when structural implementation challenges were reported (4). With additional optimization, this subsidy has potential to attain better results given that these findings were attained during a transition period when the health function was being fully devolved to the counties and the process was faced by a myriad of teething problems (24).

CONCLUSION AND POLICY RECOMMENDATIONS
This study has elucidated the determinants of choice making for place of birth among women who experienced deliveries during the era of the maternity subsidy in Kenya. The study has identified that during the time of the subsidy, more women were likely to choose to deliver in a health facility, which highlights the impact of the subsidy towards increasing universal access to skilled births and its inherent potential in reducing unwarranted complications to the mother and unborn child during birth. The findings of this study imply that the maternity subsidy has been helpful in driving more women to access institutional births. Emphasis needs to be laid on popularizing the subsidy to the potential beneficiaries, addressing the reported challenges related to its implementation and ring-fencing the allocation of resources to sustain the subsidy.
Additionally the subsidy can be expanded to cover a package of comprehensive reproductive health services.
The study has documented modifiable factors, which can empower women and strengthen decision-making and increase decision-making agency towards choosing health facility deliveries. The study contends that higher educational attainment for women and their partners, autonomy of women and engaging women in paid work positively influences decision-making. It is crucial that affirmative approaches that retain more women in education beyond primary school, create more women-friendly employment opportunities and policies that recognize and safeguard women's rights to own assets and enable women to control their own lives are enforced. Low poverty incidence and high domestic wealth are facilitative of positive choices.
To address the quandary of low skilled attendance at birth, the government should deliberately invest in national poverty eradication schemes, implement measures that enrich households with domestic resources and reduce fertility rates to reduce dependency, which can lower domestic expenditures.
Policies which address prohibitive religious or cultural beliefs, and mitigate early pregnancies and early marriages should be enforced so that women get pregnant and are married when they are not only old enough to have sufficient agency for independent decision-making but also when they can negotiate safe practices including access to the continuum of safe maternal care services. Further, policymakers should explore disruptive and innovative approaches that can expand health insurance coverage and widen access to third party payment systems to facilitate access to births in private and public facilities.