It has been hypothesized that higher autonomy of women would translate into improved health seeking behavior and consequently, into better health outcomes [6
]. However, the results of this study show that in the Nairobi informal settlements, utilization of maternal health services for delivery is not enhanced by high levels of women's overall autonomy, freedom of movement, or decision making; a finding consistent with other studies that have examined the influence of women's autonomy on various health outcomes. Using data from the demographic and health surveys, Dodoo's findings [32
] provide no support for an association between women's autonomy and decisions regarding child health in Ghana. A study in Nepal also found that the influences of women's involvement in decision-making regarding their own health or large purchases on antenatal care attendance or skilled delivery care, were rather weak, while discussion of family planning between spouses was linked to increased likelihood of receiving skilled antenatal and delivery care [6
]. Similarly, weak relations emerged between women's autonomy – measured by decision-making and freedom of movement as we do in this paper – and the use of family planning in Oman [15
]; and no consistent relationship was found between low autonomy and child mortality among Muslims and non-Muslims in India, Malaysia, the Philippines and Thailand [13
A number of other studies revealed mixed conclusions on the effect of women's autonomy on various aspects of maternal health. While women's freedom of movement appeared to be a major determinant of maternal health care utilization among poor to middle-income women in a large urban area of Uttar Pradesh, the two other dimensions investigated (control over finances and decision-making) were not significantly associated with the outcomes studied [12
]. Similarly, Saleem and Bobak [19
] showed that decision-making autonomy was associated with contraception while movement autonomy was not.
However, a number of other studies have clearly demonstrated that women's autonomy has a strong and consistent effect on reproductive health outcomes. Bloom et al. [12
] demonstrated that women's autonomy was a major determinant of maternal health care utilization among urban poor to middle-income women in a North Indian city. The study used a two-face cluster design concentrated on households within a 15-minutes walking distance to a government or charity facility where health care is given free of charge. The proximity to health services and the fact that care was provided free of charge, may suggest that women's autonomy is a major enhancer of maternal health in a context of higher geographic accessibility and lower minimal financial costs. This view that women's autonomy may only be relevant in conducive environments seems to agree with Mason who noted that women's autonomy, coupled with education and communication with spouses greatly influences use of contraception [20
]. This interpretation is only based on speculations, and requires additional, in-depth research for support. It is worth noting that studies that have shown strong linkages between women's autonomy and maternal health outcomes have been conducted in Asia, a region with cultures that are different from sub-Saharan Africa's, especially regarding kinship structures, with possible bearing on the way women's autonomy plays a role on reproductive health. It should also be mentioned that very few studies have focused on informal settlements which may have specific conditions such as lack of social networks and cohesion among its residents.
It is also important to note that not only is the influence of women's autonomy not statistically significant in the research presented in this paper, the negative sign observed in some of the coefficients in Additional file 3
suggests that the results tend to run contrary to the expectation emanating from the hypothesis that maternal health outcomes are hampered by low levels of women's autonomy. Before we dismiss the important role of women's autonomy on maternal health services utilization, and in line with the interpretation mentioned above, we examine the interaction between women's autonomy and household wealth as determinants of maternal services utilization. Our results clearly show that women's autonomy is an enhancer of maternal health service utilization among the middle-income and least poor groups in the slums of Nairobi.
Strikingly, this study shows that among the poorest, lower autonomy women tend to exhibit higher use of maternity services, a finding which is contrary to expectation. Other authors have contemplated the possibility that the relative status of women may be inversely associated with health outcomes [32
] or have speculated that the formulation of the questions related to women's autonomy may offer at least a partial explanation of the counter intuitive results [6
]. One could argue that poorest women, because they are likely to be uneducated, may have a different understanding of autonomy-related issues.
The multivariate analyses confirm the well known effects of key socio-demographic characteristics on the utilization of maternity services. The most notable effects are those of education and household wealth, as shown by various other authors [26
]. Compared to Viwandani residents, Korogocho women were about 50% as likely to deliver at a health facility in general, and more than four times as likely to do so in an appropriate one. This could be explained by the fact that appropriate facilities in Viwandani are limited and a higher number of the respondents (57%) in this study are from Korogocho which is better served with appropriate health facilities compared to Viwandani. Also, consistent with other studies, women with higher parity were less likely to deliver in well-equipped health facilities; women aged less than 25 years were the least likely to deliver at health facilities or at the appropriate ones, a finding consistent with a study which found that teenagers in sub-Saharan Africa experience poorer maternal health care than older women with similar characteristics [34
There is a strong association between the use of antenatal care services and delivery at a health facility. Interestingly, women who were advised during antenatal visits to deliver at a health facility were more likely to do so. This highlights the need for poor urban women to have such maternal health services as contraception, delivery and post-partum care alongside other health promotion messages to improve their health-seeking behavior. Access to family planning services is also crucial in improving delivery care. About 30% of the women in this study reported that they did not intend to get pregnant with the current baby, and the multivariate results clearly indicate that unplanned pregnancies were less likely to be delivered at appropriate health facilities.
Very few studies have focused on autonomy being a mediator between education and health service utilization. Results from our study indicate very modest changes in coefficients between models with and without autonomy variables, indicating that while education is a major determinant of health seeking behavior, its effect is not mediated by women's autonomy. This result agrees with Saleem and Bobak [19
] in their study on women's autonomy, education and contraception in Pakistan.