Background
Appropriate delivery care is crucial for both maternal and perinatal health and increasing skilled attendance at birth is a central goal of the safe motherhood and child survival movements. Skilled attendance at delivery is an important indicator in monitoring progress towards Millennium Development Goal 5 to reduce the maternal mortality ratio by three quarters between 1990 and 2015 [
1]. In addition to professional attention, it is important that mothers deliver their babies in an appropriate setting, where life saving equipment and hygienic conditions can also help reduce the risk of complications that may cause death or illness to mother and child [
2].
Over the past decade interest has grown in examining influences on care-seeking behavior. As cited in the "three delays" model, three main inhibitors to health care service utilisation exist: the delay in deciding to seek care, the delay in reaching an adequate health care facility and the delay in receiving adequate care at that facility [
3]. The first delay may be due to a lack of understanding of danger signs, the absence of the decision maker from the household, the low status of the woman, cost, previous unsatisfactory experience with the health care system and perceived low quality of care [
4]. Phase 2 delays may be due to distance from facility, lack of transportation, difficult terrain and the high cost of travel [
3].
Research consistently shows that high cost is an important constraint to service utilization particularly for the poor [
5‐
11]. In India studies show a very high out of pocket expenditure on delivery care, and, although the private sector is more expensive, the cost of public sector inpatient care services has increased since the 1990s [
12]. Hence, income is a major determinant of care seeking [
13]. Recent analysis of the third National Family Health Survey (2005/6) shows 13% of women in the lowest wealth quintile accessing institutional delivery care compared with 84% in the highest [
14]. The importance of proximity to health services as a factor affecting utilization has also been stressed. It exerts a dual influence on health care utilisation. Long distance can be an obstacle to reaching a health facility as well as a disincentive to even try to seek care. Rural populations are particularly disadvantaged as they often lack reliable means of transportation. A sizable proportion of maternal deaths in developing countries occur on the way to hospital; other women are almost beyond help by the time they arrive [
15]. Some studies (including in India) have found that geographical access has a greater effect on utilisation than socioeconomic factors [
16,
17], particularly in rural areas with limited service provision [
18,
19].
India's Child Survival and Safe Motherhood Programme (CSSM), launched in 1992, involved training of physicians and traditional birth attendants (TBAs), provision of aseptic delivery kits and expansion of existing rural health services to include facilities for institutional delivery i.e. supplying essential equipment to district, sub-district and first level referral facilities to deal with high risk obstetric emergencies (MOHFW 1997-8). The initiative aimed to improve the proportion of pregnant women receiving three antenatal visits, and the proportion of deliveries conducted by trained attendants. The CSSM gave way to the Reproductive and Child Health (RCH) programme in 1997, at which point the scope was widened to include other reproductive and child health services [
20]. The second five year phase of the RCH programme (RCH II) is currently being initiated and contains a comprehensive newborn health strategy that includes promotion of institutional deliveries, with cash subsidies for poor families and compensation of TBAs facilitating the process [
21]. In areas remote from facilities, improvement of home-based newborn care via auxillary nurse-midwives is envisaged [
22,
23].
An investigation of the determinants of place of delivery in rural India is the primary objective of this paper. It adds to existing work by using a logistic model with village-level random effect and data for the whole of rural India. Previous studies used standard regression techniques which do not take into account the clustered nature of multi-level data at each level and can give misleading results in terms of both central estimates and their pecision. In particular this paper aims to assess the relative importance of community access and household economic status in determining place of delivery, while also examining the influence of birth order, mother's education and region. A secondary objective is to investigate the choice of provider amongst those who opt for institutional delivery and the impact of these same factors on this decision. Although a few deliveries take place in NGO or charity hospitals, the major choice is between private - for-profit and government providers.
Discussion and Conclusions
The institutional birth rate in India is extremely low even for those living within easy reach (5 km) of a hospital. Unless the pace of change accelerates, it will take until 2025 for half of all rural births to be institutional and mid-century before 75% coverage is reached. The national goal of achieving 80% coverage by 2010 is extremely optimistic and results from the 2005-6 NFHS-3 show a continuation of the slow rise but no sign of acceleration [
30]. Maternal health needs to become a political priority [
31]. The results of this analysis (Table
1) show that institutional care seeking for child birth in rural India is currently influenced by community access, economic status, education, region and birth order. While education and region show the strongest associations, the focus of this paper is a comparison of the influences of access and economic status and results show the latter emerging as a more crucial determinant. The impact of high delivery costs and distance to services as barriers to care seeking was highlighted in the 2006 Lancet Maternal Survival Series [
32,
33].
This importance of economic factors shown by predicted probabilities in Table
1 confirms the pattern found previously in both the North and South [
13,
34‐
36]. A study in Maharashtra found that the average expenditure incurred per delivery was Rupees 512 (US$11.6). Amongst those in the lowest socio-economic group this ranged from Rs.160 (US$4.0) if it was a home delivery to Rs.230 (US$5.8) and Rs.1,039 (US$26.1) if the delivery had taken place in public or private-for-profit institutions, respectively. Cost was found to be critical in influencing the decision to seek care and the differential in the cost of private and public care was important in the choice of provider [
12]. These figures, along with analysis in this paper demonstrating the influence of financial constraints on care seeking, provides support for the government policy to promote institutional delivery by providing cash transfers of US$17 to the poor. This transfer aims to cover travel and subsistence costs for pregnant women and their accompanying family members as well as the cost of care itself. Anecdotal evidence shows that it has been leading to an increase in institutional delivery [
37]. India's growing prosperity should also accelerate progress. A further factor favouring increased use of obstetric services is fertility decline. A greater proportion of births will be first births, for whom institutional delivery is much more common than for subsequent births.
However, the importance of economic status should not be taken as grounds for dismissing the importance of geographical access. This can have a crucial influence on the second delay, delay in reaching an adequate health care facility, as cited in the "three delays" model [
3]. Its significance has previously been demonstrated on a local level [
18]. For example Stephenson & Tsui (2002) found that in Uttar Pradesh the presence of a secondary health facility increased care seeking for both pregnancy and childbirth [
38]. However, the effect of access varies by state: a study focusing on rural Andhra Pradesh, Gujarat, Bihar and Rajastan found access to health services (measured by whether a hospital was available within 5 km of the village or not) to have a statistically significant effect only in Rajastan [
36]. Using NFHS-2 data at the national level geographical access has previously been found to be a weakly significant determinant of institutionally delivery [
35]. Previous analysis has used standard regression techniques, however, not taking into account the clustered nature of the data. The random-effects method used in this study, in which community level effects have been taken into consideration, finds a more significant association. It is expected however, that failure to account for clustering would overstate significance, and it is possible therefore that significance levels differ because of some other factor (sample selection, time period, etc).
It is important to note that this logistic approach with a village-level random effect does affect associations with socio-economic status as well as distance, making it necessary and valid. While communities will obviously share distance, they are also to a lesser extent likely to share wealth. As communities share distance more precisely it may be that effects of distance are diffused more than socio-economic status, but it is correct that this happens and it is probable that studies without such controls are giving unjustified emphasis to distance.
Physical proximity does not necessarily imply uptake, however. There is also no recognised definition of what constitutes reasonable access. If a 15 km criterion is used, then nearly two-thirds of rural Indians have access. This is an admittedly crude measure because it does not take into account the availability of motorised transport and roads. As noted earlier, the presence of a sealed road in each village was recorded in NFHS-2 but was found to have little influence on uptake of services. It has also been suggested that the influence of income and education would diminish as geographical access improves but interactions between distance and all other factors were tested for and none found. Expansion of services may therefore not be sufficient to promote utilisation. Even if there is latent demand for services, poor quality and high cost can inhibit utilisation. The absence of a relevant measure of quality was a limitation of this analysis. Reluctance to use institutional services may also be a problem with many mothers preferring to deliver at home even when services are affordable, accessible and of acceptable quality [
39,
40].
In India the public sector is perceived by many to be of low quality. The absence of even primary newborn care facilities, such as warming and resuscitation equipment, is common [
41]. The private sector suffers different problems; there has been a proliferation of practitioners, some with no recognised medical qualification, but, despite the sometimes dubious quality of care, the seeking of private health care is a sign of wealth and status. Services in general need to be made more user-friendly, higher quality and the community mobilised to utilise them [
40].
The other highly significant finding is the importance of the community context in determining the use of maternal health services. This probably reflects unobserved community-level social and cultural circumstances and service characteristics. Social interactions at this level may also have an effect, influencing people's attitudes and opinions regarding care seeking.
The very strong regional differences in place of delivery that exist even after adjustment for access, economic status, birth order and education suggest that there are further unexplained factors affecting perceived desirability or preference for institutional delivery in India. Demand for services is vital for utilisation to take place and, according to Chatterjee, it is created when permission and ability coincide. Education, is certainly influential in this as higher levels are often associated with greater autonomy [
42].
Studies have found that the perceived need for care is sometimes much lower than bio-medically defined need. The belief that delivery is a natural process not requiring medical attention is thought to be particularly strong in the North [
43]. The cost of services also varies regionally and hidden costs often in the form of under the counter payments inflate the cost of institutional delivery and act as a deterrent [
12].
The results in Table
2 show that wealth is the strongest factor affecting the decision between a private-for profit or public facility amongst institutional births. After adjustment only the presence of higher education makes private delivery more likely. In the East the public sector is certainly much more heavily relied upon than in other regions, particularly the South. The East suffers from low availability of private services which may reflect greater discouragement of the private sector in West Bengal, the largest Eastern state, which for many years has been governed by a communist party. Results also suggest that heightened concern over first births is conducive to increasing demand for, and choice of, a private institution.
In India, areas very remote from services undoubtedly need better provision and in the shorter-term the outreach of skilled birth attendants (a component of RCH II) offers a compromise. However, in most areas the first priority is to increase demand and maximize utilization of existing services. Educational attainment, which generates demand, is slowly increasing but with low economic status so clearly also inhibiting use of services findings suggest that demand side financing, as is already being trialed by the government through cash payments to poor women, shows great potential for increasing rates of institutional delivery. Future analysis could usefully assist policy makers more directly in deciding where to place finite resources. This could involve exploring the comparative impact on institutional delivery of investing in either cash subsidies or the building of new facilities in underserved areas.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
This work formed part of AK's doctoral research. She carried out the data analysis and drafted the manuscript. JC supervised the thesis and was instrumental in designing the study and guiding the analysis, he made substantial comments and revisions to the draft. AS was on the advisory committee and provided particular advice and support to the statistical analysis. CR was also on the advisory committee and provided expertise on maternal health. All authors read and approved the final manuscript.