In this study, we used the methodology proposed by Pitchforth et al. in 2007  to develop a proxy wealth index. Applying this index, we showed a lower rate of utilization of EmOC at Wolisso Hospital by poorer women compared to wealthier women. Inequalities in utilization of EmOC by wealth quintile and residence are not new findings . In the case of the utilization of EmOC services, social and financial barriers play a key role in many countries [25–27].
In our study, users of EmOC were of lower parity and more likely to be urban than rural dwellers. The rural–urban inequity is partly because the hospital is located in an urban setting. Given that it is the only hospital providing comprehensive EmOC in the region, rural women are disadvantaged by geography alone. The Oromiya region is crossed by one major road, and given that most women use public transportation to reach the facility, it is likely that women living further away from the main road had less access to it.
In an attempt to reduce barriers for poorer women, maternal health services provided by the hospital are largely subsidized for all women. For instance, the hospital charges 50 Birr for a normal delivery and 150 Birr for a caesarean section (1 Birr is about 0.06 US$). On the other hand, the mean cost of transport was 87 Birr with a cost for public transport of 55 Birr. Given that public transport cost was higher than the subsidized delivery fee, transportation expenditure is likely playing an important role in preventing poorer women from using the facility. Subsidizing transportation (for instance by using vouchers) is one potential mechanism to address this barrier.
The importance of non-financial barriers, such as concerns of quality of care or cultural issues, has also been described . Wolisso Hospital is supported by a non-governmental organization (Doctors with Africa-CUAMM) and has built a good reputation in the region over the years. It began its services in January 2001, with the number of beds increasing from 83 in 2001 to 192 in 2010. Normal deliveries and cesarean sections also rose from 391 to 2,532 and from 54 to 493, respectively, over the same period. In 2010, the hospital’s bed occupancy rate was 90%. This steady growth of the hospital may indicate that quality and cultural issues are being addressed more effectively than financial barriers.
Financial barriers in accessing maternal care in Ethiopia are due to both direct and indirect costs. Indirect costs are partly due to long distances to health facilities. About 40% of the population in Ethiopia lives within 1 Km of any health facility . Most women therefore have to pay for transportation costs; both for themselves and their companions, to access maternal care. Even though the Ethiopian government has tried to address the problem of direct costs by making maternal health services free of charge as stipulated in the HSDP III, most of the facilities that are supposed to be providing free services are only partially free in practice . For instance, some women are charged for supplies used during delivery. With 39% of the population in Ethiopia living on less than 1.25 US$ a day , these costs still present a significant barrier to service access.
Importantly, the questionnaire was convenient and could be administered in less than 10 minutes. Weighted scores for each woman can be calculated and the woman categorized into a wealth group using cut-off points for weighted scores, without the need for statistical analysis. Because of these important features, the questionnaire can serve two functional roles. It can be used to identify poorer women for targeted social-assistance programs that take into account local determinants of poverty. Identification of the most deprived individuals is essential in implementing equity-focused interventions [31, 32]. In relation to access to EmOC, attempts to better target poorer women for assistance has resulted in a number of innovative approaches, such as vouchers, equity funds and community health insurance [33, 34]. Additionally, through periodic, low cost surveys, this questionnaire can be used to monitor the extent to which such programmes are contributing to the reduction of inequity in EmOC utilization.
Variables that capture the living standards are widely used in developing countries to measure the socio-economic status of households. In the Ethiopia 2005 DHS, almost 50 variables were used to construct the wealth index. In our study, we selected 5 variables that can be used to quickly assess the socio-economic status of service users in a hospital setting. One major challenge in limiting the number of variables to measure the wealth status is the issue of clumping and truncation . Clumping is a situation whereby households are grouped into small number of distinct clusters while truncation is a situation where the socio-economic scores are spread over a narrow range. Both issues result into difficulties in differentiating between socio-economic groups. This is a typical problem in rural Ethiopia  where asset ownership and access to utilities is low and housing characteristics are the same for a majority of households. Pitchforth and colleagues  therefore suggest balancing the variables to include some whose possession in the poorer groups is average. Another solution is to increase the number of variables in the index , but this is not a good option if the goal is to a have an easy-to-construct and apply tool.
Our study has a number of limitations. The original plan to collect data from all women discharged from the hospital maternity in one full year was constrained mainly by the lack of human resources. During the study period (January – August 2010), the maternity ward sometimes ran out of beds and some women who had delivered without complications were discharged directly from the delivery ward without going through the postnatal ward. A good proportion of these women were thus missed. Of the 760 women who were interviewed, we could only trace back full data on maternity admission for 532 of them (i.e. 70%), because of problems in the routine hospital health information system. The women who were not interviewed (n = 748) did not differ from the interviewed sub-population (n = 532) based on age and district of residence. However, the two groups differed by mode of delivery and duration of hospital stay: the interviewed women being more likely to have a “complicated” delivery. Women with more severe maternal conditions are more likely to stay longer in the hospital and hence more likely to have been contacted for an interview irrespective of their socio-economic status. The effect of selection bias based on socio-economic status is thus thought to be minimal.
Although the administration of the questionnaire at discharge increased the convenience of data collection, it had the disadvantage of excluding the women who died and the runaway (i.e., women who left before being discharged). However, during the study period, hospital data showed six maternal deaths and no cases of runaway women. Seasonal variations in maternity service utilization may have affected our results, as the assessment did not cover a full year.
We revalidated the tool using the 2011 DHS data of parous women aged 15–49 years and usual residents of Oromiya. The validity of the tool had not changed much but its reliability had slightly reduced but still higher than what has been reported elsewhere ; implying that the tool may not have a stable reliability over time and hence the need to revalidate it using the most updated household survey data. One weakness of using variables that reflect living standards in the context of assessing access to health services is that these measures are reflective of cumulative household wealth and fail to take account of short-run or temporary interruptions to the households . Moreover, the ownership of an asset does not reflect its quality or value. This has the potential of misclassifying individuals.
Ensuring access to EmOC to women of all socio-economic status is critical to improve maternal and newborn health and outcomes. Monitoring equity in service utilization at the point of care is indispensable given the current challenges facing poorer communities. Results from this study show that monitoring equity is feasible by applying a simple tool developed using the open-access DHS databases, internationally acknowledged and virtually ubiquitous in developing countries. The evidence of inequity that is highlighted here should prompt the implementation of effective strategies to promote equitable access to health services. Use of this tool to monitor the distribution of maternity service users by wealth quintiles is recommended. Further studies are needed to explain the contextual causes of unequal access to EmOC, in order to understand how best to address it.