This study presented an analysis of the impoverishing effect of OOP health spending and factors associated with impoverishing OOP health spending in Ethiopia. In 2011, between 1.18% and 1.19% of the total population were forced into poverty in Ethiopia due to OOP health spending. When extrapolated at the national level based on population projection for 2011 , this represents over 957,100 individuals impoverished by OOP health spending. Also, the average poverty gap has deepened for those who were already below the poverty line before paying OOP for health services. Harari, SNNP, Somali, and Amhara regions had higher incidences of impoverishment due to OOP health spending, which is higher than the national average. However, SNNP, Oromia, and Afar had a more substantial rise in the depth of poverty, part of which is due to people who were already in poverty being pushed further into poverty. Similarly, rural residents and female-headed households had a higher proportion of individuals that were pushed below the national poverty line compared to their counterparts. It is important to note that the proportion of people living below the national poverty line before deducting OOP health spending (37%) is seven percentage points higher than the World Bank’s estimate (30%). Also, pre-payment poverty rates for the regions were also estimated at higher ratios . These disparities may be due to the difference in data sources. Nevertheless, these estimates show that about a third of the Ethiopian population is living below the national poverty line.
The results of this paper are not directly comparable with other studies due to differences in methodology, poverty lines, and/or data. Nevertheless, the results appear similar to those from some studies. A recent international study showed that impoverishment resulting from OOP health spending ranged between 1% point and 4% points . An earlier study using the 2004 data showed that OOP health spending impoverished less than 1% of the Ethiopian population . While that study used older data, the more recent (2010/11) HCES dataset shows that impoverishment from OOP health spending in Ethiopia is as high as 1.19% points. Other studies in Africa found a similar pattern of impoverishment from OOP health spending [31-35]. For example, in Ghana, J Akazili, JE-O Ataguba, EW Kanmiki, J Gyapong, O Sankoh, A Oduro and D McIntyre  report that between 1.6% points and 1.8% points of the Ghanaian population were impoverished by OOP health spending in 2005/06. In eSwatini, this was estimated to be between 1.0% and 1.6% of the Swazi population in 2009/10 . B Kwesiga, CM Zikusooka and JE Ataguba  found that as high as 4% of the Ugandan population was impoverished by OOP health spending in 2009/10.
The impoverishing effect of OOP spending in Ethiopia suggests that policies such as the fee waiver and health insurance schemes  have not led to reductions in the incidence of impoverishment from paying OOP for health services. Fee waiver beneficiaries still report under-utilisation and other barriers to seeking health services . Although introduced since 2003/04, the new fee-waiver system was not fully operational nationally in 2011 with only about 6% of people living below the poverty line screened . Based on the 2011 Ethiopian population, these were only about 1.8 million people . All things being equal, any individual from the remaining population living below the poverty line (22.9 million) would be pushed further into poverty as a result of OOP health spending, if health services are needed. Fee-waiver does not also necessarily provide the safety net it is supposed to. It was revealed that fee waiver beneficiaries pay for healthcare despite being entitled to free care due to perceived poor quality of free services, unavailability of drugs and diagnostic procedures, and to avoid the social stigma of being labelled the poorest . Previous studies also show that fee-waiver and exemption systems are ineffective in providing financial protection to the poor [10, 37, 38]. It should also be noted that the fee-waiver system does not cover non-medical costs of care-seeking (e.g. transportation, food, and accommodation), and the opportunity cost of care-seeking. These costs can be substantial and can increase the burden of healthcare payments . Thus, impoverishing healthcare payments may lead to forgoing needed healthcare services, which further aggravates health problems. Households may also resort to coping strategies such as borrowing, distress sale of assets, and reducing consumption of necessities . Therefore, OOP health spending is a barrier to achieving not only the health-related SDGs but also the other goals such as ending poverty, hunger, and food insecurity by trapping people in the vicious cycle of poverty .
Similar to people below the poverty line who are pushed deeper into poverty due to OOP health spending, the burden faced by other populations – especially those near poor – should also be appreciated. Households who are not entitled to fee-waiver or covered by health insurance and are closer to the poverty line can easily fall into poverty if they incur OOP healthcare costs. Therefore, they would also be either ignoring healthcare needs or using similar coping strategies to reduce healthcare payments and thereby avoid poverty. This study showed that the risk of impoverishment is higher for rural residents than for urban residents. This can be explained by the interrelated factors of the higher burden of disease in rural areas coupled with inadequate means for paying for healthcare services due to subsistence living [41-43]. Sub-populations that are vulnerable to impoverishment from OOP health spending include households with many dependents (either large household size) and/or households with under-five children. Under-five children may need more healthcare services, which increase OOP health spending . Among household characteristics that are associated with a lower probability of impoverishment are living in households headed by a male and/or educated persons. Other studies from China, India, Nigeria, and Vietnam [45-48] reported similar results. This is an indication of gender-based differences in the burden and impact of OOP health spending to the disadvantage of women. This also highlights the importance of the social determinants of health in influencing healthcare need and thereby OOP health spending and impoverishment. Based on this study, age and having older people in a household were not related to impoverishment. However, K Koch, C Pedraza and A Schmid  and S Ahmed, S Szabo and K Nilsen  found that having older adults and a household headed by older persons increase the likelihood of impoverishment from OOP health spending. It remains unclear why these variables were not significant in the Ethiopian case.
Although the Ethiopian health insurance reforms were at their early stages when the data used for this study were collected, a study that evaluated the CBHI pilot programmes provides some insights on the affordability of the schemes. The evaluation showed that although premiums were subsidised, especially for the poorest of the poor, 39% of non-CBHI members indicated the unaffordability of premium as a barrier to enrolment. Premiums were also unaffordable for about 16% of CBHI members . In fact, between 2012-2013, more than a quarter (26%) of households in the pilot districts, mainly those in the lowest income quintile, dropped out of CBHI schemes because they could not afford to pay the monthly premiums . Thus, the estimated impoverishing impact of OOP health spending reported in this paper may be an underestimation .
Although this study used the 2011 data, the latest HCES data that were available at the time of the study, the results reported in this paper are very likely not to have changed significantly as OOP payments remained high in 2016; accounting for 37% of current health expenditure  . Based on the results of this paper, there is a need for the government of Ethiopia to aim to reduce the share of OOP health spending in overall health financing in the country. This reduction is achievable through a well-functioning pre-payment system for health services. If there is no reduction in OOP health spending, all things being equal, more than 1.2 million people will be pushed into poverty in 2020 based on poverty headcount estimates of this study and population projections for 2020 . Thus, prepayment systems need to be expanded urgently. Lessons from the CBHI pilot and other health financing reforms could be used to design prepayment systems that guarantee access to affordable and effective health services for all in Ethiopia. One challenge for providing adequate financial protection in Ethiopia is the large share (>80%) of the population engaged in either rural agrarian activities or the informal sector [42, 51]. This notwithstanding, Ethiopia can consolidate on the already declining share of OOP health spending in current health financing and the SHI framework to ensure that financial protection is realised, especially for low-income households and those that are unable to afford health services through an effective cross-subsidisation mechanism. The strength of this paper is the use of a nationally representative and the latest available HCES data. The study also presents a comparative analysis of poverty levels using both international and national poverty lines. There are also a few limitations. OOP data may have been under-reported since health expenditure was aggregated at the household level and due to recall bias. The survey also excluded those who did not seek needed health care due to unaffordability. Thus, impoverishment resulting from OOP health spending may have been under-estimated.